The effect of corporate liquidity and investor protection on the behaviour of distressed equity in Europe

UPPSALA UNIVERSITET

The effect of corporate liquidity and investor protection on the behaviour of distressed equity in Europe

DD MSc International Financial Management Department of Business Studies

ABSTRACT

This study examines the effect of corporate liquidity and investor protection on the relation between financial distress and equity returns using a European sample over the 2002-2016 period. The results show that returns are hump-shaped and decreasing for increasing default risk. This can be rationalized by corporate liquidity indicating that higher cash holdings decrease liquidity risk. Moreover, firms in countries with high investor protection exhibit a more severe decrease of returns when default risk increases relative to firms in countries with low investor protection. This is because of the legal system that allows investors to renegotiate upon distress and to more accurately price equities.

JEL classification: G12, G32, G33

Key words: financial distress, equity returns, cash holdings, investor protection

Name: Anneke Damhuis Student number: 930810-T220 Supervisor: Dr. V. Purice

Second supervisor: Dr. M.M. Kramer

Date: 12 January 2018

2

1. Introduction

Each year, 200,000 firms go bankrupt in the European Union (European Commission, 2016).

Hence, the new proposal of the European Commission aims to restructure business and investor rights favouring prevention rather than failure. However, besides new procedures and rules resulting in economic gains through governmental intervention, firms and investors should suitably assess financial distress risk. Managers need to allow for financial distress risk when establishing business policies. Moreover, the market aims to effectively price (distressed) equities.

To this extent, literature has extensively reviewed the debate about default risk and equity returns (see, e.g., Dichev, 1998; Griffin and Lemmon, 2002; Garlappi, Shu, and Yan, 2008).

However, existing empirical evidence provides a complicated view eluding a unifying and coherent explanation (Garlappi and Yan, 2011). Contrary to the general intuition, that firms facing higher default risk earn higher returns, distressed equities are commonly found to earn lower returns, widely known as the distress anomaly. While some explanations have been considered, evidence on the distress anomaly in the U.S. market is currently characterized by disagreement. Moreover, surprisingly few studies examine the distress puzzle by using non- U.S. data. Therefore, Gao, Parsons, and Shen (2015) argue that it is essential to shift the focus to new non-U.S. evidence. This study goes beyond research on U.S. data and examines the performance of distressed equity returns in Europe.

An important research direction for non-U.S. firms is to further examine the drivers of the

distress anomaly, and whether this enables us to understand the behaviour of distressed equities

outside the U.S. (Aretz, Florackis, and Kostakis, 2017). A recent study by Medhat (2014) sheds

new light on prior findings by showing that corporate liquidity is of importance to explain the

relation between equity returns and financial distress for a U.S. sample. His research is among

the first to consider corporate liquidity influencing this relation. The author finds that returns

are hump-shaped and decreasing for higher default risk. The rationale behind this finding is that

firms with default risk can decrease liquidity risk by holding precautionary cash. This study is

inspired by the research performed by Medhat (2014), but focuses on a European sample with

a deepened understanding of the theory involved in this area. As such, this study analyses the

effect of cash holdings on the equity risk of financially distressed European firms. Thereby

testing whether this has a direct impact and accounts for the observed variation in the relation

3

between default risk and equity returns, over and above the well-documented effects of size and value premiums.

The diverse European sample allows this study to relate equity returns and default risk to country-level characteristics that might influence the distress puzzle. Within the international dimension, several papers elaborate upon the cross-country effect of investor protection, which is identified as the quality of the regulatory and legal protection of a firm’s shareholders (La Porta, Lopez-de-Silanes, Shleifer, and Vishny, 1997). Significant differences exist between the U.S. and European countries in terms of financial and legal systems, and hence the protection of investors and their rights (Doupnik and Perera, 2011). These differences might lead to a deviation in pricing distressed equity and cash holding policies between European and U.S.

firms. As the analysis of Garlappi et al. (2008) highlights, the role of investor advantage, defined as the combination of investors’ bargaining power and the efficiency gained through bargaining, is essential in the determination of the relation between default risk and equity returns. Hence, this study examines whether country-level characteristics drive the default premium and if it follows that this is differently from U.S. results.

The main findings are as follows. The distress anomaly is revisited for the European sample consisting of 714 unique firms resulting in a hump-shaped and decreasing relation between equity returns and default risk. Moreover, corporate liquidity and investor protection significantly affect this relation. The findings supporting the propositions are robust. These results extend the literature on financial distress by illustrating the economic and statistical significance of corporate liquidity and country-level institutional differences in explaining the distress anomaly for European firms. Moreover, the results have several practical implications including business credit evaluations, investment guidelines, and accurately pricing financial assets when assessing default risk. Further, managers must consider the optimal level of cash to minimize liquidity risk and the strength and quality of investor rights in corporate (risk- taking) decisions.

The remainder of the paper proceeds as follows. The next section provides the main theoretical background and the opposing views and results. Thereafter, the data and methodology are described. In the fourth section, the results of both multivariate regressions and portfolio analysis are presented, followed by an extensive elaboration upon the findings.

Finally, the conclusions and limitations are given.

1 Aretz, et al. (2017) find that the magnitude of the default risk premium declines when adjusting for size and value premiums, suggesting that these features are related to default risk. This is in line with the conjectures of previous studies (see, e.g., Chan and Chen, 1991; Fama and French, 1996; and Vassalou and Xing, 2004).

4

2. Literature review

2.1 Financial distress and equity returns

A firm is financially distressed

when its value of assets is not sufficient to meet contractual interest or principal on debt obligations. These firms are assigned to have higher default risk.

The effect of default risk on equity returns is not clear because equity holders are the residual claimants on the cash flows of a firm (Vassalou and Xing, 2004). While some papers argue that higher default risk results in higher equity risk premiums, others find evidence for the distress puzzle revealing that financial distress is associated with anomalously low equity returns. Since this is one of the most ambiguous asset pricing irregularities, the subject has gained considerable attention in both theoretical and empirical research (Eisdorfer, Goval, and Zhdanov, 2014).

Chan and Chen (1991) and Fama and French (1992, 1996) were among the first to analyse the effect of financial distress on equity returns. They suggest that financial distress is a possible explanation for some of the anomalies in the cross-section of equity returns. The rationale behind these studies rests on the conjecture that investors require a higher premium for holding equities that are exposed to a higher probability of default. Chan and Chen (1991) argue that the size premium is mainly driven by firms with low market value and high leverage to justify distress risk, as these firms are more sensitive to adverse economic fluctuations. Fama and French (1992) relate the book-to-market effect to default risk as well. In a later study, Fama and French (1996) suggest that, if bankruptcy events are correlated across firms, the relative default risk of a firm might be a state variable which eventually affects asset prices in the cross-section.

More recently, scholars argue that default risk is positively priced in the market and is profoundly associated with size and value effects (see, e.g., Vassalou and Xing, 2004; Aretz et al., 2017). In this regard, distress risk is found to explain size and value effects that are anomalies in the standard capital asset pricing model. Though these papers find that a firm’s default risk is positively priced in the market, existing empirical literature, by using various measures of default risk, has not grounded consistent evidence to confirm above conjecture.

2 In this study, the terms default and financial distress are used interchangeably.

5

Moreover, several studies show that distressed firms have actually lower, not higher, equity returns.

A common interpretation of the distress anomaly is market mispricing, that is, investors are not able to fully assess the perspective of distressed firms and hence do not demand a default risk premium (Garlappi and Yan, 2008). Another clarification proposed by Campbell et al.

(2008) identifies that investors might prefer positively skewed equities, and hence are willing to hold equities with high default probabilities regardless their low returns. Furthermore, by decomposing default risk in systematic and idiosyncratic risk, scholars indicate that the systematic component drives the significant default risk premium.

An explanation that helps reconcile the conflicting interpretations of the effect of default risk on equity returns, which does not rely on capital structure, is the economic mechanism of investors’ bargaining power. Garlappi et al. (2008) recognize that equity returns are humped and decreasing in default probability due to the possibility of investor recovery upon distress.

Extending these findings, Garlappi and Yan (2011) theoretically show that investor recovery upon distress implies the non-monotonic and decreasing relation between financial distress and conditional betas, and empirically show this by using time varying betas. Moreover, Favara, Schroth, and Valta (2012) find evidence that a firm’s equity risk is reduced when the perspective of debt renegotiations is favourable for investors. Thus, several scholars identify that returns are hump-shaped and decreasing when default risk increases. These papers rest on the conjecture that investors do not require a higher return for holding distressed equities. To test the existence of the distress anomaly for the constituted European sample, this thesis revisits the empirical relation between financial distress and equity returns, resulting in the following hypothesis:

Hypothesis 1: Equity returns are hump-shaped and decreasing as default probability increases.

3 Dichev (1998) finds a negative relation between returns and default probability using both Altman’s (1968) Z- score and Ohlson’s (1980) O-score. Griffin and Lemmon (2002) document that this finding is stronger for firms with low book-to-market ratios by using Ohlson’s (1980) model. This finding is more recently confirmed by Campbell, Hilscher and Szilagyi (2008). Friewald, Wagner and Zechner (2014) show that firms with the highest default probability earn the lowest returns using Merton’s (1974) model. Gao et al. (2015) find a significant negative relation between financial distress and equity returns using Moody’s KMV, more specifically they find that the distress anomaly is more prevalent in North America and Europe, in conformity with Eisdorfer et al. (2014) who use Merton’s (1974) model.

4 Aretz et al. (2017) find conclusive evidence for the distress risk premium, consistent with earlier studies (see, e.g., Bonfim, 2009; Qu, 2008; Friewald et al., 2014; Hilscher and Wilson, 2016).

6

2.2 Corporate liquidity

Medhat (2014) complements the distress debate by separating default risk in solvency and liquidity risk. The relation between default risk and equity returns is explained by the effect of corporate liquidity for a U.S. sample. As U.S. firms are to a greater extent holding levels of cash reserves, the field of studies related to corporate cash holdings has increased considerably.

Ample of empirical evidence suggests that the precautionary motive is of most importance for this appearance and for their considerable growth since the 1980s.

At the end of 2016, firms in the S&P 500 index (excluding financial firms) held cash and cash equivalents of $1.54 trillion, just about 13% of total assets.

Such sizeable cash holdings are not specific for U.S. firms;

substantial liquidity seems to be a global phenomenon (Seifert and Gonenc, 2016). The sample in this study, consisting of 18 European countries, has average cash holdings amounting to 10.4% of total assets.

The extensive literature on cash holdings distinguishes between three main theories. The first theory is widely known as the trade-off theory arguing the existence of an optimal level of cash and cash equivalents. Therefore, managers need to make a trade-off between the marginal benefits and marginal costs of holding an extra dollar of cash (Myers, 1984). Fundamental thoughts behind this model are the transaction cost theories of Keynes (1936) and Baumol (1952). Besides the transaction motive, cash holdings can be rationalized by the precautionary motive (Opler et al., 1999). The rationale behind the precautionary motive argues that serious amounts of cash are hold by firms to prevent prejudicial shocks (Bates et al., 2009). The second theory, known as the pecking order theory, of Myers and Maljuf (1984) argues that cash holdings are established by the investment and capital expenditure policies of firms. For individual financing decisions managers prefer internal financing, such as retained earnings, to outside funding and debt is preferred over new equity when outside financing is required.

Consequently, firms will hold a certain level of cash to preserve their policies (Dittmar, Mahrt- Smith, and Servaes, 2003; Ferreira and Vilela, 2004). Third, the free cash flow theory of Jensen (1986) is influential in the debate concerning cash flows, touching upon the agency theory.

Agency issues are naturally a determinant of cash holdings and cash valuation. Managers who are not incentivized by shareholder maximization are inclined to act in their own interest. These managers are likely to maintain above average levels of cash (Pinkowitz, Stulz and Williamson, 2006). Another possibility prevails upon the irrelevance of financing policies. The seminal work

5 See, e.g., Opler, Pinkowitz, Stulz, and Williamson (1999); Ferreira and Vilela (2004); Bates, Kahle, and Stulz (2009); Acharya, Davydenko, and Strebulaev (2012); Davydenko (2013), and references therein.

6 www.yardeni.com/pub/spxratios.pdf, accessed on 14 October 2017.

7

of Modigliani and Miller (1958) argues that the choice of financing has no effect on firm value or on the availability or cost of capital. Modigliani and Miller (1958) state that cash holdings and capital structure are irrelevant based on the perfect market argument. Although this theory is widely accepted, financing clearly matters. Perfect markets do not exist and hence the theory is highly hypothetical. Therefore, there are several reasons, enclosing aforementioned main theories, why it is beneficial for firms to hold substantial amounts of cash.

More recently, empirical studies have enhanced the literature by analysing cash holdings in light of default risk. Campbell et al. (2008) argue that firms with higher cash holdings have more liquid assets available to meet interest payments, and thus are able to delay or even prevent default. They empirically show that, on average, firms filing for bankruptcy do not hold considerably less cash. Davydenko (2013) elaborates on default by distinguishing between insolvency and illiquidity – as well as economic versus financial distress. He finds that the market value of assets is of most importance in explaining the timing of default rather than illiquidity. Conversely, Acharya et al. (2012) show that firms with higher cash levels are less likely to default in the short term. The results suggest that firms holding higher cash balances in their asset and investment portfolio should face less risk. Hence, the authors suggest that precautionary savings are central to understanding the effects of cash holdings on credit risk.

Medhat (2014) confirms the theory that the precautionary motive is of most importance for substantial cash holdings, touching upon the trade-off theory. When a firm is in default, managers have to make a trade-off between holding precautionary cash to cover a coupon (not defaulting on liquidity) or acting in the interest of shareholders and voluntarily defaulting (due to a lack of insolvency) when not holding enough assets to outweigh future coupons.

Subsequently, a firm’s optimal policy regarding cash holdings depends on the following two approaches: an extra dollar invested in dividends increases shareholders’ equity while an extra dollar invested in internal cash holdings increases the probability of survival. He shows that a hump-shaped relation exists between expected returns and financial distress risk when precautionary cash is used to offset liquidity risk, by arguing that firms want to manage their cash to minimize liquidity risk by maintaining some of their earnings as precautionary cash.

More particular, an insolvent but liquid firm is empirically found to have relatively large cash

holdings. By eliminating liquidity risk, a firm will not default on illiquidity because it has

enough means to cover a coupon. So, even though a firm in financial distress is more likely to

default due to a lack of solvency, investors face lower risk when liquidity risk is eliminated

resulting in lower, not higher, equity returns. Moreover, a solvent firm is expected to have

higher and upward sloping returns as its asset value is mostly based on expected earnings.

8

Therefore, equity returns are relatively sensitive to earnings risk. A financially distressed firm, or a firm close to default, is expected to have lower and downward sloping returns because the firms’ asset value is mostly based on its level of cash holdings. This is because cash holdings are especially valuable when a firm is in financial distress since they are able to absorb shortfalls in operating profits (Dittmar, 2008). Hence, the equity value of a firm is relatively insensitive to earnings risk (Medhat, 2014). Based on this reasoning, the following hypothesis is formulated:

Hypothesis 2: Cash holdings affect the relation between financial distress and equity returns.

2.3 Investor protection

Another factor that might explain the behaviour of financially distressed equities is the quality of investor protection. A country’s legal origin is of considerable importance concerning a country’s strategy for protecting investors (Shleifer and Vishny, 1997). Investor rights are founded because shareholders are above all last claimants on a firm’s assets, and therefore have different rights than debtors. In this light of debtor rights, debt contracts need to be enforced.

Hence, legal systems are originated to protect lenders from borrowers’ default. As a result, many countries rely on courts to enforce debt contracts, generally by way of financial distress and default procedures (Djankov, Hart, McLiesh, and Shleifer, 2008). Countries in the European Union follow the ‘Regulation on Insolvency Proceedings’, 2002. In 2014, these procedures are updated by the European Commission to ‘A New Approach to Business Failure and Insolvency’ (Hillier, Ross, Westerfield, Jaffe, and Jordan, 2016). Notwithstanding these procedures, differences in investor protection exist across countries with different legal origin.

Among the first to elaborate on investor rights were la Porta et al. (1997), by focusing on

country specific legal origin accompanying the protection of investors and elaborating on the

prevalence of these rules in a considerable amount of countries. Naturally, investors are best

protected by common law countries that have large developed equity markets, like the United

States and the United Kingdom. Contrary, code law countries tend to rely more extensively on

their banking systems. These differences in financial systems can be explained by differences

in investor protection coming from legal origins and associated quality of enforcement (La

Porta et al., 1997). For instance, the German and French system require collective procedures

and court supervision, while the U.K. system requires neither. Further, the U.K. law provides

contracting parties with a relatively free implementation of the stipulated procedure

(Davydenko and Franks, 2008). Moreover, bankruptcy filings occur more frequently in

9

countries with bankruptcy regimes favouring managers and creditors rather than investors. If a country’s bankruptcy system prevents renegotiations, investors gain less from strategic default.

Conversely, when bankruptcy codes favour renegotiations, investors have incentives to renegotiate debt terms upon default (Favara et al., 2012).

Hence, some scholars conjecture that equities of distressed firms might be safer because of the possibility of deviations or renegotiations by the priority rule. Literature provides significant evidence for cross-country variations with regard to investor protection rights affecting the pricing of default risk. Garlappi et al. (2008) show that the appraisal of default risk should allow for the potential recovery for investors. They empirically find that the prospects of debt renegotiation favouring investors decreases a firm’s equity risk, and hence lowers expected returns. More specifically, the trade-off between the risk of default and the likelihood of bargaining gains in renegotiation results in a hump-shaped relation between equity returns and default risk. Extending these findings, Garlappi and Yan (2011) empirically explain the relation between default risk and conditional betas by considering shareholder recovery upon distress. Furthermore, as highlighted by Davydenko and Franks (2008), Favara et al. (2012), and Aretz et al. (2017), the default risk premium is expected to be lower in countries with the legal system allowing debt renegotiations and favouring investors’ bargaining power. However, Eisdorfer et al. (2014) identify only weak support for this theory. Additionally, Gao et al. (2015) fail to find a relation between investor protection and the default risk premium. Nevertheless, Eisdorfer et al. (2014) find evidence for the stock market development hypothesis. Common law countries have considerably more valuable stock markets, and successful stock markets mandate that investors obtain the information they need and the power to act (Davydenko and Franks, 2008). A weaker stock market, indicated by lower investor protection, is characterized by higher information asymmetry. Therefore, it is harder for investors to properly assess the true probability of default. Hence, the misevaluation of distressed equity is more pronounced in less developed equity markets.

All in all, despite the pronounced differences in investor protection among countries, the

literature finds inconclusive results when analysing the default probability. Notwithstanding,

the studies finding weak or even no evidence rely on an international sample including

emerging markets, whereas this study rests on a considerable smaller European sample. As both

the U.S. and Europe can be seen as developed markets, this thesis follows the reasoning of

Garlappi et al. (2008) to indicate that strong investor rights explain reduced equity risk

premiums. So, it can be expected that the default risk premium is lower in countries with the

10

legal system allowing debt renegotiations and favouring investors’ bargaining power.

Therefore, based on the theoretical founding, the following hypothesis is developed:

Hypothesis 3: Firms in countries with high investor protection earn the lowest returns for distressed equity.

Although some characteristics in investor rights might be an endogenous response of the legal system to variations in the economy, for the empirical analysis it is assumed that investor rights are predetermined, in line with Acharya et al. (2011).

3. Data and methodology

3.1 Data

To empirically test the formulated hypotheses a data sample is extracted from Thomson Reuters DataStream. Financial as well as accounting data is acquired through DataStream for the companies constituting the ASSET4 Europe index at an annual frequency.

When necessary, the accounting items in local currency are converted into euros using the Thomson Reuters DataStream conversion factors. The accounting variables are computed from DataStream as of the fiscal year-end of a given year t. Utilities and financial firms are excluded from the sample (SIC codes 4900-4999 and 6000-6999). Moreover, firms that went bankrupt during the sample period are also excluded. The final sample consists of 10,710 firm-year observations for 714 unique firms in 18 European countries over the period January 2002 to December 2016. All financial variables are winsorized at the 1

and 99

percentile to alleviate the influence of possible reporting errors and statistical outliers.

3.2 Sample distribution

Table 1 exhibits the sample distribution of all observations for Altman’s Z-score by year and by country. Altman’s Z-score denotes the distance-to-default, so a higher Z indicates a healthier firm.

This table indicates that most observations are available for the United Kingdom, revealing that most firms in the sample are established in the U.K. (33%). Examining Altman’s Z-score by year, the sample size increases during the period 2002-2016 as a result of more

7 Thomson Reuters Database has only yearly accounting variables. Compustat presents accounting variables at a quarterly frequency. Notwithstanding, less information, and hence less quarterly accounting data, is available for non-U.S. firms in Compustat. Furthermore, the income statement variable Earnings Before Interest and Taxes is required to define Altman’s Z-score as a measure of financial distress. In both Compustat and DataStream this variable is only available at a yearly frequency.

8 See section 3.4 for the precise measurement of Altman’s Z-score.

11

available observations. The number of firms in financial distress is moderately highest in 2009 (2.772), possibly reflecting the impact of the financial crisis. The findings of Aretz et al. (2017), who look at the number of bankruptcy filings for an international sample, also exhibit a peak in failures in the aftermath of the global financial crisis. In 2003, firms are also found to have high values of financial distress (2.947). This may be due to the dot-com bubble where many internet-based companies defaulted from 2000 to 2002. Moreover, firms are found most healthy in 2006 and 2007 with a Z-score of 3.736 and 3.891 respectively. With reference to the summary statistics of Z by country, it can be immediately observed that the distance-to-default is reasonably highest for firms located in Denmark (5.223) and Switzerland (4.488) suggesting that firms in these countries are the healthiest. A possible reason is the high stock market capitalization and bank development in Switzerland, and high bond market development in Denmark (Oxelheim, 2006). By contrast, the most distressed firms are located in Austria (2.361), Portugal (2.222), and France (2.372).

Table 1 Observations by year and country for distance-to-default. This table presents the observations by year and country based on 10,710 firm-year observations for Z as a measure of financial distress. Z refers to the distance-to-default.

Data for the sample countries is collected for 714 unique European firms over the 2002-2016 period.

Year Obs. Mean Std. Dev. Country Obs. Mean Std. Dev.

2002 487 3.374 2.965 Austria 150 2.361 1.306

2003 501 2.947 2.230 Belgium 182 3.351 3.579

2004 517 3.444 2.647 Denmark 288 5.223 3.951

2005 553 3.426 2.648 Finland 323 3.186 1.606

2006 577 3.736 2.850 France 1,008 2.372 2.058

2007 605 3.891 2.946 Germany 954 3.054 2.442

2008 625 3.456 2.808 Greece 154 2.949 3.207

2009 626 2.722 2.265 Ireland 201 3.209 2.868

2010 629 3.199 2.632 Italy 271 2.526 1.935

2011 640 3.442 3.007 Luxembourg 46 4.026 3.552

2012 650 3.223 2.737 Norway 203 2.960 2.021

2013 657 3.376 2.886 Poland 158 3.409 2.287

2014 668 3.495 3.009 Portugal 99 2.222 2.548

2015 694 3.418 3.084 Spain 356 3.291 3.480

2016 703 3.338 2.965 Sweden 709 4.054 3.710

Switzerland 628 4.488 3.207

the Netherlands 378 2.623 1.701

United Kingdom 3,024 3.559 2.673

Total 9,132 3.367 2.812 Total 9,132 3.367 2.812

12

3.3 Descriptive statistics

Descriptive statistics of the full sample and subsamples are presented in table 3. The subsamples are based on countries with high investor protection against countries where investor rights are not well protected. Deviations between firms in high and low investor protection countries can be compared by this disjunction.

To measure investor rights at the country level, the updated data and rankings from the survey by Djankov et al. (2008) is used. Table 2 depicts the classification of investor protection for each country included in the sample. The mean (median) in the sample is 3.22 (3.25) regarding the anti-director rights index, with higher values indicating higher investor protection. The mean (median) in the sample is 0.41 (0.38) for the anti-self-dealing index, with higher values indicating stronger investor protection. A country is estimated to have high investor protection when both the anti-director rights index and the anti- self-dealing index are classified as high. Else, countries are indicated as countries with low investor protection.

9 Scholars illustrate that quality of investor protection significantly affects financial distress (see, e.g., Garlappi et al., 2008; Garlappi and Yan, 2011; Favara et al., 2012; Aretz et al., 2017) and cash holdings (see, e.g., Dittmar et al., 2003; McLean, Zhang, and Zhao, 2012; Seifert and Gonenc, 2016).

Table 2 Investor protection per country. This table reports the quality of investor protection per country based on the anti-director-rights index and anti-self-dealing index obtained from Djankov et al. (2008). Total investor protection of a country is defined as high when both indices are classified as high. Otherwise, countries are classified as having low investor protection.

Anti-director rights index Anti-self-dealing index Total investor protection

Country Value Classification Value Classification Classification

Austria 2.5 Low 0.21 Low Low

Belgium 3 Low 0.54 High Low

Denmark 4 High 0.46 High High

Finland 3.5 High 0.46 High High

France 3.5 High 0.38 Low Low

Germany 3.5 High 0.28 Low Low

Greece 2 Low 0.22 Low Low

Ireland 5 High 0.79 High High

Italy 2 Low 0.42 High Low

Luxembourg 2 Low 0.28 Low Low

Norway 3.5 High 0.42 High High

Poland 2 Low 0.29 Low Low

Portugal 2.5 Low 0.44 High Low

Spain 5 Low 0.37 Low Low

Sweden 3.5 High 0.33 Low Low

Switzerland 3 Low 0.27 Low Low

The Netherlands 2.5 Low 0.20 Low Low

United Kingdom 5 High 0.95 High High

Mean 3.22 0.41

Median 3.25 0.38

13

Descriptive statistics for the test variables are shown in Table 3. A two-sample t-test is conducted to test whether there are statistically significant differences in the means for the two subsamples, see Appendix A2. The total sample encompasses 714 firms (Panel A), the subsamples deviated by high versus low investor protection include 460 (Panel B) and 254 (Panel C) firms respectively. Returns are higher for firms in countries where investors are well protected with a mean of 0.088 against a mean of 0.067 in low investor protection countries.

Moreover, Z is higher for firms in countries with high investor protection (3.601), suggesting less distressed firms in the latter. This implies that firms in countries where investors are well protected on average face less default risk. Cash is found to be higher in countries with low investor protection, in line with literature that states that cash holdings tend to be higher in countries with lower investor protection (Seifert and Gonenc, 2016). Book-to-market and market value are higher for firms in low investor protection countries.

Table 3 Descriptive statistics. This table reports the descriptive statistics of the firm-level variables used in this study.

The subsamples are based on quality of investor protection. The full sample consists of 10,710 firm-year observations for 714 unique European firms over the 2002-2016 period. Variable definitions are presented in Appendix A1. N denotes the number of observations. Financial variables are winsorized at the 1st and 99th percentiles.

Panel A: Full sample (N=714 firms)

Variables Observations Mean Std. Dev. Median Min Max

Returns 9,427 0.076 0.415 0.127 -1.355 1.077

Z 9,132 3.367 2.812 2.635 -0.397 17.775

Cash holdings 10,074 0.125 0.124 0.088 0.000 0.999

Market value 9,214 6.317 0.680 6.293 4.618 7.987

Book-to-market 9,217 -0.749 0.794 -0.721 -3.116 1.296

Panel B: High investor protection (N=460 firms)

Variables Observations Mean Std. Dev. Median Min Max

Returns 4,145 0.088 0.431 0.137 -1.355 1.077

Z 4,039 3.601 2.740 2.925 -0.397 17.775

Cash holdings 4,418 0.115 0.121 0.074 0.000 0.957

Market value 4,057 6.163 0.663 6.107 4.618 7.987

Book-to-market 4,025 -0.814 0.840 -0.785 -3.116 1.296

Panel C: Low investor protection (N=254 firms)

Variables Observations Mean Std. Dev. Median Min Max

Returns 5,282 0.067 0.401 0.118 -1.355 1.077

Z 5,093 3.182 2.854 2.436 -0.397 17.775

Cash holdings 5,656 0.132 0.127 0.098 0.000 1.000

Market value 5,157 6.438 0.669 6.444 4.618 7.987

Book-to-market 5,192 -0.699 0.753 -0.677 -3.116 1.297

3.4 Construction of variables

The following explains the measures of the variables. The Appendix to this thesis provides an

overview of (the construction of) the variables in detail (Appendix A1). The dependent variable,

14

equity returns (RET), is a common measure in existing literature when examining financial distress (see, e.g., Dichev, 1998; Campbell et al., 2008; Favara et al. 2013; Aretz et al. 2017).

The returns are accumulated into yearly log-returns to match the yearly available accounting data.

To study the relation between financial distress and equity returns an accounting-based measure of default probability is used based on the seminal contribution of Altman (1968).

Bankruptcy risk appears to be a natural measure of a firm’s financial distress (Dichev, 1998).

Garlappi et al. (2008) and Medhat (2014) rely on Moody’s KMV EDF as a measure of financial distress. However, Hilscher and Wilson (2016) document that credit ratings are relatively inaccurate measures of bankruptcy probability. Hence the measure for financial distress is derived from Altman’s (1968) Z-model, which is probably the most popular model of bankruptcy prediction. As a result, the model is extensively used in practice and empirical research (see, e.g., Dichev, 1998; Ferguson and Shockley, 2003; Hillegeist, Keating, Cram, and Lundstedt, 2004; Acharya et al. 2012). The independent variable Z is a measure of financial strength (higher Z means higher distance-to-default) and is calculated as follows:

Z = 1.2(working capital/total assets) + 1.4(retained earnings/total assets) +

3.3(earnings before interest and taxes/total assets) + (1) 0.6(market value of equity/book value of total liabilities) + (sales/total assets)

Corporate liquidity (LIQ) is used to explain that precautionary cash is used to offset liquidity risk by firms with higher probability of default. The independent variable liquidity is based upon a widely used measure, and is estimated by a firm’s cash and short-term investments relative to its total assets (see, e.g., Acharya et al., 2012; Seifert and Gonenc, 2016). The independent variable investor rights is measured by both the anti-director rights and anti-self- dealing index, based on the values of Djankov et al. (2008). Total investor protection of a country is defined as a qualitative variable with a value of zero or one. Total investor protection is classified as high (value of 1) when a country scores high on both the anti-director rights index and the anti-self-dealing index. Else, countries are indicated as countries with low investor protection (value of 0).

The regressions are adjusted for two firm-level control variables, including size (MV) and

book-equity to market-equity (BE/ME). Size is measured as the market value of equity. The

book-to-market ratio is defined as common equity divided by the market value of equity. Both

size and book-to-market equity are log-transformed (when non-negative), since these variables

15

do not have a naturally bounded distribution. These firm-level variables are included because of their well-documented association with equity returns (Dichev, 1998).

3.5 Methodology

Fama-MacBeth (1973) regressions are performed to statistically determine whether the relation between equity returns and financial distress is hump-shaped. Moreover, multiple regressions are run to determine a possible effect of cash holdings and country-level investor protection on the returns of distressed equities. Furthermore, portfolio results are presented where firms are assigned into equally weighted decile portfolios according to their default risk. An investigation of the means of the variables will illustrate whether the results are economically significant.

Besides, an examination could reveal the expected nonlinear relation between equity returns and financial distress, over and beyond the quadratic regressions.

Since this study uses panel data, correlation between the error terms and independent variables could be expected, which will cause ordinary least squares estimators to fail.

Therefore, the Hausman test is performed to examine whether there are endogenous regressors.

The Hausman test shows that fixed effects is the preferred model for the regression analysis, since fixed effects regressions control for unobserved, but constant variation cross-sectional.

Further, the assumption of homoscedasticity is tested using a modified Wald test for group wise heteroscedasticity. It reveals that there is heteroscedasticity in the sample, which leads to biased standard errors and significance levels. Moreover, a Woolridge test for autocorrelation shows that there is also autocorrelation present in the sample. To account for these two issues robust standard errors, as in Newey-West (1987), are used to correct the error terms and obtain more trustworthy results.

In the following, both linear and quadratic regressions are considered. The linear specification tests whether the correlation between equity returns and financial distress is as predicted by the model. The quadratic specification is a direct and simple test of a hump-shaped relation by assessing whether the quadratic term has a negative coefficient (Medhat, 2014).

Specifically, the following linear and quadratic regression specifications are assessed:

L: RET

= α

+ 

Z

+ 

MV

+ 

BE/ME

+ 

(2) Q: RET

= α

+ 

Z

+ 

Z

+ 

MV

+ 

BE/ME

+ 

(3)

Here, RET

is firm i’s accumulated return at year t, Z

is a measure of the firm’s distance-to-

default at year t measured by Altman’s Z-score, MV

and BE/ME

, are control variables and

16



are error terms. In the linear regression, 

estimates the marginal effect of financial distress on returns. In the quadratic specification, 

estimates the effect of financial distress, while 

estimates the shape of the relation between default risk and equity returns. Hence, the expected hump-shaped relation is examined by testing whether 

<0.

Hereafter, multiple regressions are performed to test the effect of corporate liquidity and investor protection on the main relation. Therefore, the interaction terms of corporate liquidity and investor protection are added to the regression to examine whether they significantly affect the relation between financial distress and equity returns. This results in the following:

Q: RET

= α

+ 

Z

+ 

Z

+ 

LIQ

+ 

Z

*LIQ

+ 

Z

*LIQ

+ 

MV

+



BE/ME

+ 

(4)

Q: RET

= α

+ 

Z

+ 

Z

+ 

LIQ

+ 

IP

+ 

Z

*LIQ

+ 

Z

*LIQ

+



Z

*IP

+ 

Z

*IP

+ 

MV

+ 

BE/ME

+

(5)

Here, LIQ

, a firm i’s cash holdings at time t, and IP

, a measure of a country i’s investor protection estimated by anti-director rights and anti-self-dealing, are added to the model.

3.6 Pearson correlation

Table 4 shows the pair-wise correlation matrix. It can be observed that almost all variables are

significant on a 1% level. The correlation between Z and equity returns is positive, which agrees

with the distress hypothesis that financially distressed firms earn lower returns. Corporate

liquidity is positively related to both Z and returns. Furthermore, cash holdings are negatively

related with market value. This is explained by the fact that larger firms hold relatively less

cash because large firms have better access to capital markets relative to small firms and are

more likely to attain funds more easily (Dittmar et al., 2003). Moreover, a significant negative

correlation between cash holdings and both measures of investor protection is found, consistent

with Dittmar et al. (2003). They find that firms in countries where shareholder rights are not

well protected hold up to twice as much cash as firms in countries with high investor protection,

justified by the importance of agency costs and capital market development. Substantial

correlation is exhibited when looking at the variables anti-director-rights and anti-self-dealing,

which is justified for the reason that both are measures for country-level investor protection. A

notably negative correlation between Z and book-to-market equity is found in line with Dichev

17

(1998). Even though considerable correlation exists between the variables, this does not necessarily mean there is causality.

*, ^**, ^*** denote significance at the 10%, 5%, and 1 % level respectively.

4. Results

This section presents the results of the multivariate regression models, the accomplished portfolio analysis, and finally the results of the subsamples. All specifications from the performed regressions include controls for firm size and book-to-market-equity. The tables report yearly slope-coefficients with standard errors that are adjusted for heteroscedasticity and autocorrelation as in Newey-West (1987). The second analysis of this paper includes portfolios of firms sorted by their default risk. Firms are assigned into equally weighted decile portfolios according to their probability of default, based on the Z estimates. The portfolios contain firm- observations in percentiles 10, 10 to 20, 20 to 40, 40 to 60, 60 to 80, 80 to 90 and 90 to 100 of the default risk distribution. The highest default risk equities are exhibited in Q1 and the lowest default risk equities in Q10. The analysis indicates whether firms with high default risk have uncommonly low or high equity returns relative to the predictions of standard cross-sectional asset pricing models, in line with Campbell et al. (2008). Besides this subsample of firms, a distinction is made between countries with high and low investor protection respectively.

4.1 Equity returns and financial distress

First, the effect of financial distress on equity returns is examined. Table 5 shows the results from multivariate regressions for the main relation and the interaction effects of cash holdings and investor protection. Column (1) shows that a higher Z is associated with significantly higher returns (0.010), contradicting the theory that financial distress risk carries a premium. Hence, this finding is consistent with the empirical studies finding evidence for the distress anomaly

Table 4 Pair-wise correlation matrix. This table reports the Pair-wise correlation statistics of the firm-level variables used in this study. The full sample consists of 10,710 firm-year observations for 714 unique European firms over the 2002-2016 period. Variable definitions are presented in Appendix A1. Financial variables are winsorized at the 1st and 99th percentiles.

Variables (1) (2) (3) (4) (5) (6) (7)

Returns 1.000

Z 0.006 1.000

Cash holdings 0.054^*** 0.374^*** 1.000

Anti-director-rights 0.027^** 0.064^*** -0.063^*** 1.000

Anti-self-dealing 0.017^* 0.046^** -0.082^*** 0.845^*** 1.000

Market value -0.107^*** -0.054^*** -0.088^*** -0.147^*** -0.199^*** 1.000

Book-to-market 0.171^*** -0.458^*** -0.161^*** -0.109^*** -0.087^*** -0.146^*** 1.000

18

(see, e.g., Dichev, 1998; Griffin and Lemmon, 2002; Campbell et al., 2008; Friewald et al., 2014). After including the quadratic term (Column 2), both Z as well as the quadratic term are significant at the 1% level. The quadratic term of Z is negative (-0.002), implying that the quadratic specification fits the model. So, the quadratic function has a negative second derivate significant at the 1% level, as was already expected by the model. Consistently, the visual relation between equity returns and financial distress reveals a hump-shaped and decreasing graph, see Appendix A3. The control variables are significant at the 1% level for all models.

Market value is significantly negative, in line with Aretz et al. (2017). Book-to-market positively affects equity returns at a 1% significance level, consistent with Medhat (2014). In the six specifications, the adjusted R-squared exhibits that between 13.9% and 16.3% of the dependent variable, equity returns, can be explained by the independent and control variables.

Table 6 presents portfolio results for the researched relations. An examination of the findings for the Z-sorted portfolios illustrates that firms with the highest and lowest default risk (Q1 and Q10) earn the lowest returns. Even a negative average return for the most distressed firms is exhibited (-0.021), in conformity with Campbell et al. (2008). Hence, this result indicates that the return premium is economically large for avoiding financially distressed firms. The portfolios provide economic evidence for the non-monotonic relation between equity returns and financial distress, consistent with Garlappi and Yan (2011). Moreover, there is little evidence of a size effect. Market value is lowest for the healthiest and most distressed firms. In regard to book-to-market equity, a moderately positive relation with financial distress is observed along with a difference of 1.134 between Q1 and Q10. So, the relation between returns and financial distress is more pronounced for firms with higher book-to-market equity. The most distressed firms (Q1) exhibit the highest book-to-market equity, suggesting that investors might undervalue these firms. Conversely, this finding is able to confirm the conjecture that investors do not demand a higher premium for holding financially distressed equity.

All in all, the positive coefficient for Z in the first model (column 1) suggests that default matters, consistent with empirical evidence presented by for instance Garlappi et al. (2008).

The quadratic coefficient further provides statistical support for the hump-shaped relation

between financial distress and equity returns, in compliance with Medhat (2014). These results

represent a conditional dependence of equity returns on financial distress. When analysing the

portfolios, returns are humped and decreasing when financial distress increases. The finding

that equities with high default risk tend to deliver anomalously low returns is consistent with

the results of Dichev (1998), who uses the same accounting-based measure of bankruptcy,

Griffin and Lemmon (2002) using Ohlson’s (1980) model, Garlappi et al. (2008), who obtain

19

default risk measures from Moody’s KMV, and Campbell et al. (2008) who use the reduced- form econometric model to predict corporate bankruptcies and failures.

Table 5 Regression models. This table reports coefficient estimates of the multivariate regressions with firm-level returns as dependent variable. The full sample consists of 10,710 firm-year observations for 714 unique European firms over the 2002-2016 period. Variable definitions are presented in Appendix A1. Standard errors are in parentheses, and are adjusted for heteroscedasticity and autocorrelation as in Newey-West (1987). For all models, the standard errors are adjusted for 691 clusters (company). Financial variables are winsorized at the 1st and 99th percentiles.

Independent variables (1) (2) (3) (4) (5) (6)

Intercept 1.616^***

(0.132)

1.580^***

(0.132)

1.220^***

(0.133)

1.122^***

(0.352)

1.181^***

(0.276)

1.230^***

(0.134)

Z 0.010^**

(0.004)

0.044^***

(0.009)

0.082^***

(0.011)

0.106^***

(0.033)

0.061^***

(0.017)

0.072^***

(0.012)

Z² -0.002^***

(0.000)

-0.004^***

(0.001)

-0.006^***

(0.002)

-0.003^***

(0.001)

-0.004^***

(0.001)

Cash holdings 1.248^***

(0.119)

1.257^***

(0.120)

1.231^***

(0.119)

1.244^***

(0.119)

Cash holdings*Z -0.192^***

(0.027)

-0.200^***

(0.028)

-0.183^***

(0.027)

-0.190^***

(0.027)

Cash holdings*Z² 0.008^***

(0.002)

0.009^***

(0.002)

0.008^***

(0.002)

0.008^***

(0.002)

Anti-director-rights 0.033

(0.107)

Anti-director-rights*Z -0.005

(0.007)

Anti-director-rights*Z² 0.000

(0.000)

Anti-self-dealing 0.271

(0.889)

Anti-self-dealing*Z 0.034

(0.022)

Anti-self-dealing*Z² -0.002

(0.001) Investor protection

(dummy)

0.318^***

(0.067)

Investor protection*Z 0.022

(0.014)

Investor protection*Z² -0.000

(0.001) Market value (ln) -0.247^***

(0.018)

-0.251^***

(0.018)

-0.235^***

(0.018)

-0.236^***

(0.018)

-0.235^***

(0.018)

-0.234^***

(0.018) Book-to-market (ln) 0.268^***

(0.013)

0.277^***

(0.013)

0.290^***

(0.013)

0.290^***

(0.013)

0.290^***

(0.013)

0.291^***

(0.013)

Adjusted R² 0.139 0.144 0.163 0.163 0.163 0.163

Observations 8,761 8,761 8,761 8,761 8,761 8,761

*, ^** and ^*** denote significance at the 10%, 5% and 1% level (2 tailed), respectively.

20

So, this study finds in conformity with Eisdorfer et al. (2014) that the distress anomaly is not U.S. specific. According to both multivariate regressions and portfolio analysis, hypothesis 1, stating that the relation between returns and financial distress is hump-shaped and decreasing, is statistically and economically supported. Moreover, the results show that financial distress is significantly priced in the presence of the commonly used firm control variables, and therefore Z is priced over and above some of its constituent variables. This is because Z is a transformation of firm characteristics, and hence the inclusion of these controls in the regression model cannot absorb the explanatory power of the distance-to-default measure.

4.2 Cash holdings and financial distress

The moderating effect of cash holdings on the revisited main relation is examined in this section. The second hypothesis, whether corporate liquidity affects the relation between equity returns and default risk is tested in column (3) by including cash holdings and the interaction terms between financial distress and cash holdings. Higher cash holdings are associated with significantly higher equity returns (1.248). The presence of the significant interaction terms indicates that the effect of financial distress on equity returns is different for different values of cash holdings. Subsequently, the unique effect of financial distress to equity returns not only depends on the significant coefficients of Z and Z

, but depends on the coefficient of corporate liquidity and the interaction terms with cash holdings as well. Moreover, portfolio results for the effect of cash holdings can reveal whether cash holdings help justifying the distress anomaly. An examination suggests that cash holdings are highest for most healthy firms (Q10).

More financially distressed firms exhibit lower cash holdings, however the most distressed firms (Q1) show an increase from 0.092 (Q2) to 0.103 (Q1); suggesting that cash holdings rationalize the distress anomaly.

Thus, cash holdings are found to significantly affect returns, and moreover have a

moderating effect on the relation of financial distress and returns. This finding is emphasized

by the economically significant results of the portfolios, showing that cash holdings increase

for the most distressed equities. The empirical findings are consistent with the model’s

conjecture regarding the influence of cash holdings in determining the relation between equity

returns and default risk. In particular, returns decrease and cash levels increase for firms with a

higher probability of default. Consequently, hypothesis 2, which suggests that cash holdings

affect the relation between equity returns and default risk, cannot be rejected.

21

The results are in line with Medhat (2014), who finds the same empirical evidence for U.S.

firms. So, no deviations in corporate liquidity policies between U.S. and European firms are found in this study. The findings are consistent with the prevailing insights suggesting that firms with high cash holdings should have a lower probability of default (Bolton, Chen, and Wang, 2011 and Acharya et al., 2012). The conclusive results convey that cash holdings allow managers to offset liquidity risk, thereby reducing risk exposure by investors. Particularly, firms holding higher cash balances in their asset and investment portfolio should face less risk. The findings suggest that the level of cash holdings constitute an important extent of distressed equity. Therefore, managers need to consider default risk in corporate liquidity decisions, and vice versa. More specifically, managers need to make a trade-off between paying out dividends or holding precautionary cash to outweigh future coupons.

To elucidate the distress anomaly with a firm-level variable that is more well-known to affect financial distress, leverage is enclosed in the portfolio analysis. The explicit inclusion of financial leverage allows this research to show how leverage amplifies the default risk effect.

The evidence indicates that financial distress is positively related with leverage, as is shown by the linear decrease in leverage when Z increases. Thus, the negative relation between equity returns and financial distress is more pronounced for firms with higher leverage. As presented by Griffin and Lemmon (2002), firms in the high financial distress quantile exhibit characteristics traditionally associated with bankruptcy probability, such as high leverage.

Moreover, market leverage increases significantly as probability of default increases. Although

Table 6 Portfolio results. This table reports portfolio results where firms are assigned into equally weighted decile portfolios according to their probability of default. The highest default risk equities are exhibited in Q1 and the lowest default risk equities in Q10 respectively. The full sample consists of 10,710 firm-year observations for 714 unique European firms over the 2002-2016 period. Variable definitions are presented in Appendix A1. Standard errors are in parentheses.

Financial variables are winsorized at the 1st and 99th percentiles.

Deciles

P10-P1

1 2 3-4 5-6 7-8 9 10

Returns -0.021

(0.019)

0.067 (0.015)

0.098 (0.009)

0.099 (0.009)

0.109 (0.009)

0.084 (0.013)

0.062 (0.012)

0.083

Cash holdings 0.103

(0.004)

0.092 (0.002)

0.097 (0.002)

0.102 (0.002)

0.115 (0.003)

0.156 (0.004)

0.223 (0.005)

0.120

Market value (ln) 6.301 (0.026)

6.480 (0.022)

6.358 (0.015)

6.362 (0.016)

6.277 (0.016)

6.220 (0.022)

6.254 (0.021)

-0.047

Book-to-market (ln) -0.207 (0.031)

-0.337 (0.026)

-0.495 (0.016)

-0.707 (0.015)

-0.883 (0.015)

-1.174 (0.021)

-1.341 (0.025)

-1.134

Leverage 0.689

(0.006)

0.628 (0.005)

0.531 (0.003)

0.441 (0.003)

0.341 (0.003)

0.213 (0.003)

0.175 (0.006)

-0.514

Observations 812 879 1,764 1,770 1,763 891 1,138

22

higher market leverage will not automatically result in default, a financially distressed firm will more likely be forced by its debtholders to anticipate on its declined solvency (Medhat, 2014).

4.3 Investor protection and financial distress

Investor protection is added to the main regression model to test the effect of this cross-country variable on the relation between equity returns and financial distress. In column (4), anti- director rights as a measure of investor protection is included (Table 5). The signs and significance levels of the distance-to-default and cash holding variables remain the same.

Nevertheless, the influence of anti-director rights on the regression appears to be insignificant.

Next, the anti-self-dealing index is incorporated, resulting in the same findings as when examining anti-director rights. When adding the dummy variable for investor protection, the non-interacted term gives a significantly positive coefficient of 0.318. So, it can be said at the 1% significance level that firms located in high investor protection countries earn higher returns. Notwithstanding, nothing can be said about the interaction terms of financial distress and investor protection as their coefficients are insignificant.

In table 7 the results of a more formal test on the impact of investor protection are presented

by re-running the main estimations for each subsample. Table 7 depicts regression models for

the split sample of relatively high and low investor protection. The subsets of the data are used

to examine how differences in investor protection affect the relation of returns and financial

distress. The following countries are found to have high investor protection; Denmark, Finland,

Ireland, Norway and the United Kingdom. The regression models for the low investor

protection sample consists of subsequent countries; Austria, Belgium, France, Germany,

Greece, Italy, Luxembourg, Poland, Portugal, Spain, Sweden, Switzerland, and the

Netherlands. Column 7 and 8 exhibit that the coefficient of Z is higher for firms in high investor

protection countries, indicating that in those countries default risk decreases equity returns

more. However, the quadratic term is qualitatively the same with coefficients of -0.003 and

-0.002 and significance levels of 1% for both subsamples (column 7 and 9). When adding cash

holdings and the interaction terms of Z and cash holdings, the quadratic term is -0.005 for firms

in high investor protection countries and -0.003 for firms in low investor protection countries

respectively. Hence, equity returns decrease more severely when default risk increases for firms

in high investor protection countries relative to firms in countries where investors are less

protected.

23

Moreover, a portfolio analysis for the subsamples can reveal whether investor protection helps justifying the distress anomaly. The results are shown in table 8. An examination suggests that, in line with the performed regressions, returns are moderately highest for firms in countries with high investor protection. However, for Q5-6 returns are found to be lower compared to firm-year observations in the low investor protection countries. For the portfolios with the most distressed firms, equity returns decrease substantially more for the high investor protection sample (0.092 to -0.019, against 0.056 to -0.022 for firms in the low investor protection sample).

This finding suggests that investors in countries with a high quality of investor protection require a lower default risk premium for distressed equities. The results are in compliance with existing literature asserting that the default risk premium is lower in countries in which investors have higher bargaining power and the judicial process favours debt renegotiation.

Table 7 Regression models for subsamples based on investor protection. This table reports coefficient estimates of the subsamples for the run multivariate regressions with firm-level returns as dependent variable. The full sample consists of 10,710 firm-year observations for 714 unique European firms over the 2002-2016 period. Variable definitions are presented in Appendix A1. Standard errors are in parentheses, and are adjusted for heteroscedasticity and autocorrelation as in Newey- West (1987). Standard errors are adjusted for 309 clusters (company) in the high investor protection sample and for 382 clusters (company) in the low investor protection sample. Financial variables are winsorized at the 1st and 99th percentiles.

High investor protection Low investor protection

Independent variables (7) (8) (9) (10)

Constant 1.738^***

(0.178)

1.448^***

(0.181)

1.177^***

(0.165)

0.828^***

(0.167)

Z 0.059^***

(0.011)

0.102^***

(0.017)

0.031^***

(0.011)

0.069^***

(0.012)

Z² -0.003^***

(0.001)

-0.005^***

(0.001)

-0.002^***

(0.001)

-0.003^***

(0.001)

Cash holdings 1.165^***

(0.242)

1.377^***

(0.139)

Cash holdings*Z -0.232^***

(0.070)

-0.185^***

(0.027)

Cash holdings*Z² 0.012^***

(0.004)

0.007^***

(0.002)

Market value (ln) -0.318^***

(0.027)

-0.300^***

(0.027)

-0.186^***

(0.024)

-0.176^***

(0.024)

Book-to-market (ln) 0.251^***

(0.019)

0.262^***

(0.019)

0.307^***

(0.017)

0.320^***

(0.017)

Adjusted R² 0.137 0.147 0.154 0.180

Observations 3,858 3,858 4,903 4,903

*, ^** and ^*** denote significance at the 10%, 5% and 1% level (2 tailed), respectively.