Enter your keyword

post

Financial liberalization and economic growth: an analysis of the linkages between them in developing countries

Financial liberalization and economic growth: an analysis of the linkages between them in developing countries

Free dissertation sample from the best dissertation writing services

CHAPTER ONE: INTRODUCTION

Background of the study

The relationship between financial liberalization and economic growth has long been a topic of interest and debate. Dating back to 1912, Schumpeter recognized the potential importance of financial sectors in promoting economic growth. Many followers argue that the services the financial sector provides – of reallocating resources to the highest value use without substantial risk of loss through information costs or transaction costs – are essential for economic growth. In contrast, numerous economists believe that finance is a relatively unimportant factor in economic development. For instance, Robinson (1952) contends that financial development simply follows economic growth. Lucas (1988) declares that the finance-growth relationship is “over-stressed”. Although conclusions must be stated hesitantly, a growing literature, especially in empirical studies, suggests a strong, positive relationship between the level of financial development and economic growth. There is even evidence that the level of financial development is a good predictor of future rates of economic growth, capital accumulation, and technological change (e.g. King and Levine 1993a, 1993b, Beck, Levine and Loyaza 2000, Demirguc- Kunt and Maksimovic 1998, and Rajan and Zingales 1996).

One of the main challenges for capital-scarce developing countries is to stimulate saving to increase investment and achieve high growth rates. Consequently, most of the recent policy recommendations to these countries have focused on the role of financial liberalization policies to attract capital flows, which are expected to provide additional resources to reach higher levels of capital accumulation and growth. After the 1990s, most developing countries undertook widespread financial liberalization policies and began to attract massive amounts of capital inflows. As capital flows to and from emerging countries increased continuously, the effects of these capital flows on investment became a source of discussion. It was widely accepted that the flow of resources to developing countries increases investment by reducing the cost of capital (Fischer, 1998, Summers, 2000). On the other hand, it was also argued that financial liberalization policies generate speculative capital flows and therefore do not have any significant effect on economic growth and capital investment (Bhagwhati, 1998, Rodrik, 1998).

Several empirical studies emerged to assess the impact of capital flows on economic growth. Borenzstein et al. (2001) found that FDI has a positive effect on economic growth when the level of education in the host country, a measure of its absorptive capacity, is high. Gruben and McLeod (1998) focused on cross-sectional variation by using annual data and found that there is a significant relationship between various types of capital flows and GDP growth. Bosworth and Collins (1999) analyzed the impact of capital flows on domestic investment in 58 developing countries between 1978-1998 by distinguishing types of capital flows and found that an increase of a dollar in capital inflows is associated with an increase in domestic investment of about 50 cents. Blanchard and Giavazzi (2002) studied the opening of Greece and Portugal, in the context of their joining the European Monetary Union, and documented that capital flows financed increased investment and consumption. A comprehensive study by Mody and Murshid (2005) also concluded that in the 1990s, even as liberalization attracted new flows, foreign capital stimulated less domestic investment than in the preceding decade as governments accumulated more international reserves and residents diversified by investing abroad.

The main idea behind most of these empirical studies is that by easing restrictions on capital mobility, developing countries can attract capital flows to meet their financing requirements, which helps boost investment (Fischer, 2003, Summers, 2000). However, it is observed in many developing countries that residents are actually moving scarce capital to the more advanced countries, worsening their financing problems. This process is termed “capital flight” and has come to be viewed as a major economic problem in many developing countries.

Statement of the problem

During the debt crisis in the early 1980s, it was widely discussed that the occurrence of capital flight severely constrains the development of economies by reducing investment. Since then, capital flight has been a topic of research and policy debate. Still, much of the existing literature focuses on the different measures and determinants of capital flight (Lessard and J.Williamson, 1987, Lensink et ah, 2002, Schneider, 2003a) as well as on the relation between capital flight and other macroeconomic outcomes such as low rates of growth (Varman-Schneider, 1987), increased aid flows (Collier, Hoeffler and Pattilo, 2001), high external debt (Boyce and Ndikumana, 2001, Cerra et ah, 2005), and financial liberalization (Lensink et ah, 1998). There are only a few studies investigating the effects of financial liberalization on economic growth and on the (Kadochnikov, 2005).

Since understanding and evaluating the nature and impact of financial liberalization is crucial for implementing necessary policy responses, it is important to analyze the effects of financial liberalization on economic growth. Also, it is equally important to examine whether this impact has changed over time with financial liberalization policies, because these policies have been the dominant pattern in offering solutions to macroeconomic problems in many developing countries.

In countries with capital mobility restrictions, capital flight may be very costly considering that a decline in domestic savings caused by capital flight leads to decreases in investment and growth. On the other hand, if capital flows have a positive effect on investment in developing countries, as suggested by previous empirical studies, then those countries, which adopted measures towards financial liberalization to attract foreign investment, will not experience the harmful effects of capital flight on investment to the full extent. Consequently, with open capital accounts, the negative effect of capital flight on investment is expected to be lower because the real interest rate differential between capital abundant developed countries and capital scarce developing economies would generate a spontaneous flow of funds that would provide the additional foreign savings required for new investment1. When this is the case, even if domestic savings decrease as a result of capital flight, there may be no change in the investment since a reduction in investment may be offset by an increase in capital inflows.

Research Questions

In order to assess the impact of financial liberalization on economic growth, the author will attempt to answer the following research questions:

  1. How important is financial liberalization?
  2. Does economic growth have any relationship with financial liberalization policies?
  3. W hat role do capital flows play in this context?

In the following chapters the author will analyze the financial liberalization policies from developing countries. This will help in understanding the impact of financial liberalization on economic growth.

Structure of the study

The remainder of the dissertation is structured in five chapters. In chapter two, the author will review the literature on financial liberalization policies. Chapter three explains the theoretical background of the relationship between financial liberalization and investment. This is followed by a description of the data sources and the methodology adopted and empirically examines the relationship between capital account liberalization in chapter four. Chapter five concludes the research work.

CHAPTER TWO: REVIEW OF THE LITERATURE

Introduction

Financial deregulation policies aim at reducing the cost of credit and broadening its availability. If we consider a cross-section of countries in 2000, we can in fact observe a positive association between the degree of financial liberalization and financial depth, measured as the ratio of private credit to GDP (See figure 1). Several studies have carefully shown that the liberalization of financial markets reduces financial constraints and broadens the availability of credit. For example, Laeven (2003) studies several developing countries and finds that financial liberalization reduces firms’ financing constraints. Tressel and Detragiache  (2008) study a larger group of countries over the last three  decades and find that  financial reform results in deeper financial markets,  measured as bank credit to the private  sector.

Evolution of financial liberalization

Figure 1: Evolution of financial liberalization through the  1975-2005 period for six groups of countries. Source:  Abiad et al. (2010)

Since financial liberalization increases financial depth, it has the potential of contributing to higher economic activity. Several studies have provided evidence that financial deregulation leads to higher economic growth (Galindo et al., 2002; Gupta and Yuan, 2009; Levchenko et al., 2009).  In order to resolve causality issues, these studies have documented in detail the mechanisms through which deregulation influences growth. In particular, using the methodology proposed by Rajan and Zingales (1998), these papers use industry-level data and find that financial liberalization increases economic growth by reallocating resources towards the industries that are more financially constrained.

Although a large literature finds that financial liberalization produces faster average growth, researchers have not yet determined clearly whether deregulation benefits the whole population equally, or whether it disproportionately benefits the rich or the poor. This study attempts to fill this gap.

In an attempt to understand in more detail the relationship between financial liberalization and economic growth, several papers have documented that deregulation increases growth primarily by boosting total factor productivity (TFP) (Levine, 2005; Levchenko et al., 2009; Bekaert et al., 2011). Figure 2 presents cross-sectional evidence consistent with this finding. However, since these papers use aggregate data (country or industry- level), they cannot analyze the factors leading to these TFP gains.

Wage inequality, defined as the relative wage between skilled and unskilled labor, increased substantially starting in the 1980s in several OECD countries, such as the U.S., the U.K., and several others. Although the dynamics of wage inequality have been well documented, there is still disagreement about their causes. Several explanations have been proposed, including skill-biased technical change, globalization and trade liberalization, and changes in wage setting institutions. However, little attention has been paid to the role of financial markets in this process. An interesting fact is that, at the same time that inequality began to increase, many countries dramatically liberalized their financial markets. Figure 3 shows that, in a cross-section of countries,  wage inequality is positively correlated with the state of financial liberalization. Figure 4 shows a case study, the U.S., where the dynamics of wage inequality closely followed the dynamics of financial liberalization. This evidence suggests that financial liberalization might have been an additional factor contributing to the increase in wage inequality

Wage inequality

Figure 2: State of financial liberalization and wage inequality in a cross-section of countries

In this chapter, it is argued that the liberalization of financial markets has widened the wage gap between skilled and unskilled workers and has therefore contributed to the rise in wage inequality in many developed countries. To identify the causal effect of financial liberalization on inequality, the author focuses on a theoretical mechanism through which finance affects inequality. According to theory,  financial liberalization should improve the efficiency of financial intermediation, alleviating firms’ borrowing constraints and increasing capital demand. If the production functions of firms exhibit capital skill complementarily meaning that capital and skilled labor are relative complements (as the evidence indicates) the demand for skilled labor should increase by more than the demand for unskilled labor. As a result, wage inequality, which is the relative price of skilled labor, will increase in equilibrium.

Financial liberalization should have particularly large effects when there is a large increase in capital demand and a large increase in the relative demand for skilled labor. Therefore, the effect of liberalization on inequality should be increasing in the extent of financial needs and in the degree of capital-skill complementarity (henceforth CSC). Given that industries are heterogeneous regarding these two dimensions, financial liberalization should have heterogeneous effects on wage inequality across industries.  To estimate the causal effect of the reform on inequality, the author exploits cross-industry differences in financial needs and CSC. The identification assumption is that there aren’t other concurrent policies or shocks that increase wage inequality exclusively in the subset industries with both high financial needs and strong complementarity.

Related Literature

This paper is related to several strands of the literature. First, it contributes to the recent literature on financial deregulation and inequality. Beck et al. (2010) and Jerzmanowski and Nabar (2011) use different methodologies to analyze the effect of branch deregulation on inequality in the U.S. While the former paper finds that deregulation decreased wage inequality, the latter finds the opposite result.  While they both identify the effect by exploiting  differences in timing  of deregulation  across states,  the author exploits differences in external finance and CSC across industries within a state (or country). Documenting evidence of a specific mechanism by which finance affects inequality provides a stronger test of causality. In addition, besides analyzing deregulation across states in the U.S, I analyze deregulation across a large group of countries.

This paper adds to a growing literature studying the real effects of financial liberalization. There are several papers that have studied the effects of liberalization on economic growth using international inter-industry data (Galindo  et al., 2002; Gupta  and  Yuan, 2009; Levchenko et al., 2009). These papers identify the effect of the reform by exploiting cross-industry differences in the need for external finance. Chari et al. (2009), show that average wages increase after capital market integration. However, there is no country-industry evidence regarding the effects of financial reform on wage inequality.

This paper in addition relates to the extensive literature on the determinants of rising wage inequality.  Several explanations have been proposed to explain the shift of demand against unskilled workers. In particular: skill biased technical change (Katz  and Murphy, 1992), trade liberalization (Wood,  1995), and changes in labor market institutions (Di- Nardo et al., 1996).

Empirical strategy

According to the model, financial liberalization should increase wage inequality disproportionally in industries with high financial needs and strong CSC. Based on this prediction, my identification strategy for estimating the causal impact of the policy on inequality will consist of exploiting cross industry differences in both financial needs and CSC.

Identification

There at least two potential  threats to identification. First, there could be other policies or reforms that take place at the same time than financial liberalization and could also increase inequality (e.g. trade openness or skilled biased technological change). Secondly, the decision to deregulate could be endogenous and be triggered by a third factor that could also increase inequality (e.g. banking crisis).

My strategy of exploiting heterogeneity across industries should deal with these threats for the following reason. The model provides a very specific prediction regarding the cross- industry effects of financial  liberalization: the subset of industries with high financial needs and strong complementarity should be the  most affected by the  policy. It is very hard to come up with a reasonable alternative story of another shock that delivers the exact same cross-industry prediction. In other words, I don’t disagree that there could be other policies or shocks that could be concurrent to financial liberalization or that could trigger the decision to deregulate. Neither do I disagree that the effects of these other factors could be heterogenous across industries. The identification assumption is that these factors do not increase inequality exclusively in the subset of industries with both high financial needs and strong complementarity.

Capital-skill complementarity

Since different industries have different production functions, capital and skilled labor should have a stronger degree of complementarity in some industries than others. Capital will tend to strongly substitute for unskilled workers in industries where the latter carry out a very limited and well-defined set of cognitive and manual activities, which can be accomplished by following explicit rules (routine tasks). Likewise, capital will tend to strongly complement skilled workers in industries where the latter carry out problem-solving and complex communication activities (non-routine tasks).

In order to construct a CSC index, the author estimates a skilled labor share equation for each industry. Following Berman  et al. (1994), the author assumes that capital is a quasi-fixed factor and  that skilled and  unskilled labor are variable factors. I approximate the variable cost function to a Translog function. As shown in the following equation, cost minimization under constant returns to scale yields the following share equation for each industry:

S = α + γ log(ω) + γ log(k/y),

where S denotes the share of skilled labor payment in the wage bill, i.e.  S = ws s . A positive coefficient for γ in equation implies capital-skill complementarity. Intuitively, when capital and skilled labor are relative complements, an increase in capital intensity leads to an increase in the relative demand for skilled labor, causing the wage bill share of skilled workers to increase. The stronger the complementarity between capital and skilled labor, the larger the increase in the skilled labor share. I therefore use the γ coefficient as a measure of industry-level CSC. I use data from a panel of countries across time and estimate the following equation for each industry:

Sct  = α + β log(ω)ct  + γ log(k/y)ct + ηc + ηt  + εct,

where c indicates the country, t the year, and ηc and ηt  are country and year fixed effects. I estimate this equation separately for each industry and recover the parameter γ from each estimation. I define the capital-skill complementarity index of industry  i as C SCi = γi.

To estimate equation (2), I must take into account that the variation  in log(k/y) might be not be completely exogenous. For example, a skill-biased technological shock can increase both capital intensity and the relative demand for skilled labor,  and hence the wage share of skilled labor. To obtain an exogenous variation of capital intensity, I use lagged values of the dependent and independent variables as internal instruments. I estimate equation (2) in first differences:

∆Sct = β∆ log(ω)ct  + γ∆ log(k/y)ct + ∆ηt  + ∆εct,                     (2.3)

where ∆ denotes the time difference operator,  i.e. ∆x = xt − xt−1 . The identification assumption to estimate the first-differences equation is that the error term in equation (2) is not serially correlated and that the explanatory variables are weakly exogenous (i.e., uncorrelated with future realizations of the error term). In other words, the exclusion restriction states that lagged values of capital intensity affect the wage share of skilled labor only through its effect via current capital intensity. The GMM panel estimator uses the following moment conditions to estimate the complementarity coefficient:  E[zct−j  εct] = 0 for j ≥ 2, t ≥ 3, where z = [S , log(ω), log(k/y)].

Data

Financial liberalization

In the last quarter of the twentieth century, financial markets across the world moved from government ownership or control towards greater private provision of financial services under fewer operational restrictions. Abiad and Mody (2005) document that in many cases financial reforms were triggered by shocks such as a balance-of-payments crises, which destabilized cooperation among different interest groups. Other shocks that precipitated reform were falling global interest rates and participation in IMF programs. The overall trend towards liberalization also reflected pressures generated by the need to catch up with regional reform leaders.

The data on financial liberalization used in this paper comes from Abiad et al. (2010). The  authors create a liberalization index  that runs from 1975 to 2005 and  measures the removal of government control of the financial sector. Recognizing the multifaceted nature of financial liberalization, the index is an aggregation along seven dimensions: credit controls, interest rate controls, bank entry barriers, restrictive regulations, bank privatization, controls on international financial transactions, and securities market policy.

After intersecting the Abiad et al. (2010) reform data with the EU-KLEMS dataset on wage inequality, which will be explained below, I obtain a sample of 20 countries. By the nature of the EU-KLEMS dataset, the majority of the countries are European.

To obtain precise liberalization dates, I set a threshold for the Abiad et al. (2010) index, above which a country is considered liberalized. Following the work of previous studies (Braun  and  Raddatz, 2007), the reform variable is defined to take the value of one when the country’s normalized liberalization index is above the median of the index across all countries (which corresponds to the value of 0.7) and the value of zero when the index is less than  or equal to the median.

The countries that first started liberalizing their financial markets (late 1970s) were Germany and the U.K. Eastern European countries were the last to undertake reform (late 1990s).

For most of the last century, states in the U.S. imposed various restrictions  on the ability of banks to branch within state borders and to operate in other states. Starting from 1970, several states relaxed these restrictions, allowing bank holding companies to consolidate bank subsidiaries into branches and permitting de novo branching statewide. This relaxation came gradually, with the last states  lifting restrictions following the  1994 passage of the Riegle-Neal Interstate Banking  and Branching Efficiency Act.

Kroszner and Strahan (1999) have argued that small banks fought to maintain branching restrictions, since these restrictions protected them from larger and more efficient banking organizations. Technological innovations, such as the invention of the ATM and the reduction in transportation and communication costs, allowed firms to by-pass local banks, reducing the value to the protected banks of geographical restrictions.  These technological innovations interacted with preexisting state-specific differences in the power of local banks to shape the timing of deregulation across states.

I set the date of deregulation as the date in which a state permitted branching via mergers and acquisitions through the holding company structure, which was the first step in the deregulation process.

Wage inequality

Country-level data. The data on wage inequality comes from the EU-KLEMS dataset, a statistical and analytical research project financed by the European Commission. It provides industry-level information for a group of European countries, plus a few non- European ones, on capital, labor by skill level, and labor compensation by skill level. It is a panel spanning the 1970-2005 period. Fourteen countries in the database have information on capital, labor, and compensation by skill level, and thus can be included in the estimations of the skilled labor share equations. The 20 countries listed in the previous subsection have data on labor compensation by skill level and can be included in the estimation of the effect of financial liberalization on inequality. Finally, there are 15 two-digit ISIC Rev. 3 industries providing capital and labor compensation data, for which a skilled labor share equation can be estimated. The data includes both manufacturing and non-manufacturing industries.

EU-KLEMS provides industry information on wages total hours worked by skill level (high, medium, and low). I define skilled labor as the labor force with some tertiary education (high skill level) and unskilled labor as the labor force with less than  tertiary education  (medium  and low skill levels).

State-level data. I use the Merged Outgoing Rotation Groups (MORG) files of the Current Population Surveys (CPS) to obtain wage inequality data.  The CPS is a monthly household  survey  conducted by the Bureau  of Labor Statistics to  measure labor force participation and employment, where 60,000 households per month across the U.S. are queried.

The sample period under study is 1979-2002. I include all wage workers with ages from 25 to 64. I use a consistent variable for years of education and assign workers a consistent CIC industry code using the concordance tables provided by Autor et al. (1998). Hourly wages are defined as reported hourly earnings for those paid by the hour and usual weekly earnings divided by hours worked last week for non-hourly workers. I define skilled workers as those with 13 or more years of completed education and unskilled workers as those with 12 or fewer years of education.  All results are robust to dropping the top 1%, 5%, and 10% wage earners within the high-skill group.  I aggregate the wages of all workers to the industry level by using an earnings weight that is equal to the product of the CPS sampling weight and hours worked in the prior week.

Financial dependence and complementarity indices

External financial dependence: The index is constructed using data from COMPU- STAT. Both capital expenditures and cash flow are summed up over the relevant time period (1975-2005) to compute the firm-level external financial measure. The industry- level index is then defined as the external financial dependence of the median firm for each industry.

Column (1) of table 1 depicts the external financial dependence measure for the 15 industries in the sample. As can be seen, there is substantial cross-industry variation in the index. The chemicals manufacturing industry presents the highest needs for external finance, while the wholesale trade industry presents the lowest financial needs.

Table 1: Financial liberalization and wage inequality in a cross-section of countries

Capital-skill complementary: Column (2) of table 1 reports the CSC index for each industry, together with its 95% confidence interval. CSC is statistically different from zero in all but one industry (hotels and restaurants). Capital and skilled labor are relative complements in all industries except retail trade, which is consistent with the evidence of CSC at the aggregate level (Duffy et al., 2004). All manufacturing industries exhibit CSC, which is consistent with the fact that, on average, low-skilled workers are more easily substituted by capital in manufacturing than in services, since these workers conduct more routine tasks. The industry with strongest CSC is post and telecommunications. This finding is compatible with the fact that telecommunications is an industry highly intensive in skilled labor, where computer capital strongly complements skilled workers in carrying out non-routine tasks.

It is interesting to note that the EFD and CSC indices are statistically uncorrelated. The third column of table 1 shows the product between both indices. Industries with high values of this product (e.g. manufacturing of chemicals and post and telecommunications) should be affected disproportionally by financial liberalization.

Stability of indices: Finally, I calculated the EFD and CSC indices for different time periods and the resulting ranking of industries remain unchanged. The ranking also remain unchanged if I estimate the indices with only pre-reform data. These findings support the assumption that the two industry characteristics are technologically determined and are therefore not affected by the reforms.

Contract enforcement

To further support the results found in the paper, I analyze how the effect of financial liberalization on wage inequality varies according to a country’s contracting institutions. Since financial liberalization increases financial depth particularly in countries with solid contracting institutions (Galindo et al., 2002), the effect on inequality should be increasing in contract enforcement strength.

I use the index created by Djankov et al. (2008) as a measure of country-level debt enforcement. Insolvency practitioners from several countries describe how debt enforcement proceeds against an identical firm about to default on its debt.

I divide the sample into countries with the enforcement index above and below the median of the index across all countries.  I then estimate equation for each of the two sub-samples. The results are reported in columns (1) and (2) of table 2. While the triple interaction term is highly significant for the group of high-enforcement countries (column (1)), it is not significant for the low-enforcement group (column (2)). The coefficients of both sub-samples are statistically different from each other. The evidence is thus consistent with financial liberalization being complementary to contract enforcement institutions.

Table 2: Effect of country-level financial liberalization on wage inequality

Endogeneity

A potential threat to identification would be that other policies concurrent to financial liberalization could be driving the results. To be a real threat, these potential confounding factors would necessarily have to increase inequality exclusively in the subset of industries with both high financial needs and strong complementarity.

SBTC is a shift in the production technology that favors skilled relative to unskilled labor by increasing its relative productivity.  Since SBTC increases the relative demand for skilled labor, it should also increase wage inequality. The effect might be particularly strong in industries with high CSC. Nevertheless, there is no reason to expect that the effect should be disproportionally large in industries heavily dependent on external  finance.  Moreover, the effect of SBTC on inequality is expected to be independent of the level of debt enforcement of an economy.  As a result, the finding that the effect of the reform on inequality is increasing in the level of enforcement is inconsistent with SBTC causing the rise in inequality.

According to the Stolper-Samuelson Theorem, trade opening increases the relative price of a country’s abundant factor.  Given that most countries in my sample are skill-abundant, one concern is that simultaneous changes in trade policy might be increasing the relative wage of skilled labor.  However, there is no evidence that the tariffs were reduced particularly in industries with high needs for external finance and strong complementarity. Furthermore, even though the countries in my sample have made some free-trade agreements in the last three decades, the bulk of tariff and non-tariff reductions took place at least a decade before financial deregulation (Wacziarg and Welch, 2008).

Outsourcing, which means the import of intermediate inputs by domestic firms, can also contribute to the increase in wage inequality. If firms respond to import competition from low-wage countries  by moving unskilled-intensive activities abroad, then trade can lead to an increase in the  relative demand for skilled workers in developed economies. However, the industries that show the highest propensity to outsource are not the ones that exhibit both high financial needs and strong CSC (Feenstra and Hanson, 1996).

I also consider changes in creditor rights.  When a country improves the laws that protect the legal right of investors, savers are more willing to finance firms and thus financial markets flourish.  This could increase wage inequality in industries with high finance needs and strong CSC and present a threat to my identification.  To deal with this issue, I explicitly control for these reforms by using the time-varying creditor rights index developed by Djankov et al. (2007). The results are shown in column (3) of table 3. The triple-interaction term corresponding to creditor rights reforms is not statistically different from zero and the triple-interaction term corresponding to financial liberalization remains unchanged.

Finally, another threat to identification is that a third factor that increases inequality could be triggering the decision to deregulate.  In the case of the state-level reforms, this factor could be a technological shock that improved telecommunications; for the country-level reforms, it could be a balance of payment crisis.  In either case, while these shocks could increase wage inequality, there is no reason to expect that inequality increased exclusively in the subset of industries with both high financial needs and strong complementarity.

Robustness checks

I also conduct a series of robustness checks to analyze the validity of my results. First, instead of using the original industry indices (which are continuous) I use binary indices, which impose less structure for the estimating equation. In particular, I re-define the external financial dependence and CSC index as binary variables that are equal to one if the original index is above the median of the index across all industries and zero otherwise. The treatment group now consists of industries with both indices above their respective median values; the remaining industries make up the control group. Results are presented in column (1) of table 3. The triple interaction remains significant at the 1% level. The reform increases inequality in the treatment group by 4% more than in the control group.

Table 3: Effect of country-level financial liberalization on wage inequality, robustness checks

The results are shown in column (2) of table 3. The triple-interaction term is negative and significant at the 1% level. This means that a financial reform increases inequality particularly in industries with low asset tangibility and strong CSC, as expected.

Next, I use the original Abiad et al. (2010) financial liberalization index, which is a continuous measure of reform, as opposed to the binary measure used in the main part of the paper. As can be seen in column (3) of table 3, the triple-interaction term remains positive, large, and highly significant at the 1% level. In addition, I use an alternative financial reform index, developed by Kaminsky and Schmukler (2008).  The index has been used previously to study the effects of financial liberalization on economic growth (Levchenko et al., 2009). It is the composite of three subcomponents:  liberalization in the stock market, the banking system, and freedom of international transactions. Using this reform index reduces the sample to 12 countries. Results are shown in column (4) of table 3. Again, the triple-interaction term remains positive and significant at the 1% level. The magnitude of the effect actually increases by 40%.

Finally, I analyze whether the results change if I modify the composition of countries in the sample. To study a more homogenous group of countries, I focus only on European countries, excluding Australia, Korea, and Japan from the analysis. The results are shown in column (5) of table 3. The effect remains highly significant and the magnitude does not change.

CHAPTER THREE: FINANCIAL LIBERALIZATION

Introduction

In this chapter, the author conducts a quantitative analysis in order to measure the effect of liberalizing financial markets on aggregate inequality. I simulate a financial reform using a simple two-sector general equilibrium model. The industries in the model are heterogeneous regarding financial needs and capital-skill complementarity (CSC). The economy exhibits frictions in both the capital and the labor market.

According to a back-of-the envelope calculation, the effect on aggregate wage inequality is sizable. For the country-level reforms, I find that financial liberalization increases aggregate inequality for the average country by 1.6%.  For the U.S. state-level reforms, financial deregulation increases aggregate inequality by 1.5%.  As a result, liberalization explains 20% of the increase in aggregate inequality in the U.K. during the 1980-2000 period.  Likewise, bank deregulation explains 15% of the rise in U.S. inequality during the same time period.

The model

Consider an economy that produces two goods (1 and 2) using three productive factors: capital (k), skilled labor (s), and unskilled labor (u).  There are three types of agents in the economy: firms, skilled workers, and unskilled workers.  All workers supply inelastically one unit of labor.  The aggregate supplies of skilled and unskilled labor are S and U . The skilled and unskilled wage rates are denoted by ws and wu. Firms in both industries have the same capital endowment, equal to A.  The economy is small and open and takes the relative price of goods (p1/p2) and the rental rate of capital (r) as given.

Production functions in both industries exhibit constant returns to scale and are strictly quasi-concave:

yi = fi(ki, si, ui)      for   i ∈ {1, 2}

Markets

Capital market.  The capital rental market in the economy is imperfect. There is a financial friction that has an asymmetric effect across industries.  Firms in each industry can borrow (b) only a multiple θ − 1 of their capital endowment at the international rental rate:

bi  ≤ (θi − 1)A       with    θi ≥ 1                                        (3.1)

The  multiple  θi  is separated  into  two  components,  θi  =  θ +  i.  The parameter θ captures the degree of financial repression in the economy.  The parameter i measures the asymmetry of the financial friction across industries.

Financial liberalization

Now, suppose that the government undertakes financial liberalization with the ultimate goal of fostering credit markets. A reform in the model consists of an increase in the parameter θ, which makes the borrowing constraint (3.1) less binding.   The underlying logic is that financial liberalization improves the efficiency of financial intermediation, which improves banks’ ability of screening and monitoring loans.  

Financial liberalization alleviates the borrowing constraint in the economy.  Since the borrowing constraint in industry is more binding than in industry it will benefit relative more from the reform.  As a result, capital demand will increase relatively moreinindustry

The increase in capital demand in both industries is accompanied by an increase in the demand for both skilled and unskilled labor.  Since the production functions exhibits CSC, the relative demand for skilled labor increases, leading to an increase in the relative wage of skilled labor.  Given that borrowing constraints and CSC are stronger in industry 1 than in industry 2, wage inequality will increase relatively more in industry 1.  This will produce an instantaneous outflow of relative skilled labor from industry 2 towards industry 1, in search for the higher returns to skill.  However, given that labor is not perfectly mobile, the movement will be less than that required to equalize relative wages across industries.  As a result, wage inequality will increase by more in industry 1 than in industry 2.

Level of wages

Finally, since the concept of capital-skill complementarity is by definition relative (capital increases the relative demand for skilled labor), the reduced form analysis can only inform about relative wages, not about absolute wages.  Absolute wages can either increase or decrease as the capital stock increases.

An additional  benefit of the quantitative analysis is that  it allows to analyze the effect on absolute  wages, since I have not  imposed any  structure on the  production  function besides CSC and  constant returns  to  scale.  After simulating a financial reform in the model, I find that the absolute wages of both skilled and unskilled workers increase in absolute levels (in terms of the numeraire). For the country-level analysis, the wages of unskilled workers increase by 6.4%. For the state-level analysis, unskilled wages increase by 4.1%. Therefore, according to this analysis, financial liberalization is a policy that  is Pareto  improving for employees. That is, all workers benefit from the reform, but skilled workers benefit relatively more.

Labor market institutions

While an increase in labor market flexibility decreases the differential effect of financial liberalization on wage inequality across industries, it might increase or decrease the effect on aggregate inequality. More flexibility means that more relative labor can flow from industry 2 towards industry 1, which increases wage inequality in industry 2, but also decreases wage inequality in industry 1. Given that the change in aggregate inequality is a weighted average of the change in inequality in both industries, the overall effect could go either way.

Figure 3.1 plots the model-derived relationship between labor flexibility and the change in aggregate inequality. As can be seen, for both the country-level (panel (a)) and the state-level (panel (b)) reforms, more flexibility increases the overall effect, although at a decreasing rate. For the country-level reforms, the effect converges to 3.5% for sufficiently flexible labor markets (ψ > 10). This effect is more than twice the effect calculated with the benchmark calibration (1.6%). On the other hand, the effect for the state-level reforms converges to 2% with sufficient labor flexibility, which is 30% larger than the benchmark effect (1.5%).

Figure 3: Effect of reform on aggregate inequality under different labor market institutions

Conclusions

According to a back-of-the envelope calculation, the effect on aggregate wage inequality is sizable. Liberalization explains 20% of the increase in aggregate inequality in the U.K. during the 1980-2000 period and 15% of the rise in U.S. inequality during the same time period.  In addition, I find that the absolute wages of both skilled and unskilled workers increase in absolute levels.  Therefore, according to this analysis, financial liberalization is a policy that is Pareto improving for employees.

Finally, it is important to note that the time horizon of the model is short to medium- run.  First, I have assumed that the aggregate supply of both types of labor is inelastic. In the long run, agents have incentives to acquire human capital in response to the higher returns to skill.  Secondly, I have assumed that labor is imperfectly mobile across industries.  Again, in the long-run the frictions that prevent labor from moving across industries should disappear.  In future research, it might be interesting to develop a fully dynamic model to trace both the short/medium and long-run of effects of financial reform.

CHAPTER FOUR: THE IMPORTANCE OF FINANCIAL CONSTRAINTS

Introduction

It is well known that cross-country differences in income per capita are very large. Since the work of Klenow and Rodriguez-Clare (1997) and Hall and Jones (1999), there is a growing consensus that total factor productivity (TFP) is the most important factor in accounting for these differences. An emergent literature has developed in order to under- stand why TFP differs across countries.  One particular strand, starting with Banerjee and Duflo (2005), has argued that differences in the allocation of resources across heterogeneous agents may be a significant factor in accounting for cross-country differences in TFP.

One potential source of misallocation relies on financial frictions. When an economy has an underdeveloped financial market, productive but poor firms lack the collateral required for taking out a loan.  As a result, these firms may produce at a sub-optimal scale. Reallocating capital from rich and low-productivity firms towards poor and high- productivity ones would increase the economy’s output. Failure to reallocate is referred to as resource misallocation. Such a misallocation shows up in aggregate data as low TFP.

This chapter studies the relationship between financing frictions and misallocation by focusing on the episode of financial liberalization by a group of ten Eastern European countries. Starting in the mid 1990s, these countries drastically reduced the intervention of the government in financial activities.  As a result, financial frictions were alleviated and financial depth increased. The main goal of this chapter is to use microeconomic data to analyze whether these financial reforms led to higher TFP by allowing a better allocation of capital across firms. The main contribution of our essay is to provide the first reduced form assessment, using a large firm-level dataset, on the effects of financial frictions on misallocation and TFP.

Looking at a cross-section of the ten transition economies under study in the year 2000, I observe a positive association between financial liberalization and aggregate productivity (see figure 4).  However, since these countries were transitioning from a command to a market economy, other events that took place during the same period might be driving this relationship.

Figure 4: Financial liberalization and aggregate productivity

To identify the causal impact of financial deregulation on productivity, we make use of the fact that this policy should have a stronger effect on those industries that are more financially constrained.  In particular, the reform should affect particularly those industries that have high requirements of external finance (Rajan and Zingales, 1998) and high levels of asset tangibility (Braun, 2003; Claessens and Laeven, 2003).

Exploiting differences in financial constraints across industries allows us to disentangle the effects of financial liberalization from other policies that could have taken place at the same time and that affect industries uniformly.  To allow for a differential effect of other reforms across industries, we explicitly control for a large array of reform indicators constructed by Campos and Horvath (2009).

Another threat to identification is that financial liberalization is itself a political outcome, which might be endogenous to a specific pattern of industry-level TFP growth. It is argued that this possibility is highly unlikely, since financial deregulation was largely induced by external pressures from outside governing bodies such as the European Union (EU), IMF, and OECD. Most of the sample countries were seeking EU membership, and accession imposed strict guidelines regarding financial repression. Also, many of the countries were asking for financial support from the IMF and expressed their commitment to undertake financial sector reforms in order to obtain such help. In addition, we document that financially constrained industries have relatively low political strength, which makes the possibility of reverse causality even more implausible.

In this chapter the author estimates firm-level productivity as the residual from an estimated production function. I then calculate industry TFP as the weighted average of productivity of all firms producing in that industry.

The differential effect across industries is sizable. After a large financial reform, TFP in industries with high financial dependence increases by 16% more than in industries with low dependence.  Likewise, TFP in industries with low asset tangibility increases by 29% more than in industries with high tangibility. However, these industry TFP gains could be the result of an improved allocation of capital across firms or of firms becoming individually more productive.

To identify the source driving the TFP gains, the author uses a standard industry decomposition method (Olley and Pakes, 1996).  I express industry TFP as the sum of two components: an average-productivity term and an allocation term. The first component measures the individual efficiency of firms of the industry. The second component is the within-industry cross sectional size-productivity covariance.  It measures the efficiency with which resources are allocated across firms within a sector.

Finally, the author analyzes the effect of financial liberalization across firms that ex-ante face different levels of financial constraints.  In particular, I compare domestically -owned firms with foreign -owned firms.  Since foreign firms have access to an internal capital market, they should be less constrained than domestic firms.  The effect is particularly strong in highly financially constrained industries. In high external dependent industries, the market share of domestic firms increases by 7% more than the foreign firms’ share. Likewise, in industries with low asset tangibility, domestic firms see their market share increase by 9% more than the share of foreign firms.

One particular strand of this literature focuses on financial frictions as the underlying source of misallocation (Banerjee and Moll, 2010; Midrigan and Xu, 2010; Buera et al.,2011; Moll, 2012). All these papers conduct structural estimations. They develop general equilibrium models with imperfect financial markets, calibrate the structural parameters of the model using micro-level data, and quantify the magnitude of output loss due to misallocation.  In contrast, this paper conducts a reduced form estimation of a concrete episode of a financial reform to measure empirically the effect of this policy on TFP and misallocation. A benefit of our approach is that our results do not depend on particular assumptions or parameter calibrations.  A drawback is that in order to achieve identification we can only calculate the differential effect of the policy across industries, not the aggregate effect.

The model

Consider an economy that produces two goods (1 and 2) using two productive factors: capital (k) and labor (l). There are two types of agents in the economy: firms and workers. Firms are heterogeneous regarding their productivity (z ≥ 0) and their wealth (a ≥ 0). The joint distribution of productivity and wealth is denoted by F (z, a). The mass of firms is normalized to one in each industry and entrepreneurs are not mobile across sectors. All workers supply inelastically one unit of labor. The aggregate supply of labor is L and the wage rate is denoted by w. The economy is small and open and takes the relative price of goods and the rental rate of capital (r) as given. For simplicity, we normalize the prices of both goods to one.

The production function of firms in each industry is:

y = z(kα l1−α )ν ,

where α ∈ (0, 1) and v ∈ (0, 1) is a span of control parameter that induces decreasing returns to scale.

Factor markets

The capital rental market in the economy is imperfect. There is a financial friction that has an asymmetric effect across industries. Firms in each industry can borrow (b) only a multiple θs − 1 of their wealth at the international rental rate:

bs ≤ (θs − 1)a    for  s ∈ {1, 2}

The multiple θs ≥ 1 is separated into two components, θs = θ + s . The parameter θ captures the degree of financial repression in the economy and the parameter s measures the asymmetry of the financial friction across industries. One interpretation of constraint is that the fraction of wealth that is tangible, and hence collateralizable, varies across industries. This formulation of the capital market imperfection is analytically convenient.  The parameter θ captures the degree of financial repression of the economy. By varying it, we can trace out all degrees of capital market efficiency. θ → ∞ corresponds to a perfect capital market while θ = 1 − s means that the capital market is completely shut down. The labor market of the economy is perfectly competitive, so firms equate the marginal product of labor with the wage rate.

Financial liberalization

Now, suppose that the government undertakes financial liberalization with the ultimate goal of fostering credit markets.  A reform in the model consists of an increase in the parameter θ, which makes the financing constraint less binding.  The underlying logic is that financial liberalization alleviates asymmetric information problems in the capital market by improving the screening and monitoring ability of banks. The effects of financial liberalization are the following.

Proposition 1. Total factor productivity and the size-productivity covariance increase.

As explained above, an increase in the size-productivity covariance leads to an increase in TFP. To understand the intuition behind the increase in the covariance term, we can further decompose it as follows:

Cov[log(ω(a, z)), log(z)] =   z  − αν Cov[log(fk(a, z)), log(z)]

where σ2 ≡ Var(log(z)). Intuitively, with perfect financial markets, fk (a, z) = r so the marginal product of capital is equalized across firms. As a result, Cov[log(fk (a, z)), log(z)] =0, so the size-productivity covariance is maximized and equals  σz .3  With imperfect capital markets, however, the marginal product of capital varies across firms. In particular, given a level of wealth, a more productive firm will be more constrained and face a higher shadow cost of capital.  This positive covariance lowers the size-productivity covariance, as the second term in equation becomes positive. Financial liberalization alleviates financial constraints and therefore reduces the covariance between productivity and the marginal product of capital.  This increases the size-productivity covariance and hence industry-level TFP.

This decomposition therefore delivers another testable implication. If the increase in TFP is driven by an alleviation of financial frictions, then we should observe a reduction in the within-industry covariance of the marginal product of capital and productivity after reform. In contrast, given that firms equalize the marginal product of labor to the wage rate, we should observe no changes in the covariance between the marginal product of labor and productivity.

Proposition 2. The increase in the size-productivity covariance is driven by a reduction in the covariance between the marginal product of capital and productivity.

Another intuitive measure for misallocation arising from frictions in the capital market is the within-industry variance of the marginal product of capital. For instance, Hsieh and Klenow (2009) consider a model of monopolistic competition with idiosyncratic wedges. Using lognormality assumptions for productivity and wedges, they show that TFP can be expressed as a function of the variance of the marginal product of capital. Motivated by this expression, we also test the hypothesis that this variance should decline after a financial reform. For the same argument given above, we should observe no changes in the variance of the marginal product of labor.

Comparing industries with different levels of financial constraints, a reform will have the following effects.

Proposition 3. The increase in total factor productivity and size-productivity covariance is larger in industry 1 than in industry 2.

Given that the financing constraint is more binding in industry 1 than in industry 2, in- dustry 1 will benefit disproportionately from the reform. As a result, the size-productivity covariance and therefore TFP will increase by more than in industry 2.

Proposition 4. Wages increase.  If labor is not perfectly mobile across sectors, the in- crease is larger in industry 1 than in industry 2.

Holding factor prices constant, only previously constrained firms will increase their factor demand after reform. This will lead to an increase in the demand for labor, and, with a fixed supply of labor, an increase in the wage rate.  If there is some degree of segmentation of labor markets, the wage rate will increase more in industry 1, since this industry benefits relatively more by the reform.

As explained in the previous proposition, the wage rate increases after financial liberalization.  This general equilibrium effect will force previously unconstrained firms to lower their factor holdings, and they will lose in market share. Again, the effect will be stronger in industry 1, which is financially more constrained.

Financial reforms and data

Starting in the beginning of the 1990s, Eastern European countries undertook dramatic reforms in their transition from a centrally planned economy to a free market economy. Financial liberalization reforms were a key component of the second phase of transition, which was designed to be market deepening. Financial deregulation was largely induced by external pressures from outside governing bodies such as the EU, the IMF, and the OECD. The pressure from the EU and the OECD was a result of the countries’ prospective accession to these institutions. The pressure from the IMF derived from the request for financial support.

The countries chosen for our study arise from the intersection of countries that experienced significant financial reforms since the 1990s and countries whose firms are well represented in our firm-level dataset. Since most Western European countries experienced financial reforms in the 1970s and 80s, and coverage is poor for some Eastern European countries, this intersection restricts our study to 10 transition economies: Bulgaria, Czech Republic, Estonia, Hungary, Latvia, Lithuania, Poland, Romania, Russia, and Ukraine.

Part of the same wave of enlargement was the accession of Bulgaria and Romania in 2007, which were unable to join in 2004 but constitute part of the same enlargement. To join the EU, a state needs to fulfill economic and political conditions summarized in the Copenhagen criteria. The economic criteria, broadly speaking, require that candidate countries have a functioning market economy. This required a substantial reduction of the government intervention in the financial sector. These criteria had a substantial influence on policies in Eastern European countries.  Schimmelfennig and Sedelmaier (2004) (p.671) note that “The credibility that the EU will reward rule adoption with membership, which increased significantly once accession negotiations started, emerges as the most important factor influencing the cost-benefit calculations of CEEC governments.  The massive benefits of EU membership being within close reach, the fulfillment of EU acquis conditions became the highest priority in CEEC policy-making, crowding out alternative pathways and domestic obstacles.”

For example, in 1997 the European Commission, in its report on the progress of the Czech Republic in the accession process, recommended that bank privatization and improvement of the regulatory framework and standards of governance would bring the country closer towards fulfillment of the Copenhagen criteria. It then consistently pushed for bank privatization in successive annual assessments (Vliegenthart and Horn, 2007).

In their request for financial support from the IMF, many countries expressed their commitment to undertake financial sector reforms. For example, the letter of intent (LOI) of the government of Bulgaria in 1998 stated that “Our priorities in the structural areas are to complete privatization of the state banks and enterprises and to develop and deepen financial markets.”  The LOI of Ukraine in 1998 stated that “Financial sector reform and an acceleration of privatization will also be important elements of our medium-term program.”  According to Romania’s 1999 LOI, “In the key area of banking supervision, we envisage measures to strengthen the National Bank of Romania’s supervisory and enforcement capacities and to ensure banks’ compliance with prudential regulations.”

Finally, three of the countries under study (Czech Republic, Hungary, and Poland) are members of the OECD. The OECD also imposed pressure to its prospective members. For example, following the division of the country into two republics in 1993, the Czech Republic authorities began to map the steps toward capital account liberalization. According to a number of former officials interviewed, the process of eliminating capital controls was largely driven by the country’s prospective accession to the OECD (IMF, 2005).

Reform data

The data on financial liberalization used in this paper comes from Abiad et al. (2010). The authors create a liberalization index, which runs from 1975 to 2005 and measures the removal of government control of the financial sector. Recognizing the multifaceted nature of financial liberalization, the index is an aggregation along seven dimensions: (1) directed credit, (2) interest rate controls, (3) entry barriers, (4) restrictive operational regulations, (5) privatization in the financial sector, (6) controls on international financial transactions, and (7) securities market policy.

Along each dimension, a country is given a final score on a graded scale from 0 to 3, with 0 corresponding to the highest degree of repression and 3 indicating full liberalization.

Among all countries, Ukraine presents the highest level of financial repression during this period.  In 2005, Ukraine’s index of financial liberalization was less than 70% of the maximum achievable level.  Estonia, on the other hand, is the country with the least government intervention in the financial sector. Hungary is the economy that most rapidly deregulated its financial markets during these years.  Its financial liberalization index almost doubled between 1994 and 2005.

There are several papers documenting that financial liberalization alleviates financial constraints faced by firms (Laeven, 2003) and leads to deeper financial markets (Tressel and Detragiache, 2008). If we look at our cross-section of countries (figure 4.2), we can observe that countries with more deregulated financial markets do in fact exhibit higher financial depth.

Table 4: Coverage of firms in AMADEUS dataset

AMADEUS data

The firm-level data comes from AMADEUS. AMADEUS is a commercial database pro- vided by Bureau van Dijk (BvD). It contains financial information on over five million public and privately held firms across many Western and Eastern European countries. BvD collects the data from local information providers, which in most cases are the local company registers. The database comes in yearly versions and each vintage includes up to ten years of information per firm.

Table 4 reports the coverage of firms for the ten countries in our sample.  The differences in the number of firms across countries can mainly be attributed to different filing requirements for companies. In most cases, these filing requirements are related to size criteria, or to the mode of incorporation.  The large number of Romanian firms is particularly striking in the table. This can be attributed to the exceptional coverage of 6 Our sample spans between 1994 and 2005 since our firm-level data starts from 1994 and the financial liberalization data ends in 2005.

Our sample period is also characterized by a significant increase in the number of observed firms over time. According to BvD representatives, the inclusion of small-and medium-sized enterprises has contributed significantly to this increase. Moreover, BvD has made an effort to source additional data, working together with country authorities. In our empirical analysis, our coefficient of interest will be identified by within-country variation across industries.  A potential concern might be that this increase in coverage was biased across industries.  Although this would not directly affect our main variables of interest, such as the size-productivity covariance, it would directly affect sector-level totals (e.g. total output, total number of employees). We therefore check whether this increase in the number of firms is biased across industries with different degrees of financial constraints (external financial dependence and asset tangibility). We do not find evidence of any such bias.

Due to data restrictions, the author does not attempt to provide a detailed analysis of the intensive versus extensive margin in the process of reallocation. Depending on the filing requirements, we are unable to capture entry or exit if entrants are either too small to meet the filing requirements or if they start their business in a mode of incorporation that excludes them from the requirement to file accounts.  Similarly, the author cannot distinguish between firms that exited the market and firms that fell below the size restrictions for filing or changed their mode of incorporation.

Another threat for the representativeness of our data is the well-known survivorship bias that is inherent in the construction of the AMADEUS data.  If a firm has stopped filing, it is kept in the database for four subsequent years and is then deleted. This biases the coverage towards surviving firms and has limited the time coverage of several previous studies using AMADEUS. For our study, it is essential to follow firms within a country for consecutive years. We overcome this bias by appending two versions (2006 and 2002) of the database. Firms that exited prior to 2002 and got deleted in the 2006 version of the database are present in the 2002 vintage and will therefore be included in our appended dataset.

Empirical strategy

To identify the causal effect of financial liberalization on TFP  and misallocation, we exploit variation in financial constraints across industries. For technological reasons, some industries are more constrained than others. First, some industries inherently need more external financing than others (Rajan and Zingales, 1998). As an industry’s dependence on external finance increases, the availability of outside capital becomes more important. Second, some industries operate naturally with a higher proportion of tangible assets than others (Braun, 2003; Claessens and Laeven, 2003). As an industry’s asset tangibility decreases, the amount of collateral that can be pledged for loans becomes scarcer.  As a result, financial deregulation should increase TFP particularly in industries with high needs of external finance and low levels of asset tangibility.

If there are other reforms taking place at the same time than financial liberalization that affect all industries equally, their effect will be cancelled by the cross-industry comparison. However, there might be other reforms that affect industries differentially depending on their degree of financial constraints. While identification does not require that industries have exactly the same level of financial constraints in every country, it does rely on the ranking of sectors remaining relatively stable across countries.  Given that we are working with a fairly homogenous group of transition economies, we consider this a reasonable assumption.

A final concern is that a country might have deregulated its financial market in order to accommodate the needs of a high-productivity growth industry.  If this industry is highly financially constrained, this would impose a threat to identification through re- verse causality.

The first measure of industry-level financial constraints we use is external financial depen dence (Rajan and Zingales, 1998). The index of dependence is measured as the median ratio across firms belonging to the corresponding U.S. industry of capital expenditures minus cash flow from operations to capital expenditures.  Second, we use asset tangibility (Braun, 2003; Claessens and Laeven, 2003). The tangibility index is measured as the median ratio across firms of the value of net property, plant, and equipment to total assets.

As can be seen from the table, electrical machinery is an example of an industry with very high financial needs, while one of the industries with lowest needs is food products. Likewise, basic metals exhibits one of the highest degree of asset tangibility, while medical instruments is an example of a very low tangibility industry. The two sectoral indices are negatively correlated.

Productivity decomposition

We start by measuring firm-level TFP. For each country, we compute firm-level productivity residually from the production function:

log(z)ist = log(y)ist − αk log(k)ist − αl log(l)ist,

where i denotes a firm, s a sector, and t a year. y corresponds to revenues, k to fixed assets, and l to number of employees.

In principal, we could estimate directly the parameters of equation.  The two threats for identification would be endogeneity of some of the inputs (which depend on productivity) and sample selection (exit of inefficient firms). Olley and Pakes (1996) and Levinsohn and Petrin (2003) propose methods to deal with these two threats. However, we are unable to replicate these methods since we lack data on intermediate inputs and our data has insufficient details on firm exit.

Instead of estimating the parameters, we set input elasticities equal to factor shares. We impose constant returns to scale and measure the labor elasticity for each sector as the average labor share across all of countries and years, i.e. αl = (wl/y)s . The data on labor shares comes from the UNIDO dataset. In section 4.7 we show that our results are robust to using labor productivity as an alternative measure of firm-level efficiency.

Next, we define industry-level productivity as a weighted average of firm-level productivities:

nlog(Z )st = X ωist log(z)ist,

where ωist is the share of revenues of firm i in total revenues of sector s.  We use the method of Olley and Pakes (1996) to decompose industry-level productivity into two components:

Results

We start by analyzing the effect of financial liberalization on industry output and its dif- ferent components. For this, we estimate the following generalized difference-in-differences specification:

log(Q)cst = αRefct • F ins + βxct • F ins + ηct + ηcs + cst   for  Q ∈ {Y, K, L, Z },

where Qcst  denotes either output, capital, employment or TFP of country c in sector s in year t. Refct represents the financial reform index for each country in each moment of time.  F ins  measures sector s’s level of financial constraints, which in alternative regressions we proxy with external financial dependence and asset tangibility. xct  is a vector including internal, external, and privatization reform indicators. The specification includes country×year and country×industry fixed effects. The standard errors of this and all regressions are clustered at the country level.

The coefficient of interest is α, which measures the differential effect of financial liberalization across industries with different levels of financial constraints. The coefficient is identified purely from the cross-industry variation within a country over time.

Panel A of the table reports the results using external financial needs as the measure of financial constraints, while panel B uses asset tangibility. From the table, we can observe that financial liberalization increases output disproportionately in industries with high requirements for external finance and low tangibility. The effect is statistically significant at the 5% level.

To interpret the magnitude of the effect, consider a country undertaking a large re- form, which according to Abiad et al. (2010) corresponds to an increase of the financial liberalization index by three units. The differential effect of the reform on two industries with different levels of financial constraints is α • 3 • (F inh − F inl ). The point estimate implies that liberalizing financial markets increases output in the 75th-percentile industry by financial dependence by 16% more than in the 25th-percentile industry.  Likewise, it increases output in the 25th-percentile industry by asset tangibility by 24% more than in the 75th-percentile industry.

Turning to the determinants of industry output, we find no significant effect on either capital or labor, but the effect on TFP is highly significant at the 1% level. The magnitude of the effect is also large: the differential effect across industries with different levels of financial dependence is 16% and across industries with different asset tangibility is 21%. This finding is consistent with previous evidence documenting that financial liberalization increases economic growth primarily through higher TFP (Levine, 2005; Levchenko et al., 2009; Bekaert et al., 2011).

 

CHAPTER FIVE CONCLUSIONS

If an economy exhibits perfect financial markets, productivity is the only determinant of the size of firms. If, on the other hand, the financial market is underdeveloped, wealth also plays an important role for determining a firm’s size. As a result, resources are not allocated towards their most efficient use, resulting in low aggregate productivity.

An improvement in the functioning of the financial market should weaken the link between entrepreneurial wealth and firm size. Resources should become allocated more efficiently and aggregate TFP should increase. In this essay, we focus on a specific reform which reduces financial frictions, financial liberalization.  We use a large cross-country firm-level database to analyze whether the reform increases TFP and whether the efficiency with which resources are allocated improves. To identify the causal effect of the reform on productivity and misallocation, we exploit differences in financial constraints across industries.

We find that financial liberalization indeed increases TFP particularly in financially constrained industries, i.e. sectors with high external financial needs and low asset tangibility. To understand the factors that are driving the productivity gains, we decompose industry productivity into an average-productivity term (measuring firm efficiency) and an allocation term (measuring resource allocation quality).  We find that a better allocation of resources is the main driver of the TFP gains. This result is confirmed by our finding that the reform reduces the covariance between TFP and marginal product of capital, and reduces the variance of the marginal product of capital. The finding that the market share of ex-ante financially constrained firms (domestically-owned firms) increases after financial deregulation further supports our results.

Our findings contribute to the discussion about the importance of financial frictions for aggregate TFP losses. While Buera et al. (2011) document that misallocation arising from financial frictions leads to large productivity losses, Midrigan and Xu (2010) document only a minor effect. However, both papers study the effect of financial frictions across steady states, where firms are able to accumulate internal funds and grow out of their financing constraints. In our essay, we focus on the transition to the steady state. We show that during this transition, financial frictions are an important source of misallocation and a reform that alleviates these frictions leads to large TFP gains. As such, our results show that the liberalization of financial markets can contribute to closing the gap of cross-country differences in income per worker.

The development of financial markets can affect both economic growth and income inequality. While economists have studied thoroughly the effects of financial sector policies on growth, the potentially enormous impact of such policies on inequality has been under appreciated. As documented by Demirguc-Kunt and Levine (2009), the three volumes of the Handbook of Income Distribution do not mention any possible connections between inequality and formal financial sector policies. In this essay, I argue that the deregulation of financial markets has contributed to the rise in wage inequality observed in the last three decades in several developed countries.

The author focused on a particular mechanism through which improvements in financial markets can affect wage inequality. According to theory, financial liberalization reduces borrowing constraints and increases capital demand.  If capital and skilled labor are relative complements, the relative demand for skilled labor should increase, enlarging the wage gap between skilled and unskilled workers in equilibrium. The higher the extent of financial needs and the higher the degree of CSC, the stronger the effect on inequality. The effect should be particularly strong in industries with both high needs for external finance and strong complementarity. I rank industries in these two dimensions by using a standard measure of financial needs and constructing a novel measure of the industry-level degree of CSC. I then analyze the differential effect of liberalization on inequality across industries.

The author focused on two distinct episodes of financial liberalization: country-level financial deregulation across a large group of countries and state-level bank deregulation across states in the U.S. I find that, in both episodes, liberalization led to a disproportional increase of wage inequality in industries with high financial needs and strong CSC. I also find that while the differential effect on relative wages is increasing in labor market rigidity, the differential effect on relative labor flows is increasing in labor flexibility.

References

Abiad,  A. and Mody, A. (2005). Financial  reform:  What  shakes it?  What  shapes it?  American Economic  Review, 95(1):66–88.

Abiad,  A., Detragiache, E., and  Tressel,  T. (2010).  A new database of financial reforms.  IMF Staff Papers, 57(2):281–302.

Abiad,  A., Oomes,  N., and  Ueda,  K. (2008).  The  quality  effect:  Does financial  liberalization improve the allocation  of capital?   Journal of Development Economics,  87(2):270–282.

Arellano,  M. and  Bond,  S. (1991).   Some tests  of specification  for panel  data:   Monte  Carlo evidence and an application to employment equations.  Review of Economic Studies, 58(2):277–297.

Autor,  D. H., Katz,  L. F., and Krueger,  A. B. (1998).  Computing inequality:   Have computers changed  the labor market?   Quarterly Journal of Economics,  113(4):1169–1213.

Banerjee,  A. V. and  Moll, B. (2010).   Why  does misallocation  persist?    American  Economic Journal:  Macroeconomics, 2(1):189–206.

Banerjee, A. V. and Duflo, E. (2005). Growth theory through  the lens of development economics. In Aghion, P. and Durlauf,  S., editors,  Handbook of Economic  Growth, volume 1 of Handbook of Economic  Growth, chapter 7, pages 473–552. Elsevier.

Bartelsman, E. J.,  Haltiwanger, J.  C.,  and  Scarpetta, S. (2012).  Cross-country differences in productivity: The role of allocation  and selection.  American  Economic  Review, Forthcoming.

Beck, T.  and  Demirguc-Kunt, A. (2009).   Financial  institutions and  markets  across  countries and over time:  Data  and analysis.  World Bank Policy Research  Working Paper,  No. 4943.

Beck, T., Levine, R., and Levkov, A. (2010). Big bad banks?  The winners and losers from bank deregulation in the United  States.  Journal of Finance, 65(5):1637–1667.

Bekaert,  G.,  Harvey,  C.  R.,  and  Lundblad, C.  (2011).   Financial  openness  and  productivity.

Benmelech,  E., Bergman,  N. K., and  Seru, A. (2011).  Financing  labor.  NBER  working paper,

Berman,  E., Bound, J., and Griliches, Z. (1994). Changes in the demand  for skilled labor within U.S. manufacturing:  Evidence  from the  annual  survey  of manufacturers. Quarterly Journal of Economics,  109(2):367–397.

Bernard, A. B., Redding,  S. J., and Schott,  P. K. (2011).  Multiproduct firms and trade  liberal- ization.  Quarterly Journal of Economics,  126(3):1271–1318.

Bertrand, M., Duflo, E., and  Mullainathan, S. (2004).  How much  should  we trust differences- in-differences estimates?  Quarterly Journal of Economics,  119(1):249–275.

Bhagwhati J. 1998. The capital myth: The difference between trade in widgets and dollars. Foreign Affairs, 77: 7-12.

Blanchard O, Giavazzi F. 2002. Current account deficits in the euro area: The end of Feldstein-Horioka Puzzle? Brookings Papers on Economic Activity, 2: 147-209.

Borenzstein E, Gregoria ED, Lee JW. 2001. How does foreign direct investment affect growth? Journal of International Economics, 16: 115-135.

Bosworth B, Collins S. 1999. Capital flows to developing economies: Implications for saving and investment. Brookings Papers on Economic Activity, 1: 143-180.

Botero,  J.  C.,  Djankov,  S., La-Porta, R.,  Lopez-De-Silanes,  F.,  and  Shleifer,  A. (2004).   The regulation  of labor.  Quarterly Journal of Economics,  119(4):1339–1382.

Boyce JK. 1992. The revolving door? External debt and capital flight: A Philippine case study. World Development, 20 (3): 335-349.

Braun,  M. (2003).  Financial  contractibility and asset hardness.  Mimeo, Harvard  University. Braun,  M. and Raddatz, C. (2007).  Trade  liberalization, capital  account liberalization and the real effects of financial development. Journal of International Money and Finance, 26(5):730–761.

Buera,  F. J., Kaboski, J., and Shin, Y. (2011). Finance  and development: A tale of two sectors. American  Economic  Review, Forthcoming.

Campos,  N. F. and Horvath, R. (2009).  Reform redux:  Measurement, determinants and reversals.  Working Paper, Czech National  Bank, 2009-6.

Casas, F. (1984). Imperfect  factor mobility:  A generalization and synthesis  of two-sector models of international trade.  Canadian Journal of Economics,  17(4):747–761.

Caselli,  F.  and  Coleman,  W.  J.  (2006).   The  world  technology  frontier.   American  Economic Review, 96(3):499–522.

Cerra V, Rishi M, Saxena SC. 2005. Robbing the riches: Capital flight, institutions, and instability. Working Paper 199, IMF.

Cetorelli,  N. and  Strahan, P.  E. (2006).  Finance  as a barrier  to entry:   Bank  competition and industry structure in local U.S. markets.  Journal of Finance, 61(1):437–461.

Chari,  A., Henry,  P. B., and Sasson, D. (2009).  Capital  market  integration and wages.  NBER working paper,  15204.

Claessens, S. and Laeven, L. (2003). Financial development, property rights, and growth. Journal of Finance, 58(6):2401–2436.

Correa,  R. (2008).  Bank integration and financial constraints: Evidence  from U.S. firms.  FRB International Finance Discussion  Paper,  925.

Davis,  S. J.,  Haltiwanger, J.,  Katz,  L. F.,  and  Topel,  R.  (1991).   Wage  dispersion  between and  within  U.S.  manufacturing plants,   1963-86.   Brookings  Papers   on  Economic  Activity, 1991:115–200.

Demirguc-Kunt, A. and Levine, R. (2009). Finance and inequality:  Theory and evidence. Annual Review of Financial Economics,  1(1):287–318.

Desai, M. A., Foley, F., and Hines, J. (2004).  A multinational perspective  on capital  structure choice and internal  capital  markets.  Journal of Finance, 59(6):2451–2487.

DiNardo,  J.,  Fortin, N. M., and Lemieux, T. (1996).  Labor  market  institutions and the distribution  of wages, 1973-1992: A semiparametric approach.  Econometrica, 64(5):1001–44.

Djankov,  S., McLiesh, C., and  Shleifer, A. (2007).  Private credit  in 129 countries.   Journal of Financial Economics,  84(2):299–329.

Djankov, S., Hart,  O., McLiesh, C., and Shleifer, A. (2008). Debt enforcement around  the world. Journal of Political  Economy,  116(6):1105–1149.

Duffy,  J.,  Papageorgiou, C.,  and  Perez-Sebastian, F.  (2004).   Capital-skill  complementarity? Evidence from a panel of countries.  Review of Economics  and Statistics, 86(1):327–344. EBRD  (2007).  Transition report.  European  Bank for Reconstruction and Development.

Feenstra, R.  C.  and  Hanson,  G.  H. (1996).   Globalization, outsourcing,   and  wage inequality. American  Economic  Review, 86(2):240–45.

Fischer S. 1998. Capital account liberalization and the role of IMF. In S Fischer, R Cooper, R Dornbusch, P Garber, C Massad, J Polak, D Rodrik, SS Tarapone (eds.), Should the IMF pursue capital account convertibility?, 1-10, Princeton University, Princeton, NJ.

Fischer S. 2003. Globalization and its challanges. American Economic Review, 93: 1-30.

Forteza, A. and  Rama,  M. (2006).  Labor  market  rigidity  and  the  success of economic reforms across more than  100 countries.  Journal of Policy Reform,  9(1):1384–1289.

Foster,  L., Haltiwanger, J., and Syverson, C. (2008). Reallocation, firm turnover, and efficiency: Selection on productivity or profitability?  American  Economic  Review, 98(1):394–425.

Galindo,  A., Micco, A., Ordoez,  G., and  Bris, A. (2002).  Financial  liberalization:  Does it pay to join the party?  Economia,  3(1):231–261.

Galor,  O. and Zeira, J. (1993).  Income distribution and macroeconomics.  Review of Economic Studies,  60(1):35–52.

Gorodnichenko, Y. and Schnitzer,  M. (2012).  Financial  constraints and innovation: Why poor countries  don’t catch  up.  Journal of European Economic  Association, Forthcoming.

Griliches,   Z.  (1969).    Capital-skill   complementarity.    Review  of  Economics   and  Statistics, 51(4):465–468.

Guner,  N.,  Ventura, G.,  and  Yi,  X.  (2008).   Macroeconomic  implications   of size-dependent policies. Review of Economic  Dynamics,  11(4):721–744.

Gupta, N. and Yuan, K. (2009). On the growth effect of stock market  liberalizations. Review of Financial Studies,  22(11):4715–4752.

Hall, R. E. and Jones,  C. I. (1999).  Why do some countries  produce  so much more output per worker than  others?   Quarterly Journal of Economics,  114(1):83–116.

Hamermesh,  D. S. (1993).  Labor Demand.  Princeton University  Press.

Haselmann,  R., Pistor,  K., and  Vig, V. (2010).  How law affects lending.  Review of Financial Studies,  23(2):549–580.

Haskel, J. (1998). The skilled wage premium and the growth of small firms in UK manufacturing. Labour,  12(2):221–238.

Hsieh, C. and Klenow, P. J. (2009).  Misallocation  and manufacturing TFP in China  and India. Quarterly Journal of Economics,  124(4):1403–1448.

IMF (2005). The IMF’s Approach  to Capital  Account  Liberalization. Evaluation Report,  Wash- ington,  D.C.:  International Monetary  Fund.

Jerzmanowski, M. and Nabar,  M. (2011).  Financial  development and wage inequality:   Theory and evidence.  Economic  Inquiry,  Forthcoming.

Jones,  C.  and  Romer,  P.  (2010).   The  new kaldor  facts:   Ideas,  institutions, population, and human  capital.  American  Economic  Journal:  Macroeconomics, 2(1):224–245.

Kadochnikov D. 2005. Economic impact of capital flight from Russia and its institutional context: Why capital controls cannot be a part of a pro-growth policy. International Finance 0509007, EconWPA, URL http://ideas.repec.Org/p/wpa/wuwpif/0509007.html

Kaminsky,  G. L. and Schmukler,  S. (2008).  Short-run pain,  long-run gain:  Financial  liberaliza- tion and stock market  cycles. Review of Finance, 12(2):253–292.

Katz,  L. F.  and  Murphy,  K.  M. (1992).   Changes  in relative  wages,  1963-1987:  Supply  and demand  factors.  Quarterly Journal of Economics,  107(1):35–78.

Klenow,  P.  J.  and  Rodriguez-Clare, A. (1997).  The  neoclassical  revival  in growth  economics: Has it gone too far?  NBER  Macroeconomics Annual,  12:73–103.

Kroszner, R. S. and Strahan, P. E. (1999). What  drives deregulation? Economics and politics of the relaxation of bank  branching  restrictions. Quarterly Journal of Economics,  114(4):1437–1467.

Krusell,  P.,  Ohanian, L. E., Rios-Rull,  J.-V.,  and  Violante,  G. L. (2000).  Capital-skill  comple- mentarity and inequality:  A macroeconomic  analysis.  Econometrica,  68(5):1029–1053.

Laeven,  L. (2003).  Does financial  liberalization reduce  financing constraints?  Financial Man- agement,  32(1):5–34.

Lensink R, Hermes N, Murinde V. 1998. The effect of financial liberalization on capital flight in African economies. World Development, 26: 1349-1368.

Lensink R, Hermes N, Murinde V. 1998. The effect of financial liberalization on capital flight in African economies. World Development, 26: 1349-1368.

Lessard DR, JWilliamson. 1987. Capital flight and third world debt. Institute for International Economics, Washington, DC.

Levchenko, A. A., Ranciere,  R., and Thoenig,  M. (2009). Growth  and risk at the industry level: The real effects of financial liberalization. Journal of Development Economics,  89(2):210–222.

Levine, R. (2005).  Finance  and  growth:  Theory  and  evidence.  In Aghion,  P.  and  Durlauf,  S., editors,  Handbook of Economic  Growth, volume 1, chapter  12, pages 865–934.

Levinsohn,  J.  and  Petrin, A. (2003).  Estimating production functions  using inputs  to control for unobservables. Review of Economic  Studies,  70(2):317–341.

Manova,  K., Wei, S.-J.,  and  Zhang,  Z. (2011).  Firm  exports  and  multinational activity  under credit  constraints.  Mimeo, Stanford  University.

Midrigan,  V. and  Xu, D. (2010).  Finance  and  misallocation:   Evidence  from plant-level  data. NBER  Working Paper,  No. 15647.

Mody A, Murshid P. 2005. Growing up with capital flows. Journal of International Economics, 65: 249-266.

Moll, B. (2012).   Productivity losses from financial  frictions:   Can  self-financing undo  capital misallocation?   Mimeo, Princeton University.

Moore, W. (1998).  The determinants and effects of right-to-work laws:  A review of the  recent literature. Journal of Labor Research,  19:445–469.

Olley,  G.  S. and  Pakes,  A. (1996).   The  dynamics  of productivity in the  telecommunications equipment industry.  Econometrica, 64(6):1263–1297.

Pagano,  M. and Pica,  G. (2011).  Finance  and employment. CSEF  working paper,  283. Philippon, T.  and  Reshef, A. (2011).  Wages  and  human  capital  in the  U.S. finance industry: 1909-2006. Mimeo, Stern  School of Business,  New York University.

Rajan,  R. G. and  Zingales, L. (1998).  Financial  dependence  and  growth.  American  Economic Review, 88(3):559–586.

Ranciere,  R.,  Tornell,  A.,  and  Westermann, F.  (2006).   Decomposing  the  effects of financial liberalization: Crises vs. growth.  Journal of Banking   Finance, 30(12):3331–3348.

Restuccia,   D.  and  Rogerson,  R.  (2008).   Policy  distortions and  aggregate  productivity with heterogeneous  establishments. Review of Economic  Dynamics,  11(4):707–720.

Rodrik D. 1998. Who needs capital account convertibility? In S Fischer, R Cooper, R Dornbusch, P Garber, C Massad, J Polak, D Rodrik, SS Tarapone (eds.), Should the IM F pursue capital account convertibility?, 55-65, Princeton University, Princeton, New Jersey.

Schimmelfennig,  F. and Sedelmaier,  U. (2004).  Governance  by conditionality: EU rule transfer to the candidate countries  of Central and Eastern Europe.  Journal of European Public Policy, 11(4):661–679.

Schneider B. 2003a. Measuring capital flight: Estimates and interpretations. Working Paper 194, Overseas Development Institute.

Summers LH. 2000. International financial crises:Causes, prevention and the cures. American Economic Review, 90: 1-16.

Tressel, T. and Detragiache, E. (2008). Do financial sector reforms lead to financial development? Evidence from a new dataset. IMF Working Paper,  No. 08/265.

Uzawa,  H. (1962).   Production functions  with  constant  elasticities  of substitution.  Review of Economic  Studies,  29(4):291–299.

Vliegenthart, A. and Horn, L. (2007).  The role of the EU in the (trans)formation of corporate governance  regulation  in Central Eastern Europe.  Competition and Change, 2(11):137–154.

Wacziarg,  R. and Welch, K. H. (2008).  Trade  liberalization and growth:  New evidence.  World Bank Economic  Review, 22(2):187–231.

Wood, A. (1995). How trade  hurt  unskilled workers. Journal of Economic Perspectives, 9(3):57–80.

World Development, 39(1):1–19.