1 Introduction

Business bankruptcy rates (ratio of the number of business bankruptcy filings to the number of business exits) in Spain are among the lowest in the world, which means that Spanish firms rarely enter a formal bankruptcy procedure. The goal of this paper is to explain this empirical observation, which may imply that economic agents regard the system as inefficient and try to deal with financial distress in alternative ways.Footnote 1 For that purpose we employ a large sample of Spanish, French and UK firms, finding that small businesses in Spain, unlike their European counterparts, rely on mortgage foreclosuresFootnote 2 as the main alternative to bankruptcy proceedings.

According to Table 1 Spain had the second lowest bankruptcy rate out of 26 countries, including both high-income and emerging economies, in 2006. An even more striking observation is the difference in the orders of magnitude between Spain and other developed economies: for instance, while there were around 29 bankruptcies per 100 firm exits in France and 16 in the UK, there were 0.3 in Spain. Only the deep economic crisis that Spain is currently experiencing has modestly increased the number of bankruptcy filings, but the Spanish bankruptcy rate was still one of the lowest in the world in 2010 (see Table 1).

In contrast with the low incidence of business bankruptcies, business mortgage foreclosures have soared during the crisis. While around 8,000 firms filed for bankruptcy in 2012, there were nearly 26,000 business mortgage foreclosuresFootnote 3 in the same year. Moreover, the latter figure must be considered a lower bound, since small business owners may finance their firms with loans secured on their homes (Berkowitz and White 2004) but, if lenders repossess the collateral, they will be reflected as residential foreclosures in the official statistics.

Table 1 Business bankruptcy rates around the world

However, the use of bankruptcy procedures by Spanish businesses varies widely depending on the size of the distressed firms, as shown in Fig. 1. While the rates of micro firms (businesses with less than 10 employees) were around 0.15 % in 2006 and they have just reached 1.3 % during the economic crisis, those of non-micro firms were 10.4 % in 2006 and they have increased up to 90 % during the crisis, in line with the aggregate rates of developed countries. Since micro firms account for more than the 95 % of firms in Spain,Footnote 4 they are the key drivers of the low bankruptcy rate of Spanish companies. They are also very important in terms of economic activity: they accounted for 51 % of total employment and 28 % of total value added before the economic crisis and they currently account for 39 and 25 %, respectively.Footnote 5 Finally, although the available evidence is rather limited, Spanish micro firms seem to file for bankruptcy much less than some of their European counterparts: in 2006, the bankruptcy rates for self-employed and micro enterprises were 0.01 and 0.15 %, respectively, in Spain, while those in France were 11.1 and 23 %,Footnote 6 and the bankruptcy rate for self-employed in the UK exceeded 16.2 %.Footnote 7

Fig. 1
figure 1

Bankruptcy rates by size in Spain. Data are quaterly except for the first period 04C3 (last 4 months of 2004). Rates are annualized. Source: authors’ calculations on data from the Spanish National Statistics Institute. Size is measured in terms of employees. Micro: [0,9], small: [10,49], medium and large: \(>\)50. Non-micro: \(>\)9

Spanish micro firms also have other distinct characteristics. They hold, by far, the largest proportion of mortgage loans over financial debt, as shown in Fig. 2. Filing for bankruptcy is especially unattractive for them because a significant proportion of the bankruptcy costs are fixed (Van Hemmen 2011).Footnote 8 Personal bankruptcy may apply to many of those firms regardless of their legal form, because the distinction between limited and unlimited liability may be blurred for them, partly because lenders require personal guarantees or security in the form of a mortgage on the owner’s home (Berkowitz and White 2004).

Fig. 2
figure 2

% Mortgage loans over bank debt by business size in Spain. Source: Authors’ elaboration with data from the Central Credit Register and the Central Balance Sheet Data Office, Banco de España

Consistent with those stylized facts, our hypothesis on the low business bankruptcy rates in Spain is the following. Filing for bankruptcy in Spain is very costly for both small firms and their creditors. Due to this, the capital structure of micro firms is biased towards mortgage loans (i.e., loans secured on land and buildings). Having this capital structure allows them to avoid bankruptcy by carrying out debt enforcement via mortgage foreclosures,Footnote 9 which are cheaper procedures than bankruptcy, in case of financial distress.

In order to test this hypothesis the optimal identification strategy would be to analyse the impact of substantial changes in the Spanish bankruptcy law in both bankruptcy rates and firms’ capital structure. The current bankruptcy code entered into force in 2004 after a major legislative reform. But it seems that the de facto insolvency framework barely changed because the performance of bankruptcy proceedings did not seem to substantially improve (Gutiérrez 2005; Van Hemmen 2004), bankruptcy rates did not increase after the introduction of the new code and it seems that firms’ capital and asset structures have not changed either (Celentani et al. 2010).

By contrast, our identification strategy relies on cross-country comparisons. Specifically, we compare the observed choices (choice of capital structure, choice between bankruptcy and mortgage) of Spanish firms with those of firms from countries where their bankruptcy systems are more efficient and their laws do not incentivise them to bias their capital structure towards mortgage loans. France and the UK are chosen as the comparison group because their bankruptcy rates are much higher than the Spanish ones and because of the specific features of their insolvency frameworks.Footnote 10

Our findings corroborate the proposed hypothesis. First, there is a positive and strong correlation between the ex-ante probability of default and the ratio of tangible fixed assets (the assets that can be pledged as mortgage collateral) to financial debt in the case of Spanish micro firms, suggesting that firms with risky business models bias their capital structure towards mortgage loans to avoid filing for bankruptcy in the event of default. Second, a higher proportion of tangible fixed assets over financial debt significantly decrease the probability of being in bankruptcy among Spanish micro firms in financial distress. By contrast, these two relations do not hold either for Spanish larger businesses or for firms from the other two countries.

Finally, we must stress the importance of the research question. The model of García-Posada (2013) predicts that, in the context of the Spanish insolvency framework, there is a positive relation between bankruptcy rates and welfare. The intuition is that low bankruptcy rates and low welfare are the outcome of an institutional design characterised by the low efficiency and low creditor protection of the bankruptcy system relative to those of an alternative insolvency institution, the mortgage system. In that context, firms and their creditors avoid filing for bankruptcy by heavily relying on mortgage collateral, which can be repossessed and liquidated in the event of default. The problem is that the mortgage system is not well suited for some firms, which need to bias their asset structure to have enough collateral, with the ensuing productive inefficiencies. Those firms would be better off if they had access to a bankruptcy system that worked relatively well. In other words, as the bankruptcy and mortgage systems are imperfect substitutes, the equilibrium in which only mortgage is widely used (reflected in low bankruptcy rates) is Pareto dominated by the equilibrium in which agents can choose between the two insolvency institutions (reflected in higher bankruptcy rates). His analysis also predicts that bankruptcy will be unfeasible for the smallest firms in the economy as long as some of the bankruptcy costs are fixed. As some of those firms will have to overinvest in capital assets to sign their contracts under mortgage, they will incur in productive inefficiencies. If the absence of a well-functioning bankruptcy system for those firms also reduces their growth opportunities—e.g., by hampering access to unsecured lending such as venture capital—then the current insolvency framework may help explain the firm size distribution and the low aggregate productivity of the Spanish economy. This is consistent with the evidence of Fabbri (2010) in Spain, who finds that lengthy bankruptcy procedures decrease firm size and raise funding costs and with that of Ponticelli (2012) in Brazil, who shows that congestion in bankruptcy courts substantially reduces firm-level investment and productivity.

The rest of the paper is structured as follows. Section 2 provides a brief literature overview and discusses the paper’s main contributions. Section 3 discusses some key features of the insolvency framework of Spain, France and the UK Sect. 4 focuses on data sources and sample selection criteria. Section 5 explains the empirical testing of the hypothesis. Section 6 concludes. Appendix A provides a description of the main legal concepts used in this paper and Appendix B contains some robustness analyses.

2 Contribution and related literature

This paper is mainly related to the works of Morrison (2008, 2009), and Celentani et al. (2010, 2012). Morrison (2008, 2009) studied why US small distressed firms—defined as those with 500 or fewer employees—rarely file for bankruptcy. He argued that there are cheaper procedures for these firms, such as assignments for the benefit of creditors,Footnote 11 bulk sales,Footnote 12 foreclosures and private workouts. Their implementation, however, require that neither the debtor firm nor the creditors’ file for bankruptcy. They also face, unlike the bankruptcy system, major coordination and asymmetric information problems that may hamper their use. Thus he identified the conditions under which these problems are not very important so those procedures can be implemented: small firms, with simple capital structures (i.e., low number of secured creditors) and with close and trustworthy relationships with their creditors are likely to avoid filing for bankruptcy. This paper applies a similar reasoning to the Spanish, British and French case: wherever there are cheaper alternatives to bankruptcy, the latter will only be used when parties don’t reach an agreement, becoming the residual option.

Celentani et al. (2010, 2012) were the first that studied the low bankruptcy rates in Spain. They proposed an explanation that was not immediately contradicted by a number of aggregate stylized facts. Specifically, they used the theoretical prediction of Ayotte and Yun (2009), according to which low creditor protection and low judicial ability imply low bankruptcy rates, to conjecture a wide set of activities (leverage reduction, lenders’ screening and monitoring, choice of projects that trade off return for lower risk and/or lower liquidation costs, use of mortgage collateral) in which firms and their creditors could engage to reduce the probability of bankruptcy. This paper focuses on one of their ideas, the use of mortgage foreclosures as an alternative to formal bankruptcy procedures. To the best of our knowledge, this is the first study that addresses the research question with firm-level data, which allows testing the hypothesis by means of econometric analyses.

3 Insolvency frameworks

In this section we will focus on the features of the insolvency frameworks of Spain, France and the UK related to our hypothesis, namely, the choice between bankruptcy procedures and mortgage foreclosures and the choice of firms’ capital and asset structures. We will examine the incentives to file for bankruptcy of both the debtor firm and its creditors, as alternative procedures such as mortgage foreclosures can only take place if both parties refrain from filing. For a more thorough analysis of the insolvency frameworks see Celentani et al. (2010, 2012) and Davydenko and Franks (2008).

3.1 Spain

The Spanish bankruptcy system (Ley Concursal) only had, until very recently, an insolvency procedure, the concurso de acreedores (bankruptcyFootnote 13), both for firms and individual debtors.Footnote 14 Both the debtor and the creditors may initiate the proceedings.

Bankruptcy procedures are costly and lengthy, rendering them unappealing for both distressed firms and their creditors. The direct costs of bankruptcy are high, as those procedures are complex, uncertain, involve many creditors and face high information asymmetries between the company and its lenders, requiring a great deal of intervention by the court, insolvency administrators, lawyers, etc. According to the Doing Business estimates, those costs would account for a 15 % of the firm’s total assets. As a substantial part of those costs are fixed (Van Hemmen 2008), bankruptcy procedures are especially costly in the case of small firms. The median duration of a bankruptcy process ranged between 20 and 23 monthsFootnote 15 (Van Hemmen 2008) before the economic crisis. The modest increase in the number of bankruptcy filings due to the crisis has congested the courts and lead to a dramatic increase in the length of the procedures, which ranged between 28 and 42 months in 2011 (Van Hemmen 2012).Footnote 16 Finally, as the law does not provide any debt discharge for individualsFootnote 17 and homestead exemptions are very low, individual debtors—including self-employed people and owners of small limited-liability firms that pledge personal guarantees to obtain funding for their businesses—have no incentives to file for bankruptcy.

Mortgage foreclosures are much cheaper and quicker than bankruptcy procedures, as they are quite standardised processes with a low degree of uncertainty about its final outcome. According to European Mortgage Federation (2007), their total costs are between the 5 and 15 % of the price obtained in the auction of the collateral (the percentage decreases as the sale price increases), and their usual length is 7–9 months.Footnote 18

Hence, mortgage foreclosures are an attractive alternative to bankruptcy, especially in the case of small firms. But, to make possible that a firm and their creditors use the mortgage system in case of financial distress, the firm’s capital structure must be biased towards mortgage loans and their asset structure must be biased towards assets—such as land and buildings—that can be pledged as mortgage collateral.

3.2 France

The redressement judiciaire (judicial reorganization) and the liquidation judiciaire (judicial liquidation) are the main insolvency procedures for corporations in France. As for personal bankruptcy, which may apply to both consumer and entrepreneurs, there are two different procedures: the plan de redressement (reorganization plan) and the procedure de rétablissement personnel (procedure of personal recovery). The debtor, creditors, the public prosecutor and the court itself may initiate the proceedings.

Bankruptcy procedures are relatively cost-effective. According to the Doing Business estimates, the direct costs would account for a 9 % of the firm’s total assets and the average duration in 2007 was 14.2 months (Ministère de la Justice 2010). Moreover, self-employed and small business owners may have incentives to file for personal bankruptcy as they may benefit from debt discharge in some circumstances.Footnote 19

Another characteristic of the bankruptcy system is the high dilution that mortgage credit suffers inside bankruptcy (Davydenko and Franks 2008). First, there is an automatic stay for secured creditors until the end of the procedure. Second, bankruptcy courts tend to sell the assets below their potential market prices, as they are not obliged to sell the assets to the highest bidder, but they can sell the whole company to a lower bidder that commits to preserve employment, as creditors’ approval is not required for the sale of their collateral. Third, the state places its own claims and those of employees first in priority when the collateral is sold.

In that context, mortgage creditors would like to enforce their claims outside bankruptcy via mortgage foreclosures, but they are quite slow and expensive. According to European Mortgage Federation (2007), their total costs are between the 10 and 12 % of the price of allocation and their usual length is between 15 and 25 months. As a result, the response of creditors is to rely more on some types of collateral—such as personal guarantees and accounts receivable—that can be realised directly by secured creditors and are not diluted by preferential creditors. These collateral types are used more often than mortgage collateral (Davydenko and Franks 2008).

Hence, mortgage foreclosures are not an attractive alternative to bankruptcy and mortgage collateral is not a very appealing guarantee. As a consequence, we expect mortgage loans to have little weight in the firms’ capital structure and the assets that can be pledged as mortgage collateral—such as land and buildings—to account for a low proportion of their total assets.

3.3 UK

Although various corporate insolvency procedures coexist in the UK, administration is the most important one since the entry into force of the Enterprise Act 2002Footnote 20 and bankruptcy is the most common procedure used by individuals.Footnote 21 Both the debtor and the creditors may initiate the proceedings.

Bankruptcy procedures are quite cheap and fast. According to the Doing Business estimates, the direct costs would account for a 6 % of the firm’s total assets and their average duration would be \(<\)1 year (Armour and Hsu 2012; Frisby 2006). In the case of personal bankruptcy, debt discharge is allowed one year after the end of the procedure, providing incentives to small firms and self-employed to file for bankruptcy.

Another characteristic of the UK insolvency framework is the existence of floating charges. A floating charge is a security interest over a fund of a firm’s changing assets that “floats” until it “crystallises” (converts) into a fixed charge,Footnote 22 at which point the charge attaches to specific assets. The crystallisation can be triggered by a number of events, being one of them the borrower’s default. There are two main differences between a floating charge and other security interests such as a mortgage. First, because the security “floats”, the firm remains free to purchase and sell its assets. Second, the assets of the entire business can be pledged as collateral. Those characteristics grant high flexibility to the firm’s asset structure and permit it not to be biased through certain types of assets such as land and buildings.

Mortgage foreclosures are neither significantly faster nor cheaper than bankruptcy procedures. According to European Mortgage Federation (2007) their usual length is between 8 and 12 months and their total costs are around 5 %. These facts, together with the existence of floating charges, lead us to expect a low incidence of foreclosures and a relatively low weight of mortgage loans (land and buildings) in firms’ capital (asset) structures.

Table 2 summarises the main characteristics of the insolvency frameworks of Spain, France and the UK.

Table 2 Insolvency frameworks in Spain, France and the UK

4 Data

The firm-level data come from the OECD-Orbis database, which is the result of the treatment of the commercial database Orbis by the OECD (Ribeiro et al. 2010; Ragoussis and Gonnard 2011). Orbis contains financial information on both private and publicly held companies around the world although coverage, especially of small firms, greatly varies across countries. Orbis also provides other non-financial information, such as year of incorporation, industry, legal form and status. Status is a variable that tells the legal and economic condition of the firm: for instance, if the company is active or it has ceased its operations and if it is undergoing a bankruptcy procedure or not. The data have, however, some important limitations. First, if a business shut down without filing for bankruptcy, the records do not indicate which alternative procedure the firm used.Footnote 23 Second, the status is only observed at the moment in which the data are extracted from the database, i.e., no historical records are kept. Since the data from Orbis were extracted in 2010 (December 31, 2010), we have the status of each company at that time. Finally, as Orbis is a commercial database, our sample may not be representative of the whole population. We address this potential criticism in Appendix B.

Regarding the sample selection, we use data on firms from three countries: Spain, France and the UK We only keep their financial data for 2008 because of two reasons. First, the main variable in all our analyses will be constructed using the information on status, which is only available for 2010. This makes panel data an unfeasible structure for the sample, since the variation in the main variable will happen across sections, but not across time. Second, because of the time lag in the submission of financial statements by firms, the Orbis database is characterised by a typical time lag of 2 years (Ribeiro et al. 2010), which implies that coverage (in number of companies with complete records) for 2009 and 2010 is very poor, leaving 2008 as the best choice. While this time gap could be problematic, it alleviates a simultaneity bias that may arise if the bankruptcy process or the alternative insolvency procedure affects the company’s financials (e.g. a foreclosure on some of the firm assets or a debt haircut), as our regressors will be lagged twice. We also apply some filters to clean the data. We exclude state-owned companies, non-profit organisations and membership organisations. To avoid double-counting of information we eliminate all consolidated accounts for which unconsolidated information exists. Finally, we remove inconsistent observationsFootnote 24 and extreme values. Our final sample has more than 560,000 firm-level observations.

For the empirical analyses of this paper it is crucial to distinguish between financially distressed firms and non-distressed ones. While it is probably safe to assume that all firms under bankruptcy proceedings are distressed (rarely will a healthy business file a bankruptcy petition), for the rest of observations we proxy distressed businesses as those whose interest coverage ratio (EBITDAFootnote 25 over interest expenses) is lower than 1.

We then construct several variables. Bankruptcy is a dummy variable that equals 1 if the firm was bankrupt when the data were extracted (2010). To measure the probability of default we use the Altman’s Z-Score (Altman 2000).Footnote 26 As the Orbis database does not contain specific information on mortgage loans, we need to construct a proxy for the proportion of those loans on total debt. The proposed proxy is “Tangibility”, which is computed as the ratio between tangible fixed assets (land, buildings, plant and machinery)Footnote 27 to financial debt, in percentage terms. Since tangible fixed assets are the only assets that can be used as mortgage collateral in Spain, we relate those assets with the debts they may secure. For robustness, we have carried out all this paper’s analyses with an alternative proxy that includes trade credit, namely the ratio between tangible fixed assets to total debt, reaching very similar conclusions, which is not surprising given the high correlation between the two proxies.Footnote 28

As controls, we use a dummy that equals 1 if the firm has limited liability, the firm’s age, the firm’s size—computed as the number of employeesFootnote 29—and industry dummies. According to Berger and Udell (1995) and Petersen and Rajan (1994), firm’s age captures the public reputation of the firm, since they find a negative relationship between firms’ age and interest rate premium charged by banks. Davydenko and Franks (2008) interpret age as a proxy for information asymmetries between a firm and its lenders, since they find negative impact of age on the probability of filing for bankruptcy (vis-à-vis using out-court procedures). Age may also capture coordination costs, as older firms are more likely to maintain multiple bank relationships (Hernández-Cánovas and Köeter-Kant 2008). With respect to firm’s size, small firms may file less for bankruptcy if a substantial proportion of the bankruptcy costs are fixed (Morrison 2008) or if personal insolvency laws are very severe, although the relationship between size and bankruptcy need not be linear because very large firms may prefer to avoid the adverse publicity of a bankruptcy filing. To correct for right skewness we will take logs of age and size in our statistical analyses.

Table 3 shows the descriptive statistics of the variables for the non-distressed firms differentiating by country and size class (micro and non-micro). Spanish firms are smaller and younger than their French and UK counterparts in both groups. More remarkably, their mean levels of Tangibility are substantially higher than those of UK and France in the case of micro firms (Panel A), while only slightly higher in the case of larger firms (Panel B). We obtain similar results in the case of distressed firms (Table 4). Tangibility is also higher in Spain than in the other countries when we disaggregate by industry (Table 5). This evidence supports our hypothesis that Spanish firms have their capital structure biased towards mortgage collateral as a response to the particular insolvency framework they face, as in countries where the bankruptcy system is more effective vis-à-vis mortgage and the law grants less protection to mortgage creditors relative to other secured creditors (France, UK) firms have less tangible fixed assets relative to their financial debt.

Table 3 Descriptive statistics (non-distressed firms)

5 Empirical analyses

Our hypothesis on the low business bankruptcy rates in Spain leads to two testable hypotheses regarding firms’ behaviour ex-ante (i.e., prior default) and ex-post. From the ex-ante perspective, as filing for bankruptcy is very costly, small firms with risky business models will bias their capital structure towards mortgage loans to avoid filing for bankruptcy in the event of default. From the ex-post perspective, holding mortgage debt will reduce the probability of filing for bankruptcy by a small financially distressed firm. These two implications should not occur in the case of either Spanish larger businesses or firms from the other two countries.

5.1 Ex-ante perspective: capital structure and business risk

We run within-country regressions to assess the sign and size of the relationship between the proxy for the percentage of mortgage debt, Tangibility, and the ex-ante probability of default, as measured by the Altman’s Z-score.Footnote 30 We only use, from our sample, non-distressed firms, i.e., those whose interest coverage ratio is equal or greater than 1, which are not under bankruptcy procedures and are active in the market. We split the data into two sub-samples, one for micro firms and another one for non-micro firms.

Table 4 Descriptive statistics (distressed firms)
Table 5 Descriptive statistics of tangibility by industry (distressed and non-distressed firms)

Five different specifications, where Age, Size (both in logs), the limited liability dummy and 488 industry dummiesFootnote 31 are used as controls, are estimated through Tobit regressions. The results are displayed in Tables 6 (micro firms) and 7 (non-micro). The coefficient on the Z-score is negative for Spanish micro firms. As a lower Z-score represents a higher probability of default, the negative sign means that riskier firms rely more on mortgage collateral, suggesting that those firms bias their capital structure towards mortgage loans to avoid filing for bankruptcy if they experienced financial distress. The impact is also economically significant: a unit-decrease in the Z-score would increase Tangibility between two or three percentage points, depending on the specification. By contrast, the coefficient on the Z-score is positive in the rest of cases: firms with risky business models usually have little collateral to pledge, as their main assets are know-how, intellectual property -often unregistered- and firm-specific human capital and machinery.Footnote 32

Table 6 Determinants of tangibility (micro firms)
Table 7 Determinants of tangibility (non-micro firms)

5.2 Ex-post perspective: capital structure and bankruptcy risk

A first descriptive check can be found in Table 8, where we split our sub-sample of distressed firms into bankrupt and non-bankrupt for each size class and each country. In the case of micro firms (panel A), distressed non-bankrupt firms have much higher levels of Tangibility in Spain, while the opposite occurs in France and the UK. Non-micro firms follow the same pattern, but the positive gap between bankrupt and non-bankrupt in Spain is now smaller. As expected, non-bankrupt Spanish firms are smaller and younger for both size classes, while those patterns are not so clear in France and the UK.

Table 8 Descriptive statistics by size and BANKRUPTCY status (unweighted sample)

A more thorough test consists of running within-country regressions to assess the sign and size of the relationship between the proxy for the percentage of mortgage debt, Tangibility, and the probability of filing for bankruptcy by a financially distressed firm in each country, once other determinants are controlled for. We split the data into two sub-samples, one for micro firms and another one for non-micro firms. In analytical terms what we estimate is the following model:

$$\begin{aligned}&P({{ Bankruptcy}_{i}/{ FinancialDistress}})\\&\quad =f({{ Tangibility}_i ,{ Control}1_i ,\ldots ,{ ControlK}_i ,u_i}) \end{aligned}$$

Notice that we do not face a sample selection bias by only keeping the financially distressed firms. Denoting \( S_{i}\) as a selection indicator that equals 1 if the observation is included in the sample and 0 otherwise, and icr the interest coverage ratio: \(S_{i}=1\) if icr\(_{i}<\)1; \(S_{i}=0\) if icr\(_{i}\ge \)1. As long as icr is uncorrelated with \(u\), the unobserved that factors that influence the decision to file for bankruptcy conditional on being in financial distress, our sampling mechanism \(S\)(icr) will be exogenous. As our dependent variable, BANKRUPTCY, is measured in 2010, while the interest coverage ratio icr is measured in 2008, it seems safe to assume that BANKRUPTCY cannot have any influence on icr, implying \(E\)[\(u_{i}/S\)(icr\(_{i})\)] \(=\) 0.

In the case of micro firms, the first set of results is shown in Table 9, which displays OLS regressionsFootnote 33 for the probability of bankruptcy. Five specifications, where Age, Size (both in logs), the limited liability dummy and 14 dummies for industryFootnote 34 are used as controls, are shown for robustness. The table reveals that Tangibility is negatively correlated with the probability of bankruptcy in Spain, while positivelycorrelated in France and the UK.

Table 9 Marginal effects (%) for the probability of bankruptcy in micro firms (OLS)

However, we expect the estimates of Table 9 to be biased due to the endogeneity of capital structure, as explained in the previous section. In other words, as firms’ capital structure and the mechanism used to deal with insolvency are (ex-ante) jointly chosen by firms, we face a simultaneity bias. Moreover, we expect Tangibility to be measured with error because tangible fixed assets are valued at their acquisition (historical) cost, which may differ from their market/collateral values. To solve these problems we use as instrumental variable (IV) the average industry level of Tangibility—where industry is defined at 4 digits of disaggregation, leading to 473 different classes—for each size class (micro and non-micro).Footnote 35 We expect this IV to be uncorrelated with any unobserved determinant of the probability of bankruptcy of a single firm because no firm chooses the asset and capital structure of its industry counterparts. Moreover, there is a positive and sizeable correlation between the IV and the endogenous regressor—as reflected by first-stage regressionsFootnote 36—since companies for the same industries tend to have similar levels of tangibility.

The selected IV estimator is two-stage least squares (2SLS). We prefer not to use IV probit as our main estimator because its consistency relies in some strong assumptions such as conditional normality of the endogenous regressor (Wooldridge 2002) that do not seem to hold in our case. Despite the well-known caveats of the linear probability model (heteroskedasticity, fitted probabilities out of [0, 1]), it requires weaker assumptions and it usually provides good approximations of the marginal effects (Angrist and Pischke 2009).Footnote 37

The results for the estimation via 2SLS are displayed in Table 10. In the regressions for the Spanish subsample, the marginal effects of Tangibility are negative and highly significant, and they are substantially higher than those estimated without instrumenting the regressor. They are also economically significant. A 1 % increase in Tangibility—a small change, as its mean equals 85 % and its standard deviation 101 %—decreases the probability of filing for bankruptcy by a Spanish micro firm by around 0.03 %.Footnote 38 As the unconditional probabilityFootnote 39 of those firms is 3.6 %, the estimated semielasticity is 0.83 %, which is a sizeable effect. By contrast, the marginal effects of Tangibility are positive and significant both in France and the UK. This latter result is consistent with the results of Davydenko and Franks (2008) on their study of French, UK and German firms that defaulted on their bank debt. They find that higher levels of collateral imply a significantly higher incidence of bankruptcies and a somewhat higher probability of liquidation, suggesting that banks use formal bankruptcy procedures to force a sale of collateral in those countries. This is not the Spanish case, as there is an alternative insolvency procedure, a mortgage foreclosure, through which collateral can be more efficiently liquidated.

With respect to the control variables, size has a positive impact in the probability of bankruptcy in the three countries, suggesting that the fixed costs of bankruptcy procedures deter very small firms from using them, as argued by Morrison (2008). Age has a negative effect in the case of UK, suggesting that lower information asymmetries and higher reputation concerns incentivise older firms to avoid bankruptcy. By contrast, it has a positive impact in the Spanish and French subsamples, probably capturing the fact that, as older firms have more lenders, higher coordination costs reduce the chances of non-bankruptcy procedures. Limited liability has a positive effect in three countries—although not significant at 10 % in the UKFootnote 40—consistent with the idea that, when the debtor may lose part of its personal wealth, she has fewer incentives to file for bankruptcy, even when some partial discharge—as in France and in the UK—may be granted.

The case of non-micro firms is analysed in Tables 11 and 12. As a benchmark, Table 11 shows the (biased) OLS estimates for the three countries, which reveals the same patterns as for micro firms: negative correlations in Spain and positive correlations in France and the UK.

Table 11 Marginal effects (%) for the probability of bankruptcy in non-micro firms (OLS)

By contrast, the consistent IV estimates in Table 12 show that Tangibility has a positive impact on the probability of bankruptcy of Spanish non-micro firms in 3 out of the 4 specifications and it is not significant in specification (4), suggesting that it is not a robust determinant. The case of French and UK larger firms is similar to the one for micro: Tangibility has a robust positive impact in the probability of bankruptcy.

With respect to the control variables, size has a positive impact in the probability of bankruptcy in Spain—as it was the case in the subsample of micro firms—but a negative one in France and the UK. A possible interpretation is that the fixed costs of bankruptcy proceedings deter small and very small firms from using them but, in the case of quite large firms, other factors, such as the reputational loss of managers, make filing for bankruptcy less appealing. As in the case of micro firms, age has a positive effect in the Spanish subsample, but a negative one in the UK and no robust impact in France.

Table 12 Marginal effects (%) for the probability of bankruptcy in non-micro firms (2SLS)

5.3 Robustness analysis of the ex-post perspective: private workouts and subsample of firm exits

In our sample of distressed firms we have two types: bankrupt and non-bankrupt. The former consists of firms under bankruptcy proceedings (i.e., still operating in the market) and firms that have been liquidated after a bankruptcy procedure (i.e., they have exited the market). The latter consists of companies that are still operating the market under financial distress and companies that exited the market while they were financially distressed. An alternative explanation for the negative impact of Tangibility on the probability of filing for bankruptcy by a Spanish micro firm is that firms with high levels of tangible fixed assets relative to their levels of financial debt still have assets they may pledge as mortgage collateral to get new loans or refinance their current ones. In that case, they would avoid bankruptcy by surviving and staying in the market thanks to a private workout, rather than exiting via a foreclosure.

We have two objections to this view: one is logical; the other one is based on empirical evidence. First, in France, due the high dilution that most secured creditors suffer inside bankruptcy (see Sect. 3.2), they may be willing to make debt concessions in a private workout to deter the debtor from filing for bankruptcy. By contrast, secured creditors are unlikely to be held up by a debtor in Spain because, while there is an automatic stay over the enforcement of some secured credit in bankruptcy, it is very limited in time and uncertain in scope.Footnote 41 In fact, the LLSVFootnote 42 index (La Porta et al. 1998, updated by Djankov et al. (2007)), which measures the protection of secured creditors in bankruptcy in a scale from 0 (lowest protection) to 4 (highest), assigns 3 to Spain while 0 to France. Hence it seems implausible to explain the large differences in the bankruptcy rates of small firms in Spain and France in terms of the relative incidence of private workouts. Since those rates are low in Spain and high in France, workouts should be abundant in Spain and rare in France, while our reasoning suggests the opposite.

Second, from the empirical point of view, we address this potential criticism by keeping in the sample only those firms that exited the market. We construct a new dependent variable, bankruptcy2, which takes the value 1 if the firm left the market after a bankruptcy procedure and 0 otherwise. Our main results are the same: Tangibility has a negative and significant impact on the probability of being bankrupt in the case of Spanish micro firms (see Table 14).Footnote 43 This effect is not present in Spanish larger firms, since the correlation is not different from zero in our OLS estimates (see Table 15) and the causal impact is not robust to several specifications in our IV estimates (see Table 16). By contrast, Tangibility has a positive impact on the probability of leaving the market after bankruptcy in the case of French firms of both size classes, while there is no effect in the case of British firms.

Table 13 Marginal effects (%) for the probability of bankruptcy EXIT in micro firms (OLS)
Table 14 Marginal effects (%) for the probability of bankruptcy EXIT in micro firms (2SLS)
Table 15 Marginal effects (%) for the probability of bankruptcy EXIT in non-micro firms (OLS)
Table 16 Marginal effects (%) for the probability of bankruptcy EXIT in non-micro firms (2SLS)

6 Conclusions

Spain had, before the current economic crisis, one of the world’s lowest business bankruptcy rates, i.e., the number of business bankruptcy filings divided by the number of business exits. Only the crisis has modestly increased the number of bankruptcies, but the Spanish bankruptcy rate is still one of the lowest in the world. This fact is driven by the behaviour of micro firms—the majority of Spanish firms—which rarely file for bankruptcy when dealing with financial distress.

This paper presents and tests a hypothesis that attempts to explain this empirical finding. According to this hypothesis, filing for bankruptcy in Spain is very costly for both small firms and their creditors. Due to this, the capital structure of micro firms is biased towards mortgage loans (i.e., loans secured on land and buildings). Having this capital structure allows them to avoid bankruptcy by carrying out debt enforcement via mortgage foreclosures, which are cheaper procedures than bankruptcy, in case of financial distress.

To test this hypothesis our identification strategy relies on cross-country comparisons. Specifically, we compare the observed choices (choice of capital structure, choice between bankruptcy and mortgage) of Spanish firms with those of firms from countries where their bankruptcy systems are more efficient and their laws do not incentivise them to bias their capital structure towards mortgage loans. France and the UK are chosen as the comparison group because their bankruptcy rates are much higher than the Spanish ones and because of the specific features of their insolvency frameworks.

Our findings corroborate the proposed hypothesis. First, there is a positive and strong correlation between the ex-ante probability of default and the ratio of tangible fixed assets (the assets that can be pledged as mortgage collateral) to financial debt in the case of Spanish micro firms, suggesting that firms with risky business models bias their capital structure towards mortgage loans to avoid filing for bankruptcy in the event of default. Second, a higher proportion of tangible fixed assets over financial debt significantly decrease the probability of being in bankruptcy among Spanish micro firms in financial distress. By contrast, these two relations do not hold either for Spanish larger businesses or for firms from the other two countries.

We must stress the importance of the research question. Bankruptcy procedures and mortgage foreclosures are not perfect substitutes, and the underutilization of one of them—reflected in low bankruptcy rates—may lead to efficiency losses and lower welfare (García-Posada 2013). The reason is that the mortgage system is not well suited for some firms, which need to bias their asset structure to have enough collateral, with the ensuing productive inefficiencies. Those firms would be better off if they had access to a bankruptcy system that worked relatively well. If the absence of a well-functioning bankruptcy system for those firms also reduces their growth opportunities—e.g., by hampering access to unsecured lending such as venture capital—then the current insolvency framework may help explain the firm size distribution and the low aggregate productivity of the Spanish economy. This is consistent with the evidence of Fabbri (2010) in Spain, who finds that lengthy bankruptcy procedures decrease firm size and raise funding costs and with that of Ponticelli (2012) in Brazil, who shows that congestion in bankruptcy courts substantially reduces firm-level investment and productivity.

This paper is a first step towards understanding how agents respond to the Spanish insolvency framework and their implications for the real economy. Further work is required in two directions. First, better data without the limitations of our sample, especially regarding information on mortgage loans and the alternative procedures to formal bankruptcy, could be collected. Second, the impact of the low efficiency of the bankruptcy system vis-à-vis mortgage foreclosures on the performance of Spanish firms should be empirically analyzed.