EntErprisE survEys EntErprisE notE sEriEs 68273 productivity Total Factor Productivity Across the Developing World 2011 Federica Saliola and Murat Seker T otal factor productivity (TFP) is a crucial measure of efficiency and thus an important indicator for policymakers. Using micro level data from manufacturing industries in 80 developing countries, this note analyzes TFP performance at the firm-level. Among the countries surveyed during the same period across multiple regions—Eastern Europe and Central Asia, Latin America, Africa, and Asia—Hungary, Peru, Ethiopia and Indonesia have the highest aggregate productivities. A comparison of average productivities in each region shows that Moldova, Nicaragua, Ethiopia and Indonesia have the highest values among the countries surveyed. This note EntErprisE notE no. 23 also discusses separate estimates of TFP values obtained at the industry level. These industry-level estimates are the most useful for policymakers in that they reveal comparative advantages of specific industries within countries. In the garments and chemicals industries, Brazil has the highest average productivity among all the countries surveyed. Introduction Data description In the last three decades, many studies have analyzed The World Bank’s Enterprise Surveys1 provide a unique the relative contribution of factor inputs and technical source of information that can be used to measure TFP progress to economic growth. Since the seminal work of across a large set of developing countries. The data used Solow (1957), total factor productivity—defined as the for TFP analysis in this note cover manufacturing firms efficiency with which firms turn inputs into outputs—has in 80 countries from different regions of the world.2 All been considered as the major factor in generating growth. data used in this analysis were collected from surveys The availability of firm-level data allowed researchers conducted since 2006, with the exception of India which to investigate the reasons behind the vast dispersion in was surveyed in 2005. The regional coverage of the productivity performances across firms which led to the countries is presented in Table 1. The table also shows establishment of policies that would improve productivity the number of firms that are included in the analysis from and eventually generate growth. Some early examples of each region. firm-level productivity analyses are Bailey, Hulten, and Campbell (1992) and Bartelsman and Dhrymes (1998) for U.S. manufacturers and Roberts and Tybout (1996) for a Table 1 Number of countries in each region World Bank Group number of developing countries. Research on the comparison of productivity performances Region # of Countries # of Firms across countries has been limited due to the unavailability Sub-Saharan Africa (AFR) 25 5,582 of a homogenous data source. This note aims to fill this South Asia and East Asia 9 5,439 gap. It uses a data set which has been collected through and Pacific (Asia) surveys conducted across a large number of developing Eastern Europe and 25 2,872 Central Asia (ECA) countries. The homogenous nature of the data provides Latin America and the a unique opportunity to compare average productivity 15 5,514 Caribbean (LAC) performances of firms across industries, countries and Middle East and North regions. 6 2,005 Africa (MENA) Total 80 21,412 Source: Enterprise Surveys. of the analysis with these alternative TFP measures are Table 2 Industries included in the analysis discussed in Saliola and Seker (2010).5 That study showed ISIC Code Two-digit Industry Percentage that TFP estimates obtained from all specifications are 15 Food 20.9 positively and highly correlated with each other. 17 Textiles 8.8 The productivity values are estimated separately for 18 Garments 15.2 each country, while controlling for industry differences by 24 Chemicals 8.4 including dummy variables for each industry group listed 26, 27 Non-Metallic & Basic Metals 7.2 above. All monetary values are converted into U.S. dollars Fabricated Metal & and then deflated by GDP deflator in U.S. dollars (base year 28, 29 12.3 2000).6 For each variable used in the estimation, values that Machinery - Other Manufacturing 27.2 are three standard deviations away from the mean value Source: Enterprise Surveys. for each country are excluded from the analysis. These outlier tests are performed at the country level. Firms The data cover all the major two-digit manufacturing that have material cost-output or labor cost-output ratios industries according to the International Standard that are three standard deviations away from the mean are Industrial Classification (ISIC), revision 3.1. For this also excluded from the analysis. In addition, Afghanistan, 7 analysis some industries are combined to achieve a Albania, Burkina Faso, Kosovo, Malawi, Niger and West sufficiently large number of observations (Table 2). Bank and Gaza were excluded since at least one of the Industries were grouped together based on similarities in variables required to compute TFP was not available for the type of activity and factor intensity. The group “Other at least 30 percent of the manufacturing firms surveyed. Manufacturing� is a residual category that includes all When the data is collected, each firm is assigned firms that are outside the six major industry groups. The a sampling weight in order to allow the data to be concentration of firms in six major industry groups is the representative at the country level. These weights are not 8 result of a sample design, used in most countries, where used in the TFP analysis because the variables to measure selected industries were targeted to facilitate industry-level TFP are not available for all firms included in the surveys. analysis. Hence the composition of the sample adopted in the empirical analysis to measure TFP might not reflect the Estimating total factor productivity actual composition of firms in the manufacturing sectors. A Cobb-Douglas production The un-weighted sample for which function with three factors of TFP analysis could be performed is production—capital, labor and In the 2008–2009 defined as the productivity sample. The intermediate goods—is used to sample, Indonesia has data coverage issue raises the question estimate TFP. Firm sales are used 3 whether the productivity sample over- the highest aggregate or under-samples firms in certain size to measure output; the replacement value of machinery, vehicles and productivity and Brazil groups. In order to test this difference, equipment is used to measure has the highest average size distribution measured in terms of capital; labor is assessed by the total employment levels in the productivity productivity. sample is compared to the distribution compensation of workers including wages, salaries and bonuses; and in the full sample obtained by using the intermediate goods are determined by the cost of raw survey weights (which is defined as the weighted sample). materials and intermediate materials. TFP is estimated as The weighted sample includes the productivity sample the residual term of the production function. and the rest of the firms for which TFP could not be The TFP values used in this note are compared with the estimated and it is representative of the manufacturing values obtained from five additional production function sector in each country. In general, the distribution of the specifications. These specifications are three variations of productivity sample mirrors relatively well the distribution the Cobb-Douglas production function; a transcendental of the weighted sample. In countries where there is a logarithmic (trans-log) production function with capital, reasonable difference (more than 10 percentage points labor and materials as input factors; and a non-parametric in any size group), small firms (less than 20 workers) cost-based Solow residual method.4 The first variation of are slightly under-sampled in the productivity sample. In the Cobb-Douglas production function adds energy costs a few countries like Indonesia, Nepal, Uzbekistan and to the input factors; the second variation uses only labor Guatemala this difference is around 30 percent. and capital as input factors; and the third uses value added as the dependent variable instead of total sales. Details 2 Estimation results Eslava et al. (2004) estimate the production function at The coefficients obtained from the estimation using a industry level rather than country level. This could also Cobb-Douglas production function can be interpreted as play a role in explaining the different elasticities. input factor elasticities; they show the responsiveness of Cross-country analysis sales to changes in the levels of each input factor used in production. In the estimation of the production function, Using the factor elasticities obtained above for each raw materials and intermediate goods have the highest country, firm-level TFP values were computed. Firms’ elasticity in 52 of the 80 countries.9 In 51 countries, labor productivity levels are weighted by their output shares in has the second highest level of elasticity after materials. order to compute aggregate productivity. Output shares are The average elasticity values across countries are 0.10 calculated as the ratio of each firm’s sales to aggregate for capital, 0.46 for labor, and 0.54 for materials. Figure sales in the country. Hence, when weighted productivities 1 presents elasticities for select countries. The share of are aggregated to compute the aggregate productivity, a capital is lowest in Indonesia with a value of 0.02 which firm with higher production has a larger contribution than means that a 10 percent increase in capital is associated a firm with low production. Simple average productivities with an increase in output of just 0.2 percent. For each are also presented in order to see how an average firm country, the sum of the three factor elasticities is around performs in each country. one. This corresponds to the assumption of the Cobb- The years in which the surveys were conducted vary Douglas production function. in the data. This difference can contribute to variation in The input factor elasticities obtained from the productivity performances across countries. For analytical estimation yield comparable results to several other purposes, countries were grouped in two cohorts— studies. Using firm-level data from Colombia covering the those surveyed in 2006–2007 and those surveyed in years 1982-1998 and using the same 2008–2009 (44 and 36 countries estimation method as above with respectively). The cross-country four input factors—capital, labor, In the 2006–2007 sample, comparison in this section uses data energy and materials—Eslava et al. Peru has the highest from countries that have relatively (2004) find factor elasticities of 0.08, large sample sizes. Comparison of 0.24, 0.12 and 0.59. The estimation aggregate productivity average and aggregate productivities using Enterprise Surveys data for and Nicaragua has shows noticeable differences across Colombia from 2006 yields the factor the highest average countries. A country with a high elasticities in respective order of average productivity level could have 0.09, 0.48, 0.07, and 0.46. Hallward- productivity. quite low aggregate productivity or vice Driemeier, Iarossi and Sokoloff versa. This discrepancy between the (2002) calculate these elasticities as 0.15, 0.30, 0.24 and two measures could be caused by the differences in the 0.31 for Malaysia using firm-level data covering 1996- size distribution of the samples. Small sample size in a 1998. In our results for 2007 Malaysian data, these values particular size group, which is more likely to be the case are 0.03, 0.48, 0.10, and 0.51 respectively. Differences in for large firms, could cause noticeable differences across these elasticities could be a result of changes in the time both TFP measures. Another reason for this discrepancy period studied or differences in the definition of capital.10 is the variation in average productivity levels of firms in different size groups. If small firms are much more Figure 1 Factor elasticities for selected countries 1.2 1 Factor Elasticities 0.8 0.6 0.4 0.2 0 Brazil Chile Egypt, Arab Ethiopia Indonesia Mexico Mozambique Nigeria Philippines Russian Syrian Arab Turkey Ukraine Republic Federation Republic ■ Capital ■ Labor ■ Material Source: Enterprise Surveys. 3 Figure 2 Aggregate and average productivity of countries in 2008–2009 ■ Aggregate TFP ■ Average TFP 0.34 0.11 0.24 0.06 Aggregate log (TFP) Average log (TFP) 0.14 0.04 0.01 Syrian Arab Republic Mongolia Madagascar Vietnam Croatia Ukraine Brazil Egypt, Arab Rep. Bulgaria Madagascar Philippines Indonesia Serbia Russian Federation Kazakhstan Turkey -0.06 0.04 -0.16 -0.26 -0.09 Source: Enterprise Surveys. productive than large firms in a country, then this country lowest in this country. The difference between average might have high average productivity but low aggregate and aggregate productivities could be caused by how productivity relative to other countries. productivity is distributed among firms at different size Figure 2 shows aggregate and average productivity levels. Productive large firms make a large contribution values in the countries that were surveyed in 2008-2009 to aggregate productivity. However, this difference and that had at least 100 firms for which TFP could be could also be caused by the distribution of firms in the estimated. Among these countries, Indonesia has the productivity sample. For example, in Nicaragua the share highest aggregate productivity followed by Turkey. The of large firms in the sample is 3.5 percent (only 9 firms), picture is quite different for average productivity. Brazil has one of the lowest shares in the 2006–2007 period. These the highest average productivity among these countries. large firms have very low productivities which drag the Serbia, which has the lowest aggregate productivity level, aggregate productivity to lower levels as compared to has an average productivity that is higher than the average average productivity. While the firm-size distribution for productivities in Indonesia or Turkey. Nicaragua is representative of the population (the share The same analysis is performed for those countries of large firms is 4.5 percent in the weighted sample), the that were surveyed in 2006–2007 and that had more than small number of observations causes the big discrepancy 200 firms for which TFP could be estimated (Figure 3).11 between the two TFP measures. Peru has the highest aggregate productivity among these countries. However, average productivity is among the Figure 3 Aggregate and average productivity of countries in 2006–2007 0.35 ■ Aggregate TFP ■ Average TFP 0.07 0.25 0.05 0.15 0.03 Aggregate log (TFP) Average log (TFP) 0.5 0.01 -0.05 -0.01 Zambia Guatemala Senegal Colombia South Africa Ecuador Tanzania Mozambique Morocco India Thailand Nigeria Angola Nicaragua Kenya Uganda Ghana Argentina Boliva El Salvador Chile Mali Malaysia Pakistan Ethiopia Mexico Peru -0.15 -0.03 -0.25 -0.05 Source: Enterprise Surveys. 4 Table 3 Countries with high and low productivity levels ECA 2008/09 LAC 2006 AFR 2006/07 ECA 2008/09 LAC 2006 AFR 2006/07 Mean 0.18 Mean 0.01 Mean -0.02 Mean 0.03 Mean 0.03 Mean 0.02 High values of aggregate TFP High values of average TFP Hungary 1.50 Peru 0.32 Ethiopia 0.24 Moldova 0.07 Nicaragua 0.05 Ethiopia 0.04 Romania 1.16 Mexico 0.28 Botswana 0.23 Kyrgyz Rep. 0.06 Honduras 0.05 Zambia 0.04 Uzbekistan 0.64 Chile 0.11 Mali 0.12 Serbia 0.06 Panama 0.04 Namibia 0.04 Kyrgyz Rep. 0.50 Panama 0.11 Rwanda 0.11 Kazakhstan 0.06 Guatemala 0.04 Swaziland 0.03 Georgia 0.31 El Salvador 0.10 Ghana 0.05 Macedonia, FYR 0.05 Paraguay 0.03 Burundi 0.03 Low values of aggregate TFP Low values of average TFP Bulgaria -0.09 Ecuador -0.13 Tanzania -0.12 Latvia 0.02 Bolivia 0.02 Rwanda 0.01 Belarus -0.10 Colombia -0.15 South Africa -0.14 Azerbaijan 0.02 Colombia 0.02 Angola 0.01 Latvia -0.11 Uruguay -0.19 Senegal -0.16 Croatia 0.02 Chile 0.02 Mali 0.01 Slovak Rep. -0.19 Guatemala -0.19 Swaziland -0.19 Romania 0.01 Argentina 0.01 Mauritania 0.01 Serbia -0.27 Honduras -0.34 Zambia -0.24 Hungary 0.01 Peru 0.01 Ghana 0.01 Source: Enterprise Surveys. country is the second highest in the region. In this region all Regional analysis countries except Brazil were surveyed in 2006. In the AFR region, 21 of the 25 countries included in the analysis were The rich coverage of data from the ECA, LAC and surveyed in 2006–2007. Among these countries, Ethiopia AFR regions allows performance of regional-level analysis has the highest aggregate and average productivity levels. (Table 3). Using all countries for which TFP could be estimated, countries are ranked according to their On the other hand, Zambia has the lowest aggregate aggregate and average productivity levels. In the ECA productivity but the second highest average productivity. region Hungary has the highest aggregate productivity The other four countries in this region—Cameroon, Côte which is followed by Romania and Uzbekistan. However, d’Ivoire, Madagascar and Mauritius—were surveyed in the ranking for average productivity is quite different. 2009. The country with the highest aggregate productivity Among the large economies in the region—Ukraine, in this group is Côte d’Ivoire (with a TFP of 0.76) followed Turkey, Russia, Bulgaria and Kazakhstan—Turkey has the by Madagascar (with a TFP of -0.04). highest aggregate productivity level. The spread of average productivity distributions shows In the LAC region, Peru has the highest aggregate variation across these three regions (Figure 4).12 The productivity, followed by Mexico. The least productive dispersion in the AFR region is the smallest among the country is Honduras although average productivity in this three. The standard deviation of TFP values in AFR is 0.39 whereas it is 0.64 and 0.71 in LAC and ECA respectively. Figure 4 Box plot of TFP distribution Figure 5 Cumulative density distribution in three regions 1 .8 Cumulative density of TFP .6 .4 .2 4 0 -4 -2 0 2 Average log (TFP) -2 0 2 4 ■ AFR ■ ECA ■ LAC Average log (TFP) ■ AFR ■ ECA ■ LAC Source: Enterprise Surveys. Source: Enterprise Surveys. 5 The difference in log productivity levels between the highest average productivity in the food industry, has the 5th and 95th percentiles in the AFR region is 1.2, which highest aggregate productivity. In addition, in the garments corresponds to a TFP ratio of 3.3.13 These ratios are 7.4 in and chemicals industries, Brazil shows the lowest aggregate LAC and 9.4 in ECA. Figure 5 shows the cumulative density productivity but the highest average productivity. The Arab of average TFP in each region.14 The graph indicates that Republic of Egypt has the highest aggregate productivity all regions had similar average productivity. Moreover, in the chemicals industry and it ranks second to last in the productivity distribution in ECA and LAC are more food and garments. Comparison of average productivities spread out than the distribution in the AFR region. This shows that Turkish manufacturers have the second lowest means that the number of firms with very high and very productivity in all three industries. low productivity in these two regions is higher than the As mentioned earlier, the discrepancies between number in the AFR region. average and aggregate productivities could be caused by In Asia, there are five countries surveyed in 2009— differences in firm-size distributions within the samples Indonesia, Mongolia, Nepal, Philippines and Vietnam. and average productivity levels at different size groups. The average TFP value of these countries is 0.03. Nepal For example, the Philippines has the highest average has the highest aggregate productivity level (0.38) which productivity in the food industry, but exhibits the lowest is followed by Indonesia (0.27). The lowest aggregate aggregate productivity level. In the Philippines sample, the productivity is observed in Vietnam (-0.004). Comparing share of firms with more than 100 employees in the food average productivities, Indonesia has the top ranking industry is relatively small (9 firms) and they have relatively (0.05), followed by the Philippines (0.04). low productivity. Table 4 presents a comparison of aggregate and average Industry analysis TFP for the group of countries surveyed in 2006–2007.16 The manufacturing industries listed in Table 2 are likely Chile has the highest aggregate productivity in the food to have different production technologies. Therefore, industry. Bolivia shows the highest aggregate productivity separate estimations at the industry level are not only in garments while Peru is the country with the highest desirable but they could be useful in understanding average productivity (Figure 7). Morocco has highest differences in firm performance as well as revealing aggregate productivity in chemicals although the average comparative advantages within countries. productivity is second to last. Mexico exhibits relatively Industry-level estimates of TFP values are presented good performance in garments and chemicals industries. only for those countries that had at least 45 observations Firms in Mexico have the third highest aggregate in each selected industry—food, garments and chemicals.15 productivity and the fourth highest average productivity in The countries for which industry-level TFP values could garments, and the second highest aggregate productivity be computed in the 2008–2009 period are presented in and the third highest average productivity in chemicals. Figure 6. The cross-country comparison of aggregate productivities shows that Brazil, which has the second Figure 6 Aggregate and average productivity of countries in 2008–2009 in food, garments and chemicals industries 1.2 ■ Aggregate TFP ■ Average TFP 0.84 0.64 Aggregate log (TFP) 0.7 0.44 Average log (TFP) 0.24 0.2 0.04 Brazil Turkey Mongolia Russian Federation Vietnam Indonesia Egypt, Arab Rep. Philippines Russian Federation Indonesia Turkey Vietnam Egypt, Arab Rep. Brazil Egypt, Arab Rep. Philippines Indonesia Turkey Brazil -0.16 -0.3 -0.36 -0.8 Food Garments Chemicals -0.56 Source: Enterprise Surveys. 6 Table 4 High and low productivity levels of countries in 2006-2007 in the food, garments and chemical industries Food Garments Chemicals Food Garments Chemicals High Values of Aggregate TFP High Values of Average TFP Chile 0.44 Bolivia 0.32 Peru 0.31 Nicaraugua 0.08 Peru 0.05 Morocco 0.04 Malaysia 0.24 Guatemala 0.26 South Africa 0.21 El Salvador 0.05 El Salvador 0.04 Mexico 0.03 Kenya 0.23 Mexico 0.09 Ecuador -0.12 Pakistan 0.05 Zambia 0.04 Chile 0.03 Low Values of Aggregate TFP Low Values of Average TFP Tanzania -0.35 Tanzania -0.37 Mexico -0.16 Mali 0.01 Nigeria 0.01 Malaysia 0.02 Uruguay -0.37 El Salvador -0.38 Morocco -0.26 Ghana 0.01 Mali 0.01 South Africa 0.02 Honduras -0.51 Peru -0.42 Chile -0.22 Malaysia 0.01 Ghana 0.01 Peru 0.01 Source: Enterprise Surveys. Conclusion lowest in this country. The regional analysis shows some variation across ECA, LAC and AFR regions in terms This note provides an analysis of the total factor of average productivity distributions. The dispersion of productivity for firms in developing countries from total factor productivity in AFR is the smallest among the different regions of the world using the World Bank’s three regions. The analysis across industries shows how Enterprise Surveys. It presents cross-industry, cross countries vary in the productivity performances of their country and regional productivity comparisons. Indonesia industries. In 2008–2009 Brazil stands out for having the has the highest aggregate productivity among the highest average productivity in the garments and chemicals countries that were surveyed in 2008–2009, followed by industries and the second highest average productivity Turkey, while Brazil has the highest average productivity. in the food industry. Among the countries that were Among the countries that were surveyed in 2006–2007, surveyed in 2006–2007, Mexico exhibits relatively good Peru has the highest aggregate productivity among these performance in garments and chemicals. countries. However, average productivity is among the Figure 7 Aggregate and average productivity of countries in 2006–2007 in the garments industry 0.40 0.1 ■ Aggregate TFP ■ Average TFP 0.1 -0.10 Aggregate log (TFP) 0.0 Average log (TFP) -0.60 -0.1 Bolivia Guatemala Mexico Ghana Bulgaria Nigeria Argentina Chile Mozambique Mali Morocco South Africa Thailand Kenya Zambia Colombia Tanzania El Salvador Peru -1.10 -0.1 Source: Enterprise Surveys. 7 Notes 12. A box plot graphically displays the distribution/spread for a set of data. The three vertical lines of the box itself correspond to the 1. The data used in this study as well as the methodology used in 25th, 50th, and 75th percentiles. The formulas used to construct data collection and sample construction are available at www. the box's "whiskers" correspond to J. W. Tukey's Exploratory enterprisesurveys.org Data Analysis (1977). The dots that appear outside the whiskers 2. The countries included in the analysis, by region, are: Eastern correspond to actual data values and visually indicate how many Europe and Central Asia (ECA): Armenia; Azerbaijan; Belarus; data points are in the lower/upper extremes of the distribution. Bosnia and Herzegovina; Bulgaria; Croatia; Czech Rep.; Estonia; Macedonia, FYR; Georgia; Hungary; Kazakhstan; Kyrgyz Rep.; 13. TFP(95pct) =e1.2=3.3 Latvia; Lithuania; Moldova; Poland; Romania; Russian Federation; TFP(5pct) Serbia; Slovak Rep.; Tajikistan; Turkey; Ukraine; Uzbekistan; Middle 14. The upper and lower tails of the cumulative density graphs are East and North Africa (MENA): Algeria; Egypt, Arab Rep.; Jordan; trimmed in order to have a better illustration of the central part of Morocco; Syrian Arab Rep.; Yemen, Rep.; Latin America and the the TFP distribution across regions. Caribbean (LAC): Argentina; Bolivia; Brazil; Chile; Colombia; 15. These industries were chosen due to their relatively higher coverage Ecuador; El Salvador; Guatemala; Honduras; Mexico; Nicaragua; across countries. The analysis was also performed for textile, non Panama; Paraguay; Peru; Uruguay; South and East Asia and Pacific metallic and metal and machinery industries. Additional results are (Asia): India; Indonesia; Malaysia; Mongolia; Nepal; Pakistan; available upon request. Philippines; Thailand; Vietnam; Sub-Saharan Africa (AFR): Angola; 16. Among the countries surveyed in 2006–2007, 30 countries meet Botswana; Burundi; Cameroon; Côte d’Ivoire; Congo, Dem. Rep.; the 45 observations criterion in the food industry, 19 countries in Ethiopia; Ghana; Guinea; Guinea-Bissau; Kenya; Madagascar; Mali; garments and 8 countries in chemicals. Mauritania; Mauritius; Mozambique; Namibia; Nigeria; Rwanda; Senegal; South Africa; Swaziland; Tanzania; Uganda; Zambia. Indicator Surveys (IS) were excluded because of the small size of References the sample. Bailey, M. N., C. Hulten, and D. Campbell. 1992. “The Distribution of 3. The Cobb-Douglas production function specification used in Productivity in Manufacturing Plants.� Brookings Papers on Economic the estimation is , where K is capital, L is labor and Activity: Microeconomics. 4:187-267. M is material input. The exponents represent factor elasticities. Bartelsman, E. J. and P. J. Dhrymes. 1998. “Productivity dynamics: 4. In the non-parametric Solow residual method, output elasticity of U.S. manufacturing plants, 1972–1986.� Journal of Productivity each input factor is calculated as the cost share of that input in total Analysis 9(1):5–34. cost. TFP is estimated as the residual of the production function, Eslava, M., J. Haltiwanger, A. Kugler, and M. Kugler. 2004. “The making use of the calculated elasticities. Effects of Structural Reforms on Productivity and Profitability 5. The paper is available from the authors upon request. Enhancing Reallocation: Evidence from Colombia.� Journal of 6. Exchange rates and GDP deflators are obtained from World Development Economics 75(2):333-371. Development Indicators, World Bank. Hallward-Driemeier, M., G. Iarossi, and K. L. Sokoloff. 2002. “Exports 7. In total, 3,381observations (out of 24,793) were identified as and Manufacturing Productivity in East Asia: A Comparative outliers. Analysis with Firm-Level Data,� NBER Working Paper No: 8894. 8. Please refer to the Methodology page of the Enterprise Surveys Roberts, M. and J. Tybout. 1996. Industrial Evolution in Developing website for more information: Countries: Micro Patterns of Turnover, Productivity and Market Structure. http://www.enterprisesurveys.org/Methodology New York: Oxford University Press. 9. Elasticity values of the 80 countries are available upon request. Saliola, F. and M. Seker. 2010. “Productivity analysis using micro level 10. In this study the value of capital stock is measured by the data from Enterprise Surveys,� Working Paper, Enterprise Analysis replacement cost of machinery, vehicles and equipment. Unit, World Bank. 11. More countries were surveyed in 2006–2007 than in 2008–2009 and Solow, R. 1957. “Technical change and the aggregate production many of the countries in the 2006–2007 survey had sample sizes function.� Review of Economics and Statistics 39(3): 312-320. above 100 observations. Hence 200 observations is used as a cutoff only to make the graph easier to read. The Enterprise Note Series presents short research reports to encourage the exchange of ideas on business environment issues. The notes present evidence on the relationship between government policies and the ability of businesses to create wealth. The notes carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this note are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. 8