Report No. 40717-LAC Economic Performance in Latin America and the Caribbean: A Microeconomic Perspective (In Two Volumes) Volume I: Main Findings June 22, 2007 Finance and Private Sector (LCSPF) and Chief Economist Office (LCRCE) Latin America and the Caribbean Region Document of the World Bank LCRCE Officeofthe ChiefEconomist, LatinAmerica andthe Caribbean Region(WorldBank) LLC LimitedLiability Company M Imports MA Moving-average MENA MiddleEast andNorthAf5ca MEX Mexico MYS Malaysia NBFI NonBank FinancialInstitution NIC Nicaragua NTB Non-tariff barriers Nx Non-exporters Obs. Observations OECD Organisationfor Economic Co-operationandDevelopment OLS Ordinary LeastSquares PAN Panama PER Peru PISA Programfor InternationalStudent Assessment PRY Paraguay R&D ResearchandDevelopment SA SouthAsia SLV El Salvador SME SmallandMediumEnterprises SSA Sub-SaharanAfrica TFP Total FactorProductivity THA Thailand TI TransparencyInternational TOT TermsofTrade Trini. & Tobago TrinidadandTobago UK UnitedKingdom URY UWFY USA/ US/ U.S. UnitedStatesofAmerica USD UnitedStatesDollars USPTO UnitedStatesPatentandTrademarkOffice Venezuela, RB RepublicaBolivarianade Venezuela WDI WorldDevelopmentIndicators WDR WorldDevelopmentReport WEF WorldEconomic Forum X Exports ZAF SouthAfrica Vice President: PamelaCox ChiefEconomist: GuillermoPeny SectorDirector: Ernest0May SectorManager: LilyChu Task Managers: PabloFajnzylber J. LuisGuasch, and HumbertoLopez ii TABLE OF CONTENTS: Acknowledgements ............................................................................. Executivesummary ............................................................................. V vi Economicperformancein LatinAmerica andthe Caribbean: A microeconomic perspective I. Introduction.................................................................................. 1 111 A first passat the investment climate inLatinAmerica............................... I1 .. Investment and TFP in Latin America................................................... 6 IV. The investment climate inLatinAmerica: narrowingthe field...................... 8 13 V Getting some sense of policy priorities .................................................. 25 VI 31 VI11 What are the determinants offinancialaccess inLatinAmerica?.................... VI1 Corruption................................................................................... ....Improvingthe investment climate....................................................... 34 IX Innovationandtechnological development............................................. 38 43 X.. Exporters do itbetter................................................................ 46 ... 111 LISTOFTABLES: Table 1: Impactof the investment climate on labor productivity........................... Constraints to productivityand growth............................................... 15 Table 2: 18 Table 3: Impactof the investment climate on TFP.......................................... 20 Table 4: Impactof the investment climate on wages........................................ 21 Table 5: Education and the investment climate............................................... 23 Table 6: Top three priorities for firms movingto the 75* percentile o f same industry 30 Determinants ofbribe payments in Latin America................................ andfirmsize............................................................................. Table 7: 37 Table 8: 38 Access and court quality............................................................... Determinants ofbribe payments in LatinAmerica (11) ............................. Table 9: 42 Table 10: Are R&D expenditures related to productive innovation? Direct versus reverseregressions..................................................................... 45 Table 11: Productivityand wage exporter premiums in LAC................................ 47 LISTOFFIGURES: Figure 1: (deviations with respectto global trends)........................................... Per capita growth in LatinAmerica and the Caribbean 1963-2006 2 Figure 2: Per capita growth inLatinAmerica andthe Caribbean 1963-2005 3 Growth and investment (1970-2005)................................................ Gross capital formation (% of GDP)............................................. (deviations with respectto selected country groups)............................. Figure 3: 7 Figure4: 7 Figure5: TFP in Latin America Comparative performance................................. Vulnerabilities have been dramatically reduced................................... . 8 Figure6: 9 Figure 7: Cost o f inadequateinfrastructure.................................................... Businessenvironment. ................................................................ 11 Figure 8: 12 Figure9: 12 Figure 10: Firmswith I S 0 certification per million workers, LAC andthe OECD....... Domestic credit to the privatesector (% of GDP) ................................. 13 Figure 11: Contributionto growth in Latin America........................................... 25 Figure 12: Potentialproductivity gains associatedto different scenarios.................... 27 Figure 13: 28 Figure 14: Aggregate labor productivity gains, by firm size................................... Aggregate labor productivity andwage gains, by country........................ 31 Figure 15: FinancialAccess inLatinAmerica.................................................. 39 Figure 16: Financial Access in LatinAmerica (firm size) ...................................... 41 iv ACKNOWLEDGEMENTS Economic Performance in Latin America and the Caribbean: A Microeconomic Perspective is the product of a collaborative effort o f two units of the Latin American Region of the World Bank: the Chief Economist Office and the Finance and Private Sector unit of the Poverty Reduction and Economic Management Group. The report was prepared under the general guidance of J. Luis Guaschby a team led by Pablo Fajnzylber and Humberto Lopez and comprising Veronica Alaho, Carlos Caceres, Carlos Casacuberta, Wendy Cunningham, Nkstor Gandelman, Daniel Lederman, Norman Loayza, Inessa Love, Marcel0 Olarreaga, Ana Maria Oviedo, Guido Porto, Eliana Rubiano, and Naotaka Sugawara. Guillermo Beylis, Edwin Goiii, Tatsuji Hayakawa and Naotaka Sawadaalso collaborated with processing o fthe survey data. Excellent advice has been received from Guillermo Perry and especially from our peer reviewers, Alvaro Gonzdlez and Mary Hallward-Driemeier. Special thanks are also due to Luis Servkn for providing us with the econometric discussion in Chapter 2 for discussionsinthe author workshop. To all of them we are gratehl without implication. V EXECUTIVE SUMMARY Over the past 3 years, the Latin America and the Caribbean region has shown signs of significant economic improvement. Between the end of 2003 and the end of 2006 Chile, the Dominican Republic, Ecuador and Peru reached per capita growth rates above 3.5 percent whereas Argentina, Uruguay and Venezuela grew at annual per capita growth rates above 7 percent. Median per capita growth in the region in the last three years has averaged 2.7 percent and would now be at a level not seen since the mid 1970s when the region experienced growth rates of around 3 percent. Taking into account that over the last 30 years median per capita growth inthe region has been about 1.13 percent and has rarely averaged more than 2.5 percent over a 3 year period, this achievement is nothing short o f remarkable and could probably be interpreted as the result o f an acceleration in structural growth rates, perhaps due to the significant reforms implemented by a large numberof countries of the region over the decadeofthe 1990s. However, there is little reason to be complacent about the current situation. First, even if the region managed to sustain the current trends, the corresponding growth rates would be insufficientto reduce the current share of people living on less than US$2 a day (about 25 percent) over the short run. Second, while the region's recent performance is quite impressive by historical standards, so far it is only a three year run, driven, to a significant extent, by a number of external factors, such as a favorable global environment, high commodity prices, the voracious appetite for raw materials and intermediate goods from countries such as China andIndia and so on. Hence, it is unclear how much longer such a set o f favorable conditions will persist. Third, and perhapsmore important in this context, the improvements on the growth front observed in Latin America in recent years are less remarkable when considered in a global context. Infact, Latin America and the Caribbean is falling behind important reference groups such as the average Middle Income or East Asian country. Thus, despite the recent significant absolute improvements, LatinAmerica and the Caribbean are getting relativelypoorer. All these elementspoint to the fact that LatinAmerica andthe Caribbean is yet to live up to its potential in terms of economic growth and in matching the development achievements of other regions, such as East Asia, that have managed to halve poverty rates in the last decade. This regional study aims at contributing to the identification o f both the causes of that lackluster performance and the policy actions needed to improve the region's achievements. True, the past few years have witnessed the emergence of a considerable empirical literature that has focused on the determinants of economic growth in the region. In fact, there are a number of important lessons that emerge fi-om these studies regardingthe priorities that policy makers will have to tackle to unleash the growth potential of the region. For example, policy interventions aimed at improving the infrastructure of the region seem promising. Similarly, improving both the "quantity" and "quality" o f education also appears as promising particularly when complemented with a strong push for technological development and innovation, so as to move up on productivity andon the value chain. Other important findings o f recent studies basedon a macroeconomic perspective include the need to take actions aimed at hrther increasing vi the region's integration into the global economy while at the same time deepening Latin America's relatively shallow financial markets. This study complements that body o f knowledge by taking an alternative microeconomic, firm-level approach to explain productivity and economic performance. In particular, instead of the usual cross country econometric methodologies used to uncover the determinants of per capita GDP growth, we focus on the analysis o f the investment climate determinants of productivity and wages at the firm level. To that end, we take advantage o frecently collected Enterprise Surveys covering more than 10,000 firms from 16 Latin American countries. This microeconomic approach is warranted since, after all, the aggregate economic performance o f countries depends on firms' employment, investment and production decisions, which in turn are a reflection o f the market and policy environment in which firms operate. The report's micro perspective also allows for a better understanding of the linkages between the business environment and firms' responses, which facilitates the identification of where and how governments ought to intervene to improve economic performance. What are the main lessons emerging from our analysis? We would like to highlight six mainmessages. First, the governance agenda, encompassing the strengtheningof the rule o f the law and the improvement of institutions and regulatory frameworks, emerges as a key policy priority for improving firm performance in Latin America. This is reflected in robust links between, on one hand, low levels of corruption and crime and increased regulatory compliance and, on the other hand, higher levels of firm productivity andwages. In particular, we have given special attention to the possible determinants and policy levers associated with combating corruption in the region. The Enterprise Surveys indicate that about 40 percent o f Latin American firms report paying bribes when doing business with the government or requesting a public service, and there are countries where this figure is above 75 percent. Beyond the ethical reasons to fight corruption, the results in this report also indicate that the direct economic costs of corruption are large, with a strong negative impact on firm performance, workers' earnings and, therefore, on countries' potential growth and poverty reduction prospects. Moreover, the indirect effects are equally damaging since corruption adversely affects the effectiveness of government actions and programs directed at improving the various elements o f the business environment, for instance by correcting market failures. Our analysis indicates that in order to explain the corruption problem in the region one has to particularly look at the existing governance and regulatory framework, and at the credibility o f the judiciary. Countries should assess their regulatory frameworks to distinguish between regulations that are justified by public interests or market failures and those that have become anachronistic, or merely protect private interests and transfer rents to specific interest groups. Complementarily, countries should also consider modernizing their regulatory frameworks from an administrative and procedural perspective. Indeed, even well motivated and designed regulations may suffer from vii implementation problems, particularly if their enforcement is not even-handed or the administrative procedures needed for the private sector to prove compliance are costly or cumbersome. Thus, a good regulatory framework should, to the extent possible, provide minimal discretion to the responsible public servants and use information and communication technologies to reduce users' costs, increase transparency and reduce opportunities for corruption. Inthis context, administrative simplification, together with efforts to professionalize the civil service could be effective tools to lower the cost and improve the efficiency of regulations, and therefore lower the incidence of corruption. Incidentally, progress on the regulatory front could not no only help reduce corruption but also, as highlighted inthe last Flagship Report on Informality of the Latin American ChiefEconomist Office, contribute to reducing regulatory non-compliance inthe region. However, in order to eradicate corruption and informality, deterrent incentives may also be needed. In other words, whereas the appropriate remedial policy mix will clearly depend on country circumstances, reducing corruption will probably require a combination of "carrots" and "sticks," the latter being associatedwith increasedefforts to detect corrupt practices and regulatory non-compliance, and higher penalties in the form of fines and sanctions for those caught performing irregularities - including both public servants and firms. Similarly, improvements in the judicial system that increase the likelihood of punishment for all parties involved in corruption (i.e. other measures to increasethe expected cost o f entering in such type of agreements) are also likely to lead to lower corruption levels. Itis also worth mentioningthat progress on the regulatory andjudicial fronts could lead, as an important by-product, to improvements in corporate governance which in turn could positively affect firm performance. In particular, the evidence shows that businesses are more likely to be incorporated in countries with efficient legal systems, strong shareholder and creditor rights, low regulatory burdens and corporate taxes, and efficient bankruptcy processes. This is important in this context because our analysis shows that, even after controlling for a number of other factors, incorporated firms tend to operate more efficiently thanunincorporated firms. A second important finding i s that expanding access to credit also appears as a relevant policy priority. As discussed in the report, a well hnctioning financial system is an important component o f the investment climate and in fact, firms with access to credit tend to have significantly higher productivity, are more likely to innovate and less likely to engage in corruption. Moreover, the region seems to have significant room to improve on this front as it is well behind high income countries - and some other developing regions o f the world - both in terms o f financial deepening and in terms of access to financial services. Access problems seem to be particularly important for small and medium firms who have consistently lower usage of financial products than large firms. Interms of policy options to tackle this challenge, the report finds, first, that progress on the judiciary front would have a positive impact on increasing not only financial depth but also financial breadth and, second, that actions directed at increasing the former (depth) would also lead to progress in terms of the latter (breadth). In particular, the V l l l ... report's findings indicate that the quality of the courts has a consistent positive effect both on measures of access, such as the probability of having a checking account or using bank credit, and financial depth, such as the number of credit products. Relevant indicators ofprogress inthis front include increases in the usage of the courts for dispute resolution and increases in the share of cases that result in actual court judgments. In addition, the report finds that increased financial depth tends to be associated with increased loan maturity, larger size of loans relative to sales, and more likely use of land and buildings as collateral. Finally, we find that small and medium firms benefit more than their larger counterparts from increasesinloan size and maturity. Our third main finding is that innovation, technological development and humancapital remain at the core of the agenda o f reforms aimed at moving up in the value chain and securing the elusive goal of higher sustainedgrowth. Indeed, our findings corroborate the beliefs (and other existing evidence) of the critical role o f human capital and innovation for productivity, growth and poverty alleviation. With regard to human capital, the report combines data from Enterprise and Household Surveys to assess the impact of workers' educational levels on firm productivity. Our findings indicate that human capital, together with technology adoption and to a lesser extent training, has a significant impact on productivity. In light of those findings the report assesses the determinants of product innovation by firms. The main findings are that the investment climate matters for product innovation and specifically, that market failures appear to affect the propensity of firms to introduce new products, that indeed R&D and licensing efforts are correlated with product innovation, and that trade-policy distortions matter for productive innovation. In particular, the report shows that regulatory reforms that facilitate firm entry are desirable to increase competition and knowledge diffusion, they also have the effect o f facilitating imitation and thus reducing firms' propensity to introduce new products. This implies that, especially after trade and regulatory reforms, governments should consider the implementation of well monitored programs to stimulate private-sector product innovation. The report also shows that exports and the trade environment are significant determinants of firms' propensity to introduce new products. Inparticular, an increased incidence o f exporters is associated with a higher probability of undertakingproduct innovations by individual firms. Moreover, we show that low and concentrated -as opposed to dispersed -tarifflevelsleadto higher product innovation ratesatthe fmlevel. Thereport also sheds some light on the debate on the effects of FDIon product innovation. Inparticular, the results suggest that on the one hand foreign-owned firms are not more likely to introduce new products than their domestic counterparts in developing countries - they settle in these countries to produce specific products. On the other hand, foreign-owned firms are likely to produce goods that are different from those o f their local competitors - even ifthey are not new on a global scale -which may lead to spillovers effects that help local f m s introduce new products. i x Fourth, the findings of this regional study complement the extensive evidence on the critical impact of infrastructure on growth and poverty. While the Enterprise Surveys are not optimally suited to evaluate the impact of infrastructure on firm productivity, they provide evidence on the significant direct costs to firms of operating with poor public infrastructure. For example the survey shows that the direct and indirect costs of power outages represent about one percent of annual firm's sales. Other transport and trade facilitation factors contribute towards the high logistic costs faced by firms in Latin America, near 25 percent o f product value. All these costs adversely affect output levels and render certain products and firms uncompetitive, thus limiting growth and competitiveness. Improving the delivery and efficiency of infrastructure services and modernizing the corresponding legal and regulatory frameworks are thus key priorities for Latin America. Inparticular, inorder to secureuniversal coverage and support moderate to highgrowth, investment levels need to increase considerably from the current 2 percent to levels of about 4 to 6 percent of GDP. Moreover Latin America needs to increasethe efficiency o f its expenditures in the area of infrastructure and in this regard improved governance and modern regulatory frameworks play a critical role, so as to ensure the right selection of projects, proper evaluation and oversight, and transparentprocurement processes. F$h, the report shows that, after controlling for other factors, firms that export tend to perform better: they are more productive, pay higher wages and are more likely to innovate. As a result, reorientation of economic activity towards the export sector should lead to higher growth, a finding that is in line with an extensive strand of the academic literature that argues that countries that are more open to trade tend to grow faster. This area is particularly important in the region because there are a large number of countries that are negotiating bilateral trade agreements with the US and the European Union and that are relying on these agreements tojump start economic growth. From a policy point of view, however, the report shows that it is important to distinguish between two non-mutually exclusive explanations for the existence of exporter premiums. First, it is possible that better firms with high levels o f productivity are the ones that tend to become exporters - the so-called "self-selection" hypothesis. On the other hand, however, the correlation between productivity and exports could be due to the fact that firms could become more productive as a result o f engaging in exports - they would "learn by exporting". Incases where the self-selection explanation is the dominant one, policy makers should focus their efforts on the internal determinants o f productivity growth, On the other hand, if learning-by-exporting is the dominant channel linking exports to productivity, governments ought to invest in export promotion activities. Moreover, in this case, particular emphasis should be placed on large and medium domestic firms that are not yet exporting, and on the promotion ofexport consortia that allow smaller firms to export. The evidence presented in this report indicates that self selection is much more generally observed, which implies a focus on actions aimed at improving the determinants of productivity growth. However, direct export promotion policies could be X more relevant ineconomies with small domestic markets, where we find that learning-by- exporting tends to be important, probably because ofunexploitedeconomies o f scale. Finally, our sixth main finding is that the benefits of an improved investment climate are not limitedto firms, as their workers benefit significantly as well. Indeed, we find that in the context of high levels of corruption and crime, or where governance and institutional quality issues lead to high levels o f regulatory non-compliance, firms tend to pay lower wages per worker. On the other hand, audited and incorporated f m s , those with access to the financial sector and companies with better technology appear to pay higher wages. In consequence improving the investment climate should be taken into account in the design and implementation of poverty reduction and development strategies. Similarly on a related matter, our results also indicate that the much feared globalization process, seems to play in favor or workers. Controlling for other factors, exporters tend to pay higher wages than non exporters, and so do foreign owned firms incomparison with their domestic counterparts. This report contributes to the understanding o f the determinants of economic performance and the appropriate policies to improve it. The report's value added is derived from the adoption of a microeconomic perspective, and the use o f fm level data for analyzing the determinants of firm behavior and performance. In some cases our analysis stresses areas that have also been stressed by others such as the need to make progress on the trade openness and financial deepening fronts, or the importance of fostering technological development and innovation. The report's main contribution in this regard is related to an attempt to better understand the policy options that are available to makeprogress along those fronts. Inparallel, we also stress other aspects that have received less attention and that yet we do believe are keeping the region behind. More specifically, our results first suggest that the prevalence o f the rule of law, as reflected in low levels o f corruption and crime, is a critical element of the investment climate, in that it has a robust impact on firm productivity and wages. Similarly, good governance and regulatory frameworks, as illustrated by low levels o f regulatory non-compliance and a highproportiono f firms that chose the incorporated form and/or have externally audited financial statements, are associated with higher levels of productivity and wages. These robust links between economic performance and the quality of governance and institutions make the reform agenda to improve Latin America's standings in these areas a top priority to unleash the region's growth andpoverty reduction potential. xi EconomicperformanceinLatinAmerica and the Caribbean:A microeconomic perspective I.Introduction Do we need another growth study focusing on the Latin America region? After all, over the past three years, the Latin America and the Caribbean (Latin America hereafter) economy has been performingvery nicely. Between the end of 2003 and the end of 2006 Chile, the Dominican Republic, Ecuador and Peru reached per capita growth rates above 3.5 percent whereas Argentina, Uruguay and Venezuela grew at annual per capita growth rates above 7 percent. Median per capita growth in the region in the last three years has averaged 2.7 percent and would now be at a level not seen since the mid 1970s when the region experienced growth rates of around 3 percent. Taking into account that over the last 30 years median per capita growth in the region has rarely averaged more than 2.5 percent over a 3 year period, this achievement is nothing short of remarkable and could probably be interpreted as the result of an acceleration in structural growth rates, perhaps due to the significant reforms implemented by a large number of countries of the region over the decade of the 1990s. So, does this mean that policy makers and more generally development practitioners focusing on Latin America should not be concerned with economic growth any longer? The truth is that there are few reasons to be complacent about the current situation. First, even if the region managed to sustain the current trends,' the corresponding growth rates would be insufficient to achieve fast poverty reduction over the short run. According to Perry et al. (2007), today LatinAmerica's (headcount) poverty rate (using a US$2 PPP a day poverty line) is close to 25 percent. This, together with an average growth elasticity of poverty o f around 1.4 for the region (see Gasparini et al., 2007) would imply that on current trends the region would need about 25 years to achieve a poverty rate that is below 10percent. The needto achieve faster growth rates is even more pressingwhen we rely on national statistics to measure poverty, in which case the regional poverty rate would be around 40 percent (see Figure2.1 inPerry et al., 2007). Second, while the region's recent performance is quite impressive by historicalstandards, so far it i s only a three year run driven, to a significant extent, by a number of external factors, such as a favorable global environment, high commodity prices, the voracious appetite for raw materials and intermediate goods from countries such as China and India and so on. Inthis regard, it is unclear how much longer such a set of favorable conditions will persist. 'Accordingto the World Bank's 2007 GlobalEconomic Prospectsper capitagrowthis expectedto levelat the 2006 levelin2007and2008. 1 Third, andperhapsmore important inthis context, the improvements on the growth front observed in Latin America are consistent with those observed in the rest o f the world. Figure 1plots the deviations of Latin America's annual growth rates with respect to the median growth rate of the world (excluding Latin America).* When this deviation is positive (negative) it implies that the region was growing faster (slower) than the typical country in the world (as captured by the median). Thus this figure presents Latin American growth estimates fiom which we have somewhat eliminated the effect o f the internationalbusiness cycle. Figure1. Per capitagrowthinLatinAmerica andthe Caribbean1963-2005 (deviationswith respectto globaltrends) 3.0 1 2.0 1.o A 0.0 -1.0 -2.0 -3.0 -4.0 I -5.0 U -6.0 I Note: Annual mediangrowth rate ofthe regionminusannualmedian growth rate ofthe rest ofthe world. The series havebeensmoothedwith abackward looking 3 year MA. Source: Own calculations basedon data from WDI Inspection of Figure 1 suggests several interesting facts. First, even in the 1960s and 1970s when Latin American growth rates were at historical maxima, the region was falling behind with respect to many other regions (i.e. it does not seem to have fully exploited the opportunities presented by the global economy). Second, for a number of years duringthe lost decade o f the 1980s, the region experienced growth rates that were 4 to 5 percent lower than those observed globally further contributing to the backwardness of the region. Notice that this would imply a cumulative loss of about 25 percent o f GDP for the median country in the region with respect to the median country in the world. Third, after the recovery of the early 1990s, the relative performance of the region with respect to the rest of the world has again been disappointing. In fact, over the past 10 years Latin America has been loosing ground with respect to the median country o f the rest o f the world at an averageo f 1percent per year. In other words, the current situation seems to somewhat mirror that in the 1960s and 1970s when despite having high growth rates by historical standards the region was falling behind and at least inrelative terms getting poorer. The results remain basically unchanged when comparisons are based on the means of the respective groups ratherthanonmedians. 2 True, the previous discussion could be biased because o f the reference group chosen. After all, the "rest of the world" includes developed and developing countries; countries rich in natural resources and net importers of commodities, countries that have experienced important crisis over the past decade and economies that have had smooth rides. To explore this hrther, Figure 2 presents the equivalent to Figure 1 but using different reference groups. The figure compares the performance of the Latin American region against the OECD (panel A); the group o f middle income countries (panel B); a region that went through a profound crisis inthe late 1990s such as East Asia (panel C); and a group of natural resource abundant economies3outside the Latin American region (panel D). Figure2. Per capitagrowthinLatinAmericaand the Caribbean1963-2005 (deviationswith respectto selected countrygroups) PanelA. OECD PanelB. MiddleIncomeCountries II ,I 8 4.0 3.0 2.0 2.0 1.O 1.o 0.0 0.0 -1.0 -1.0 -2.0 -2.0 -3.0 -3.0 -4.0 -4.0 -5.0 -5.0 -6.0 -7.0 ' I7 PanelC. EastAsia PanelD. NarturalResourcesAbundant 4.0 4.0 3.0 1 2.0 1.o 1 .o 0.0 0.0 -1.o -1.0 -2.0 -2.0 -3.0 -3.0 -4.0 V -5.0 II -6.0 -6.0 -7.0 -7.0 I I Natural resource abundant countries are selected on the basis o f the Leamer index and more specifically o f the average index over 1990-1999 for each country. Whenever this index is positive it implies that net exports ofnatural resources are positive andtherefore the country is classified as resource abundant. 3 Figure 2 suggests a picture that at least for the last few years and for some o f the reference groups is more nuanced that the one emerging from Figure 1.4 As Latin America's growth rates have recovered after 2003, the region seems to be closing the existing gap inper capita growth rates with respect to East Asia and the natural resource abundant economies and in the case of the OECD to have already closed it. Thus once again this figure indicates that the recent recovery ingrowth rates has to be recognized. Yet, despite these gains, the reading of the different panels is that in relative terms (i.e. taking into account developments in other regions of the world) the growth rates that we are observing in the region these days cannot be considered exceptional at all and that in fact there is still ample room for fitrther improvement, particularly given the significant differences in initial conditions. Moreover, as noted in Chapter 1of Volume 11, with the exceptions of Brazil and Paraguay in the 1970s and Chile in the late 1980sand 1990s, it i s difficult to find episodes where a Latin America country has been performing above global trends in a sustained fashion. Inother words, the lack of growth sustainability has been a trademark of the LatinAmerican region. Against this background, this regional study contributes to the understanding o f the growth process in Latin America. There are a large number of studies o f economic growth in Latin America. Just to name a few among the most recent: FernandezArias and Montiel (2001), Calderdn and Schmidt-Hebbel (2003), Calderdn and Serven (2003), De Gregorio and Lee (2003), Blyde and FernandezArias (2004), and Loayza, Fajnzylber and Calderdn (2005). As a result o f these efforts we have learned important lessonsregarding the role played by factors such as human capital, public infrastructure, financial development, or openness to trade inthe growth process. However, most of the existing literature on the topic has relied on cross country regressions and so far, at least in the Latin American context, much less effort has been devoted to understanding the constraints faced by firms to invest and innovate. After all, aggregate productivity is not only the result of aggregating the productivity o f individual firms but also one o f the main elementsbehindeconomic growth (Loayza, Fajnzylber and Calderdn, 2005). Moreover the use of a micro-firm level approach i s warranted since firms are the ones directly responding to their environment, using and managing factors of productions, creating employment and producing output. Analyzing the actions and outcomes at the firm level allows for a better understanding of the causality between, on one hand, the market and policy environment in which firms operate and, on the other hand, f m s employment, investment and production decisions. In addition, a better understanding of the linkages between the business environment and firms' responses facilitates the identification of where and how governments ought to intervene to improve economic performance. In fact, even if the "macro context" is in good shape (which is arguably the case in LatinAmerica) it could be that the "micro context" that determines the ability of firms to thrive still needs attention (which as shown in this regional study appears to be the case inLatinAmerica). Between the early 1960sand the late 1990sFigure 2 is to a large extent consistent with Figure 1. With the exception of the natural resource abundant countries in the 1960s and 197Os, and a few years in the early 199Os,LatinAmerica hasbeenconsistentlyloosinggroundwith respectto almost anyreferencegroup. 4 With this motivation, unlike the above mentioned growth studieswhich use cross country econometric methodologies to uncover the determinants of per capita GDP growth in the region, this report focuses on the analysis of the investment climate determinants of productivity and wages at the firm level. To that end, we take advantage of Enterprise Surveys prepared by the International Finance Corporation (IFC) of the World Bank covering more than 10,000 firms from 16LatinAmerican countries. The study is structured in two volumes. The present Volume summarizes the report's main findings, which are described in detail in the seven chapters o f Volume 11. In Chapter 1we motivate the report with a review of how the region fares in a number o f aggregate indicators such as investment, the relation between investment and growth, and total factor productivity. One conclusion that emerges from this chapter is that not only the region has relatively low levels of investment, but the impact of those investments appears to havebeen rather low ina number of countries. A second conclusion is that the region also scores rather low on its capacity to efficiently use and improve inputs o f production, as revealed by relatively low rates of total factor productivity growth. In our view both these elementsjustify a carefbl look at the investment climate -defined as the factors that shape the opportunities and incentivesforjrms to investproductively, create jobs and expand (World Development Report, 2005) in the region. Infact, a first pass at four broad areas o f the investment climate (governance and regulation and institutional quality; infrastructure; access to finance; and education, innovation and technology) reveals that the regionhas substantial roomto improve. Chapter 2 further delves into the investment climate issue, but the focus here is on estimating econometrically the main elements of the investment climate that are behind firm performance. To that end, the chapter exploits the information contained in Enterprise Surveys performed for Argentina, Bolivia, Colombia, Brazil, Chile, Costa Rica, Ecuador, El Salvador, Guatemala, Honduras, Mexico, Nicaragua, Panama, Peru, Paraguay and Uruguay. The chapter also looks at whether the benefits o f the investment climate extent beyond firms and explores the impact o f investment climate attributes on wages. Chapter 3 provides country specificity. Buildingon Chapter 2, it presents the results of a "what if' exercise where we simulate the impact on firm productivity of sharp increases ina number of investment climate areas. Whereas a number of caveats should be kept in mindwhen performing this type of simulations, the results are usefbl for discussing the prioritizationof investment climate reforms at the country level. Chapters 4 to 7 move beyond identifying the main investment climate determinants of firm performance in LatinAmerica, to investigating the policy levers that could be used to act on a number of those determinants. Specifically, we focus on corruption, access to finance, technology and innovation, and the role o f the external sector. We do not investigate other important investment climate determinants such as infrastructure and informality since these two areas have been the focus o frecent studies ay the World Bank (Calderdn, Easterly and Servkn, 2003; Guasch, 2004; Perry et al, 2007) 5 InChapter 4 we look at the incentives anddeterminants of corruption usinga framework similar to that in Svensson (2003), where corruption patterns are explained by a combination o f hypotheses: control rights (where corruption is the result of public officials exploiting the regulatory framework to gain bribes), bargaining strength (where the amount o f bribes paid by firms are positively related to the profitability of the firm), and "grease the wheels" effects (where firms bribe officials to obtain businessgains). Similarly in Chapter 5 we look at the determinants of firm access to financial services. A well fbnctioning financial system is an important component of the investment climate and hence of a growth strategy. Yet, few studies have looked at the determinants o f access to credit. Questions addressed in this chapter include: which firms tend to lack access and hence could benefit from targeted policies? What is the relationship between financial deepening and financial access? What is the role playedby the legal system?To the best of our knowledge this is one ofthe few attempts at looking at these issues. Chapter 6 focuses on innovation at the firm level and exploits an expanded database o f Enterprise Surveys from 36 countries. More specifically, the chapter explores the determinants of firms' innovation patterns or the elements that may lead a firm to introduce new products and the barriers that can be found in trying to do it. This is a critical area for policy makers interested inboosting productivity and growth bymeans of accelerating the path of technological progress and moving up the value chain. The chapter adopts a broad view (a reflection o f the many areas playing an important role) to cover aspects related to regulation (especially entry and exit), quality o f institutions, investment inR&D, fm ownership, trade environment, etc. Chapter 7 looks at the external sector. One result that typically emerges from cross country growth studies is that there exists a positive relationship between exports and growth. More recently, the empirical literature on the micro-determinants of export activity at the firm level has found that firms that export are "better": they have higher productivity growth, hire more workers (particularly skilled workers), and pay higher wages (see for example Chapter 2 o f Volume 11). However, despite different attempts to correct for potential reverse causality from productivity and wages to exports, there is limited knowledge on whether the robust link between those two types o f variables i s driven by the fact that exporting improves firm performance, or by the fact that better performing firms are the ones that tend to become exporters. Chapter 7 looks at this issue and generates evidence on the causality of exports to wages using a country case study approach. 11. InvestmentandTFPinLatin America As shown in Figure 3, the only geographical region with investment rates (as measured by the ratio of gross capital formation to GDP) below those o f Latin America in 2004 is Sub-SaharanAfrica which, not unlike Latin America, exhibits investment ratesthat hover around 20 percent. At the other extreme, we have East Asia and the Pacific with investment rates close to 35 percent o f GDP, followed by the Middle East and North M i c a where gross capital formation representsbetween 25 and 30percent o f GDP. Latin 6 America's investment levels are also lower than the averages rates found inboth low and middle income countries whereas they are comparable to those of high income countries. I Figure3. Gross capital formation (% of GDP) 4 0 1 EM: East Asia and Pacific; ECA: Europe and Central Asia; LCR: Latin America: MENA: Middle East and North Africa; SA: South Asia; SSA: Sub-Saharan Africa. Data refer to 2004. Source: WDI2006 Figure4. Growth and investment (1970-2005). PanelA. Latin America PanelB. Rest ofthe world 5 6 g 4 .+5 2 cE -2 O p, -4 4 -8 -8 GCF/GDP GCWGDP I' Source: Owncalculations using WDI data. Moreover, not only are Latin America's investment rates low but they also seem to generate a relatively small pay-off in terms of increased growth. In fact, investment and GDP growth rates over 1970-2005 do not show any significant positive correlation (Figure 4, Panel A). This is in contrast to what is observed when we repeat the exercise usingthe rest of the world (Figure 4, PanelB)where investment and growth appear to be positively correlated. Complementing this finding, the report shows that Latin America's total factor productivity (TFP) performance has also been relatively weak. Although in principle there are a number of caveats that apply to standardTFP growth measures (Le. 'Chapter 1 of Volume I1presents similar results for different subperiods. 7 to the different versions o f the Solow residual), our calculations (Figure 5) show that with the exception of the 1971-75 and 1991-95 periods, Latin America's TFP growth rates havebeen below those of most o f the other regions. Figure5. TFP in LatinAmerica. Comparativeperformance. 6 1 71-75 76-80 81-85 86-90 91-95 %-00 01-05 ILACeaMENA0EAP SA 61 SSA IOECD Source: Own calculations. In summary, as noted above, we find that Latin America performs rather poorly on the two factors that may explain differences inper capita growth rates across countries. First, the region's aggregate investment rates are relatively low and exhibit a small pay-off in terms of increasing growth. Second, beyond factor accumulation, the region also scores rather low on productivity growth. 111. A first passat the investmentclimateinLatinAmerica There is no doubt that from a macro perspective, Latin America has made significant progress over the past few years. Comparing the situation today with that inthe early and mid 199Os, Latin American countries have better macroeconomic frameworks, are more open to trade, have more developed financial sectors, and invirtually all countries public spending in health has increased and education coverage has expanded. Although this should not be understood as indicating that there is not progress still to be made in these areas (e.g. openness to trade in Latin America is still very low by any standard that we use and financial markets are very shallow), those significant improvements in the macroeconomic framework have inall likelihoodcontributed positively to GDP growth. For example, as shown in Figure 6, the average primary fiscal balance in the region has increased from 0.1 percent of GDP in 2002 to 1.5 percent of GDP in 2006. Similarly, public debt has been declining since 2002 and would now be around 40 percent of GDP. On the external front, reserves have increased from 35 percent to 40 percent of M2 whereas the current account balance has moved from a 2 percent deficit over 1990-97 to a 1 percent surplus over 2003-06. As a result, the region is today much less vulnerable thanwhat itusedto be a few years back. 8 Figure6. Vulnerabilitieshavebeen dramaticallyreduced PanelA. FiscalBalance PanelB. PublicDebt Q 2 - 1 - % 0 - I, 8 50 4-1- g40 g8 - -2 Y 30 a -3 - -4 - k 20 -5 J 10 o . . . . . . . . . . . . . . . . . moverall Balance Primary Balance ~~ ~ PanelC. Reserves PanelD. Current account balance 50 5 - 45 4 - 40 fl2 8 3 - 30 35 73 2 - % 25 28 20 %I- I 15 f 0 I O 2 - 1 - 5 -2 - 0 -3 - EAP LAC ECA OECD ECA LAC EAP W 1990-97 W 2003-06 1990-97 2003-06 But a strong macroeconomic framework is not sufficient to predict sustained high growth. It needs to be complemented by a good microeconomic environment driven by reforms in the broad investment climate regime. Unfortunately, however, those themes have not received the attention devoted to macroeconomic stability issues, and the corresponding policy agenda has developed very unevenly in the Region. It is only recently that countries have begun to focus on microeconomic reforms and the improvement o f the investment climate, in the pursuit of that elusive sustained growth. All of the usual estimations and rankings show the Latin America region lagging in microeconomic reforms and well below the average (WEF 2006, DBR 2006). Given the large development impact of those reforms (WDR 2005), it can be argued that the poor investment climate in Latin America could be responsible for its weak overall economic performance, and in particular for its relatively low rates o f investment and total factor productivity growth. Clearly, in principle there is a wide range o f microeconomic factors that may affect productivity and investment, so that ifwe are going to focus on the investment climate it may help structuring the discussion around a reduced key number of areas such as: (i) the quality of governance, institutions and regulations, (ii)the availability and quality o f physical infrastructure, (iii)the development of the financial sector and firms' access to 9 financial services, and (iv) education, training and the national innovation system. We now discuss the rationale for focusing on these areas, payingspecial attention to the Latin American context. III.1Governanceandinstitutionalquality The quality of governance and institutions can be addressed from several perspectives focusing for instance on the prevalence o f the rule of law, as reflected inthe incidence o f corruption and crime andthe efficiency of thejudicial system, or on the nature and extent of regulatory burdens imposed on the private sector. Inprinciple, all of these areas may have an important potential impact on firm performance, which operates through increases in firms' transaction costs and by means o f limiting firm entry and distorting competition. Inpractically all o fthese areas, however, LatinAmericaperforms poorly. For example, ifwe look at the Ease o f Doing Business Indicators produced by the World Bank, Latin America is the region with the largest number of both procedures (12) and days (66) required to start a new business (Figure 7). This compares with 9 procedures and 56 days in East Asia, or 7 procedures and 24 days in high income countries. Even Sub-Saharan Africa performs better than Latin America with respect to the number o f procedures (11) and days (64) required to start a business. Latin America also scores poorly when we focus on the time required to enforce contracts - 470 days, compared to 423 in East Asia and 282 in high income countries - and in the financial and ownership information that companies are mandated to disclose to the public (a measure o f investment protection). Indeed, the average LatinAmerican country obtains a score of 4 inthe DoingBusinessDisclosure Index, compared to 5 for other low and middle income countries and6 for high income countries. Similarly, according to the IMF, Latin America is the region with the highest (i)total number of taxes paid by businesses (49); (ii) number of hours per year neededto prepare, fillanpaytaxes(549); and(iii) paidas apercentageofprofits (54.5 percent). This taxes compares to 40 taxes, 398 hours and 48.5 percent of profits in the average low and middle income country, and 18 taxes, 181hours and 38.8 percent ofprofits inthe average highincome country. 10 Fimre 7. Businessenvironment Panelk Procedures to start a business Panel B. Days required to start a business i o- 12 - 60 - 10 - I I Panel C. Days requiredto enforce a contract Panel D.Disclosure index* 500 1 6l 300 "11.11111, 200 250 .- c, I * Thedisclosure index measures the degree to which investorsareprotected throughdisclosure of ownership andjnancial information. It rangesfiom 0 to 7and higher values indicate more disclosure. Source: WDI 111.2Infrastructure As documented by Calderbn, Easterly and Serven (2003), the 1980s and 1990s saw a widening of the infrastructure gap (both in terms of quantity and quality) between Latin America and successfbl East Asian countries. This is particularly important in this context because there are several studies that have found a strong cross-country association between investments in infrastructure and economic growth (Easterly and Rebelo, 1993; Esfahani and Ramfrez, 2003). Inaddition, infrastructure has been found to have a significant contribution to output, with rates o f return that are sometimes found to be larger than those o f non-infrastructure capital (Gramlich, 1994, Calder6n and ServCn, 2004a). Similarly, Guasch(2004) has shown that poor infrastructure and limited transport and trade services tend to increaselogistics costs-which are about 25 percent ofproduct value in LatinAmerica versus 9 percent inOECD countries (Figure 8) - as well as render otherwise competitive products uncompetitive and limit rural production and people's access to markets, which adversely affects poverty and economic activity. 11 Figure8: Cost of Inadequate Infrastructure 35 - 30 - 25 - 20 - 15 - 10 - 5 - 0 -- I I I Losses en route ('YOof Logistic costs ('YOof Inventory levels (index agric. goods not value of products) OECD=lO) reachingthe market) IIOECDOLAC~ Source: Guasch(2005) 111.3 The development of thefinancial sector Well hnctioning financial systems ameliorate the problems created by information and transaction costs and help allocate resources across space and time. Not surprisingly, a large number of studies have found a robust relationship between the level of financial development and long-run growth (Levine, Loayza and Beck, 2000; Loayza, Fajnzylber, and Calderdn 2005). And again in this area Latin America seems to be well behind not only the high income countries but also most o f the other regions of the developing world, at least interms of the ratio of domestic credit to GDP (Figure 9). Figure 9. Domestic credit to the private sector (% of GDP) I lSO 1 160 14.0 120 100 80 60 40 20 0 I E4P ECA LAC MENA SA SSA High Income I11990120041 Source: WDI III.4 Education, Innovation and Technology Innovationand human capital are major drivers of growth and productivity and as noted above a large share of cross-country differences in per capita income are driven by differences in total factor productivity, generally attributed to technological development 12 and innovative capacity (Hall and Jones, 1999; Dollar and Wolf, 1997). As noted by Prescott (1998) one of the main candidates to explain large international income and productivity differences is resistance to the adoption of new technologies and to the efficient use o f current operating technologies, which in turn is conditioned by the institutional and policy arrangements each society employs. Moreover, productivity growth is intrinsically related to the availability of a skilled workforce and to the extent to which countries invest in research and development (R&D), either to generate new technologies, to absorb innovations generated by others (De Ferranti et al., 2003; Lederman and Maloney, 2003), or to counteract the forces of diminishingreturns inother accumulable factors such as physical capital (Lucas, 1988). Once again, however, the available evidence indicates that the Latin America region has considerable progress to make also in this area. For example, Latin American countries invest less than 0.5 percent of GDP on R&D whereas most OECD countries have R&D investment rates o f around 2 percent. Similarly on the educational front, the region has made significant progress lately, but yet there manyissuesregarding quality ofeducation. Onproduct quality, Figure 10shows the number of Latin America and OECD firms with I S 0 9000 and 14000 certification, per million workers. Usingthat as a proxy for quality adoption, we see that Latin American countries are way behind, with about one tenth of the number of certified firms in the OECD (after normalizing by the size of the labor force). This particularly worrisome considering the evidence that firms' investments in quality upgradingare often the entry point into innovation. Figure 10: Firmswith I S 0 certification per millionworkers, L A C and the OECD 165 I I I I I I S 0 9000 I S 0 14000 I S 0 14000 Source: I S 0 (2005) IV. The investmentclimate inLatin America: narrowing the field The previous discussion suggests that Latin America could benefit from improvements in several attributes of the investment climate. However, in practice not all o f those attributes where the region is lagging are likely to be equally important. Just as an example, it would be difficult to expect that natural resource abundant economies like those in Latin America invest in R&D as much as the East Asian countries that mainly 13 focus on manufacturing activities (Maloney and Rodriguez-Clare, 2006).6 That is, different investment climate attributes may weight differently for firms in different sectors and country contexts.' Thus one natural question in this context might be: Why notjust ask firms what are the main problems they face with the investment climate? Infact, the idea o f directly asking f m s i s the underlying premise of the Enterprise Surveys prepared by the International Finance Corporation (IFC) o f the World Bank (see http:llwww.entemriseswevs.org/). These surveys, which were launched in 2001 under the name Investment Climate Surveys*, cover 94 countries and over 26,000 firms and measure "business perceptions on the biggest obstacles to enterprise growth, the relative importance of various constraints to increasing employment and productivity, and the effects of a country's investment climate on its international competitiveness." More specifically, Enterprise Surveys collect data from managing directors, human resource officials and other relevant staff of a sample o f companies on more than 200 firm aspects (some of which are subjective assessments of constraints facing firms and some of which are objective data related to the firm) on a variety o f topics ranging from bureaucracy and corruption, courts and crime, access to finance, informality, infrastructure, innovation, taxes, and international trade. Chapter 2 exploits the information contained in the 16 Latin American surveys to determine which investment climate attributes are the most important determinants of firmperformance inthe region. Table 1 summarizes the countries' top 5 obstacles as perceived by firms and reveals that some "obstacles" tend to repeat inalmost every country. For example, corruption appears to be a priority in 15 countries (all but Chile) and in 10 of the countries it is either the first or the secondpriority. Somewhat surprisingly, macroeconomic instabilityalso appears as an important priority in all 16 countries. This despite the progress made by the region on the macro front noted above. Other areas that tend to be mentioned frequently by Latin American firms as major or very severe obstacles to growth are informal competition (13 countries), tax rates (9 countries), and the availability and/or cost of finance (8 countries). In six countries the latter is the obstacle mentioned most frequently by firms. Also mentioned, although less frequently, are issues related to infrastructure (electricity) and crime - each variable appearing among the top issues inrespectively 6 and 5 countries. 'Note, however, that a number o f natural resource abundant economies such as Norway, Finland, New Zealand and so on have made a very successful transition to technologically developed economies through a coherent government strategy that has led to major sustained investments in R&D, oRen surpassing 3 ercent o fGDP. It must be noted though that for the case o f R&D Maloney and Rodriguez Clare (2006) note that even correcting for output composition they cannot rule-out an innovation shortfall inthe case ofLatinAmerica. 'In turn, Investment Climate Surveys builton the WorldBusinessEnvironment Surveys launchedin 1999. The latter reliedon smaller samples andmainly on perception aspects. 14 T ]le I.Constraintsto ProductivityandGrowth Countrv Year Firms' Perceptions:Major Obstaclesto Growth Argentina 2006 0 Taxrates 0 Corruption Macroeconomic instability Education Labor Regulations Bolivia 2006 0 Corruption 0 Practiceso finformal competitors Macroeconomic instability 0 Transportation 0 Electricity Brazil 2003 0 Access to finance (availability & cost) Taxrates Macroeconomic instability Corruption 0 Tax administration Chile 2004 0 Accessto finance (availability & cost) Labor Regulations Macroeconomic instability 0 Education Practiceso finformal competitors Colombia 2006 0 Practiceso finformal competitors Corruption Electricity Macroeconomic instability 0 Taxrates CostaRica 2005 0 Accessto finance (availability & cost) 0 Macroeconomic instability Practiceso finformal competitors 0 Corruption Taxrates Ecuador 2003 0 Access to finance (availability & cost) 0 Macroeconomic instability 0 Corruption 0 Practiceso finformal competitors 0 Tax rates ElSalvador 2003 0 Crime 0 Practicesofinformal competitors Access to finance (availability & cost) Corruption 0 Macroeconomic instability 15 Tablc 1. Constraints to Productivity and Growth (cont.) Country Year Firms' Perceptions: Major Obstacles to Growth Guatemala 2006 0 Corruption 0 Electricity 0 Macroeconomic instability Practiceso finformal competitors 0 Crime Honduras 2003 0 Accessto finance (availability & cost) 0 corruption 0 Crime 0 Macroeconomic instability 0 Practiceso finformal competitors Mexico 2006 0 Corruption 0 Macroeconomic instability 0 Taxrates 0 Practiceso finformal competitors 0 Electricity Nicaragua 2003 0 Access to finance (availability & cost) 0 Corruption 0 Macroeconomic instability Practiceso finformal competitors 0 Crime Panama 2006 0 Electricity 0 Corruption 0 Crime 0 Taxrates 0 Macroeconomic instability Paraguay 2006 0 Corruption 0 Practiceso finformal competitors 0 Macroeconomic instability 0 Education 0 Access to finance (availability & cost) Peru 2006 0 Corruption 0 Practiceso finformal competitors 0 Macroeconomic instability Tax administration 0 Education Uruguay 2006 0 Taxrates 0 Practicesofinformal competitors 0 Macroeconomic instability Corruption 0 Electricity Source: OwnE xlationsusingEnterprise Surveys. Clearly, whether the perceptions of the firms correlate with the performance of individual firms is another issue. Thus in Chapter 2 we explore econometrically how different investment climate attributes affect firm performance. The econometric methodology used in the chapter builds on the work of Escribano and Guasch (2005) and Escribano, Guasch, and de Orte (2006) for Guatemala, Honduras and Nicaragua, and Chile 16 respectively.' However, we refine those works by restricting the initial potential set of explanatory variables to those that are both objective and can be interpreted unambiguously. Moreover, rather thanworking on a country by country basis we pool the data o f the different firms together. This is relevant in this context because if investment climate conditions affect all the firms in a country in a similar fashion, then it is likely that individual analyses will not be able to capture the relevance of that particular investmentclimate attribute. Table 2 presents the results of our basic econometric exercise undertaken in Chapter 2. The key findings are that, as expected, the four broad areas o f the investment climate discussed above have an impact on the performance of Latin American firms. More specifically we find that the rule of law appears as a critical element o f the investment climate. In our econometric models there are three variables pointing in that direction. One is corruption, measuredthrough f m s ' payment of bribes to secure public contracts, and to obtain public services (electricity, telephone, etc.). This should not be surprising because corruption is costly and highly distortionary" (Shleifer and Vishny, 1993). The two other variables related to the rule of law are the share of sales lost to crime and the preventive security expensesby firms. We also find evidence o f the important role played by the regulatory framework. For example, we find that the level o f regulatory compliance (as measured by the share of sales declared for tax purposes) has a negative effect on both labor and total factor productivity. Note that the degree of regulatory compliance is likely to be capturing several elements such as the quantity and quality o f regulations and the ability of public officials to enforce existing regulations in a non-discretionary way. As discussed below the regulatory framework is also an important determinant of corruption, so that progress inthis areawould likely have a double pay-off interms of firm performance: by directly improving firm productivity and through its positive effects on other investment climate attributes that also affect firm performance. Similarly, if we think of the proportion o f firms with external audits or that choose the corporate form as reflections of a country's legal system's efficiency, strong shareholder and creditor rights, and low regulatory burdens," then our results also suggest that a better regulatory framework will lead to better firm performance. These findings are in line with Perry et al. (2007) who argue that there are good reasons to expect important overall productivity gains derived from improvements to the regulatory framework. Escribano and Guasch methodology relies on a "general to specific" approach similar to the one postulated by Hendry and Mizon (1978) where to avoid missing variables bias one starts with a very eneral model andthen dropsvariablesrecursivelyaccordingto whether they are significantor not. Shleifer and Vishny (1993) argue that the secrecy associated to corruption makes it much more distortionarythan taxesbecauseofthe neededeffortsto avoiddetection andpunishment. " For instance, Demirguc-Kunt, Love, and Maksimovic (2006) argue that firms are more likely to be incorporatedincountrieswith a well-hctioning legalsystem. 17 Table 2. Impactof the investmentclimateon labor productivity Dependentvanable: Log01salesper worker - . .. - - . 11) (2) (3) 14) (5) OLS ZSLS ZSLS ZSLS ZSLS Substitution N N Y N Y Instruments Sample Sample Weighted Weighted averages averages averages averages Bribes -0.016 -0.515** -0.469*** -0.467** -0.454*** [0.044] [0.201] [O. 1171 [O. 1991 [0.114] Security -1.116*** -2.002 -3.556*** -2.222 -3.616*** [0.338] [1.631] [1.168] [1.6371 [1.175] Crime -2.969*** -10.644*** -7.476*** -11.746*** -6.164** * [0.736] [3.032] [1.7641 [3.294] [1.9301 Regulatorycompliance 0.335*** 1.687*** 1.136*** 1.702*** 1.048*** [0.060] [0.282] [0.157] [0.293] [O. 1601 Audited 0.272*** 0.486** 0.273** 0.483** 0.377*** [0.043] [0.205] [O. 1281 [0.208] [O. 1313 Incorporated 0.290*** 0.35 1** 0.338*** 0.376 ** 0.415*** [0.051] [O. 1721 [O. 1OS] [O. 1721 [O. 1091 Infrastucture quality 0.899** 3.030* 1.880** 2.770* 2.079** [0.397] [1.573] [0.906] [1.600] [0.930] Aces to finance 0.204*** 0.002 0.209 0.048 0.275* [0.047] [0.253] [0.166] [0.257] [0.171] Technology 0.169*** 0.213 0.194 0.229 0.314** [0.046] [0.237] [0.126] [0.243] [0.135] Training 0.102*** -0.012 0.306*** -0.004 0.234** [0.038] [0.199] [O. 1121 [O.2021 [O. 1081 Uniqueestablishment -0.339*** -0.305*** -0.253*** -0.302*** -0.237*** [O.OS 11 0.092** * [0.059] [0.0381 [0.060] [0.0391 Logof firm's age 0.121*** 0.095 *** 0.091*** 0.089*** v1 [0.026] [0.029] [0.016] [0.029] [0.016] 3 Foreignowned tr 0.356*** 0.244*** 0.400*** 0.247* ** 0.386*** [0.078] [0.085] [O.OSO] [0.086] [O.OSO] 0.255 * ** 0.224*** 0.181*** 0.222*** 0.165*** [0.045] [0.0671 [O.0381 [0.068] [0.0391 20-99 employees 0.032 -0.044 -0.031 -0.054 -0.064 [0.044] [0.088] [0.0501 [0.090] [0.051] loo+ employees -0.098 -0.282* -0.222* ** -0.298* -0.278*** [0.070] [0.152] [0.085] [O. 1561 10.0861 Observations 4,207 4,207 10,535 4,207 10,535 Note: The table reports the results of regressing the log of sales per worker on the investment climate attributes and firm controls inthe first column. Appendix A describes in detail the different variables used in the regression. Column (1) OLS; (2) 2SLS using region-industry-firm size sample averages as instruments; (3) 2SLS usingregion-industry-firmsize sample averages as instrument andreplacingmissing observationswith the instrumentalvariable; (4) 2SLS usingregion-industry-firmsize weightedaverages as instruments; (5) 2SLS using region-industry-firm size weighted averages as instrument and replacing missing observations with the instrumental variable. 2SLS Instruments are constructed using a minimum threshold of 10 observations. All regressions include country, industry and region dummies. Robust standarderrors inbrackets. * significant at 10%; ** significant at 5%; *** significantat 1% 18 Beyond our findings for governance and institutional quality, we also argue that deficiencies in the physical infrastructure are an important component of the investment climate. Our estimates indicate that inji-astructure, as measured by the firms' losses due to power outages are negatively and significantly related to labor productivity as measuredby firms' sales per worker. On the finance fiont, our models indicate that firms that are not credit constrained either because they have access to credit through bank finance or because they report not having a need for credit, exhibit larger levels of labor productivity. And finally, on the skills and technology front firms that use their own website to communicate with clients and suppliers have significantly larger levels of labor productivity and so do firms that offer training to their workers. We also study the impact of investment climate on Total Factor Productivity (TFP). Since the capital stock and intermediate inputs costs variables, which are needed to estimate TFP, have a relatively large number of missing values, we also re-estimate the labor productivity model on the same sample for which TFP measures are available. We find that corruption, crime, regulatory compliance, and the incidence of firms with externally audited financial statements retain their significance. However, results in other areas are weaker, largely due to the composition of the sample, which can be readily seen by comparing columns 1and2 in Table 3. 19 Table 3. Impactof the investmentclimateon TFP Dependent variable: Logof Logsales value Der worker (11 (2) 2SLS 2SLS Substitution Y Y Instruments Weighted Weighted averages averages Bribes -0.489*** -0.487*** [0.134] [0.130] SeCUrity -2.588** -1.829 [1.311] [1.211] Crime -5.076** -6.05 1*** [2.300] [2.251] Regulatory compliance 1.404*** 1.575*** [0.212] [0.203] Audited 0.299* 0.374** [0.155] [O. 1511 Incorporated 0.123 0.125 [O. 1401 [0.135] Infrastucture 1.242 1.461 [1.0841 [1.0581 Acces to finance 0.09 1 0.015 [0.193] [O.1881 Technology -0.142 0.073 [0.172] [0.170] Training 0.209 0.004 10.1521 10.1461 Unique establishment -0.252** * -0.292*** [0.048] [0.047] Logoffirm's age 0.1IO*** 0.078*** [0.021] [0.021] v1 Foreign owned 0.293*** 0.331*** [0.062] [0.061] !3 0.273*** 0.261*** [0.049] [0.048] 20-99 employees -0.003 0.037 [0.068] [0.068] 100+ employees -0.049 -0.02 [0.110] [0.110] Logcapitalper worker 0.026*** [0.003] Observations 6,880 6,8 80 Note: Column (1) reports the results of regressingthe log of value added per worker on the investment climate attributes and firm controls. 2SLS using region-industry-finn size weighted averages as instruments and replacing missing observations with the instrumental variable. 2SLS instruments are constructed using a minimum threshold of 10 observations. Column (2) is as column (5) in Table 2 but restricting the sample to the one used in column (1). All regressions include country, industry and region dummies. Robust standard errors in brackets * significant at 10%; ** significant at 5%; *** significant at 1% 20 Table4. Impactof the investmentclimate onwages Dependentvariableis logofwage (1) (2) (3) (4) ZSLS ZSLS ZSLS ZSLS Educ not Educ Educ not Educ Substitution Y Y Y Y Instruments Sample Sample Weighted Weighted averages averages averages averages Bribes -0.306* ** -0.312** * -0.211** -0.213** [0.098] [0.099] [0.099] [0.0991 Security -0.822 -0.838 -0.847 -0.857 [0.810] [0.813] [0.823] [0.824] Crime -1.938 -1.842 -3.551* -3.505* [1.732] [1.716] [2.041] [2.033] Regulatory compliance 0.570* ** 0.558*** 0.652*** 0.640*** [0.133] [0.135] [O. 1341 [0.137] Audited 0.449*** 0.458*** 0.397*** 0.398*** [0.115] [0.114] [0.117] [0.117] Incorporated 0.220** 0.223** 0.174* 0.177* [0.086] [0.087] [0.091] [0.091] Infiastucturequality 0.302 0.236 0.191 0.162 [O.7551 [0.771] [0.774] [0.786] Acces to finance 0.178 0.192 0.278* 0.284* [O. 1391 [0.140] [O,1461 [0.146] Technology 0.287** 0.297*** 0.2 18* 0.224* [0.113] [0.114] [O. 1201 [0.121] Training - 0.038 0.026 0.007 -0.006 [0.102] [O. 1031 [0.098] [0.099] Uniqueestablishment -0.069** -0.064* -0.082* * -0.079** [0.033] [0.034] [0.034] [0.034] Logof firm's age 0.089*** 0.091 *** 0.090*** 0.092*** [0.014] [0.014] [0.014] [0.014] Foreignowned 0.216*** 0.209*** 0.228*** 0.224*** rA [0.044] [0.0451 [0.045] [0.045] 'E 'E; Exporter B 0.118*** 0.115*** 0.129*** 0.128*** [0.034] [0.034] [0.035] [0.035] 20-99 employees -0.128*** -0.124*** -0.111** -0.107* * tz [0.046] [0.0461 [0.047] [0.047] 1OW employees -0.268*** -0.268*** -0.235*** -0.232*** [0.075] [0.075] [O. 0761 [0.077] Educ avge worker 7-12yrs 0.081** 0.111 0.087*** 0.121 [0.0331 [0.117] [0.033] [0.115] Educ avge worker 13+yrs 0.124** 0.420** 0.124** 0.301 f0.0571 [0.210] [0.0571 [0.2 151 Observations 10,205 10,205 10,205 10,205 Note: The table reports the results of regressing the log of wages per worker on the investment climate attributes and firm controls in the first column. Columns (1) and (2) use 2SLS usingregion-industry-firm size sample averages as instruments; Columns (3) and (4) use 2SLS using region-industry-firm size weighted averages as instruments. Education is instrumented in columns (2) and (4). All regressions include country, industry andregion dummies. Robust standard errors in brackets. significant at 10%; * ** significant at 5%;*** significant at 1% 21 Does a better investment climate benefit firms only? Our analysis indicates that average wages tend to rise with improvements in several investment climate attributes. In particular, Table 4 shows that companies that face higher levels of corruption, crime, or where governance and institutional quality issues lead to high levels o f regulatory non- compliance tend to pay lower wages per worker. On the other hand, audited and incorporated$rms, those with access credit and companies with better technologyappear to pay higher wages. Thus, it appears that improving the investment climate has a positive impact not only on firm performance but also on their workers. On a related matter, our results also indicate that the much feared globalization process, seems to play infavor or workers. Controlling for other factors, exporters tendto pay on averagehigher wages than non exporters (see also Chapter 7 of volume 11). Similarly, firms with some degree of foreign ownership also tend to pay higherwages. Most of these findings are robust to the exclusion of firms with 100 plus workers and firms with foreign capital from our samples, which we interpret, following Dollar et al. (2005), as evidence that our results are not driven by the self-selection of better performing firms into good investment climate areas - smaller domestic f i s being considerably less mobile than the excluded group of firms. Inaddition, our results are in most cases robust to the use of estimation methods that control for the influence of outliers -the only exceptions beingthe impact ofsecurity expenditures and audits. The results reported in Chapter 2 are also consistent with the findings of Hallward- Driemeier and Stewart (2004) and World Development Report (2005), who show that smaller firms stand to benefit the most from improvements in the investment climate. In particular, the productivity o f small firms tends to be more affected by corruption as well negatively affected by the quality of infrastructure and benefit more from the use o f new as by the need to invest more in security. Similarly, smaller firms seem to be more technologythan larger firms. Thus inline with the discussion on the impact of investment climate attributes on average wages, the central topic o f this report is not an issue o f interest only for big corporations but rather to a muchwider audience. To further explore the impact of human capital on fmperformance, we complement our Enterprise Survey data with information on workers average educational levels taken from household surveys.12 Table 5, which reports the results of this exercise, indicates that when we bring education from the household surveys into our empirical model, it appears to belong to boththe productivity andwage equations. l2The data on education in the Enterprise Surveys tend to cover only the average educationlevel of the work force (a variable which can be expected to be measured with significant noise) and the degree of educationof the manager. See Chapter 2 for details ofhow we merge this data with information taken from household surveys, as well as on how we control for potential reverse causality from productivity to education. 22 Table 5. Educationandthe investmentclimate Dependentvariable is log(sa1es per worker) log(averagewage) (1) (2) (3) (4) 2SLS 2SLS 2SLS 2SLS Substitution Y Y Y Y Instruments Sample Weighted Sample Weighted averages averages averapes averages Bribes -0.430*** -0.41I*** -0.265*** -0.171* [0.117] [0.114] [0.098] E0.0991 Security -3.677*** -3.760*** -0.932 -0.978 [1.169] [1.178] [0.814] [0.829] Crime -7.505*** -5.713*** -1.851 -2.683 [1.771] [1.831] [1.742] [1.933] .3 3 3 1.066*** 0.970*** 0.528*** 0.600*** [0.157] [0.159] [0.133] [0.134] 3 Audited E Regulatorycompliance 3is 0.274** 0.382*** 0.444*** 0.423*** [0.128] [0.131] [0.115] [0.117] Incorporated 0.366*** 0.441*** 0.244*** 0.217** 4- 8 [0.104] [0.108] [0.086] [0.090] Infrastucturequality 1.778** 2.067** 0.327 0.272 m A> Acces to finance e, [0.905] [0.928] [0.752] [0.768] 0.26 0.334* 0.204 0.308** [0.168] [0.172] [0.141] [0.147] Technology 0.211* 0.335** 0.298*** 0.244** [0.126] [0.135] [0.113] [0.120] Training 0.277** 0.221** 0.036 0.019 [0.113] [0.109] [0.101] [0.097] Average education(#) 0.028*** 0.026*** 0.014** 0.015** [O.0081 [0.008] [0.007] [0.007] Unique establishment -0.255*** -0.239*** -0.072** -0.079** [0.038] [0.038] [0.034] [0.034] Logof firm's age 0.093*** 0.088*** 0.087*** 0.087*** m [0.016] [0.016] [0.014] [0.014] 3 o Foreignowned 0.422*** 0.409*** 0.240*** 0.248*** 43 [0.051] [0.051] [0.046] [0.046] 0.184*** 0.167*** 0.123*** 0.129*** [0.038] [0.038] [0.034] [0.035] 20-99 employees -0.043 -0.078 -0.139*** -0.132*** [O.OSO] [0.051] [0.046] [0.046] 100+ employees -0.228*** -0.290*** -0.270*** -0.255*** [0.085] [0.087] [0.075] [0.076] Observations 10,444 10,444 10,113 10,113 Note: The table reports the results of regressing the log of saledwages per worker on the investment climate attributes and firm controls in the first column. Columns (1) and (3) use 2SLS using region- industry-firmsize sample averages as instruments; Columns (2) and (4) use 2SLS using region-industry- firm size weighted averages as instruments. All regressionsinclude country, industry andregion dummies. Robuststandarderrorsinbrackets. * significant at 10%;** significantat 5%; *** significantat1% 23 From a policy perspective we would like to note that improving the various investment climate attributes that we have identified as being relevant for boosting firm performance will require cooperation by many actors. Aspects related to regulatory matters (both framework and compliance), crime, and security appear under the government responsibility and therefore one would expect appropriate interventions to correct existing deficiencies. However, there are other areas, such as those related to corporate governance, the use of modern technologies, the adoption of quality standards and the provision of worker training where progress will require a combination of public intervention andprivate action. True, the quality of private sector governance may be a reflection o f creditor rights and this in turn may be related to regulations. Yet, pretending that improvements on the corporate governance front will be easily achieved only through additional government intervention instead of through a private sector pushis not likely to be very realistic. And true, the adoption o f technology or the implementation of training programs can and should be encouraged by the government through a number of different vehicles like technical support (especially for smaller firms with limited resources) and financial incentives such as partial subsidies, tax deductions, or special treatment of training expenses. But once again, unless companies see technological upgrades and training activities as an integral component of their businessplan it will be difficult to achieve any meaningfbl result. It is also worth noting that while we find that our selected econometric model contains six investment climate variables that could bethought as belonging to the governance and institutional quality group, the infrastructure, technology, and finance groups are represented more meagerly. In other words, while our results can be taken as an indication that improving the investment climate in Latin America will require interventions along several fronts, they mainly stress the importance o f improving the quality of institutions and of the regulatory framework. As a possible caveat to our findings it is worth notingthat it is difficult to discern whether they are a reflection of the reality of the region (i.e. governance and institutional factors are critical) or instead whether they are driven by the fact that the Enterprise Surveys are more appropriate to measure some investment climate aspects (e.g. those related to governance and institutional quality) than others. Our take, is that perhaps the second alternative is more realistic. Thus, we interpret our results as being complementary to those of aggregate cross-country studies. For example, Figure 11 shows Loayza, Fajnzylber and Calderh's (2005) estimates of the contributions to growth of progress in a number of policy areas between the first and second halves of the 1990s (Figure 11). Inspection of this figure indicates that in order to understand the recent Latin American growth experience one has to pay attention to four main areas: infrastructure, education, international trade, and financial development. Moreover, Loayza et al. (2005) fail to uncover robust links between growth and the quality of governance and institutions, perhaps because o f the aggregate nature o f the data. However, the results o f that macro study and those obtained with our own micro-based approach can be seen as complementing and enriching each other, namely by helping uncover different relevant 24 growth determinants thanks to their use of different types o f data and empirical methodologies. In particular, we interpret our findings as suggesting that the need to make progress in the governance and regulatory front should be put at the top of the region's policy agenda, perhaps even as a previous step to making progress on other priority areas, including education, infrastructure or financial development. Figure11.Contributionto growthinLatinAmerica Source:Owncalculations usingdatafrom Loayza, Fajnzylber andCaldercjn(2005). The figure reportsthe contribution to the change ingrowthrates between 1991-95 and 1996- 99 inthe medianLatinAmerican country. V. Gettingsome senseof policypriorities So far we have analyzed the impact o f the investment climate on firms' productivity in LatinAmerica and found that there are several areas where progresswill likely lead to a better firm performance and hence faster growth. However, should all countries focus on the same areas or instead each o f them will have to follow a different path? After all, there are important differences across countries that must be kept in mind when formulatingpolicy recommendations. For example, reducing corruption can be extremely useful in all countries but the sense o f urgency and the returns to any potential intervention will likely be different in say Chile where only 6 percent of f m s report paying bribes and in say Ecuador where as many as 68 percent of firms pay bribes (see Chapter 4). Similarly, makingprogress on the regulatory framework may lead to higher firm productivity and growth, but itwould not be very reasonableto stress it inthe same way in Panama where 92 percent o f firms are incorporated and 85 percent have external audits than in Nicaragua where the same shares are 2 and 27 percent respectively. Or consider our variable capturing the use o f technology. In Argentina 73 percent o f firms use the web, whereas in Honduras only 22 percent do. Once again, it does not seem reasonable to emphasize the need to make progress in this area in these two countries equally. 25 Following this line o f argument, one can simulate the potential gains associated with progress in each particular area by merging two pieces of information. The first is the magnitude of a marginalimprovement ineach investment climate attribute. The second is the potential for improvements in the attribute. Regarding the first element of this exercise, we borrow from Chapter 2 and use the estimated coefficients of our basic model (column 5 of Table 2 above) as the marginalimpact of each attribute. As for the potential gains in each attribute, in Chapter 3 of Volume I1we experiment with five alternative scenarios. The first two scenarios are constructed by taking as reference the average firm in a specific country. Inthis respectwe use Irelandand Chile as benchmarks. We chose Chile based on the fact that this country is the one that grew the most in the 199Os, and it has often been cited in policy circles as a successful case o f policy reforms. Moreover, Chilean firms exhibit the best indicators o f investment climate conditions of all 16 Latin American countries in our ~amp1e.l~We chose Ireland because it appears as the best performer inour global sample estimates (Table 11of Chapter 2).We decided to compare the performance of LatinAmerican countries with a country from another region because it may bepossible that Latin American countries are similar in many ways, reducing the potential scope for improvements. As will be seen below, the potential gains associated with the Ireland benchmark are much larger than the ones associated with the Chilean benchmark. The other three scenarios take as a "benchmark firm" a hypothetical firm located at the 7 5 percentile o f the Latin American firms. In the most optimistic scenario we take as ~ potential gains the difference between the value o f an attribute for a particular firm and the value corresponding to a firm situated in the 75thpercentile of the entire sample of Latin American firms. This means that we compare the average firm ineach country with this same "benchmark firm." Two additional scenariosthat we consider take into account some firm characteristics. For example, in the fourth scenario our simulation consists in computing the potential gains for the average firm in each country when moving to the region's 75* percentile o f each investment climate attribute within the same industryand firm size range. Finally, a fifth scenario is given by calculating the difference between the value of eachattribute and that corresponding to a firm inthe top 75thpercentile not only within the same industry and size range, but also within the same country. Figure 12 shows the average productivity gains of movingto those five differentbenchmarks. l3Clearly, ifone operatesmechanicallywith thesesimplerules it wouldimply that ifacountry is already inabetter situation than the reference group, the gains would be negative.Giventhat this doesnot seem a very realisticpolicy recommendation, we exclude those caseswhen we compute aggregategains. 26 Figure 12: Potentialproductivity gains associated to different scenarios Regulatorycompliance Audited& Incorporated Crime & Security Bribes Technology Training Infrastructurequality Access to fmance 0 5 10 15 20 25 30 m a Ob Oc .d .e Notes: Benchmarks (a) Ireland (same industry, size), (b) Chile (same industry, size), (c) 75thpercentile of the entire sample, (d) 75'hpercentile ofthe same industry andfirm size, and (e) 75'hpercentile ofthe same country, industry and firmsize. Inspection o f this figure indicates that in general the biggest gains are associatedwith the first (Ireland) and third (75th percentile of the entire sample) scenarios, while the most modest gains are associated with the fifth of these scenarios. Moreover, although there are important quantitative differences across the various scenarios the qualitative results are very similar (with the exception, in the first and last scenarios, o f the estimated gains for the variables on audited and incorporated firms). To simplify our presentation, we merge the contributions of crime and security, and the contributions of audited and incorporated firms. Overall, the investment climate areas that yield the largest gains are related to the regulatory framework (regulatory compliance, audited, and incorporated), and to the rule of law (crime and security, and corruption). They are followed by improvements on the technological front and in terms o f the quality of infrastructure. Somewhat surprisingly, at the end o f this classification we can find access to finance which as noted in the previous section was identified by firms as one of the top five obstacles to growth inall countries. 27 Figure 13: Aggregate Labor Productivity and Wage Gains, by country PanelA. Labor productivity gains Nicaragua Honduras Ecuador El SBlvador Brazil P W P Y Guatemala Mexico Bolivia Panama Peru Colombia Uruguay CostaRica Argentina Chile I I I I I 0 20 40 60 80 100 120 140 PanelB. Wage gains Nicaragua Honchras Brazil Ecuador Paraguay Mexico El SBlvador Guatemala Peru Uruguay Panama Bolivia Colombia CostaRica Argentina Chile , I , I 0 20 40 60 80 100 120 140 Notes: The figure depictsthepotentialpercentagegainsthat couldbe achievedif firms were to move to the 75fhpercentileofthe same industry andfirm size. Clearly, what is of most interest in the context of this exercise is the calculation of the various potential gains at the country level. In this regard, Figure 13 presents the accumulated potential productivity (Panel A) and wage (Panel B) gains from improvements in all areas under the third scenario discussed above. The four countries with the largest potential productivity gain are Nicaragua, Honduras, Ecuador and El Salvador (the first three countries could more than double their labor productivity). Brazil, Paraguay, Guatemala andMexico are the next four countries, with potential gains 28 in labor productivity ranging between 82 and 92 percent. Next, Bolivia, Panama, Peru and Colombia could obtain an average productivity gain o f almost 60 percent. Finally, Uruguay, Costa Rica, Argentina, and Chile would gain between 17 and 52 percent. Panel B shows the wage gains under the same benchmark. The average gain for the region is of 45.1 percent (no mean feat!), ranging from 8.9 percent in Chile to 74.1 percent inNicaragua. With few exceptions, the order o f countries from largest to smallest wage gain is similar to the order according to labor productivity gains. Brazil is now among the top four countries with largest wage gains, while El Salvador is inthe second group. Uruguay and Colombia also switched positions between the last two groups. The previous analyses can be disaggregated at the country and investment climate attribute level to generate a list of top priorities by country. This is done in Table 6 in terms of labor productivity gains. Inspection of this Table indicates that aspectsrelated to the regulatory.framework are critical. For example regulatory compliance appears as the top priority in 10 o f the 16 countries in our sample, and it is a secondor third priority for the other 6 countries; audited and incorporated appear among the top three priorities in 14 countries. This is also the case for crime and security. More specifically, the enhancement o f the country's legal system's efficiency, strong shareholder and creditor rights, and low regulatory burdens, measured by our variables audited and incorporated, are the top priority inHonduras, Nicaragua, Peru and Uruguay. Reductions in crime and in preventive security expenses could yield the highest gains in Chile and Ecuador. The reduction of bribes and corruption is also an important source to achieve important productivity gains for the region. Inparticular, it appears as the second priority for Argentina, Bolivia, Brazil and Ecuador, and as a third priority for Paraguay and Peru. Improvements in infrastructure quality (measured by losses due to power outages), technology (measured by web use) and the provision of training programs appear as a top 3 priority in a few cases. Infrastructure quality is a top 3 priority in Costa Rica and Panama. The latter could gain the same from improvements in technology and training. One final exercisewe would like to discuss here is related to the differential impact o fthe investment climate depending on firm size. As noted above, there is some empirical evidence indicating that smaller firms would benefit more than proportionally from improvements in the investment ~1imate.l~Chapter 3 in Volume I1explores this issue in the context of our simulation exercises and finds that, even if one ignores the fact that f m s with less than 20 workers stand to gain more from improvements in the investment climate - that is even if one uses the same coefficients for all firms - the potential productivity gains are much larger for smaller firms, which reflects the fact that, in comparison with their larger counterparts, they tend to be located farther away from the region's best-practices. Thus, on average the productivity of small firms would increase by 76 percent ina scenario ofprogress to the region's 75* percentile ofthe same industry and firm size, compared to increases o f 69 percent and 49 percent for medium (20-99 workers) and large firms (more than 100workers), respectively. l4Hallward-Driemeierand Stewart (2004), World Development Report (2005). 29 Table6: Top threeprioritiesfor firms movingto the 75fhpercentileof same industry andfirmsize ~ Country Order ofpriorities 1st 2nd 3rd Argentina Regulatorycompliance Bribes Crime& SecurityExpenses Bolivia Regulatorycompliance Bribes Audited &Incorporated Brazil Regulatorycompliance Bribes Audited &Incorporated Chile Crime& SecurityExpenses Audited & Incorporated Regulatorycompliance Colombia Regulatorycompliance Audited &Incorporated Crime& SecurityExpenses Costa Rica Regulatorycompliance Crime& SecurityExpenses Infrastructurequality Ecuador Crime& SecurityExpenses Bribes Regulatorycompliance ElSalvador Regulatorycompliance Audited &Incorporated Crime& SecurityExpenses Guatemala Regulatorycompliance Crime& SecurityExpenses Audited & Incorporated Honduras Audited & Incorporated Regulatorycompliance Crime& SecurityExpenses Mexico Regulatorycompliance Audited &Incorporated Crime& SecurityExpenses Nicaragua Audited & Incorporated Regulatorycompliance Crime& SecurityExpenses Panama Regulatorycompliance Crime & SecurityExpenses Technology, Infrastructure quality, Training Paraguay Regulatorycompliance Audited & Incorporated Bribes Peru Audited & Incorporated Regulatorycompliance Bribes Uruguay Audited & Incorporated Regulatorycompliance Crime& SecurityExpenses Mean Regulatorycompliance Audited & Incorporated Crime& SecurityExpenses Median Regulatorycompliance Audited & Incorporated Crime& SecurityExpenses Interms of the areas that yield the largest benefits, Figure 14 indicates that firms of all sizes would benefit the most from improvements inthe regulatory framework. Moreover, the estimated gains would be bigger for small firms (an increase in productivity o f 26 percent, against 22.6 and 17.4 percent for medium and large firms, respectively). Somewhat related, improvements in our audit and incorporated variables appear as the second priority for small firms, with potential productivity gains of 16 percent. Medium and large firms would obtain productivity gains of 13 and 9 percent from reductions in crime and security expenses, respectively. Recall, that we take audit and incorporated as reflecting the country's legal system efficiency, shareholder and creditor rights, and 30 regulatory burden. The potential gains derived from makingprogress inother areas of the investment climate are very similar for small and larger firms. Figure14: Aggregatelabor productivitygains, byfirm size 30 25 20 15 10 5 0 Bnbes Crime& Regulatory Audited& Infrastructure Accessto Technology Training Security compliance Incorporated quality finance Snrall(49) Medium(2b99) 0 Large(1W) Notes: The figure depicts the potential percentage gains that could be achieved if firms were to move to the 75* percentile o fthe same industry and firmsize. VI. Improvingthe investmentclimate Inthe context ofthis regional study, the type of knowledge neededto establish priorities for action (e.g. is reducing graft a priority?) and the type of knowledge needed to attack those priorities (e.g. what are the determinants o f graft?) are different. Thus we now turn to the type of interventions that policymakers could take to improve their country's investment climate conditions. True, there are several areas o f those discussed in previous sections, such as infrastructure provision or regulation, where recent studies focusing on the Latin America region have shed light on how to proceed. For example, in a recent RegionalStudy o f the World Bank's LatinAmerica and Caribbean region, Easterly and ServCn (2003) assessed the quantity and quality of infrastructure services in the region concluding that, at least when compared with the successhl economies of East Asia, the Latin American region was considerably behind, even though the latter region had been ahead of Asia as of 1980. During the last two decades of the 20* century many countries in the region undertook sector reform and introduced private sector participation in, among others, the infrastructure sectors through management contracts, concessions and privatizations. Whereas these reforms were motivated by a desire o f improving efficiency and quality o f service, as well as by the fact that by fiscal constraints prevented the public sector from expanding coverage. However, public sector retrenchment without a comparable increase in the involvement of the private sector in infrastructure provision has resulted in lower stocks of infrastructure overall. 31 Easterly and Servh argue that this outcome was in part the result o f the public and private sector having often played complementary rather than competing roles, and in part the result o f an inappropriate regulatory framework and faulty concession design and oversight. Indeed, although the opening up of infrastructure to the private sector occurred under diverse regulatory environments across the various countries inthe region, inmany cases the opening up took place before appropriate regulations and regulatory bodies had been createdor when existing capacity and experience to administer regulations was still very limited. Looking forward these frndings suggest that improvements to the infrastructure of the different countries in the region will require on the one hand a reassessment of the appropriate level of public sector investment in infrastructure. As it is discussed in Chapter 1of Volume I1the low levels of public infrastructure investment are not just a problem of the past. In fact, over the 2003-2005 period public infrastructure investment were at the lowest level of the past 25 years. It will also require a regulatory fiamework that the private sector perceives as transparent and supportive of private sector involvement. In this regard, the available evidence indicates that the existence of a regulatory body per se does not have much effect on private participation, but it does affect the quality o f investments and the level of the public's and the users' benefits from those investments (Guasch, 2005). However, regulatory predictability and credibility and overall quality and governance o f regulation do appear to play a critical role from the perspective of the private sector, and as a determinant o f sector performance (Andres, Guasch, Foster and Haven, 2007; and Andres, Guasch and Lopez Azumendi, 2007). Similarly, the recent Flagship report of the Bank's Latin American region for 2007 (see Perry et al. 2007) has devoted significant attention to how a country's regulatory framework may affect the observed levels of firm productivity. Inthis regard, Perry et al. (2007) conclude that burdensome regulations, excessive red tape, and costly bureaucratic requirements are likely to act as a brake to formal entrepreneurial activity and hence be behindhighlevels ofinf~rrnality'~well as low productivity and economic growth. as Although the appropriate policy mix will clearly depend on country circumstances, Perry et al. (2007) suggest a combination of carrots-and-sticks. Countries should assess their regulatory frameworks to distinguish regulations that are indeed justified by public interest or market failures from those that appear "anachronistic", or that merely reflect private interests and are actually aimed at transferring rents to specific groups. Complementarily, countries should also consider the regulatory framework problem from an administrative and procedural perspective. A good regulatory framework could still suffer from poor and costly enforcement (because of cumbersome procedures, or lack o f qualified personnel) or untimely implementation, which leads to increased costs for the private sector. Similarly a good regulatory framework should provide minimal discretion to the responsible public servants and use on-line IT-e tools as much as possible to reduce users costs, increase transparency and reduce opportunities for corruption. Thus The resultsinLoayzaandRigolini(2006) suggest that informalityincreaseswiththe regulatoryburden, butdecreaseswith the efficiencyinthe provisionofpublicservices. 32 administrative simplification together with a professionalization o f the civil service can be effective tools to lower the inherent costs of regulations. But to eradicate corruption that might not be enough. Incentives to deter corruption are also needed. Countries should also consider "sticks", such as increasing efforts to detect regulatory non-compliance, and raising the costs of regulation infringement by increasing fines and sanctions for those caught in irregular situations - i.e. both public servants and f m s . Similarly, improvements in the judicial system that increase the likelihood o f punishment for all parties involved in corruption (increasing the expected cost of enteringinsuch type ofagreements) are likely to leadto lower corruption levels. On the crime front, the World Bank's LatinAmerica and Caribbean region has also paid significant attention to the issue. After all, controlling for other factors the Latin American region seems to have a higher incidence o f robberies than any other region over the period 1970-1994. In fact a Regional Study by Pablo Fajnzylber, Daniel Lederman, and Norman Loayza (1998) analyzed empirically the determinants of crime rates in Latin America building on a theoretical model of incentives to crime that emphasized the role played by economic variables. Fajnzylber, Lederman, and Loayza's model assumed that criminals acted rationally andthat decisions to commit crimes would be the result o f an implicit cost benefit analysis. Several messages and policy lessons emerged from that regional study. First, crime rates appear to be highly persistent and show significant inertia. In other words, there may be a tendency for crime to self-perpetuate in the absence of interventions to attack it. Second, income inequality -an issueparticularlyrelevant inthe Latin American context16- appears to be positively correlated with crime rates. Third, deterrencemeasuresalso appear to play an important role in explaining crime rates. And fourth, drug activity seems to be also associated with violent crime, a reflection perhaps of the disputesthat the highprofits associatedto drug activities creates. What would be the policy implications of these findings? On the one hand, the high persistency of crime rates would imply that policy makers facing a crime wave should act sooner rather than later to avoid getting stuck into a high crime rate equilibrium. On the other hand, the findings on the determinants of crime would suggest, as inthe case o f the regulatory framework, a strategy based on carrots-and-sticks, the former been linked to increasing income opportunities in the formal sector and lowering inequality. In this regardthe past few years have witnessed the emergence of a type of social programs, the conditional cash transfers, which in additionto beingwell targeted appear to contribute to lowering inequality levels. For example, inan analysis performedby Kruger et al. (2006) for Brazil, conditional cash transfer programs are estimated to account for up to a third of the declines in income inequality observed in Brazil between 1995 and 2004. So called "sticks", on the other hand, would be aimed at sanctioning criminals through effective investigative andprosecuting units, as well as fighting organized crime. l6 it As is well known (see for example Perry et al. 2006), Latin America is together with Sub-Saharan Africa the mostunequalregion inthe world. 33 Yet, there are other areas of the investment climate where our knowledge is much more limited. For example, what are the determinants of graft and who must pay bribes? Similarly, what are the determinants of access to credit? Or on the technological front, who adopts/adaptsnew technologies? What should policy makers do to take advantage o f the links between exports and firmperformance? We now focus on these various issues. VII. Corruption The finding that corruption is a barrier to firmperformance and growth is consistent with a large number of cross-country econometric studies which have found that poor governance -heled, among others, by rampant corruption- negatively affects economic outcomes such as per capita income, growth, infant mortality, literacy, macroeconomic stability, and so f0rth.I7 All this leads to the conclusion that reducing the incidence of corruption is a significant priority and should be placed at the top of the reform agenda. However from a policy perspective, the critical issue is identifying and deciding which type o f actions need to be implemented to attack the problem. After all, the existence or not o f bribe payments isjust an outcome that reflects a wide range o f factors and country circumstances that facilitate an environment o f corruption rather than a policy variable that can be changed at the will o f policy makers. Infact, from a policy perspective if the objective is attacking corruption, it is extremely important to identify those factors and country circumstances that are behind corruption practices, so that appropriate effective interventions can be designed. Against this background, what does the existing literature tell us? The past few years have witnessed the emergence o f a booming literature (mainly based on cross country aggregate regressions) on the causes of corruption trying to explain why the incidence o f corruption is higher in some countries than in others. For example, Montinola and Jackman (2002), Treisman (1999), and Brunetti and Weder (1998) explore the role of democracy and freedom of speech; Montinola and Jackman (2002), and Evansand Rauch (1996) look at the impact o f a professional civil service and at pay levels o f public officials; World Development Report 1997 and Ades and Di Tella (1996) focus on the relationship between a strong judiciary system and corruption; Goel and Nelson (1998) and Lapalombara (1994) analyze whether a higher degree of government involvement in the economy leads to more corruption; finally, Brunetti and Weder (1998), Treisman (1999), and k i t e and Weidmann (1999) addressed the role of competition and openness to trade. On the whole, the main lessons of the previous studies would suggest that there are a number of policy interventions that may lead to lower levels of corruption. For example, transparency in the form o f laws protecting free speech and independent media outlets would likely result ina more free press a factor that may increasethe probability ofbeing caught in corrupting activities (and hence a deterrent to corruption). Having professional civil servants that are well paid would also reduce the incentives they have to undertake extortion. Limiting discretional powers of public servants reduces their leverage to extract bribes. Extensive use o f e-government, placing procedures on line, implementing "Seeforinstance,Mauro(1995), Kaufmannet al. (1999), andLoayza et al. (2004,2005) 34 administrative silence procedures, also limits corruption practices. On the macro front, keeping public spending at reasonable levels and opening the economy to trade and foreign competition would also be steps inthe right direction. From a microeconomic perspective, Kaufinan and Wei (1999) use data fiom three worldwide firm-level surveys to examine the relationship between bribe payment, managementtime wasted with bureaucrats, and cost of capital finding that firms that pay more bribes are also likely to face higher costs of capital. Similarly, Hellman, Jones, and Kaufinann (2000) rely on the Business Environment and Enterprise Performance Survey (BEEPS) data for a number of transition economies and find that new firms trying to compete in a market dominated by established incumbents in states that under provide contract and property rights may try to buy them directly from public officials (i.e. engage incorruption activities) to compensate for existingweaknesses inthe overall legal framework. Svensson (2003) uses a survey of Ugandan firms to study "who pays bribes and how much." Svensson proposes an explanation for the pattern of corruption and Uganda and concludes that firms having more contact with the government, more profitable firms, and firms with lower bargaining power (i.e., firms with no outside option) are more likely to pay bribes. In Chapter 4 of Volume I1we rely to a large extent on the conceptual framework put forward by Svensson (2003), to explain corruption patterns in Latin America. In this framework the incidence of corruption can be explained by the variation o f regulation across industries (i.e. the "control rights" hypothesis) and by the bargaining position of the firm, where the f m ' s ability to pay plays a crucial role (i.e. more solvent firms may be a weaker bargaining position). In other words, public officials would demand bribes and firms would have to follow to a different extent depending on the leverage that they have and on the cost of not paying. Too often, however, the problem of corruption is attacked from the sole perspective o f corrupt public officials rather than fiom that o f the firms' willingness to bribes public officials. However, there is an alternative (admittedly less romantic) view o f corruption in which firms are not just "victims" enduring corruption, but actually adopt corrupt practices themselvesto speed up procedures, avoid red tape, and in general to "grease the wheels" of the administrative system. If being caught in corrupt behavior was costless (moral considerations apart), profit maximizing firms would be willing to pay bribes when the gains in time or costs exceed the bribe amounts. However, since firms that are caught behaving irregularly are often punished in one way or another, one needs to also factor the chances of being caught and the cost associated to the fine (monetary or else) and/or loss ofreputation. But the incidence of bribes is also affected by the incentives on the side of the public servants. That is under an environment where corruption is condoned (implicitly) andnot systemically subjected to oversight and sanctions, public servants will have incentives to engageinthat behavior. Thus public servantswill take into account the gains from asking or accepting bribes against the fines or sanctions if caught. Finally, within countries, firms that have more information on who are the officials that are more amenable to corruption will also face a lower value o f being caught and fined and hence will have 35 more incentives to pay bribes. Thus this differentiates firms' attitudes and actions towards "greasing the wheels" and can explain why some firms do engage in that type o f behavior andwhy others do not. Building on the previous discussion our empirical strategy relies on the following econometric model: where j=firm; c=country; i=industry; Fregion; and s=size. Pr(bribe) is the probability that the firm in question pays bribes, and X is a set of controls that aim at capturing the three hypothesis discussed above: "control rights," bargaining," and "grease the wheels." More specifically among the control set we include the following variables. First, we have the degree o f regulatory compliance variable used in previous chapters which should capture both the impact of quantity and quality o f regulations (as measuredby the percentage of sales declared for tax purposes) and would be aimed at capturing forces underlying the control rights hypothesis. Our prior is that more compliance will be associated with less bribes. Second, we include two variables to account for the possibility o fbargaining by the firm, namely whether the firm has access to credit by the financial sector which would give us an idea of the solvency of the firm and hence o f hture profitability and in line with Svensson (2003) the capital stock per worker of the firm. If the bargaining hypothesis holds, we would expect these variables to carry positive parameters (i.e. more solvent firms should pay more bribes). Third, we include two additional variables to account for the grease the wheels hypothesis. These are whether the firm is foreign owned (a measure of the degree o f information on officials amenable to being corrupted) and the court's enforcement power, and here we would expect both variables to carry negative parameters. Fourth, we experiment with alternative variables that could also capture the three potential determinants o f corruption discussedabove. Table 7 reports some basic results. They indicate that, as predicted by the control rights hypothesis, the degree of regulatory compliance is negatively associated with the probability of payingbribes. Similarly, as predicted by the grease the wheels hypothesis, foreign owned firms have a lower probability o f paying bribes. Inthis line of reasoning, courts that have enforcing power also appear to contribute to lower corruption levels. In contrast, we do not find evidence of the bargaining hypothesis: more solvent firms, indeed, appear to pay less bribes. Although we cannot say why this is so, one possible interpretation o f this result is that more solvent firms have more to lose from the negative reputational effects associatedwith the possibility ofbeingcaught. 36 Table7: DeterminantsofbribepaymentsinLatinAmerica (1) (2) (3) (4) (5) (6) Regulatorycompliance -0.002*** -0.002*** -0.002*** -0.002*** -0.002*** -0.002*** [6.99] [6.95] [6.80] [6.71] [6.79] [6.79] Logofcapitalperworker 0.001 0.001 0.001 0 0.001 0.001 [0.50] [ O S ] [0.62] [0.41] [0.62] [0.62] Foreignowned -0.086*** -0.086*** -0.084*** -0.088*** -0.084*** -0.084*** [3.11] [3.13] [3.05] [3.19] [3.05] [3.02] Access to finance -0.058*** -0.058*** -0.063*** -0.058*** -0.058*** [2.77] [2.77] [3.02] [2.77] [2.78] Trusts courts'enforcementpower -0.050*** -0.049*** -0.050*** -0.050*** [3.15] [3.07] [3.15] [3.16] Requestedpublic services 0.071 *** [4.56] Government is main client -0.014 [0.33] Numberoftax inspections 0 [0.46] Logof firm's age -0.018* -0.017* -0.017 -0.011 -0.017 -0.017 [1.761 [I.68] [1.63] [1.051 [I.62] [1.631 Exporter -0.015 -0.013 -0.013 -0.019 -0.013 -0.012 [0.81] E0.701 [0.69] u.041 [0.70] [0.68] 20-99 employees 0.014 0.019 0.017 0.008 0.017 0.018 [0.80] [1.051 [0.97] [0.46] [0.98] [0.99] 1OW employees -0.03 -0.024 -0.022 -0.036 -0.023 -0.022 [1.24] [0.99] [0.91] [1.47] [0.91] [0.87] Observations 4870 4870 4870 4870 4870 4870 Note: The table reports the results of a probit regression of the frequency of bribe payments on the variables in the first column. The coefficients reported correspond to marginal effects, that is, they reflect the (percentage) change inthe probability of paying bribes ifthere is a one (percentage) point increase in the explanatory variable. All regressions include country, industry, andregion effects. Robust z statistics in brackets. * significant at 10%;** significant at 5%; *** significant at 1% Do these results change when we account for the effect o f the overall country attitude- environment- towards corruption? To explore this issue we experiment with country variables such as the World Economic Forum's Business Costs o f Corruption.'* Our null hypothesis is that bribe payments are more frequent in countries with a wider "acceptance" of corrupt practices. Table 8 reports the results. The results are virtually the same o f those in Table 7, butnow we also find that the probability of firms payingbribes i s very muchinfluenced by the environment. '*In Chapter IVwe also experiment with the IndexofEconomic Freedomobtaining similar results. 37 Table 8. Determinantsof bribepaymentsin Latin America (11) (1) (2) (3) (4) (5 ) (6) Regulatorycompliance -0.002*** -0.002*** -0.002*** -0.002*** -0.002*** -0.002*** [8.46] [8.41] [8.23] [8.12] [8.20] [8.23] Foreignowned -0.075*** -0.076*** -0.073*** -0.076*** -0.073* **-0.073*** [3.60] ~3.641 [3.51] [3.63] [3.5 11 [3.49] Access to finance -0.077*** -0.077*** -0.082*** -0.077* **-0.077*** [4.43] [4.45] [4.74] [4.45] [4.45] Trusts courts' enforcementpower -0.067*** -0.066*** -0.067*** -0.067*** [5.30] [5.24] [5.31] [5.30] Requestedpublic services 0.075*** [6.05] Government is mainclient -0.029 [0.99] Number oftax inspections 0.000 [0.32] Logoff m ' s age -0.01 1 -0.010 -0.010 -0.005 -0.010 -0.010 [1.41] [1.30] [1.28] [0.61] [1.251 [1.271 Exporter -0.001 0.001 0.002 -0.005 0.001 0.002 [0.09] [0.05] [O. 111 [0.33] [0.07] [0.11] 20-99 employees 0.024* 0.030** 0.028** 0.019 0.028** 0.029** [1-77] [2.18] [2.06] [1.36] [2.05] [2.07] 1OW employees -0.030 -0.022 -0.020 -0.037* -0.021 -0.020 [1.571 [1.14] [1.061 [1.921 [1.071 [1.031 Businesscosts of corruption (WEF)'" -0.116** -0.117** -0.103** -0.100** -0.104** -0.103** [2.34] [2.36] [2.06] [2.01] [2.07] [2.07] GDP per capita 0.000 0.000 0.000 0.000 0.000 0.000 [0.03] [0.01] [0.001 [0.06] [0.02] [O.OO] GDP per capita growth 0.023 0.021 0.022 0.022 0.021 0.022 10.731 [0.67] [0.68] [0.701 r0.651 [0.68] Pseudo R-2 0.23 0.24 0.24 0.24 0.24 0.24 Observations 7273 7273 7273 7273 7273 7272 Notes: significant at 10%; * ** significant at 5%; *** significant at1%. Robust z statistics inbrackets. Reported coefficients are marginal effects. Region andindustrydummies are includedinall regressions. (a) Measured on a 1-7 scale, higher values indicatelower costs. Source: World Economic Forum. GDP per capitais PPP inconstant2000 USD. VIII. What are the determinantsof financialaccessin LatinAmerica? Well hnctioning financial systems ameliorate the problems created by information and transaction costs and help allocate resources across space and time. Not surprisingly, a large number of studies have found a robust relationship between the level o f financial development and long-run growth (Beck, Levine, and Loayza, 2000; Loayza, Fajnzylber, and Calderh, 2005). Financial development affects capital accumulation and technological innovation through at least five channels: by facilitating risk management, by reducing the costs of acquiring information about new investment opportunities, by simplifying corporate control over managers, by mobilizing savings and by facilitating exchanges and thus promoting specialization and innovation (Levine, 1997). Yet, while significant attention has been devoted to financial deepening issues, muchless effort has been devoted to the equally critical issue of access to credit - or financial 38 breadth. Chapter 5 relies on the new wave of Enterprise Surveys to explore this topic. It starts by exploring the extent o f financial access in the countries under analysis, together with the determinants of access by firms with a special focus on differences inaccess by small and mediumenterprises. Clearly, a critical element in this context regards access measurement. The analysis in Chapter 5 focuses among others on the following aspects: whether the firm has a checking account (checking);whether the firm has either overdraft, loan, line of credit, or any bank financing for working capital or investment (credit); whether the firm is financial unconstrained (unconstrained) inthe sense that it has not applied for a loan that has been rejected or has not applied for a loan for reasons other than "don't need a loan". On the basis of these three variables we also construct an access index that takes a value of 0 for firms with no checking account, no credit and constrained and a value of 3 for those firms with checking account, credit andbeingunc~nstrained.'~ Figure 15: Financialaccessin LatinAmerica PanelA. Checking PanelB. Credit I PanelC. Unconstrained PanelD. Access index I 1 I Source: Owncalculations basedon Enterprise Surveys. Figure 15 presents these indicators for the countries under analysis. It indicates that there i s significant heterogeneity across countries and across indicators. For example, looking l9 The analysis inChapter 5 is restricted to the most recent wave o f enterprise surveys for 8 Latin American countries. Indeed, the survey instrumentsusedinprevious waves do not allow for the construction of all the variables usedinthe analysis. 39 at the region as a whole 85 percent of the firms have a checking account, but only 68 percent have credit. With respect to the unconstrained index, it indicates that more than on-fourth of Latin American firms would be constrained. As for country differences, it is worth noting that while in Argentina, Colombia and Panama virtually all firms have a checking account, in Mexico only 55 percent of firms have it. More dramatically in Mexico, only 27 percent of firms have access to credit. This would be in contrast with Colombia or Peruwhere 9 out of 10firms have credit. Not surprisingly when we look at our access index, Mexico appears as the country with more access problems (an average score of 1.5). The rest o f the countries have indices above 2 and range between 2.2 in Uruguay and 2.6 in Colombia, Panama and Peru. The averagefor the region is 2.2. Another interesting aspect regards the differences in access by firm size. As indicated by Figurel6, regardless of the indicator that we consider larger firms have more financial access. For example, only 60 percent of small firms have access credit. This would be in contrast to 80 percent among the large firms. Judging from the access index indicator, small firms would have a score o f2.1, whereas large firms score 2.5. To address the question of what determines access to finance, Chapter 5 relies on regression analysis and explores the relative role of different firm characteristics. The main findings are as follows: Subsidiaries of larger corporations have more access and report lower access obstacles. There is no difference between domestic and foreign firms in terms of access to credit, checking accounts, beingunconstrained or inthe access index. There is no difference inaccess indicators for firms ownedby females either. Firms with certified products have more checking accounts, but no significant difference for other access characteristics. Exporters are more likely to have a checking account and any credit, but are also slightly likely to be more constrained (significant only at 15 percent) and they report higher Access Obstacle. Perhapstheir need for finance exceeds their usage o f funds. Firmswith owned land, audited financial statementsand limited liability (i.e. LLC) have more access to finance. Among different industries, it is found that Garments and Food products have less access (the results for Garments are more significant), while no consistent differences are observed for other industries. Among different countries, Colombia, Panama and Peru score higher than the rest o f the countries (i.e. they have most access) even after controlling for a large number of firm characteristics. Bolivia (the omitted category) falls in the middle andMexico is on the lowerend for objective measures ofaccess. 40 Figure 16: Financialaccessin LatinAmerica (firm size) PanelA. Checking PanelB. Credit 90 - 80 - 70 - 85 - I Small Medium Large I I Small Medium Large PanelC. Unconstrained PanelD.Access index 2.6 1 85 1 80 - 2.5 - 2.4 - 75 - 2.3 - 70 - 2.2 - Small Medium Large Small Medium Large Source: Own calculations basedon Enterprise Surveys. Perhaps more interestingly from a policy perspective it appears that access tends to improve with the quality of the courts. For example, Table 9 presents the results from estimating an econometric model relating a number of access indicators to court quality measuresand several firm controls (some o f which have been reviewed above). Each cell in this table corresponds to the slope of a different court quality measure (i.e. each cell correspondsto one regression) related to the use of courts to resolve conflicts (usage), to the share of firms (out o f those that used courts) that received ajudgment on their case (judgment made), and to the share of firms that got their judgment enforced (enforced). We also compute a simple average o f those indicators to obtain the variable court objective. Alternatively we also experiment with four subjective measures based on the perception of the firms: fair, quick, aflordable, and enforceable (the tiles are self- explanatory) plus the averageo f those four indices that we refer to as court subjective. 41 Table 9. Access and Court quality 1 2 3 4 Checking Credit Uncons. Access index Court Objective 0.38 0.46 -0.1 0.29 [0.01]*** [O.OO]*** [0.11l8 [O.OO]*** Court Usage 0.37 0.34 -0.14 0.21 [O.OO]*** [O.OO]*** [O.OO] *** [O.OO]*** Court Judgment Made 0.19 0.25 0.04 0.18 [0.03]* * [O.OO]*** [0.49] [O.OO] *** Court Enforced -0.02 0.07 0.01 0.04 [0.94] [Oh41 [0.77] [0.76] Court Subjective 0.18 0.19 0.05 0.15 [0.011** [O.OO]*** [0.15]" [O.OO]*** Court Fair 0.02 0.1 0.05 0.08 [0.89] [0.37] [0.29] [0.32] Court Quick -0.19 -0.18 0.13 -0.07 [0.12Ia [0.1lIa [O.001 *** [0.46] Court Affordable 0.33 0.28 0.01 0.22 [O.OO]*** [O.OO]*** [0.90] [O.OO]*** Court Enforceable 0.26 0.26 0.03 0.2 ro.oii** ro.ooi*** * significant L a [0.41] L . . at 10%;**significant at 5%;***[0.01]*** significant at 1% Inspection of Table 9 indicates that objective measures of court quality are strongly correlated with access indicators. In fact, all access indicators respond to court quality, except the measureo funconstrained. Among the individual indicators, we findthat Court Usage and Court Judgment Made have a significant positive influence on access indicators (again, except Unconstrained), however the percent of judgments enforced, Court Enforced, does not seem to matter. This could be due to lower variation in the enforcement rates and some missing data as not all regions have enforcement data (we only observe enforcement rates when respondents say that they have used the courts and the courts have made ajudgment). The subjective measures of court quality perceptions are also positively related to access, especially the measures of Court Affordable and Court Enforceable. Court Fair is not significant and Court Quick is even negative (expect for Unconstrained). The aggregate Subjective Court index is significantly relatedto access as well. For comparison purposes all court measures in Table 9 were standardized to have mean zero and standarddeviation o f one. This makes the magnitude comparison much simpler. We note that Court Objective index has about twice as large an effect inmagnitude than Court Subjective. For example, one standard deviation improvement in Court Objective will increasetheAccess Zndex increasesby about 0.3, whereas the same increasein Court Subjective will increaseAccess Zndex only by 0.15. 42 Another determinant of access appears to be the level o f financial development o f the country. In fact, the results of the analysis in Chapter 5 indicate that higher financial development results in larger amounts of loans for small and medium firms. Moreover, higher financial development i s also related to longer loan maturities, particularly for small and medium sized firms. In other words as financial development increases, small and medium f m s appear to find it easier to obtain loans and these loans appear to be grantedinbetter conditions (at leastinterms of loan maturity). IX. Innovation and technologicaldevelopment It is so widely recognized that innovation is a key driver of economic growth that it is almost clichC to say so. Some studies reveal that much of the widening gap between rich and poor countries is due to differences not in capital investment but in technological progress. For example, according to Hall and Jones (1999) and Dollar and Wolf (1997), roughly half of cross-country differences inper capita income and growth are driven by differences in total factor productivity, generally associated with technological progress. Easterly and Levine (2003) also argue that productivity differences explain the lion's share of global income differentials. To the extent that productivity is driven by innovation, both patentable and non-patentable, then we can infer that innovation has become an important ingredient in the new growth agenda. Furthermore, there is empirical evidence that the rates o f return to investments in R&Dcan be very high(Hall andJones, 1998). Yet, in spite of the extensive literature on the importance o f expenditures in research and development (R&D) and science and technology policy to innovation, the distinction between adoption and invention in developing countries should lead us to explore numerous other areas that may posebarriers to the emergence of innovative firms. Infact, there is an emerging literature on what can be called "product" innovation, which focuses on the introduction o fnew products by firms. Hausmann and Rodrik (2003), for example, present a theoretical framework where market failures affecting the introduction of new export products in developing countries might be more severe than those affecting innovation in the developed countries, because in the latter most innovations can be patented, thus providing at least a partial institutional solution to the appropriability problem that inhibits private sector innovation. In developing countries, where most innovations are probably not patentable, other policy instruments would need to be devised to stimulate private-sector investments in product innovation. A related theoretical literature has emphasized the role o f entrepreneurship that is responsible for commercializing research outputs, which are then reflected in the introduction of new products (Michelacci, 2003). Even in the context of high-income countries the determinants of product innovation across firms might be different from those of patentable innovation. Criscuolo, Haskel, and Slaughter (2005) find inapanel of firms from the UnitedKingdom that the correlates o f patents and product innovation are different, particularly with respect to the role played by linkages between firms and universities, the latter being more important for 43 patentable innovations. Another example is the study by Aghion et al. (2006) that found that the response of U.K. firms (measured by productivity changes and patenting) to increased competition (due to the regulatory reforms o f the Thatcher government) was different across firms, depending on their distance to the technological frontier (proxied by the productivity gap with respect to the most product firms in each industry). Yet we still have much to learn about the empirical correlates o f product innovation in developing countries. Against this background, Chapter 6 examines the empirical determinants of firm-level product innovation in a large sample o f manufacturing firms, covering at least 36-60 developing countries, 8 manufacturing industries, and totaling thousands of firms, depending on the empirical model. It must be noted, however, that the focus of many o f the studies in this area has been on the determinants o f R&D investments (Park, 2004; Hall, 2002; Nelson, 2000; among others) under the understandingthat R&Dis a driver o f innovation and because R&D data are usually widely available while innovation data much less so. However R&D remains an intermediate input, while innovation is the final and desired outcome. Chapter 6 takes advantage of the fact that the Enterprise Survey data provide information on firms' innovation outputs, which allows for directly investigating the determinants ofproduct innovation at the firm level. More specifically, Chapter 6 addresses two questions: First, is R&D investment correlated with product innovation in developing-country firms? If so, then the "D' in R&D - investments in product development - might be an important correlate of the propensity to innovate by f m s in developing countries even when such innovations are notpatentable. Second, is there evidence of market failures that wouldjustify government involvement to raise private-sector investments inproduct innovation? Inthe presence o f market failures, aspects of the investment climate associated with the extent of market competition can have unexpected effects on the private firms' propensity to introduce new products, especially among firms that are farthest from the global technological frontier. For instance, regulatory reforms that facilitate the entry o f firms could raise the prospectsof imitation, thus leading entrepreneursto reduce their innovation expenditures. To answer these questions Chapter 6 presents estimates of reduced-form models o f product innovation that consider the possibility that R&D expenditures and perhaps the sales variables that was recorded in the firm surveys are measured with error (or in an alternative model licensing). This is important because if they are, the standard direct regression model with product innovation as the dependent variable and the R&D/Sales variable (our proxy for research-cost intensity derived from the theoretical model) might be biased, possibly suffering from attenuation bias if the measurement error is random. To assess the influence ofmeasurement errors the chapter follows Leamer (1978, chapter 8) and in addition to the direct estimates it also reports reverse regression models. Inthis approach the dependent variable becomes the variable o f interest and the variable o f interest the dependent variable. Thus, if the innovation variable is measured accurately, whereas the R&D/Sales or licensing are measured with error, then the inverse o f the estimated coefficients from the reverseregression is the "true" partialcorrelation between product innovation and R&D/Sales or licensing. 44 Table 10. Are R&D Expenditures Related to Productive Innovation? Direct versus Reverse Regressions (1) (2) (3) (4) DepedentVariables --> New Product R&D/Sales New Product Licensing EstimationMethod-3 dProbit dTobit dProbit dProbit R&D/Sales 0.019 (0.71) Licensing 0.127 (3.41)** Newproduct 0.202 0.038 (2.32)* (3.64) ** Employees (log) (firm-level) 0.084 1.404 0.085 0.05 (4.30)** (2.62)** (3.66)** (3.44)** EmployeesY (log) (firm-level) -0.003 -0.295 -0.005 -0.002 (1-25) (2.50)* (1.80) (1-40) Foreignownership(firm-level) (d) 0.002 0.003 -0.009 0.124 (0.12) (0.17) (0.54) (9.31)** Capacityutilization (firm-level) 0.001 -0.171 0.001 0 (1.74) (a) (1.56) (0.85) Export status (dummy) (firm-level) (d) 0.081 0.032 0.082 0.012 (3.87)** (2.01)* (3.74)** (1.70) Average years of educationof employees (log) 0.025 0.005 0.03 0.014 (1.86) (0.20) (1.76) (2.66)** Observations 11924 11924 18587 10898 Countries 59 60 60 CensoredObservations ___59 3876 --- --- Industry andSurvey-YearDummies Yes Yes Yes Yes Countrv Dummies Yes No No Yes Note:Robustz statistics inparentheses;standarderrorsare clusteredaroundcountries. * significant at 5%; ** significant 1%.Country, industry, andsurvey-yeardummiesare not reported. at What are the results o f this exercise? According to Table 10, which reports the marginal coefficients, or the elasticities calculated at the sample mean when we look at the direct regression relating the probability of including a new product to R&D/Sales, our variable of interest is not statistically significant and the point estimate o f the elasticity is negligible. However, the results from the reverse regression models (presented under column 2) indicate that the estimated coefficients are highly significant. Furthermore, its inverse implies a rather large partial correlation between R&D/Sales and product innovation. In fact, the elasticity of the probability of introducing a new product with respectto R&D/Sales would be about 5. Columns 3 and 4 report the corresponding estimates for an alternative exercise with a different variable: licensing. The literature on innovation has paid much attention to the adoption of foreign technologies and hence this would justify using this variable as an alternative innovation input. In this case, the direct regression results suggest that licensing is positively correlated with product innovation. Nevertheless, the implied 45 marginal effect estimated with the reverse regression model (Column 4) is significantly larger, thus also suggesting the there might be measurement errors in the licensing variable as well and that direct regressions estimates may be downwardbiased and hence underestimate the effect of licensing. On the second question above (i.e. on whether there is evidence of market failures that would justify government involvement to raise private-sector investments in product innovation), Chapter 6 argues that trade policies and the investment climate play an important role for product innovation. In particular, an increase in the number o f exporters is associated with a higher probability of undertaking product innovations. Moreover, where tariff levels are low and concentrated - as opposed to dispersed - innovation rates at the firm level also tend to be higher. With regard to FDI, the results suggest that foreign-owned firms are not more likely to introduce new products than their domestic counterparts in developing countries, but they are likely to produce goods that are different from those of their local competitors - even if they are not new on a global scale - which may lead to significant positive spillovers effects. Finally, the chapter shows that whereas regulatory reforms that facilitate firm entry are desirable to increase competition and knowledge diffusion, they also have the effect of facilitating imitation and thus reducing firms' propensity to introduce new products. This implies that, especially after trade and regulatory reforms, governments should consider the implementation of well monitored programs to stimulate private-sector innovation. X. Exporters do itbetter Many countries in the Latin American region have recently been negotiating free trade agreements with the U.S. and the European Union, among other important trading partners. Many o f them have put those agreements at the center of their development strategies, which is not surprisinggiven the largebody o f evidence linkingtrade openness to productivity and growth. Inthis context, Chapter 7 o f Volume I1explores one specific channel through which trade openness can produce positive development results, namely the link between firms' exporter status and higher wages and productivity. This is confirmedby the evidence reported inChapter 2 andit is further illustrated in Table 11. This table reports estimated productivity and wage exporter premiums (based on Enterprise Survey data) for each o f the countries under analysis in this report and for a pooled sample o f all the 16 countries. These parameters are obtained by running simple OLS regressions with TFP and average wages as dependent variables on indicators capturing exporter status - whose coefficient is reported in the table as well as firm age, size, foreign ownership, unique establishment, log o f capital per worker, and region and industry dummies (we also included country dummies when running the pooled regression). Inspection of Table 11 suggests that in 14 of the 16 Latin American countries in the sample and inthe pooled sample there is positive total factor productivity (TFP) exporter premium, signaling that exporters are more productive. Similarly, in 15 o f the countries and in the pooled sample there is a positive wage exporter premium, indicating that 46 exporters pay higher wages. True, only around half o f these positive exporter premiums are statistically significant (7 in the case of TFP and 11 in the case of wages), which indicates that these are not estimated very precisely. However, for Latin America as a whole the productivity exporter premium is of around 40 percent, and exporters tend to pay wages that are on average 23 percent higher than those paidby firms of the same size and age, and operating inthe same regions, industries and countries. Table 11.Productivityandwage exporter premiumsin LAC Premium L o g o f Premium Log o f country TFP wages Country TFP wages L A C 0.339*** 0.203*** Guatemala 0.153 0.347** [0.035] [0.026] [0.148] [0.143] Argentina 0.597*** 0.09 Honduras 0.22 0.208* [0.111] [0.074] [O. 1631 [0.126] Bolivia -0.353 * 0.084 Mexico 0.001 0.124 [0.193] [O. 1631 [0.132] [0.096] Brazil 0.478*** 0.274*** Nicaragua -0.02 -0.019 [0.070] [0.054] [O. 1091 [0.080] Chile 0.14 0.188 Panama 0.024 0.257** [0.138] [0.133] [0.3621 [O. 1251 Colombia 0.304*** 0.275*** Paraguay 0.485 0.052 [0.083] [0.071] [0.297] [O. 1791 Costa Rica 0.011 0.538*** Peru 0.520*** 0.297** [0.009] [O. 1771 [0.160] [O. 1213 Ecuador 0.091 0.017 Uruguay 0.678*** 0.465* ** [0.259] [O. 1731 [O. 1851 [O. 1051 ElSalvador 0.357*** 0.305** * 10.1261 [0.0921 Note: The estimates inthe table are based on an OLS regression of either TFP or the log o f wage per worker on an exporter dummy (to which the reported parameter corresponds) and on a set o f control variables to capture firm, region and industry characteristics. Standard errors are provided in brackets and percent level, ** * stands for statistical significance at the 5 for statistical significance at the 10percent level and *** for statistical significance at the 1percent level. What is behind these exporter premiums? There are two non mutually exclusive explanations: either firms become better as they engage in exporting (learning-by- exporting) or better firms become exporters (self-selection). Under the first hypothesis exporting improves productivity. The most common explanation for this, known as "learning-by-exporting," is that exporters acquire information from foreign customers on how to improve product designs, their manufacturing processes or the quality o f their goods and services. In addition, foreign demand may allow domestic firms -particularly insmall countries- to take advantageofunexploited economiesofscale. Under the second hypothesis the best firms self-select into export markets. One rationale for this i s that there exist important entry barriers into exporting markets due to higher costs associated with selling in foreign markets (transport, but also distribution, marketing and even production costs when firms need to adapt their product to foreign 47 standards). Thus, only the more productive firms become exporters, and the observed advantagepre-date their entry into foreignmarkets. These two hypotheses are obviously not mutually exclusive, but depending on which force is more important, the policy implications can be very different. On the one hand, export promotion activities, which are quite common in Latin America, are oftenjustified on the basis o f the "learning-by-exporting" explanation. On the other hand, the self- selection explanation would suggest that policy makers should focus their efforts on the internal determinants ofproductivity growth. Thus, what does the literature say regarding the relative importance of each hypothesis? The truth is that there is no clear-cut answer in the existing literature regarding the relative strength o f the "self-selection" versus "learning-by-exporting" hypotheses. By nature this literature is country specific, and depending on the country examined, the studies seem to reach different conclusions. A review o f 54 studies that look at the productivity premium associatedwith export activity presented inChapter 7 o fVolume I1 indicates that in 86 percent of the cases exporters are found to be more productive than non-exporters and that inmost o f these studies -with rare exceptions- there is evidence of self-selection: good firms become exporters, suggesting that penetrating foreign markets may require higher productivity. About 60 percent of the studies test the learning-by-exporting hypothesis, but the evidence is mixed on this front. Half of the studies finds support for the learning-by-exporting hypothesis, and the other half find no evidence that there are post-entry-into-exports differences in productivity growth between new exporters andnon-exporters. Thus, on the whole, the messages coming from the literature are that (i) exporters indeed are more productive that non-exporters; (ii) firms do tend to self-select into the export market, (iii) exporting does not always improve productivity (or only half of the time); and (iv) there is substantial heterogeneity across countries in terms of the relevance of the learning-by-exporting hypothesis. A critical element that appears to be behind this heterogeneity is the size of the country. In other words, we are more likely to find that exports cause productivity premiums in small countries. On the contrary, there is no evidence indicating that the level of development, degree of trade openness, or investment climate affect the tendency to learn by exporting. This finding would somehow give prominence to the economies of scale rationale for productivity premiums rather than to the knowledge acquisition hypothesis. Taking all this information together suggests that export promotion activities may be a better idea in smaller countries, whereas in larger countries policy makers may want to consider activities more aimed at promoting across-the-board productivity improvements. Evidence elsewhere indirectly suggests that the latter is probably more important in developing countries. Indeed, a recent paper by Lederman, Olarreaga and Payton (2006) indicates that indeveloping countries returns to on-shore exportpromotion activities such as technical assistance and training for (large) domestic firms on how to enter foreign markets were much larger than returns to off-shore export promotion activities such as country image, fair participation, and other marketing activities abroad, such as foreign 48 offices. They also find that export promotion activities should focus on large domestic firms that are not yet exporting rather than on established exporters in order to maximize their impact on aggregate exports. These findings are consistent with the idea that self selection plays an important role inthe explanation o f export premiums. Evidence for the role o f larger firms in this context also emerges from a country study for Uruguay discussed inChapter 7, which shows that more than 80 percent of the output differential between exporters and non exporters is due to differences insize. On the link between exports and wages, Chapter 7 presents a review of 30 studies that have explored wage premiums associated with export activity. This review indicates that in two-thirds of the studies there is evidence of an overall wage premium and in fact, in all but two studies, there is evidence o f large skilled wage premiums, whereas unskilled workers in the export sector benefit from a premium in only 45 percent of the studies. When we relate these findings to country characteristics we find once again that wage premiums are more likely to be observed in small countries. This again provides tangential evidence to the importance for f m s in small countries to be able to take advantage of unexploited economies o f scale in world markets. Also, unskilled wage premiums are more likely to be observed in skilled abundant countries. This may be due to the fact that exporters may need to pay higher wages to attract the rare factor (unskilled workers) into their f m s incountries which are relativelyabundant in skilled workers. Unfortunately, the issue of causality mentioned above (i.e. whether firms that pay more tend to be exporters or whether it is exporters who tend to pay more) is seldom addressed in this literature. While many of these papers are based on panel data that allow for controls of fixed effects and unobserved heterogeneity at the firm level, the issue o f causality from exports to wages remains largely unsolved. In this regard, Chapter 7 presents new evidence on the causality of exports to wages based on a country study for Argentina that exploits the Brazilian devaluation o f 1999 to identify a causal relationship between exports and wages. Argentina and Brazil are major trade partners and the Brazilian devaluation had a large impact on Argentine exports. The combination of the panel dataset and the devaluation shock is an important instrument to address this problem because it is possible to consider the same firm before and after the devaluation and to see how the wages paidby this firm changewhen there is an exogenous change in exporting opportunities. Thus these data provide us with the opportunity to determine causality from exports to wages. What are the messages that emerge from this exercise?The major conclusion of this work i s that while exporting to high income countries improves wages, the ratio o f exports to sales does not affect them. This means that exporting per se is not really a significant channel towards higher wages, but exporting to high income countries is. That is, what appears to matter is the composition o f exports. These results which are robust to a number of departures from our basic specification, also appear to be important in magnitude. One hypothesis of why this is so is that exports and export destinations allow firms to make higher profits (by selling to more profitable markets) that are henceforth shared 49 with the workers. A second hypothesis is that higher income countries have a higher valuation for high quality exports so that exporting to those countries requires more skills, higher wages, and a higher skill premium. Can we use our results and data to explore which of these two hypotheses seem to be more relevant inArgentina? We do not have information on wages by skill levels but we do have the composition o f skills o f the firm's employment. We can thus ask whether exporters, and in particular exporters to high income countries, have a different skill composition. The results in Chapter 7 suggest that whereas the ratio of exports to sales does not seem to have an effect on the composition of skills at the firm level, exporting to high income countries do matter. In fact, the intensity of use o f skilled labor in firms that export to high income countries is between 10and 14percent higher than in other firms. These results provide some support to those hypotheses that require increases of the skill composition following exports (especially to high income countries), like the quality story. Indeed, in the profit sharing model, there is actually no reasonto expect increasesinthe skill composition following a higher exposure to export markets. Onthe whole our analysis seems to favor the quality hypothesis over the profit sharing hypothesis. 50 References Andres, L.A., J.L. Guasch, and S. LopezAzumendi(2007). "Infrastructure, Regulatory Governance, and Sector Performance, The World Bank. Andres, L.A., J.L. Guasch, V. Foster, and T. Haven(2007). WhatReally happenedwith private Sector Participation in Injhastructure. The World Bank. Ades, A., and R. DiTella (1996). "The New Economicsof Corruption: A Surveyand SomeNew Results," inLiberalization and theNew Corruption, B. Harris and G. White (eds.), IDS Bulletin. Aghion, P., R. Blundell, R. Griffith, P. Howitt, and S. Prantl (2006). "The Effects of Entry on Incumbent Innovation and Productivity." Harvard University, Cambridge, Massachusetts,http://post.economics.harvard.edu/faculty/aghion/papers.html. Beck, T., R. Levine, andN.Loayza(2000). "Finance andthe sources of growth," Journal ofFinancia1 Economics, 58(1-2): 261-300. Blyde, J. andE.Fernandez-Arias (2004). "Why Does Latin America Grow More Slowly?," Inter-AmericanDevelopmentBank,mimeo. Brunetti,A. andB.Weder (2003). "A FreePressis BadNews for Corruption," Journal of Public Economics, 87: 1801-1824. Calderh, C., W. Easterly, and L.ServCn (2003). "Infrastructure Compressionand Public Sector Solvency in LatinAmerica" inEasterly and Servtn (eds.) TheLimits of Stabilization-Infrastructure,Public DeJicits and Growth in Latin America. Calderbn, C. andK.Schmidt-Hebbel(2003). "Macroeconomic policies andperformance inLatinAmerica," Journal of InternationalMoney andFinance,22(7): 895-923. Calderdn, C., and L. Servtn (2003). "The Output Cost of LatinAmerican's Infrastructure Gap", inEasterly and ServCn (eds.) TheLimits of Stabilization-Infiastructure, Public DeJicits and Growth in Latin America. Criscuolo, C., J. Haskel, and M. Slaughter (2005). "Global Engagement and the Innovation Activities of Firms," de Ferranti, D., G. Perry, I.Gill, J. Guasch, W. Maloney, C. Sbchez-Pdramo, and N. Schady (2003). Closing the gap in education and technology. The World Bank, Washington, DC. De Gregorio, J. and J.-W. Lee(2003). "Economic Growth and Adjustment inEastAsia and Latin America," Documentode TrabajoNo. 245, Banco Centralde Chile. Demirguc-Kunt,A,, 1.Love, andV. Maksimovic (2006). "Business environmentandthe IncorporationDecision," Journal of Banking andFinance, 30: 2967-2993. Dollar, D., and E. Wolf (1997). "Convergence of Industry Labor Productivity among AdvancedEconomies, 1963-1982," inEdwardN.Wolff (Ed) TheEconomics of Productivity. UnitedKingdom: Elgar. Easterly, W. and R. Levine (2003). "Tropics, Germs and Crops: How Endowment Influenced EconomicDevelopment," Journal of Monetary Economics, 50( 1): 3-39 51 Easterly, W. and S. Rebelo (1993). "Fiscal Policy and Economic Growth: An Empirical Investigation," Journal of Monetary Economics, 32: 417-458. Easterly, W., and L. Servtn (2003). The Limits of Stabilization: Znfiastiwcture, Public Deficits and Growth in Latin America. Stanford University Press, Palo Alto, CA and the World Bank, WashingtonD.C. Escribano, A. andJ.L. Guasch(2005). "Assessing the Impact ofthe InvestmentClimate on Productivity Using Firm-LevelData," World Bank Policy ResearchWorking Paper No. 3621. Escribano, A., J.L. Guasch, andM.de Orte (2006). "Investment Climate Assessment (ICA) on Productivity and on Allocative Efficiency: Effects on Exports, ForeignDirect Investment, Wages and EmploymentAnalysis Basedon FirmLevel Data from 2001- 2003," mimeo, The World Bank, Washington, DC. Esfahani, H., and M. Ramirez (2003). "Institutions, Infrastructure and Economic Growth," Journal of Development Economics 70: 443-77. Fajnzylber,P., D. Lederman, andN.Loayza(2002). "Inequality andViolent Crime," The Journal of Law and Economics, 45: 1-39. Fernhdez-Arias, E., and P. Montiel(2001). "Reform and Growth in LatinAmerica: All Pain, No Gain?," IMF StafSPapers,48( 3): 522-546. Gasparini, L., F. Gutierrez, and L.Tornarolli (2007). "Growth andincomepoverty in LAC: evidence from householdsurveys," Review of Zncome and Wealth, forthcoming. Goel, R. andM.Nelson(1998). "Corruption and GovernmentSize: A Disaggregated Analysis." Public Choice 97: 107-120. Gramlich, E.(1994). "Infrastructure Investment:A Review Essay," Journal of Economic Literature, 32:1176-1196. Guasch J., (2005). Granting and Renegotiating Infiastructure Concessions: Doing it Right. World Bank Institute DevelopmentStudies Report28816. Hall, R. (2002). "Industry Dynamics with Adjustment Costs," NBER Working Papers 8849 Hall, R., andC. Jones (1998). Hall, R., and C. Jones (1999). "Why do some Countries Produce so much more Output per Worker thanothers?," The Quarterly Journal ofEconomics 114: 83-116. Hallward-Driemeier M.andD. Stewart(2004). "How Do InvestmentClimate Conditions Vary Across CountriesandTypes of Firms?," mimeo, The World Bank. Hausmann, R, and D. Rodrik (2003). "Economic Development as Self-Discovery." Journal of Development Economics, 72: 603-633. HellmanJ., G.Jones, and D. Kaufmann(2000). "Seize the State, Seize the Day: State Capture, Corruption, and Influence inTransition Economies," World BankPolicy ResearchWorking PaperNo. 2444. 52 Hendry, D.andG. Mizon (1978). "Serial Correlationas aConvenientSimplification, Not aNuisance:A Comment on a Study of the Demandfor Money by the Bank of England," TheEconomic Journal, 88(35 1): 549-563. InternationalOrganizationfor Standarization(2005). htb://ww.iso.org/ Kaufmann, D.,A. Kraay,. and P. Zoido-Lobath (1999). "Governance Matters," World BankPolicy ResearchWorking PaperNo. 2196. KaufmannD.and S. Wei (1999). "Does `Grease Money' Speed Up the Wheels of Commerce?," NBERWorking PaperNo. 7093 Kruger, D., R. Soares, and M.Berthelon, Matias(2006). "Household Choicesof Child Labor and Schooling: A SimpleStructuralModelwith Application to Brazil," mimeo, University ofMaryland. LaPalombara,J. (1994). "Structural and Institutional Aspects of Corruption," Social Research 61(2), pp. 325-50. Learner, E. (1978). SpeciJcation Searches: Ad Hoc Inference with Nonexperimental Data.New York, JohnWiley & Sons. Lederman, D.and W. Maloney (2003). "Research and Development(R&D) and Development," The World Bank Policy ResearchWorking Paper Series 3024. Leite, C. andJ. Weidmann (1999). "Does Mother Nature Corrupt?NaturalResources, Corruption, and, EconomicGrowth," IMF Working Paper WP/99/85. Loayza, N.,P. Fajnzylberand C. Calder6n(2005). Economic Growth in Latin America and the Caribbean. StylizedFacts, Explanations, andForecasts. The World Bank, Washington, DC Loayza, N.,Oviedo, A. M.and Servbn, L.(2004). "Regulation andMacroeconomic Performance," The World Bank Policy ResearchWorking Paper No. 3469. Loayza,N., A.M. Oviedo, and L. ServCn (2005). "The Impactof Regulationon Growth and Informality: Cross-Country Evidence," World Bank Policy ResearchWorking Paper No. 3623. Loayza, N.andJ. Rigolini (2006). "Informality Trends and Cycles," World Bank Policy ResearchWorking Paper No. 4078. Maloney, W. and Rodriguez Clare (2006). "Innovation Shortfalls," mimeo, The World Bankand IADB. Washington, DC. Mauro, P. (1995). "Corruption and Growth," The QuarterlyJournal of Economics, 110: 681-7 12. Michelacci, C. (2003). "Low Returns inR&D Dueto the Lack of EntrepreneurialSkills," Economic Journal, 113: 207-225. Montinola, G. andR.Jackman(2002). "Source of Corruption: a Cross-country Study," BritishJournal of Political Science, 32: 147-170. Nelson, R. (2000). TheSources of Economic Growth. HarvardUniversity Press 53 Park, W. (2002). "Institutions andIncentivesfor R & D: Implications for LAC Economies." Studypreparedfor World Bank, LatinAmerica andthe Caribbean Flagship2002. Perry, G., 0.Arias, H. Lopez, W. Maloney and L.ServCn (2006). "Poverty Reduction andGrowth: Virtuous andVicious Circles," The WorldBank. Perry, G., W. Maloney, 0.Arias, P. Fajnzylber, A. Mason andJ. Saavedra (2007). "Informality: Exit and Exclusion," The World Bank. Prescott, E. (1998). "Needed: A Theory ofTotal Factor Productivity," International Economic Review, 39(3): 525-5 1. Rauch, J. andP. Evans(2000). "Bureaucratic StructureandBureaucraticPerformancein Less DevelopedCountries," Journal of Public Economics, 75(1): 49-7 1. Shleifer, A. and R. Vishny (1993). "Corruption," The Quarterly Journal of Economics, 108(3): 599-617. Svensson, J. (2003). "Who Must Pay Bribes and How Much? Evidence from a Cross SectionofFirms," The Quarterly Journal of Economics, 118(1): 207-230. Treisman, D. (1999). After theDeluge, University ofMichigan Press, Ann Arbor. World Bank (2006). Doing Business in 2006: Creatingjobs, World Bank and InternationalFinanceCorporation, Washington, DC. World Bank (1997). WorldDevelopment Report 1997: TheState in a Changing World, The World Bank, Washington, DC. World Bank, (2005). WorldDevelopment Report: A Better Investment Climatefor Everyone. World Bank, (2006). WorldDevelopment Report: Equity and Development World Bank(2007). WorldDevelopment Indicators. World Economic Forum (2006). "The Latin America Competitiveness Review 2006. Pavingthe Way for RegionalProsperity," Geneva, Switzerland 54