Comércio ao Vivo:

Report No. 36267-TH Thailand Investment Climate, Firm Competitiveness and Growth June 14, 2006 Poverty Reduction and Economic Management Sector Unit East Asia and Pacific Region Document of the World Bank THAILAND: INVESTMENT CLIMATE. FIRM COMPETITIVENESSAND GROWTH TABLE OF CONTENTS Executive Summary………………………………………………………………………………………………………… i Chapter 1: Growth Performance and the Business Climate…………………………………………….. 1 A. Introduction…………………………………………………………………………………………………………. B. Thailand’s Business ClimateandFirmPerformance ………………………………………………… 1 C. Thailand’s Business Climate………………………………………………………………………………….. 6 8 E. Conclusions………………………………………………………………………………………………………… 45 D. Determinantsof FirmPerformance……………………………………………………………………….. 27 References………………………………………………………………………………………………………………. 46 Chapter 2: Regional Investment Climate and FirmPerformance ………………………………….. A. Introduction……………………………………………………………………………………………………….. 47 47 B. Overview of RegionalDevelopmentinManufacturing……………………………………………. 48 52 D. Determinants of RegionalCompanyLevelTotal FactorProductivity (TFP): Company C. PIC Survey and Descriptive Statistics …………………………………………………………………… 59 E. Conclusions……………………………………………………………………………………………………….. Characteristicsand the InvestmentClimate……………………………………………………………. References………………………………………………………………………………………………………………. 63 64 Chapter 3: Supplying Skills for Competitiveness………………………………………………………….. 65 A. Introduction……………………………………………………………………………………………………….. B. Assessing Thailand’s Skills Adequacy: The Macro-Evidence…………………………………. 65 C. Assessing Thailand’s Skills Adequacy: The Micro Evidence………………………………….. 67 69 D. Determinants of Skills Adequacy …………………………………………………………………………. 80 E. ConclusionsandPolicy Implications…………………………………………………………………….. References………………………………………………………………………………………………………………. 84 87 Chapter 4: StrengtheningTechnologicalCapabilities……………………………………………………. A. What i s Thailand’s CurrentTechnologicalPerformance?………………………………………… 89 91 B. TechnologicalCapabilities Index (TCI): Towards a BroaderDefinition of Technology101 . ………………………………………………………….. References…………………………………………………………………………………………………………….. C ConclusionsandPolicy Recommendations 116 118 Chapter 5: ICT and FirmPerformance inThailand……………………………………………………. References…………………………………………………………………………………………………………….. 119 137 Appendices 138 Chapter 2 Appendix 1:………………………………………………………………………………………………….. Chapter 1Appendix……………………………………………………………………………………………………… 166 168 186 Chapter 4 Appendix …………………………………………………………………………………………………….. Chapter 3 Appendix:…………………………………………………………………………………………………….. Chapter 2 Appendix 2:………………………………………………………………………………………………….. Chapter 5 Appendix……………………………………………………………………………………………………… 191 196 Appendix – SamplingDesignandMethodology………………………………………………………………. 198 List of Tables Table 1.1: Sources of Growth. 1977-2004 …………………………………………………………………… 3 Table 1.2: Sources of Growth, Major Sector, 1977-2004……………………………………………….. Table 1.3: Sources of Growth inEast Asian Economies, 1975-2000……………………………….. 4 5 Table 1.4: Number of Daysto ObtainLicenses/Permits/Approvals/Certificates to Start a Table 1.5: Number of Daysto ObtainDifferent LicensesA`ermits/Approvals/Certificates … 19 Business………………………………………………………………………………………………………………. 18 Table 1.7: Number of Days to Process Application for Different Export Incentives …………20 Table 1.6: Number of Days to ObtainDifferent Approvals and Documents……………………. 19 Table 1.8: CorrelationamongFirmPerformanceMeasures ………………………………………….. 30 Table 1.9: Correlates of FirmPerformance…………………………………………………………………. 32 36 Table 1.11: Benefits from RelaxingSkill Shortages……………………………………………………… Table 1.10: Correlates of FirmPerformanceUsingthe MOIPanel…………………………………. Table 1.12: Business Climate and FirmPerformance…………………………………………………… 41 Table 2.1: FirmCharacteristicsby Region………………………………………………………………….. 44 Table 3.1: Scientists andEngineersinR&D (per millionpeople) ………………………………….. 53 68 Table 3.2: EstimatedSheepskinEffects inThailand andMalaysia………………………………… 72 Table 3.3: EstimatedSheepskinEffects inThailandand Malaysia, SelectedWorker 74 77 Table 3.5: Skills Mismatch-Thailand PICS Worker Survey 2004…………………………………. Table 3.4: Workers’ Self Assessment of Skills Adequacy…………………………………………….. Characteristics(Dependent Variable: Individual LogHourly Wage)………………………….. Table 3.6: EducationMismatch………………………………………………………………………………… 79 Table 3.7: TIMSS Score…………………………………………………………………………………………… 79 81 Table 3.8: PISA Score……………………………………………………………………………………………… 81 Table 4.2: Rankingof Technological Readiness and InnovationSub-Indexes……………….. Table 4.1: Selected TechnologicalInput Indicators……………………………………………………… 92 101 104 Table 4.4: TCI for the ThailandPES Sample…………………………………………………………… Table 4.3: Illustrative Matrix of TechnologicalCapabilities by La11(2002) ………………….. 105 107 113 Table 4.7: Workforce Skills RegressionEstimates…………………………………………………….. Table 4.6: TCI RegressionEstimates……………………………………………………………………….. Table 4.5: DescriptiveStatistics of TCI inThai Manufacturing…………………………………… Table 4.8: TFP RegressionEstimates……………………………………………………………………….. 114 115 Table 5.2: ICT and Innovation………………………………………………………………………………… Table 5.1: TFP Regressions ……………………………………………………………………………………. 122 123 Table 5-3: Constraintsto Introducingor ExpandingIT Use Considered “Important” Table 5.4: Policy Matrix………………………………………………………………………………………….. or “Very Important” …………………………………………………………………………………………… 124 136 List of Figures Figure 1.1: Sources of Growth inthe Total Economy. 1977-2004…………………………………… 3 Figure 1.2: The Ratio of Private Investmentto GDPhasFallen Dramatically since the Crisis ………………………………………………………………………………………………………. 7 Figure 1.3: Panel A. FDIto GDP Ratio inThailand. 1980-2003 Panel B: Thailand’s Share inTotal FDIinto East Asia and Pacific. 1980-2003 …………………………………………… 7 Figure 3.4: Thailand Compares Favorably to other Countries inTerms of Regulatory Panel B. Percentage of Management’s Time Spent Dealing with Regulation………………… Burden: Panel A. Number of Days to Clear Customs for Exports 9 Figure 1.5: Comparison of Business Climate Constraints inThailand. Malaysia and 9 Figure 1.6: Concerns about Business Climate -Results from Closed Question……………….. Indonesia ………………………………………………………………………………………………………………. 11 Figure 1.7: Firms’ Concerns about Business Climate -Results from Figure 1.8: Political Stability and Rule of Law Governance Indexes……………………………… Open-Ended Question …………………………………………………………………………………………… 11 Figure 1.9: Governance Indexes for Thailand: 1996, 1998, 2000, 2002 and 2004…………….13 13 Figure 1.10: Regulation of Labor Markets and Regulation of Entry……………………………….. Figure 1.11:Why are Thai and Malaysian FirmsOver-staffed or Under-staffed? …………….. 15 16 Figure 1.12: HiringProcedures are the Biggest Labor Regulations Obstacle for Thai Firms. 17 Figure 1.13: Thai FirmsFace aLesser Regulatory BurdenThan Many other Countries …….21 Figure 1.14:Thailand Performs Well inTerms of Property Registrationbut Figure 1.15: Government Effectiveness and Regulatory Quality Governance Indexes……….22 Worse inAreas of Contract Enforcement ………………………………………………………………… 23 Figure 1.16: Control of Corruption Governance Index………………………………………………….. 23 Figure 1.17: Thailand RanksPoorly Relative to Other Countries inTerms of Key Infrastructure Concerns: Panel A: Frequency of Power outages PanelB: Number of Days Figure 1.18: Thailand Lags Behind Other Countries inTerms of Electricity Supply………….25 to Obtain an Electricity Connection ………………………………………………………………………… 25 Figure 1.19: Although More Frequent, Power Outages Appear to Cost Thai Firms Comparatively Less than in Other Countries ……………………………………………………………. 26 Figure 1.20: Telecommunications are also Poor inThailand by International Standards Panel B: Frequency of Phone Interruptions Last Year ……………………………………………….. Panel A: Number of Days to Obtain a Phone Connection 26 Figure 1.21: Thailand Lags Malaysia, South Korea and Singapore inMobile and Fixed Telephone Penetration…………………………………………………………………………………………… 27 Figure 1.22: MedianLabor Productivity (Value-Added per Worker in2001 U.S. dollars) in Different Industries ………………………………………………………………………………………………. 28 Figure 1.23: MedianLabor Productivity (Value-added per Worker in 2001 U.S. dollars) in Garments and ElectronicsElectrical Appliances Industries………………………………………… 28 Figure 1.24: Business Climate Concerns by FirmAge and Size: PanelA: Age; PanelB: Size…………………………………………………………………………………… 38 Figure 1.25: Business Climate Concerns by Export Status and Ownership: PanelA: Exporter Figure 1.27: Distributionof Skill M i x across Industries: Food Processing and Textiles…….40 Figure 1.26: Business Climate Concerns by Usage of Computer-Controlled Machinery…….39 Status; Panel B: Foreign-Ownership Status ……………………………………………………………… 39 Figure 1.28: Distributionof Skill M i x across Industries: Garments and Auto-Parts …………..40 Figure 1.29: Distributionof Skill M i x across Industries: Electronics/Electrical Appliances and Rubber/Plastics ………………………………………………………………………………………………. 41 Figure 1.30: Distributionof Skill M i x across Industries: Wood/Furniture and MachineryEquipment…………………………………………………………………………………………… 41 Figure 1.31: Benefits from Relaxing Skills Shortages and Number of Days to FillVacancy 42 Figure 2.1: Regional Share of Manufacturing GDP, 1981-2004…………………………………….. for a Professional………………………………………………………………………………………………….. Figure 2.3: Regional Herfindahl Index, 1996/7 and 2001/2…………………………………………… Figure 2.2: Spatial Distribution of ManufacturingEmployment, 1996/7 and 2001/2 ………..49 50 51 Figure 2.4: SpatialDistributionof Employmentof PICS Industries……………………………….. 51 Figure 2.5: Firms’ Opinions of InvestmentClimate inOther Regions Relative to Their Own ……………………………………………………………………………………………………….. 54 Figure 2.6: Aspects of Regulationthat FirmsRate as Representinga “Major” or “Very Severe” Constraint on Their Operations ………………………………………………………………….. 56 Figure 2.7: Bangkok. Centraland the East Considered the Regionswith the Best Business Climate; the South Considered the Worst…………………………………………………… 57 Figure 2.8: FirmsEstimate that ProductionCosts CanVary by Almost 40 percent Figure 2.9: InvestmentClimate Residual…………………………………………………………………….. Dependingon the RegionalBusinessclimate …………………………………………………………… 57 Figure 2.10: Low-TechInvestmentClimate Residual…………………………………………………… 62 62 62 Figure 3.1: Skills are a Major Constraint for Thai Firms………………………………………………. Figure 2.11: High-TechInvestmentClimate Residual…………………………………………………… Figure 3.2: ThailandPerformsPoorly on Secondary Education…………………………………….. 65 68 Figure 313: MeanLogHourly Wage by Years of FormalEducation………………………………. 70 Figure 3.4: Returns to University Educationare ConsistentlyHigher Than for Lower Levels of EducationinThailand …………………………………………………………………………….. 71 Figure 3.5: Vacancies for Skilled Workers andProfessionalsAppear Hardto Fillin ThailandComparedWith Other Countries………………………………………………………………. 73 Figure 3.6: FillingVacancies for SkilledWorkers andProfessionalsTakes a Relatively 75 Figure 3.7: InadequateSkills are the Key Driver of Vacancies ……………………………………… Long Time inThailand………………………………………………………………………………………….. Figure 3.8: The Extentof English Skills Mismatch……………………………………………………… 76 Figure 3.9: Managers inThailandBelieveTheir Workers HavePoor Skills……………………. 78 80 Figure 3.10: Training Dynamics………………………………………………………………………………… 82 Figure4.1: R&D ExpendituresandLevelof Development…………………………………………… 93 Figure4.2: Percentageof ManufacturingEstablishmentsReportingR&D Figure4.3: Numberof Researchers andLevelof Development…………………………………….. Activities in2002…………………………………………………………………………………………………. 94 Figure4.4: FirmsEmploying Staff Exclusively for R&D in2002-2003 (percent)…………….95 Figure4.5: Patents Issuedby the U S to SelectedEast Asian Countries………………………….. 95 96 Figure 4.6: PatentsFiled inthe United States ……………………………………………………………… Figure 4.7: HighTechExportsi s a MisleadingIndicator of TechnologicalPerformance……97 Figure4.8: Kernel DensityPlotsof TCIby Region……………………………………………………. 98 Figure4.9: Kernel DensityPlotsof TCIby Industry………………………………………………….. 110 110 111 Figure 5.1: KernelDensities of the Employment Share of Skilled Workers. Sample of Figure4.10: TCIby Size, Ownership,Export Status, and IndustrialEstates…………………… 125 Figure 5.2: KernelDensities of Employment Share of SkilledWorkers. Sample LargeEstablishments ………………………………………………………………………………………….. 125 Figure 5.3: Computer Use by Size of Firm2001……………………………………………………….. of Small-medium-sizedEstablishments…………………………………………………………………. Figure 5.4: ICT Use by Sector ………………………………………………………………………………… 126 Figure 5.5: MainLines per 100Peopleand GDPper CapitaPurchasingPowerParity…… 126 129 Figure 5.7: BroadbandPrices per Month, 2003 …………………………………………………………. Figure 5.6: Cost of Business Telephone ConnectionandMonthly SubscriptioninUSD…. 130 130 Listof Boxes Box 3.1: MeasuringSkills for Work ………………………………………………………………………….. Box 2.1: BusinessCase Studies of NortheastExporters……………………………………………….. 58 69 Box 3.2: The Thai EducationSystem…………………………………………………………………………. 83 ACKNOWLEDGEMENTS This report is the fruit of the continuous collaboration between the National Economic and Social Development Board (NESDB) and the World Bank. It i s written by a team at the World Bank which includes Albert Zeufack (Team Leader), Magdi Amin, Kirida Bhaopichitr, Amadou Dem, Ana Margarida Fernandes, Tenzin Norbhu, Kaspar Richter, Charles Udomsaph, Khuankaew Varakornkarn, and consultants who include Barry Bosworth (The Broolungs Institution) and Paul Welsh. The team received overall guidance from Kazi Matin, World Bank Lead Economist for Thailand. Peer Reviewers from the World Bank are Aart Kraay, William Malone, and Shahid Yusuf, Productionof the report was supportedby Hedwig Abbey andFraser Thompson. The report is basedon data from the Productivity and Investment Climate Survey (PICS), a large survey of 1,385 establishments in Manufacturing and 100 in ICT, and 1,385 workers, conducted under the supervision of the Foundation for Thailand Productivity Institute (FTPI) with the help of the World Bank between March 2004 and February 2005. The survey covers six regions in Thailand and nine industries. The PICS included interviews with CEOs, Chief Financial Officers, Human Resource managers, and workers. Additional sources of information were used to complement the PICS. The Productivity and Investment Climate Survey instrument was designed by Albert Zeufack. Charles Udomsaph managedthe database for the report team. The implementation of the survey and the production of the report benefited greatly from the collaboration of Government officials inThailand. The survey was funded by the NESDB as part of the National Competitiveness Committee’s agenda. The team would like to thank in particular Dr. Phanit Laosirirat and the team at (FTPI) for the hard work accomplished during the survey, Dr. Ampon Kittiampon, Mr. Arkhom Termpittayapaisith, Ms. Wilaiporn Liwgasemsarn, Mr. Panithan Yamvinij, Mr.Thanin Paem, Ms.Pojanee Artarotpinyo, and Ms.Montathip Chanpum (NESDB) and the Steering Committee for the Productivity and Investment Climate Study for their valuable comments on the report and support of the report’s productionprocess untilits completion. Also, the team i s indebted to Dr. Piyanuch Wuttisorn (NESDB) and Mr.Natapat Lopraditpong (FTPI) for their excellent assistance all along the process, and their valuable inputs, including during their visit to Washington DC for the preparation of the report. THAILAND: INVESTMENTCLIMATE, FIRM COMPETITIVENESSAND GROWTH EXECUTIVESUMMARY 1. Thailand’s private investment recovery after the 1997-98 Asian financial crisis has been sluggish. Until 1996, Thailand’s growth was characterized by high rates of total investment. Private investment recovery after the crisis has been slower than in past recoveries and the current rate of such investment remains low despite a strong economic recovery after 2001. In 2004, private investment reached only about 15 percent of gross domestic product (GDP), compared with an average of more than 25 percent inthe periodprior to the crisis. 2. The nature of Thailand’s growth has also changedinrecentyears. Prior to the financial crisis, as with most East-Asian economies, growth was dominated by highrates of capital accumulation. Between 1977 and 1996, total factor productivity (TFP) growth contributedan average of 1.6 percent per year to the growth of output compared with four percentage points from capital and 1.6 percentage points from increased employment. Improvements in labor quality, as measured by gains in educational attainment represent 0.3 percent per year, a relatively minor source of growth. However, since 1999, Thailand’s growth has been driven by the increasing employment of its large reserves of underemployed rural labor in the manufacturing sector. As competition from neighboring countries in low-skill manufacturing increases, it i s likely that future demand for unskilled labor (UL) will decrease. To sustain strong economic growth and competitiveness in the medium-to-long-run, Thailand may need to embark on a different strategy. On the one hand the country should continue on the path of increasing private investment, but on the other hand, it should make greater efforts to strengthen the contribution of TFP growth to output growth by emphasizing innovation, and developing skills and technological capabilities. 3. This report addresses the following questions: (i) can be done to bolster What private investment? (ii)What should be done to ensure a more efficient geographic distribution that respects agglomeration effects and economies of scale but which i s not biased by public sector distortions? (iii)How to improve firms’ technological and innovation capabilities? Together these three issues constitute the core of Thailand’s competitiveness agenda, a center piece of the country’s development strategy. i A. Boostingprivate investment and productivity by improving the businessclimate. 4. Improving the business climate is key to reducing risks and costs, as well as increasing investment, productivity, and the competitiveness of the Thai economy. A good business climate provides firms with the incentives to invest, innovate and grow while providing individuals with incentives to invest in skills valued by dynamic and growing firms. This study finds that Thailand’s investment climate i s favorable by international standards. It i s better than that of China, India, Braziland most neighboring countries, but not as good as that of Malaysia. Thailand cannot be contented with this performance. Its objective should be to improve the investment climate further so as to compete more effectively in the region and the world, if it i s to realize its long-term vision. 5. Thai firms have confirmed that their operations are still hinderedby three major constraints: heavy regulatory burden, shortages of skills and deficient infrastructure especially outside Bangkok. As shown in Figure 1, more than 60 percent of the 1,385 firms surveyed have identified regulatory burden as a top obstacle to doing business. About 50 percent of firms identified skill shortages and about 40 percent, infrastructure deficiencies as the two other most bindingconstraints to their operation and performance. Skill shortage i s a key obstacle to operations, productivity and growth of firms in all regions, while deficiencies in infrastructure and support services are most serious for firms in the worst-performing regions. The majoi- obstacle for firms in the better performingregions such as Bangkok and Vicinity i s heavy regulation. Figure 1:Major BusinessClimate Concernsfor ThaiFirms Regulatory Burden Skilled Labor Shortage Infrastructure and Support Services Dissatisfaction with Economic Situation 0 10 20 30 40 50 60 70 Percent o f Firms Identifying Issues as One of Three T o p Obstacles 6. Specific issues of regulatory burden that are identified by firms include tax regulations and/or rates, bureaucratic burden, labor regulations, import and customs regulations, and ownership regulations. Firms in Thailand also experience a large degree of uncertainty associated with the time needed to obtain licenses, permits and authorizations. Thailand’s performance in infrastructure i s particularly weak in respect of electricity and telephone-related indicators. Firms face a higher frequency of power 11 .. outages and a longer time to obtain electricity connections than firms in comparator countries. Also, firms inThailand suffer more interruptions and experience a longer delay in obtaining a fixed telephone connection than firms in neighboring countries. The concerns about skills are pervasive across firm sizes and are felt equally by exporters and non-exporters, domestic and foreign-owned firms, as well as by all Thai regions. This i s in part because the nature of labor-skills demanded is changing as the manufacturing sector transitions from mainly labor intensive products to higher technology products. These concerns about shortages of skilled labor (SL) are consistent with the evidence on relative educational attainment. For example, the completion rate of secondary education inThailand, which was 4.1 percent in 2000 compared with 23.6 percent inMalaysia and 17.5 in the Philippines, suggests the secondary education has failed to keep pace with Thailand’s growing needs. B. Increasingproductivity throughefficient geographic distributionof activities. 7. The manufacturing activity has increasingly shifted from Bangkok and vicinity to the East and Central regions. Of all South-East Asian capital cities, Bangkok has the highest concentration of a nation’s economic activities. Nevertheless, the role of Bangkok and vicinity as Thailand’s factory-hub has clearly declined over the last 25 years, and at a faster rate after the crisis. From an output share of 65 to 70 percent in the 1980s, Bangkok and vicinity share fell to 46 percent in 2004. The East’s contribution has exceeded Bangkok’s since 1996, and the Central’s contribution topped Bangkok’s share since 2003. The role of Bangkok as Thailand’s factory hub has declined. The shift i s significant from the perspective of both output and the employment, see Figure 2. Figure 2: RegionalBreakdownof Thai ManufacturingGDP 1981-2004 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% , 881 1982 1883 ~~ l1 InFigure 1.8 (Figure 1.9), these marginsof error are shown by the vertical (horizontal) bars corresponding to 90 percent confidence intervals. If the margins of error for an indicator across two years are mostly non-overlapping, thenthereis a significant changeinthe indicator. 13 1.25 However, Thai firms’ concerns with the overall economic situation do not appear to be structural and can be rationalized by several factors.12First, managers likely considered non- macroeconomic factors as part of the overall economic situation: (i) the SARS crisis (which was ending when the PICS was conducted); (ii) the outbreak of avian influenza; and (iii) Thai the Southern region unrest.l3Indeedthe aforementioned deterioration of the Political Stability Index and the Rule of Law Index are very likely associated with the Southern region unrest. Second, although macroeconomic performance was solid in 2004, it was worse than in 2003 in terms of GDI and investment gr0~th.l~ Moreover, the rise in interest rates, and particularly oil prices in 2004, likely led to a decline in business confidence. Third, since the Asian crisis, Thailand engaged in negotiations of a multiplicity of free-trade agreements with a very diverse set of trading partners that are still far from concluded. This likely introduced uncertainty in the business climate, particularly for current and potential exporters, and for foreign-owned firms. Fourth, the concerns about the overall economic situation are felt very strongly by firms in the garments industry. The end of the Multi-Fiber Agreement (MFA) quotas in January 2005 and the ensuing increase in competition from Chinese exports in developed country markets were likely at the forefront of garment firms’ managers concerns in the PICS. Additional support to the claim that Thai managers’ concerns about the overall economic situation are not structural i s found in the high percentage of Thai firms — 59 percent — that report in the PICS that they expect to make substantial increases in investment in order to increase capacity or improve quality in a near future. We now delve in more detail into each of the three crucial business climate obstacles. Skills Shortages 1.26 Concerns about shortages of S L are consistent with internationalevidence on educational attainment. Data from Barro and Lee (2000) suggest that the percentage of Thailand’s population aged 25 and older who have at least completed secondary education i s substantially lower than the norm for its income level. This gap has narrowed slightly over time, but only 15.3 percent of Thailand’s population had secondary or higher education in 2000, which i s still well below Malaysia (29.9 percent) or the Philippines (3 1.9 percent). The shortage of S L inThailand will be analyzed in detail in Chapter 3. The importance of the SL shortage as an obstacle to Thai firms will be confirmedbelow by the analysis of labor market regulations. l2Also, note that the PICS questions related to the overall economic situation are phrased in a very general manner: e.g., macroeconomic instability and economic policy uncertainty. Thus, it i s difficult to know exactly what each manager considers when answering those. In fact, macroeconomic instability and economic policy uncertainty are also among the top three concerns of firms surveyed inBrazil, China, Indonesia, Malaysia and the Philippines. Since these countries exhibit very diverse macroeconomic behavior and policy uncertainty, this suggests that one cannot draw policy conclusions from such concerns. l3Ultimately, these factors had minimal impacts on growth. l4In 2003, Thailand exhibited the highest GDP growth since the crisis (6.7 percent) coupled with a reduction in external vulnerabilities, very strong export growth (17 percent) and private investment growth (18 percent). 14 Regulatory Burden 1.27 Government effectiveness, in terms of the quality of policies and regulations, has far- reaching influence on the cost of doing business and thus on manufacturing firm operations and performance. While regulations play a role in mitigating market failures, if they are excessive, complex and costly, they can become harmfulto firms. In Thailand, a large percentage of firms complain about general as well as specific aspects of the bureaucratic burden as a major obstacle to their operations and growth. Concerns with bureaucratic burden are not unique to Thailand, however. They are also crucial for firms in China, Indonesia, Malaysia and the Philippines, according to the PICS. 1.28 Labor regulations are part of the regulatory and bureaucratic burden faced by firms which may increase their costs and reduce their flexibility. According to the 2004 Doing Business indicators published by the World Bank and shown in Figure 1.10, Thailand’s labor regulations are not overly restrictive when compared with those in India or Indonesia, but they are quite restrictive when compared with those in China, South Korea, Malaysia or Singapore.” This finding suggests that even in sectors where Thailand has a comparative advantage but particularly in low-end manufacturing industries, labor market flexibility can hamper the competitiveness of its firms in global markets where firms from countries with more flexible labor markets compete and may adjust more easily to constantly changing demands. Figure 1.10: Regulation of Labor Markets and Regulation of Entry . 2M 1 I I zoo i . I 180 25th Percentile IIndex of Labor Market $ 160 8Y II Regulations =26.2 . DIndon& Brah . .m I aa . 1M II E . ……. m 3100 . II 9 .I I .. . :Inaia …. . . .. v) 80 I . 8 na h I . . .——-.-.. 60 ..I D 40 . … .. … .. … I1. m . 1.Philippiw rn m China . m 25th Percentile Malayda I. LThai1ar;d. ‘ O’ Start.Business = 26 Days 20 –L—17–$..-.-ICPPeB.-.–:–D . …. .:. -I——. Sin a ere .)%. … … m . 0 I !I 0 10 Index of Overall Restrictiveness of Labor Market Regulations90 20 30 40 50 60 70 80 100 Source: World BankDoing Business Indicators2004. l5This index is basedon information aboutthe difficulty inhiringandfiring workers andthe rigidity of regulations on working hours. The specific labor market aspects covered by the index are the availability of part-timeand fixed- term contracts, requirements on working time, minimumwage laws and minimumconditions of employment. 15 1.29 Specific information from the PICS provides additional insights on labor market regulations in Thailand. Firms were asked whether they were over- or under-staffed and the reasons why, Their responses are summarized in Figure 1.11, On average, Thai firms are under- staffed, Le., they would like to have a workforce three percent larger than their current size.16 However, a majority of managers believe their firms are currently operating at their optimal employment levels.” Labor regulations do not play a role in explaining firm over-staffing, as the majority of firms points out to the expectation of an upturnin sales as the reason for being over- staffed. Labor regulations are more relevant in explaining firm under-staffing, but shortage of skilled workers and the anticipation of a downturn in sales are the most important reasons for that. Also shown inFigure 1.11are the responses to a similar question posed to Malaysian firms inthe 2002 PICS. It is very clear that labor marketregulations regardingeither firing or hiringof workers are much more relevant in Malaysia than in Thailand as an explanation for the over- or under-staffing of firms. Figure 1.11:Why Are Thai and MalaysianFirmsOver-staffed or Under-staffed? Reasonsfor Being Over-Staffed Reasonsfor BeingUnder-Staffed 1 9 0 1 & – Difficulty in Shortage of skilled Anticipationof a Laws and regulations on Anticipationof anupturn employing local workers dormtuminsales firing of workers insales workers 1 Wlhailand OMalaysia 1 1UThaiiand OMalaysia 1 1.30 Figure 1.12 shows Thai and Malaysian firms’ rating of five specific labor regulations according to the severity of each to their operations and growth. Hiring procedures for local workers are a major obstacle for more than 20 percent of firms in Thailand, which i s a much higher fraction of firms than in Malaysia. However, most Thai firms do not consider other types of labor regulation to be problematic while many Malaysianfirms do so.18Overall, these findings l6This could be viewed as an attitude of caution, given Thai firms concerns with the overall economic situation. l7Specifically, 707 out of 1385 Thai firm managers say that they would like to have a workforce exactly the size of their current workforce. l8The importance of labor regulations in Thailand differs across industries. Dealing with hiringproceduresfor local workers is a stronger concern for firms in the garments and rubbedplastics industries, while dealing with those procedures for foreign workers is more important concern for firms in the food processing industry. Finally, an inflexible salary scale for skilled workers is a more significant obstacleto firms inthe garments industry. 16 provide two different messages. On the one hand, they reinforce the importance of S L shortage as an obstacle for Thai firms. On the other hand, they suggest that while Thailand has more restrictive labor market regulations than some of its competitors, for firms in the PICS, labor market regulations are not a fundamental problem, particularly when compared with firms in Malaysia. This does not imply, however, that a detailed evaluation of Thai labor market regulations i s not necessary, rather it suggests that other aspects of government regulation may pose a more bindingconstraint to firm operations. Figure 1.12: HiringProceduresare the BiggestLabor Regulations Obstaclefor Thai Firms 2- 45.0 d’e:E 40.0 – m :'[ 35.0 – 30.0 – 25.0 – n n Hiringproceduresfor Hiringproceduresfor Layoff proceduredcostLimitsontemporary Intlexiblesalary scale local workers foreign workers of retrenchment hiring for skilled workers 1 mThailand OMalarsia 1 1.31 Regulations affecting the entry and exit of firms reduce the competition faced by incumbent firms and their incentives to become more efficient. According to the 2004 Doing Business indicators shown in Figure 1.10, the time required to start a new business in Thailand (33 days and eight steps) i s close to that in Malaysia (30 days and nine steps) and lower than that inChina (41 days and 12 steps), the Philippines (50 days and 11steps) and Indonesia (151 days and 12 steps) but much higher than that in Singapore (eight days and seven step^).’^ This finding suggests that the Thai business climate i s hospitable to the creation of new businesses. However, these statistics also point to clear examples, such as Singapore, that Thailand may have to emulate to make the process of entry of new firms fully efficient and streamlined. 1.32 We complement the Doing Business de jure regulations on starting a new business with information from the PICS on the regulatory burden to open a business in 2003-2004, such as the number of weeks necessary for firms to obtain licenses, permits, approvals or certificates from l 9The number of steps required to start a businessi s not shown inFigure 1.10, those figures are available at http://rru. worldbank.org/DoingBusiness/. 17 different government levels and agencies, summarized in Table 1.4. The advantage of considering the reports of firms in the PICS on the ease of opening a business in Thailand i s that they reflect the institutional environment in which the de jure regulations are applied.20In Thailand, specific agencies such as the Food and Drug Administration take on average longer than the central or local governments in processing licenses, permits, approvals or certificates needed for a firm to begin operating. The coefficient of variation shown in Table 1.4 captures one crucial dimension of the regulatory burden, which i s the uncertainty associated with the length of time required to obtain any of the licenses, permits, approvals or certificates. Interestingly, in Thai firms’ experience, the number of weeks needed to obtain approvals or certificates from the specific agencies or the local government i s more uncertain than the number of weeks needed to obtain documents from the central government. For all types of institutions, the time to obtain approvals or certificates i s the most uncertain. A possible interpretationfor the coefficient of variation is, for the case of approvals or certificates from local government, that: (i) isa32percentchancethatitwilltakeaThaifirmfiveormoreweekstoobtainthem, there rather than the two-week average delay; and (ii) there i s a five percent chance that it will take a Thai firm eight or more weeks to obtain them, rather than the two-week average delay.” Finally, it is worth pointing out that in the PICS, Thai firms report having to do more periodic renewals of permits similar to those needed to start operations, than of licenses, approvals or certificates. Only eight percent of Thai firms resort to agents or consultants for help in processing government documents. This i s a small fraction of firms when compared with those in Malaysia or Indonesia. This finding could signal either: (i)less complexity of Thai government regulations; (ii) more capability of firms in dealing with those; or (iii)less availability or visibility of such services inThailand. Table 1.4: Numberof Weeks to ObtainLicenses/Permits/Approvals/Certificatesto Start aBusiness Avg. St. Dev. Coeff. Variat. Median N.Obs. Central Government (e.g. Dept. of Licenses 3 2 0.6 4 29 Industrial Works, Ministryof Permits 5 7 1.3 4 107 .-.–..————._____1____1__1__1_1_–..———. ~ . . . Commerce, etc) AppovaldCertif. ~ ~ ~ ~ ~4~ . ~ ~ . . . . . ~ . 3 1.3 2 83 LocalGovernment (e.g. municipality, Licenses 4 n.a. n.a. 4 1 local administration, etc) Permits 6 8 1.4 4 21 .-.—.-..-.–.—–_————–11–1.—..—-.– AppovaldCertif. ___.__________——_____1__111___1 2 3 1.5 1 16 Specific Agencies (e.g., Food and Drug Licenses 4 1 0.2 4 2 Admin., Dept. of Fisheries, Royal ForestDept., Dept. of Medical Permits 5 3 0.7 4 20 Sciences, etc. ApprovaldCertif. 10 17 1.7 4 8 1.33 In addition to its role in regulating labor and business entry, the Thai government routinely interacts with manufacturing firms through the requirement of licenses and permits for firms’ day-to-day operations, as well as through inspections related to taxes, customs, and other health, safety and environmental standards. Table 1.5 shows the number of days needed to obtain 2o For accuracy purposes, we consider only the responses of firms that experienced the bureaucratic burden of opening a business inthe recent past, i.e., firms that are less than five years old at the time of the PICS. 21 Assuming that the number of weeks to obtain an approvakertificate from the local government across Thai firms i s normally distributed, the probability of realizing a value larger than the average plus one standarddeviation (5 weeks) is 32 percent. 18 licenses, permits, approvals or certificates from various government agencies, and thus captures crucial dimensions of the regulatory burden experienced by firms in operating their business. These agencies are involved inprocedures of varying complexity, so the average number of days taken to provide the documents i s expected to differ substantially across agencies. Licenses from the Department of Industrial Works take the longest to obtain. Again, the most interesting aspect in Table 1.5 is the high degree of uncertainty that Thai firm managers face regarding when to expect the various documents, as captured by the coefficient of variation. Licenses from the Ministry of Commerce or the Land Office are the most uncertain to obtain. In particular, that uncertainty means that: (i) there i s a 32 percent chance that it will take 30 or more days to obtain a license from the Ministry of Commerce, rather than the average 10 days; and (ii) there i s a five percent chance that it will take 50 or more days to obtain that license.22 Table 1.5: Numberof Daysto ObtainDifferentLicensesPermitdApprovald Certificates Avg. St. Dev. Coeff. Variat. Median N.Obs. Ministry of Commerce 10 20 1.9 2 964 Departmentof IndustrialWorks 17 22 1.3 7 926 Immigration Department 10 14 1.4 3 80 LandOffice 13 22 1.7 3 131 LocalGovernment 10 15 1.6 2 419 1.34 Table 1.6 shows the number of days needed to obtain additional approvals and documents in Thailand, Malaysia, Indonesia and the Philippines. In Thailand, operating licenses and approvals for construction take much longer to obtain than import permits. Operating licenses are also the most uncertain documents to obtain. The degree of uncertainty faced by Thai firms in obtaining these documents i s generally larger than that of Malaysian firms, but much lower than that faced by firms inIndonesia and the Philippines. Table 1.6: Numberof Daysto ObtainDifferentApprovalsandDocuments Thailand Avg. St. Dev. Coeff. Variat. Median N.Obs. Approval for Construction 36 40 1.1 30 242 ImportPermit 13 18 1.4 7 164 Operating License 39 58 1.5 30 188 Malaysia Avg. St. Dev. Coeff. Variat. Median N.Obs. Approval for Construction 82 107 1.3 50 267 ImportPermit 27 50 1.9 14 286 Operating License 36 55 1.5 14 447 Indonesia Avg. St. Dev. Coeff. Variat. Median N.Obs. Approval for Construction 50 110 4.8 15 88 ImportPermit 11 14 1.6 7 67 Philiuuines Avg. St. Dev. Coeff. Variat. Median N.Obs. Approval for Construction 27 84 9.7 10 82 ImportPermit 13 21 2.6 7 73 OperatingLicense 25 68 7.3 7 170 22This statement is verified assuming that the number of days to obtain a permit from the Ministry of Commerce across Thai firms is normallydistributed. 19 1.35 Finally, Table 1.7 shows information on the time needed to obtain export incentives. Although only a small number of Thai firms requested such incentives, they can be crucial for initiating or continuing firms’ export activities, and — as will be shown later in the chapter — this has important positive effects on firm performance, The responses in Table 1.7 again show significant variation in the predictability of the time required to process and receive the application for the various export incentives. The application for e-commerce export promotion i s by far the most uncertain of all export incentives. Namely, this uncertainty suggests that there i s a 32 percent chance that it will take 178 days or more to obtain the e-commerce export promotion incentive instead of the 33 days on average.23A more in-depth analysis of the explanations for such variability and possible improvements i s crucial for Thailand to take further advantage of its export potential inmanufacturingproducts. Table 1.7: Number of Days to Process Application for Different Export Incentives Avg. St. Dev. Coeff. Variat. Median N.Obs. E-commerceExportPromotion 33 145 4.4 7 56 Promotionof ThailandBrands 41 61 1.5 30 62 ExportsPromotionto new market(2003-2005) 18 19 1.o 10 39 Distribution Networking 32 48 1.5 23 12 Arrange sale promotionwith Departmentstore 17 13 0.7 7 11 Exportone stop service 15 13 0.9 11 14 PrimeMinister’s ExportAward2003 (PM Award) 21 20 1.o 14 13 DeductiononCost of DevelopingWebsites 28 16 0.6 30 10 Tax Incentivesfor OffshoreTradingVia Websites 30 31 1.o 23 8 1.36 From an international perspective, on some dimensions Thai firms face a relatively light bureaucratic burden, as was shown in Figure 1.4. Thailand, with only 1.8 percent of senior management time spent dealing with regulations in a typical year, ranks substantially better than all other comparator countries. InThailand, the average number of days to deal with customs for exports of 1.5 i s also very low, when compared with, for example, more than five days inBrazil, China, India and the Philippines. This apparent efficacy of customs for exports in Thailand needs to be maintained as it can give a competitive edge to Thai exporters in world markets. Within Thailand, Appendix Tables 4B-4D show that managers of firms in food processing, textiles and wood & furniture industries spend much less time dealing with regulations than managers of firms in electronics & appliances and machinery & equipment industries. It i s likely that the stronger export orientation of the latter industries and consequent need for customs and other export-related forms accounts for this difference in bureaucratic burden. This i s somewhat confirmed by the finding that managers of foreign-owned firms are significantly more burdened with regulation. 1.37 Figure 1.13 Panel A shows that in a given year, Thai firms suffer from a lower burden of inspections and required meetings with officials than firms in China, India, Malaysia and the 23 This statement is verified assumingthat the number of days to obtain an e-commerce export promotion incentive across Thai firms is normally distributed. 20 Philippines. Those statistics are based on visits by inspectors from all types of government agencies. InThailand about half of the time spent by firms in inspections and visits i s spent with officials from the revenue department. The next most time consuming inspections are from the IndustrialWorks Department. From an international perspective, Thai firms face a less favorable situation regarding the number of days required to clear customs for imports, as shown inFigure 1.13 Panel B. Thailand, where imports customs take on average five days to clear, ranks worse than Malaysia. Since imports of inputs and machinery are crucial for the modernization of the production process by manufacturing firms, namely those competing in global markets, the issue of customs clearance for imports in Thailand deserves further analysis to determine where efficacy improvements can be achieved. – Figure 1.13: Thai FirmsFace a Lesser Regulatory BurdenThan Many Other Countries – — China B d Brazil Philippines Malaysia CbiM = Phillppiws India = India Indonesia Thailand Thailand- Indonesia I Malaysia 0 5 10 1s Panel B 1.38 The regulatory and bureaucratic burden faced by Thai firms i s also illustrated with two additional measures from the 2004 Doing Business database on contract enforcement (Le., the number of days requiredto resolve a typical contract dispute such as collecting a badcheck) and on time taken for property registration, shown in Figure 1.14.24In terms of property registration, Thailand performs extremely well with a wait time of only two days, ranking above all comparator countries and only below Norway. In contrast, Thailand scores substantially worse on contract enforcement. Thailand’s 390 days to solve a payment dispute place it well into the third quartile of all countries. In particular, that figure is substantially longer than 75 days for Korea, 241 days for China and 300 days for Malaysia. Firms in the PICS report that court cases to resolve a dispute over payments take on average 420 days but can take up to 1,680 days. On the one hand, Figure 1.14 suggests that Thai firms’ concerns about the regulatory burden revealed by the PICS are not related to property registration. This also complements the finding inAppendix Table 1Athat access to landis amajor constraint for only 2.5 percent of Thai firms. On the other hand, Figure 1.14 points out to contract enforcement as another specific aspect of the Thai business climate that requires further in-depth analysis. It i s crucial to understand and 24More specifically, the contract enforcement measure consists of the number of days taken from the moment that the plaintiff files the lawsuit incourt untilthe moment of actual payment. The property registration measureconsists of the number of days taken to fulfill all the procedures needed to transfer the property title from the seller to the buyer, where the buyer i s a firm acquiring land and a building inthe country’s mainbusiness city. 21 address the causes for the lengthy time period taken to enforce contracts in Thailand. In the PICS, 52 percent of Thai firms report having had payment disputes with clients concerning delay or suspension of payments, return shipments or cancellation of shipments in 2002-2004. Firms report that 22 percent of these payment disputes were resolved by court action, which i s a figure quite similar to that in Malaysia (20 percent of payment disputes), but much higher than that in China (5.4 percent), Indonesia (one percent) and the Philippines (2.2 percent). Hence, the legal system inThailand seems capable of enforcing property rights, and indeed an important fraction of firms resort to it for business disputes. The potential improvements and benefits that may be identified from a more in-depth analysis of the Thai legal system may be linked to its efficiency, namely related to the time taken to resolve court cases. Figure 1.14: Thailand Performs Well interms of Property Registration butworse inareas of ContractEnforcement 2M – -. I 200- II 180- I 25th Percentile Payment Dispute 160- f . pc 140- [s”5 . h lZO- IIIII- IISolving= 250 Days -Ma’aysia. I 100- I” 9 + 3 r 80- ‘1. ‘ cn B —. I. .Indiam . n 25th Percentile Registering Property —————=24 Days 0 200 400 600 800 lo00 1200 1400 1600 Days to Solve a Payment Dispute Source World BankDoing BusinessIndicators2004. 1.39 International comparisons based on some of the governance indicators of Kaufmann, Kraay and Mastruzzi (2005) support and complement Thai firms’ concerns in the PICS about regulatory and bureaucratic burden. Figure 1.15 suggests that Thailand has clearly better government effectiveness than the median across all countries, however it ranks far behind Korea, Malaysia and Singapore. In the dimension of regulatory quality, Thailand ranks only slightly above the cross-country median, but performs much worse than Malaysia or Brazil. Figure 1.16 shows that Thailand has approximately the median value for the Control of Corruption Index, but its ranlungi s worse than Brazil, Malaysia and Korea. 1.40 The GCR i s final data source that reinforces and confirms the importance of bureaucratic burden as a serious obstacle to Thai firms. Thailand worsened its ranking in the Public Institutions Index to 45 out of 104 in 2004 from 37 out of 102 in 2003. This finding reinforces 22 the need for improvements in the efficiency of the bureaucracy and cuts in red tape to enhance manufacturingfirms’ productivity. It i s interesting to note that the major problems in Thailand’s business climate identified by the managers surveyed for the GCR are bureaucratic red tape, corruption, and lack of an educated workforce, all of which exactly confirm the findings in the PIcs.25 Figure 1.15: Government Effectivenessand Regulatory Quality Governance Indexes Government Effectiveness-2004 d ‘1 Regulatory Quality 2004 – I HlQH LOW 209 Countries 204 Countrles Figure 1.16: Controlof Corruption GovernanceIndex. Control of HlQH Infrastructure and Support Services 1.41 Government effectiveness in providing infrastructure services i scrucial for manufacturing firms to operate in a low-cost environment. A large fraction of Thai firms points to infrastructure and support services as a serious obstacle for their operations and growth. Indeed, Thailand’s rankingrelative to comparator countries on infrastructure-related indicatorsin the PICS i s poor, as shown in Figure 1.17. Except for India, where firms experience on average 210 days of power outages per year, Thailand has the highest frequency of power outages. This 25See Thailand Economic Monitor, October 2004. 23 suggests that electricity supply for Thai manufacturing firms i s quite unreliable. Thai firms also report a lengthy delay to obtain an electricity connection, only surpassedby the time needed by firms to obtain that connection in India. Figure 1.18 provides an international comparison of electricity generating capacity that confirms the concerns of Thai firms in the PICS. Although generating capacity inThailand had a sixfold increase since 1980 — and actually increased about 20 percent between 2000 and 2002 — the per capita generating capacity of Thailand still lags well behind that of Malaysia, South Korea and Singapore. These indicators suggest that addressing the bottlenecks in Thai infrastructure services i s crucial to enhancing the competitiveness of Thai firms. Power outages can lead to losses in production and sales by generating idle capacity and damage of raw materials, goods-in-process and even machinery & equipment. Although power outages are more frequent in Thailand, they lead to lower production losses than in com.parator countries, as shown in Figure 1.19 Panel A. Moreover, the PICS indicates that each power outage lasts on average 2.4 hours in Thailand, which i s lower than in other countries for which comparable data i s available. Firms often respond to power outages by acquiring and running their own generators. Appendix Table 3A documents the percentage of firms with generators in Thailand — 17 percent — which i s lower than in most comparator countries. This hints at the possibility that the power outages captured by the PICS in 2004 were abnormally high, andthus firms didnot take steps to counteract those. 1.42 Telecommunications are considered to be a serious obstacle to operations and growth for 11percent of Thai firms. Figure 1.20Panel A indicates that Thai firms experience a longer delay in obtaining a fixed-line telephone connection than firms in Malaysia, China or the Philippines. Additionally, firms in Thailand suffer from more frequent interruptions of fixed-line telephone services than firms in Malaysia, Indonesia, the Philippines or Brazil, as shown in Figure 1.20 Panel B. Intheir business operations and contacts, Thai firms may circumvent these problems by relying on cellular phone connections. However, Figure 1.21 shows that Thailand still lags in terms of mobile phone penetration, when compared with Malaysia, South Korea and SingaporeSz6 Also, fixed-line penetration inThailand i s low when compared with China. 1.43 Two additional elements of general infrastructure services are water supply and transportation. Thailand does not perform particularly well on water-related indicators. Thai firms wait on average 23 days to obtain a connection, which i s much longer than firms in Malaysia, Indonesia or Brazil, as shown in Figure 1 in the Appendix. Also, water supply interruptions — four times in a given year — are lower in Thailand than in Malaysia or the Philippines, but are higher than in Indonesia or Brazil. Finally, 40 percent of water use by Thai firms comes from their own sources rather than from the public grid, compared with less than five percent in Malaysia (see Figure 2 in the Appendix). Transportation problems are a serious obstacle for about 14 percent of Thai firms. Figure 1.19 Panel B shows, however, that shipment losses (as a percentage of sales) are much lower for firms in Thailand than for their counterparts inChina andIndonesia. Also, transport disruptions are less frequent inThailand than inMalaysia or Indonesia, as shown by Figure 2 inthe Appendix. 26The statistics shown inFigure 1.21show mobile and fixed-line phone penetration across countries aggregateboth residential and industrialconsumers. 24 Figure 1.17: Thailand Ranks Poorly Relative to Other Countries inTerms of Key – InfrastructureConcerns Frequency ofPower Outages (NumberofTimes Last NumberofDays to Obtain811Electricity Conwetion Year) Ma’aysia Philippines iI China Indonesia 0 5 10 15 M ole: India with210 days is not showndue to space constraints. 0 20 40 60 80 Panel A Panel B Figure 1.18: Thailand Lags Behind Other Countries inTerms of Electricity Supply 2 1.8 1.6 1.4 1.2 1 0.8 0.6 0.4 – 0.2 – 0 – iource:US.Energy InfomationAgency. 25 – Figure 1.19: Although More Frequent, Power OutagesAppear to Cost Thai FirmsComparatively Less Than inOther Countries – PercentofPmductionLostDue to PowerOutages India Philippines Indonesia = Brazil = china Thailand Malaysia 0 2 4 6 8 10 – – Figure 1.20: Telecommunicationsare also Poor inThailand by International Standards — Sumber ofDays to ObtainaPhone Connection FrequencyofPhone InterruptionsLastYear I India Thailand Indonesia Malaysia Thailand = Brazil B M Philippines ‘hilipines = China Malaysia ndonesia I I 0 10 20 30 40 50 60 70 0 1 2 3 4 5 6 7 Panel A Panel B 26 Figure 1.21: Thailand Lags Malaysia, SouthKorea andSingapore inMobile and FixedTelephone Penetration 140 120 100 80 60 40 20 0 0FixedPhone Lines Mobile Phones Source: EIUCountry Indicators, 2004. D. DETERMINANTSOFFIRMPERFORMANCE 1.44 This section explores the links between firm performance and its determinants in Thailand. It i s interesting to consider first some cross-country comparisons of one measure of firm performance — labor productivity — for various industries using data from PICS.27Figures 1.22 and 1.23 show that Thailand has much lower labor productivity than Brazil and Malaysiain food processing, textiles, garments and electronics industries.28The same cross-country ranking applies to wages per worker, and the differences in labor productivity can be linked to differences in the capital intensity with which firms operate in each of these industries and countries as shown in Appendix Figures 3-6. Figures 1.22 and 1.23 deliver a mixed message on Thai competitiveness. On the one hand, Thailand still commands a reasonable labor productivity premium relative to competitors in world markets such as China and Indonesia in low-tech industries like food processing and garments. However, Thailand’s labor productivity difference relative to these competitors i s much smaller in the electronics & electrical appliances industry. 27Details on the calculation of labor productivity and average wages are providedinthe Appendix. The analysis is performed for four separate industries since countries differ significantly in the industrial composition of their PICS samples. Moreover, within countries, industries differ substantially in average labor productivity. 27 Figure 1.22: MedianLabor Productivity (VAL in2001U.S. dollars) inDifferentIndustries Figure 1.23: MedianLabor Productivity (VAL in2001U.S. dollars) in Garments and Electronics/Electrical Appliances Industries 1.45 The remainder of this section explores the links between Thai firm characteristics, performance and the business climate. Larger firms, exporting firms, firms with FDIand firms usingmore computerized machinery exhibit stronger performance than other firms, measuredby VA and TFP. There are important differences in the business climate faced by the better performing firms relative to the worse performing firms. The major obstacles for the better performing firms, large firms, exporter firms, foreign-owned firms and firms with more computerized machinery are heavy regulation and bureaucratic burden. Irrespective of location, industry or characteristics, skill shortages are a major problemfaced by firms intheir operations. Measuring Firm Performance 1.46 Inorder to promote stronger productivity growth, it is importantto understand the factors that influence firm performance in Thailand and how the business climate differs for firms with different characteristics. Firm performance can be measured using several indicators. This chapter relies on three performance indicators obtained using the data from the PICS. Details on how the indicators are constructed are providedinthe Appendix. 0 VAL measuresthe productivity of a single-factor of production: labor. 28 TFP is a multi-factor productivity measure that represents the efficiency of the firm in transforming inputs into output. TFP captures technology, managerial quality and government policies, among others. 0 SGindicates which firms expand andwhich contract.” 1.47 While VAL and SG are observable indicators of firm performance, TFP i s not observable. We obtain measures of TFP for Thai firms as residuals from a production function estimated following Levinsohn and Petrin (2003) techniques that correct for the potential endogeneity between the choice of inputs and firm productivity, i.e., the fact that firms chose their inputs with knowledge of their productivity that i s unknown to the econometrician. The production function considered assumes that output i s produced combining four inputs: SL; UL; intermediates; and capital and i s estimated separately for each of eight industries usingtwo years of data for each firm (2001 and 2002).30Thailand’s Minister of Industry (MOI) matched some of the firms inthe PICS to firms in their survey and kindly provided us with data for two additional years — 1999 and 2000 — for roughly 230 firms. Thus, we also obtained a different set of TFP measures from a production function estimated usingthis longer panel datasetq31 what follows, In we will be referring to the TFP measures obtained for the entire PICS sample, unless otherwise specified. 1.48 The three performance measures — TFP, VAL and SG — capture different dimensions of firm performance. Table 1.8 shows the correlations among firm performance measures in 2002 controlling for fixed differences across industries and regions. All measures are positively correlated, but the magnitude of the correlations varies substantially. VAL and TFP exhibit the highest correlation with a coefficient of almost 0.5. The correlation between TFP and SG i s very low, and that between VAL and SG i s also low. These findings suggest that although the measures are consistent in classifying firms as good or bad performers, they are sufficiently different to make it informative to pursue the analysis of the determinants of performance for all three measures. *’Ifmarkets ~~ are relatively free from distortions (e.g. no obstacles to entry, exit and growth of firms) and the firms that grow are those most efficient, then sales growth is also a good indicator of productivity. 30 Further details on the production function estimation are provided in the Appendix. The eight industries considered are food, textiles, garments, auto-parts, electronics and electrical appliances, rubber and plastics, wood and furniture and machinery and equipment. 31Inthis case the estimation was performed separately for two broadly-defined industries: low-tech industries (food, textiles, garments and wood and furniture) and high-tech industries (auto-parts, electronics and electrical appliances, rubber and plastics and machinery and equipment). 29 Table 1.8: Correlationamong FirmPerformance Measures TFP Sales Value- N.Obs.= 1033 Growth Added per Wnrker TFP 1 Sales Growth 0.047 1 Value-AddedDer Worker 0.483*** 0.129*** 1 Note: *** indicatessignificance at the 5% confidence level. The correlations are obtainedbasedon the residuals fromregressionsof eachperformance measure on inudstry andregiondummies, excluding outliers. Correlatesof Firm Performance 1.49 Firms differ substantially in their performance. At any point in time, there are firms growing, others contracting, firms with highproductivity and others with low productivity. The focus in this section i s to understand how different firm characteristics account for the observed heterogeneity in firm performance in Thailand. The theoretical and empirical literature i s quite well developed in this area. For example, Navaretti and Venables (2005) survey a large literature that finds a positive link between foreign ownership and firm productivity as foreign ownership allows for technology transfer, while studies such as Bernard and Jensen (2001) and Fernandes and Isgut (2005) suggest that exporting i s associatedwith higher productivity. 1.50 A regression framework is usedto relate the three measures of firm performance in 2002 to a set of firm characteristics: age; size; export status; foreign ownership status; measures of technology use and innovation; industry affiliation; and regional location.32The patterns of association between firm-level characteristics and performance are presented inTable 1.9. 1.51 The findings from the regressions in Table 1.9 are informative and intuitive, but one needs to be cautious in their interpretation. Those regression results suffer from two potential econometric problems: (i)reverse causality; and (ii) multicollinearity. Reverse causality i s possible since some firm characteristics whose effect on performance i s being estimated may themselves be affected by performance. For example, exporting activity may have a positive effect on firm performance but concurrently, exporting activity may be influenced by firm performance, i.e., only the best performing domestic firms are able to engage in exporting activities. This reverse causality problem cannot be addressed convincingly since the PICS i s mostly cross-sectional. Thus, the findings from the regression analysis are indicative of a correlation between performance and characteristics but not of a causal relationship. A multicollinearity problem may result from the correlation among determinants of performance. Such correlation makes it more difficult to isolate the effect of a given characteristic on firm performance when many other correlated characteristics are also included in the regression. For Thai firms, the various firm characteristics included as determinants of performance are significantly correlated among themselves, as shown in the Appendix. Thus, we complement the analysis of the regressions of firm performance that include simultaneously all firm characteristics by partial regressions of firm performance on each firm characteristic at a time, controlling for industry and regional fixed effects. The unreported results from the partial 32 The empirical specification and the measurement of firm characteristicsare described inthe Appendix. 30 regressions are very similar to those shown in Table 1.9 but they tend to be stronger and more significant, given the multicollinearity problem. For example, the effect of exporting on firm performance i s much stronger when the exporter dummy i s the only characteristic includedinthe regression than when other characteristics such as firm size are included. Since exporters are generally larger than non-exporters, including firm size reduces the sign and significance of the exporter variable. 1.52 A final note of caution regarding the results. Table 1.9 relates to the comparison of firm performance across industries which can be problematic for VAL and TFP. For example, the comparison of VAL across industries simply reflects differences in capital intensity across industries. Hence, we do not focus on cross-industry comparisons of VAL and TFP. However, we do include in the regressions the industry affiliation of each firm. The rationale behind this i s that the distribution of firm characteristics differs across industries and we want to analyze results that are obtained controlling for these fixed differences. The differences in SG across industries can be analyzed as they are more c~mparable.~~Relative to firms in the food industry, firms in the auto-parts industry and the rubber & plastics industries have much higher SG, respectively, 14.3 and 12.4 percent higher.34Firms in the textiles industry and in the electronics &electrical appliances industryalso have significantlyhigher SGthan firms inthe foodindustry. 1.53 Firmage is associated with better performance. The results in Table 1.9 show that, controlling for all other characteristics, older firms have higher TPF and VAL but lower SG. The finding on TFP mirrors those obtained by Jensen, McGuckin and Stiroh (2001) for U.S. manufacturingfirms. 1.54 Firm size measured by employment is strongly positively associated with firm performance. Larger firms exhibit significantly higher TFP. If one considers two hypothetical firms, A and B, where firm A i s twice as large as firm B, firm A has TFP that i s 8.6 percent higher than TFP of firmB, “ceteris Firmsinour estimating sample have a median size of 133 employees, and 23 percent can be classified as small following Thailand’s MOI d e f i n i t i ~ n .The negative effect of size on VAL shown in Table 1.9 i s obtained almost by ~ ~ construction, as VAL i s defined with total employment as a denominator. In unreported regressions of firm performance on size dummies instead of a continuous size variable and other firmcharacteristics, VAL for medium and large firms is higher than VAL for small firms. Large firms grow more slowly, although not significantly so. The positive link between firm size and TFP is robust to changes inthe definition of size. The regression results are qualitatively similar when size i s measured by firm capital or by employment inthe initial year of plant operations. 1.55 The finding on size and productivity performance should be interpreted with caution. First, the correlation between firm size and TFP may be spurious as TFP is estimated using revenues as the measure of output. Large firms are likely to have market power and charge 33 For simplicity, the coefficients on the industry and regiondummy variables are not shown in Table 1.18. 34 Note that these coefficients refer to sales growth between2001 and 2002. 35 TFP and firm size are in logarithms, so the coefficient on size indicates the percent change in TFP per 1percent change in firm size with all else constant, i.e. “ceteris paribus”. The figure of 8.6 percent mentioned in the text is obtained as 0.086*100 percent, where 100percent indicates that firm A is 100percent larger than firmB. 36 The MOI defines small firms as those with fewer than 50 employees, medium firms as those with 50 to 200 employees and large firms those with more than 200 employees. higher prices. This will be reflected in higher revenues, hence in higher measured TFP. Second, our cross-sectional results cannot disentangle the direction of causality between size and firm performance. In fact, it i s possible that firms grow because they are more efficient, and not the reverse. The fact that we obtain similar findings when firm size i s measured by employment in the initial year of firm operations somehow mitigates this causality issue. However, we also note that though one needs to be cautious about drawing policy implications from the estimated effect of size on performance, it i s likely that large firms derive benefits from other sources such as being better able to cover fixed costs, having easier access to productivity-enhancing activities such as the use of training and technology institutes (see Chapters 3 and 4). Table 1.9: Correlates of FirmPerformance Regressors Total Factor Sales Growth Labor Productivity Productivity FirmAge 0.004*** -0.003** 0.011*** (0.002) (0.001) (0.003) Current Employment 0.086*** -0.015 -0.02 (0.015) (0.009) (0.025) Exporter Dummy (more than 10%) 0.077*** -0.015 0.266*** (0.028) (0.022) (0.060) ForeignOwnership Dummy 0.124** * -0.031 0.428*** (0.032) (0.026) (0.063) Capital Vintage (% Mach.Under 5 Years) 0.046 0.084″” 0.125 (0.047) (0.038) (0.093) % Computer-Controlled Machinery 0.142* ** -0.027 0.283*** (0.048) (0.035) (0.096) R&DSpendingDummy 0.017 0.01 0.056 (0.031) (0.022) (0.057) Industry Dummies Yes Yes Yes RegionDummies Yes Yes Yes N.Observations 1033 1033 1033 R-squared 0.98 0.06 0.21 Notes: OLS estimation is used. Robust standard errors are in parentheses. ***, ** and* represent significance at the 1,5 and 10percent confidence levels, respectively. 1.56 Exporter firms perform significantly better than non-exporter firms according to TFP and VAL. The regressions relate firm performance to integrationin international markets by including a dummy variable for exporter firms. In our estimating sample, 53 percent of firms are exporters.37With all else equal, exporter firms have an average 30.5 percent higher VAL than non-exporter firms.38Exporter firms are generally more capital intensive, but they also exhibit, on average, 8 percent higher TFP than non-exporter firms. Exporters have lower SG than non- 37We define exporters as those firms that export more than 10percent of their output. 38For the exporter dummy variable with all else constant, ln(VAL exporters) – ln(VAL non-exporters) = ln(VAL exporters / VAL non-exporters) = 0.266.This means that VAL exporters / VAL non-exporters = exp(0.273) = 1.305. This is the value of 30.5 percent mentioned in the text. The 8 percent difference for TFP of exporter firms relative to non-exporter firms i s obtained similarly, and so i s the 13.2 percent difference for TFP of firms with FDIrelative to domestic firms below. 32 exporters, but the difference is not significant. In the Appendix, we present the results from additional regressions that include export intensity instead of a dumrny variable identifying exporters.39Those results indicate that a firm with the average export intensity (62 percent) has a TFP that i s higher by 5.6 percent than afirm that does not export.4o 1.57 The performance of Thai exporters in terms of productivity i s significantly stronger than that of non-exporter firms, a finding which parallels those obtained for manufacturing firms in other developed and developing countries by, e.g., Bernard and Jensen (1999) for the U.S., Fernandes and Isgut (2005) for Colombia and Kraay (1999) for China.41 The stronger performance of exporter firms may be due to their exposure to fierce competition ininternational markets and to more advanced technology and the more stringent demand for quality by their foreign buyers. While it i s likely that over time this “knowledge” and productivity advantage i s transmitted and diffused as exporters interact with other firms, for Thailand, it seems that the effects are still strong despite the fact that Thai firms have been participating heavily in global markets for manufacturing products for almost two decades. This finding suggests that although the Thai economy i s already characterized by a large degree of openness and integration into global markets, promoting the entry of more firms into export markets, or the expansion of those already in, i s likely to increase the competitiveness of the Thai economy. 1.58 Foreign-ownedfirms perform significantly better than domestic firms in terms of VAL and TFP. Thai firms with FDIhave VAL that is on average 53.4 percent higher than that of domestic firms. While these foreign-owned firms also have higher capital intensity, they have higher TFP (13.2 percent higher) than domestic firms.42This result i s consistent with the findings by Aitken and Harrison (1999) for firms in Venezuela and Arnold and Javorcik (2005) for firms in Indonesia. Firms with FDI have lower SG than domestic firms but the difference is not significant. The findings for TFP suggest that the facilitation of FDI inflows into Thailand may contribute to an improvement in the productivity of the domestic economy. We also show in the Appendix the results from regressions that include the foreign ownership share instead of a dummy variable identifying foreign-owned firms. Those results indicate that a firm with the average foreign ownership share (63 percent) has a TFP that i s 14.4 percent higher than a fully domestic firm.43 1.59 Firmswithlarger fractionsof computer-controlledmachineryhavehigherVAL and TFP. Technologically more advanced firms are expected to exhibit better performance. Two 39Those regressions include also the share of capital that i s foreign-owned instead of a dummy variable for foreign ownership. 40The average export intensity of 62 percent is obtained considering only firms that have some positive level of exports. If zeros were included, the average export intensity would be 39 percent. 41Infact, the TFP of exporter firms is evenhigher than that of non-exporters inthe partial regression where only the exporter dummy variable, industry effects and region effects are included. By adding other firm characteristics correlated with the exporter dummy variable (namely firm size) to the regression, the magnitude of its coefficient is reduced. 42For the foreign ownership dummy variable with all else constant, ln(TFP foreign-owned) – ln(TFP domestic) = ln(TFP foreign-owned / TFP domestic) =0.124. This means that TFP foreign-owned / TFP domestic = exp(0.124) = 1.132. This is the value of 13.2 percent mentioned in the text. The 53.4 percent difference for VAL is obtained similarly. 43The average foreign ownership share of 63 percent is obtained considering only firms that have some foreign ownership. Ifzeros were included, the average foreign ownership share would be 17 percent. 33 measures related to the technology used by firms are included in the regressions: the fraction of firm machinery that is computer-controlled; and the fraction of firm machinery that is less than five years old. Firms with a larger fraction of their machinery that i s computer-controlled have higher VAL and TFP. For example, the increase in TFP from a firm with no machinery controlled by computers to a firm with the average percent of machinery controlled by computers (36 percent) i s 5.1 percent.44This effect i s quite substantial, but it i s even stronger in the partial regressions of TFP that include only this measure of technology use. In such regressions, the increase in TFP due to an increase from no computer-controlled machinery to the average level of computer-controlled machinery i s 11.4 percent. Sales of firms that have more modern machinery grow significantly faster than sales of firms with older machinery. The increase in SG from a firm with no machinery less than five years old, to a firm with the average percentage of machinery less than five years old (33 percent), i s 2.8 percent. Firms with more modem machinery have higher TFP and VAL but the effects are not significant. A possible interpretation for the lack of significance of these effects i s as follows. The percentage of new machinery basically identifies machinery purchased since the Asian crisis. The strong depreciation of the Thai Baht during this period led to a marked decline in imports of machinery. Thus, for many Thai firms recent machinery purchases must have been of Thai origin. It i s well established, however, that machinery imported from industrialized countries embodies a more advanced technology than domestic machinery for a middle-income country such as Thailand. Hence, for a Thai firm, having more new machinery in 2004 does not necessarily mean a strong quality improvement in the machinery relative to slightly older machinery which might have been imported. Consequently, the effects on productivity are only weakly positive. 1.60 A measure related to the innovative capacity of firms is also included inthe regressions: a dummy variable for firms that engage in R&D activities. In the estimating sample, only 23 percent of firms report positive expenditures on R&D. Firms doing R&D have higher TFP, VAL and SG, but none of the effects are significant. In the partial regressions of TFP and VAL on the dummy variable for R&D spending and industry and region dummy variables only, the effects are much stronger and significant: i.e., firms doing R&D have TFP that i s 11percent higher than firms not doing any R&D.45However, given the strong correlation between R&D and firm size, exporter and foreign-ownership dummy variables, it i s understandable that when all firm characteristics are included in the regression, the coefficient on the R&D dummy loses its significance. Overall, these findings suggest that technology and innovation are importantly associated with better firm performance. The issues of technology use and adoption by Thai firms are analyzed in detail in Chapter 4. 1.61 Large firms, exporter firms, foreign-owned firms and firms using more computerized machinery are more productive in Thailand. There is an important overlap between size, exporter status and foreign ownership status of firms: 89 percent of exporters are medium or large firms; 97 percent of foreign-owned firms are medium or large firms; and 75 44 The average percentageof computer-controlled machinery of 36 percent is obtained considering only firms that have some positive percent of their machinery controlled by computers. If zeros were included, the average ercentageof computer-controlled machinery would be 20 percent. Inthose unreported regressions, for the R&D dummy variable with all else constant, ln(TFP firm doing R&D) – ln(TFP firm not doing R&D) = In(TFP firm doing R&D / TFP firm not doing R&D) = 0.104. This means that TFP firm doing R&D / TFP firm not doing R&D = exp(0.104) = 1.11. This is the value of 11percent mentioned in the text. 34 percent of foreign firms are exporters. Overall, our results suggest that each of these firm characteristics as well as their combination play an important role of firm characteristics inexplainingdifferences infirmperformanceinThailand. 1.62 We should note that the importance of these firm characteristics in understanding differences in firm performance i s also obtained in regressions that control for more ,disaggregated product categories: four-digit ISIC industries. This means that firms in relatively similar lines of business perform better if they are larger, exporters, foreign-owned or use more advance technology. 1.63 The findings on size, exports and foreign-ownership are confirmed using the MOI firm-level panel. The results in Table 1.10 for TFP, VAL and SG for firms in the M O I panel confirm our findings in Table 1.9 using the cross-sectional PICS. Larger firms are more productive and so are exporters and foreign-owned firms.46 The production functions estimated for the M O I panel include total employment as the labor input.47Thus, we also include a measure of labor quality — the average number of years of education of the workforce in 2002 — as a determinant of performance and find that firms with a more educated workforce are significantly more productive. 46These regressionsdo not includeexactly all the firm characteristics inTable 1.9 due to a lack of data. 47The MOIpanel datadoes not allow for a separationof employment into skilled and unskilled components. 35 Table 1.10: Correlates of FirmPerformance using the MOIPanel Regressors Total Factor Sales Labor Productivity Growth Productivity Size (Current Employment) 0.028** -0.073*** -0.057** (0.013) (0.023) (0.027) Exporter Dummy (more than 10%) 0.021 0.056 0.298*** (0.025) (0.054) (0.08 1) ForeignOwnership Dummy 0.103*** 0.054 0.578*** (0.035) (0.060) (0.088) Capital Vintage (% Mach.Under 5 Years) -0.073 0.024 0.138 (0.058) (0.109) (0.143) R&D SpendingDummy -0.012 -0.026 0.007 (0.023) (0.053) (0.072) Average Years of Education of Workers 0.026** 0.039** 0.063*$ (0.011) (0.016) (0.027) RegionDummies Yes Yes Yes Industry Dummies Yes Yes Yes Year Dummies Yes Yes Yes N.Observations 647 639 649 R-squared 0.97 0.05 0.31 Notes: OLS estimation is used. Robust standarderrors are inparentheses. ***, ** and * represent significance at the 1, 5 and 10 percent confidence levels, respectively. Sales growth i s not available for 1999, thus 1999 data on TFP and labor productivity i s not included inthe estimation. Business Climateand Firm Performance 1.64 In this section, we examine the role of the business climate in understanding differences in firm performance. We follow three different approaches to measure the interactionbetween performance and the major business climate concerns for Thai firms. First approach: The main idea underlying this approach is to identify the differences in the business climate faced by good performers relative to those faced by bad performers. For that purpose, we split the sample into groups of firms according to each of the characteristics that affect performance in the previous section, and we identify the differences in the business climate faced by firms in these different groups. In particular, we group firms according to their age, their size, their export status, their FDIstatus and their use of computer-controlled machinery. Second approach: We use firm profit maximization within the production function framework to quantify how one of the most binding constraints for Thai firms — skills shortages — affects their output. Third approach: Using the MOI panel, we estimate regressions of firm performance on quantitative measures of the business climate taken as industry-regionaverages. 1.65 These three approaches are followed as an alternative to the estimation of regressions of firm performance on business climate perceptions as these might suffer from reverse causation 36 * problems. For example, firms with low SG or low TFP are very likely to “blame” the business climate for their bad year (e.g., identify “insufficient demand” for their products as a main obstacle to their operations). FirstApproach 1.66 Figures 1.24-1.26 relate the differences incharacteristics associatedwith better firm performance with the business climate.48 Younger firms (which the earlier analysis found have comparatively lower TFP and VAL) are more concerned about SL shortages and infrastructure and support services as suggested by Figure 1.24 Panel A. However, the burden of regulation imposes a constraint for both young and old firms alike – about 60 percent of firms (regardless of age) state that regulatory burden i s a principal concern for their operations. 0 Mediumand large firms in Thailand (which were found to perform better in terms of VAL and TFP) are more likely to point to regulatory burden as representing a major obstacle to their operations as shown in Figure 1.24 Panel B. In particular, large firms face a heavy regulatory burden in terms of customs and import regulations and foreign currency regulations but also in terms of inspections from various agencies. Medium firms are slightly more concerned with infrastructure problems. Export firms (which were also found to have better performance according to VAL and TFP) are much more concerned with the regulatory burden than non-export firms. Figure 1.25 Panel A shows that 64 percent of exporter firms point to regulatory burden as being a main obstacle to business whereas only 52 percent of non-exporter firms feel likewise. Given the nature of their operations, export firms voice particularly strong concerns about customs regulations, foreign currency regulations and import regulations but also labor regulations. Moreover, exporter firms spend a significantly larger number of days dealing with inspections from various public agencies. 0 Foreign-owned firms (which have significantly higher V A L and TFP than domestic firms) are much more .likely to consider regulatory burden as an obstacle to their operations than domestic firms. Figure 1.25 Panel B indicates that 67 percent of foreign-owned point out to those issues as major constraints compared with only 55 percent of domestic firms. Foreign-owned firms express very strong concerns about customs regulations, foreign currency regulations and import regulations. Foreign- 48Figures 1.24-1.26are constructed using the answers from the open-endedquestion where firms list the three main obstacles they face for doing business in Thailand. The fractions shown in the figure obtained as 100 minus the percentage of firms that do not report a given issue or groups of issues being part of their top three constraints to doing business in Thailand. As mentioned earlier, dissatisfaction with the economic situation aggregates insufficient demand for products and import competition. Regulatory burden aggregates ownership regulations, tax regulations/or high taxes, labor regulations, foreign currency regulations, import regulations, regulations for starting a new business and bureaucratic burdens. Infrastructure and support services aggregates lack of business support services, inadequatesupply of infrastructure and utility prices. 37 owned firms spend significantly larger numbers of days dealing with visits and inspections from different types of agencies, and their managers spend a significantly larger share of their time dealing with bureaucratic burden. Finally, firms that use more computer-controlled machinery (which were found to perform better in terms of VAL and TFP) are much more concerned about regulatory burden than firms with little or no computerized machinery. Also according to Figure 1.26, infrastructure and support services are a more important concern for firms with more computer-controlled machinery, although the difference i s relatively small. While in all locations, industries and for all firm characteristics, skill shortages are a main obstacle to operations and growth, there are fundamental differences in the perceptions of the business climate for the firms with better performance relative to the firms with worse performance. Better-performing firms (i.e., those that are larger, export, are foreign-owned or have more computer-controlled machinery) voice strong concerns about regulation as an obstacle to their operations. This finding is crucial, as it shows that regulatory impediments are particularly harmful since they are felt mostly by the best-performing firms in the Thai economy. Figure 1.24: BusinessClimate Concernsby FirmAge and Size Old BSLUlrdLbor ShorIage RLluhtory Burden Oldr.abuehm and SuppartSrvlcls . Skilled Labor Shortage BRcgulatoy Burden 0Inrm~tructurtand SupportSrrvlrls 38 Figure 1.25: BusinessClimate Concerns by Export Status and Ownership 7 0 , I 80 Forrigmownod Domgtk .Sklibd LPboiShortage Regulatory Burden .Skilied LaborShortage HReguletory Burden Olnbutrveturrand SupportServlns 0lnlrutrueture andSuppDrtServices Panel A Panel B Figure 1.26: BusinessClimate Concerns by Usage of Computer-Controlled Machinery 707 SecondApproach 1.67 Shortage of skills i s a pervasive problem for firms in Thailand. We develop below a set of simple “micro” estimates to quantify the output costs of skill shortages based on the production functions that were estimated for firms inthe PICS sample.49 1.68 Ifthere arenoskill shortages, afirmthat maximizesprofits hires slulled workers untilthe marginal product of a skilled worker i s equal to its payments (skilled worker wage). If the firm i s constrained, i.e., it cannot hire as many skilled workers as it desires due to skill shortages, the marginal product of skilled workers i s larger than the skilled worker wage. This difference can be viewed as the potential benefit to the constrained firm interms of increased sales/profits that it can obtain if skill shortages are red~ced.~’In the PICS, we have data on average wages for skilled workers across industries. Also, we are able to calculate the marginal product of skilled workers using firm-level production data and the production function parameter estimates used in the TFP calculations, described earlier. Finally, we are able to identify constrained firms as those having a skill mix (their number of skilled workers divided by their total number of workers) that i s lower than the optimal skill mix for their indu~try.’~ In Figures 1.27-1.30, we 49A detailed description of the framework usedis providedinthe Appendix. 50More specifically, this difference representsthe benefit from being able to hire one additional skilled worker. 51The optimal skill mix is calculated for each industry based on a formula derived from profit maximization and using the production function parameter estimates (see the Appendix for details). 39 show the distribution of skill mix (or the S L share) across firms in each industry. In each of the graphs, a dotted line represents the value of the optimal skill mix in that industry. It i s very clear that across industries, most firms have a skill mix that i s lower than the optimal skill mix. 1.69 We consider the following experiment: suppose that skills shortages are reduced (e.g., due to a large number of college graduates entering the labor market in Thailand) so each constrained firm can increase the number of slulled workers to reach the optimal skill mix for its industry. The benefit from this increase in skilled workers i s calculated for each constrained firm. The total benefit for an industry i s the sum of individual firms’ benefits. This total benefit i s then divided by the sum of current sales of those constrained firms and the results are shown in Table 1.11. Overall, the potential benefits from relaxing the skill constraints are very large in most industries, averaging 15 percent of sales. Figure 1.27: Distribution of Skill Mix Across Industries: Food Processingand Textiles Distributionof SkilledLaborShares across Distributionof SkilledLaborShares across Firms – FoodProcessing F i m Textiles – -. 45- I “” I I I 50 – I I I ?L c @ – II 30- IIIII I Figure 1.28: Distribution of SkillMix Across Industries: Garments and Auto-Parts Distributionof SkilledLaborShares acrossFirms. Distributionof SkilledLabor Shares across Garments Firms – Auto-Parts 50 1 I I I 40 Figure 1.29: Distribution of Skill M i x Across Industries: Electronics/Electrical Appliances and RubberD’lastics Distributionof Skilled Labor Shares across Distributionof SkilledLaborShares across Firm – Electronics/Elec.Appliances Firms RubberPlastics – Figure 1.30: Distribution of Skill M i x Across Industries: Wood/Furniture and Machinerymquipment Distributionof SkilledLaborShares across Firms – Wood/Furniture Distributionof Skilled LaborShares across Firms – MachineryEquipmnt 30 __ II 40- – 1 I I II t 20 -c251 II I I I I P 15- II L 10- ! IIII I S – , I , Table 1.11: Benefits from Relaxing Skill Shortages Industry Benefit from reducing skill shortages as % sales FoodProcessing 8.2 Textiles 14.1 Clothing 10.7 Auto-parts 4.6 Electronics andElectrical Appliances 3.6 RubberandPlastics 27.7 Wood Products andFurniture 44.8 Machinery andEquipment 7.8 1.70 InFigure 1.31, we plot the percentage sales gains from relaxing the skill constraints by industry against the average number of days taken to fill the latest vacancy for professionals in the industry. The industries in which an increase in firms’ skilled employment would have the 41 greatest benefits in terms of increased sales are also those which have more bindingS L shortage constraints, as suggestedby longest time needed to fill a vacancy for a professional. 1.71 Overall, this evidence suggests that a decrease in skills shortages can have substantial benefits. From the standpoint of individual firms that are unable to hire as many skilled workers as they would like, the benefit in terms of higher sales could be on the order of 15 percent. A problem with these “micro” estimates i s that they focus exclusively on the costs to individual manufacturing firms of skill shortages, ignoring other potential benefits from expanding the supply of skilled workers. If an increase in the supply of skilled workers allowed Thailand to expand into sectors that are intensive in skilled workers, there could be substantial increases in average earnings as more and more skilled workers would be able to achieve higher earnings. Inthe Appendix, we quantify these potential “macro” benefits. Chapter 3 of this report will discuss skill shortages and policies to alleviate them inmore detail. Figure 1.31: Benefitsfrom Relaxing Skills Shortagesand Number of Daysto FillVacancy for Professional . . Clothing Wood . ProductslFurnit 7 – RubberiPlastics 6 – MachinerylEquipmen FoodProfessing . 6 – . Textiles ElectronicslElec. 5 – Appliances – ~ ~~~ 0 5 10 15 20 25 30 35 40 45 50 % Sales Gainfrom ReductioninSkiuShortages ThirdApproach 1.72 The collection of PICS firm-level data across countries by the World Bank has resulted in various Investment Climate Assessments and studies that investigate how elements of the business climate impact firm productivity. This effort i s still in an exploratory phase, and thus several methods are currently in circulation, all of which have strengths and weaknesses. An in- depth review of the different methods is beyond the scope of this chapter and the interested reader can refer to studies such as Dollar, Hallward-Driemeier and Mengistae (2005), Escribano and Guasch (2005), Haltiwanger and Schweiger (2005) and Eifert, Gelb and Ramachandran (2005). Essentially, the studies are of two varieties. The first variety obtains measures of firm TFP from a production function using OLS, Olley and Pakes (1996) or Levinsohn and Petrin (2003) estimation techniques. The effect of business climate indicators on these TFP measures is 42 estimated in a regression framework.52Some studies consider the business climate indicators at the firm-level, However, that may result in small sample sizes since not all firms report the information on all indicators. That approach also suffers from reverse causality problems, Le., the fact that there may be a correlation between firm-level unobservable influences on TFP and the incidence of business climate constraints or firm reactions to business climate constraints. To address this endogeneity problem, Dollar, Hallward-Driemeier and Mengistae (2005) have considered averages of the business climate indicators taken at the region-industry level. The argument i s that such averages represent the business climate that i s “exogenous” to a given firm. However, even considering those averages, the regressions may suffer from other econometric problems. On the one hand, if the regressions include “too many” business climate indicators, these are likely to be correlated with one another and their estimated effects may be biased due to multicollinearity. On the other hand, if the regressions include “too few” business climate indicators, the corresponding coefficients may suffer from an omitted variable bias. The second variety consists of the OLS estimation of a productionfunction that includes directly the business climate indicators, along with inputs. The rationale for this approach i s that business climate variables can be used as observable firm-specific fixed effects, which help in controlling for the endogeneity of inputs in the production function. This type of studies also considers region- industry averages of the business climate indicators and includes a very large number of indicators in the initial estimation and progressively drops those that are not significant and whose elimination does not change substantially the signs on other variables.53 1.73 InTable 1.11,we show the results from regressions of firm performance measures for the MOIpanel on business climate variables taken as an average for industry-region cells to mitigate problems of reverse causality.54The regressions also include firm characteristics as in Table 7 and industry and region dummy variables to control for unobserved fixed industry and regional characteristics. A crucial assumption that i s made for estimation by the aforementioned studies — and that we also make — i s that the business climate i s constant over the time periodof the PICS, while inputs and output vary over time. This assumption i s made due to the nature of the PICS that collects data for the business climate at only one point in time, while it collects three years of data on production variables. However, the most important question for policy-makers is: What is the effect of changes in the business climate on firm performance? For such analysis, one needs to wait for a second or third wave of PICS to be conducted, namely in Thailand, so that time-varying measures of performance can be linked to time-varying businessclimate indicators and a more econometrically sound procedure can be used. 1.74 Given this crucial caveat as well as the econometric problems discussed earlier, the results in Table 1.12 need to be taken with caution as indicative of a link between the business climate and performance, but no meaning should be attached to the estimated coefficients. Firm productivity i s significantly weaker when longer time periods to clear customs for exports are faced. Also, firm performance i s negatively associated with several aspects of the bureaucratic 52The studies also consider other measures of firmperformance such as labor productivity, sales growth, investment or employment growth. 53 Multicollinearity problems may still be present in this approach. Moreover, another problem is that while the rationale for this approach i s that business climate variables can be used as observable firm-specific fixed effects, this i s complicated by the businessclimate variables entering the regressions as region-industry averages. 54 We cluster the standard errors in the regressions to account for the fact that the same value of the variable is repeated across firms inaregion-industry cell. 43 burden such as the number of inspections and the time spent by managers dealing with regulation. Some problematic aspects of infrastructure have a negative impact on firm performance such as the number of power outages, the production lost to power outages and the number of days to obtain an electricity connection experienced in the region and industry. A strong positive correlation i s found in firm performance and the percentage of production costs that transport and logistics costs represent. This may seem counterintuitive if one thinks of such regions and industries as those operating in more remote locations with worse access to markets. However, that variable may also be proxying for dynamic regions and industries with more exports which we find to be positively linked to performance. Finally, regions and industries with higher fraction of informal financing of investment have weaker performance. Table 1.12: BusinessClimateandFirmPerformance Regressors Total Factor Sales Growth Labor Productivity Productivity Time. to FillVacancy for Professional (region-industry) 0.461*** 0.218** -0.196 (0.076) (0.098) (0.273) Vacancies for SkiiledWorkers as % of Total (region-industry) 1.366*** 0.522 0.979 (0.433) (0.354) (1.547) Days to Obtain Import Permit (region-industry) 0.079*** -0).144*** 0.151** (0.014) (0.023) (0.058) Days to ObtainOperating License (region-industry) 0.099 -0.174** 0.728*** (0.059) (0.076) (0.208) Number of Inspections (region-industry) -0.018 -0).184*** 0.233** (0.027) (0.037) (0.101) Days to Clear Customs for Exports (region-industry) -0.532*** -0.277*** 0.084 (0.037) (0.058) (0.152) Percent of ManagerTime Dealing with Regulation (region-industry) -0.004 0.047* -0.296*** (0.018) (0.024) (0.073) Number of Power Outages (region-industry) -0.037** 0.030* -0.098* (0.016) (0.015) (0.053) Production Lost due to Power Outages (as % sales) (region-industry) -0.018 -0).186*** 0.383*** (0.028) (0.036) (0.107) Days to Get Electricity Connection (region-industry) -0,364 -0.973*** 2.316*** (0.222) (0.265) (0.808) Transportnogistics Costs as % of Costs (region-industry) 14.608*** .15.983*** 71.190*** (3.248) (4.290) (12.721) Percent of Sales with PaymentsOverdue (region-industry) -0).016 0.035 -0.134 (0.023) (0.025) (0.080) Percent of InvestmentFinanced from InformalSources(region-industry) -0).131*** 0.300*** -0.747*** (0.041) (0.056) (0.164) Size (Current Employment) 0.039** -0).063*** -0.007 (0.015) (0.017) (0.049) Exporter Dummy (more than 10%) 0.032 0.056 0.365* (0.042) (0.056) (0.181) ForeignOwnership Dummy 0.150*** 0.031 0.63I*** (0.050) (0.065) (0.142) Capital Vintage (%Mach.Under 5 Years) -0.1 -0.031 0.157 (0.083) (0.149) (0.270) R&D SpendingDummy -0.016 0.016 0.018 (0.026) (0.059) (0,138) RegionDummies Yes Yes Yes Industry Dummies Yes Yes Yes Year Dummies Yes Yes Yes N.Observations 591 585 593 R-squared 0.97 0.07 0.35 … Notes: OLS estimation is used. Standard errors clusteredby region and industry are in parentheses. ***, ** and * represent significance at the 1, 5 and 10 percent confidence levels, respectively. Sales growth is not available for 1999,thus 1999dataonTFPand labor productivity is not included inthe estimation. 44 E. CONCLUSIONS 1.75 Improvements in TFP have played a relatively modest role in Thailand’s past growth performance. For Thailand’s future growth, there i s a needto continue to investing inhuman and physical capital. However, a very crucial drive will come from improving productivity growth. 1.76 A good business climate is required to provide firms with the incentives to invest, innovate and grow. A good business climate will also provide individuals with incentives to invest in skills valued by dynamic and growing firms. Thailand’s business climate presents some good elements compared with many regional competitors, but it does exhibit some key vulnerabilities: (a) shortages of skilled workers; (b) dissatisfaction and uncertainty about the overall economic situation; (c) regulatory burden and; (d) infrastructure and support services. Skill shortages and regulatory burden have a particularly harmful effect. Evidence from Thai firm-level data indicates that the most dynamic firms, i.e., those that are larger, foreign-owned, or export or use more computerized machinery, are most adversely affected by these factors. Making progress inthese four areas should be the focus of policy efforts inthe short- to medium- runinorder to provide the foundations for sustained productivity improvements and growth into the future. 1.77 This chapter points to several key areas of policy relevance that require further investigation and in-depth analysis. First, firms identify regulatory burden as a key weakness in the business climate in Thailand, but this report does not analyze this issue in depth. A detailed assessment of key elements of the regulatory environment and the bottlenecks it creates, such as labor market regulations, customs procedures for imports and the legal system, i s crucial to inform future policy moves to ease regulatory burden. The joint World Bank-Bank of Thailand “Assessment of Regulatory Burden on Private Businesses in Thailand” i s an important step in that direction, and its findings will be an important complement to this report. 1.78 Second, while this chapter suggests that productivity improvements are crucial for Thailand’s future growth performance, the evidence on determinants of productivity at the firm level based on the Thailand PICS i s only a first step. The main problem with this survey i s that it observes firms only at a single point in time. Additional follow-up surveys will be needed to collect data on how firms respond over time to different policies and opportunities, which inturn will shed more light on the effectiveness of various policy interventions on firm performance. Additionally, the analysis of trends and determinants of productivity growth for industries in Thailand, which i s absent from this chapter, would benefit from the collection of census data covering every single establishment engaging in manufacturing activity in Thailand at regular intervals, inthe same way that a 1996 census was collected. 1.79 The remainder of this report delves into some of the key vulnerabilities in Thailand’s business climate in more detail. Chapter 2 provides a more detailed analysis of regional differences inindustrial activity and the business climate. Chapter 3 analyzes in greater detail the skill shortages felt by Thai firms and outlines policy responses. Chapter 4 investigates issues of technology adoption and innovation in Thai firms and evaluates the effectiveness of policy interventions designed to expand technology use. Chapter 5 focuses on information technology use and services. 45 References Aitken, B.andHarrison, A., 1999.”Do Domestic Firms BenefitfromDirectForeign Investment? Evidence from Venezuela,” American Economic Review 89,605-618. Arnold, H.andJavorcik, B., 2005. “Gifted Kids or Pushy Parents?Foreign Acquisitions and Plant PerformanceinIndonesia,” World Bank Policy ResearchWorking Paper. Bernard, A. andJensen, J., 1999. “Exceptional ExportPerformance:Cause, Effect, or Both?,” Journal of International Economics47, 1-25. Bosworth, B., 2005. “Thailand’s Long-Run Growth 1977-2002,” Backgroundpaper for Thailand: Growth Investment Climate, and FirmCompetitiveness. Dollar, D.,Hallward-Driemeier, M., andMengistae, T., 2003. “Investment Climate and Firm PerformanceinDeveloping Economies,” World Bank Policy ResearchWorking Paper. Escribano,A. and Guasch, J., 2005. “Assessing the Impact of the Investment Climate on Productivity UsingFirm-LevelData: Methodology and the Cases of Guatemala, Honduras and Nicaragua,” World Bank Policy ResearchWorking Paper. Fernandes, A. andIsgut, A., 2005 “Learning-by-Doing, Learning-by-Exporting, and Productivity: Evidence fromColombia,” World Bank Policy ResearchWorking Paper 3544. Haltiwanger, J. and Schweiger, H.,2005. “Allocative Efficiency andthe Business Climate,” Mimeo, University of Maryland. Jensen, J., McGuckin, R., and Stiroh, K.,2001. “The Impact of Vintage and Age on Productivity: Evidence from US.Manufacturing Plants,” Review of Economicsand Statistics 83,323-332. Kraay, A., 1999. “Exports andEconomic Performance:Evidence from a Panelof Chinese Enterprises,” Revue D’ Economie du Development 1-2, 183-207. Levinsohn, J. and Petrin, A., 2003. “Estimating ProductionFunctions UsingInputsto Control for Unobservables,” Review of Economic Studies 70, 317-341. Navaretti, G.and Venables, A,, (eds.) 2004. Multinational Firmsinthe World Economy. Princeton UniversityPress. World Bank. 2003. Improvingthe Investment Climate inChina. World Bank. 2002. Improvingthe Investment Climate inIndia. World Bank. 2004a. MalaysiaFirmCompetitiveness,Investment Climate and Growth. World Bank. 2004b. Improvingthe Competitivenessof Cities by Improvingthe Investment Climate: Ranking 23 Chinese Cities. World Bank. 2004c. India: Investment Climate and Manufacturing Industry. World Bank. 2004d. Indonesia: Investment Climate Assessment. 46 2. REGIONALINVESTMENTCLIMATEAND FIRM PERFORMANCE A. INTRODUCTION 2.1 This chapter examines the regional distribution of the manufacturing sector in Thailand. There are two main sections. The first section examines changes in the regional composition of manufacturing VA since the early 1980s, then investigates the changes in the spatial distribution and sectoral concentration of manufacturing employment from 1996/7, just before the Asian crisis, to 2001/2, well into the economic recovery. Two important trends emerge. First, manufacturing activity has increasingly shifted from Bangkok and the Vicinity to East and Central regions. Second, the composition of the manufacturing sector has shifted towards more high-tech industries. 2.2 The second section examines what i s driving these shifts in the spatial distribution of economic activity. Building on the insights of Chapter 1that stressed the crucial role played by the investment climate in influencing firm performance, the role of the investment climate in explaining the varied regional performance in Thailand is examined. Using the 2004/5 PICS data, regional variations in the Investment Climate are firstly examined and related to the current spatial distribution of economic activity in Thailand. Given that the investment climate data i s available only for a cross-section of firms at a single point of time, it i s impossible to explain the changes in economic activity inThailand over time, but it does have the ability to potentially explain current regional performance. 2.3 The results of this analysis show that investment climate concerns vary widely across regions in Thailand. Bangkok, Central and the East are considered to have the best business climates overall, which appears mainly attributable to their better provision of infrastructure. The regulatory environment in Bangkok i s considered an important constraint to business however. The South i s generally considered the region with the worst business climate. Furthermore, there i s preliminary support both in the regional TFP data and firm perceptions that regional differences in the local business climate can have a substantial impact on firm performance. 2.4 In order to further examine the role of investment climate in explaining regional disparities in firm performance, a TFP regression i s estimated that controls for simultaneity bias stemming from endogenous location choices (i.e. more productive firms may choose better locations making it difficult to disentangle the effects of this from the effects of location on productivity) and endogeneity of input choices. Using this framework, we examine the potential of investment climate indicators in explaining firm performance. The methodology employed i s similar to Dollar et a1 (2003), Escribano and Guasch (2004) and Haltiwanger and Schweiger (2005). 2.5 The results indicate that investment climate variables have a large role in explaining firm-level productivity (up to 40 percent in some cases). Company and product characteristics explain the rest. 47 2.6 We also estimate the fraction of the observed productivity gap between regions in Thailand that can be explained by the investment climate and other factors, based on the coefficients of these productivity regressions. 2.7 This analysis shows that while the investment climate explains a significant part of the productivity gap, clear differences in product and company characteristics across regions seem to account for more of the gap. Yet, these differences in product and company characteristics across regions may well reflect responses of firms to problems with the local business climate. B. OVERVIEW OFREGIONALDEVELOPMENTINMANUFACTURING 2.8 Manufacturing inThailand has grown inimportance over the last 25 years. Inspite of the investment slump and the Asian crisis, the manufacturing sector expanded its share in GDP. The sector i s now approaching two-fifths of GDP, compared with one-third of GDP before the Asian crisis and just over one-fifth of GDP in the early 1980s. The expansion i s closely linked to the boom in exports, which increased from about one-fifth of GDP in the early 1980s to about 45 percent before the Asian Crisis and now contributes close to two- thirds of GDP. There has been a significant shift in the composition of exports towards higher VA items since the financial crisis. We observe a decline in the exports of textile and garments products, compensated by a surge in exports of electrical machinery & parts, non- electrical machinery & parts, and vehicles & parts. Such high-tech categories of exports accounted for 44 percent of Thai export earnings in 2004. 2.9 While the importance of manufacturing has grown, the role of Bangkok and the immediate vicinity (which we refer to as the “Vicinity”) as Thailand’s factory hub has declined (see Figure 2.1). During the 198Os, their combined share of national GDP was between 65 to 70 percent. By the time of the Asian crisis, it had fallen to just more than 50 percent, and it now stands at 46 percent. Most of the decline was due to Bangkok, which suffered the brunt of the adjustment triggered by the Asian crisis. Since the early 198Os, Bangkok’s contribution to Thailand’s manufacturing VA has fallen by 45 percent, and the contribution of the immediate surrounding areas has contracted by 15 percent. B y the same token, the manufacturing shares of Central’s six provinces almost tripled, and it doubled for the East’s eight provinces. These two sub-regions contributed just under one-fifth of manufacturing GDP in 1981 and now account for twice as much. The East’s contribution has exceededBangkok’s since 1996, and Central’s has surpassedBangkok’s share since 2003. 48 Figure2.1: Regional Share of ManufacturinginGDP (1981-2004) 2.10 B y contrast, the Northeast, North and South do not appear to have benefited from the expansion of the manufacturing sector beyondBangkok and Vicinity. Both the North and the Northeast currently contribute only four percent to the sector’s VA, which i s the same fraction as in the early 1980s. The South did even worse, and its share contracted to three percent from five percent over this period. The trends for the Northeast and especially the North look more encouraging since the late 1980s, with an average annual GDP growth rate in manufacturing of 10 percent, about two percent above the national average. However between 1999 and 2004, the annual growth rate dropped to 6.4 percent in the North and 4.4 percent in the Northeast, compared with a national average of 6.5 percent. In any case, whatever expansion has taken place over the last decade and a half has occurred from very low levels. 2.11 As manufacturing VA has shifted from Bangkok and Vicinity to other areas, the geographical distribution of companies and their work force has also changed. Combining the 1996/7 and 2001/2 manufacturing censuses, we get a detailed picture of the dynamics in the spatial spread of employment among business establishments from just prior to the Asian crisis to well into the recovery. Figure 2.2 plots circles of employment for enterprises with 10 workers or more, where the size of the circles corresponds to employment levels. This covers just over half of all manufacturing employment. There is a strong concentration of employment in and about the Bangkok area. Other parts of the countries are dominated by gaps, representing areas without manufacturing employment. Out of Thailand’s 7,400 tambons (i.e. localities), only 2,700 hadmanufacturing establishments in 1996/7. There were also significant regional disparities in 1996/7. For example, there were manufacturing establishments in four-fifths of Bangkok’s tambons and over half of Center’s tambons, but only in one-third of the North’s and the South’s tambons and only one-fifth of the Northeast’s tambons. 49 Figure 2.2: SpatialDistributionof ManufacturingEmployment, 1996/7 and 2001/2 .` d c t’q .I 2.12 Five years later, the clustering of companies inthe extended Bangkok area increased even further. Almost all of the 154 Bangkok tambons and more than three fifths of the 1932 Central tambons offered manufacturing employment. But the Northeast, where the share increased to just under one third, and the North, with an increase to two fifths, also saw more manufacturing employment. Only the South did not experience any increase. The employment numbers tell a similar story for the Central area. Between 1996/7 and 2001/2, Bangkok’s employment share declined to 23 percent from 28 percent, and the Central area’s share increased to 60 percent from 56 percent. The Northeast’s employment share increased by two percent while the North’s and South’s contributions remained unchanged. 2.13 Much of Thailand’s employment i s concentrated in a few sectors. Low-tech manufacturing, food products and beverage, wearing apparel, textile and furniture were the largest employers in 1996/7, accounting for over two fifths of all manufacturing jobs and remaining among the five most important sectors in 2001/2. The spatial pattern varies widely by industry. Figure 2.4 shows the employment maps for the eight industries covered in the 2004/5 PICS, which includes Thailand’s biggest employers. The concentration of employment in individual industries exceeds that of employment overall. The Northeast i s most represented in wearing apparel, electronic parts, textiles, food processing and furniture, which account for three-fifths of total employment. 2.14 The employment maps suggest important differences in diversification across Thailand. Figure 2.3 shows the Herfindahl Index by region before and after the Asian crisis. It equals the sum over the squared employment shares of all four-digit industries. A higher index implies a larger concentration of four-digit industries within regions. In general, the South and the Northeast lag behind other regions in terms of diversification. Among the 125 four-digit 50 manufacturing industries in Thailand, 29 expanded and 40 declined, while the rest remained largely unchanged. This structural change in products has increased the concentration of products in the North, Northeast and South overtime. As a result, the manufacturing sector became more concentrated interms of employment across four-digit industries. Figure 2.3: RegionalHerfindahlIndices (1996/7 and 2001/2) 2 5 0 0 2 0 0 0 1 5 0 0 1 0 0 0 5 0 0 0 I B K K V I C I C e n t r a l I E a s t I N o r t h ,`.N o r t h e a s t S o u t h 2.15 I s diversification within regions linked to the expansion of manufacturing employment? Focusing on the changes between 1996/7 and 200112, the rise of Central’s employment share would indicate it does, while the contraction in Bangkok’s share and the rise in the Northeast’s share would indicate it does not. Indeed, relating growth rates in overall manufacturing employment over this period to the Herfindahl Index in 1996/7 across Thailand’s 845 amphoes (districts) gives no clear relationship. Figure 2.4: Spatial Distributionof Employment of PICS Industries A. FoodProcessing B.Textile I C. WearingADDarel D.AutoParts 51 E.Electronic Parts & Amliances F.Rubber &Plastics G. Wooden Furniture & Products H.Machinery&Equipment f C. PIC SURVEY AND DESCRIPTIVESTATISTICS PIC Survey 2.16 In the previous section, the evidence suggests that considerable shifts in economic activity (measured both in terms of GDP and employment) have taken place in Thailand in recent years. InChapter 1, support was found for the influence of the investment climate on firm- level performance. In this section, we utilize the PICS dataS5to analyze whether the investment 55 The PIC survey consists of 1,385 firms and covers six regions; North, NorthEast, Central, Bangkok and Vicinity, East and South, and eight industries. 52 climate can also explain the current regional disparities in economic performance documented earlier inthis chapter.56 Descriptive Statistics 2.17 In Table 2.1 below, some descriptive evidence is provided to illustrate how firm characteristics vary across regions in Thailand. Several interesting results emerge.57The oldest firms are in Bangkok, which also has the lowest initial employment size of newly formed firms. The percentage of firms engaged in exporting activities i s highest in the South, but surprisingly there does not appear a strong positive correlation between exporting activities and the share of foreign ownership. Finally, it i s worth noting that TFP (details of estimation can be found in section D) i s considerably higher in Bangkok than other regions, while the South and the Northeast have the lowest TFP. In Box 2.1, we provide several case studies of businesses established in the Northeast in order to understand the dynamics of this region inmore detail. Table 2.1: FirmCharacteristicsby Region Source: Investment Climate Survey (2004), World Bank 2.18 Inorder to examine how the investment climate differs between regions, a series of graphs i s presented in Figure 2.5 below. Firms were asked to rate various elements of the investment climate in their own region relative to that in other regions. In order to improve interpretability, the results were transformed to the following: 2 if the surveyed firm believes that particular aspect of the investment climate i s superior in a given region compared with its own region; 1 if they are considered approximately the same; and 0 if the firm considers its own region to have a superior performance in that aspect of the investment relative to another region. The results were then averaged for each region. Thus, a value above 1suggests that the surveyed firm considers that region to have a superior investment climate to its own region. A value below 1 suggests the surveyed firm considers its own region to have a superior investment climate (according to that particular dimension). 56Unfortunately given that the PICS data is only available for a single point intime, we are unable to examine the role of investment climate ininfluencing these changes inregionaleconomic performance over time. ”The Appendix to this chapter provides further informationregarding the construction of these variables. 53 Figure 2.5: Firms’ Opinions of Investment Climate inOther Regions Relative to Their Own I Overall InvestmentClimate I 1 1 Bangkok Central East Lowr Uppr North Nonh East Nwth h t Power Supply Telecommunications Y c 2 , I F 2 2 1 f Bangkok Central East Lowr Upper Nwth South Bangkok Central East Lower Upper Nwth Swth NwthEasI NwthEasl Nwth East Nwth Eaal Transportation Availabilityof Suppliers 5 Y I 8 – 1 4 > I i Bangkok Central East h e r Upper NwUl Nwth East Nwlh East Bangkok Central East b w r Upper ‘ North South NorthEast NwlhEast 2.19 Examining the results presented inFigure 2.5, Bangkok, Central and the East are viewed as having a superior investment climate to other regions. Interestingly, inTable 2.1, these regions were also found to have the highest average TFP in Thailand, providing preliminary support for the role of the investment climate in explaining these regional disparities in economic performance. Focusing on particular aspects on the investment climate, Bangkok, Central and the East were perceived to offer the greatest advantages over other regions in Thailand in the areas of power supply, transportation, telecommunications and the availability of s~ppliers.~’Apart from the latter issue, all of these areas are related to infrastructure concerns. 58The other issues sampled — access to land and labor quality — showed little inter-regional variation and as such are not shown. 54 2.20 From the above analysis, it i s clear that Bangkok, and to a lesser extent, Central and the East, offer advantages in terms of infrastructure over other parts of the country. Another key aspect of the business climate concerns regulation. In Figure 2.6 below, various aspects of the regulatory environment affecting firms are highlighted for these three regions in order to examine how this aspect of the investment climate impacts on firms. 2.21 In terms of regulation, the key constraint appears to be customs and trade regulation, which nearly 40 percent of all firms consider a major or very severe constraint. Overall, the next most pressing regulatory concern i s considered to be tax administration, followed by corruption, labor regulation and business licensing. 2.22 However these overall concerns mask important inter-regional differences in the aspects of the investment climate that are considered the most constraining to firm performance. Interestingly, surveyed firms in Bangkok (on average) consider all aspects of the regulatory environment highlighted in this figure to be of more concern than firms in the East or Central regions. The differences are greatest inthe areas of customs and trade regulations and corruption. Central and the East by contrast perform comparatively well in these aspects of business regulation. While the absence of time series measures of these investment climate variables precludes analysis of the impact of these effects on the evolution of industry inthese regions, it i s interesting to speculate whether this combination of reasonable infrastructure and regulatory environments i s linkedto the expansion of the manufacturing share in Central and the East. 2.23 The perception of firms that Bangkok, Central and the East have superior investment climates i s given further support in Figure 2.7 below. Firms were asked to name the region with the best and worst business climates. Bangkok, Central and the East were clear choices as regions with the best business climate, while the South was considered clearly the region with the worst businessclimate. 2.24 While firms may perceive regions as having very different business climates, a key question that follows from this i s whether these differences in investment climate actually matter for firm performance. This question will be addressed in more detail in the next section of this chapter, but some preliminary support for the importance of the business climate on firm performance i s provided in Figure 2.8 below. Firms were asked to estimate how much their production costs would be decreasedincreased if they shifted to the business climate with the besuworst business climates respectively. While the measurement error implicit in this question i s obviously substantial, it i s nevertheless interesting to note the degree to which firms consider the business climate to impact on their performance. Shifting to the region with the best business climate in Thailand i s estimated to reduce costs by 10percent, and moving to the region with the worst businessclimate i s estimated to increase production costs by almost 30 percent. 55 Figure 2.6: Aspects of Regulationthat FirmsRate as Representing a “Major” or “Very Severe” Constraint on Their Operations Customsand Trade Regulations $ 450 ‘” 400 {‘E 350 300 p 8 8 250 -1E 200 150 -: loo 5 0 e. 0 0 Bangkok East Central Overall Corruption Tax Administration ‘0 25.0 ‘0 30.0 g” -{ .5 0 20.0 gt” 25.0 I 15.0 .!8- p810.0 I Y ; E 5.0 -E 5.0 Y # 0.0 # 0.0 Bangkok Eaet ce”lr3 Overall Bangkok East Central Overall Business Licensingand Operating Permits Labor Regulations b 14.0 gb 12.0 1 E : 10.0 8.0 EFt 54 8.0 – -1 4.0 Y 2.0 # 0.0 # 0.0 Bangkok Central Overall Bangkok Central Overdl 56 Figure 2.7: Bangkok, Centraland the East Considered the Regionswith the BestBusiness Climates; the South Considered the Worst Number of FirmsConsideringRegionto havethe Best Business Number of FirmsConsideringRegionto havethe Worst Environment BusinessEnvironment 450 0 800 u)4000 700 3500 u) ii 3000 600 ‘s 2500 5 500 8 200.0 O 400 5 150.0 300 1000 =!200 500 = 100 0 0 0 Figure 2.8: FirmsEstimate that Production Costs Can Vary by Almost 40% Dependingon the RegionalBusinessClimate Firms Perceptions of the Percentage Decrease / Increase in Production Costs if Located in Region with Best/Worst Business Environment -s 35 C 30 0 *- 25 u) 20 m 15 e e -c e- a a 5 o ce m -5 ae -10 -15 I 2.25 It is also interesting to examine regional variations when industries are divided along high-technology and low-technology dimension~.~’ In the Appendices to this chapter in Tables 2.2-2.7, regional variations in firm characteristics and investment climate perceptions are shown for these two sub-groups of firms. Surprisingly, a S L shortage i s viewed as a more important constraint in low-technology firms than high-technology firms. However bureaucratic burden, tax issues and infrastructure concerns all appear more pressing for high-technology firms than 59High-technology industries include: auto parts; electronic parts & appliances; rubber & plastics; and machinery & equipment. Low-technology industries include: food processing; textiles; wearing apparel; and wood furniture and products. 57 low-technology firms, which reaffirm the importance of Thailand addressing these important business climate constraints if they hope to climb the technology ladder. Box 2.1: BusinessCase Studies of Northeast Exporters The PICS analysis indicates that enterprise in the Northeast have lower productivity than Bangkok-based firms. Yet, while the Northeast manufacturing base is small, it includes many foreign-owned and export- oriented companies. What businesses locate in the Northeast? What products do they produce? This box gives eight examples of export-oriented firms, both local and foreign-owned. These firms produce agricultural (fruit and vegetable flour and sweetener, Jasmine rice, chicken products) and industrial (iron roofs, cargo services, trucks and carpets) goods and services. Three features stand out. First,the Northeast attracts companies through access to raw materials, workers and land as well as BO1incentives. Second, the products either build on traditional activities (rice, livestock and fabrics), service the storage and transportation needs resulting from the distance to consumer markets (cargo and trucks), or supply housing materials for the more than 20 million regional population. Third, while all companies consider lack of skilled workers a serious constraint, daily workers constitute a large part of the work force and only some firmsuse wages to attract qualified employees. Thaisun Food Product Co. is a Japanese and Taiwanese company producing canned fruits and vegetables. It is located in Nong Khai province in 1988, attracted by the availability of labor and the special BO1 investment incentives. While originally exclusively export-oriented, it now also sells to the Thai market. The company still buys its raw materials from the same five local suppliers as at the beginning of its operations. Their mainproblem is the lack of qualified workers. Corn Products Amardass is an agricultural joint venture, founded by the 2001 merger of a Thai firm and a US company, which holds 80 percent of the stock. The two main products are flour and sweetener, about 70 percent of which are exported. The company chose the Northeast for raw materials (mainly cassava) and BO1 investment promotions. Among the 300 employees, 20 percent are daily workers, receiving wages slightly above the minimumwage. The company is sometimes confronted with labor shortages. Deimos Holding Co. Ltd. was founded in 2000 in Nakhon Ratchaseemato provide cargo services to their U.S. parent Effen Food Co. Ltd, which provided half of the founding capital. Services include storage of goods, packaging and truck forklift logistics. The company has expanded to 32 permanent employees from 10as of 2004. Monthly wages are on average Bt7,600 (2003 figures). Bluescope Lysagh, Ltd. is a subsidiary of an Australian multinational steel company. Founded in 1988 as a coated steel factory in Prathumthani, the company has since expanded to Khonkaen, Rayong, Chiangmai and Songkhla. Specializing iniron roof construction and installment, about 70 percent of it goods are sold domestically. The main obstacle to expansion of this medium-size company is the lack of qualified workers, even though the company pays wages above the levels of civil servants and other private employers. The company provides basic in-house training to workers. Chiameng Rice MillCorporation Co., Ltd, founded in 1937, originated from the first rice millin Bangsue, Bangkok. By 1955, the company, which started exporting jasmine rice under the brand name “Golden Phoenix”, i s widely recognized in Thailand for its swan logo and “Hong Thong” rice. This family company, with its headquarters in Bangkok, has four branches and produces a full range of jasmine rice products. The company has invested continuously in modernizing its technology, which it considers superior to its competitors. The main foreign competition includes basmati-rice producing companies from India and Indonesia. There is little competition inthe input market as the company purchasesdirectly from farmers, selecting high-quality unhusked rice for which it pays a price premium. The location of the Sri Sa Ket branch was chosen for raw material supplies and land area. This branch employs about 150 workers with a salary of about Bt6,000 to Bt6,500. The main problemis lack of SL. 58 Kawna Kaisod Co.Ltd., a Thai company established in 1981 with BO1support, began chicken exporting in 1993. It belongs to a group of five companies, whose activities range from breeding and broil& farms, a frozen chicken factory and a chicken feather factory. The company invests regularly in upgrading of its technology and exports about 80 percent of its products to Europe and Asia. Raw materials are sourced mostly from local suppliers. Since its foundation, employment has increased almost every year to reach 1,400 employees in 2004. Only 15 percent of the workers are on monthly contracts. The main obstacle in employment is the lack of qualifications of the employees. A part of the profits is reinvested to avoid the needfor bank loans. Khon Kaen Cho Thawee Co., Ltd is a truck and trailer company and was founded by a former rice mill owner. The owner relied on the know-how of his foreign-trained sons to expand its business over the last 30 years. In the mid-l990s, the company entered a joint venture with Emil Doll Gmbh to upgrade its technology. The foreign company selects raw materials and is incharge of boosting productionefficiency. The most important products today are trailers and half-trailers that can handle heavy freight on every type of road. Some trailers have anti-vibration systems as well as axle systems that allow every wheel to turn freely enabling them to carry manufactured concrete weighing up to 160 ton per block. The company is careful to select only high-quality inputs, most of which are imported from abroad. Four fifths of the customers are foreign and much of the domestic customers are state-owned companies. The company expanded to 300 employees after the merger from about 200 permanent employees in 1994, before it declined to 150 workers after the Asian crisis. Today, it has recovered to 400 employees. Salaries, which were cut by about 20 percent in 1997, are on average Bt8,300. The mainproblem is lack of qualified labor. The company reinvests most of its profits, but also takes out bank loans to fundcapital investment. Carpet Maker Co., Ltd. is a joint venture of Thai business people, established in 1985 in Khonkaen. Originally producing silk, in 1987 it started to produce carpets for domestic market and began in 1998 to focus on exports. Today, exports account for 80 percent of total sales. In contrast to its technologically more advanced two main competitors, the company specializes in high-quality handmade production. Cloth and wool raw materials are imported from Europe, while the glue is ordered from a domestic supplier. The company had 335 employees in 2004 compared with 270 employees in 2001. About 60 percent are monthly employees, receiving a salary of about Bt6,000. One important obstacle is to find qualified employees for executive-level responsibilities. The company draws on bank credits to respond to new funding needs, in addition to credits from raw material suppliers. Source: Kaenmanee, Sumeth, Anongnuch Thienthong and Phumsith Mahasuweerachai. Case Studies of Business inthe NorthEasternRegion of Thailand. Khonkaen University. 2005. D. DETERMINANTSOFREGIONALCOMPANYLEVELTFP: COMPANY CHARACTERISTICSANDTHEINVESTMENTCLIMATE 2.26 In the previous section of this chapter, some preliminary evidence for the impact of regional differences in investment climate on economic performance was found. In this section, we examine this issue in more depth by analysing the impact of the investment climate on productivity. 2.27 The objective i s to estimate TFP at the firm-level and then relate it to firm structures, four-digit product effects and investment climate conditions amongst other factors. In order to estimate TFP, we require consistent estimates for returns to SL, UL and capital (K) in VA. Ideally the estimated returns should be firm-specific but there are insufficient degrees of freedom (time series) to do this. Instead, the regressions are estimated at the industry level by region. Again, due to data limitations, we aggregate to two broad industries (high-technology and low- technology) and region taxonomies, defined below. TFP i s estimated as follows: 59 for company i in time t (2001 & 2002), industryj (i.e. low-technology and high-technology) and region k (Bangkok Region +Two Central Provinces & Outside Bangkok Region – Two Central Provinces). 2.28 Consistent estimation of production function parameters requires dealing with two inter- related estimation biases — simultaneity bias and selection bias. Simultaneity bias arises out of the fact that input demands are, in part, determined by the manager’s knowledge of productivity levels. Selection bias stems from the notion that productivity affects a firm’s decision to locate in a particular region. We adapt an estimation algorithm developed in Olley and Pakes (1996) to deal with both of these biases. The benefit of the Olley and Pakes (1996) approach i s that there i s a structural model of the unobservable productivity effect that implies that given the optimal investment dynamics of enterprises and a set of observable state variables, one i s able to control effectively for the omitted unobservable using non-parametric techniques. We use the Olley and Pakes (1986) estimation routine to estimate our parameters on inputsinorder to derive TFPi,.60 2.29 We report our estimates of PI, p2and p3 in Appendix Tables 2.8 and 2.9. We use the OP3 estimates for outside of Bangkok and within Bangkok in low and high-tech industries. Allowing for endogenous selection into Bangkok we estimate: TFPit = VAit – .35 Slit – .35d i t-. 13kit inhigh-tech industries. TFPi,= VAit – .31 slit – -37ulit -.22 kit inlow-tech industries. Allowing for endogenous selection to locate outside of Bangkok we estimate: TFPit = VAit – .25Slit – .37~lit-.34 kit inhigh-tech industries. TFPit = VAit – .27 Slit – .36ulit -.34 kit in low-tech industries. 2.30 The elasticities with respect to S L are lower than for UL but one should note that the skilled share of the labour force i s small and hence the average part of the elasticity i s also small, giving an overall small elasticity. The marginal effect i s still higher. 2.31 We see from Tables 2.8 and 2.9 (included in the appendix) that mean productivity in Bangkok in both low- and high-tech industries i s higher. On average, the difference in TFP between Bangkok and other regions i s about 30 percent. High-tech industries have higher mean productivity within each location. 2.32 In our estimation of firm-level productivity, TFP, we allowed for selection to location effects non-parametrically (a selection equation i s estimated, driven by investment climate variables among other factors. This probability, along with other factors, enters the non- parametric structure to control for our omitted variable, TFP, to get consistent estimates of the observable inputs in the production function. Then we back out TFP). As a second step, we want 6o See Appendix I1for further details. 60 to decompose parametrically our estimated productivity into the effect of firm characteristics (e.g. employment size at birth, age, share of skilled workers, share of local workers, share of raw materials from local sources, presence of imports, presence of exports and presence of foreign ownership), four-digit product effects and the investment climate, among other factors (time dummies, Heckmanlambdacontrolling for regional selection effects). 2.33 Our expected TFP returns from these factors are documented in the first column of Tables 2.10 and 2.11 (presented in the appendices) for Bangkok companies in high- and low-tech industries, respectively. The TFP returns for companies outside Bangkok in high- and low-tech industriesare respectively documented inthe first columns of Tables 2.12 and 2.13 (shown inthe appendices). For parsimony reasons, we only report the investment climate dummies that are significant, but we include all the variables in each regression outlined inthe data appendix. 2.34 High-Tech IndustryinBangkok: The results suggests that firms perform better when they are more intensive in skilled workers, the more they use local raw materials, the more they import raw materials, are export oriented and foreign-owned. In terms of the investment climate constraints, firms perform worse when they report problems with a shortage of Slulled Workers, Access to Foreign Credit, Macro Instability, Obtaining Land/Buildings, Currency Regulations, Lack of Business Support, Utility Prices, Official Corruption and Political Instability. 2.35 High-Tech Industryoutside Bangkok: The results suggest that firms perform better when they are more intensive in skilled workers, use fewer local workers and are foreign owned. In terms of the investment climate constraints, firms perform worse when they report problems with Telecommunications, Tax Rates, Access to Land, Labor Regulations, Anti-competitive Practices, and Supply of Infrastructure. 2.36 Low-Tech Industry in Bangkok: Firms perform better when they are bigger, older, more intensive in skilled workers, the more they use local workers, the more they use local raw materials, export oriented and foreign-owned. In terms of the investment climate constraints, firms perform worse when they report problems with Political Instability, Access to Credit, Macro Instability, Lack of Business Support, Utility Prices, Competition from Imports and Crime and Theft. 2.37 Low-Tech Industry OutsideBangkok: Firms perform better when they are older, use less local workers, export oriented and foreign-owned. Interms of the investment climate constraints, firms perform worse when they report problems with Telecommunications, Access to Land, Labor Regulations, Tax Administration, Macro Instability and Access to Credit. 2.38 In Figures 2.9-2.11 the distribution of firm-level TFP residual explained by investment climate factors i s documented by region. Investment climate constraints explain a good proportion of this variation in the residual TFP. These factors are more important in the low- tech industries and outside of Bangkok and in some instances, account for up to 40 percent of firmTFP, mainly inthe Northeast. 61 Figure2.9: InvestmentClimateResidual 3 CCMaU pmv Figure2.10: Low-TechInvestmentClimateResidual N3Rh Centrs1-2 Pro” Bangbk+2 central prm’ 3 1 1 1 2 1 0 East Northeast South E :3 Figure2.11: High-TechInvestment Climate Residual 2 I 0 E : 2.39 This analysis shows that while the investment climate explains a significant part of the productivity gap, clear differences inproduct and company characteristics across regions seem to account for more of the gap. Yet, these differences inproduct and company characteristics across regions may be endogenous to the local business climate. Further research i s requiredto explore these potentialindirect influences of the investment climate on firm performance. 62 E. CONCLUSIONS 2.40 In this chapter we have documented regional disparities in economic performance in Thailand and have analyzed the potential role of the business climate in explaining these disparities. 2.41 The descriptive analysis provides evidence that the manufacturing activity has increasingly shifted from Bangkok and Vicinity to Central and the East. As the investment climate data are only available for a single cross-section in time we are unable to link these trends over time to changes in regional investment climates. However, we can gain insights of the role of the investment climate in influencing current regional disparities in economic performance. 2.42 The descriptive analysis shows that investment climate concerns vary widely across regions in Thailand. Bangkok, Central and the East are considered to have the best business climates overall, which appears mainly attributable to their better provision of infrastructure. The regulatory environment appears to represent an important constraint on firms inBangkok relative to firms in other regions of the country. The South i s generally considered the region with the worst business climate. Furthermore, there i s preliminary support both in regional TF’P data and firm estimates that regional differences in the local business climate can have a substantial impact on firm performance. 2.43 In order to further examine the role of investment climate in explaining regional disparities in firm performance, a TFP regression i s estimated that controls for location selection and input choice endogeneity bias using the methodology developed by Olley and Pakes (1996). The results indicate that investment climate variables have a large role in explaining firm-level productivity (up to 40 percent in some cases) and productivity gaps across regions. 63 References Dollar, D., Hallward-Driemeier, M.andMengistae, T. “Investment Climate andFirmPerformancein Developing Economies” World Bank, 2003 Escribano, A. and Guasch, J.L., “Assessing the Impact of the Investment Climate on Productivity Using Firm-Level Data: Methodology and the Cases of Guatemala, Honduras and Nicaragua” World Bank, Nov. 2004. Haltiwanger, J.C. and Schweiger, H.”Allocative Efficiency and the Business Climate”, University of Maryland, 2005. Olley, G.S. andPakes, A., “The Dynamics of Productivity inthe Telecommunications Equipment Industry”, Econornetrica64, 1996: 1263-1297. 64 3. SUPPLYINGSKILLS FOR COMPETITIVENESS A. INTRODUCTION 3.1 Firms in Thailand identify the lack of appropriate skills as one the most binding constraintsto doingbusiness. As shown inChapter 1,more than a thirdof Thai firms surveyed report inadequate worker skills as a major obstacle to their activities. Concerns about skills of the labor force are pervasive and are felt equally by exporters and non-exporters, by domestic as well as foreign-owned firms. These concerns affect firms in all regions of Thailand, including the North East, which may suggest a changing nature of the labor demand as the manufacturing sector transitions to high-technology products. Figure 3.1 below shows that, irrespective of size and industry, firms are affected by this constraint61, though it i s particularly strong in the garments industry. A skilled workforce, not just raw rural labor, seems to be what firms need at this stage of Thailand’s development. Improving the skills of the labor force is the necessary condition for Thailand to move from low-tech assembler to high-tech manufacturer and escape the strong competition from Bangladesh, China and Vietnam on the low-endof manufacturing. Figure 3.1. Skills are a Major Constraint for Thai Firms Percento P h IdenWyiq Skills andEducationofAvailable Percentof FlmIQnUtylngSWUS andEduutianof Available Warltelsas a WoAen as a “Severe” or “Very Severe” Obstacle 60.0 “Severe” or”Very Severe” Obstacle 50.0 40.0 30.0 20.0 10.0 nn snau Medium 3.2 This chapter reassesses Thailand education and skills needs in light of future aspirations of the country. It examines the level, distribution and quality of shlls and competencies of workers in Thailand, as well as the match between these qualifications and those required by businesses. The chapter reviews macro evidence to uncover the extent of the Thai gap in skills performance. Then it uses PICS data to highlightthe disequilibria inthe market for skills. The chapter conjectures that the widespread perception of skills inadequacy relates, in fact, to two distinct types of disequilibria in the market for skills. The first i s skills shortage, and the second i s skills mismatch or skills deficiency. There i s a shortage when there are not enough people available on the labor market to do the jobs that are available in firms. Skills mismatch 61This figure uses the results of the closed-end question, see chapter 1. 65 refers to a situation where firms’ existing staff do not have the skills they need to do their job effectively. 3.3 Convergent sources seem to suggest that a key failure in the market for skills in Thailand is a combination of insufficient supply of educational output and the mismatch between what firms want and the type of skills supplied by the education system. First, unemployment has been declining over time. The Thai unemployment rate has declined to 2.1 percent in 2004 from 3.1 percent in 2001. Second, in a survey conducted by the National Statistical Office (NSO) among the unemployed graduates in 2002, about 57 percent of the respondents cited “lack of experience and the fact that their qualifications do not meet with employers’ demand” as the main reason why they have remained unemployed.62Third, Chapter 1 has shown that skills shortages are causing a skills mismatchinfirms as firms are operating with a ratio of skills to unskilled workers which i s lower than the optimal level in their industry. Econometric estimates suggest that this wrong skills mix at the firm level results in severe output losses, with an average firm losing nearly 15 percent of its output. 3.4 Making extensive use of the worker survey data from the PICS, we find compelling evidence of both skills shortages and mismatch in Thai manufacturing sector. Firms pay large wage premiums to tertiary education graduates and to workers who receive training in technical skills, indicating extreme levels of excess demand for the highest S L in Thailand. In addition, the gap between returns to secondary and tertiary education has not narrowedovertime. While the return to a high-schoolcertificatei s only 5.3 percent, the marginal effect of receiving a university-level degree i s nearly 35 percent. Furthermore, the incidence and intensity of hard-to- fill vacancies are very high. Nearly 80 percent and 95 percent of Thai manufacturing plants surveyed have had vacancies for professionals and production workers respectively in the last two years. Incidence of professional vacancies i s by far greater than in Malaysia and Indonesia and the Philippines. Also, it takes more than six weeks in Thailand to fill a vacancy of a skilled production worker or a professional, which i s longer than in any other benchmark country. The skills mismatch i s evidenced both from the employee and the employer’s perspectives. First, in relation to what employers consider as the desirable skills, Thai managers consistently rank their workers as having poorer skills. A large percentage of managers from the PICS rate Englishand IT skills of their Thai employees as poor — a greater share than seen in Malaysia. Nearly 60 percent of managers rate the English skills of their local professional workforce as poor. More 40 percent rate their IT skills as poor. Second, we provide an employee’s perspective, a self- assessment of their skills’ adequacy. Findings are fully consistent with managers’ assessment. More than 72 percent of the 14,000 Thai employees interviewed identify English language proficiency as the most severe constraint in doing their job. IT skills are a distant second with respectively seven percent and six percent of managers and professionals identifying it as the skill they lack the most. Skills shortages and mismatch could hinder Thailand’s ambition to sustain competition inhigh-tech manufacturing. They could also prevent the country to transition to a service-led model of growth. 3.5 The Thailand PICS i s a rich dataset particularly suited to the analysis of skills issues. The survey included interviews with CEOs, Human Resource Managers and workers. Employers 62Survey on Working Situation and Unemployment of Middle and Upper Class 2002, National Statistical Office (NSO) 66 were asked about their experience in filling vacancies and their experience with deficiencies in the quality of their existing workforce. Employees were asked independently about the skills they lack most for doing their jobs as well as about the adequacy of their field of education as it relates to the work they do. Also, the survey offered insights into how firms and workers use technology and the public institutions set up by the Thai Government to provide these services. The worker survey module of the ThailandPICS provides information on the demand-side of the labor market and individual-specific information needed to properly assess the shortage or mismatch of skills inThailand. 3.6 The next section reviews the macroeconomic evidence to uncover the extent of the gap in skills performance. Section 3 uses PICS data to highlightthe disequilibria inthe market for skills and explores the intensity of the shortages and mismatch. Section 4 provides an examination of the key drivers of skills adequacy, including education and in-service training. Section 5 provides a summary and suggests the policy implications of the findings. B. ASSESSINGTHAILAND’S SKILLS ADEQUACY: THEMACRO-EVIDENCE The Stockof HumanCapital Education Completion rates 3.7 Despite a remarkablesuccess in achieving universal primary education and a long tradition of tertiary education, Thailand’s education stocks still lag behind its level of development and its regional competitors. Chapter 1 has shown that the percentage of Thailand’s population with at least secondary education completed i s substantially lower than the norm for its income level. Further investigation reveals that the source of Thailand’s gap in education stock i s its poor secondary education completion performance. Disaggregating the above described data by secondary and tertiary levels, it appears that secondary education i s the source of Thailand’s gap in education stock. Figure 3.2 shows that Thailand completion rate for secondary education i s by far lower than the norm for its income level. The Thai completion rate was 4.1 percent in 2000 compared with 23.6 percent in Malaysia and 17.5 in the Philippines. Secondary education has failed to keep pace with economic development inThailand, marginally increasing from 3.5 percent in 1990 to 4.1 percent in 2000.63This modest change clearly did not match the increase in demand for skilled workers due to technological upgrading starting in the 1990s (see Zeufack, 1999), especially in automobile parts, electronics and machinery & equipment industries as shown in Chapter 4 of this report. 63We plot the relationship betweenreal per capita GDP (on the horizontal axis) and the percentageof the population aged 25 and older who have completed only secondary education. Data for comparator countries refers to 2000, while data for Thailand are shown for 1980, 1990,and 2000. 67 Figure 3.2: Thailand PerformsPoorly on Secondary Education I I t i I 6 7Nature4Logarilhmof8RealPerCapltaGDP9In2OM) 10 Source: RobertJ. BarroandJong-WhaLee, “InternationalDataon EducationalAttainment: UpdatesandImplications,” NBERWorkingPaper No. 7911,Cambridge, MA, September 2000; World Bank, World Development Indicators 2004. Author’s Calculation. Stock of Scientistsand Engineers. 3.8 The percentage of engineers and scientists in the (adult) population is very low. An alternative indicator of the gap in the stock of S L i s the number of engineers and scientists engaged in R&D. As shown in Table 3.1, Thailand lags behind all benchmark countries. In 1997, Korea had 2245 engineers and scientists engaged in R&D per million inhabitants against 74 for Thailand. This gap may limit the innovative capacity of Thailand. Table 3.1: Scientists and Engineers inR&D (Per Million People) Country 1992 1993 1994 1995 1996 1997 1998 1999 2000 Malaysia 85 114 92 160 Korea 2033 2642 2233 2193 2245 2009 2160 2319 China 353 351 454 473 387 420 545 Japan 5678 5148 6301 5369 4909 4962 5160 5196 5095 India 133 157 Singapore 2182 2699 2832 2991 3215 4140 Thailand 113 118 74 u s 3729 3730 3863 4099 Source:SIMA, World DevelopmentIndicators (April 2003). The Flow of Human Capital 3.9 Thailand has achieved a remarkable success in its primary education. In 2003, for primary level, the gross enrollment rate stood at 104.4 percent, transition rate to lower secondary lever was at 92.5 percent, the retention rate at 89.5 percent and literacy rate at 95.7 p e r ~ e n t ~ ~ , ~ ~ . 64 FromEducation inThailand 2004, pre aredby ONEC r65etentioneRatethe official school-agepopulationcorrespondin to the same level of education in given school-year. tGross of Enrollment Rate is total enrofment in a specific level of education, regardless of age, expressed as a a g school-year whoisare expectedto reach final year of primary education, Le. Grade 6. the percentage of a cohort of students enrdled in the first grade of primary education in a given 68 Net primary enrollment also showed the rising trend towards universal primary education, rising to 98 percent in200266from 92 percent in 1994. Expansion of primary education has been across all groups of population, regardless of income, location or gender. Especially, across income group, the trend shows that the net enrollment rate for the poorest quintile has been rising continuously over time, to 98 percent in 2002 from 89 percent in 1994, proving the success of the expansion to reach the poor household. 3.10 There has been a recent expansion of secondary education coverage, though not across all population groups. The overall gross enrollment rate for 2002 was 77 percent, after a long period of stagnation at 30 percent. However, the poorest group of population still does not benefit from the expansion of the secondary education as the gap between the richest and the poorest quintile has been quite stable over time. Other education indicators have also improved since the crisis. Transition rate from primary to lower secondary education rose to 93 percent in 2003 from 87 percent in 1999. Retention ratios for both upper and lower secondary are higher than their levels in 2002 (80 percent). Further expanding secondary education is critical to fulfill the country’s agenda oncompetitiveness. Expansionofsecondary education coverage is, indeed, a condition for countries to fully reap technological spillovers from FDI and trade, see Gill & al. (2002). C. ASSESSING THAILAND’S SKILLS ADEQUACY: THEMICROEVIDENCE 3.11 The previous section presented evidence on shortage of skilled workers from the supply side. This section explores the incidence of skills shortages and mismatch, using micro data and adopting a Work-Centered Approach (WCA), see Box 3.1. First we document evidence of skills shortages, and then we explore the presenceof skills mismatch. 66The Net Enrollment Rate is calculated by using data from Socio Economic Survey which is collected every two years. 69 Skills Shortages WagePremiums 3.12 One of the most effective ways to assess the extent of skills shortages is to estimate wage premiums, which capture how much the entrepreneurs are willing to pay for the scarce factor. Wage analysis provide strong evidence that tensions at the high end of the Thai labor market have resulted in high wage premiums to workers with tertiary education, reflecting the severity of skills shortages and the high value that managers place on skilled workers. The returns to education are estimated using data from the Workers Survey of the PICS for manufacturing establishments. The results based on a semi-parametric regression are presented in Appendix Table 1A and depicted in Figure 3.3. The estimates in Appendix Table 1A show clear increases in the slope for tertiary education, specifically from the 13th year of schooling and beyond. Figure 3.3 shows the mean log hourly wages by years of formal education, where the curve connects the average of predicted values of log hourly wages (for each year of formal education) from the semi-parametric estimation of log hourly wages on formal education. Premiums to workers with more than 16 years of schooling are much higher than those paid in Malaysia, indicating extreme levels of excess demand for the highest S L in Thailand. Also, it i s important to note that wage premiums for tertiary education graduates have been high for the past 10 years. As shown in Figure 3.4, studying successive labor force surveys since the early 199Os, it appears that the gap between returns to upper secondary (light blue line) and university education (yellow line) has been consistently large and has not narrowedovertime. Figure 3.3: MeanLog Hourly Wage by Years of Formal Education (Relativeto comparisongroup with 9 or less completedyears of formal education.) 120 f 100 / 8fi -z R 80 rnE ._ 1Y 60 1 CI 40 b” 20 12 CompletedYears of Formal Education 13 14 15 over 16 Source: Thailand Productivity and InvestmentClimate Survey 2004; Malaysia Productivityand InvestmentClimate Survey 2002. 70 Figure 3.4: Returns to University Education are Consistently Higher than for Lower Levels of EducationinThailand 13.000 12,000 11.ooo 9,ooO 8.000 7,000 6,000 5.000 4,000 3,000 2,000 1,000 FQl A91 FQP A92 F93 A93 F94 A94 F95 A95 FQ6 A98 F97 A97 F98 A98 FQQ A99 FOO PO0 FO1 PO1 FO2 W2 F03 pD3 F04 PO4 +None +Less than Primary +Pnrnary +Lower Secondaty +Upper Secondary -Vocationel Unimrsily -e-Olsrali Source:Thailand Labor Force Surveys, NorthEast Study. 3.13 The return to completing secondary school is low, and wage premiums to educational attainment only accrue to those receiving a college degree. Table 3.2 presents the direct estimates of the sheepskin effects for the sample of workers in the PICS dataset. The coefficients in Column 4 indicate that the return to a high-school certificate i s statistically significant but only 5.9 percent; the marginal effect of receiving the Por Wor Sor above a high- school certificate i s 6.3 percent but not statistically significant, and the marginal effect of receiving a university degree above a high-school certificate i s 36.9 percent and statistically ~ignificant.~~ Within the context of a simple human capital investment model, a person can choose either to study or get a job — the above results suggests no monetary incentive to attend high school if one believes that college is unaffordable or impossible. In other words, the decision of the individual to invest in his or her own human capital i s solely focused on whether he or she will ultimately attend and graduate from college and obtain a degree. 3.14 Because college degrees seem to be the only educational “signals” that are rewarded, there is no incentive to complete high school if graduatingfrom university is not feasible. A high-school certificate i s not viewed as a goal in terms of higher wages, but rather a means to a college degree. Therefore, choosing work to gain experience and skills may be more valuable than staying the extra years in secondary school and receiving a high-school certificate.68 Conversely, as seen in the regression results for Malaysia, Column 8, shows that the wage premium for receiving a high-school certificate i s `11.1percent and statistically significant; there are monetary rewards for attaining a high-school certificate in Malaysia. This may explain the differences in secondary as well as tertiary education attainment levels nationally in both Thailand and Malaysia. Given the tension on the Thai labor market, improving the quality of 67The percentage increase in wages associatedwith a dummy variable coefficient i s calculated as exp(P) – 1. Within a screening model of education, the employer is the uninformed first mover and uses academic credentials to identify potential employees with desirable traits that cannot be directly observed. In a screening equilibrium, workers of higher type choose more education, and employers are willing to pay higher wages to more educated employees. 71 secondary education i s likely to increase the demand for secondary graduates and their wage premium, providing a strong incentive to complete secondary school. It should be noted that, while returns to education are a good indicator for policy, governments should adopt a diversified strategy to educational investments at the national level rather than simply usingrates of return evidence to make decisions about how to expand education, see Hawley (2004). Table3.2. EstimatedSheepskinEffectsinThailandandMalaysia [Dependent Variable: individual log hourly wage] [DependentVariable: individuallog hourly wage] Thailand Malaysia (1) (2) (3) (4) (5) (6) (7) (8) HighSchool Diploma – 0.226″ 0.057′ 0.389” 0.105″ (0.034) (0.025) (0.030) (0.024) ~ morWor-Chor Marginal Effect Over High School DiplomalPor Wor Chor PorWor Sor 0.155″ 0.061 (0.047) (0.033) CollegeDegree 0.566″ 0.314″ 0.289” 0.225” (0.069) (0.042) (0.040) (0.031) Years of ComaletedSchooling 9 or less -0.433″ .0.215″ -0,.l50″ -0.098″ -0.566″ -0.521″ -0.148″ -0.143″ (0.012) (0.034) (0.010) (0.025) (0.029) (0.029) (0.022) (0.022) 10 -0.170′ -0.036 -0.055 -0.026 -0.298″ -0.253″ -0.005 0.003 (0.084) (0.090) (0.052) (0.054) (0.044) (0.044) (0.031) (0.031) 11 -0.001 0.001 -0.049 -0.051 -0.114” -0.079” -0.026 -0.017 (0.104) (0.103) (0.068) (0.068) (0.029) (0.029) (0.021) (0.021) 12 ref. ref. ref. ref. ref. ref. ref. ref. 13 0.033 -0.094 0.09 0.035 0.248″ 0.154″ 0.081″ 0.061′ (0.136) (0.141) (0.080) (0.081) (0.037) (0.036) (0.027) (0.027) 14 0.335″ 0.179” 0.126″ 0.071′ 0.482” 0.175” 0.128″ 0.053 (0.015) (0.047) (0.013) (0.034) (0.039) (0.043) (0.029) (0.033) 15 0.456″ 0.258′ 0.246″ 0.155′ 0.698” 0.288″ 0.237″ 0.103″ (0.107) (0.108) (0.061) (0.065) (0.044) (0.048) (0.034) (0.038) 16 0.799″ 0.238″ 0.371″ 0.073 0.869″ 0.338″ 0.333″ 0.125” (0.015) (0.070) (0.013) (0.042) (0.046) (0.055) (0.038) (0.043) more than 16 1.352” 0.835″ 0.598″ 0.325″ 0.980″ 0.442″ 0.359″ 0.130″ (0.060) (0.081) (0.030) (0.047) (0.046) (0.055) (0.034) (0.041) Potentialexperience 0.054” 0.053″ 0.028” 0.028″ 0.061″. 0.062” 0.029″ 0.029″ (0.002) (0.002) (0.001) (0.001) (0.002) (0.002) (0.002) (0.002) Potentialexperience2x10-2 -0.074” .0.073″ -0.041″ -0.041** -0.091″ -0.092″ -0.048″ -0.048″ (0.004) (0.004) (0.003) (0.003) (0.005) (0.005) (0.003) (0.003) Constant 3.040″ 2.819″ 2.927″ 2.875″ 1.227″ 1.164″ 0.857″ 0.844″ (0.014) (0.036) (0.017) (0.029) (0.032) (0.032) (0.030) (0.030) Worker characteristics No No Yes Yes No No Yes Yes Establishmentfixedeffects No No Yes Yes No No Yes Yes Observations 13,458 13,458 13,458 13,458 7,812 7,812 7,812 7,812 Numberof Firms 1,383 1,383 1,383 1,383 893 893 893 893 AdjustedR z 0.461 0.469 0.574 0.576 0.309 0.332 0.481 0.487 Total Returns to Por Wor Sor/College Degree over High School Diploma PorWorSor – 0.38 1** 0.118″ (0.056) (0.039) CollegeDegree – 0.791″ 0.371″ 0.678″ 0.330″ (0.076) (0.046) (0.044) (0.035) Note:Robuststandardm nare denotedinparentheses. * denotes significance at 5% level; ** denotes significance at 1% level. Source:Thailand Frductivity and Investment Climate Survey 2004; Malaysia Pmductivity andInvestment Climate Survey 2002. Authofs calculations. 72 TrainingPremiums 3.15 The return to training is substantial. Workers with basic IT skills (e.g. printing invoices), intermediate (e.g. word processing and e-mail) and advanced (e.g. programming) computer skills earn 9.9, 18.9, and 29.1 percent higher wages, respectively, than those with no computer skills. Workers who lack Englishproficiency in doing their jobs earn 2.6 percent lower wages than those with sufficient Englishskills, see Table 3.3 below. Workers who have received formal training from their current employers in the areas of marketing and managemendquality technologies earn 12.5 and 4.5 percent higher wages, respectively, than those who have not received any training from these current employers. Those who received training from their previous employer have 4.8 percent higher wages than those who did not receive training from their previous employer, whereas the return to outside training i s not statistically significant. Hard-to-Fill Vacancies 3.16 The presence of hard-to-fill vacancies at the firm level is one of the most used indicators of skills shortages in the literature. As shown in Figure 3.5 below, among PICS establishments, nearly 80 percent of manufacturing plants had vacancies for professionals in the last two years. Incidence of professional vacancies i s by far greater than in Malaysia and Indonesia (about 50 percent) and the Philippines (about 25 percent). Even India experiences a lower professional vacancy rate of approximately 70 percent. Furthermore, the incidence of skilled production worker vacancies in Thailand i s about 95 percent. Naturally, this i s greater than the incidence in Malaysia (about 80 percent) and India (about 75 percent). It i s also much higher than inIndonesia (about 50 percent) and the Philippines (about 40 percent). Shortages for professionals are experienced by almost all establishments in food processing, auto parts, electronics & electrical appliances and machinery & equipment. Nearly all establishments across the eight industries have experienced vacancies for skilled production workers in the last two years. Figure3.5: Vacanciesfor Skilled Workers and ProfessionalsAppear Hard to Fillin Thailand Compared with Other Countries Last2 Years Bangladesh B d – Thailaod – Thailand B m d r Malaysia India Bangladesh Malaysia India Indonesia ‘ Indonesia Philipines Philipines 73 74 3.17 Furthermore, it takes longer in Thailand than in any other benchmark country to fill a vacancy of a skilled production worker. More than six weeks are required to fill a vacancy for a professional or skilled production worker in Thailand. While Malaysia also experiences long search times to fill vacancies for professionals and skilled production workers, Thailand i s at a disadvantage i s terms of identifying suitable workers when compared with most other competitor nations. To fill a vacancy for a professional in the Philippines, India, or Bangladesh, it takes less than four weeks, and in Indonesia, it takes less than three weeks. In Bangladesh, filling a vacancy for a skilled production worker takes less than two weeks. Less than three weeks are needed in Indonesia, the Philippines and India. Longest times in Thailand are experienced by establishments in garments (more than eight and 12 weeks for professionals and skilled production workers, respectively); shortest times are experienced by establishments in electronics & electrical appliances (less than six and four weeks for professionals and skilled production workers, respectively). Figure 3.6: FillingVacancies for Skilled Workers and ProfessionalsTakes a Relatively Long Time inThailand – Time to FillV x m y 5rSkilledProductionWo&er in (weeks) Last2 Years – B d – Thailand Malaysia Philipines India Bangladesh Indonesia 3.18 The lack of appropriate basic and technical skills is the main reason of these vacancies. Firms were asked why they have difficulties filling the vacancies, as i s shown in figure 3.7 below. The fact that applicants do not have the required basic and technical skills i s the main reason why incidence and intensity of vacancies are so high. More than 80 percent of the managers cited these two reasons. These percentages are higher than in Malaysia, where approximately 65 percent of managers responded with these two reasons. In contrast to Malaysia, Thai managers do not feel that applicants demand very high wages (approximately 25 percent versus nearly 60 percent). 75 Figure 3.7: Inadequate Skills are the Key Driver of Vacancies Percentage o f Firms Pointing Out to Each Factor as O n e o f Main3 Causes f o r Vacancies 90 80 70 60 50 40 30 20 10 0 Applicants Univ. Not ApplicantsNot Applicants Not N o Applicants HighTurnover Demandvery Produce Have Required Have Required forUwkilled of New Recruits HighWage Sufncient Basic Skills Technical Skills Workers Number Position Graduates 1 Thailand 0 Malaysia 1 I s There a SkillsMismatchinThai manufacturing? 3.19 Several sources point to the existence of skills shortages and mismatch on the Thai labor market.Inthis section, we use the PICS to explore the skills mismatch both from the employee and the employer’s perspectives. Mismatchfrom an employee’s perspective:A self-assessment of skills’ adequacy 3.20 English language proficiency is the skill that workers lack the most in doing their job. Workers were asked to list the three skills that they lack the most in doing their job. Table 3.4 shows that more than 72 percent of workers identify English language proficiency as the most severe constraint in doing their job. This number i s even much higher than in Malaysia, where 48 percent of workers cited English as the most severe constraint. Among Thai workers, IT skills are a distant second in terms of percentages, with approximately four percent of the workers identifying it as the skill they lack the most. Logically, managers and professionals are the groups with the largest percentage of workers identifying IT skills as the skill they lack the most, 7.1 and 6.2 percent, respectively. 76 Table 3.4: Workers’ Self Assessmentof Skills Adequacy Number and percentage (inparentheses)of workers by occupation What is the first mostimportantskill workers lack indoing theirjob? (percentageinparentheses) Skill Unskilled Non- Management Professionals Production Production production Apprentice Total Workers Workers Workers Englishlanguageproficiency 180 671 1,505 4,626 2,277 763 10,022 (58.4) (67.4) (74.8) (74.5) (69.1) (74.6) (72.4) Professionalcommunicationskills 11 30 65 197 106 43 452 (3.6) (3.0) (3.2) (3.2) (3.2) (4.2) (3.3) Socialskills 11 21 41 152 63 27 315 (3.6) (2.1) (2.0) (2.5) (1.9) (2.6) (2.3) Teamworking 6 23 56 225 78 47 435 (2.0) (2.3) (2.8) (3.6) (2.4) (4.6) (3.1) Leadershipskills 10 32 43 130 116 21 352 (3.3) (3.2) (2.1) (2.1) (3.5) (2.1) (2.5) Time managementskills 11 49 37 90 95 12 294 (3.6) (4.9) (1.8) (1.5) (2.9) (1.2) (2.1) Adaptability 8 10 18 47 28 9 120 (2.6) (1.0) (0.9) (0.8) (0.9) (0.9) (0.9) Creativity/innovation skills 6 16 44 156 75 16 313 (2.0) (1.6) (2.2) (2.5) (2.3) (1.6) (2.3) Numericalskills 14 23 47 165 104 26 379 (4.6) , (2.3) (2.3) (2.7) (3.2) (2.5) (2.7) Problemsolving 7 18 32 110 76 16 259 (2.3) (1.8) (1.6) (1.8) (2.3) (1.6) (1.9) IT skills 22 62 80 180 157 23 524 (7.1) (6.2) (4.0) (2.9) (4.8) (2.3) (3.8) TechnicaVprofessional skills 12 30 42 126 87 16 313 (3.9) (3.0) (2.1) (2.0) (2.6) (1.6) (2.3) Other 10 11 2 8 34 4 69 (3.3) (1.1) (0.1) (0.1) (1.0) (0.4) (0.5) Total 308 996 2,012 6,212 3,296 1,023 13,847 (100.0) (100.0) (100.0) (100.0) (100.0) (100.0) (100.0) Source : ThailandProductivityand InvestmentClimateWorker Survey 2004. 3.21 Englishlanguage deficiency is present across all occupations and regions. Figure 3.8 clearly shows that nearly three out of four workers in Thailand lack the English skills needed to do their jobs. This result i s quite worrisome. The highest percentages are found among skilled and unskilled production workers and apprentices; nearly 75 percent of workers in these groups lack English skills required for their jobs. The extent of English mismatch for production workers i s likely to translate into productivity gaps, as they are the backbone of any manufacturing plant. This negative impact i s especially true regarding the lack of English proficiency among skilled production workers as they are the employees assigned to operate machinery, read and interpret the manuals and then instruct unskilled production workers. Re- enforcing the English curriculum in secondary schools and as the language of training in technical vocational schools would be useful. Similarly, workers in all regions exhibit the massive deficiencies inEnglishskills. 77 Figure 3.8: The Extent of English Skills Mismatch Share of workers rankingEnglish language proficiency as the first skill they lack the most indoing their job 80 50 – b40- L 30 – 20 – 10 – 0; Source:Thailand Productivityand lnvcsfmnt Climate Worker Survey 2004. 3.22 While workers in Thailand clearly lack English language skills in doing their jobs, many believe they have the skills needed to adapt to labor market changes. From Table 3.5, fewer than seven percent of workers in the total sample are not properly equipped. The industries with the highest proportion of workers with skills mismatch i s auto parts (12.3 percent) followed by rubber & plastics (8.6 percent) and textiles (8.3 percent). These findings are much brighter than in Malaysia, where more than one third of workers believedthey didnot have the skills needed to adapt to labor market changes. This surprising finding, that requires further investigation, may capture the optimistic nature of the Thai workers. 78 Table 3.5: Skills Mismatch ThailandPICS Worker Survey 2004. – Share of manufacturing workers by education and industry who declare not having the skills neededto adapt to labor market changes Total FoodProcessing Textiles Garments Autoparts Degree 2.3 0.9 1.2 2.7 5.1 Por Wor Sor 3.8 2.0 3.3 7.6 6.2 Upper secondarypor Wor Chor 6.5 2.1 7.0 7.3 15.7 Lower Secondary 8.7 3.5 6.9 9.0 19.5 PriIKlly 9.1 2.6 13.1 9.0 18.6 Total 6.6 2.2 8.3 7.7 12.3 Electronics & Electrical Rubber & Wood & WOO( Machinery & Appliances Plastics Furniture Equipment Degree 1.9 2.8 4.4 0.8 Por Wor Sor 4.6 4.3 1.2 1.o Upper secondaryh’or Wor Chor 5.O 8.9 4.7 1.8 Lower Secondary 8.7 11.5 10.0 2.4 Primary 12.6 9.8 8.8 2.8 Total 5.1 8.6 7 4 … 1.8 Source : Thailand Productivity and InvestmentClimate Worker Survey 2004. 3.23 The study also uncovers a serious education mismatch for those with education attainment less than a university degree. Analysis of employee data suggests that the skills shortage i s leading managers to a sub-optimal hiring policy, as employers are forced to hire fewer qualified workers. As shown in Table 3.6, more than 27 percent of workers with high school (Por Wor Chor) believe they are over-utilized injobs requiring a diploma. Similarly, more than 30 percent of workers completing only lower secondary education are doing a job that requires at least an upper secondary level education. This mismatch could cause productivity losses at the plant level. Table 3.6: EducationMismatch What level of education is more appropriate for your work? Degree Por Wor Sor Upper secondary/ Lower Por Wor Chor Secondary Primary None Degree 81.87 27.02 12.06 3.3 2.49 5.61 Por Wor Sor 12.95 52.29 15.18 7.36 3.67 1.87 Upper secondaryPor Wor Chor 3.75 14.67 48.75 20.64 12.86 6.54 Lower Secondary 1.02 4.01 15.82 49.3 27.29 12.15 Primary 0.35 1.88 7.07 16.66 49.13 48.6 None 0.07 0.13 1.12 2.74 4.56 25.23 Source :Thailand Productivity and InvestmentClimateWorker Survey 2004. Mismatch in relation to what employers consideras the desirable skills. 3.24 Thai managers consistently rank their workers as having poorer skills. A large percentage of managers from the PICS rate Englishand IT skills of their local employees as poor — greater share than seen in Malaysia. Nearly 60 percent of managers rate the English skills of their local professional workforce as poor. More than 40 percent rate IT skills as poor. Ratings are worse for local skilled production workers, with more than 90 and 80 percent of managers 79 rating their English and IT skills as poor, respectively. Compared with Malaysia, this i s quite high. Only about 12 and 20 percent of managers in the Malaysia PICS sample rate the English and IT skills of their local professional workforce as poor, respectively. Similarly, about 65 and 70 percent of managers inthe MalaysianPICS sample rate the Englishand IT skills of their local skilled production workers as poor, respectively. Figure 3.9: Managers inThailand Believe Their Workers Have Poor Skills 60 I O 40 30 10 10 0 D. DETERMINANTS SKILLSADEQUACY OF Assessing the quality of the educationaloutput 3-25 Thai secondary education students score lower than average in international tests. The latest international test for quality of education, the “Trend in Mathematic and Science Study” (TIMSS), revealed poor results for Thailand. Table 3.7 below shows raw scores for math and science. Five East-Asian countries scored higher than international average. These countries include Hong Kong, Japan, South Korea, Singapore and Taiwan. While Malaysia seems to be in the middle of the pack, Thailand scored a little lower than international average, along with Indonesia and the Philippines. 80 Table 3.7: TIMSS Score 1995 1999 Country Math Science Math Science Hong Kong 568.89 509.73 582.06 529.55 Indonesia 403.07 435.47 Japan 581.07 554.47 578.6 549.65 South Korea 580.72 545.78 587.15 548.64 Malaysia 519.26 492.43 Philippines 344.91 345.23 Singapore 608.59 580.35 604.39 567.89 Taiwan 585.12 569.08 Source: TIMSS 3.26 In addition, latest results from the Program for International Student Assessment (PISA) suggest that Thai students perform below the OECD average scores and poorly compared with peers in other Asian countries. The test i s conducted by the Organization for Economic Co-operation and Development (OECD), which measures how 15-year-old students in more than 40 countries perform in math and reading. Altogether, these results suggest that the quality of secondary education output in Thailand may be sub-optimal. Reforming curriculum to adapt to changing market conditions may be most appropriate, see World Bank, 2000. Table 3.8: PEA Score 1 2003 I math I science Reading I I Japan 553 548 598 Korea 552 538 534 Hong Kong 558 539 510 Indonesia 361 395 382 I Macao-China 1 528 I 525 I 498 I Source: PISA, (OECD) 3.27 More worrisome is the fact that fewer students are enrolling in vocational education. In a tight labor market, vocational education plays a crucial role by providing very specific technical skills to workers. However, in Thailand, the worrisome fact i s that the ratio of student at the upper-secondary level who opt for vocational education instead of general education has declined to 34 percent in 2001 from 46 percent in 1994. A recent synthesis study on vocational education also found that most vocational curriculum are not flexible, obsolete and not fitted to the needs of the employer^.^^ In addition, for the first job, the salaries of vocational education’s degree holders tend to be 25 to 30 percent lower than their peers with general education. 69″A Synthesisof Studies inVocational Education from 1993-1998″ by Krismant Whattananarong. 81 In-Service Training 3.28 Given the poor quality of the educational output, managers compensate by training workers. This situation translates into very high training incidence among Thai manufacturing establishments. Also, Thailand has exhibited high training incidence for many years. To complement the data on training from the PICS, we have computed training incidence among matched establishment in the MOIdataset. As seen in Figure 3.10 below, approximately 80 and 70 percent of Thai establishments have consistently offered in-house and outside training to skilled production workers, respectively since 1999. Figure 3.10: Training Dynamics Percentageof FirmsinMOIPanelTraining Skilled ProductionWorkers 30 20 10 1999 2000 2002 InternalTraining0 Extern1Training Educationand Training Policy inThailand 3.29 The Royal Thai Government (RTG) has a strong commitment to education. More than 20 percent of the total budget has been allocated to the education sector over the last decade. The 1997 Constitution ensures the “equal right to receive fundamental education for the duration of not less than 12 years which shall be provided by the State thoroughly, of quality, and without charge, ” paving the way for universal access to 12 years of education “of quality” for all Thai children. In 1999, the National Education Act (NEA) was promulgated to serve as a fundamental boost for education provision and administration. The NEA raises the compulsory education to nine years from six years, enforcing all parents to have their children in school until they graduate from lower secondary level. The NEA i s currently being implemented and i s paving the way for reform in the Thailand education system. The education reform covers five aspects: (i) education system reform; (ii)learning reform; (iii)administrative and structural reform; (iv) personnel reform; and (v) reform of resources and investment of education. More recently, the 15-year National Educational Plan (2002-2016) has been enacted with a focus on all aspects of the quality of life. 82 Box 3.2: The Thai EducationSystem Thailand’s education system has adapted to help meet the new economic challenges faced by the country. The government required students to complete four years of compulsory schooling until 1978, and six years thereafter. Between 1960 and 1978, secondary education consisted of general and vocational streams. Vocational education had in principle two levels, a lower level consisting of two to three years, and an upper vocational stream of two or three years. General secondary education consisted of two levels up until 1978, including a three-year lower secondary degree and a two-year upper secondary degree. After 1978, vocational education was redefined to include an upper vocational stream, offering a parallel track to the upper secondary degree that lasts three years. The system included a higher education vocational degree stream generally considered equivalent to university and lastingbetween two and four years. Since 1978, general secondary education has consisted of two streams, a lower and upper secondary degree lasting three years. Secondary education was gradually transformed into comprehensive secondary education, thereby incorporating more vocational and professional training at the secondary level. At the same time, the government eliminated central school leaving exams, freeing individual schools to create innovative local programs of study. However, local schools did not have the resourcesto fashion new programs, and inany case, parents and students generally look towards secondary school as preparation for college (Orapin, 1991and Sirilaksana, 1988). This division between secondary and vocational education is commonplace. Germany offers the most prominent example of a “dual” secondary education system, providing students with either vocational apprenticeships or college preparatory schooling. Many countries, such as Korea, Singapore or Indonesia, maintain separate school systems for vocational education at the secondary level (Brand, 1998; and Gill, Fluitman and Dar, 2000). Thailand’s higher education system, divided until recently into a number of four-year colleges and approximately 30 teacher-training colleges, i s almost entirely controlled by the Government. In addition to college and teacher-training programs (now called Rajaphat Colleges), the Government has a variety of specialized nursing and military schools considered equivalent to the university degree. Although the country has a few very prestigious private colleges, they enroll a small percentageof university students, and only emerged inthe 1980sas a seriousalternative to public universities. In 1989, 24.4% of university studentswere enrolled in private universities. In comparison, in 1983, only 12.7% were enrolled in private schools (Myers & Chalongphob, 199lb). Thailand’s government began the post-war period with a strong commitment to education and managed to achieve universal primary education in the 1980s. In 1961, only 77.4% of primary school age children were in school. By 1990, 99% of the primary school age population was enrolled in primary school. Lower secondary school enrollments grew from 13.7% of the eligible age group in 1961 to 37% in 1980, and 49% in 1994. In 1999, the secondary school enrollment rate stood at 79%, illustrating the progress that the government made in providing lower secondary education. Enrollment rates in higher education have grown rapidly in Thailand as well. Thai enrollment in higher education stood at 1.3% of the school age population in 1970, rising to 22% in 1994. Source: Hawley (2004) However, there is still a need for more coordination among the nine government agenciesif quality of the education system is to be improved. Currently, nine different ministries are providing training courses for various target groups. However, most of the courses are still supply driven andmost teachers do not have direct experiences inrespected indu~try.~’ ‘O”Vocational Education Reform inThailand” by Chinnapat Bhumirat. 83 3.30 The Vocational Education Act may provide a platform to improve both the curriculum and system. The draft Vocational Education Act contains measures that could help in addressing the issues of quality of output. The act will support networkmg among education institutions, industry and communities in order to ensure that the curriculum responds to market signals and to provide effective incentives for private participation. The act will also support quality assurance. For example, vocational teachers will be required to have a license. A vocational education qualification framework will be established. In addition, financial support will be provided to graduates from compulsory education to obtain vocational training of at least one year before entering the labor market. 3.31 Also, efforts have been made to improvingpost graduate skill training. In addition to the Office of Vocational Education Commission, the Department of Skill Developmentunder the Ministry of Labor i s the key agency in providing skill development and training, mainly to workers in the labor market. Courses have been designed to meet three target groups: those who just enter labor market; those already in the labor market; and those that would like to change jobs. 3.32 A Skill Development Fund (SDF) has been established to promote private sector involvement intraining. Based on the Skill Promotion andDevelopment Act 2002, the SDF has been established to support skill development. The fund requires all establishments with more than 100 workers to provide training to at least 50 percent of their workers, otherwise, such establishment will have to contribute one percent of the monthly base wage for each untrained workers to the fund. The owner of the establishment, by abiding to the fund’s requirement, will be waived from income taxes that arise from training provided in their work place. In addition, employers can also seek loans from the Labor Skills Development Fund to subsidize training or skill testing programs. A Skill Development Promotion Committee has also been established. The committee will consist of representatives from the Ministry of Labor, Ministry of Finance, Ministry of Industry, Ministry of Education (MOE), Ministry of Information Communication and Technology (MICT), Ministry of Natural Resource and Environment, Bureau of Budget, Board of Investment, Tourism Authority of Thailand and the Board of Trade. Combinedwith the input from relevant stakeholders, this committee will oversee the SDF, ensure the quality of training, setting the standards for skill development and promoting skill competition. E. CONCLUSIONSANDPOLICYIMPLICATIONS 3.33 This chapter has investigated the level, distribution andquality of skills and competencies of workers in Thailand, as well as the match between these qualifications and those required by businesses. Following Yusuf (2003), this chapter has reassessed Thailand’s education and skill needs inlight of future aspirations of the country. 3.34 The chapter finds compelling evidence of both skills shortages and mismatch in Thai manufacturing sector. Firms pay large wage premiums to tertiary education graduates, and to workers who receive training in technical skills, indicating extreme levels of excess demand for the highest S L in Thailand. In addition, the gap between returns to secondary and tertiary education has not narrowed over time. While the return to a high-school certificate (Por Wor Chor) i s only 5.3 percent, the marginal effect of receiving a university-level degree i s nearly 35 percent. Within the context of a simple human capital investment model, a person can choose 84 either to study or get ajob — these results suggest no monetary incentive to complete high school if one believes that college is unaffordable or impossible. This may explain the gap insecondary education performance in Thailand. Furthermore, the incidence and intensity of hard-to-fill vacancies are very high. Nearly 80 percent and 95 percent of Thai manufacturingplants surveyed have had vacancies for professionals and production workers respectively in the last two years. Incidence of professional vacancies i s far greater than in Malaysia and Indonesia (about 50 percent) and the Philippines (about 25 percent). Also, it takes longer (more than six weeks) in Thailand than in any other benchmark country to fill a vacancy of a skilled production worker or a professional. When asked about the reasons for these vacancies, more than 80 percent of the managers cited the fact that applicants lack appropriate basic and technical skills. 3.35 Policies to address the shortage of skilled workers in Thailand should target a further expansion of secondary education in order to increase the local supply of skilled workers (professionals and sub-professionals). This will be critical to fulfill the country’s agenda on competitiveness. Expansion of secondary education coverage i s a condition for countries to fully reap technological spillovers from foreign direct investment (FDI) and trade. Specific measures could include enforcing the NEA decision of nine years compulsory education or raising the compulsory number of years of education, especially in rural areas and in poorer groups of the population. An increase in the secondary education attainment rates in rural areas will have the advantage of equipping the workforce with appropriate skills where they live, which could then lead to more firms settling outside Bangkok, and a slower migration of unskilled workers towards Bangkok. As the shift towards high-tech industries continues, unemployment rates among unskilled workers will likely rise, constituting a serious challenge for cities. Along the same lines, the RTGcould accelerate the creation of more public secondary institutions in rural areas and encourage more participation of the private sector to secondary education. 3.36 The skills mismatch is evidenced both from the employee’s and the employer’s perspectives. First, inrelation to what employers consider as the desirable skills, Thai managers consistently rank their workers as having poorer skills. A large percentage of managers from the PICS rate English and IT skills of their Thai employees as poor — greater share than seen in Malaysia. Nearly 60 percent of managers rate the English skills of their local professional workforce as poor. More than 40 percent rate IT skills as poor. Ratings are worse for Thai slulled production workers, with more than 90 and 80 percent of managers rating their English and IT skills as poor, respectively. Only about 12 and 20 percent of managers in the Malaysia PICS sample rate the English and IT skills of their local professional workforce as poor, respectively. Second, we provide an employee’s perspective, a self-assessment of his or her skills’ adequacy. Workers were asked to list the three skills that they lack the most in doing their job. Findings are fully consistent with the managers’ assessment. More than 72 percent of the 14,000 Thai employees interviewed identify English language proficiency as the most severe constraint in doing their job. In Malaysia, only 48 percent of workers cited English as the most severe constraint. Englishlanguage deficiency i s present inThailand across all occupations and regions. Among Thai workers, IT skills are a distant second with approximately four percent of the workers identifying it as the skill they lack the most. 3.37 Quality of educational output is the key determinant of skills mismatch. However, convergent sources point to the fact that the quality of secondary education output in Thailand 85 may be inadequate by international standards. Thai Secondary Education Students score lower than average in international tests. The latest international test for quality of education, the TIMSS, revealed poor results for Thailand. Also, the latest result from PISA suggests that Thai students perform below the OECD average scores and poorly compared with peers in other Asian countries. More worrisome i s the fact that fewer students are enrolling in vocational education. In a tight labor market such as the one in Thailand, vocational education may play a crucial role by providing very specific technical skills to workers. However, in Thailand, the ratio of student at the upper secondary level who opt for vocational education instead of general education has declined to 34 percent in 2001 from 46 percent in 1994. As a consequence, college degrees seem to be the only educational “signals” that are rewarded on the Thai labor market. Therefore, there i s no incentive to complete high school if staying the course until graduatingfrom university is not an option. 3.38 Improving the quality of secondary education is likely to increase the demand for secondary graduates and their wage premium, providing a strong incentive to complete secondary school. Specific measures could include improving the content of education for the skills that are most needed (English language proficiency and ICT skills). The RTG could consider introducing/reinforcing English in primary school and aim at teaching one or more courses in English in secondary school, both general and vocational in five years. In the long run, English language could become compulsory in all skills development centers and for Diploma level. This would imply training more English teachers and improving the training of existing ones. 3.39 Employer-provided training is the other channel through which workers accumulate firm specific skills. Given the poor quality of the educational output, managers in Thailand have been compensating by training workers themselves. The establishment of a Skill Development Fund (SDF) with mandatory contributions i s as such a good decision. However, a stylized fact in the training literature i s that SMEs, who train the least in all countries, end up subsidizing the training by large firms, as they will be forced to contribute to a fund mostly used by large firms. Investigatingthe reason why SMEs train fewer workers than large firms and how to effectively raise their training incidence could be an interesting follow-up investigation. Given the low external training incidence, the Thailand Research Group could encourage external training, especially using more suppliers to train workers. That would increase the level of technology linkages between the Thai domestic firms and multinational suppliers, one of the weakest links in the Thai technological capability chain as shown in the next chapter of this report. 86 References Acemoglu, D.2003. “Patterns of Skill Premia.” Review of Economic Studies. Forthcoming. Becker, G. S. 1964. Human Capital. NBER,New York. Freeman, R. and Shettkat, R. 2001, “Skills Compression,Wage Differentials, andEmployment:Germany vs. the US.” Oxford Economic Papers. 3: 582-603. Gill, I.S., Fluitman,F.andDar, A., Editors. 2000. Vocationaleducationandtrainingreform: Matching skills to markets andbudgets, The World Bank, Washington,DC. Haskel, J. and Martin, C. 1993. “The Causes of Skills Shortages in Britain”. Oxford Economic Papers. 451573-588. Hawley, J. D. 2004. “Changing returns to educationin times of prosperity and crisis, Thailand 1985- 1998” Economics-of-Education-Review.June 2004; 23(3): 273-86. Middleton, J., Ziderman, A. and Adams, A. 1993. “Skills for Productivity: Vocational Educationand Training inDevelopingCountries.” World Bank. Mincer, J. 1974.Schooling,experience, andearnings, NationalBureauof EconomicResearchand ColumbiaUniversityPress, New York. Myers, C.N. and Chalongphob, S. 1991. Educationaland economic development:issues and options for policy and reform. In: 1991 Year End Conference on “Educational Options for the Future of Thailand”, The ThailandDevelopmentResearchInstitute, Bangkok, Thailand. Orapin, S. 1991.Three moreyears inschool: parents’ opinionsandproblems.In: 1991Year End Conferenceon “EducationalOptionsfor the Futureof Thailand”, The ThailandDevelopment ResearchInstitute. Psacharopoulos, G. 1994.Returnsto investmentineducation: a globalupdate. World Development22 9, pp. 1325-1343. Schady, N.T. 2003. “Convexity and SheepskinEffects inthe HumanCapitalEarningsFunction:Recent Evidencefor Filipino Men.” Oxford Bulletin of Economic and Statistics. 65(2): 171-196. Sirilaksana, C. K. 1988.The educationsector inThailand: problems, policy dilemmas, and the roleof the government (DiscussionPaper Series No. 95). Bangkok:Facultyof Economics,Thammasat University. The WorldBank. 1995.Prioritiesand strategies for education: a World Bankreview, The World Bank, Washington,DC. The World Bank.2000. Thailand: Secondaryeducation for employment, The World Bank,East Asia and Pacific Region. The WorldBank. 2004. A BetterInvestmentClimatefor Everyone.World DevelopmentReport2005. The World Bank.2004. EastAsia Update, East Asia andPacificRegion, November9, 2004. 87 TheWorld Bank.2004. ThailandEconomicmonitor,BangkokOffice, October Yusuf, S. 2003. InnovativeEastAsia: The future of Growth,The World BankandOxford University Press. Zeufack, A. G. 1999. “Employer ProvidedTrainingunderOligopolisticlabormarkets” DECRG. The World Bank. 88 4. STRENGTHENINGTECHNOLOGICALCAPABILITIES 4.1 The role and importance of technological progress in industrial development i s now well established and well illustrated by countries such as Korea and Singapore in East Asia. These countries have rapidly expanded their national and FTC required for process and product innovation and are currently able to produce and export complex manufactured goods. In contrast, other countries in the region such as Indonesia, the Philippines, Malaysia and Thailand are still lagging far behind these regional leaders in terms of technological capabilities. The daunting challenge facing these countries today i s to rapidly follow the steps of their successful neighbors in order to gear their economy towards a higher growth path and transition to a more sophisticated level of industrialdevelopment. 4.2 Strengthening bothnational as well as FTC, is essential for industrialdevelopment if Thailand is to move from labor-intensive to more technology-intensive industries. National technological capability (NTC) can be defined as the aggregation of FTCs, taking into account linkages between firms and other economic agents, combinedwith all the factors and institutions that support the “technology” enabling en~ironment.~~ The nature of these factors naturally differs across countries and can play a significant role in explaining differences in NTC. At a more microeconomic level, FTC encompasses the skills, knowledge and experience that enterprises must accumulate in order to operate technologies efficiently, and more importantly, to enable innovation in both their products and processes.72 It i s at the firm-level where empirical analysis can help identify sources of low technological capabilities, and derive appropriate policy responses. 4.3 Little, however, is known about firm-level technological performance in Thailand. Few studies on industrial technology in Thailand have either fully analyzed the technological factors that affect productivity and competitiveness at the firm-level or tested econometrically the influences affecting capability acquisition of these firms. Indeed, a systematic analysis of a firm’s technological performance has been difficult, if not impossible, due to the lack of appropriate data, as evidenced by the dearth of research on the subject. Fortunately, the PICS provides an information-rich dataset that i s a prerequisite for such a study and enables the investigation of key factors that explain the heterogeneity of technological capabilities and performance across manufacturing establishments inThailand. 4.4 For this reason, the chapter develops the analytical tool of a TCZ inorder to quantify and summarize differences in FTC and to assess technological development in Thailand. FollowingWignaraja (1998,2002), TCI draws on the taxonomy developed by La11(1992), which 71 La11 (2000) defines NTC as “the complex interaction of skills, experience, and effort that enables a country’s enterprises to efficiently buy, use, adapt, improve and create technologies. It comprises the non-market system of interfirmnetworking and linkages, way of doing business, andthe web of supporting institutions”. 72La11(2002). 89 identifies and categorizes FTC into investment, production and linkages activities. TCI permits useful comparison of the technological capabilities across firms and enables econometric analysis of the influences on the acquisition of FTC. TCI captures and integrates a variety of objective and subjective information into coherent measures of a firm’s capacity to establish, operate and transfer technology. TCI provides a composite measure of technological capabilities composed of information about firm-level technological behavior that i s provided by the rich data collected inthe PICS. 4.5 Empirical evidence presented in this chapter suggests that poor technological performance in Thailand is related to poor production technological capabilities and weak linkages, which both lead to less innovation. TCI reveals that all but three industries (electronics & electrical appliances; machinery & equipment; and auto parts) are below average, which casts doubt on the use of capital intensity as a measure of technological depth and points to the necessity of a more comprehensive approach to technological development. Moreover, linkages are the poorest component of TCI. Further results show that SME, non-exporters and domestically-owned establishments are lagging far behind large establishments, exporters and establishments with FDI, respectively, in the acquisition of technological capabilities as evidenced by the TCI. Regression analysis indicates that the specific components of the TCI that are most correlated with higher TFT are training, new machinery & equipment, computerization and website use ininteracting with suppliers and clients. 4.6 For these reasons, Thai policymakers must focus on fostering the environment that promotes and enables the development of FTC, namely through investment in human capital, linkages and competition. The process of building FTC ‘has four features that are particularly relevant and useful inguiding (a) FTC building is an incremental and cumulative process. Enterprises cannot instantaneously develop the capabilities needed to handle new technologies; nor can they makejumps into completely new areas of competence. Instead, they proceed in an incremental manner, buildingon past investments in technological capabilities and moving from simple to more complex activities. Therefore, education and training play crucial roles inthe development of FTC. (b) Firms rarely acquire capabilities in isolation. FTC building must involve linkages and cooperation between economic agents, both private and public, local and foreign. New technologies, such as e-mail and website use in interactions with suppliers and buyers, can facilitate and lower costs of such linkages and quickly lead to further integration of economic agents. Interaction and interdependence between agents lead to collective learning, and thereby, linkages are a fundamental feature of technological capability building that policymakers shouldfacilitate. (c) The development of FTC is the outcome of investments undertaken by firms inresponseto internaland external stimuli. Specifically, competitionis conducive to higher investments inFTC, with competitors from abroad possibly the most potent inducement to skill and technology upgrading. Providing a metric such as industry- 73See Wignaraja(2002). 90 specific TCIs as a benchmark for firms to measure its technical skills, knowledge and experience against those of its competitors may encourage more investment in technological capabilities. Success in building FTC determines industrial success. Differences in the efficiency with which FTC are created may be a leading source of differences inTFP between countries. 4.7 The chapter is organized as follows. The next part (Section A) will describe the operating environment of firms in Thailand by presenting available national indicators, which provides evidence that Thailand i s technologically challenged. Evidence of technology use by firms based on data from the PICS will also be presented to corroborate the national indicators. Section B will then introduce and discuss the Technological Capability Index (TCI) that i s developed from PICS data. It will also explore the relationship between FTC and TFP through regression analysis. Section C will conclude with a discussion of policy recommendations. A. W H A T I S THAILAND’S CURRENT TECHNOLOGICAL PERFORMANCE? 4.8 Thailand’s current technological performance is not impressive. Available data indicate that the country does not have the number of researchers, the level of R&D expenditures, nor the number of patents filed inthe United States that are consistent with its level of development. Focusing on the fourth indicator that i s commonly used in the literature — high tech exports, Thailand’s performance seems quite good. However, this indicator i s misleading since these exports do not reflect the country’s actual technological capability, but as discussed below, mainly reflect its ability to attract FDI into capital-intensive activities involved in the assembly of high-tech products. The following sections present selected indicators from the PICS as well as the technological indicators mentioned above as two distinct sets: input and output indicators. These metrics provide a good snapshot of technological performance by Thai manufacturing establishments and will also help benchmark and better gauge performance in the future after the completion of an ICA update. (a) Technological Input Indicators 4.9 Technological input indicators are those that reflect the country’s ability to use, develop and absorb technology. They are part of the broader technological capabilities defined earlier and, generally include, among others, indicators such as the number of graduates from university and training institutes, Research and Development (R&D) expenditures and research staff. The PICS provides data on the two latter indicators, which will be discussed below. Chapter 3 discusses the country’s number of graduates and other education-related issues. R&D Expenditures 4.10 At the national level, R&D expenditures in Thailand are far below not only the levels of regional competitors, but also the expected level for its income Table 4.1 shows that the 74 Ideally, cross-country comparisons of technological indicators should take into account that countries are at different stages on the technological path, and therefore, indicators should be normalized in some manner before comparisons are made. Unfortunately, lack of dataprohibits such weighting from being undertaken. 91 level of R&D expenditures as a percentage of GDP in Thailand i s far below Singapore and, to a lesser extent, China and Malaysia. Further, a simple regression with available data from 2002 for relevant countries i s used to derive the predicted value of R&D expenditures as a percentage of GDP.75The results are shown inFigure 4.1, which includes data for Thailand from 1996 and 2002 and indicate that the level of R&D expenditures in Thailand has indeed been consistently below expected for its level of income.76 These results warrant in-depth sector and industry- specific analyses, as well as an assessmentof past government strategies to stimulate R&D. Table 4.1: Selected Technological InputIndicators Researchers in Expenditures R&D per for R&D millionpeople (% GDP) 1996-2002 1996-2002 China 633 1.2 Malaysia 294 0.7 Singapore 4,352 2.2 Thailand 289 0.2 Note: Data are for the latest year available. Source: World Bank (2005). 4.11 Inthe PICS sample, only a smallpercentage of Thai manufacturing establishments are engaged in R&D activities. Overall, 16.1 percent of manufacturing establishments report R&D activity in 2002. The difficulty in interpreting this percentage i s due not only to the absence of a temporal benchmark, but also to the lack of knowledge about the nature, quality, and sophistication of such activities. Nevertheless, the sample i s rich enough to draw specific results for several categories of firms based on ownership, industry and export status. The percentage of overall manufacturing establishments reporting R&D activities and across ownership, industry, and export status groups are presentedinFigure 4.2. 4.12 Based on the findings from the Chapter 1 that R&D is positively correlated with TFP and assuming that TFP growth is equivalent to greater technological performance, foreign-owned establishments and exporters seem to contribute more to lifting the country up the technological ladder.77 Indeed, foreign-owned establishments and exporters are more 75 Data on R&D by productive enterprises (a better indicator of technological performance than total R&D, which includes non-industrial R&D) are not readily available. 76For a more rigorous analysis and interpretation, Figure 4.1 should normally only take into account countries with comparable economic and industrial structures since the relevance and level of R&D depends on these structures. This i s not possible due to lack of data. The figure should therefore be interpreted with caution. However, notwithstanding these concerns, it gives an approximate sense of the country’s standing. This remark also applies to Figure 4.3 which shows the number of researchers. 77 Inthe future, a comprehensive panel database would be useful to better assess the contribution of R&D to TFP growth in Thailand in order to better guide policy recommendations. Several studies have investigated the contribution of TFP in other countries. Comin (2004), for example, argues that R&D is actually a small part of TFP growth. 92 likely to report R&D activities. 78Approximately 23 percent of foreign-owned establishments, compared with 14 percent of domestically-owned ones, reported R&D expenditure in 2002. Food processing and electronics & electrical appliances are the industries with the highest percentages of firms reporting R&D. When considering only the sample of foreign-owned establishments, textile and food processing are the two industries with the highest percentages reporting R&D activities, at 36 percent and 35.3 percent, respectively. Food processing and clothing have the highest percentages of domestically-owned establishments reporting R&D activities, with 28 percent and 18 percent, respectively. In addition, 23 percent of exporters versus nine percent of non-exporters report R&D activities. For both exporters and non- exporters, food processing and electronics and electrical appliances are the industries with the highest percentage of establishments reportingR&D activities. re 4.1: R&DExpenditures and Level of Devel “1 Israel States Source: World Development Indicators 2005. Author’s calculations. Researchers 4.13 At the national level, Thailandalso falls behind regional competitors interms of the number of researchers — a technological input indicator that reflects the capability of a country to build a knowledge base, and hence, approach or challenge its technological frontier. The aggregate data on the number of researchers per million people presentedinTable 4.1 show that Thailand considerably lags behind China and Singapore. As illustrated in Figure 4.3, the number of researchersinThailand i s below not only the level of regional competitors but 78 In the future, a comprehensive panel database would be useful to better assess the contribution of R&D to TFP growth in Thailand in order to better guide policy recommendations. Several studies have investigated the contribution of TFP in other countries. Comin (2004), for example, argues that R&D is actually a small part of TFP growth. 93 also the expected value given its per capital GDP. This deficit in the number of researchers has been persistent as the number of researchershas increased only slightly since 1996. Figure 4.2: Percentageof ManufacturingEstablishmentsReportingR&DActivities in2002 I Food Processing 40 1 rn Textile 35 clothing 30 25 0 Autorotlveparts 20 Electronics/ Oectrical Appliarces 15 Rubberand plastics 10 rn Wood ProdKtsand Furniture 5 DMachllEfyand 0 Equipment % of firms with % of the foreign Yo of the non Yo of exporting Yoof non exporting Total R&D firms with R&D foreign firms with firms with R&D firms with R&D R&D I 1 Source: Thailand PICS 2004. 4.14 PICS data indicates that manufacturing establishments in Thailand use researchers, but the extent of their use in each industry varies with the ownership status. The results from the PICS presented inFigure 4.4 indicates that 21 percent of establishments employ staff exclusively for R&D (compared with 17 percent in Malaysia), with no considerable difference between foreign- and domestically-owned establishments (23 percent and 20 percent, re~pectively).~~ However, domestically-owned establishments have the highest percentage of staff employed exclusively for R&D (about 30 percent) in the industries that are more “technology-based”: auto parts; electronics & electrical appliances; and machinery & equipmentag’ Incontrast, for foreign establishments, only one “technology-based” industry (machinery & equipment) i s among the top three industriesemploying the highest percentage of staff exclusively for R&D. 79The questions asked inthe two surveys are not quite the same. The Thailand survey asks if the firm has employed staff exclusively for desigddoing innovation/R&D in2002-2003. The Malaysiasurvey only asks for 2001. “Technology-based” industries are identified using the OECD classification of high tech versus low-tech industries. 94 Figure 4.3: Number of Researchers and Level of Development I Finland JapanDenmark M*@dfth Africa Source: World Development Indicators 2005. Author’s calculations Figure 4.4: FirmsEmploying Staff Exclusively for R&Din2002-2003 (%) IFoodProcessing 45, ITextile 40 0 Clothing 35 0Automtive Parts 30 25 IBectronics/Eectr cal Appliances 20 Rubber and 15 Plastics 10 IWoodProducts and Furniture 5 0 Machinery and 0 quiprent Overall Foreign firms Non-foreignfirm ITotal Source: Thailand PICS 2004. (b) Technological Output Indicators 4.15 Technological outputs are the result of a successful combination of FTC and the institutional framework supporting these capabilities. They can therefore be used to assess the efficiency of the technological inputs and, more generally, to give a sense of technological performance. At the national level, the most commonly used technological output indicators are: high-tech production; technology-based exports; scientific publications; and patents. At the firm level, the PICS data only provides information on patents, specifically the number of utility models, patents and copyright protected materials filed during the last two years. National-level 95 and firm-level indicators taken together can provide a sense of technological output performance inThailand. Patents 4.16 Inthe PICS, 10percentof Thai establishmentssurveyeddeclaredthat they didfile a patent during the last two years. Approximately the same percentage of establishments in the Malaysia PICS declare filing patents. Again, there are differences across ownership and export status groups. Among foreign-owned establishments, 13 percent declared filing patents, compared with 10percent of domestically-owned establishments. The percentage i s also higher for exporters (12 percent) than for non-exporters (nine percent). The industries inThailand with the highest percentages of establishments filing patents are those that are expected to be more high-tech based: auto parts (16.6 percent); electronics & electrical appliances (13.9 percent); and machinery & equipment (13.6 percent). The leading industries are slightly different inMalaysia, where electronics (18.4 percent), wood & wood furniture (12.4 percent) and auto parts (12.1 percent) are the industries with the highest percentages. (See FiguresA4.1 and A4.2 inAnnex.) 4.17 Innovation performance in Thailand can more accurately be assessed using the number of patents filed in the United States. Cross-country comparison using patent indicators from the respective PICS data remains difficult since national requirements for filing patents are not necessarily the same and the commercial value of these patents i s also unknown. A comparison of the number of patents filed inthe United States can provide clearer assessment of the relative performance of countries since the requirements are the same for all applicants everywhere. Figure 4.5 shows that Thailand (along with Malaysia) has trailed far behind China and Singapore; the number of patents has been persistently very low over the 2000 to 2004 period. This poor performance partially explains why the number of United States’ patents issued to Thailand i s well below expected levels given its per capita GDP, as depicted in Figure 4.6. Figure 4.5: Patents Issued by the US to Selected EastAsian Countries -China (Hong Kong) +China (Mainland) Malaysia +- Sinpapore +Thailand 2000 2001 2002 2003 2004 I Note: Data include utility, design, plant and reissue patents. Source:United States Patent and Trademark Office (USPTO), Annual Report 2004. 96 Figure4.6: Patents Filedinthe UnitedStates Technology-basedExports (High-Tech Exports) 4.18 High-tech exports are one of the macro indicators regularly used for technological performance analysis. More specifically, the share of high-tech exports in total manufactured products exports can give an indication of the degree of sophistication and technological capabilities of the manufacturing sector. This share is expected to increase with industrialization. 4.19 Initialinspectionsuggests that Thailand is well above its high-techexports potential, as seen inFigure 4.7.’l Although this can reflect good technological performance, this indicator can be misleading for different reasons. First, the same high-tech products exported can be the outcome of different processes for different countries; a process involving sophisticated design and fabrication as opposed to a simple, relatively low-skilled based assembly process which i s clearly less productivity-enhancing. Second, the nature or compositionof high-tech exports may vary from country to country. In order to use this indicator effectively, one needs to conduct industry-specific analyses looking at the relative importance of three categories of firms: (i) firms assembling simple products; (ii) designing products while learning how to innovate; firms and (iii) firms designing products and conducting R&D for product innovation.82 Unfortunately, ‘*Electronics & Electrical Appliances, Auto Parts and Machinery & Equipment are the industries that have contributed the most to this performance. These sectors account for approximately 45 percent of total manufactured exports and explain more than 50 percent of manufactured exports growth during the last decade (Bank of Thailand, 2005). 82 See Hobday (1995) for a discussion of this classification and Hobday (2000) for a review of case studies of countries that have successfully managedtheir transitionfrom low-techto high-tech status. 97 such data are not currently available.83 Efforts by the government to set up a mechanism to systematically collect, update and analyze these type of indicators inthe future are needed. Figure 4.7: High-Tech Exports is a MisleadingIndicator of TechnologicalPerformances4 f 75 – 0 Philippines 6 0 – Malaysia ‘4 1 45 – E p Thailand 3 0 – 0 #B 3 OChina Indonesia 1 5 – 0 0 Mexico Brazil 0.00 20.00 40.00 60.00 80.00 100.00 Manufactures exports (% ofmerchandiseexports) Source: World Bank (2005). 4.20 Looking closer, Thailand appears to be more an “assembler” than a “designer”. The PICS survey reveals that Thailand relies heavily on unskilled production workers, who are more likely to be assemblers. Seventy-five percent of the workers in the manufacturing sector are unskilled production workers, whereas 11percent are skilled production workers (compared with 49 and 31percent, respectively, in Malaysia).85Both capital-intensive and high-tech exports can therefore be misleading technological performance indicators and should be interpreted with caution. Thailand at this stage i s probably benefiting from its relatively cheap labor, as are China and the Philippines, which encourage the relocation of labor-intensive activities from other countries moving up the technological ladder as well as inward FDIs. To illustrate this point, Japanese automobile companies are relocating to Thailand. Honda and Toyota, for example, have assembly factories in the country. In the electronics industry, Advanced Micro Devices ( A m ) and Sony Semiconductor have established factories that are involved in assembling computer products and other electronics devices. In addition, Canon Engineering i s an example of a Japanese machinery & equipment company that has located inThailand. 83Also, information on import content of exports would help in better characterizing high-tech exports, but that would require detailed data on all the imported parts used by different industries, which are not available. For example, it is difficult to find data on all the parts imported to make a specific type of automobile. Also, it is difficult to know whether the final products are exported or sold locally. 84Data are from the World Development Indicators database (World Bank, 2005), which defines high-tech exports as “products with high R&D intensity, such as aerospace, computers, pharmaceuticals, scientific instruments, and electrical machinery”. The database has its limitations since it also does note take into account assembly activities. 85Thailand and MalaysiaPICS data. 98 4.21 While Thailand remains very attractive to FDI, the challenge is to attract FDI that relies not only on importing ready-made technology, but also promotes local R&D and strengthens local learning and technological capabilities of domestic firms. The attractiveness of Thailand i s illustrated by the 2003 A.T. Kearny FDI Confidence Index, which ranks Thailand as the 16th most preferred destination for FDIin the world, above South Korea and Malaysia. The contribution (and origins) of FDIto high-tech export performance, as well as its linkages to local firms and its role its enhancing national and FTC deserves further investigation. 4.22 International competition is both a threat to export markets and a stimulus for technological upgrading. As Thailand becomes increasingly challenged by the entry of new competitors in the global marketplace, the country needs to strengthen its NTC, learn by doing, buildits own brand images and upgrade to more sophisticated products as Korea, Singapore and Taiwan have successfully accomplished during their export-led industrialization processes. The increase in high-tech exports i s a good sign, but this i s clearly insufficient to spark productivity- enhancing technological progress in the country. Forward-looking strategies that focus on skills development, importing of commercial firm-level R&D and access to foreign technology should be implemented. The second part of this chapter goes beyond the experience of successful regional competitors and focuses on identifying the potential key drivers of technological upgradingand learning usingthe PICS data. (c) WEF Technology Indicators 4.23 This subsection will briefly present the technology indicators derived from the Global Competitiveness Index (GCI), which was recently developed by the WEF and i s the most rigorous and comprehensive index currently available.86 It i s grounded on a solid theoretical framework and builds on surveys including both qualitative and quantitative data. The index i s composed of sub-indices related to 12 pillars of growth including, among other factors, institutions, macroeconomic stability, human capital, technological readiness and innovation. Given the focus of the chapter on technology, this section will only look at the relevant two pillars: (i)the technological readiness pillar, which takes into account the level of technology available to firms rather than the ability of a country to innovate; and (ii) the innovation pillar, which looks at a country’s ability to innovate and, hence, i s relatively more important for countries that are close to the knowledgefrontier.87 86 Inthe first GCR (GCR 2001-2002), a GCI was presented. The index was derived from hard data and an Executive Opinion Survey. Its construction was based on three pillars (institutions, macroeconomic stability and technology readiness). The latest GCR presents a new and more comprehensive index, the Global CompetitivenessIndex, which is based on 12 pillars, including the three pillars of the initial GCI. (See GCR 2005 for a detail description of the indexes.) The new GCI is aimed at unifying the Business Competitiveness Index and the GCI previously developed by the WEF. 87 The technological readinesspillar includes: technological readiness; firm-level technology absorption; quality of competition in the ISP sector; laws relating to ICT; cellular telephones; internet users; and local availability of specialized researchand training services. The innovation pillar includes: quality of scientific research institutions; company spending on R&D; universityhndustry research collaboration; government procurement of advanced technology products; availability of scientists and engineers; utility patents; IPR protection; and capacity for innovation (GDP – exports+ imports). 99 4.24 Prior to presenting the indices corresponding to these pillars, it is important to note that GCI takes into account the differences across countries in their stage of development. Indeed, countries are not necessarily at the same stage of development, and therefore, the relative importance of different pillars in explaining competitiveness and economic performance may vary. Based on this, the weighting of the pillars in the index changes with a country’s level of development.” For instance, the innovation pillar i s given a lower weight for countries below a certain income threshold. This i s a practical approach that highlights the most important pillars for each country and helps facilitate country-specific policy recommendations. The index i s at its early stage of development, but can reasonably be used for cross-country comparison. Its technological readiness and innovation sub-indices can complement the input and output indicators and present an accurate depiction of the relative technology performance of Thailand. 4.25 From a sample of 160 countries, Thailand ranks 46 and 42, respectively, in terms of technological readiness and innovation. The country i s doing relatively well when compared with countries classified at the same level of development (Le., countries in italics below Thailand in Table 4.2).89 Indeed, only Brazil i s ranked as more technologically ready than Thailand. Compared with other East-Asian countries, Thailand generally fares well, but still lags behind Malaysia and Singapore. On innovation, Thailand ranks higher than in technology readiness, but actually falls behind several countries in the same comparison group. 4.26 Overall, technological performance inThailand is not impressive. Improvement must be made for Thailand to catch up with other leading regional competitors. The remainder of this chapter adopts a microeconomic perspective and investigates the relationship between technological capabilities and firm performance. Empirical findings will help identify the most pressingissues intechnology development facing Thai firms and guide policy recommendations. Technological non-core economies are those where innovation is not crucial for growth. They are relatively (compared with technological core economies) distant from the technological frontier, and can benefit more from technology adoptiodtechnology transfer, copy and imitation. The index for the core technology economies consist of two sub-indices: innovation; and ICT. 88Chapter 1.3 of the 2005 GCR presents a discussion on the maximum likelihood-based weighing methodology usedto compute the GCI. 89Countries at the same level of development are those with real per capita GDP between US$2,000 and US$3,000 and do not export more than 70 percent inprimary goods (GCR, 2005). 100 Table 4.2: Ranking of TechnologicalReadinessand InnovationSub-Indexes Technological readiness index index China 54 26 Indonesia 64 33 Philippines 63 76 Malaysia 28 25 Singapore 7 11 Thailand 46 42 Brazil 42 32 Bulgaria 62 84 Peru 75 94 Romania 53 53 Tunisia 47 34 Source:GCR (2004) and (2005). B. TCI: TOWARDS A BROADERDEFINITIONOF TECHNOLOGY 4.27 Industrialtechnologicaldevelopment is not a process that can be promotedinstantly or easily by investing in new equipment or imported technology, but rather requires conscious and continuous investments by firms in their own technological capability. That is, simply purchasing new machinery or entering into a partnership with an M N C i s insufficient for establishments in Thailand to catch up with global leaders. A growing literature stresses the difficult firm-specific processes involved in building technological capabilities and argues that enterprises have to undertake conscious investments to put technology to productive use.9o Transfer necessarily requires learning because many aspects of technologies are tacit; technological knowledge i s difficult to embody in hardware or written instructions. The process of getting a new technology into production requires the development of new skills and information. Mastery of new technologies, ultimately, can only be acquired through concerted effort, skill upgradingand investments intraining, R&D activities and extensive experience. (a) Constructing the Firm-Level TCI 4.28 For the analysis, a functional index — TCZ — is developed in order to quantify and summarize differences in FTC. I t serves as useful tool for the assessment of technological development in Thailand. TCI permits useful comparison of the technological capabilities across firms and enables econometrig analysis of the influences on the acquisition of FTC. TCI captures and integrates a variety of objective and subjective information into coherent measures of a firm’s capacity to establish, operate and transfer technology. TCI provides a composite measure of technological capabilities composed of information about firm-level technological behavior that i s providedby the rich data collected in the PICS. 90 See Pack and Westphal (1986), Katz (1987), La11 (1992),Bell and Pavitt (1993),Pietrobelli (1997), Wignaraja (1998),Romijn (1999),Wignaraja (2002),and Metcalfe (2003). 101 4.29 Following Wignaraja (1998, 2002), TCI draws on the taxonomy developed by La11 (1992), which identifies and categorizes FTC into Investment, Production and Linkages activities. Table 4.3 recreatesthe illustrativematrix of the major technical functions involvedin the development of FTC from La11(1992). `The columns categorize the major FTC by function, and the rows identify the degree of complexity or difficulty as measuredby the nature of activity from which the technological capability arises.91 The advantage of this framework i s that it provides a clear continuum of technical functions that are necessary for the development of FTC from the time new technology enters a given firm to when it exits to other firms andinstitutions. 4.30 As seen in Table 4.3, La11 (1992) identifies three broad categories of FTC — Investment, Production and Linkages: (0 Investmenttechnologicalcapabilitiesare skills and information required before the investment i s undertaken and those needed for carrying out the investment itself. These activities include project preparation, technology identification and transfer and workforce training. The understanding gained by the operating firm of the basic technologies involved affect the efficiency with which it later operates its facilities. A crucial factor in the development of technological capabilities i s the stock of skilled employees and additions to this stock by training, both in-house and externally. (ii) Productiontechnologicalcapabilitiesrangefrom basic skills such as adoption, operation and maintenance to more advanced knowledge such as adaptation, improvement and equipment “stretching” to ultimately the highest and most demanding technical proficiencies of research, design and innovation. This category covers both process and product technologies as well as the monitoring and control functions included under industrial engineering. Process technological capability includes quality control, maintenance, plant layout, inventory control and improvements in equipment and processes. Product technological capability includes mastering product design and specifications, improving existing products, developing new products and licensing product technology. The skills classified under this production category determine not only how efficiently and effectively given technologies are operated and improved, but also how successfully in-house efforts are utilized to absorb technologies bought or imitatedfrom other firms. (iii) Linkages to technological capabilities are the skills needed to exchange information, knowledge and technology with component or raw material suppliers, buyers of output, subcontractors, consultants, service firms and technology institutions. These linkages affect not only the productive efficiency of the firm, but also the diffusion of technology throughout the economy and the deepening of the industrial structure, both of which are essential to industrial 91Since the simplicity or complexity of a particular function is difficult to judge a priori, this categorization i s simply indicative rather than exact. The technological capabilities matrix i s not intended to represent a necessary sequence of learning since different firms do adopt different technologies using different sequences. See Lail (1992). 102 development. Such extra-market linkages play a significant role in promoting productivity increase and buildingtechnological capability at the national level. 4.31 TCI is constructedby mapping questions available in the PICS onto the taxonomy of FTC developedby La11 (1992) described above, with the scoring system shown in Table 4.4. TCI i s composed of 27 separate technical activities. Investment technological capabilities are represented by six separate technical activities covering investment planning, technology transfer and workforce training. Production technological capabilities are represented by 14 separate technical activities, which range from common process engineering tasks (such as upgrading machinery & equipment, introducing new technology, and I S 0 9000 quality management status) to product engineering tasks (such as improving existing products, introducing new products and researching and developing new designs). Linkages to technological capabilities are represented by seven separate technical activities involving technological interactions with buyers of output, suppliers and research institutions. A single point i s given for each technical activity the firm has performed; for higher levels of IT-related investments and computer-controlled machinery, an additional point i s scored. Therefore, each firm is ranked out of a total technological capability score of 29, and the result is normalizedto give a value between 0 and lq9* ~ 92Weighting complex activities more than simple ones was considered in the construction of TCI. However, the very nature of technological learning through the accumulation of experience in problem solving aided through formal researcheffort or aided by external inputs dictates that mastery would proceed from simpler to more difficult activities. For example, an establishment undertaking an advanced activity such as in-house process innovation would have also completed the basic task of assimilating process technology. Therefore, that particular establishment would have gained two points and further weighting would skew TCI in favor of establishmentswith high levels of technological capabilities. 103 U E . . . 0 . . . . . cr 0 Y xsva BZVIa3EulI3LNI AZIX37IdEuO3 do 33HI33a 104 Table4.4. TCI for the ThailandPICS Sample INVESTMENT PRODUCTION LINKAGESWITHIN ECONOMY Expect to make a substantial What percentageof your production Technology innovationdeveloped increase ininvestment inorder to machines is computer controlled? (iii-7) incollaboration with other firms? increasecapacity or improve (iii-18-1) quality? (iii-4) [None=O] [>O and e30 Percent=l] [Yes=l] [No=O] [Yes=l] [No=O] [L30Percent=2] Technology innovation developed What percentageof your next Employ staff exclusively for desigddoing incollaboration with universities? investment will be I T related? innovation/R&D? (iii-12) (iii-18-2) (iii-4-36) [Yes=l] [No=O] [Yes=l] [No=O] [None=O] [>O and e10 Percent=l] Subcontract R&D project to other Technology innovationdeveloped [LlOPercent=2] companiesor organizations? (iii-13) incollaboration withresearch institutions? Training of workforce to [Yes=l] [No=O] (iii-18-3) implement technology transferred from parent establishment? Paid royalties? (iii-14) [Yes=l] [No=O] (iii-19-2) [Yes=l] [No=O] Technology innovationdeveloped [Yes=l] [No=O] incollaboration other Planningto introduce new designs or institutions? (iii-18-5) Ifasupplier to aMNC and products inthe next 2 years?(iii-IS) learned new technology from that [Yes=ll [No=Ol MNC, was it explicitly via MNC [Yes=l] [No=O] licensing, training, quality Ifasupplier to aMNC, did you certification programs? Upgraded machinery & equipment? learn any new technology from (iii-20b) (iii-16-1) that MNC?(iii-2Ou) [Yes=l] [No=O] [Yes=l] [No=O] [Yes=l] [No=O] Runformal in-house training Entered new markets due to processor Use e-mail ininteractions with programs for employees? product improvements incost or quality? clients and suppliers? (viii-28u) (x-20) (iii-16-2) [Yes=l] [No=O] [Yes=l] [No=O] [Yes=l] [No=O] Use website ininteractions with Send employees to formal Filed any patent/ utility models or clients and suppliers? (viii-28b) training programs runby other copyright protected materials? (iii-16-3) organizations? [Yes=l] [No=O] (x-23) [Yes=l] [No=O] [Yes=l] [No=O] Developed a new product line? (iii-16-4) [Yes=l] [No=O] (Table4.4 continued on nextpage) 105 Table4.4. (cont.1 INVESTMENT PRODUCTION LINKAGESWITHIN ECONOMY Upgraded an existingproduct line? (iii-16-5) [Yes=l] [No=O] Introducednew technology that has substantially changedthe way the main product i s produced? (iii-16-6) [Yes=l] [No=O] Adaptation or R&D of technology transferred from parent establishment to suit local conditions? (iii-19-1) [Yes=l] [No=O] Have you received any government incentives to conduct technological innovation andR&D?(iii-21) [Yes=l] [No=O] Has your firmreceived any I S 0 (e.g. 9000,9002, or 14,000) certification? (iii-26) [Yes=l] [No=O] Note: Codes for variables inThailand PICS 104dataset denoted inparentheses. (b) TechnologicalCapabilitiesinthe ThailandPICS Sample Kernel Density Analysis 4.32 The data suggests a wide variation in TCI scores between manufacturing establishments inThailand across regions, industries, size, ownership, export status andindustrial estates. Table 4.5 presents the frequency distribution of the technological capability scores for the 1,385 establishments in the Thailand PICS sample. Figures4.8,4.9, and 4.10 graphically represent the distribution of TCI across regions, industries, size, ownership, export status and industrial estates with kernel density plots, which are useful tools allowing the visualization of the differences in technological capabilities scores across groups. Tables A4.3, A4.4 and A4.5 in the Annex present the statistics across the various groups noted above. These tables also provide average TCI scores as well as TCI for each of the activities: Investment, Production, and Linkages. 106 4.33 TCI confirms the macro analysis showing that there is much room for improvement in the technological capabilities of Thai manufacturingestablishments. More than half the sample do less than half of the 27 activities that composed TCI (score of 0.5) and only three percent of the sample have high levels of technological capabilities (scores more than 0.8) as seen in Table 4.5. The average TCI score i s 0.413 with a standard deviation of 0.222 and median of 0.435. Moreover, almost a fifth of the PICS establishments do fewer than five of the 27 technological capabilities activities (TCI scores between 0 and0.2). 4.34 Decomposing TCI into the three main types of activities shows that Thai manufacturing establishments are lagging in Production, and especially, Linkages to Technological Capabilities. Average Production TCI score i s 0.364 (5 of 14 activities) and average linkages TCI score i s 0.309 (two of seven activities). Fewer than one percent of all PICS establishments have highlevels of Production and Linkages (scores more than 0.8). Nearly a quarter of establishments engage in fewer than three of 14 Production technological capabilities activities, and more strikingly, more than 40 percent of the sample do only one or no technical activities related to Linkages. As Linkages are crucial in the development of technological capabilities, policy should endeavor to facilitate and develop the economic and political environment that foster cooperation, interaction and interdependence between economic agents. Table 4.5: DescriptiveStatisticsof TCI inThai Manufacturing(n=1,385) Mean 0.413 Std.Dev. 0.222 Median 0.435 OverallTCI InvestmentTCI ProductionTCI LinkagesTCI Mean 0.426 0.364 0.309 Std. Dev. 0.266 0.220 0.208 Median 0.500 0.357 0.333 Frequency distribution of TCI Over TCI OSTCI10.2 0.2cTCI10.4 0.4cTCI10.6 0.6eTCI10.8 0.8cTCI51 Percentage 19.4 30.3 26.1 21.3 3.0 InvestmentTCI OITCI50.2 0.2cTCI10.4 0.4eTCI10.6 0.6eTCI10.8 0.8cTCI11 Percentage 26.7 22.9 20.6 18.1 11.7 ProductionTCI 05TCI10.2 0.2cTCI10.4 0.4cTCI10.6 0.6cTCI10.8 0.8eTCI11 Percentage 23.5 29.7 32.7 13.4 0.8 LinkagesTCI OITCI10.2 0.2eTCI50.4 0.4eTCI10.6 0.6eTCI10.8 0.8cTCI11 Percentage 42.2 16.1 20.8 10.2 0.7 Source : ThailandPICS 2004. Author’s calculations. 4.35 Regionally, establishments located in the East and Central score highest and those in the North, Northeast, and especially, the South, score lowest on the TCI scale. 107 Establishments inBangkok and Vicinity on average score somewhere inbetween these two groups. Figure 4.8 shows the individual kernel densities of TCI for each of the six regions in Thailand. The mass of density for the East and Central lies to the right of the mass of density of all other regions. This implies that the probability of observing a manufacturing establishment with a high TCI score i s greater in the East and Central. The establishments located in the East and Central are consistently rankedfirst and second, respectively, interms of average T C I scores overall (0.495 and 0.451) and ineach of the three areas of activity: Investment (0.529 and 0.471); Production (0.428 and 0.396); and Linkages (0.370 and 0.333), respectively. Establishments in Bangkok follow with overall average TCI score of 0.415, and also in Investment (0.411), Production (0.370), and, surprising, Linkages (0.3 16). Establishments located in the Northeast region have the poorest average Investment and Linkages TCI scores of 0.303 and 0.202, respectively. Establishments located in the South have the poorest average ProductionTCI score of 0.217. 4.36 Across industries, establishments in electronics & electrical appliances, auto parts and machinery & equipment score highest and those intextiles, clothing, rubber & plastics and wood & wood furniture score lowest on the TCI scale. Figure 4.9 shows the individual kernel densities of TCI for each of the eight selected manufacturing industries in Thailand. The mass of density for the electronics & electrical appliances, auto parts, and machinery & equipment industries lies to the right of the mass of density of all other industries. This implies that the probability of observing a manufacturing establishment with a high TCI score i s greater in these three industries. The establishments in electronics & electrical appliances, auto parts and machinery & equipment are consistently rankedfirst, second and third, respectively, interms of average TCI scores overall and ineach of the three areas of activity: Investment (0.616,0.559 and 0.457); Production (0.479, 0.468 and 0.430); and Linkages (0.404, 0.378 and 0.347), respectively. Establishments in wood & wood furniture score the worst in all three areas, with averages of 0.265, 0.27 1 and 0.241 in Investment, Production and Linkages, respectively. (Establishments in textiles are tied for the worst average Linkages TCI score.) Establishments in clothing have the second worst average Investment TCI score of 0.340. Establishments inrubber & plastics have the secondpoorest averageProductionTCIscoreof 0.294. 4.37 Large establishments on average score higher on the TCI scale than SMEs, and a large part of the disparity can be attributed to differences in Investment TCI scores, specifically the gap between the incidence of in-house and outside training across these two groups. The average TI scores for large establishments and SMEs are 0.529 and 0.343, respectively. Much of the gap between the size groups i s traceable to differences in investment in technological capabilities. Large establishments and SMEs have an average Investment TCI score of 0.553 and 0.348, respectively. More specifically, the percentage of large establishments and SMEs running formal in-house training are 86.6 percent and 51.9 percent, respectively, and sending employees to formal outside training programs are 78.2 percent and 52.0 percent, respectively. (See Figure 4.10.) 4.38 Similarly, foreign-owned establishments on average score higher on the TCI scale than those domestically-owned, and most of the disparity can be attributed to differences in Investment TCI, specifically the gap between the incidence of training due to technology transfer from parent establishments as well as the incidence of learning new technology from MNCs. The average TI score for foreign- and domestically-owned establishments are 108 0.541 and 0.368, respectively. Similar to the findings for size groups, much of the gap between the foreign- and domestically-owned establishments i s traceable to differences in investment technological capabilities. Foreign- and domestically-owned establishments have an average Investment TCI score of 0.590 and 0.368, respectively. More specifically, the percentages of foreign- and domestically-ownedestablishments training its workforce due to implementationof technology transferred from the parent establishments are 45.0 and 3.1 percent, respectively. The percentages of foreign- and domestically-owned establishments learning new technology through licensing, training, and quality certification programs offered by MNCs buyers are 35.8 and 9.2, respectively. (See Figure 4.10.) 4.39 Exporters as well as establishments located within industrial estates on average score higher on the TCI scale than non-exporters and those located outside industrial estates, respectively. The average TCI score for exporters and non-exporters are 0.465 and 0.344, respectively. The gap in average scores for each of the three areas of activity between exporters and non-exports are approximately similar. Similarly, the average TCI score for establishments located within and outside industrial estates are 0.477 and 0.402, respectively. The gap in average scores between these two groups i s somewhat smaller and approximately similar for the three areas of activity. (See Figure 4.10.) RegressionAnalysis 4.40 Given that many of the above characteristics of manufacturing establishments are correlated, a more rigorous investigation using regression analysis is conducted to test for the influence of several factors on the level of technological capabilities. The following model i s estimated by OLS: TCI,=a PILARGE, &EXPORTER, P3FOREIGNi p,IND -ESTATE, ++psIMPORTED-TECHNOLOGq+&CAPITAL + + + +C { k ~ +CA,REGION,,~ ~ ~ k+p7AGE, -VINTAGE, ~ ~ ~ ~ +E,, , 113 k 1 where LARGE,EXPORTER,FOREIGN,IND-ESTATE, CAPITAL-VINTAGE, AGE,INDUSTRY and REGION are defined as above, and IMPORTED-TECHNOLOGY i s a dummy variable that i s equal to 1 if the establishment imported new machinery & equipment from a developed country, and equal to 0 otherwise. Separate regressions are estimated for TCI, Investment TCI, ProductionTCI and Linkages TCI. Results are presentedinTable 4.6. 109 Figure 4.8: Kernel Density Plots of TCI by Region 0.1 0 2 03 0.4 0.5 0.6 0.7 0.8 0.9 TechnologicalCapabilitiesIndex Source: ThailandPICS 2004. Author’s calculations. Figure 4.9: Kernel Density Plots of TCI by Industry 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 TechnologicalCapabilitiesIndex -clothing -Auto “`””FoodParts Processmg -Textiles -Wood &WoodFurmture -Electronics & ElectricalAppliances -Rubber &Plastics -Machinely &Equipment Source:Thailand PICS 2004. Author’s calculations. 110 3 4.4 1 TCI regression estimates show that the positive correlation between establishment size, export status and ownership found in the comparison of kernel densities still hold when controlling for several other establishment characteristics. As seen in Column 1 Table 4.6, the coefficients on LARGE, EXPORTERAND FOREIGN are both positive and highly statistically significant. The coefficients on IMPORTED-TECHNOLOGY and CAPITAL VINTAGE are also significant and positive. However, the coefficient on the industrial estate dummy variable in not significantly different from 0. The same relationship between the above establishment characteristics hold in the regression of Investment TCI, Production TCI, and Linkages TCI, as seen in Columns 2,3 and4, respectively. 4.42 In contrast to the findings in the previous section, establishments in Bangkok and Vicinity on average have higher TCI scores than all other regions when controlling for all the establishment characteristics in the full model. As seen in Column 1 of Table 4.6, the coefficients on each region measure the difference between average TCI scores of establishments inits region versus those inBangkok andVicinity, holding all other establishment characteristics inthe model constant. All coefficients are negative, and these results are statistically significant except for the coefficient on the East region. This pattern holds in the Production TCI and Linkages TCI regressions, as seen in Columns 3 and 4 of Table 4.6. Investment TCI for establishments in Bangkok and Vicinity i s higher than that for establishments in the Northeast and South, but not significantly different to the Investment TCI for establishments in the North, Central, and East regions, as seen in Column 2 of Table 4.6. Differences in technology across regions therefore capture firm characteristics such as size, export status and foreign ownership. As seen in Chapter 2 of this report, firms in the East and Central are larger and more export-orientedthan those inBangkok and Vicinity. 4.43 The pattern in TCI across industries is similar to results found in the previous section with one notable difference: establishments in rubber & plastics now have on average higher TCI than food processing, clothing, textiles and wood & wood furniture. When several establishment characteristics are taken into account, establishments in rubber & plastics on average score 4.3 percent higher on the TCI scale than establishments in food processing, as seen in Column 1of Table 4.6. This finding reconciles the apparent discrepancy between the classification of rubber & plastics as a high-tech industry in Chapter 2 and the results of the kernel density analysis. As seen in Column 2, this finding i s even strong when the dependent variable i s Investment TCI, for which establishments in rubber & plastics score on average 0.108 higher than those in food processing, and this result i s highly statistically significant. The coefficient on rubber & plastics i s not significantly different from that on food processing in the regression with Production TCI and Linkages TCI as dependent variables. Electronics & electrical appliances, auto parts and machinery & equipment still remain the industrieswhere establishments are scoring the highest on the TCI scale. 4.44 Consistent with the finding in Chapter 3, English language proficiency and IT skills of the workforce are strongly correlated with the technological capabilities of the establishment. Given the importance of worker skills inthe development of FTC and the finding of skills shortage in the previous chapter, skill quality of the workforce i s introducedin Equation [13 to shed light on the relationshipbetween TCI and worker skill. Dummyvariables indicating establishment self-assessed “very poor” rankings of 12 different workforce skills are addedto the full model. Results are presented inTable 4.7. Only the coefficient on “very poor” IT skills is 112 statistically significant and negative in all four equations. The coefficient on “very poor” English proficiency is negative and statistically significant in regressions with overall TCI and Production TCI as dependent variables, marginally significant in the Linkages TCI regression and not significant inthe Investment TCI regression. Table 4.6: TCI Regression Estimates (1) (2) (3) (4) [Dependent Variable] TCI InvestmentTCI ProductionTCI Linkages TCI Large 0.147′” 0.162’;” 0.132″‘ 0.095″‘ (0.011) (0.014) (0.012) (0.012) Exporter 0.065″‘ 0.051″‘ 0.042″‘ 0.097″‘ (0.011) (0.014) (0.012) (0.012) Foreign 0.068″‘ 0.100″‘ 0.052′” 0.037″‘ (0.013) (0.017) (0.014) (0.013) Industrial Estate 0.002 0.014 -0.001 -0.003 (0.014) (0.018) (0.014) (0.015) ImportedNew Machinery 0.056″‘ 0.068″‘ 0.053″‘ 0.025’ from DevelopedCountry (0.014) (0.018) (0.016) (0.015) CapitalVintage 0.001*” O.OOl** 0.001″‘ 0.001″‘ (0.000) (0.000) (0.000) (0.000) Age 0.000 0.000 0.000 0.001 (0.001) (0.001) (0,001) (0.001) Industry [Omitted Comparison Group: Food Processing] Textiles -0.053″ -0.022 -0.059″ -0.043″ (0.022) (0.026) (0.024) (0.021) Clothing -0.054″ -0.035 -0.068″‘ -0.015 (0.022) (0.026) (0.024) (0.021) Auto Parts 0.136″‘ 0.179″‘ 0.108*** 0.089* * (0.023) (0.028) (0.024) (0.023) Electronics& Electrical 0.104″‘ 0.172″‘ 0.065″‘ 0.076″‘ Appliances (0.021) (0.026) (0.023) (0.021) Rubber& Plastics 0.043** 0.109″” 0.013 0.026 (0.020) (0.024) (0.022) (0.021) Wood & Wood Furniture -0.065″‘ -0.067″ -0.065″‘ -0.029 (0.023) (0.027) (0.024) (0.023) Machinery& Equipment O.lOl*** 0.116″‘ 0.087″‘; 0.067″‘ (0.022) (0.026) (0.023) (0.022) Region [Omitted Comparison Group: Bangkok & Vicinity] North -0.059′” -0.024 -0,044’ -0.100″‘ (0.022) (0.027) (0.025) (0.023) Central -0.040″ -0.02 -0.039″ -0.041″ (0.018) (0.021) (0.019) (0.018) East -0.017 0.007 -0.025 -0.014 (0.016) (0.020) (0.017) (0.016) Northeast -0.131*** -0.134″‘ -0,101*** -0.130″” (0.023) (0.026) (0.027) (0.022) South -0.153′” -0.074*** -0.166″‘ -0.124″‘ (0.020) (0.024) (0.022) (0.021) Constant 0.274″‘ 0.239″‘ 0.267″‘ 0.187′” (0.023) (0.028) (0.025) (0.023) Observations 1,374 1,374 1,374 1,374 Adjusted R-squared 0.389 0.344 0.297 0.254 Robust standarderrors inparentheses. * significant at 10%; ** significant at 5%; *** significant at 1% Source : Thailand Productivity and InvestmentClimate Survey 2004. Author’s calculations 113 Table 4.7: Workforce Skills Regression Estimates (1) (2) (3) (4) [Dependent Variable] TCI InvestmentTCI ProductionTCI LinkagesTCI ‘Very Poor” Quality of Workforce Skill in: Englishlanguage -0.031″‘ -0.016 -0.034″‘ -0.022. proficiency (0.012) (0.014) (0.012) (0.012) ProfessionalCommunication -0.022 0.014 -0.035 -0.019 (0.022) (0.028) (0.025) (0.021) Social -0.003 -0.02 0.000 0.009 (0.025) (0.030) (0.026) (0.025) Teamwork -0.011 0.039 -0.038 0.008 (0.038) (0.038) (0.042) (0.035) Leadership -0.009 -0.014 0.004 -0.032 (0.020) (0.024) (0.021) (0.022) Time Management 0.01 0.006 0.003 0.025 (0.024) (0.031) (0.025) (0.025) Adaptability -0.004 -0.051 0.015 0.003 (0.031) (0.035) (0.033) (0.032) CreativitylInnovation 0.013 0.022 0.003 0.021 (0.022) (0.026) (0.023) (0.023) Numerical 0.016 -0.017 0.031 * 0.007 (0.018) (0.024) (0.018) (0.018) ProblemSolving -0.001 -0.012 0.014 -0.027 (0.026) (0.031) (0.028) (0.026) IT -0.047″‘ -0.045*** -0.047′” -0.027″ (0.012) (0.015) (0.013) (0.013) TechnicaVProfessional -0.034 -0.015 -0.034 -0.035 (0.021) (0.026) (0.022) (0.022) EstablishmentCharacteristics Yes Yes Yes Yes Observations 1,374 1,374 1,374 1,374 AdjustedR-squared 0.411 0.353 0.319 0.266 Notes :EstablishmentcharacteristicsincludeLarge,Exporter,Foreign, IndushialEstate,ImportedTechnology, CapitalVintage, Age, Industry,andRegion. . . . . A constantis also includedinthe regression. Robuststandarderrorsinparentheses. *significant at 10%; **significant at 5%;*** significantat 1% Source :ThailandProductivityand InvestmentClimate Survey 2004. Author’scalculations (c) TCI Matters for TFP This section presents the results of regression analysis examining the relationship between: TCI and TFP93.The following model i s estimated using OLS: ln(TFP),= a+@CI, PILARGE, P2EXPORTER, ,8,FOREIGN, p41ND-ESTATE, + + + + +p5IMPORTED-TECHNOLOGY;+&CAPITAL -VINTAGE, p7AGE, + where LARGE, EXPORTER, FOREIGN, IND-ESTATE, CAPITM-VINTAGE, AGE, INDUSTRY,REGION and IMPORTED-TECHNOLOGY are defined as above. Two alternative specifications for TCI are used: TCI and TCI decomposed into Investment TCI; ProductionTCI; and Linkages TCI. Results are present inTable 4.8. 93See Chapter 1and Chapter 1Appendix for the methodologyto computeTFP. 114 4.45 As can be seen in Table 4.8, TCI is strongly correlated with TFP, and all of this relationship i s being driven by Investment TCI. The results in Column 1of Table 4.8 show that the coefficient on TCI i s 0.264 and statistically significant at the one percent level. A 10percent increase inTCI i s associated with a 2.64 percent increase inTFP; moving to 1from 0 on the TCI scale i s associated with a 26.5 percent increase in TFP, holding all other covariates constant. The results in Column 2 of Table 4.8 show that the coefficient on Investment TCI i s 0.228 and statistically significant at the one percent level, whereas the coefficients on Production TCI and Linkages TCI are not statistically significant. A 10 percent increase in Investment TCI is associated with a 2.28 percent increase in TFP; moving to 1from 0 on the Investment TCI scale i s associatedwith a 22.8 percent increase in TFP, holding all other covariates constant. Table 4.8: TFP Regression Estimates (1) (2) lDeDendent Variable1 ln(TFW 1dTFP) TCI 0.270*** (0.089) InvestmentTCI 0.225*** (0.080) ProductionTCI 0.041 (0.098) Linkages TCI 0.045 (0.093) Large 0.402*** 0.395*** (0.035) (0.036) Exporter 0.213*** 0.215*** (0.034) (0.034) Foreign 0.155*** 0.147*** (0.039) (0.039) IndustrialEstate 0.068* 0.067* (0.037) (0.038) ImportedNew Machinery 0.109** 0.105** from DevelopedCountry (0.047) (0.047) CapitalVintage 0.001 0.001 (0.001) (0,001) Age 0.006*** 0.006*** (0.002) (0.002) Constant 2.176*** 2.175*** (0.074) (0.074) Industry Yes Yes Region Yes Yes Observations 1,047 1,047 Adjusted R-squared 0.984 0.984 Robust standard errors in parentheses. * significantat 10%;** significant 5%; *** significant at at 1% Source :Thailand Productivity and Investment Climate Survey 2004. Author’s calculations 115 4.46 Estimation of a model with TCI disaggregated and specified as 29 separate activities shows that running formal in-house training i s the only investment activity significantly associated with higher TFP. Computerizing and upgrading machinery and receiving I S 0 certification are the only production activities significantly associated with higher TFP, and using a website in interactions with clients and suppliers i s the only linkages activity significantly associated with higher TFP. Establishments running formal in-house training programs have, on average, 12.4 percent higher TFP than those that do not. Establishments that use computers to control 30 to 100 percent of their production machinery, have upgraded their machinery & equipment in the last two years and have received any I S 0 (e.g. 9000, 9002 and 14,000) certification have, on average, 12.0, 10.8 and 7.9 percent higher TFP, respectively. Establishments that use a website in interactions with clients and supports have, on average, 9.5 percent higher TFP. (See Table 4.2.10 in the Annex for detailsSg4)Chapter 5 will examine more closely the positive effects of website use and ICT ingeneral on firm productivity. C. CONCLUSIONS AND POLICY RECOMMENDATIONS 4.47 Given the deficiencies in technological capabilities, the typical firm in Thailand will use the same technology less efficiently than its counterparts in competitor countries with better trained labor. While all activities listed inthe matrix of technological capabilities do not have to be performed for every industrial venture, there i s a core set of basic competencies in each of the three categories that have to be internalized by the firm to ensure successful commercial operation. If a firm i s unable by itself to decide on its investment plans or selection of equipment processes, or to reach minimum levels of operating efficiency, quality control, equipment maintenance and cost improvement, or to adapt its product designs to changing market conditions, or to establish effective linkages with reliable suppliers, it i s unlikely to be able to compete effectively in open markets. Moreover, the basic core of skills, knowledge and experience must grow over time as the firm undertakes more complex tasks. The hallmark of a technologically “mature” firm is the ability to identify prerequisite skills for efficient specialization in technological activities, to extend and deepen this knowledge with experience and effort and to draw selectively on other economic agents to complement the existing capabilities of the firmsg5 4.48 As a result, knowing where the firmstands on the technological ladder relative to its main competitors is vital information that managers require to increase the productivity and growth of the firm. This competitive stimulus can trigger initiatives and the adoption of measures deemed appropriate to catch-up with industry leaders. Providing a metric such as industry-specific TCIs as a benchmark for firms to measure its technical skills, knowledge and experience against those of its competitors may encourage more investment in technological capabilities. For instance, if a manager sees that most firms in the industry are running formal in-house training or receiving I S 0 certification, then there i s naturally competitive pressure for the firm to follow in undertaking these activities. The challenge for policymakers will be to set up a mechanism that can systematically collect and analyze such data required to update and – 94The percentageincreasein wages associatedwith a dummy variable coefficient i s calculated as exp(p’ – 1. ) 95See La11(1992). 116 disseminate industry-specific TCIs to Thai firms. These indices will complement the available country-level technological indicators that are useful for international comparison, but cannot guide the formulation of industry-specific policies. 4.49 The low intensity of linkages highlighted in this chapter should also be addressed. Firms rarely acquire capabilities in isolation, and FTC building must depend on linkages and cooperation with other firms and various support institutions. Interaction and interdependence between agents lead to collective learning, and thereby, linkages are a fundamental feature of technological capability building that policymakers should facilitate. Strengthening linkages (firm-to-firm collaboration, joint projects between firms and research institutions and universities, partnerships with MNCs, etc.) would help Thai firms better tap into the existing stock of global knowledge and move up the technological ladder. Facilitating access and ability of firms to acquire and use e-mail and websites for interactions with suppliers and buyers will also increase the linkages among firms. The specific needs of firms as well as the weaknesses of the national innovation system in which they are striving should be carefully investigated to guide the formulation of actionable policy measures. 4.50 The importance of skills cannot be stressed enough. Even with the best institutional framework, including all the factors that have been proven to be essential for technological development in the literature (e.g., access to technology finance, incentives for R&D, strong IPR regime, etc.), it would be difficult to improve the technological performance of firms without workers that have sufficient education, skills and training. These fundamental economic agents are the key drivers of technological capabilities within firms. The strong correlations between very poor IT skills and English language proficiency and lower TFP are empirical evidence pointing to the crucial role that human capital plays inthe development of FTC. Both Korea and Singapore, which have rapidly climbed to the frontier of world technology, have a labor force with high levels of education and skills, and their development experience can’ inspire the formulation of responsive shll-upgrading and education-increasing policies in Thailand. Given the investment climate constraint of S L shortage identified by PICS establishments and the country’s deficit in the completion rate at the secondary-education level, a particular focus on enhancing the quality of high-school education as well as vocational training would certainly contribute to facilitating the development of technological capabilities. 117 References Bell, M., and Pavitt, K. 1997. “Technological Accumulation and Industrial Growth: Contrasts between Developed and Developing Countries.” In Archibugi, Daniele and Michie, Jonathan, (Editors), Technology, Globalisation and Economic Performance, Cambridge: Cambridge University Press, 157-210. Comin, D. 2004. “R&D: A Small Contribution to Productivity Growth.” Journal of Economic Growth, 9(4): 391-421. Hobday, M.2000. “East versus SoutheastAsian Innovation Systems: Comparing OEM and TNC- led Growth inElectronics.” InKimL.and Nelson R. (Editors). Technology,Learning, & Innovation, Cambridge UniversityPress,129-169. Hobday, M. 1995. Innovation in East Asia: the Challenge to Japan. Aldershot, England: Edward Elgar. Katz, J.M. 1987. “Domestic technology generationinLDCs: a review of researchfindings.” InJorge M. Katz (Editor), Technology Generation in Latin American Manufacturing Industries. Basingstoke: Macmillan, 1-30. Lall, S. 1992. “Technological Capabilities andIndustrialization.” WorldDevelopment, 20(2): 165-186. 2000. “Technological ChangeandIndustrialization inthe Asian Newly Industrializing Economies: Achievements and Challenges”. InKimL.andNelsonR. (Editors). Technology,Learning, & Innovation, Cambridge UniversityPress,13-68. Metcalfe, S. 2003. “Science, technology and innovation policy.” In Ganeshan Wignaraja (Editor), Competitiveness Strategy and Industrial Performance in Developing Countries: A Manual for Policy Analysis. London: Routledge, 95-130. Pack, H.and Westphal, L.E. 1986. “Industrial strategy and technological change: theory versus reality.” Journal of Development Economics, 22(1): 87-128. Pietrobelli, C. 1997. Industry, Competitiveness and Technological Capabilities in Chile: A New Tiger from Latin America. London: Macmillan. Romijn, H. 1999. Acquisition of Technological Capability in Small Firms in Developing Countries. London: MacmillanPress. Wignaraja, G. 1998. Trade Liberalisation in Sri Lunka: Exports, Technologyand Industrial Policy. London: Macmillan. 2002. “Firm Size, Technological Capabilities and Market-oriented Policies inMauritius.” Oxford Development Studies,30(1):87-104. Wignaraja, G. and Gerrishion I.1999. “Adjustment, technological capabilities and enterprisedynamics in Kenya.” In Sanjaya La11 (Editor), The Technological Response to Import Liberalisation in SubSaharanAfrica. London: Macmillan, 57-111. 118 5. ICTAND FIRMPERFORMANCE INTHAILAND 5.1 This chapter investigates the relationship between ICT and firm performance. While the hype concerning the “New Economy” has largely subsided, the evidence concerning the economic impact of ICT has strengthened, largely through the use of firm-level data that empirically demonstrates a strong relationship between ICT and measures of firm performance, includingTFP. 5.2 While studies employing growth accounting methodologies in the first half of the 1990s failed to distinguish the impact of ICT at the aggregate level, more recent studies have determined, over a fairly wide range of countries at different levels of development, that ICT diffusion has contributed significantly to GDP growth and productivity growth. Inthe UnitedStates, the evidence suggests that the doubling of labor productivity growth to 2.5 percent in the second half of the decade from 1.3 percent from 1973 to 1995, while falling over the same period in Europe, was largely attributable to the faster diffusion of ICT in the United States relative to Europe. This holds both for manufacturing and services. Triplett and Bosworth found that ICT contributed to labor productivity growth broadly across most service industries due to capital deepening, but that much of the productivity impact had already been captured before 1995 and was therefore less pronounced during the post-1995 increases. Buildingto an extent on Bosworth, van Ark and T i m e r (2003) found that the benefits of ICT are sector-specific, greater in manufacturingthan inservices but were nonetheless positivein services. 5.3 Based on existing literature, this chapter conjectures that ICT impacts firm performance through three main channels: skills, innovation and networking. Studies have identified a close relationship between ICT use and labor productivity growth in Germany, Netherlands, Canada and A~stralia.’~Countries with a high share of skilled ICT workers in the workforce had higher investment in ICT than other countries. Bechetti, et. a1 (2003) found that the contribution of ICT to productivity growth i s better understood by decomposing it into software and telecommunications. Software investment increases demand for skilled workers thereby altering the skill mix in firms in favor of higher labor productivity (learning gains). Tan (2000) provides evidence on Malaysia to suggest that firms alter the skill mix prior to ICT adoption, and that these learning gains increase with years of experience with ICT. After four years, the learning gain for firms providing training was 31%, while for others 14%. 5.4 The literature suggests that ICT investments facilitate sharing of information — –particularly those that increase the ability of firms to innovate. Telecommunication investments influence the creation of new products andor processes by making information about market preferences more available to the firm and reducing 96OECD (2003) 119 the time it takes to convert knowledge of consumer preferences to products. Innovation – – creating new products, entering new markets, or introducing new technologies — can raise productivity levels by: (a) better serving customer’s needs, therefore enabling higher prices, which raise the ratio of output to inputs; and (b) introducing processes that reduce inputcosts (such as introductionof systemsthat reduce the cost of quality defects). 5.5 ICT enables networking, which reduces search, contractingand coordination costs. ICT are being increasingly applied to integration and optimization of material, information and financial flows between a particular type of network — a supply chain.97 Traditional inter- and intra-firm manual or semi-automated processes are replaced by partnership networks enabled by sharing of information, processes, decisions and resources. The key to effective supply chain management i s transparency, to enable information to substitute for inventory stocking as a hedge against ~ o l a t i l i t y . ~B y ~ sharing inventory and sales information levels across a supply chain, each participant i s able to lower inventories to levels required to fulfill immediate orders and to have the basis for anticipating demand based on the flow of orders through the supply chain. Lower inventory means lower working-capital requirements. Anticipating demand enables more efficient production scheduling to minimize switchover costs, reduce overtime and better manage humanresources and equipment. 5.6 ICT can enable firms to participate in regional and global production networks. Yusuf (2003) claimed that the future of the East Asia region i s “linked inexorably with its ability to ensure that ICT i s mastered by and readily accessible to the broad mass of the population. For East Asian firms in the highly competitive export industries that supply the bulk of the region’s income, profits are frequently a function of membership in international networks that depend on ICT-based products and use ICT to coordinate production, expedite delivery and embark on collaborative design and research. Firms, regions and economies that lag in their adoption of ICT will certainly find their competitive position eroded and may even by shut off from new commercial opp~rtunities.’~ 5.7 The fourth channel through which ICT has affected performance is as a production sector — essentially through contributing a dynamic export sector to the economy. The ICT industry, famously characterized by Moore’s Law — the circuit density of microprocessors doubles every 18 months — i s marked by rapid technological change, innovation and growth. For countries with an ICT producing sector (software, hardware or services), the productivity gains associated with ICT have impacted the overall economy even if the sector i s a relatively small share of GDP. An OECD study of Finland, Korea and Ireland suggest that about one percent of labor productivity growth from 1995 to 2001 was due to the strong performance of ICT manufacturing sectors. Thailand has invested considerably inthe possibility of developing an export sector. “Supplychainsincludepartssuppliers,manufacturers,distributors,logisticsserviceproviders, wholesalers and retailers engagedinthe delivery of aparticular product or service. 98″Information Sharing ina Supply Chain,” HauL.Lee and Seungjin Whang, Research Paper SeriesNo. 1549,GraduateSchool of Business, Stanford University. 99Shahid Yusuf, Innovative East Asia: The Future of Growth. World Bank (2003). 120 5.8 The key question for Thai policymakers is to make policy choices that maximize the potential gains through these channels. This requires first an understanding of the extent to which the relationship between ICT and performance has taken place in Thailand. Has ICT investment increased firm performance? And more specifically, through increased skills, innovation or networking? To the extent that patterns observed in more advanced countries are not taking place in Thailand, understanding the likely causes can contribute to policy formulation. The PICS Results on ICT Use and Productivity 5.9 For several reasons, empirical observations of productivity gains from ICT are not straightforward, and recent firm-level surveys have been difficult to discern. Several studies have suggested a significant time lag, such that productivity gains from ICT investment occur three to seven years after ICT investments take place. Second, ICT as an enabling technology i s embedded inmany forms of capital investment and depends on complementary investments, changes in behavior and process. Third, disinflation of hardware costs complicates productivity estimates. 5.10 Despite these measurement difficulties, the PICS results suggest a positive and significant role of ICT useon firmperformance. Controlling only for industryand region, e-mail and website use i s strongly correlated with higher TFP (See Column 1of Table 5.3). However, when establishment characteristics including size, export status, ownership, industrialestates, technology measures and establishment age are added to the model, only website use remains statistically significant. (See Column 2 of Table 5.3.) In the full model with controls for establishment characteristics, industry, and region as well as firm-specific technological capabilities, website use remains statistically significant. Website use i s associated with a 9.5 percent higher TFP on average. (See Column 3 of Table 5.2.’0°) p)- – loo Percentageincreasein wages associatedwith a dummy variable coefficient is calculated as exp ( 1. 121 Table 5.1: TFP Regressions (1) (2) (3) JDependent Variable] ln(TFP) ln(TFP) In(TFP) Use e-mail ininteractionswith clientsandsuppliers 0.139*** -0.009 -0.013 (0.038) (0.034) (0.035) Use website in interactionswith clients andsuppliers 0.207*** 0.114.’; 0.091** (0.042) (0.036) (0.035) Large 0.427’** 0.346*** (0.034) (0.037) Exporter 0.205*** 0.188′” (0.034) (0.034) Foreign 0.178″” 0.132′” (0.038) (0.041) IndustrialEstate 0.068* 0.078** (0.037) (0.039) ImportedNew Machinery 0.132*** 0.094** from DevelopedCountry (0.046) (0.047) CapitalVintage 0.001 0.001 (0.001) (0.001) 0.006*** 0.005*** (0.002) (0.002) Constant 2.704**’ 2.249*” 2.249*** (0.048) (0.067) (0.067) TCI No No Yes Industry Yes Yes Yes Region Yes Yes Yes Observations 1,047 1,047 1,047 AdjustedR-squared 0.978 0.984 0.985 Nofe :TCI include 27 variablesrelatedto firm-level technologicalcapabilities. See Table 4.2.10inthe Annex to Chapter4. Robuststandarderrors inparentheses. *significant at 10%;**significant at 5%; *** significant at 1% Source : Thailand Productivity and InvestmentClimate Survey 2004. Author’s calculations ICT Access and Innovation 5.11 The literature suggests that ICT investments facilitate sharing of information — — particularly those that increase the ability of Thai firms to raise their productivity levels through innovation. Innovation — creating new products, entering new markets, or introducing new technologies — can raise productivity levels by: (a) better serving customer’s needs, therefore enabling higherprices, which raises the ratio of output to inputs; and (b) introducing processes that reduce input costs (such as introductionof systems that reduce the cost of quality defects). 5.12 PICS data for Thailand suggest that ICT use is strongly correlated with innovation, even when controlling for firm size, exporters, foreign ownership and industry. Those firms who usede-mail were far more likely to spend on R&D, introduce new products, markets and technologies. Firms that had a website were far more likely to enter new markets, file patents and introduce new technologies. 5.13 The PICS data suggest that increasing ICT use on the margin increases the probability of innovation, even when controlling for firm characteristics such as 122 size, exporters, foreign ownership and industry. Those firms who used e-mail were more likely to spend on R&D, introduce new products, markets and technologies. Firms that had a website were more likely to enter new markets, file patents and introduce new technologies. Table 5.2: ICT and Innovation Regressors R&D New Entry into PatentFiling New Product ProductLine New Spending Products New Markets Dummy LineDummy Upgrade Technology Dummy Dummy Dummy Dummy Introduction Dummy EmailDummy 0.089*** 0.129*** 0.100*** 0.004 0.124*** 0.103*** 0.103*** (0.025) (0.031) (0.032) (0.019) (0.033) (0.029) (0.034) WebsiteDummy 0.039 0.055* 0.100*** 0.041** 0.018 0.041 0.072** (0.026) (0.032) (0.033) (0.019) (0.035) (0.031) (0.035) Current Employment 0.069*** 0.032*** 0.048*** 0.017** 0.080*** 0.081*** 0.123*** (0.010) (0.012) (0.013) (0.007) (0.013) (0.012) (0.014) ExportIntensity 0.039 -0.041 -0.01 0.020 -0.012 -0.009 -0.028 (0.027) (0.033) (0.034) (0.019) (0.035) (0.031) (0.036) ForeignOwnership Share -0.04 0.054 0.001 (0.004) 0.065* 0.015 0.034 (0.027) (0.036) (0.037) (0.019) (0.038) (0.034) (0.038) IndustryDummies Yes Yes Yes Yes Yes Yes Yes RegionDummies Yes Yes Yes Yes Yes Yes Yes Observations 1033 1033 1033 1033 1033 1033 1033 Robuststandarderrors inparentheses * significantat 10%;** significantat5%;*** significantat1% ICT, Skills and Performance 5.14 The data strongly suggest that ICT adoption increases demand for higher skills. First, it is clear that inadequate ICT skills impacts firm performance. Forty five per cent of manufacturing firms rated the IT skills of their shlled production workers as “very poor”. “Very poor” IT skills are significantly correlated with lower TFP, and this negative correlation between “very poor” IT skills and TFP i s mainly seen in establishments producing machinery & equipment. 5.15 Second, it is clear from the PICS that firms perceive the lack of knowledge and availability of trained ICT personnel and the lack of experienced consultants to provide or design ICT-based solutions as important or critically important constraints for introducingor expanding ICT use. This was consistent across different sized firms. Overall, the lack of ICT-skilled or ICT-knowledgeable employees seems be the larger concern when compared with cost and the expected returns. The Malaysia PICS also shows a similar view. This reflects the fact that firms understand that skilled- labor i s a critical input to ICT adoption and its use as an integral part of its business operations. Firms that realize this may be most risk averse when thinking to adopt ICTs. Even a firm with enough financial resources and the motivation to adopt an ICT product or service, would not be able to launch or sustain its ICT use in their daily business operations without adequate supply of ICT-skilled employees. 123 Constraints Small Medium Large Highcost of ITequipment andmaintenance 20% 18% 18% I Lack of knowledge and trained IT personnel I 42% I 35% I 33% I Low returns to investments inIT 15% 12% 11% Lack of experienced consultants to provide or design IT-basedsolution systems 39% 37% 32% IT-based systems do not improve productivity 20% 13% 13% 5.16 PICS provides evidence of a positive relationship between website use and skill upgrading. Figures 5.1 and 5.2 show the kernel densities of the logarithm of the employment share of skilled workers for large establishments (solid line) and SMEs (dotted line) that do and do not use website in interactions with clients and suppliers. Because growth of establishment size relies disproportionately on the employment of unskilled workers, the sample i s divided into large and S M E subsamples. As can be seen in Figures 5.1 and 5.2, the mass of density for website users lies to the right of the mass of density of website nonusers. This implies that the probability of observing a higher share of skilled workers (skill upgrading) i s greater for website users. The difference i s greater among SMEs than inthe subsample of large establishments. 5.17 This result is robust and striking when establishment characteristics are taken into account. The counterfactual density (dashed line), which factors in export status, ownership, industrial estate, foreign technology transfer, capital vintage, industry and region of each establishment, inboth Figures 5.1 and 5.2 lies further to the left of the density of the unmatched establishments that do not use website, especially so for the subsample of large establishments. Therefore, the difference in the masses of the actual density for users and the counterfactual density for nonusers with similar observable characteristics suggests that website use impacts skill upgrading. For large establishments, the impact of website use i s quite noticeable throughout the distribution, whereas for SMEs, website use exerts much influence on skill upgrading in the lower and middle parts of the distribution, but diminishes for high shares of skilled workers. (See Appendix for details on methodology.) 124 Figure5.1: KernelDensitiesof the EmploymentShare of SkilledWorkers. Sampleof Large Establishments ‘1 -4 -3In(shareof skilled workers) -2 -1 0 ___________Website —–MatchedNo Website No Website Figure 5.2: KernelDensitiesof EmploymentShare of Skilled Workers. Sample of SMEs ‘4 s 0 I I I 1 -3 In(shareof skilledworkers) -2 -1 0 I__.__.____.Website —–MatchedNo Website No Website ICT andNetworkedProduction 5.18 The PICS did not appear to suggest a strong correlation between ICT use and participation in such networks. There appears to be no correlation between what we traditionally understand as regional production networks — automotive and electronics — and increased electronic sales. However, these two sectors clearly use email and internet more than other sectors. 5.19 The study finds that the Internet is usedmostly for basic intra-firm and inter-firm communications. Impact of e-commerce on productivity and SG i s therefore yet to be seen. If this i s the case for Thailand, further studies are necessary to identify the specific constraints firms face in adopting e-commerce, whether domestic or international. PICS data also show that currently, firms only use websites (21 percent of firms) to advertise 125 their products only as much as they advertise through the media (21percent of firms) and exhibitions (23 percent of firms). The findings of an OECD country survey (2004) confirm that most SMEs do not adopt e-commerce, if the benefits do not outweigh the costs of systems development and maintenance. 5.20 The PICS findings are consistent with other observations of Thailand’s performance. Wiboonchutikula (2002) found that SMEs have more subcontracting work than large firms, and that subcontracting represented 3.3, 2.8 and 1.4 percent of sales for small, medium and large firms respectively. Subcontracted work i s most concentrated in clothing, wood, plastic, mineral, automotive and non-electrical machinery. Considerable progress has been made in raising awareness among SMEs of the potential benefits of information technology. However, utilization of ICT by the majority of SMEs i s for basic e-mail communication and administrative applications such as payroll and bookkeeping. Few SMEs have taken the further step of integrating ICT into their business operations or engaged ine-commerce. Figure 5.3: Computer Use by Size of Firm2001 120% 100% 80% 60% 40% 20% 0% Computers Internet E-Commerce Production Source: Ministryof industrySurvey (2001) Figure 5.4: ICT use by Sector 1 80 70 60 50 li 40 % 8 30 20 10 0 Email Website E-Sales 126 Can the ICT Industry Drive Productivity Growth? 5.21 The Software Industry Promotion Agency (SPA), established under the MICT in 2003 to support the software industry has launched a number of initiatives, including establishing a one-stop service center. The centre facilitates visas and work permits for skilled ICT workers, promotes activities to improve animation IT skills and encourages firms to develop IT solutions for SMEs. The National Software Park was also established as a government agency under National Science and Technology DevelopmentAgency to stimulate the development of Thai software industry cluster. It is, however, still early to assess the impact of these initiatives and organizations on Thailand’s software sector development, especially with regard to software exports. Thailand’s neighbors, including India, China and the Philippines, have comparative advantages over Thailand and lead in almost all ICT market segments, including business process outsourcing, software and application development and outsourcing and hardware production. Thai domestic demand for ICT services i s however growing (to 103 billion baht in 2004 from approximately 57 billion baht in 2001), and there i s potential for the Thai software sector to provide IT services for the domestic economy. Conclusions and Policy Implications 5.22 ICT i s associated with increasing firm performance in Thailand. However, the only strong relationship contributing to this impact observed through the PICS i s through innovation. The two other expected means for improving firm-level performance — skills and interfirm networking — do not seem to be key means by which ICT contributes to firm-level performance. What are the implications for these findings? The most obvious one i s the importance of increasing access to information, given its critical importance of enabling productivity growth and innovation. Given the large share of demand of the ICT sector driven by government, the quality of execution of ICT use by the government will influence the private sector significantly. Policies to Increase the Contributionof ICT Skill Development. 5.23 Policies to increase the firm-level impact of ICT networks. Three reasons may explain why ICT does not appear to be contributing to increased firm-level performance through networking. Telecommunications regulatory policy has not facilitated widespread, low- cost, broad-band access in Thailand. The networking phenomenon leverages the network effect, as defined by Robert Metcalf. It describes an exponentially increasing value of networks in proportion to the number of participants. To the extent that prices limit broadband access, network effects are suppressed. Current government policy has not supported adoption of networked production. Rather, the most visible public policies have focused on specialized niches such as multi-media and government services to citizens (People’s PC, smart-card, etc.). 127 Networked firms require complementary investments to make networking valuable. Whereas the international community i s benefiting from these developments, Thailand has an underdeveloped third-party intermodal logistics industry. Shipments are not trackable through intermodal systems (involving transfers between modes of transport, such as rail-truck-sea) due in part to the relative underinvestment in the public rail system and other transport segments. Third party logistics exists only on a limited basis for export shipments. In addition to logistics, financial settlement systems must adapt to the new environment. InThailand, there i s little available on-line credit, and there are high deposit requirements (500,000 baht) to initiate an on-line account, and there are no electronic settlement mechanisms. Business-to-business e-payments require guaranteed funds, speed, low fees and effective settlement mechanisms. IncreasingBroadbandAccess to Firms 5.24 The regulatory environment for telecommunications i s underdeveloped. As a result, Thailand’s fixed teledensity i s low compared with other low-middle income countries as indicated inFigure 5. Poor performance in the sector can be attributed to the slow pace of reforms. The sector regulator, the National Telecommunications Commission was formally established only inNovember 2004 after lengthy delays in the appointment of commissioners. The incumbent state-owned operators, TOT Corporation and CAT Telecom, have been corporatized but privatization has not yet taken place. The concession structure currently in place i s not suitable for a competitive multi- operator environment and has inhibited competitive new entry and investments in the sector. However, the fully competitive mobile market in Thailand i s gradually impacting the local- and long-distance segments, improving the general level of competition in the telecom sector. 5.25 The November 2004 appointment of commissioners to the National Telecommunications Commission (NTC) was a first step towards undertaking much needed telecom sector reforms. The NTC clearly has a pressing agenda of improving competition in the Thai telecommunications market. Key issues for NTC to address include: (i)implementing an interconnection regime that allows for fair competition; (ii) taking the lead in the management of limited telecommunications resources including numbers and the radio spectrum; and (iii) establishing and implementing a tariff regime. To further develop a license regime, NTC could help initiate conversion of telecommunications concessions into licenses. 5.26 Figure 5.5 shows that Thailand i s very close to the regression line, i.e. close to what one would expect for a country with its per capita GNP. Whereas teledensity in China, Vietnam and India seem to surpass this expectation. 128 Figure 5.5: MainLines per 100people and GDPper Capita Purchasing Power Pari1 y-0.8824~ 4.634: – Y R’ =0.7689 5.00 B 1@ 6 -7 -2 00 LNGDP per capita Source: ITU(2003) 5.27 The cost of business telephone connections, monthly subscription costs and broadband are also comparatively higher in Thailand, as indicated in Figures 5.6 and 5.7. ICs data analysis’O’ shows that the quality and cost of telecommunications infrastructure affects ICT use by firms, with firms more likely to use ICT, when the quality of services i s better. Affordable broadband services provides firms, especially SMEs, with the advantage of access to high-speed communications and the ability to reach a global marketplace, previously available only to large firms. Broadband competition inboth inter-modal (cable, DSL, wireless and fiber) and inter-operator i s key in lowering prices for services and increasing take-up by business and individual users. Currently, all ISPs must have an international link through CAT Telecom’s international gateway (NIX). Restrictions on ISPs from reselling bandwidth also provide no competition in Thailand’s relatively high-priced leased-line market. Moreover, foreign equity participation limits in the sector capped at 49 percent have deterred growth by almost 25 percent and impacteddirect prices by almost 26 percento2.While recent MICT initiatives have led to a reduction in broadband prices, sector liberalization i s the most effective manner inwhich to lower prices for telecommunication services. 5-28 Poor telecommunications sector performance has resulted in lower quality and higher costs for telecommunication services. Businesses in Thailand had to wait an average of 22.3 days to obtain a telephone connection, compared with 8.8 days, 13.2 days and 12.5 days inMalaysia, the Philippines and China, respectively. 101 The Role of ICTinDoingBusiness,( WorkingPaper – Qiang, Clarke, Halewoodand ,Gomez, 2005) 102Cost of ServicesTradeRestrictionsinThiland, World Bank, 2004 129 Figure 5.6: Cost of BusinessTelephone Connection and Monthly subscription inUSD I 2o n Business Connection (US$) IBusiness monthly subs. (US$) Source: lTU(2003) 5.29 The RTG i s interested in enabling broadband access, but the underlying foundation of market infrastructure does not exist to achieve this efficiently. The vast majority of telephone access in Thailand i s mobile, of which 80 percent i s through pre- paid phone cards. Without liberalization and effective regulation, any effort to bridge the digital divide will be more expensive, less sustainable and will be paid through either taxes or higher costs by Thai citizens. This i s because the regulatory certainty does not exist to maximize the contribution of the private sector before any subsidy. State-owned enterprises continue to play a substantial operational role in the sector, and the power of competition has not been appliedto extending telecom services. Figure 5.7: Broadband Prices Der Month. 2003 Thailand Sri Lanka China – Malaysia Korea Singapore Percent of monthly income (GNI) –t per 100 kbit/s Source: ITU(2003) UpdatingPolicy to Focuson Private Sector Performance 5.30 ICT policy in Thailand has been characterized by its comprehensiveness — including e-society, e-commerce, e-industry, e-citizen and e-government. While the vision i s admirable, it reflects a possible lack of selectivity or tight strategic focus. To the 130 extent that such a focus exists, the emphasis has recently been on popularizing access of individuals to ICT rather than firm performance through innovation or networks. 5.31 The MICT, established in October 2002, i s responsible for formulating Thailand’s ICT policy and plans, promoting and developing ICT activities and implementing national ICT projects. While this move has rationalizedthe ICT policy-making function of the government into one agency, there are several other agencies also involved in different ICT programs and projects, including the National Information Technology Committee, which reports to the Prime Minister’s office, andthe National Electronics and Computer Technology Center ( NECTEC), which reports to the Ministry of Science, Technology and Environment (MOSTE). While some of the policy overlap has been improved, policy execution remains a significant issue. Recent problems with the national Smart Card project are indicative of insufficient attention to project management, monitoring and supervision, including interagency coordination. 5.32 Government policies, while focusing on the development of a knowledge society have also paid special attention to software development, especially in the area of digital content and games. The ICT Master Plan (2002-2006) established a software production target of 90 billionbaht by 2006, with 75 percent of the amount being exports. However, in2004 the software market was worth approximately only 17.9 billion bahtlo3. 5.33 The government-led low-cost personal computer project, the People’s PC, which was started in 2003, led to extraordinary growth in the consumer marketlo4 due to increased user awareness and vendor competition. It also led to lower prices for operating systems and software and increased penetration of open-source systems in Thailand. Further, the People’s PC program has led to an increase in computer ownership in Thailand, and vendors and local assemblers have experienced pricing pressures and lower profits. While by some accounts, actual shipments were below target volumes set by the government, the People’s PC project has boosted IT awareness and computer penetration in Thailand, with approximately 11.1% of households owning PCs in2004 comparedwith 5.1% in2001. 5.34 E-Government. Thailand is in the midst of a highly aggressive government restructuring program which started in 2002. One of the results of this restructuring was the establishment in October 2002 of the MICT. One of the five major themes supporting Thailand’s vision of achieving a knowledge-based society, the government would like to see 70 percent of services provided electronicallyby 2005 and 100percent by 2010. This particularly the Revenue Department within the Ministry of Finance — Somemoved ahead year, ICT was to be used to develop integrated back-offices. agencies — have very quickly toward these goals, including through the successful Government Financial Management Information System and e-revenue. The vast majority of the government i s not yet ina position to guarantee e-service delivery. lo3Source: NECTECThailandICT Indicators2005. lo4Source-IDC, September2004 131 5.35 ICT cities,softwareparksandtechno parks. The government has undertaken a number of simultaneous initiatives designed to create localized ICT clusters in cities such as Chiang Mai, Phukhet and Khon Kaen. To date, these projects lack a coherent shared vision with the private sector, academia and local communities, as well as low levels of awareness on what the government plans for these cities.’05 5.36 The World Bank recently supported an assessment of the e-Citizen portal in order to illustrate some of the challenges of achieving a citizen-focused, service-oriented ICT tool. The assessment revealed some key difficulties: a lack of a citizen-focused service culture in some parts of the government; a shortage of skilled staff; a lack of collaboration and sharing of skills across government; and technology and processes to support interoperability. 5.37 The RTG has also taken a number of initiatives in the area of e-government. These include implementation of e-procurement, smart cards for Thai citizens, e- government portal, e-filing of taxes and online information services among others. Government spending on IT has seen significant increases in the last five years, and potentially the government, as a model user of ICT services, could drive demand for ICT services in Thailand. While evidence i s anecdotal, the private sector has begun to express concerns with many of the projects. Among them: e The People’s PC project lacked effective maintenance infrastructure to support PCs with hardware or software flaws. Supply chain planning problems caused several vendors to hold excessive parts inventory and lose substantial sums in order to participate. These are correctable problems, and the program has been replicated in several countries; e The Goodnet project, aimed at providing a safe Internet experience, initially generated high enthusiasm and participation, but the private sector i s now questioning the benefits of participation due to a lack of follow-through; e Key local stakeholders to the ICT cities projects — Chiang Mai and Phukhet — have become disillusioned with the lack of clarity and progress; e The Smart Card project — arguably the world’s largest multi-purpose smart card program of its kind — was launched with few staff and a lack of cooperation from some ministries. Other initiatives, such as the e-Citizen portal, do not appear to have dedicated staff. 5.38 As a result of the large project portfolio, few resources were left to establish the foundations for ICT development which would have made those projects more effective, particularly those with a cross-agency component. Key competencies/functions were either not built or, in the case of SIPA, delayed. The MICT has a small, capable and dedicated internal staff, but undertaking complex tasks such as e-government, enterprise architecture or, indeed, any of the 15 projects it has committed to requires a larger lo5DefinitionalMissionReportregardingICT andE-GovernmentProjectsinThailand. Preparedby Global Resourcesfor US Trade andDevelopmentAgency (USTDA). 132 number of middle-management staff with sophisticated policy, project management and coordinationskills.’o6 5.39 MICT has achieved important successes: e SPA has been established andi s now developing its business plan; e The overall paradigmfor ICT development now relies much more heavily on the private sector, rather than government as the driving force; e The ministry has demonstrated the value of a public-private partnerships through the People’s PC program, though which hardware i s now more affordable; e There has been progress intelecommunications reform; and e The ministry has raised awareness of ICT as a tool for development. 5.40 Butinother areas progress hasbeenbelow expectations. Slower progress has been made in: e E-government, perhaps the most important ICT agenda, which remains piecemeal and fragmented; e A governance framework to support cross-agency e-government initiatives (e.g., an Office of Government CIO). Important projects such as the Prime Minister’s Operations’ Center and the Smart Card program, which require not only interoperability but coordinated decision-making across agencies, cannot be effectively delivered; e Project management practices necessary for ICT to actually have an impact on people’s lives. These include change management, performance evaluation, collaboration across agencies and project execution skills. Recommendations 5.41 Inspite of constraints, observers inboththe ICTindustry andother business sectors see strong opportunities: e The cost of hardware software and connections i s becomingmore affordable; e The development of a dynamic and motivatedICT sector, plus the emergence of a younger generation of ICT-aware S M E managers and entrepreneurs; e The potential for rapid innovation in a number of key business segments, with firms being “pushed” by large buyers and e-government, or “pulled” by incentives and business opportunities; e A more proactive government pursuinga range of policies andincentives to promote ICT. 5.42 A number of key interventions might acceleratethe pace by which the private sector can leverage ICT to enhance productivity. lo6 toScale:Establishingeffectivegovernanceof Thailand’se-developmentagenda.PolicyNotefor Going Ministry of ICT. MagdiM.Amin, World Bank, 2004. 133 5.43 Humancapital development. EffectiveICT is closelyrelatedto the quality and quantity of human capital. Both inand outside government, project manager skills are lacking, as well as programmers inkey skill areas. Liberalizingthe hiringof foreign ICT professionals i s a useful first step, but inthe long runthere will needto be a much higher volume of skilled managers. Scaled-up project manager training will involve a combination of technical and managerial capacity building. 5.44 Digital literacy and technical and management skills are key issues for ICT diffusion. The private sector i s addressing some of the gaps. CISCO Systems (the leading global provider of routers and other network technology), has an Internet Engineering Support program which trains ISPs on the use of CISCO equipment. Sun Microsystems and IBMhave affiliate programs with Chulalongkorn University. 5.45 The most important gap from an e-commerce perspective i s the lack of managerial skills. Large firms and SMEs alike need to transform organizations culturally and operationally and to realign internal processes so they can fully exploit the speed and networking possibilities of IT. This i s an area that merits substantial public support inthe form of training, establishment of centers of excellence about specific competencies and demonstration projects. 5.46 Moreover, with English still the dominant language of the Internet, English language and IT skills provision should be fostered in the education system, including vocational training. The role of the private sector in providing IT skills training in countries such as Indiahas been critical in supplyingrelevant skills. Public-private sector collaboration in Thailand i s necessary to identify the specific IT skill needs and private sector-led provisioning of relevant training and education programs. Further, the success of e-government programs will also depend on the ability of civil servants to effectively utilize IT. Therefore, the government will have to continue providing relevant IT training programs for civil servants. Support for ICT training targeted at SMEs i s also important, including training programs that focus on increasing managerial skills in effectively integratingIT into business processes. 5.47 Liberalizing telecom markets is the most important short- to medium-term goal. The issue of scaling-up and mainstreaming technology will ultimately be solved by markets. Markets work best when impediments to competition are removed, and when regulatory certainty, predictability and flexibility coexist. Once markets are liberalized, innovation in targeting subsidies to the achievement of social objectives makes sense. Subsidies pay for specific outputs to pushbeyond what market can provide (per connection, per telecenter, etc.). 5.48 A policy priority is to improve the quality and availability of critical information and communications infrastructure. A key bottleneck to infrastructure development has been limited telecommunications sector reforms. With the establishment of NTC, it i s now important to implement reforms that unleash further competition inthe telecommunications sector. Priorities include: (i) resolving the issues that exist between the state-owned andprivate-sector telecommunication operators by convertingthe current concession contracts into licenses that allow all operators to operate on a level playing 134 field; (ii) completing the planned privatizationof TOT and CAT; (iii) implementing a fair pro-competitive interconnection regime; (iv) allowing for competitive new entry in the international long distance segment and V O P services; and (v) removing licensing requirements for the entry of new ISPs inthe market. 5.49 Government-wide information architecture can help in “liberating” public sector information to be used for public value through ensuring interoperability. The first step in defining architecture — taking stock of where the government i s today technologically, has been completed. The next step i s to leverage industry efforts to categorize electronically-transmitted data over the Internet, so that standard web browsers can replace proprietary hardware or software. Second, Extensible Markup Language (XML) standards have been diffused to such an extent that communications platforms are interoperable. Architecture allows systems serving different functions to exchange information. Standards are critical to Thailand’s success in ICT diffusion, as they allow systems serving different purposes or from different vendors to work together, thereby enhancing network effects and reducing risk. The importance of these network effects creates a special type of public good — an interoperable architecture. To facilitate e- commerce, standards are being defined in such areas as data interchange, electronic payments, security and networking. 5.50 Development of standards. Standards are critical to Thailand’s success in e- commerce, as they allow systems serving different purposes (finance and inventory) or from different vendors to work together, thereby enhancing network effects and reducing risk. The importance of these network effects is that they create a special type of public good — an interoperable architecture. To facilitate e-commerce, standards are being defined in such areas as data interchange, electronic payments, security and networking. Standards development i s currently narrowly focused within NECTEC and the MOC. The private sector i s becoming increasingly active. Inparticular EANThailand promotes adoption of common standards for bar-coding and e-business. The RTG could facilitate interoperability by supporting the development of an industry-led standard-setting body that can dialogue with international standard-setting organizations. The National Institute of Standards and Technology (NIST), for example, i s working with an industry consortium, OASIS to develop, classify and disseminate XML standards. 5.51 Venture Capital. Thai developers have a strong role to play in facilitating e- commerce. The government i s supporting innovation through the establishment of the Software Park and SPA. However, entrepreneurial firms face a host of formal or informal barriers which illustrate the cross-cutting nature of e-commerce development: 0 Under the current Public Company Act, it i s not possible to issue shares without par value or issue treasury stock. The first problem limits the use of equity as an employee incentive because it limits the number of shares a firm issues at founding to the number of shares it can afford to buy at par. Because of the limited number of issued shares (small float), share prices can escalate to the point that they cannot be used as incentives. The inability to authorize shares and retain them as treasury stock means that an option, if granted, would require an open market purchase of the shares and then a transfer to the employee. To further 135 compound the problem, the employee’s reward i s immediately taxable in cash. Furthermore, stock warrant pools are limited to five percent of a company’s float; Entrepreneurs often lack skills requiredto prepare business plans capable of being financed by venture capitalists. Entrepreneurs often try to go from self-financing to raising venture capital without having had sufficient time to further develop their concepts. 5.52 The IT revolution offers significant challenges for Thailand to overcome. Yet the innovations in the private sector, the quality of people in both the government and the private sector working on ICT and the relevance of the technology to the private sector provide reason for optimism. Table 5.4: Policy Matrix Targetedinvestments e-logistics – ICTSkills (SeeChapter3) 136 REFERENCES Becchetti, L.et.al, “Investment, Productivity andEfficiencies: Evidence at FirmLevel Usinga StochasticFrontier Approach.” Centre for InternationalStudies on EconomicGrowth. Working PaperNo. 29. Kluwer, I. U.2002. “Small andMedium Sized FirmsinThailand: RecentTrends.” and Small FirmDynamisminEast Asia.” Tan, H. 2000. Malaysia: Technologyand Skill Needs. World Bank. Timmer, M.P., Ypma G. andvan Ark, B. October 2003. “IT inthe EuropeanUnion: DrivingProductivity Divergence?”ResearchMemorandumGD-67, Groningen Growth andDevelopmentCenter. Triplett, J. E.andBosworth, B.P. September 2003. “Productivity Measurement Issues in Services 1ndustries:”Baumol’s Disease” Has Been Cured” Federal ReserveBank of NewYork EconomicPolicy Review.” 137 Chapter 1 Appendix – A. BusinessClimateIndicatorsfrom the ThailandPICS The analysis in Chapter 1 relies on business climate indicators calculated for Thailand as a whole and on indicators disaggregated across regions, industries and types of firms (small, medium, large, exporters, non-exporters and foreign-owned, domestic firms). The description of the indicators below i s valid for either type of disaggregation of the data. Some of these indicators are shown in Appendix Tables 1A-5D, while others are shown in Chapter l.’07 Since the summary statistics for these business climate indicators do not suffer from serious outlier problems, all firm observations in the PICS are used for calculating the means and standard deviations of the quantitative indicators. Perceptionsof the BusinessClimateby Firms Closedquestion Part I- question V.l: we calculate the number of firms that respond 3 or 4 (severe or very severe obstacles) to each of the items 1-18 in the question. We divide this number of firms by a common denominator to obtain the percentage of firms complaining about each type of obstacle. The common denominator i s the maximum number of non-missing responses among items 1-18. Open-ended question Part I- question V.2: for each of the 22 items on the list (constructed post-survey), we calculate the percentage of firms that pointedit out as either the main obstacle, the second main obstacle or the third main obstacle to doing business inThailand. InfrastructureIndicators Part I- question VI.14 a): for each firm the number of power outages, times with insufficient water supply, interruptions of fixed telephone service and transport disruptions per month i s multiplied by 12 values and averaged across firms to generate the frequency of power outages, insufficient water supply, interruptions of fixed telephone service and transport disruptions in 2002-2003. Part I- question VI.15: the average of this variable indicates the percentage of production lost due to power outages in 2003. Part I- question VI.16: the proportion of firms answering Yes to the question indicates the percentage of firms that own a generator. Part I- question VI.13: the average of this variable indicates the percentage of production that i s lost inshipment due to breakage, theft or spoilage. Part I- question VI.6 1)-3):the average of these variables indicates the number of days neededto obtain telephone, electricity and water connections. lo7For each disaggregation and eachquestion, a “*” beside the figure indicates that the averageresponsein the subcategoryof interest is statistically significantly different from the average for Thailand as a whole. 138 Regulatory and AdministrativeBurden Indicators Part I question VI.9: the average of this variable indicates the percentage of time that – senior management spent dealing with regulations ina typical year. Part I– question VI.7 first column: we calculate a total number of days spent dealing with inspections (summing the number of days spent dealing with visits from revenue department, social security office, immigration division, department of industrial works, local authority) and average it across firms. Part I- question VI.7 second column: we sum the costs of fines from revenue department, social security office, immigration division, department of industrial works, local authority, divide this value by the firm’s sales in 2002 and average the ratio to obtain the value of fines from inspections as a percentage of sales in 2002. Part I- question VII.3.a. first item: the average of this variable indicates the average number of days needed to clear customs for exports by Thai firms. Second item: the average of this variable indicates the longest number of days needed to clear customs for exports by Thai firms. Part I- question VII.4.a. first item: the average of this variable indicates the average number of days needed to clear customs for imports by Thai firms. Second item: the average of this variable indicates the longest number of days needed to clear customs for imports by Thai firms. GovernanceIndicators Part I- question V.6: we calculate the proportion of firms that respond 1 or 2 (fully disagree or disagree in most cases) to the question to obtain the percentage of firms that do not have confidence inthejudiciary. Part I-question V.5.al.: the average of this variable indicates the percentage of payment disputes resolved by firms incourts in 2002-2003. Indicators on the Regulatory Burden to Opena New Businessin2002-2003 Part I- question VI.2.: the averages of cells in the first, third and fifth columns of the table indicates the average number of licenses/permits/approvals or certificates needed to start a business in 2002-2003 by the central government, local government and specific authorities. The averages of cells in the second, fourth and sixth columns indicates the average time to obtain those documents from the central government, local government and specific authorities. In calculating these averages, we restrict the sample to be constituted only by firms that were created in 1993 or after (these younger firms have a better perception of the burden requiredto open a new firm). 139 Reasonsfor Finns Being Over-stufled or Under-stafled Part I- question IV.l.a. and 1V.l.b.: we calculate the percentage of firms that answers Yes to each of the items 1-5 in question IV.1.a. and to each of the items 1-4 of question 1V.l.b. Specific Labor Regulations Part I- question IV.2: we calculate the percentage of firms that respond 3 or 4 (severe or very severe obstacle) to each of the items (a)-(e) inthe question. Time to Get Licenses, Permits or Approvals from Specific Agencies PartI-question VIS.: the average, median and standard deviation of the number of days to get licenses from each of items 1-5 are calculated excluding values of 0. Part I-question VI.6.: the average, median and standard deviation of the number of days to get licenses from each of items 4-6 are calculated excluding values of 0. Part I- question VII.1.: the average, median and standard deviation of the number of days to process the application for each export incentive (items 1-9) are calculated excluding values of 0. Transport logistics Part I- question III.34.: the average of this variable indicates the percentage of production costs that are transportation and logistics. Finance Part I- question VIII.23.: the average of this variable indicates the percentage of annual sales that i s tied inoverdue payments. Part I- question IX.23.: the average of this variable indicates the percentage of new investments that i s financed usinginformal sources. Figures 1-2 show some infrastructure indicators, which are described in the main text of Chapter 1. B. Labor Productivity, Average Wages and CapitalIntensity usingPICS datafor Several Countries In order to obtain a simple measure of competitiveness across countries, we calculate labor productivity, average wages and capital intensity for four major industries (food processing, textiles, clothing and electronics & electrical appliances) in Thailand 140 and several other countries where a PICS was conducted.108For each firm in an industry and country, we obtain VA in local currency units as total sales minus materials and energy costs. This i s calculated using data for the latest year available in the PICS (e.g., for Malaysia it i s 2001, for Thailand it i s 2002). Then, VA i s converted to constant 2001 prices using a manufacturing price deflator obtained from the World Development Indicators for each country. VA at constant 2001 prices i s then converted into U.S. dollars using the average exchange rate between the local currency and the U.S. dollar in 2001 from International Financial Statistics (IMF).We construct labor productivity for each firm in a given industry and country by dividing the VA at 2001 U.S. dollars by the total number of permanent workers inthe firm. We then obtain median labor productivity by industry and country, eliminating from the calculations firms with negative VA and firms with VA inthe top and bottom five percent of the VA distribution for each industry and country. These values are shown in Figures 22 and 23 in Chapter 1.'” The calculations used for average wages are relatively similar. Total wages in local currency units are transformed into constant 2001 prices by using the same manufacturing price deflator and then converted into U.S. dollars using the average exchange rate. These are then divided by the total number of permanent workers to obtain wages per worker in 2001 US.dollars for each firm in an industry and country. Median wages per worker are calculated eliminating the top and bottom five percent of the wage distribution in the industry and country. Finally, capital intensity i s constructed using total assets in 2001 U.S. dollars (obtained in a similar way as VA and wages) divided by the total number of permanent workers. Median wages per worker and median capital intensity for all countries and the four industries are shown inAppendix Figures 3-6. C. Firm-Level PerformanceIndicators Production Data The analysis of firm performance is based on a sub-sample of manufacturing firms from the PICS for which the values of major variables are not considered to be outliers and for which data on all production variables i s available. The original total number of firms in the survey before selection i s 1385. We define an observation (firm) to be an outlier for variable X, if its value i s larger than the mean of X in the corresponding industry by more than three standard deviations of X (in the industry) or if its value i s smaller than the industry mean by more than three standard deviations. The variables for which this outlier rule i s applied are:’looutput-labor ratio; capital-labor ratio; materials-labor ratio; energy-labor ratio; labor share in revenue; materials share in revenue; energy share in revenue; and total intermediates share in revenue.”‘ For the inputrevenue shares, two additional rules are applied: i)all firms with labor or materials revenue shares larger than 1 are classified as outliers; and ii)all firms with materials shares or total intermediates shares smaller than 0.1 are classified as outliers. These ~~_______ ‘OsWe do not provide details on the specific question numbers used to create each variable given the large number of different questionnairesconsidered, however, that information is available uponrequest. log andIndiahadtwo wavesof PICSindifferentyears.Foreachofthosecountries, thedatafromthe China two surveys is combined, once all nominal values are converted into 2001 U.S. dollars. ‘loThe definitionof eachof the variables entering these ratios andof the shares is providedimmediately below. . Note that outliers are identified for the ratios and shares ineach of the two sample years 2001 and 2002. 141 outliers are dropped from the calculation of industry means and standard deviations required to apply the three standard-deviation rule above. Applying these outlier rules, the sample size i s reduced by 193 firms. The production function estimation generates problems for the coefficient on intermediates in two of the industries (garments and machinery & equipment), thus we eliminate some additional outlier firms in those industries by applying a more stringent outlier criterion for materials-labor ratio, energy- labor ratio, materials share in revenue, energy share in revenue, total intermediates share inrevenue of values larger than the mean of X inthe industry by more than two standard deviations of X (in the industry) or values smaller than the industry mean by more than two standard deviations. Applying this additional criterion, our final sample includes 1155 firms in 2002. The main production variables are defined as follows: Output for years 2002 and 2001 i s given by total sales in the table of Part IIA — question IX.13 deflated by a price deflator from the Bank of Thailand at the four- digit ISIC level. Materials for years 2002 and 2001 are given by direct material cost inthe table of Part IIA — question IX.13 deflated by a price deflator for intermediate materials for industry obtained from the Bank of Thailand. Energy for years 2002 and 2001 i s given by the sum of electricity expenditures and fuels and other energy expenditures in the table of Part IIA — question IX.13. Both types of expenditures are deflated by corresponding deflator from the Bank of Thailand. S L for years 2002 and 2001 i s given by the sum of the number of management, professionals, skilled production workers and nonproduction workers in the table of Part IIB– question X.8. UL for years 2002 and 2001 is given by the number of unskilled production workers inthe table of Part IIB — question X.8. Capital stocks for years 2002 and 2001 are given by net book value of machinery &equipment inthe table of PartIIA — question IX.14. Inputrevenue shares usedto determine the outliers are definedas: Labor revenue share: calculated as the sum of total wages and salaries paid to director and officers and the total wages and salaries paid to production workers inthe table of Part IIA — question IX.13 divided by total sales in 2001 and 2002 from inthe table of Part IIA -question IX.13.’13 Materials revenue share: calculated as direct material cost dividedby total sales in 2001 and 2002, both from the table of PartIIA — question IX.13. Whenever one of the subcategories of skilled or unskilled workers is missing we assume that the firm does not employ that type of worker and we replace that missingvalue by a 0. Also, to avoid havingto drop firms that do not employ any unskilled workers from the production function estimation where the inputs enter as logarithms, we consider ULto be log(UL+l). Whenever one of the subcategories of wages of slulled or wages of unslulled workers i s missing we assume that the firm does not employ that type of worker and replace that missing value of wages by a 0. 142 Energy revenue share: calculated as electricity plus fuels and other energy divided by the total sales in 2001 and 2002, both from the table of Part IIA — question IX.13. Total intermediates share: calculated as direct material cost plus electricity plus fuels and other energy divided by total sales in 2001 and 2002, both from the table of PartIIA – question IX.13. Measures of Performance The three performance measures used in the analysis are SG, value- added per worker andTFP. The first two measures are defined as follows: i) SGisthelogarithmicchangeinsalesbetween2001and2002; ii)Value-added per worker (VAL)istheratioof VA (defined as sales minusdirect material cost minus electricity minus fuels and other energy) to total employment defined as the sum of skilled and unskilled workers; iii)TheTFPmeasuresrelyonthefollowingmodel. For firm iinindustryj, the technology i s described by a Cobb-Douglas production function with Hicks-neutral technical change (in logarithmic form): where output q i s produced combining the capital stock k, total intermediates (raw materials plus energy) me, S L s and UL u. v i s residual firm productivity and can be decomposed as follows: (D2) V: = W: +E: where w i s a component of firm productivity that i s known to the firm manager and possibly affects input choices but i s unknown to the econometrician and E i s a random shock to outputlproductivity that i s realized after input choices are made (therefore it i s not correlated with input choice^).”^ We obtain estimates of the parameters(a, p,y,4)estimating Equation (Dl) separately across industries using ordinary least squares (OLS) techniques. The OLS production function parameters are presented in Appendix Table 6.’15 Since firms that are more productive will hire more skilled and UL and use more intermediates in order to produce more, OLS estimates of the corresponding production function parameters that do not account for such endogeneity are biased. As an alternative to OLS estimation, we follow Levinsohn and Petrin (2003) to obtain production function parameters that correct for the endogeneity of input choices. The basic specification is given by Equations (Dl) and (D2). The main idea i s that firms choose their variable inputs (SL, ULand intermediates) with knowledge of their productivity w whereas capital i s a quasi-fixedinput costly to adjust. 114More specifically, the conditional expected values verify the following equalities: / si1]=O, / 14J=0,E[&it/ m+,]=O andE[&it/ kiI]=O. ’15Note that for production function estimation, the data for 2001 and 2002 for all firms in the industry i s used (as long as the firms are not outliers and have no missing production variables). However, in the analysis of correlates of performance across Thai firms, we focus on the measures of performance in 2002 only. 143 The demand for variable inputsby a firmcan be derivedfrom profit maximization and depends on capital k and on productivityw. In particular, the demand for intermediates i s given by mei, = me(w, ,kit),116,117 Under fairly general technical conditions (described in detail inLevinsohn and Petrin (2003)), it i s possible to invert the intermediates demand function and express the unobserved (to the econometrician) firm productivity as a function of two variables (intermediates andcapital) that are observable: @, =w h , ,7k, 1 The crucial idea behind this estimation method i s to use this proxy for productivity to control for the endogeneity of input choices with respect to productivity. The estimation proceeds intwo stages. Stage 1: Replacing the proxy equation for productivity into Equation (D2) and this equation itself into Equation (Dl)we obtain: or (D4) qit = y e si, + 4 uit h(me,, kit)+ ci, – + where ,h(me,, k, ) = a.mei, +p uj, w(me,, k, ) + The function h(.) groups all the terms on intermediates and capital. Its functional form i s unknown but we can approximate it using a polynomial in intermediates and capital. Adding the terms of a fourth-degree polynomial in me and k to substitute for h(.) inEquation @4), it can be estimated by OLS to provide consistent parameter estimates for S L and UL. Note that one also obtains an estimate for the unknown function h(mei,, kit)that is usedinStage2. Stage 2: An additional assumption is required to obtain the coefficients on intermediates and on capital. We assume that productivity i s serially correlated: it follows a Markov process wit = E[wit / wit-,] +vi,.Also, the mainidentification assumption in Stage 2 i s that capital i s slow to adjust: it may adjust to the expected part of productivity conditional on lagged productivity E[wi, / wit-,],but it does not adjust to the unexpected shock vi,.So, one can derive the following moment condition that identifies the coefficient on capital: 118 (D7) E[vit + /kit]=0 Notethat for simplicity of notation, we drop the superscriptj inthe rest of the section, but the estimation is performed for each industry j separately. Ideally, the function me(.) should vary across years or periods (Le., me,(u),,,k,,) to account e.g., for changes in the price of materials, other inputs or output. However, since we only use two years of data we consider a function me(.) that does not vary across years. E[$,,+ Elt/ klt]=E[qlt / kit] +E[clt / kit].The / k,,]=0. T h e definition ofEltimplies that EIElr / klt] = 0. So E[q1t+ Elt / klt]= 0. mainidentificationassumption says that E[qlt 144 A separate moment condition is needed to identify the coefficient on intermediates. The condition for intermediates exactly parallel to that for capital cannot be used since intermediates are correlated with productivity (both the expected and unexpected parts): E[vit+ /mi,]# 0.11′ But one can derive a moment condition for lagged intermediates. Intermediates in year t-1 are chosen by the firm without knowledge of the productivity shocks realizedonly in year t, so: The residual vi,+gi, in the moment conditions is obtained replacing productivityuitby its Markov process and switching sides for some of the terms in Equation (Dl): Note that in Equation (D9), we replaced the coefficients on skilled and UL by their estimated values from Stage 1, and we included an estimate for the expected value of productivity conditional on lagged productivity. This conditional expected value i s a function of wit-l. We call it g(w ) but its functional form i s unknown. We approximate It-] this g(.) function by locally weighted least squares of an estimate of wit on an estimate OfUit-l 120 * Sample analogs for the two moment conditions in Equations (D8) and (D9) are obtained for all firms and a general method of moments (GMM) criterion function i s constructed. The estimates for the coefficients on intermediates and capital are those making the sample analogs of the moment conditions as close to 0 as possible:Iz1 An iterative procedure needs to be used to minimize this function starting from some candidate initial parameters a and p (e.g, those obtained by OLS). The production 119Although / me,,] i s 0, E[+,, me,] / i s different from 0. Specifically, the estimate for the expected productivity conditional on lagged productivity is given by a LWLS regression of an for Wit given by (w,,4 Eit) = yir -f s,, -4 * uir-a*.me, -pestimate A *..k,, on an estimate for wit-l given by &if-l =h(me,,_,,kit-l)-a*.meit-l -p *k,l-l. Note that both these estimates depend on the A parametersof interest inStage 2: a and p. 12’The sample analogs for the moment conditions are summed across firms i (the index i in the first summation symbol). For each firm the moment conditions are summed from the second year of available data on since the procedure uses lagged inputs (the index t in the second summation symbol). No moment condition can be computed in the first year of available data for a firm. 145 function parameters obtained by this estimation method are presented in Appendix Table 7. The OLS coefficients on variable inputs are expected to be upward biased since they do not control for endogeneity. Generally, the coefficients on most inputs in Appendix Table 7 are lower than those inAppendix Table 6.'” As mentioned in the main text of Chapter 1, we also obtain a different set of production function parameter estimates based on a sample of about 230 firms that are part of the PICS but were also matchedto a longer panel dataset from MOIand for which we have data for 1999-2002. For this sample of firms, the production function estimation follows the same techniques as presented above, with two differences: (i)estimation i s performed separately across two broadly-defined industries: high-tech industry (auto- parts, electronics & electrical appliances, rubber & plastics and machinery & equipment) and low-tech industry (food, textiles, garments and wood & furniture); and (ii) the labor input is total ernpl~yment.”~The corresponding production function parameter estimates are shown inAppendix Table 8. Since TFP measures obtained by OLS are likely to be biased, we rely in Chapter 1 on the analysis of TFP measures obtained by the Levinsohn and Petrin (2003) method. These TFP measures are obtained as the residual from Equation @1) i.e., the difference between output and inputs weighted by the estimated production function parameters. Dataon Correlates of Performance In the regressions analyzing the determinants of firm performance, several firm characteristics are used: Region dummy variables are constructed based on the region code indicated for each firm as its location (six regions: North; Central; Bangkok; East; Northeast; and South.) Industry dummy variables are constructed based on the PICS industry classification which i s approximately the two-digit ISIC classification (eight industries: food processing; textiles; garments; auto-parts; electronics & electrical appliances; rubber & plastics; wood products & furniture; and machinery & equipment). Age is defined as 2004 minus the responseto PartI– question 1.1. Size i s measured by the logarithm of firm total employment (the sum of skilled and unskilled employment). lZ2Although one expects an upward bias in the OLS coefficients on skilled and UL and intermediates, theoretically, one can expect either an upward or downward bias in the OLS capital coefficient (see Levinsohnand Petrin (2003)). lZ3The estimation is performed separately for these two broadly-defined industries since the number of observations is not sufficient to estimate a separate production function for each of the eight industries considered inthe estimation that uses the entire sample inthe PICS. 146 Exporter dummy i s defined to be equal to 1if in Part IIA — question VIII.9 the percentages given for sales exported directly and sales exported indirectly in2002 sumto more than 10percent of total sales. FDIdummy is definedto be equal to 1if inPart I– question 1.4.1the percentage of the firm owned by the private sector foreign i s any positive number. Technology and innovationvariables o computer-controlled machinery is the percentage given in Part I– question 111.7; o vintage of capital is the percentage of machinery & equipment of the firm that i s less than five years old given inPart I– question III.6; o R&D dummy is defined to be equal to 1 if the firm has positive expenditure on design and R&D inPart IIA — question IX.10. Determinants of Firm Per$ormance The regression specification usedto analyze the determinants of firm performance i s the following: (D11) p; = 8, dl age, + +d2 size, +d3 exp, +d4.FDZ, +d5.comp, +d6. Kvint ,+d6 .RD, + Z j +Ir+& where piis a measure of firm performance in 2002 and the determinants of firm performance are firm age age, , firm size size, , firm export status exp, , foreign ownership of the firm FDI,, three measures of technology use or innovation comp, ,Kvinti , RD, , I’are two-digit industry effects and I’are region effects. The regressions are estimated by OLS and standard errors that are robust to heteroskedasticity are computed. As an alternative to dummy variables identifying exporters and foreign- owned firms, we estimate a regression similar to (D11) but including export intensity and foreign ownership share as determinants of firm performance. The results are shown in Appendix Table 10. As mentioned in the main text of Chapter 1, we also estimate partial regressions that include only one firm characteristic at a time, together with industry and region fixed effects, to analyze the individual effect of each performance correlate. For example, when firmsize is the only characteristic included, the partialregressionis as follows: (D12) p i = do /z sizei +I’ v: + tI’ + Commentson the Analysis of Determinants of Firm Performance Although we do not report the results, we also estimated equation (D11) using TFP obtained from the OLS parameter estimates as a measure of firm performance. The results are generally close to those reported in the main text obtained for TFP using the parameters from Levinsohn and Petrin(2003) estimation. 147 InAppendix Table 9, we show the correlationamong the firm characteristics included inthe regressions of TFP, SG and VA for the sample of 1033PICS firms used inthe estimation. Firm size i s significantly positively correlated with the exporter dummy, the foreign ownership dummy, the percentage of computer-controlled machinery and the R&D dummy. Business Climate and Firm Performance: Costs of Skill Shortages The framework of production function and profit maximizationby firms can be used to analyze the problem of shortage of skilled workers. A firm makes its input choices in order to maximize profits as follows: where Q=AKaMEBSYUrepresents output produced according to the production function whose estimation we discussed before, P i s the price at which the firm sells its output, r i s a rental rate paid on capital, PMEi s the price of intermediates, ws i s the wage paidto a skilled worker, and wu i s the wage paid to an unskilledworker. Ifthefirmfacesnoconstraints, ithiresskilledworkersuntilthefirstorder optimization condition i s ~atisfied.’~~ S i s chosen such that: (D14) Marginal Product of S = y – P . Q I S = w s If firmisconstrained,thatisitcannothireasmanyskilledworkersasdesired, the this optimization condition is not verified. For the (smaller than desired) number of skilled workers that i s feasible for the firm to hire the condition i s infact: (D15) Marginal Product of S= y.P.Q/S > w s The difference between the marginal product of skilled workers and the corresponding wage represents the cost (in terms of outputhales lost) to the firm of not being able to hire one additional skilled worker. Alternatively, one can think of this difference as the benefit from reducingthe skill shortages. The optimal skill mix for an industry i s derived from profit maximization, combining Equation (D14) with the corresponding first order condition for UL:125 (D16) (-)*= s +S u Y*WU yaw, + 4 ‘ W S 124This occurs when the derivative of the profit function with respect to S equals 0. The interpretation is that the firm hires more skilled workers as long as the marginalproduct of each additional skilled worker is larger than or equal to the wage the firm needs to pay hidher. 125That first order condition is given by: Marginal Product of U= 4 P- Q/ u= w . * 148 Replacing y and 4 by the values from the production function estimation the optimal skill mix can be computed for each industry. We analyze the effect of the following experiment. Suppose that skill shortages are reduced (e.g., due to a large number of college graduates entering the labor market) and constrained firms can hire more skilled workers inorder to approach the optimal skill mix in their industry. The firms that are constrained in terms of the number of skilled workers that they can hire are defined as those having an actual skill mix -smaller s +S u than the optimal skill mix intheir industry. For each constrained firm, the benefit interms of increased sales from increasing its number of shlled workers to get closer to the optimal skill mix i s given by: where the first term inbrackets in the product represents the additional benefit per skilled worker and the second term in brackets represents the increase in the number of skilled workers that makes the skill mix of the firm become equal to the optimal skill mix in its industry. Using the value of y estimated in the production function and using information in the PICS on wages paid to skilled and unskilled workers, Equation (D17) can be calculated for each constrained firm. We sum the product in Equation (D17) for all constrained firms in each industry and divide that sum by the sum of sales of those constrained firms in each industry, call this ratio To obtain the final potential benefits in terms of increased sales from reducing skills shortages, we divide the ratio A by two. The rationale for this final step i s as follows. The graph below represents the marginal product of S L as a function of the number of skilled workers. For simplicity it i s assumed to be a linear function. Also shown in the graph i s an horizontal line representing the skilled wage WS. 126Equation (D17) does take negative values for some of the constrained firms. We disregard these negative values and focus on constrained firms for which the benefit of reducing the skill shortages are positive. 149 Mg- Prod. of s MPsEw WS S S”” S We denote by S the current number of skilled workers of a constrained firm and by Snewthe number of slulled workers that the firm can hire under the experiment described before. The benefit to the firm from this reduction in skill shortages (the product in Equation (D17)) i s given by the area of the square abcd (this represents additional output). We are able to calculate the area of this square for each firm and then sum across firms in an industry. However, the benefit to the firm in terms of additional output from a reduction in skill shortages i s given only by the shaded triangle bad. If the marginal product of S L was truly linear, the area of bad would be equal to half the area of abcd. Therefore, as an approximation, we assume that the marginal product is linear and we divide the product in Equation (D17) calculated for each firm by two to give the benefit from relaxing the skill constraint. InAppendix Table 11, we show the actual skill mix in each industry covered by the PICS and the optimal skill mix calculated following Equation (D16) as well as average wages of slulled workers and unskilledworkers. “Macro” Benefitsfrom IncreasingSupply of Skilled Workers Some potential “macro” benefits could come from the relative expansion of sectors that employ skilled workers more intensively in Thailand. Thailand trades large volumes with the rest of the world, thus, increases in the fraction of skilled workers could encourage production and exports in sectors using skilled workers intensively. This shift to skill-intensive industries would likely be accompanied by an increase in average wages, as more and more workers would need to earn higher wages in line with their improved skills. This potential increase in average wages can serve as a measure of the potential “macro” benefits of increasing the supply of skills inThailand. The size of these “macro” benefits in terms of higher wages would depend on: (i) the fraction of skilled workers in the total workforce; (ii) the current premium of skilled worker wages relative to unskilled worker wages; and (iii) the extent to which this skilled premium falls as the overall supply of skills in the workforce increases. We obtain (i)and (ii) from the PICS.’27 In manufacturing, on average, 33 percent of the workforce i s skilled (i.e. managers, professionals, skilled production workers and non-production workers). The 150 PICS statistics in Appendix Table 11suggest that skilled worker wages are roughly 2.5 times unskilled worker wages. Item (iii) i s more difficult to quantify precisely, but the experience of other East Asian economies analyzed in Hsieh (2002) suggests that the decline in the wages of shlled workers relative to unskilled workers has been modest, even as the supply of skilled workers has increased dramatically. In Singapore, for example, there were about 50 workers with no university education for every university- educated worker in the mid-1960s. B y the early 199Os, this ratio had fallen to 1O:l. Despite this dramatic increase in the education of the workforce, the relative wage of university-educated workers fell by only 20 percent. Appendix Table 12 provides some simple estimates of the possible order of magnitude of these “macro” benefits of skill upgrading. The first column shows a range of possible values for the elasticity of the skill premium with respect to the relative supply of skilled workers, i.e., the percent change in the ratio of skilled worker wages to unskilled worker wages that accompany a one percent increase in the ratio of skilled workers to unskilled workers. The range for these values i s taken from Hsieh (2002) who focused on the experience of other East Asian countries and found the elasticity of the skill premium with respect to the relative supply of skilled workers to be generally small, ranging from about -10 percent in Taiwan (China) and Hong Kong SAR to about -20 percent in Singapore. That elasticity was much larger, however, in Korea at -50 percent. The second column shows the percent increase in average wages inmanufacturingdue to a 20 percent increase in the share of skilled workers in the total workforce.’28If the skill premium i s not affected by changes inthe supply of skilled workers, average wages could increase by seven percent. This value reflects a purely compositional effect derived from the fact that there would be more skilled workers earning higher skilled wages. However, it is likely that as a result of the increase in the supply of skilled workers, the skill premium would decline. The other rows of the second column of Appendix Table 12 show the estimated effect on average wages for different values of this elasticity. For moderate values of this elasticity, the increase in average wages falls to six or four percent, and it falls to one percent if the elasticity i s equal to the high value observed in Korea. 12′ The values provided in the secondcolumn of Appendix Table 12 are basedon the following calculation. Let ws and wu denote skilled worker wages and unskilled worker wages, and let ys be the fraction of the workforce that i s skilled. Average wages are therefore w = wsys – +wu -Ys) (1 . The elasticity of -w with respect to ys is — a i Ys – — Ys , where E is the aYs w Ys (ws /wu 1+ (1- Ys 1 elasticity of the skill premium with respect to ys ,and we assume that the unskilled wage does not change with the increasein skills supply. 151 APPENDIX TABLE 1A: GeneralConstraintsto Operation(ClosedQuestion) – InternationalComparisons Firmsevaluatingconstraintas “major” or “very severe” (%) Elecinclty 25 6 14.8 22.3 334 28 I 28.9 20.3 17.3 10.1 4.6 0.5 TRosponption 13.8 11.6 16.4 18.3 19.4 12.4 19.3 8.4 4.2 3.8 0.0 Access to Land 25 8 1 130 148 163 9 1 197 60 38 98 37 Tax Adminisiniion 22 3 12.4 23.0 25.1 23 7 26 3 659 33.1 4.5 31.8 5 9 ~ ~ M d T n d c R e . ~ ~ 198 146 15.7 21.7 21.1 12.8 325 8.9 3.8 11.6 05 LaborRCguldtiOns II4 14.3 259 24.7 194 16.6 567 87 4.2 3 3 2 1 Skillr andEduwionofAvailabkWorkem 30.0 24.7 18.9 11.9 26.7 12.4 39.6 12.8 23.8 9.9 4.3 Business Licensing and Operating Pernuis 7 4 109 20.5 13.5 15.9 13.4 29.7 5.8 112 14.6 32 A c ~ I o ~ ~ t l c ~ I 13.6 13.9 17.5 13.5 24.1 18.1 59.5 17.3 12.1 20.3 8.2 hcccss to ForeignCredit IS 7 6 28.5 COS orplnanci 14.5 17.0 48.2 23.0 21.6 20.1 82.3 28.2 5.5 13.0 U.4 EconomicPolicy Uncenaini) 29.1 209 50.1 29.5 280 20.6 75.8 538 12.0 315 1 1 8 ~ w m ilnuabilityc 37.4 24.1 41.5 38.4 26.0 15.7 74.7 53.7 8.9 28.5 10.1 Cormpiion 18.3 124 22.0 35 2 22 4 37.3 66.9 23 7 5 4 13.7 6 1 Cti~ne,TkftandDimrda 10.3 10.6 17.3 265 15.7 15.6 520 129 6.5 12.4 3.3 Anii-Competitie or M o m l prdcticcs 20.1 118 17 3 21.3 17.6 17 3 56 I 227 15.8 I4 8 8.0 Numbcrof Plants 1385 893 711 647 973 1730 1636 514 170 506 188 152 APPENDIX TABLE 1B: General Constraintsto Operation(Closed Question) -ByRegion Firmsevaluatingconstraintas “major”or “verysevere”(96) Business Licensing and Operating Permits 1.4 11.0 2.7 * 8.2 6.3 5.6 8.4 Access to Domestic Credit 13.6 12.3 8.1 * 14.7 13.0 16.9 12.1 Anti-Competitive or InformalPractices 20.1 24.1 11.5 * 24.3* 14.6* 11.3* 14.0 Numberof Plants 1385 73 148 194 192 11 107 153 APPENDIX TABLE 1C:General Constraints to Operation(Closed Question) -ByType of Firm Telecommunications 11.4 7.2 * 11.0 14.6 * 13.3 9.3 11.0 12.5 Electricity 25.6 18.2 27.3 28.7 26.7 24.2 24.6 28.5 Transportation 13.8 9.5 * 13.8 16.7 15.1 12.3 12.7 17.2 Access to Land 2.5 2.0 3.1 2.3 3.2 1.8 2.5 2.6 Tax Rates 24.4 23.1 25.4 24.3 25.2 23.5 24.4 24.4 Tax Administration 22.3 21.3 23.4 21.8 24.7 19.7 22.1 23.0 CustomsandTrade Regulations 19.8 12.7* 18.8 25.5 * 25.1 * 13.9* 18.1 25.0* Labor Regulations 11 4 10.7 11.4 11.9 12 2 10.5 12.0 9.6 SkillsandEducationof AvailableWorkers 30.0 29.7 30.8 29.3 31.6 28.2 29.9 30.2 BusinessLicensingandOperatingPermits 7.4 6.1 7.9 7.7 7.5 7.2 7.3 7.6 Acccss toDornesticCredit 13.6 18.2* 14.1 10.0* 11.9 15.4 15.0 9.3 * Access to ForeignCredit 1.5 1.2 I.o 2.3 I.8 1.2 1.4 1.7 cost of Financing 14.5 18.2 15.9 10.7 * 12-8 16.5 15.6 11.3 EconomcPolicy Uncertamty 29.1 32.3 29.5 26.6 27.4 30.9 30.5 25.0 Corruption 18.3 20.2 18.4 169 18.0 18.6 18.4 17.1 Numberof Plants 1385 347 516 522 729 656 1041 344 APPENDIX TABLE 1D: General Constraints to Operation(ClosedQuestion) -By Electricity 25 6 29 1 22 6 185 * 310 27 1 356 * 208 158 * Number of Plants 1385 179 186 168 145 166 239 125 177 154 APPENDIX TABLE 2A: GeneralConstraintsto Operation(Open-EndedQuestion)-By Region FirmsthatpointedalternativesasoneoftopthreeobstaclestodoingbusinessinThailand(8) Skilled Labor Shortage 47.9 46.6 49.3 48.6 49.5 53.5 35.5* Tax Regulations and/or HighTaxes 27.7 21.9 28.4 30.5 27.1 12.7 * 21.5 BureaucraticBurden 22.7 24.7 26.4 22.2 26.0 9.9 * 23.4 29.1 * 17.2 5.6 * 3.7* InadequateSupply of Infrastructure 11.9 13.7 16.2 7.9 * 16.7 26.8 * 15.9 Labor Regulations 9.5 13.7 9.5 9.4 7.3 9.9 10.3 APPENDIX TABLE 2B: GeneralConstraintsto Operation(Open-EndedQuestion) By – Skilled LaborShortage 41.9 48.4 48.1 47.5 46.9 49.1 48.3 46.8 Tax Regulationsand/or HighTaxes 27.7 28.8 29.7 25.1 27.4 28.0 26.6 31.1 Bureaucratic Burden 22.7 18.4 * 22.1 26.2 25.7 19.5* 20.9 28.2* Competitionfrom Imports 22.1 27.1 * 22.5 18.4* 18.2 * 26.4* 24.7 14.2* Utility Pnces 149 13.5 15.9 14.9 12.8 17.4 14.0 17.7 InadequateSupply of Infrastructure 11.9 7.8 * 12.6 14.0 12.5 11.3 10.6 16.0* Labor Regulations 9.5 6.9 8.3 12.3 11.8 6.9 * 9.1 10.5 Official Corruption 7.9 6.9 10.1 6.5 6.0* 10.1 8.5 6.1 155 APPENDIX TABLE 2C: General Constraintsto Operation (Open-EndedQuestion) -By Industry Firmsthat pointedalternatives asoneoftopthree obstaclesto doingbusinessinThailand(S) SkilledLabor Shonage 41.9 41.5 52.2 66.1 42.1 46.4 42.1 49.6 39.0 Tax Regulations and/or mghTaxes 27.7 190 23.7 214 37.2 31.9 25 9 25.6 39.0 Competitionfrom Imports 22.1 5.0 40.9 23.2 24.8 19.3 11.7 13.6 39.0 Uhllty Prices 14.9 15 6 19.9 5.4 14.5 15.1 24.3 88 10.2 Labor Regulations 9.5 12.3 6.5 155 9.0 9.6 8.8 12.8 2.8 HighInterestRates 8.1 6.7 5.4 5.4 11.7 4.8 15.5 6.4 6.2 Official Corruption 79 8.9 9 1 77 5.5 7 2 117 2.4 7.9 NumberofPlants 1385 179 186 168 145 166 . 239 125 117 APPENDIX TABLE 3A: Infrastructure Indicators International Comparisons – -. Froquencyofpoweroutager(timsolastyear) 16.8 9.1 4.4 6.0 5.0 201.0 4.6 2.7 3.8 5.6 2.0 Roduftioolost inshipment (%) 0.5 0.1 23 na. 1.2 n.a 1.0 na n.a n.a. n.e No ofddystoobtamatelephoneconneclion 22 3 8 8 26.6 13.2 12.5 63 3 18.2 I 5 2 4 144 7.8 No.ofdaystoobtainaneleetricitycormecticm 26.4 11.0 14.6 8.2 18.2 72.1 25.6 0.9 1.8 5.9 33 No of days IOobtam a water connccuon 221 9.8 131 na. n.a na. 16.0 n a na. n.a. n.a. APPENDIX TABLE 3B: Infrastructure Indicators-By Region Frequencyof poweroutages(times lastyear) 16.8 21.o 15.5 13.6* 19.9 18.3 34.8 * Production lost due to power outages (7i) 1.5 0.9 * 1.4 1.6 1.9 0.9* 1.5 Haveowngenerator (%) 16.0 42.5 * 23.6′ 7.9′ 12.5 18.3 51.4* Productionlostinshipment(96) 0.5 0.6 0.3 * 0.6 0.4 0.5 0.5 No.of days to obtain a water connection 22.1 33.5 18.8 25.1 14.5* 13.4* 18.5 156 APPENDIX TABLE 3C: Infrastructure Indicators -By Type of Firm Frequencyof poweroutages (times last year) 16.8 13.7* 18.7 17.0 17.2 16.4 16.9 16.5 Production lostinshipment(46) 0.5 0.7 0.6 0.4 * 0.5 0.6 0.6 0.4 * No. of days to obtain a telephone connection 22.3 15.5 28.7 20.3 21.2 23.7 22.8 21.1 No. of days to obtain a water connection 22.7 20.8 23.1 23.3 21.9 23.4 22.1 24.4 APPENDIX TABLE 3D: Infrastructure Indicators -By Industry Production lost due to power outages (%) 1.5 10 * 1.4 1.3 1.7 1.5 2 0 * 1.2 2.0 Reductiontostinshipmnt (46) 0.5 0.4″ 0.4 0.2 * 0.4 0.5 0.8 1.0 0.7 No. of days to obtaina water connection 22.7 29.6 27.8 14.3* 24.9 12.6 * 24.7 26.6 21.9 APPENDIX TABLE 4A: Regulatory Burden and Administrative Delays International – Comparisons … (a) Nwnberof day&spentininspectionsor mtingswith offlfialsfrmndifferent 4.4 12.1 1.0 7.4 30.9 6.3 13.7 aa ILB n.a n . ~ sgeocies (b) Total cost (as % of sales) associated with inspections or meetings with officials 1.1 1.1 0.1 0.0 0.0 n.a 0.5 n.a n.a n.a n.8. from different agencies Impom: Avg uumbcrof d a y 10 clear customs 4 6 3 1 5 8 9 1 7 5 7 1 138 37 1 6 6 9 3 1 Longc~tnumberofdaysmclesrcuswms 9.7 7.0 11.5 16.6 12.2 11.5 32.4 5.6 4.2 13.6 6.9 Exuorts: Avg numberof days10 ckar cystoms 1.5 2.1 4.1 6.6 5.5 5.3 8.4 1.9 1.7 8.5 2.0 Longestnumbcrofda)sIo~learcusr~~s 3 2 4 3 7 5 107 8 1 7 9 169 3 0 57 166 4 2 157 APPENDIX TABLE 4B: Regulatory Burdenand Administrative Delays By Region – Senior management’stime spent dealing with’regulations (96) 1.8 1.3 1.0* 1.9 2.1 2.6 1.9 Inspections: (a) Numberof days spent ininspectionsor meetingswith officials fromRevenue Dept, Social Security Office, Immigration 4.4 3.9 4.9 4.0 4.2 5.5 6.3* Division, Dept. of IndustrialWorks, Local Authorities (b)Total cost (as % of sales) associated with inspections or meetings with officials 1.1 2.0 0.4 1.5 0.1 * 1.6 0.3* fromRevenueDept., Social Security Office, I m g r a t i o nDivision, Dept. of Industnal Works, Local Authorities Avg. number of days to clear customs Exports: Longest number of days to clear customs 3.2 3.0 2.7* 3.5 3.9 2.5 2.0* APPENDIX TABLE 4C: Regulatory Burden and Administrative Delays-By Type of Firm Seniormanagement’s time spent dealing with regulations (5%) 1.8 2.0 2.0 1.5 1.6 2.0 1.7 2.3* Insvectwns: (a) Nm’ of days spent ininspectionsor meetingswith officials from Revenue Dept-, Social Security Office, Immigration 4:4 3.3* 4.1 5.4 * 5.2* 3.4* 4.2 4.9* Division, Dept. of IndustrialWorks, Local Authorities (b)Totalcost (as %of sales) associated with inspectionsor meetings with officials 1.1 3.4 0.1 * 0.4* 2.0 1.5 0.1* from RevenueDept., Social Security 0.6 * Office, ImmigrationDivision, Dept. of IndustrialWorks, LocalAuthorities Exports: Longestnumberof days to clear customs 3.2 2.9 3.0 3.4 3.2 3.4 3.2 3.3 158 APPENDIX TABLE 4D: Regulatory Burdenand Administrative Delays By Industry – .- 3 8 dE Senior managemnt’s time spent dealing with ngulatioos(%) 1.8 0.6* 1.1 . 1.3 2 2 3.2* 2.1 0.6′ 3.1 * Insnectinns: – -r– – – (a) Numlmof days spent ininspectionsor meetingswith officials fromRevenue apt., socialsecurity office,tmmigrati~n 4.4 6.8* 4.3 5.2 3.2* 4.1 3.8 * 4.7 3.0 * Division, Dept of IndustrialWorks, Locat Authorities @) Total cost (as % of sales) associated with inspectionsor meetings with officials 1.1 0.3* 0.0* 3.4 0.5 0.0 * 0.6 1.9 2.6 fromRevenueDept., Social Security Office. ImmierationDivision. Deot. of IndusmialWorks, Local Authonties Exports: Avg. numberof days toclear customs 1.5 1.6 1.5 1.5 1.7 1.3* 1.4 1.3 i.8 Longestnumberof days to clear customs 7 2 3.0 3.4 3.2 3.4 3.2 2.4* 3.6 4.0 * APPENDIX TABLE 5A: Governance International Comparisons – bonfideaceinrhcjudiciary (96 disagree) 12.5 12.0 21.0 16.5 7.4 15.5 22.5 33.1 28.6 65.3 45.6 Percentof payment disputes resolvedin the courts 22.3 20.1 1.0 2.2 5.4 n.a 1.2 35.5 19.9 11.9 17.2 APPENDIX TABLE 5B: Governance By Region – Confidence in thejudiciary (% disagree) 12.5 15.1 5.4 14.1 9.4 9.9 15.9 Percent of payment disputes resolvedin the courts 22.3 n.a. 27.5 21.7 2.0 ‘ 32.7 0.2 * APPENDIX TABLE5C: Governance ByType of Firm – a b .Y 9E !i t ! l a W B zw Percent of payment disputes resolved inthe courts 22.3 23.7 22.4 21.4 24.9 21.0 23.2 15.2 159 APPENDIX TABLE5D: Governance By Industry – Percentof payment disputes . . resolved inthe courts 22.3 34.3 28.0 18.9 4.5* 1.2* 18.4 38.5 11.2 APPENDIX FIGURE 1: PanelA. Number of Daysto Obtaina Water Connection PanelB.Frequencyof InsufficientWater Supply. PanelA PanelB NumberofDays to Obtaina Water Connection FrequencyofInsufEcient Water Supply LastYear Philipines Thailand Malaysia Brazil Thailand ndonesia Indonesia Malaysia Brazil 0 5 10 15 20 25 0 2 4 6 8 1 0 1 2 APPENDIX FIGURE2: PanelA. Percentageof Water Use ComingfromOwn Sources. PanelB.Frequencyof TransportDisruptions. PanelA PanelB PercentageofWater Coming hornOwnSources FrequencyoflrausportDisruptionsLast Year I Malaysia ndonesia Thailand [ndonesia ‘hilipine s Thailand Brazil Philipines Malaysia Brazil 0 10 20 30 40 50 60 70 0 0.5 1 1.5 2 2.5 3 160 APPENDIX FIGURE3: Median Wages per Worker in2001U.S. Dollars inFood Processing and Textiles Industries. APPENDIX FIGURE4: MedianWages per Worker in2001U.S. Dollars inGarments and ElectronicdElectrical Appliances Industries. Waws DCI Worker. Garmcnb –I APPENDIX FIGURE5: Median CapitalIntensity in2001U.S. Dollars inFoodProcessing and Textiles Industries 161 APPENDIXFIGURE6: MedianCapitalIntensity in2001U.S. DollarsinGarmentsand ElectronicsDIlectricalAppliancesIndustries. C.plbl lnlrnrlty. G.rmrnb Capital Iotewity EkeeooMlechkdAppbom . 4woT- 1 1-1 APPENDIX TABLE 6: ProductionFunctionParametersEstimatedby OLS Industry N. Coeff. Coeff. Coeff. COeff. Returns to Obs. Skilled Unskilled Intermediates Capital Scale Labor Labor FoodProcessing 299 0.100 *** 0.054 *** 0.794 *** 0.050 *** 0.998 (0.021) (0.011) (0.025) (0.017) Textiles 324 0.149 *** 0.055 *** 0.773 *** 0.042 *** 1.020 (0.018) (0.012) (0.020) (0.008) Clothing 235 0.102 *** 0.048 *** 0.769 *** 0.035 *** 0.954 (0.019) (0.012) (0.019) (0.013) Auto-parts 219 0.136 *** 0.049 *** 0.777 *** 0.057 *** 1.019 (0.029) (0.014) (0.028) (0.017) Electronics and Electrical Appliances 277 0.080 *** 0.094 *** 0.796 *** 0.026 ** 0.996 (0.020) (0.017) (0.017) (0.012) Rubber and Plastics 358 0.158 *** 0.038 0.831 *** 0.026 *** 1.053 (0.031) (0.024) (0.012) (0.009) Wood Products and Furniture 206 0.234 *** 0.101 *** 0.758 *** 0.012 1.105 (0.024) (0.015) (0.024) (0.011) Machinery and Equipment 245 0.129 *** 0.040 *** 0.799 *** 0.033 *** 1.001 (0.024) (0.013) (0.017) (0.007) Notes: The first column shows the number of firm-year observations included in the estimation. Robust standard errors are in parentheses. ***, ** and * represent significance at the 1, 5 and 10 percent confidence levels, respectively. The hypothesis of constant returns to scale i s rejectedfor Clothing (decreasing) Rubber and Plastics and Wood Products and Furniture (increasing). 162 APPENDIXTABLE7: ProductionFunctionParametersEstimatedbyLevinsohnand Petrin(2003)Methods Industry N. Coeff. Coeff. Coeff. Coeff. Returns to Obs. Skilled Unskilled Intermediates Capital Scale Labor Labor FoodProcessing 299 0.104 *** 0.059 *** 0.760 *** 0.080 * 1.003 (0.024) (0.015) (0.103) (0.042) Textiles 324 0.104 *** 0.038 *** 0.480 * 0.080 0.702 (0.021) (0.014) (0.277) (0.097) Garments 235 0.082 *** 0.038 ** 1.Ooo*** 0.000 1.120 (0.031) (0.018) (0.280) (0.093) Auto-parts 219 0.122 *** 0.044 ** 0.700 *** 0.140 *** 1.005 (0.034) (0.019) (0.117) (0.040) Electronics and Electrical Appliances 277 0.059 ** 0.069 *** 0.480 0.060 0.667 (0.026) (0.023) (0.352) (0.080) Rubber and Plastics 358 0.149 *** 0.030 0.700 *** 0.070 0.949 (0.046) (0.037) (0.234) (0.072) Wood Products and Furniture 206 0.220 *** 0.091 *** 0.470 0.080 0.861 (0.033) (0.025) (0.311) (0.084) Machinery and Equipment 245 0.114 *** 0.030 1.000 *** 0.000 *** 1.144 (0.032) (0.019) (0.141) (0.042) Notes: The first column shows the number of firm-year observations included in the estimation. Bootstrapped standard errors are in parentheses. ***, ** and * represent significance at the 1, 5 and 10 percent confidence levels, respectively. The hypothesis of constant returns to scale i s rejectedfor Textiles (decreasing), Garments and Machinery and Equipment (increasing). APPENDIX TABLE8: ProductionFunctionParametersEstimatedby OLSandby LevinsohnandPetrin(2003) Methodsfor the MOIPanel Industry N. Coeff. Coeff. Coeff. Returns to Obs. Labor Intermediates Capital Scale OLS Estimates High-Tech Industry 412 0.131 *** 0.783 *** 0.087 *** 1.000 (0.020) (0.017) (0.011) Low-TechIndustry 448 0.165 *** 0.763 *** 0.058 *** 0.986 (0.016) (0.023) (0.016) Levinsohn and Petrin (2003) Estimates High-Tech Industry 412 0.122 *** 0.790 *** 0.070 0.982 (0.026) (0.244) (0.089) Low-Tech Industry 448 0.156 *** 0.490 *** 0.190 *** 0.836 (0.018) (0.208) (0.100) Notes: The first column shows the number of firm-year observations included in the estimation. In the OLS Estimates panel, robust standard errors are in parentheses. In the Levinsohn and Petrin (2003) Estimates panel, bootstrapped standard errors’ are in parentheses. ***, ** and * represent significance at the 1, 5 and 10 percent confidence levels, respectively. The hypothesis of constant returns to scale cannot be rejected for OLS and Levinsohn and Petrin (2003) estimates. 163 APPENDIX TABLE 9: Correlation Among FirmCharacteristics FirmAge FirmSize Exporter Foreign Capital % Computer R&D N.Obs. = 1033 (Current Dummy Ownership Vintage (% Controlled Spending Employment) Dummy mach. under Machinery Dummy 5 years old) FirmAge 1 FirmSize(Employment) 0.213*** 1 Exporter Dummy 0.033 0.421*** 1 ForeignOwnership Dummy 0.101*** 0.287*** 0.251*** 1 Capital Vintage (% mach. under 5 years old) -0.221*** 0.014 0.032 -0.021 1 % of Computer-Controlled Machinery 0.004 0.312*** 0.084*** 0.183*** 0.068*** 1 Dummy for R&DSpending 0.046 0.258*** 0.150*** 0.062*** -0.012 0.107*** 1 Note: *** indicates significance at the 5% confidence level. The correlations are obtained for the sample of 1033 firms used in the regressions of TFP, sales growthand value-added per worker. APPENDIX TABLE 10: Correlates of FirmPerformance Regressors Total Factor Sales Labor Productivity Growth Productivity FirmAge 0.004* ** -0.003 ** 0.012* ** (0.002) (0.001) (0.003) Current Employment 0.083* ** -0.012 -0.017 (0.015) (0.009) (0.025) Export Intensity 0.091 ** -0.073″ * 0.228″ ** (0.036) (0.030) (0.072) ForeignOwnership Share 0.229″”” -0.015 0.684*** (0.051) (0.041) (0.095) Capital Vintage (% Mach.Under 5 Years) 0.047 0.086** 0.121 (0.046) (0.038) (0.093) % Computer-Controlled Machinery 0.137*** -0.027 0.275″ ** (0.048) (0.035) (0.097) R&D SpendingDummy 0.022 0.013 0.072 (0.031) (0.022) (0.058) Industry Dummies Yes Yes Yes RegionDummies Yes Yes Yes N.Observations 1033 1033 1033 R-squared 0.98 0.07 Notes: OLS estimation i s used. Robust standarderrors are inparentheses. ***,0.21 ** and * represent significance at the 1, 5 and 10percent confidence levels, respectively. All regressions include a constant. 164 APPENDIX TABLE 11: Skill Mix and Average Wages Across Industries N. Avg. Skill Optimal Avg. Skilled Avg. Ratioof obs. Mix Skill Mix Labor Wage Unskilled Skilled to Industry (inBaht) LaborWage Unskilled (inBaht) Wage FoodProcessing 149 0.31 0.41 196007 77786 2.5 Textiles 162 0.27 0.52 175331 70012 2.5 Clothing 117 0.31 0.51 133754 63636 2.1 Auto-parts 109 0.52 0.57 164731 78319 2.1 ElectronicsandElectricalAppliances 138 0.32 0.25 199034 78446 2.5 Rubber andPlastics 179 0.28 0.62 152653 50719 3.0 WoodProductsandFurniture 103 0.35 0.53 138912 64764 2.1 MachineryandEquipment 122 0.39 0.61 167139 69047 2.4 Total 1079 0.33 0.50 167474 68501 2.5 APPENDIX TABLE 12: Increase inAverage Wages Dueto IncreaseinShare of Skilled Workers Elasticityof Skill Percent Increasein Average Premiumto Relative Wages as a Result of a 20 Percent Supplyof Skilled Increaseinthe Share of Skilled Workers Workers 0 7% -0.1 6% -0.2 4% -0.5 1Yo -0.6 0% 165 Chapter 2 -Appendix 1 The Thailand PICS provides us with a rich data set to undertake the following analysis. PICS surveyed 1,385 firms surveyed from March 2004 to February 2005 (response rate of 40 percent). The survey covers six regions: North; North East; Central; Bangkok and Vicinity; East; and South. It also covers eight industries: food processing; textiles; wearing apparel; auto parts; electronic parts & appliances; rubber & plastics; wood products & furniture; and machinery & equipment. Comparison of PICS sample with the NSO updated 2004 frame shows that the FTPI frame seriously under-represents SMEs. The data i s biased towards the larger companies, and hence we see a high incidence of exporting and foreign ownership. The PICS survey involves face-to-face interviews with the CEOs, human resource managers and a sample of workers. The interview with CEOs provides comprehensive coverage of firm’s governance structure as well as business practices such as investment, employment and skills provision, infrastructure, access to government services and access to finance. Firms also report their impressions of the existing investment climate. We are able to construct a panel on the basis of retrospective questions. For most variables we have information for the years 2001 and 2002. Production Variables There i s data for three years in the accounting questions IX.13 2000, 2001 and 2002; the data that we have for employment in question X.1 also refers to three years (2001, 2002 and 2003) but only two years overlap 2001 and 2002. This means that for our production function estimation these will be the two years of data potentially available. I)Real VA We estimate production functions with VA as the dependent variable. To obtain a measure of VA we should deflate (PPI by industry) these nominal values of total sales and take away deflated expenditure on raw materials (direct material cost electricity+ fuel +purchased parts cost+ and other energy). We would need to deflate this nominal quantity to get a quantity of intermediates using Bank of Thailand price deflators by stage of processing “Intermediate materials for industry”. 2) Labor variables We do not exactly have skilled and unskilled workers, but we can use the production versus non-production distinction that is widely used for work on United States’ data, and assume that production workers are the unskilled workers, while non-production workers are the skilled. 3) Capital We have balance sheet information that gives us a book value of capital (book price machinery equipment). This i s our proxy for the capital stock. Inaddition, we have a question on productive capital investment. We use first difference in the capital stock to replace missingvalues inthe productive capital investment responses. 4) Investment Climate Variables: Companies are asked to judge on a closed four-point scale how problematic the following factors are for the operation and growth of your business: telecommunications; electricity; transportation; access to land; tax rates; tax administration; customs and trade regulation administration; labor regulations; skills and education of 166 workers; business licensing and registration; access to domestic credit; access to foreign credit cost of financing (e.g. interest rates); economic policy uncertainty; macro-economic instability (inflation, exch. rate); corruption; crime, theft and disorder; and anti-competitive practices (e.g. monopoly). A dummy i s created for each company. The dummy is equal to 1for scoring greater than or equal to 3, otherwise 0. Inaddition, companies are askedto name the threebiggest obstaclesto doingbusiness in Thailandin order of importance: ownership regulations; tax regulations and/or hightaxes; SL shortage; labor regulations; obtaining land and buildings; foreign currency regulations; lack of business support services; inadequate supply of infrastructure; utility prices; inadequate access to credit; import regulations; high collateral requirements; high interest rates; insufficient demand for my products; competition from imports; crime and theft; official corruption; regulations for starting a business; bureaucratic burden; political instability; and lack of insurance (product liability). The dummy i s equal to 1if a company selects an obstacle inthe top three, otherwise 0. To deal with errors in reporting we exclude outliers. Our criteria first groups firms into eight broad industries. To be considered an outlier, we take the set of variables used in our empirical analysis. If any variable has a value larger than the mean, and i s more (or less) than three standard deviations of the industry-year mean, we exclude the firm. For the labor and materials revenue shares, two additional rules are applied: i)all firms with labor or materials revenue shares larger than one (or 1.1) are classified as outliers; ii)all firms with materials shares smaller than 0.05 are classified as outliers. These outliers defined by i)and ii)areactuallydroppedfromthecalculationofindustry-yearmeansandstandarddeviations required to apply the three standard-deviation rule above. 167 Chapter 2 -Appendix 2 METHODOLOGY FORESTIMATINGCOMPANY LEVEL TFP The basic idea i s to get consistent estimates for return to SL, UL and capital (K) in VA (VA). We are going to estimate them at the industry level by region. Again, due to data limitations, we aggregate to two broad industries and region taxonomies, defined below. Our aim i s to estimate TFP at the company level, as follows, for company i in time t (2001 and 2002), industry j (low-tech (food processing, textiles, clothing and wood products & furniture) & high-tech (auto parts, electronics appliances, rubber & plastics and machinery & equipment) and region k (Bangkok region + two central provinces and outside Bangkok region – two central provinces). Our goal i s to estimate consistent production function parameters dealing with two interrelated estimation biases. Simultaneity bias arises out of the fact that input demands are, in part, determined by the manager’s knowledge of productivity levels. Selection bias stems from the notion that productivity affects the decision to select into a particular location. We adapt an estimation algorithm developed in Olley and Pakes (1996) to deal with both of these biases. Why use Olley and Pakes (1996) to deal with simultaneity bias? The benefit of the Olley and Pakes (1996) approach i s that there i s a structural model of the unobservable that suggests the optimal investment dynamics of enterprises, given the observable state variables, should allow one to control effectively for the omitted unobservable using non-parametric techniques. We use the Olley and Pakes (1986) estimation routine to estimate our parameters on inputsto get our measureof TFPi, (See Appendix IIfor details) (24 VAit =PO+Plslit +PZUlit +P3kit + wit + vir We assume that investment sequences, !if,chaseperformance and are short-run decisions that are mainly determined by long-run state variables such as the observable stock of physical assets, kif and the unobservable productivity type of the company, wif.,,Weassume that ii, = hit(Uia kit) and more importantly can be inverted and differentiated, wit = hit(ii, kit). (2b) = plslir +P 2 Ulir+ Vit(iit,kit)+ vir where pidiit, kit) = Po +VAit + /?3kif hit(Uit, kit) and i s proxied with a third-order polynomial in iif and kit. vir i s our regression error, and we find estimates of PIand PZestimating equation (lb) usingOLS. These are the elasticities of VA with respectto skilled and UL,respectively. While this allows for simultaneity bias, we also model the probability of being in a particular location, pif, given the companies productivity type among other sets of characteristics, Xi, (3) PrIL = 1 wit+l, kir+l)Xit}= pt(& kt,Xir) I To estimate unbiased estimates of P1 andP2 allowing for both selection and simultaneity bias one should proxy for Vir(iit,kif)andPit with a thirdorder polynomial in iit, kit and Pit. Finally, we estimate ourP3 the elasticity of VA with respect to capital usinga non linear least square estimator, A-m A A h j = o m=O 168 We proxy the third term on R.H.S.of the equation with a thirdorder polynomial inestimates of hitand Pit ( locationprobability), where the estimate of hidoit) kit)= vidiit, kit) – Po – P&ir. We include time and industry dummies inour regressions. The above may be re-written to allow for intercept shifts for each year and our sub-industries. 169 170 171 172 173 174 175 176 177 178 rl o L n * o m r l m r l . . 0 mu3 Lnw o o d r l d m m m . . . . . . . . m rlrl w m 9 . I I I I I 179 — m u , . . 2 2 rl 0o 0d 0c 0 0 ~ 2 : *p .+ . . n m ‘ , crlrll .n . r I . . . ‘ I ” I I I I I I 180 9 cu m m w w r l m m m . . . . I I I I I H a ) H k a) 181 Pa rn a, u .rl u C V Id 0 G 182 z 2 3 3 2 2 r- r- Q G 3 G T 9 r ?W 9 x v, W 3 2 2 r? Q 4 3 3 2 o! o! w – w 00 00 c? 9 m m m 9 3 3 2 $1 c? 2 2 2 2 3 3 3 2 r- 9 9 Q 0 Q v) .3 Yu Ecd v) 4 P 8 v) 0 sa: 3.3 Y x 183 w – w – w ? ? T ? v3 0 0 0 0 0 0 008 8v, N z-8 E? *3 3 u 5 gc M G0 184 51 ?1? c ? 9 1 3 3 3 1 1 1 49 ” ? 3 ” 2 1 1 4 3 .-Y E M 0 !i M 3.3 e IM3 M 2a 0 M 0 aEI M M .3c 0 a 3 .i Y 5c F4 b 8 3 s0 a0 a Q) 185 Chapter3 Appendix – Table 1A. Returns to education. [Dependent Variable: individualloghourly wage] Thailandt Malaysia 10 0.117′ 0.143″ (0.050) (0.027) 11 0.099 0.122″ (0.068) (0.015) 12 0.148” 0.148″ (0.010) (0.022) 13 0.238** 0.229’* (0.080) (0.025) 14 0.273.. 0.276″ (0.014) (0.028) 15 0.392′; 0.385’. (0.061) (0.034) 16 0.516” 0.480.. (0.015) (0.037) morethan 16 0.743** 0.506″ (0.031) (0.034) Constant 2.780″ 0.709** (0.017) (0.025) Worker characteristics Yes Yes Establishmentfixedeffects Yes Yes Observations 13,476 7,812 Number of Firms 1,385 893 AdjustedR z 0.574 0.481 Note : Standarderrors are denotedinparentheses. denotes significanceat 5% level; ** denotes significanceat 1%level. TCoeftIcients are estimated using OLS. Worker characteristics include potential experience, potential experience squared, tenure, tenure squared, distance from workplace and dummy variables for occupation (manager, professionals, skilled production workers, unskilled production workers and nonproduction workers), study abroad, computer skills (none, basic, moderate andcomplex), people skills, content of training (production technologies, marketing, information technology, management/quality technologies, intellectual property, safety procedures, language skills and others), training received from previous employer, outside training, lack of English language proficiency, unemployment in last two years, non-Thai ethnicity and martial status interacted with gender (single male, single female, married male, and married female). The specificationfor the Malaysian sample i s similar. Source:Thailand PICS 2004; Malaysia PICS 2002. Author’s calculations. 186 Table 2A: ThailandLaborForceand GDPby Sectors LaborForce I GDP I Numbers Ratio (thousand) MillionBaht Ratio rota1 33,719.3 I I lOO%l 5,930,362 lOO.OO%l Agriculture 13,907.6 41.25% 595,004 10.03% Agriculture, Hunting and Forestry 13,451.5 39.89% 491,026 8.28% Fishing 456.1 1.35% 103,978 1.75% Non-Agriculture 19,811.7 58.75% 5,335,358 89.97% Mining and Quarrying 48.2 0.14% 154,564 2.61% Manufacturing 5,270.0 15.63% 2,060,447 34.74% Electricity and Water Supply 95.1 0.28% 190,946 3.22% Construction 1,848.3 5.48% 175,586 2.96% Wholesaleand Retail Trade; Repair of Motor Vehicle 5,171.3 15.34% 914,328 15.42% Hotels and Restaurants 2,120.7 6.29% 300,414 5.07% Transport, Storage and Communications 1,043.8 3.10% 465,204 7.84% Financial Intermediation 286.8 0.85% 202,151 3.41% Real Estate, Renting and BusinessActivities 554.2 1.64% 177,890 3.00% Public Administration and Defense; Compulsory Social Security 934.4 2.77% 262,081 4.42% Education 981.9 2.91% 220,938 3.73% Health and Social Work 514.0 1.52% 106,803 1.80% Other Community,Social and Personal Service Activities 670.1 1.99% 96,184 1.62% Private Households withEmployedPersons 272.9 0.81% 7,822 0.13% Source: National Income from National Economic and Social Development Board (NESDB) and Labor Force Survey from National Statistical Office (NSO). 187 E – ,% Q H a g = N N I- m 00 2 vi vi 00 7 vi -* 2 m m vi m W 2 vi a m 3 m m 9 W- N 3 2 vi 00 N N F 9 i m N m z N 3 3 t; Ni m m o vi c? m m N o 3 c4 x m 3 3 -8 W I- m m m t 188 Time To FillVacancies Of Professionals And Skilled ProductionWorkers By Industry – Textiles – – B r a d- Thailand – Thailand Malaysia Malaysia 1 Philipines 3angladesh India India Brazil Philipines Bangladesh Indonesia 0 2 4 6 8 – – Garments — n m e to FillVacuncy Por Skilled Production Worker in Last 2 Years Garments (in weeks) – Brazil Thailand Thailand Malaysia Malaysia Brazil Bangladesh Philipines Philipines Bangladesh = India India Indonesia Indonesia 0 2 4 6 8 1 0 1 2 1 4 — FoodProcessing – n m e to FillVacancy for Skilled Production Worker in Last 2 Years Food (inweeks) – – Brazil ‘Ihailand Thailand I Malaysia Malaysia Philipines India India = Philipines Brazil Bangladesh Bangladesh 0 2 4 6 8 10 0 2 4 6 8 189 — — Electronics – n m e to FillVacancy for Professional in Last 2 Years- %e to FillVacancy for Skilled Production Worker in Electronicsmec. AWances (in weeks) Last 2 Years Electronicsmee Appliances (in weeks) – Brazil 1 Malaysia Malaysia “hailand “hailand India Philipines Brazil = India Philipines = = Indonesia Indonesia Bangladesh Bangladesh Machinery & Equipment %e to FillVacancy for Professional in Last 2 Years- Tlme to FillVacancy for Skilled Production Worker in MachinerylEquipment (in weeks) Last 2 Years Machinery/Equipment (inweeks) – Brazil Brazil Thailand Malaysia Malaysia ‘Ihailand India Indonesia Indonesia India I 0 5 10 15 20 25 30 35 40 45 0 2 4 6 8 10 12 Auto Parts n m e to FillVacancy for Professional in Last 2 Years- lime to FillVacancy for Skilled ProductionWorker in Auto-Parts (inweeks) Last 2 Years Auto-Parts (inweeks) – Brazil Brazil “hailand lhailand e Malaysia Malaysia India India Indonesia Indonesia 0 5 10 15 20 25 30 35 40 0 2 4 6 8 10 12 190 Chapter 4 Appendix – FigureA4.1: EstablishmentsFilingPatentsby Ownership andExporter Status 25 m%of totalfirms 8 % of forebnfirm %of nonforeipn firm x of exportlngflrm 8 % of nonexportmgfirms n ” Y .10 5 0 FigureA4.2: EstablishmentsFilingPatentsinthe Last Two Years by Industry (%) Firms filing patentshtility models or copyright protected materials In the last two years (YO) – – _ _ _ _ _ _ _ _ — _ _ _ _ _ – _ _ _ _ _ _ _ _ _ _ _ Food Textile Clothing Automotive Electronics Rubberand Wood Machinery OVERALL processing parts Plastics products and furniture 191 TableA4.3: FrequencyDistributionof TCI andMeanScores by Region North Central Bangkok East Northeast South Frequency distribution of TCI ogc110.2 20.6 23.7 16.1 12.0 36.6 38.3 0.2O and O and <30 percentof production machines i s computer controlled 0.015 (0.037) 30-100percentof 0.113"' (0.041) employ staff exclusively for desigddoing innovation r&d -0.053 subcontract project to othercompanies or organizations 0.001 (0.076) paidroyalties -0.063 (0.059) planning introduce new designs productsinthe next 2 years 0.003 (0.034) upgradedmachinery & equipment 0.103'' (0.043) enterednew marketsdue processor product improvementsincost quality -0.01 filed any patent>

Continue lendo em https://documents.worldbank.org/curated/en/200381468123899136/text/362670TH.txt