Enlarging the Contracting Space: Collateral Menus, Access to Credit, and Economic Activity

A large literature has looked at links between creditors' rights and financial development, documenting a positive relationship between creditor protection and the size of credit markets (e.g., La Porta et al. 1997, 1997; Levine 1998, 1999). Notably, existing analyses are often conducted with country-level data, making it difficult to identify how different elements of financial contracting ultimately work in promoting credit expansion. Against the backdrop of this research, credit reforms in emerging economies have focused on the allocation of control rights over assets in liquidation as a way to facilitate contracting. We extend this literature by providing evidence on the micro-level foundations of the connections between the contracting environment and credit expansion.

Theory suggests that access to credit in imperfect capital markets depends on the menu of assets that can be credibly offered as collateral by borrowers (Aghion and Bolton 1992; Hart and Moore 1994). In this paper, we study reforms that emphasize the idea of enlarging the contracting space (larger “collateral menus”) available to parties as a way to promote secured credit transactions. We look not only at credit taking by firms, but also at how reform-enabled credit expansion ultimately affects real aspects of economic activity. Recent reform in Romania is considered a textbook example of change that enlarged the menu of assets that can be used as collateral in credit transactions (Fleisig, Safavian, and de la Peña 2006). We thus start our analysis with the Romanian setting as a basis to understand how the reform worked and to analyze the economic responses it triggered. We subsequently generalize our analysis to a broader group of Eastern European countries, some of which witnessed similar legal changes, but at different time points.

Throughout the 1990s, Romania's Commercial Code only allowed for secured transactions involving “immovable assets” (e.g., land and buildings). For practical purposes, “movable assets” (e.g., machinery and equipment) could not be pledged as collateral. Around the world, firms' movable assets comprise about half of their total fixed assets (Alvarez de la Campa 2011); yet in Romania, as well as in other Eastern European countries, firms had a hard time using those assets to secure credit. In light of a legal framework that favored debtors, Romanian firms faced difficulties in financing investment in machinery and equipment through alternatives, such as leasing and unsecured lending with contractual provisions. In 2000, the Romanian government implemented Law 99, which transformed the framework in which debt contracts could be written, giving Romanian firms a new alternative for financing machinery and equipment.

In essence, Law 99 eliminated the possessory nature of security interests. As an intended consequence, the contracting framework no longer required the physical transfer of movable assets to the creditor and made it possible for firms to give creditors “substitute” assets (e.g., similar assets or cash equivalents) if mutually agreed. It also introduced a uniform electronic system of real-time information on seniority of security interests over movable assets. Law 99 gave lenders much larger latitude in the origination of credit facilities (e.g., varying syndication, subordination, and maturity structure), allowing for more loans to be taken out of a new class of legally pledgeable assets.

From an empirical standpoint, the Romanian setting is unique in identifying the types of assets that could allow for credit expansion under the collateral reform. As the law made it possible to pledge movable assets as collateral for the first time, it would affect firms that make intensive use of machinery and equipment in their operations. Immovable assets, on the other hand, were used as collateral before 2000, and the reform had no bearing on contracts secured by real estate assets. This institutional wrinkle helps us pinpoint the link between collateral menus and access to credit.

To estimate the effects of the collateral-menu-enlarging reform, we take advantage of the fact that some sectors of the economy naturally use more machinery and equipment than do others. We conduct a difference-in-differences test in which we contrast firms operating in sectors with high versus low demand for movable assets, before and after the passage of the law. To minimize potential confounders (e.g., concurrent credit supply shocks), we benchmark the results from this test against a test that measures pre-post reform changes along the high versus low demand for immovable assets, which were not contemplated by the legal reform. We gauge the external validity of our inferences by extrapolating our tests to other Eastern European countries. While these economies are comparable in relevant dimensions, the timing of the collateral reforms was not contemporaneous (or did not happen at all), owing to idiosyncrasies affecting the speed of various political processes.¹ This time variation in the wave of reforms across Eastern Europe allows us to exploit both within-country and cross-country contrasts.

Our base tests show that firms operating in sectors more intensive in overall tangible assets (the sum of movable and immovable assets) observe an increase in their leverage ratios after the reform. As we break these effects across movable and immovable assets, however, we find that only those firms operating in sectors intensive in movable assets observe an enhancement in their ability to borrow. We look not only at the amount of debt firms raise, but also at the likelihood firms start using debt in the first place. On this front, albeit statistically weaker, our results point to an expansion of access to credit: firms operating more movable assets are more likely to abandon their previous “zero-leverage” status. We also find that high-movable-assets firms accumulate less cash in their balance sheets. Whether firms operate more or less immovable assets, in contrast, does not have any effect on their use of debt financing nor on their savings policies following the reform.

The increase in credit access that stems from operating more movable assets is economically sizable. Controlling for key capital structure determinants, such as firm size, age, and profitability, a firm operating in the top quartile of the movable assets sectoral distribution observes an increase in its leverage ratio by 3.7 percentage points more than its counterpart in the low movable-assets ranking following the reform. This is a significant number when one considers that the average debt-to-asset ratio of Romanian firms is just 10.5%; a 35% increase relative to the baseline. Using the same comparison, the proportion of zero-leverage firms drops by 32 percentage points more in the high-movable-assets category (or 57% of the sample mean) after the reform. This latter result is statistically weaker for the average firm. As we investigate this effect in more depth, however, we find that it is particularly significant for smaller firms in the economy.

Our analysis goes further in showing how changing the ability to sign secured debt contracts can have far-reaching implications for corporate outcomes across Eastern Europe. Using data from ten countries in the region, we confirm that firms with more movable assets raise more debt and save less cash after collateral-menu-enlarging reforms. We also show that they invest more in fixed assets, which allows for more debt capacity. To gauge the effect of the spur in capital investment, we consider a number of additional outcomes. First, we examine if firms changed their labor demand and find that firms hire more workers. We look at measures of profitability and find that they also increase for firms with more movable assets following reforms. Finally, we examine if the increase in tangible assets and labor usage leads to changes in productivity. We find that firms with more movable assets observe an increase in total factor productivity. Our findings imply that firms not only raise more funds and grow more as a result of their enhanced debt capacity, but also seem to establish a better asset mix.

Looking at the economy-wide consequences of collateral reforms, we document important reallocation effects. Sectors that make more intensive use of movable assets witnessed a stark increase in their share of aggregate fixed assets in Romania: from 37% to 52% between the pre- and post-reform periods. These same sectors witnessed a significant increase in their share of employment in the economy, from 31% to 38%. Similar patterns are found for Latvia and Poland, which reformed their legal contracting environment to allow for borrowing against movable assets in 1999 and 1998, respectively.

We subject our results to a long battery of checks. Among others, we falsify our experiment by testing for the introduction of “placebo reforms” in 2000 in the countries that share borders with Romania (Bulgaria, Hungary, and Ukraine), as well as Romania's largest trading partner (Italy). None of these countries passed such reforms in or round 2000, yet one could worry that underlying economic, geopolitical, or technological factors may have allowed firms in some movable-intensive sectors to gain more access to debt starting in 2000. We find no significant changes in the credit capacity of firms with movable assets in these placebo countries. As we detail below, our analysis also considers sectoral dynamics (e.g., sensitivity to business cycle) and utilizes alternative econometric methods (e.g., matching estimation) to ensure the robustness of our results.

Because of data limitations, we are not able to fully observe the extent to which movable assets were used to secure credit through other alternatives, such as leasing prior to a collateral reform. However, our experimental design shows strong evidence that firms operating more movable assets borrowed more and invested more as a result of a reform that altered the terms of standard secured lending. Simply put, we show that allowing firms to collateralize machinery and equipment enlarged the debt contracting space, with positive consequences for access to credit and real economic activity.

Only a small literature has analyzed the impact of sudden changes in the contracting environment using detailed, country-specific firm data as we do here. Lilienfeld-Toal, Mookherjee, and Visaria (2012) and Vig (2013) look at reforms in India that empowered creditors in seizing assets of defaulting firms. They find that strengthening enforceability led to a decline in borrowing, especially for smaller firms, which is arguably a counterproductive outcome from a policy perspective. Assunção, Benmelech, and Silva (2013) study a reform in Brazil that simplified the sale of repossessed cars used as collateral for auto loans. The authors find that the change led to more lending, eventually allowing for riskier borrowers to obtain loans for more expensive cars. Looking at the reform of Napoleonic Code-based security laws in France, Aretz, Campello, and Marchica (2015) find that capital-intensive, small, start-up firms located in rural areas increased their borrowing with value-enhancing effects at both the micro- and macroeconomic levels. These studies highlight the difficulties policy makers face in predicting the outcomes of credit reforms.

Our paper is closer in spirit to Haselmann, Pistor, and Vig (2009), who study both bankruptcy and collateral reforms in Eastern Europe and find that such legal reforms led to more lending. Our paper extends their work by identifying some of the mechanisms through which collateral reforms led to higher firm borrowing. As Eastern European reforms expanded the collateral menu to include movable assets, we exploit within-country variation in asset-type utilization intensity around the reforms to identify that firms operating more movable assets borrowed more. Moreover, we show that beyond increasing credit, collateral reforms ultimately triggered changes in firm-level real-side outcomes. Our paper also relates to Calomiris et al. (2015), who document that loan-to-values of loans collateralized with movable assets are lower in countries with weak collateral laws. Our paper complements their work by analyzing the problem from the point of view of the firms, instead of creditors, and by looking into within-country law changes, rather than comparing laws across countries.

Our paper also adds to the literature that looks at the impact of collateral on leverage ratios. Among recent studies, the emphasis has been on variations in the value (Gan 2007), quantity supplied (Campello and Giambona 2013), or salability (Benmelech 2009) of assets that are used as collateral. Our study is different as it identifies the impact of the enlargement of the contracting space–what is accepted as collateral–on access to debt financing. In this way, our results are important for policy makers, who cannot alter asset liquidation values or their supply in secondary markets, but can alter collateral menus as a way to enhance financial contractibility. Our paper also stands out in that real-side outcomes, such as productivity, labor, or profitability, are only rarely examined in conjunction with the impact of collateral on access to credit.²

1 Institutional Setting

Research from the World Bank points to Romania as a textbook example of a country enacting changes that significantly enhanced the law of secured transactions (Fleisig, Safavian, and de la Peña 2006). In this section, we start by describing Romania's reform. We then discuss collateral reforms across a broader group of Eastern European countries.

1.1 Collateral reform in Romania

Firms operating in well-developed credit markets typically face three alternatives for financing investment in machinery and equipment. The first alternative is secured lending. The firm obtains a loan from the creditor and purchases the equipment, pledging it as collateral in a legally binding contract. The creditor has a “security interest”—not ownership—over the equipment, which it can enforce in the event of default. The second alternative is leasing. The leasing contract guarantees the use of the equipment in exchange for regular payments from the lessee (firm) to the lessor (creditor).³ One advantage of leasing is that ownership still lies with the lessor, which allows for faster recovery in default. At the same time, moral hazard problems often arise from the separation of ownership and control. The third alternative is unsecured lending, with contractual provisions that indirectly try to make sure that the liquidation value of the equipment accrues to the creditor (e.g., a combination of asset sales restrictions and antilayering provisions).

Prior to 2000, Romania lacked a legal framework that allowed firms and creditors to write secured loan contracts over machinery and equipment. Specifically, two major codes governed secured transactions in Romania at the time: the Civil Code and the Commercial Code.⁴ Under codified law, a creditor could secure a loan by creating a security interest over immovable assets (“mortgage”) and over movable assets (“pledges”). Critically, however, pledges took the form of possessory interests. This meant that the law did not allow parties to establish security interests over movable assets without transferring actual possession to the creditor. In practical terms, a firm could not pledge a piece of equipment to a third party and still physically control it (keep it under its roof). In other words, the firm could not write a standard secured loan contract with a bank. In principle, firms could still finance machinery and equipment investment through leasing or unsecured credit. However, as explained above, in light of a legal framework that favored debtors, the market for leasing was bleak at that time.⁵ The country's adherence to the traditional (French-style) pledge system made it difficult for firms to use a host of its productive assets as collateral in standard debt contracts. As we explain below, the 2000 reform allowed nonpossessory secured lending over movable assets, giving Romanian firms a new alternative for financing machinery and equipment.

Throughout the 1990s, the European Bank of Reconstruction and Development (EBRD) advocated for secured transactions reforms in Eastern Europe. In January 1999, the Center for the Economic Analysis of Law (CEAL), with the support of local attorneys and the World Bank, drafted a proposal on the regulation of security interests over movable assets. Shortly thereafter, in May 1999, the Romanian parliament passed Law 99, whose Title VI contained the “Legal Treatment of Security Interests in Personal Property.” The new law was molded after Article 9 of the United States' Uniform Commercial Code, seen as the state-of-the-art legislation on secured transactions over movable assets. Law 99 came into full force in December 2000.

Romania's Law 99 vastly expanded the range of assets that could serve as collateral. It introduced a broad system of security interests and derogated the old pledge regime. Importantly, the law allowed parties to establish security interests over movable assets without having to transfer possession of the asset to the creditor. The law also introduced the “Electronic Archive of Security Interests in Personal Property,” a fully automated system of registration of security interests over movable assets that instantaneously files into a database notices that a security interest has been taken over a movable asset.⁶ The law further awarded creditors legal powers to repossess collateral in the event of default without court intervention, as long as a breach of the peace did not occur.

Figure 1 plots the time evolution of the number of filings into the Electronic Archive (left vertical axis). The archive's entries have grown exponentially since its inception in 2000. The movable assets archive system received 65,000 filings in 2001 and rose to 360,000 filings by 2005. As of 2005, cumulative gross filings amounted to roughly one million. The notice of the security interest does not require filing the amount of the obligation secured, and hence the amount of secured credit cannot be determined from the number of filings. Nonetheless, several other indicators are consistent with a rapid and large increase in the volume of credit granted to firms after the 2000 reform. For example, the number of borrowers reported in the Central Bank's debtor registry rose from 18,000 in 2000 to more than 100,000 in 2005 (de la Peña and Fleisig (2004)). Along these lines, Figure 1 displays the evolution of the total volume of corporate bank credit as a share of GDP from 1996 to 2005 (right vertical axis). The fraction of corporate credit to GDP nearly tripled between 2000 and 2005, rising from 7% to 20%.

1.2 Collateral reform in other Eastern European countries

We extend our study to a cross-country analysis by collecting information on collateral laws for nine additional Eastern European economies. While these economies are comparable on a number of dimensions–they inherent much of the same legacy of centralized, socialist-oriented, large manufacturing base–the passage of collateral reforms was not contemporaneous, owing to various idiosyncrasies. Table 1 reports the relevant details of the collateral reforms observed in the countries in the region during our sample period 1996–2005. Four of the countries had reformed their collateral legal frameworks by 1996, the first year of our sample period: Bulgaria, Estonia, Hungary, and Lithuania. Two other countries reformed in the middle of our sample period: Latvia passed the “Law on Commercial Pledge” in 1998 (effective in 1999), and Poland passed the “Law on Registered Pledge” in 1996 (effective in 1998). The reforms in Latvia and Poland were similar to the reform passed in Romania: they allowed parties to establish security interests over movable assets without having to transfer possession to the creditor, and they created a collateral registry to register such security interests. Three remaining countries did not implement reforms over the time period we study: Czech Republic, Ukraine, and Russia.

In conducting our tests under a treatment effects framework, we generalize our findings by integrating and comparing outcomes for firms operating in these different sets of countries.

2 Data and Empirical Strategy

2.1 Data

We use firm-level information from Amadeus, a commercial dataset compiled by Bureau van Dijk. Amadeus contains financial statements from millions of companies operating in 35 European countries. Amadeus collects data from local information providers, which in most cases are the local registries. In Romania, all joint stock companies, partnerships, and limited liability companies are required to file their financial statements to the Romanian National Trade Register Office. As a result, Amadeus's data coverage of Amadeus of Romania, covering the majority of privately held firms in the country, is particularly comprehensive. The filing requirements for other Eastern European countries are less strict, leading to less comprehensive coverage.

Our basic outcome variable is leverage, which should be plausibly affected by changes in the menu of assets firms are able to offer as collateral for debt taking. We measure Leverage as the ratio between total debt and the book value of assets. We define ZeroLeverage as a dummy variable equal to one if a firm has no leverage and zero otherwise. We also glean additional insights into firms' borrowing by looking at their savings behavior; in particular, their need to carry cash balances. We define Cash as the ratio of cash holdings and cash equivalents to total assets. Our base analysis controls for the standard determinants of capital structure that are available in the data (e.g., Rajan and Zingales 1995; Lemmon, Roberts, and Zender 2008). We measure Size as the log of total assets; Age is the number of years the firm is in operation; Profitability is the ratio of earnings before interest and taxes to total assets; and OverallTangibility is the ratio of fixed assets (i.e., property, plant, and equipment) to total assets.

We also study the effect of collateral reforms on a set of real-side corporate outcomes. Investment is the change in fixed assets scaled by lagged fixed assets; Employment is the number of employees; total factor productivity (Productivity) is the residual from a Cobb-Douglas production function;⁷ and Sales is the log of sales. Following the literature on asset tangibility and leverage, we focus on manufacturing firms (e.g., Campello and Giambona 2013). We winsorize variables at the upper and lower 1% to avoid the impact of extreme outliers.

Table 2 reports summary statistics of Leverage, ZeroLeverage, and Cash for each of the ten countries in the sample during the 1996–2005 period. The mean value of Leverage in Romania is 10.5%. Average leverage varies across the region, ranging from 4.4% in Russia to 18.9% in Latvia. The fraction of zero-leverage firms also varies across countries. In Romania, on average, 57% of the firms in the sample are financed entirely with equity. The country with the highest fraction of zero-leverage firms is Russia (73%), and the country with the lowest fraction of zero-leverage firms is Latvia (13%). Firms in Romania hold, on average, 7.9% of their assets in cash. The average cash ratio ranges from 2.6% in Ukraine to 15.9% in Estonia. The average firm in the sample is young and small, consistent with private sector enterprises in transition economies. For example, the average Romanian sample firm is seven years old, has total assets worth $1.8 million (in 2000 U.S. dollars), and hires 65 employees.

2.2 Test strategy

Because a collateral reform allows firms to pledge movable assets as collateral, it should particularly benefit firms operating in sectors that make intensive use of assets such as machinery and equipment. To identify the effect of a reform, we take advantage of the fact that some sectors are inherently more intensive than others in their demand for machinery and equipment.

We exploit ex ante variation in asset-type demand that stems from the nature of firms' production processes and conduct a difference-in-differences test around the passage of the reform. To do so, we rank manufacturers according to a movable assets demand index (explained shortly). We then assign those firms operating in sectors at the high end of the sectoral ranking to a “treatment group.” The “control group” consists of firms at the bottom of the ranking. Next, we calculate the pre- versus post-reform difference in the outcome variable of interest (e.g., Leverage) for the treated group, doing the same for the control group. Finally, we calculate the difference between these two group differences. Our estimation accounts for both individual firm and year fixed effects. As we discuss below, we provide a number of checks on the validity of our strategy.

2.3 Sectoral asset indices

In a legal framework in which movable assets are considered “dead capital,” the use of movable assets in firms' production processes is likely to be a distorted representation of the underlying demand for those assets. In particular, it is likely that movable assets are under utilized. As such, even if Amadeus provided data on the observed use of movable assets before the reform, we could not use that information to make predictions about the impact of the collateral reform. Instead, we need to gauge firms' desired use of movable assets. To do so, we must identify comparable firms whose use of movable assets are unconstrained by the severe legal frictions in place before the reform.

2.3.1 Index construction

With this differentiation in mind, we construct the movable assets index using data on manufacturers from the United States as follows. First, we follow Campello and Giambona (2013) in identifying information on the decomposition of firms' fixed assets between: (1) machinery and equipment and (2) land and buildings. This information is conveniently available for the 1984–1996 period in the Compustat database; that is, it contains data on manufacturers' asset mix for the period prior to the collateral reform. For each individual firm, we compute the time-average ratio of machinery and equipment to total assets. Next, we follow the guidelines of the International Standard Industrial Classification (ISIC) and divide the sample into 48 three-digit sectors. For each sector, we calculate the movable assets index as the median of the movable assets-to-total assets ratio of the firms operating in that sector. We do the same calculation for the land and buildings-to-total assets ratio, thus computing an immovable assets index. Likewise, we use the fixed assets-to-total assets ratio to compute an overall tangibility index. Overall tangibility equals 34% of total assets, on average. Movable assets, in turn, represent 54% of the ratio between fixed assets and total assets. The correlation between the movable and immovable assets indices is positive- but low (only 0.3).

Figure 2 plots the movable assets index for the 48 sectors examined. The figure reveals a substantial degree of cross-sectoral variation in the utilization of movable assets. Manufacturing of precious metals, domestic appliances, and furniture are examples of sectors with low intensity in movable assets used in production. In these sectors, machinery and equipment amounts to about only 10% of total assets. In contrast, the manufacturing of structural metals, glass, and paper represent examples of sectors with a high use of movable assets. In these sectors, machinery and equipment amounts to well over 30% of total assets.

As we take our tests to other economies in Eastern Europe, we study the relative size of movable-intensive sectors in the countries in our sample. We find that the fraction of aggregate fixed assets allocated to sectors above the top quartile of the movable assets index in each sample country is as follows: Bulgaria 53%, Czech Republic 40%, Estonia 51%, Hungary 38%, Latvia 40%, Lithuania 46%, Poland 38%, Romania 40%, Russia 49%, and Ukraine 46%. The relatively narrow band in which these fractions are distributed may be unsurprising given that our sample countries operated as central planned economies for several decades leading up to the time period we study.

2.3.2 Index utilization

Our approach does not require that the value of the index in each sector is the same in the United States and Eastern Europe. The approach only assumes that the sectoral ranking of demand for movable assets is sensibly similar.⁸ For example, the manufacture of paper products demands an intense use of large mills (heavy machinery and equipment), regardless of whether the factory is operated in the United States, Western Europe, or Eastern Europe. On the other hand, precious metal production is relatively less dependent on machinery, with most hard assets associated with the plant, again independent of the country in which firm operates. As we restrict our attention to traditional manufacturing activities in countries with sizable industrial sectors, our working hypothesis appears to be plausible.

Because we have data on overall tangibility for Romanian firms (that is, movable plus immovable assets), we can compare sectoral indices based on total asset tangibility across U.S. and Romanian manufacturers as a way to assess the plausibility of our strategy. Indeed, we can make that comparison with any other country that serves a reasonable benchmark for credit-unconstrained financing. Our prior belief is that the observed asset mix of Romanian firms before the 2000 reform was distorted away from an optimal benchmark due to legal constraints. The collateral reform, in turn, should make Romanian firms more able to utilize an optimal asset mix, converging to the asset mix observed in the United States or other credit-unconstrained economies.

To check this conjecture, for each year between 1996 and 2005, we calculate the sectoral overall tangibility index using Romanian data, as well as data from the United States and Germany. In Figure 3, we plot the evolution of the correlation coefficient between the Romanian- and U.S.-based sectoral rankings of overall tangibility. Prior to the 2000 reform, the correlation between both rankings is quite stable, hovering around a value of 0.35. After the reform, the correlation increases steadily, and by 2005 it is equal to 0.42. In the same figure, we plot the correlation between the Romanian- and German-based rankings of overall tangibility.⁹ The correlation between the Romanian- and German-based rankings exhibits a very similar behavior. The correlation is stable prior to 2000, after which time it starts to steadily increase. Figure 3 suggests that prior to the reform, firms in Romania displayed tangible-to-total asset ratios that were quite different than those of comparable, credit-unconstrained firms based in either the United States or Germany. Following the collateral reform, Romanian firms' asset mix started to more closely resemble the asset mix of foreign-based benchmark firms.

3 Credit Expansion

3.1 The baseline empirical model

A unique characteristic of the collateral reform is that it affected only movable assets, not immovable assets. This provides for an extra identification wrinkle in our difference-in-differences setting. In particular, one concern with our baseline test is that there could be a concurrent credit supply shock at the time of the reform affecting all tangible assets, both movables and immovables. Because immovable assets were allowed to be pledged before and after the reform, we address this concern by adding an interaction term to Equation (3) between $Post$ and a dummy that equals one (zero) if the firm belongs to a sector in the top (bottom) quartile of the immovable assets sectoral index. We denote this variable as $HighImmovableAssets$⁠.

3.2 Parallel trends

Our difference-in-differences strategy assumes that, in absence of the reform, the change in the outcome variables of interest would have been the same for firms in the treated and control groups. Accordingly, it is important to check whether trends in the outcome variables for both groups were similar (“parallel”) prior to the reform in Romania. We do so looking at the evolution of changes in leverage ratios, the fraction of zero-leverage firms, and cash holdings before the reform. Panel A of Table 3 reports the results for Leverage. The difference between the change in leverage for the treated and control groups is not statistically different from zero. This holds for all pre-reform horizons we consider, going back all the way to the beginning of our sample period. Panels B and C show similar patterns for ZeroLeverage and Cash for the two comparison groups. In all, there are no discernible differences in trends for either debt or cash ratios for our sample firms in the high- and low-movable-assets categories before the collateral reform.

3.3 Leverage ratios in Romania

Table 4 reports the results for Leverage in Romania. To build intuition, we start by estimating the effect of the collateral reform across sectors with different intensities in overall asset tangibility, which includes all types of tangible assets (movables and immovables). The estimates in Column (1) show that the reform increased leverage in firms operating in sectors with high overall tangibility by 1.2 percentage points more than in firms in low-tangibility sectors. This base result is statistically significant, but economically confounded because not all types of fixed assets were affected by the reform. Accordingly, we break the overall tangibility effect into its different components. In particular, because the collateral reform only boosted the pledgeability of movable assets, there should only be an effect in sectors that are intensive users of movable assets. This is what we find. According to Column (2), the collateral reform increased firms' leverage in movable-intensive sectors by 2.4 percentage points more than in sectors in which firms operate fewer movable assets. That is, for firms of the same size, age, profitability, and even overall tangibility, those that operate in sectors that have a higher use of movable assets observe a markedly higher use of debt financing following the collateral reform. In Column (3), we include the interactions between the Post dummy and both the movable-assets and immovable-assets indicators.¹² The results confirm that the effects are only observed for firms operating in sectors intensive in movable assets. The reform increased leverage of firms in movable-intensive sectors by 3.7 percentage points more than in firms in nonintensive sectors. The effect is highly significant and of sizable magnitude: it amounts to 35% of the average sample leverage (⁠$=3.7%/10.5%$⁠).

3.4 Leverage ratios across Eastern Europe

We report the results for Leverage in Table 5. As before, we first estimate the effect of the collateral reform across sectors with different intensities in overall asset tangibility. Column (1) shows that collateral reforms, on average, increased leverage for firms in high-overall-tangibility sectors by 3.6 percentage points more than in firms in low-tangibility sectors. When we break the overall tangibility effect into its different components, we again find that the effect is working entirely through the movable-assets component. According to Column (2), the typical collateral reform increased leverage in firms operating in movable-intensive sectors by 4.1 percentage points more than in firms in nonintensive sectors. From Column (3), we also observe that the immovable assets coefficient is not statistically different from zero, but the movable coefficient becomes slightly larger (the effect is 4.7 percentage points). In all, the results we obtain when analyzing the entire Eastern European region resemble those of the more focused Romanian data analysis.

3.5 Propensity to access credit

The evidence above shows that firms operating more movable assets carry more debt in their balance sheets after a collateral reform. From the point of view of promoting access to credit, it is important to know whether firms that previously did not use debt (“zero-leverage firms”) are able to resort to this type of financing after the reform. To gauge this effect, we re-estimate Equation (3) using a dummy that equals one if the firm has no debt in its balance sheet and zero otherwise (ZeroLeverage) as dependent variable. Because the dependent variable is binary, we estimate a linear probability model.¹⁶ Columns (1) and (2) of Table 6 report the results for Romania, and columns (3) and (4) report the results for the full sample of ten countries.

The results in Table 6 suggest that a collateral reform reduced the probability of a firm having zero leverage in the movable-intensive sectors (see Column 1). In Column (2), we control for the interaction between the post-reform dummy and the immovable-assets dummy. The coefficient on the movable dummy remains negative and becomes twice as large in magnitude: the probability of a firm having zero leverage in movable-intensive sectors decreased by 32.4 percentage points more than in nonintensive sectors. This is a sizable magnitude, accounting for 57% of the average fraction of firms with no leverage in the sample (=32.4%/57.2%). The effect is estimated less precisely, nonetheless, suggesting that we should examine more closely the sample variance of this effect. We conjecture that this could be reflecting an heterogenous effect of the reform on firms. This is a particularly important examination because it allows us to gauge the distributional effects of the reform. These types of distributional effects are often a relevant metric with which policy makers evaluate reforms.

To explore this dimension, we divide Romanian firms in movable-intensive sectors into deciles according to size. Figure 5 reports the distribution of the fraction of zero-leverage firms within each size bin for the pre- and post-reform periods (panels A and B). Before the reform, 83% of the firms in the smaller size bins had no debt in their balance sheets. This fraction declines as we move toward larger size bins. After the reform, the fraction of zero-leverage firms declines across all size bins, but the effect is concentrated primarily in the smaller size bins (deciles 1 through 7). Panels C and D replicate the results for the sectors that are not intensive in movable assets. The panels confirm that the effects of the reform are only present in sectors that intensively use movable assets. Notably, the contrast of zero-leverage firms across high- and low-movable-assets sectors disappears with the reform (compare panels B and D).

Columns (3) and (4) of Table 6 reports the zero-leverage results for our complete set of Eastern European countries. In this case, even after controlling for the immovable-assets dummy, we observe a significant effect of the reform on firms operating in movable-intensive sectors. Collateral reforms introduced in Eastern Europe, on average, reduced the probability of a firm having zero leverage in movable-intensive sectors by 21% more than in nonintensive sectors (Column 4).

3.6 Cash savings

Intuition suggests that firms with an enlarged capacity to borrow need to carry less cash in their balance sheets, because carrying cash is expensive if firms have easy access to credit (Acharya, Almeida, and Campello 2007). We study the effect of a collateral reform on corporate liquidity to better characterize our results. Savings capture the “dual” of debt, and using this alternative proxy as a dependent variable helps us guard against common endogeneity concerns in our standard leverage tests.

We report the results for regressions featuring the ratio of cash to assets as the dependent variable (Cash) in Table 7. Columns (1) and (2) report the results for Romania. According to the estimates, the reform reduced cash holdings of firms operating in movable-intensive sectors by 1.9 percentage points more than of firms not making intensive use of those assets (Column 1). The effect becomes larger (2.6 percentage points) after controlling for the immovable-assets dummy (Column 2). This is a sizable effect, accounting to 33% of the average cash-to-asset ratio in the sample (=2.6%/7.9%). Columns (3) and (4) report the results for the full sample of Eastern European countries. The results are in line with what we observe for Romania. In all, our estimates imply that better contracting terms for movable assets seem to make these assets more liquid and firms responded by moving away from hoarding cash.

4 Real Economic Activity

4.1 Real effects of access to credit

Having established that collateral reforms increased access to credit, we take our analysis one step further and look at the real-side implications of these changes. Looking at how financing decisions impact real corporate outcomes like investment and efficiency sets our study apart from the literature and highlights the policy relevance of our findings. To conserve space, we focus our analysis on the full sample of Eastern European countries.¹⁷ We estimate the cross-country specification in Equation (5) for each real outcome of interest.

Table 8 reports the results. From this table onwards, we report results returned from the specification that controls for both the movable-assets and immovable-assets indicators. Column (1) of Table 8 shows that collateral reforms in Eastern Europe, on average, increased the investment rate in fixed assets in firms operating in sectors intensive in movable assets by 3.8 percentage points more than in sectors that do not demand those assets. The magnitude of the effect is sizable, amounting to more than 60% of the average sample investment rate (=3.8%/6.3%). The results in Column (2) suggest that after a collateral reform, firms in movable-intensive sectors hired 2.6% more workers. According to Column (3), the productivity of firms in sectors with high movable assets usage increased by 3.4 percentage points. Column (4) shows that profitability also increased by 3.4 percentage points. Finally, Column (5) shows that sales increased by 2.6 percentage points more in sectors intensive in movable assets. For all these variables, the effect is uniform (i.e., indistinguishable) across sectors with different immovable-assets intensities.