Successful turnarounds in bankrupt firms? Assessing retrenchment in the most severe form of crisis

Abstract

During economic downturns, firms file for bankruptcy in an effort to attempt a ‘turnaround’. The objective of this study is to assess the effectiveness of retrenchment strategies in the context of bankruptcy, as the most severe form of crisis. We conducted a longitudinal analysis of a sample of 868 bankrupt Spanish firms during the period 2004–2017. The empirical results show that stakeholder support and deep cost retrenchment increase the likelihood of survival and performance recovery, while aggressive layoffs are detrimental for turning bankrupt firms around. Surprisingly, intense asset retrenchment had no significant effects on firm survival and also pushed performance downwards. The findings suggest that retrenchment should not be regarded as a general remedy for firms suffering the most severe of crisis. Bankrupt firms should focus on restoring stakeholder relationships and reducing superfluous expenses, while making employees redundant or selling assets should be evaluated carefully when attempting a turnaround.

Introduction

The liquidation of firms is a pervasive problem in modern society. Over the past decade, the dramatic socioeconomic effects of liquidation were made evident in several cases involving bankrupt firms (General Motors, Toys ‘R’ Us, Schlecker). Therefore, research on bankruptcy and survival is essential and, in light of this, turnaround strategies are of interest to bankrupt firms (Franks and Sussman, 2005; Pandit, 2000; Trahms et al., 2013).

Previous studies have traditionally identified two main stages of the turnaround process i.e., retrenchment and recovery (Pearce and Robbins, 1993). The retrenchment stage has been viewed as the first essential step in a turnaround process (Robbins and Pearce, 1992), aiming to stabilize the decline through divestments, asset disposals, cost-cutting or layoffs. These stabilizing measures provide the basis and resources for subsequent recovery (Arogyaswamy et al., 1995; Barker and Duhaime, 1997).

However, empirical findings regarding the efficacy of retrenchment are incomplete and heterogeneous, and few studies have identified the most effective means by which to implement retrenchment actions (Lim et al., 2013). Why is there such a controversy regarding the efficacy of retrenchment? Strategic contingency theory suggests that context matters (Arogyaswamy et al., 1995; Hofer, 1980), and the appropriateness of turnaround strategies will depend on the origin of the decline (external vs internal) and the situation of the firm (high vs low severity) (Ndofor et al., 2013). In terms of the severity of the decline, previous research found that retrenchment actions are expected to improve the likelihood of a turnaround in firms that suffer a high severity crisis (Hambrick and Schecter, 1983; Hofer, 1980; Morrow et al., 2004; Pearce and Robbins, 1993; Robbins and Pearce, 1992). In opposition to that view, other scholars argued that the immediate benefits of retrenchment fail to outweigh its long-term strategic costs (Barker and Duhaime, 1997; Barker and Mone, 1994; Boyne and Meier, 2009; Castrogiovanni and Bruton, 2000) and also proposed a balanced adoption of retrenchment and recovery actions (Dolz et al., 2019; Schmitt and Raisch, 2013). To test both of these conflicting views, we examined the role of retrenchment in turning around firms during the most severe form of crisis, namely, bankruptcy.

In contrast to previous turnaround studies, this paper focuses on insolvency (status) and bankruptcy (formal procedure), a more severe situation than the commonly studied case of financial distress (Altman and Hotchkiss, 2006). While financial distress refers to firms that experience difficulties meeting payments (Gilson, 2010), a firm is defined as insolvent when it is actually unable to repay its debts. Bankruptcy is the formal procedure filed by an insolvent firm. Such a procedure is known as concurso de acreedores in Spain or Chapter 11 in the United States (Altman and Hotchkiss, 2006). Given that bankruptcy is the most severe situation for a firm, retrenchment should be associated with survival and turnaround. However, not all retrenchment actions have the same effect, and the pervasive impact that layoffs (Datta et al., 2010; Santana et al., 2017) or indiscriminate asset disposals (Ndofor et al., 2013) can have on a firm, may make these measures counterproductive in respect to recovery from bankruptcy. Furthermore, the role of stakeholders and its importance during bankruptcy (Pajunen, 2006) are considered.

To gain a deeper understanding of the phenomenon, this study overcomes the traditional success-failure dichotomy (Åstebro and Winter, 2012), and identifies three possible outcomes, namely, liquidation, marginal survival, and success. Liquidation is the complete failure of the firm, marginal survivors are firms that did not turn performance around, and successful firms are those that both survived and recovered their performance. Thus, the analyses differentiate between firms that failed or survived, but also distinguish between those that did or did not recover their prior performance. To the best of our knowledge, no previous studies assessed the evolution of performance in Spanish firms within the context of the bankruptcy procedure. Nonetheless, this issue has generated interest, as some large well-known firms, such as Martinsa-Fadesa, Pescanova, Reyal Urbis, and Blanco, entered into bankruptcy proceedings over the past few years.

The results of the multinomial logistic regression (MLR) analysis, which was carried out using a sample of 868 Spanish bankrupt firms during the period 2004–2017, contributes in several ways. First, our study provides a partial test of the contingency theory of turnaround as we examined the association of retrenchment actions with performance decline in firms suffering the most severe form of crisis (bankruptcy). We found that cost retrenchment is associated with survival and turnaround, but layoffs decreased the probabilities of survival and turnaround of the firm. Thus, retrenchment should not be regarded as a general remedy for firms suffering the most severe crisis. In addition, stakeholders prove to be key players during bankruptcy proceedings. The particularities of Spanish bankruptcy regulation are also assessed.

Second, the findings provide a guide for the managers of bankrupt firms to adopt measures that not only ensure survival, but which also make their firms successful. The finer-grained operationalization of turnaround outcomes and the MLR analyses allow us to distinguish between those firms that merely survive and those that also succeeded. In particular, the results show that survival is explained by structural factors as size, age, debt, and the severity of decline, while success is mainly associated with restoring stakeholder support and reducing superfluous costs. Layoffs are not a recommended action since they decrease the likelihood of survival and turnaround, and selling assets has pernicious effects in terms of improving performance.

Third, the results provide guidance to improve the efficacy of bankruptcy regulations. In addition, by focusing on Spanish firms, we incorporate a relevant, albeit neglected, context, as the majority of previous turnaround studies focused on US firms.

The paper is structured as follows: In the next section, the hypotheses are developed. The “Methodology” section describes the data and methodology employed to test the effects of stakeholder support and retrenchment on bankrupt firms. In the “Results” section, the results of the analyses are reported and discussed. Finally, the last section presents the conclusions drawn from our study.

Conceptual model and hypotheses

Retrenchment strategies, bankruptcy, and stakeholders

In their seminal article, Robbins and Pearce (1992) addressed the relevance of retrenchment to turn declining firms around. The study employed a sample of textile firms that experienced a minimum of two years of decline in respect to return on investment (ROI) and return on sales (ROS) relative to the industry average following a period of prosperity. Successful turnarounds were defined as those firms that subsequently became prosperous and achieved increasing, above industry average ROI and ROS for two consecutive years. Two types of retrenchment were defined i.e., cost retrenchment (net reduction in total costs) and asset retrenchment (net reduction in total assets). Robbins and Pearce found that both cost and asset retrenchment were positively correlated with turnaround performance. Furthermore, the strongest correlation was found for firms that experienced higher severity declines. As a result, the study concluded that “retrenchment was a critical strategic element in attaining turnaround” (Robbins and Pearce, 1992:303).

However, subsequent research on retrenchment has produced contradictory results. Barker and Mone (1994) suggested that retrenchment may be a consequence of decline rather than a means for turnaround. Alternative studies found that whilst cost and asset retrenchment may be appropriate within mature industries, it is argued that these actions are inappropriate in other contexts. Morrow et al. (2004) question the universality of retrenchment in high growth and innovative environments. Furthermore, the effectiveness of certain cost and asset retrenchment actions are believed to be dependent upon the firm's rent-creation mechanism (Lim et al., 2013). Finally, the implementation of retrenchment (Castrogiovanni and Bruton, 2000; Sudarsanam and Lai, 2001), including its timing, (Tangpong et al., 2015) can affect outcomes. One way to produce consistent results could be to study homogeneous samples in terms of the turnaround situation (Pandit, 2000). In this respect, bankruptcy is a promising framework (Collet et al., 2014).

As far as we know, no prior scholars studied the effectiveness of turnaround strategies in bankrupt firms. Literature from post-bankruptcy performance (Altman and Hotchkiss, 2006; Gilson, 2010; Hotchkiss, 1995) focuses on firms emerging from bankruptcy, and suggests that nearly half of the firms experienced performance losses and even re-entered the procedure. However, this approach is different to the one that is employed in this study, the latter of which assesses the adoption of retrenchment during the bankruptcy procedure and how it affects performance during bankruptcy. To achieve this aim, the contingency theory perspective of turnaround was selected as the proper framework (Arogyaswamy et al., 1995; Hofer, 1980). According to this framework, retrenchment is expected to be more closely associated with performance turnarounds in the case of firms suffering a highly severe crisis. However, the pervasive effects that aggressive measures, such as layoffs (Datta et al., 2010; Santana et al., 2017) or indiscriminate asset disposals (Ndofor et al., 2013) may have on firm performance should be considered, despite the expected positive impact of retrenchment. Furthermore, recent studies suggest the need to balance retrenchment and recovery actions during decline (Dolz et al., 2019; Schmitt and Raisch, 2013).

How is retrenchment expected to work for bankrupt firms during the procedure? Having considered previous contradicting findings regarding retrenchment and the framework of bankrupt firms, we believe that it is necessary to study the effectiveness of retrenchment by adding stakeholder support and the intensity of response (Arogyaswamy et al., 1995). Previous studies have scarcely studied the interrelationship that exists between these variables. In summary, the literature proposes that such variables evolve in the following way: (i) when a firm experiences a performance decline, its relationship with stakeholders deteriorates due to worsening results (D’Aveni, 1989); (ii) to address this situation, the firm must realign its stakeholders’ expectations with the new situation in order to obtain their support for the adoption of necessary measures to stabilize the crisis (Filatotchev and Toms, 2006); (iii) the firm will need to adopt somewhat drastic measures to stabilize the decline and improve profitability (Robbins and Pearce, 1992); (iv) once such measures have been adopted, a virtuous cycle is activated, such that an improvement in performance will encourage the support of stakeholders (Pajunen, 2006). In short, the proposed hypotheses are formed on the basis of stakeholder support and response intensity, as well as their impact in the event of an extremely severe crisis.

Hypotheses

One of the first consequences of decline is a deterioration in relationships with stakeholders (Arogyaswamy et al., 1995). In fact, D’Aveni (1989) defined bankruptcy as a situation that results from a lack of organizational legitimacy, such that stakeholders withdraw their support. Therefore, to achieve a successful turnaround, the firm must renew its relationship with stakeholders. This dynamic was described by Filatotchev and Toms (2006) who proposed the “realignment stage”, at which the firm enhances its ties with critical stakeholders, particularly those who will support the adoption of necessary measures in order to survive and regain profitability. Thus, prior to taking any action, the firm must reorient stakeholders’ expectations to ensure that they are in line with the current situation to restore their support. In this situation, it may be the case that stakeholders come to play a critical role, particularly as the firm's survival depends on their support after a survival-threatening decline (Pajunen, 2006). In the event that the firm is able to restore their support, stakeholders will provide funds, commercial credit or capital (Trahms et al., 2013). Some firms that illustrate the relevance of stakeholders include General Motors or Schlecker. Thus, the first hypothesis is proposed as follows:Hypothesis 1

Ceteris paribus, (not) obtaining stakeholders’ support increases (decreases) the probability of turnaround success for bankrupt firms.

Although the financial press depicts cases of bankrupt firms and the way that they confront the process (e.g. store closings by Toys ‘R’ Us or massive employee layoffs at LyondellBassell), few studies have examined the intensity of the strategic response. While prior authors suggested that intensity could be a relevant variable (Hofer, 1980; Lim et al., 2013; Robbins and Pearce, 1992), it has not been explicitly studied by previous research. As Lim et al. (2013) highlighted, little is known about what, how, and when retrenchment strategies should be adopted in turnaround situations. From this perspective, previous literature offers some intuitive support for the idea that the intensity of response may be relevant for the turnaround outcome, whether positive (Hofer, 1980; Robbins and Pearce, 1992) or negative (Weitzel and Jonsson, 1989).

Nonetheless, the impact of retrenchment strategies is expected to be contingent upon its content. Thus, a retrenchment strategy cannot be studied as a whole. Retrenchment measures were classified into three areas, namely, costs, employees, and assets, as previously observed by turnaround scholars (Tangpong et al., 2015). For all three strategies, hypotheses were developed on the basis that bankrupt firms experience a highly severe situation (Robbins and Pearce, 1992) and, as such, the intensity of retrenchment responses will affect firms in a particular manner.

In respect to cost retrenchment, previous studies found both positive and negative consequences for subsequent turnaround potential. Nonetheless, cost-cutting strategies were found to be beneficial in studies which used firms in mature or declining industries (Morrow et al., 2004; Robbins and Pearce, 1992; Tangpong et al., 2015), while high growth environments were not suitable for cost retrenchment strategies (Ndofor et al., 2013). In this context, bankrupt firms are expected to find themselves in a somewhat dramatic situation, and a decisive stabilization of their operations will be needed (Arogyaswamy et al., 1995). Consequently, as suggested by Robbins and Pearce (1992), intense cost-cutting measures are expected to increase the likelihood of achieving a successful turnaround. Thus, this hypothesis is formulated as follows:Hypothesis 2a

Ceteris paribus, more (less) intense cost retrenchment measures will positively (negatively) impact the probability of turnaround success for bankrupt firms.

Human resources are an area that offers greater opportunities for cost savings in declining firms. In particular, small- and medium-sized firms (SMEs) are limited by their scale, and lack the resources to undertake measures that are more complex reducing the size of the labour force (Chowdhury and Lang, 1996). Reducing labour costs while maintaining sales leads to an increase in productivity. As such, firms are tempted to adopt this measure. However, the immediate advantages of reducing the number of employees might outweigh the harm that results from a drastic reduction in the firm's labour costs (Tangpong et al., 2015). Labour liabilities, such as decreased involvement or trust, (Boyne and Meier, 2009; Datta et al., 2010; Santana et al., 2017) often arise. Therefore, the desired turnaround results will not be attained. As an intense reduction in the labour force may harm a firm's potential to recover in the future, the following hypothesis is proposed:Hypothesis 2b

Ceteris paribus, more (less) intense labour cost retrenchment measures will negatively (positively) impact the probability of turnaround success for bankrupt firms.

Asset retrenchment is one of the most relevant measures that declining firms adopt in turnaround situations (Bruton et al., 2003). In fact, selling assets can contribute to reducing debt in over-leveraged firms (Altman and Hotchkiss, 2006). Robbins and Pearce (1992) concluded that asset retrenchment was necessary when the severity of decline was high or extreme, which is observed in the case of bankruptcy. Hofer (1980) furthermore proposed that firms whose sales are well below the break-even point (i.e., fixed costs are greater than sales) should aggressively retrench assets. Nonetheless, asset retrenchment is usually determined by the time and resources available to the firm when it is confronted with decline. An excessive reduction in assets threatens the future competitiveness of the firm (Ndofor et al., 2013). Hofer (1980) restricts asset retrenchment to “the level needed to meet the firm's cash flow needs for the next three or six months”, though logically, excessive asset retrenchment will cause the firm to lose its competitive advantages. Consequently, the intensity of asset retrenchment is relevant for the turnaround process, and depending on the intensity of the retrenchment, it can have a positive or negative impact on turnaround probability. Therefore, the following hypothesis is proposed:Hypothesis 2c

Ceteris paribus, more (less) intense asset retrenchment measures will negatively (positively) impact the probability of turnaround success for bankrupt firms.

Methodology

Empirical context: the Spanish bankruptcy procedure

Empirically, this study focuses on bankrupt Spanish firms for the period 2004–2017. Bankruptcy figures in Spain dramatically increased during the economic crisis, reaching a peak in 2013 as GDP sank during a second recession (Fig. 1). Martinsa-Fadesa, Reyal-Urbis, Pescanova, Fagor or Blanco are only the tip of the iceberg for a population over 50,000 firms that went bankrupt in Spain, which represented nearly 65 billion assets in the period 2004–2017 (Van Hemmen, 2018). What is a bankruptcy? Why focus on Spain? In this study, “bankruptcy” is distinguished from “financial distress” and “insolvency”. While “bankruptcy” (concurso de acreedores) is a formal in-court procedure, “insolvency” is an economic and financial status. Insolvent firms are those which cannot repay their debts (Gilson, 2010). The bankruptcy procedure permits only two solutions: (1) survival, upon having reached an agreement with creditors and (2) liquidation and dissolution of the firm. Therefore, the bankruptcy system is one of the means by which to address insolvency.

Figure 1.

Bankruptcies and GDP growth in Spain (2005–2017).

We selected Spain, while considering the substantial influence of the institutional context on the bankruptcy system (Davydenko and Franks, 2008). Spain is characterized by, at least, two aspects. On the one hand, firms rarely enter into a bankruptcy procedure (García-Posada and Mora-Sanguinetti, 2012). On the other hand, the effectiveness of the bankruptcy system is much lower than that of other developed countries (García-Posada and Vegas, 2016). This is the reason why in Spain, only 6% of bankrupt firms achieve survival, while the USA Chapter 11¹ is associated with survival rates of around 24% (Altman and Hotchkiss, 2006). Thus, it is important to examine the measures undertaken by this minority of Spanish firms in order to extend the findings to the wider majority.

The structure of the Spanish bankruptcy procedure is illustrated in Fig. 2. Given that the focus of this study is placed on formal bankruptcy, the out-of-court procedure of pre-bankruptcy (preconcurso) was excluded, as those firms do not meet the definition of ‘bankrupt’. First, a firm becomes insolvent and files for bankruptcy (1). In this case, the firm can go into liquidation (2–4) or attempt a turnaround (3). The first step (i.e., liquidation) was discarded in this study, as such firms do not adopt any strategy that is relevant to the research object. If an agreement with creditors is reached, the turnaround attempt will be successful and the firm will achieve temporary survival (5). If creditors do not agree with the viability plan, the firm is liquidated and the turnaround will fail.

Figure 2.

The bankruptcy procedure.

The present study assumes that bankrupt firms can reverse their situation throughout the bankruptcy procedure, i.e., from the declaration date until the turnaround attempt. During the procedure, firms attempt to restore their stakeholder support and adopt corrective measures.

We move beyond the traditional survival-liquidation dichotomy, and classify survival in terms of two possible outcomes, namely, marginal and successful (Trahms et al., 2013). Thus, we seek to enrich the contributions of the study by examining three outcomes, as opposed to two, in order to facilitate a more thorough understanding of resilient firms (i.e., successful survivors). Marginal survival is defined as a situation whereby the firm, despite having survived the bankruptcy procedure, fails to restore its previous performance level in respect to return on assets (ROA), whereas successful survival occurs when the firm, rather than merely surviving, recovers its pre-bankruptcy ROA value. In summary, three possible turnaround outcomes were defined: liquidation, marginal survival, and success.

The role of the banking industry in bankruptcy

Spain suffered two consecutive recessionary periods since 2008 and, as a result, the number of firm bankruptcies substantially increased, from around 1000 cases in 2008 to nearly 10,000 in 2013, which was the peak year of bankruptcies declarations (INE, 2018). According to institutional analysts (Banco de España, 2014; García-Posada and Vegas, 2016), bankruptcies in Spain were propelled by both the economic recession and the specific crisis of the banking industry. Indeed, the Spanish banking system underwent a radical change due to the increase in non-performing loans and the internal inefficiencies of some saving banks (cajas de ahorros) and credit unions (cooperativas de crédito). As a result of such dynamics, entities were forced to merge into bigger ones. Saving banks and credit unions either became commercial banks or were acquired by them. During this process, which took place from 2009 to 2013, the banking industry drastically reduced credit concessions to non-financial firms (more than 18%), further aggravating firm's economic troubles (García-Posada and Vegas, 2016). In this context, saving banks and credit unions were no longer closely linked to the territories (Comunidad Autónoma or, at a lower level, provincia) in which they were created (such as Bancaja in Valencia or Catalunya Caixa in Catalunya). Consequently, small and medium sized enterprises (SMEs), bound to their regions of origin, were the most affected by this concentration process, given that their key source of financing came from saving banks or credit unions (García-Gallego and Mera, 2016).

What was the effect of this process for bankruptcies? Researchers (Franks and Sussman, 2005; Gilson, 2010) found that firms with a higher proportion of banking debt were more likely to resolve their financial distress by way of out-of-court procedures. However, the Spanish banking industry was dealing with its own internal troubles and, as a result, pushed a substantial number of firms towards bankruptcy (García-Posada and Vegas, 2016).

Sample

The Spanish bankruptcy law came into force in 2004 (García-Posada and Vegas, 2016). Thus, the sample was drawn from the population of Spanish firms that entered bankruptcy during the period 2004–2017 (Fig. 2). At the time of data collection (i.e., 31 December 2018), the cut-off date for data was the end of 2017, since no complete data track was available for procedures that commenced in 2018. Accordingly, firms that reached a particular outcome (i.e., liquidation, marginal survival, or success) between 2004 and 2017 were included in the sample. In order to avoid gathering data related to failures that were due to liability of newness, firms were required to have been operating for at least five years (Thornhill and Amit, 2003). In addition, it was required that the procedure lasted more than one year in order to calculate the variations needed for computing variables.

Data were extracted from the Sistema de Análisis de Balances Ibéricos (SABI) database. SABI stores financial information obtained from the annual accounts of two million Spanish firms and half a million Portuguese firms, which are collected from the Public Commercial Register (Registro Mercantil). SABI offers the possibility to filter firms that have filed for bankruptcy at some time in their business life (“incidences/current status/concurso”). Data regarding the bankruptcy declaration and the outcome of the procedure were obtained from the Public Bankruptcy Register (Registro Público Concursal). The Register contains data including the declaration date, survival, and/or liquidation. It is possible that a selected firm was still engaged in a bankruptcy procedure, though only if it was in the process of liquidation. In that instance, its turnaround attempt would have been complete, though the formal proceeding had not yet been completed. Firms that had simultaneously entered the bankruptcy procedure and gone into liquidation were also discarded.

State-owned firms and sports clubs were excluded from the study. Public or state companies need a turnaround procedure that differs significantly from that of privately held firms (Jas and Skelcher, 2005). Sports clubs were also left out given that they are characterized by certain features that make their exclusion advisable (Rico and Puig, 2015). Initially, 1042 bankrupt firms were identified using the SABI database. Among these, 883 satisfied the criteria for inclusion. Firms with missing data were also excluded. The final sample consisted of 868 firms, among which 357 (41.1%) were liquidated, 381 (43.9%) marginally survived, and 130 (15.0%) were successful. The sample is biased towards bankrupt firms that survived (58.9% vs 6% of the entire population of bankrupt Spanish firms). The same problem was found in similar studies of Spanish firms (Pozuelo et al., 2013; Van Hemmen, 2009), given the difficulties in obtaining data from firms that ceased their operations. Provided that the majority of bankrupt firms (around 94%) are liquidated, few data are available to study the actions adopted by them during bankruptcy.

Variables

Dependent variables

Outcome

Traditionally, the dependent variable used in turnaround studies is a binary variable based on financial indicators (e.g. ROI, ROS or ROA) (Morrow et al., 2004; Ndofor et al., 2013; Robbins and Pearce, 1992). However, Åstebro and Winter (2012) found that the binary logistic model is normally mis-specified, and the multinomial model offers further alternatives for analysis. Thus, a dummy variable seems inappropriate to capture the nuances of turnaround in a bankruptcy procedure. In addition, Boyne and Meier (2009) found that the study of turnaround required the use of indicators that are suitable for the context that is being analyzed. Consequently, a combination of bankruptcy- and turnaround-related indicators were developed for this study. During bankruptcy, firms can either achieve survival or be liquidated. Furthermore, surviving firms may or may not restore their performance and achieve a minimum level above which it could be asserted that they were turned around. Thus, three possible outcomes were defined, i.e., liquidation, marginal survival, and successful survival.

‘Liquidation’ is defined as the failure outcome i.e., the firm does not survive. In this event, the variable took the value of ‘1’. For the other two outcomes, we followed prior turnaround studies (Ndofor et al., 2013) to examine ROA and change in ROA between the year preceding the declaration of bankruptcy and the year that preceded the reaching of an outcome. If a firm diminished its ROA, or if it increased ROA but it was negative in the year prior to reaching an outcome, it was defined as a ‘marginal survivor’. In this case, the variable was assigned a value of ‘2’. ‘Successful survival’ is used to describe a firm that achieved survival, improved its previous ROA, and showed a positive ROA at the year preceding the reaching of an outcome i.e., the firm was turned around. In this case, the variable was assigned a value of ‘3’.

Change in ROA

In order to measure the robustness of our findings, we also tested the proposed hypotheses against one of the most common measures of turnaround potential, which is the change in ROA (Ndofor et al., 2013; Tangpong et al., 2015). The change in ROA was measured as the difference between ROA in the year preceding the bankruptcy declaration and ROA in the year preceding the outcome.

Independent variables

Stakeholder support

Bankrupt firms are predominantly SMEs. Consequently, data or information about stakeholder support, such as market capitalization or debt rating, are scarce or non-existent (Tangpong et al., 2015; Xia et al., 2016). Considering this limitation, the capital variation during the bankruptcy procedure was utilized as a proxy variable to represent stakeholder support. An increase in capital during the bankruptcy proceeding indicates that stakeholders have reduced debts and that the firm has raised additional capital, thus achieving better results through contract renegotiations or the receipt of grants (Castrogiovanni and Bruton, 2000). In this case, capital variation can be an appropriate indicator of whether or not the firm is successful in gaining the support of external actors. If capital increased during bankruptcy, the variable took the value of ‘1’. If capital was either maintained or reduced, the variable took the value of ‘0’.

Intensity of response

This variable captures the intensity of retrenchment with respect to costs, employees, and assets. To measure retrenchment for each variable, we calculated the percentage variation that occurred between the year preceding bankruptcy and the year prior to having reached an outcome. To measure cost retrenchment, the sum of selling, general and administration costs (SGA) and interest was taken (Robbins and Pearce, 1992). The reduction in the number of employees was used to measure retrenchment in relation to employees (Tangpong et al., 2015), and the firm's tangible assets reduction (i.e. plant, equipment and machinery) was used to address asset retrenchment strategies (Morrow et al., 2004).

Control variables

Size

Previous research has addressed the influence of firm size on its survival prospects, concluding that larger and older firms are more likely to subsist (Cater and Schwab, 2008; Schmitt and Raisch, 2013; Thornhill and Amit, 2003). Large firms are able to raise unsecured capital, and assets generated by such borrowing provide collateral for additional borrowing (Routledge and Gadenne, 2000). As such, these firms can enjoy not only a wider resource-base, but a higher level of slack resources. In addition, some authors (e.g. Altman and Hotchkiss, 2006; Camacho-Miñano et al., 2015) acknowledged the existence of an economy of scale with respect to bankruptcy costs. These costs are comparatively higher for SMEs than for larger firms, as a significant proportion of the expenses associated with the proceeding are fixed. Consequently, larger firms are in a better position to enter into a bankruptcy proceeding than smaller firms (Van Hemmen, 2018). Size was captured as the log of total assets in the year that preceded the bankruptcy declaration.

Age

The age of the firm has a direct influence on its likelihood of survival during a bankruptcy procedure. Thornhill and Amit (2003) found that younger firms have a higher tendency to fail due to deficiencies in managerial knowledge, while older firms may be unable to adapt to sudden environmental changes. Age is also an indicator of strategic resources stock (Wild and Lockett, 2016), and older firms may be better able to survive financial shocks than younger ones which have an inferior buffer of resources and capabilities. In this study, age was measured as the number of years between the firm's establishment and the bankruptcy declaration. The natural log of age in the year preceding the bankruptcy declaration was employed in the analyses.

Debt

Insolvency is the inability to repay debts. Thus, a control variable capturing the debt burden of the firm is common in financial distress and bankruptcy studies (Altman and Hotchkiss, 2006; Gilson, 2010). As Routledge and Gadenne (2000) found, firms with a higher degree of leverage are expected to have a lower probability of survival due to the complexity of directing the firm towards a financially viable form. For the purposes of the study, debt was captured as the sum of long-term debt and short-term loans-to-total assets, in the year prior bankruptcy declaration.

Severity of decline

Severity of decline is one of the most commonly examined variables in previous turnaround research (Lim et al., 2013; Morrow et al., 2004; Robbins and Pearce, 1992; Schmitt and Raisch, 2013), and several measures were proposed to capture it. The severity of decline determines which turnaround strategy is more suitable to the firm's situation. While low severity situations may require only cost retrenchment, firms that experience highly severe declines should also adopt asset retrenchment (Hambrick and Schecter, 1983; Hofer, 1980; Robbins and Pearce, 1992). Consistent with previous studies, severity of decline has been captured by employing an indicator of financial health. Our sample includes mostly privately held firms as well as those from non-manufacturing industries. Therefore, the Altman Z-score (Altman et al., 2017) was selected, given its validity for both private and public firms, as well as for manufacturing and non-manufacturing ones. The Z-score was calculated in the year preceding the bankruptcy declaration. The model is defined as follows:

where

• X₁ = (Current assets − Current liabilities)/Total assets

• X₂ = Retained earnings/Total assets

• X₃ = Earnings before interests and taxes/Total assets

• X₄ = Equity/Total liabilities

Industry

The industry in which a bankrupt firm operates has commonly been used as a control variable, as this factor contributes significantly to the final outcome (Lim et al., 2013; Morrow et al., 2004; Robbins and Pearce, 1992). While some studies attempted to homogenize the industry represented by their samples (Bruton et al., 2003; Ndofor et al., 2013; Robbins and Pearce, 1992), others combined a variety of industries, and controlled for this variable (Lim et al., 2013; Morrow et al., 2004; Tangpong et al., 2015) in order to capture the differences between them. In the current study, this variable was captured as 3-digit SIC code dummies (López-Gutiérrez et al., 2012). We collected the main SIC code in the year prior to the bankruptcy declaration, i.e., 168 in total.

Territory

The location of bankrupt firms has a substantial impact on the outcome (Van Hemmen, 2009). Several factors linked to the territory, such as the experience or workload of the Commercial Court (Juzgado de lo Mercantil), and the idiosyncrasy of financial entities or concentration on determined activities (Mora-Sanguinetti and Spruk, 2018) are expected to affect the final solution. To account for these factors, 49 Spanish provinces (NUTS²-3) dummies were introduced in our analyses.

Year

The Spanish bankruptcy law was enacted at the end of 2004 and has underwent significant modifications ever since. In particular, 2011, 2014 and 2015 were years in which serious changes were introduced into the law in order to promote the survival of firms, which folded after the economic crisis (García-Posada and Vegas, 2016). Given the substantial modifications adopted by the Spanish bankruptcy law following its enactment, we controlled for the year in which a firm was declared bankrupt. This variable enabled us to control for any unknown or unobserved macro-level variable that may have influenced the bankruptcy outcome.

Results

The means of the variables involved in the analyses are evaluated in Table 1. With respect to size, both marginal survivors and successful firms are larger than firms in liquidation. In addition, a statistically significant difference was found among them. Successful firms and marginal survivors were also older than liquidating firms, and no difference was found between them. A statistically significant difference was found between the three groups in respect to stakeholder support. As expected, successful firms had a higher proportion of stakeholder support than marginally successful firms and firms in liquidation. No difference was found when considering severity of decline, which is likely due to their common experience of bankruptcy.

Table 1.

Mean comparison between bankruptcy outcomes.

Variable	Mean			t-statistics
	Liquidate	Marginal	Success	Liquidate vs marginal	Liquidate vs success	Marginal vs success
Size	14.77	16.08	15.79	−9.95^***	−5.41^***	1.72^*
Age	1.20	1.31	1.28	−5.78^***	−2.55^***	1.54
Debt	0.94	0.68	0.71	4.74^***	3.70^***	−0.75
Severity	2.22	3.78	3.56	−1.20	−0.62	0.35
Stakeholders	0.12	0.16	0.55	−1.57	−11.28^***	−9.68^***
Costs	0.56	0.50	0.56	2.37^**	−2.28^**	−4.08^***
Employees	0.47	0.32	0.36	5.71^***	2.96^***	−1.05
Assets	0.25	0.19	0.19	2.88^***	1.92^*	0.00

*

p < 0.10.

**

p < 0.05.

***

p < 0.01.

With respect to the measures adopted, Table 1 also shows significant differences between the three groups. Cost retrenchment intensity is significantly different between the three groups. Successful firms retrench more intensely than marginal survivors and liquidating firms, while liquidating firms reduce costs more intensely than marginal survivors. An inverse relationship was found for employees and assets. Successful firms and marginal survivors adopted weaker employee retrenchment measures, with no significant differences between them. Finally, successful firms implemented less intense asset retrenchment measures, and such measures were significantly weaker than those adopted by firms in liquidation. No significant differences were found between marginal survivors and successful firms.

As anticipated, the territory is expected to have a relevant impact on turnaround outcomes. Table 2 illustrates the liquidation, marginal survival, and success rates classified using Comunidad Autónoma (NUTS-2). Madrid, Cataluña, and Comunidad Valenciana account for more than half of the cases of bankruptcy (i.e., 498 out of 868 firms). The highest liquidation rates are observed in Baleares and Murcia (67.9%) followed by Comunidad Valenciana. These three regions were particularly affected by the restructuring of the banking sector. Cataluña shows a liquidation rate of 8.3%, which is well below the overall rate (41.1%). Navarra (80.0%) and Cataluña (73.1%) show the highest proportion of firms that survived marginally, while Baleares (21.4%) shows the lowest. Finally, Asturias shows the highest success rate (40.0%), and both Extremadura and Navarra had no successful firms in their territories.

Table 2.

Outcome rates by Comunidad Autónoma (NUTS-2).

Comunidad Autónoma	Liquidation	Marginal	Success	Number of firms
Andalucía	57.4%	27.7%	14.9%	47
Aragón	29.0%	64.5%	6.5%	31
Asturias	20.0%	40.0%	40.0%	15
Baleares	67.9%	21.4%	10.7%	28
Canarias	38.7%	38.7%	22.6%	31
Cantabria	33.3%	33.3%	33.3%	6
Castilla y León	50.0%	40.0%	10.0%	50
Castilla-La Mancha	41.7%	41.7%	16.7%	12
Cataluña	8.3%	73.1%	18.7%	193
Comunidad Valenciana	60.4%	24.8%	14.9%	101
Extremadura	42.9%	57.1%	0.0%	7
Galicia	46.6%	44.8%	8.6%	58
La Rioja	20.0%	60.0%	20.0%	5
Madrid	53.4%	33.3%	13.2%	204
Murcia	67.9%	17.9%	14.3%	28
Navarra	20.0%	80.0%	0.0%	5
País Vasco	38.3%	44.7%	17.0%	47
Number of firms	357	381	130	868

Moreover, the year in which a bankruptcy was declared was used as an expected relevant variable for the analyses. As shown in Table 3, the majority of bankruptcies were registered during the period 2011–2014 (574 out of 868), when the second recession took place in Spain. With the exception of the first years of the bankruptcy law (2004–2006 inclusive), in which a few cases are found, liquidation and survival rates have evolved showing opposite trends. Marginal survival represented 60% and 72% of outcomes in 2007 and 2008, respectively, but at the end of 2017, these firms represented only 17% of the sample. Similarly, successful firms represented 33% of outcomes in 2007 but just 8% in 2017. Conversely, liquidated firms represented 17% of the sample in 2008 and increased year after year to 75% in 2017. The sample may contain bankruptcies that commenced in recent years which have not yet reached an outcome, but the increase in the rate of firm liquidations is notably clear.

Table 3.

Outcome rates by year of bankruptcy declaration.

Year	Liquidation	Marginal	Success	Number of firms
2004	50%	50%	0%	2
2005	0%	100%	0%	1
2006	0%	71%	29%	7
2007	7%	60%	33%	15
2008	17%	72%	11%	36
2009	11%	64%	25%	61
2010	30%	56%	15%	61
2011	43%	41%	16%	102
2012	36%	49%	15%	146
2013	44%	43%	13%	206
2014	58%	28%	14%	120
2015	52%	36%	12%	58
2016	68%	20%	12%	41
2017	75%	17%	8%	12
Number of firms	357	381	130	868

Table 4 presents descriptive statistics that include mean values, standard deviations, and Pearson correlations. With the exception of the correlation between severity and debt (-0.79), there are no correlation or multicollinearity concerns, particularly as significant Pearson statistics are scarce and lower than 0.5 (Hair et al., 2006). In fact, the majority of correlations are observed for the dependent variable, as expected. The correlation between severity and debt is also expected, given the variables used for their calculation.

Table 4.

Descriptive statistics and correlations.

		Mean	S.D.	1	2	3	4	5	6	7	8	9
1	Outcome	1.74	0.70
2	ROA change	−0.39	4.06	0.09^***
3	Size	15.50	1.86	0.26^***	0.11^***
4	Age	1.26	0.29	0.14^***	0.01	0.24^***
5	Debt	0.79	0.72	−0.15^***	0.04	−0.22^***	−0.07^**
6	Severity	3.11	16.43	0.04	−0.07^**	0.00	−0.06	−0.79^***
7	Stakeholders	0.20	0.40	0.31^***	0.07^**	−0.01	0.05	0.12^***	−0.12^***
8	Cost	0.55	0.34	0.03	0.14^***	−0.05	0.04	0.10^***	−0.11^***	0.06
9	Employees	0.39	0.37	−0.15^***	−0.06^*	0.06	−0.07^**	0.01	0.00	−0.09^***	0.18^***
10	Assets	0.21	0.28	−0.09^***	−0.14^***	0.05	−0.05	−0.01	0.01	−0.13^***	−0.01	0.24^***

*

p < 0.10.

**

p < 0.05.

***

p < 0.01.

In line with the proposed model, we used the multinomial logistic regression (MLR) model to assess the effects of the independent variables on the probability that each of the three turnaround outcomes (i.e., liquidation, marginal survival, and success) would be achieved. This is a particularly appropriate and commonly used technique when firms are confronted with outcomes involving multiple options (Hair et al., 2006). In a multinomial logit model, one of the dependent variables is selected as a case of reference, or base case. In our study, three MLR analyses were conducted to compare the three outcomes of the bankruptcy procedure, the two first using liquidation as the base case, and the third using marginal survival as a reference. Non-ordered multinomial regressions were run, assuming that the dependent variable was not purely ordinal and outcomes were independent (Åstebro and Winter, 2012).

Table 5 illustrates the results obtained. To display the results of the MLR, six models were defined. Models 1, 3, and 5 include the prediction results when using only control variables, while Models 2, 4, and 6 include all of the variables. The MLR analysis also includes the marginal effects of the defined categories for the independent variables, thus ensuring that the contribution of each can be assessed. This reflects a more fine-grained analysis and permits more precise conclusions. The chi-square test shows that all the models are highly significant, and Models 2, 4 and 6 increased both the significance and accuracy of Models 1, 3, and 5. Model 2 has the higher explanatory power when compared to the other two outcomes (i.e., success vs liquidation and success vs marginal survival).

Table 5.

Multinomial logistic regression results.

	Marginal survival vs liquidation		Success vs liquidation		Success vs marginal survival
	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6
Stakeholder support		0.63^* (0.44)		3.73^*** (0.55)		3.10^*** (0.47)
Cost retrenchment		0.19 (0.43)		2.30^*** (0.69)		2.11^*** (0.65)
Employees retrenchment		−1.55^*** (0.43)		−0.61^* (0.58)		0.94 (0.55)
Asset retrenchment		−0.32 (0.53)		0.71 (0.72)		1.03 (0.67)
Size	0.66^*** (0.11)	0.62^*** (0.12)	0.59^*** (0.13)	0.62^*** (0.15)	−0.07 (0.12)	0.01 (0.13)
Age	1.26^** (0.55)	1.29^** (0.59)	0.85 (0.67)	−0.07 (0.76)	−0.41 (0.56)	−1.26^* (0.68)
Debt	−0.64^* (0.37)	−0.93^** (0.37)	−0.81^* (0.47)	−1.46^*** (0.53)	−0.17 (0.44)	−0.52 (0.48)
Severity of decline	−0.02 (0.01)	−0.03^** (0.01)	−0.03^* (0.02)	−0.02 (0.04)	−0.01 (0.02)	0.01 (0.03)
Constant	1.19 (2059.12)	1.06 (2084.46)	−27.01 (4365.56)	−29.91 (4599.41)	−28.19 (4108.83)	−30.97 (4099.95)
3-Digit activity dummies (168)	Included	Included	Included	Included	Included	Included
Territory dummies (49)	Included	Included	Included	Included	Included	Included
Year dummies (14)	Included	Included	Included	Included	Included	Included
Observation	868 (738)	868 (738)	868 (487)	868 (487)	868 (511)	868 (511)
Log-likelihood ratio	820.53	713.19	820.53	713.19	820.53	713.19
R^2 (Cox and Snell)	0.66	0.70	0.66	0.7	0.66	0.7
R^2 (Nagelkerke)	0.76	0.81	0.76	0.81	0.76	0.81
R^2 (McFadden)	0.53	0.59	0.53	0.59	0.53	0.59
Chi square	934.91^***	1042.24^***	934.91^***	1042.24^***	934.91^***	1042.24^***
Correctly classified	79.00%	85.90%	79.00%	81.90%	79.00%	82.30%

Note: Standard errors in parenthesis.

*

p < 0.10.

**

p < 0.05.

***

p < 0.01.

The left panel of Table 5 (Models 1 and 2) illustrates the comparison between liquidation (reference) and marginal survival. The estimated coefficients, significance, and standard error (in parenthesis) were calculated. As such, it was possible to interpret the magnitude of the relationship between the dependent variable and the independent and control variables. A positive sign indicates that the independent variables contribute to increasing the probability of marginal survival with respect to liquidation (i.e., reference variable). While size and age increased the probability of marginal survival;liquidation, debt and severityof decline reduced it. In terms of stakeholders, having the support of stakeholders significantly increased the probability of survival. With respect to intensity of response, the only variable that showed a significant, and negative relationship was retrenchment in employees.

The central panel of Table 5 (Models 3 and 4) presents the results of the MLR which compare liquidation (reference) and successful survival. In this case, the only statistically significant control variables were size, with the expected positive impact, and debt, which increased the likelihood of being liquidated rather than being successful. Having stakeholder support was clearly positive and significant, as hypothesized. In terms of managerial actions, intensecost retrenchment proved significant and positively contributed to success, while the opposite was found in the case of employees retrenchment. Asset retrenchment strategies had a positive but non-significant coefficient.

Finally, the right panel of Table 5 (Models 5 and 6) depicts the results comparing marginal survival (reference) and success. Among the control variables, only age showed a negative and significant impact on the likelihood of success compared to marginal survival. Stakeholder support showed the expected positive effect, while the only managerial action that influenced results was cost retrenchment, with a significant and positive coefficient.

Therefore, Hypothesis 1 (stakeholder support) was fully supported, as obtaining stakeholders’ support increased the likelihood of surviving and being successful in the case of a bankruptcy procedure. Hypothesis 2a (cost retrenchment) was partially supported, as it showed a positive expected effect when comparing success vs liquidation and marginal survival, though not between marginal survival and liquidation. Hypothesis 2b (employee retrenchment) was supported, as an expected negative impact was found when comparing marginal survival and success with liquidation, but not between the two survival outcomes. Thus, hypothesis 2b received partial support. Finally, Hypothesis 2c (asset retrenchment) was not supported in any case.

To test the robustness of the multinomial model, we ran an ordinary least-squares regression (OLR) using the ROA change as the dependent variable (Tangpong et al., 2015). The results are shown in Table 6. In this case, size remained the only significant control variable, while the rest became irrelevant. Stakeholder support also showed a positive but non-significant relationship with performance change. The intensity of retrenchment responses showed the proposed impact. Cost retrenchment clearly improved ROA, while excessively reducing employees and assets was detrimental to increasing performance.

Table 6.

Ordinary least squares regression results.

DV: ROA change	Model 7	Model 8
Stakeholder support		0.25 (0.32)
Cost retrenchment		1.76^*** (0.42)
Employees retrenchment		−0.54^* (0.36)
Asset retrenchment		−1.82^*** (0.50)
Size	0.27^*** (0.09)	0.54^*** (0.08)
Age	−0.31 (0.52)	−0.51 (0.49)
Debt	0.17 (0.36)	0.22 (0.31)
Severity of decline	−0.02 (0.01)	−0.02 (0.01)
Constant	−4.35^*** (1.51)	−4.44^*** (1.45)
3-Digit activity dummies (168)	Included	Included
Territory dummies (49)	Included	Included
Year dummies (14)	Included	Included
Observation	868	868
R^2	0.14	0.27

Note: Standard errors in parenthesis.

*

p < 0.10.

** p < 0.05.

***

p < 0.01.

In summary, the proposed hypotheses received substantial support. More specifically, stakeholder support (Hypothesis 1) was decisive in discriminating between the three outcomes, given that successful firms were supported more frequently than marginal survivors and liquidating firms. However, this variable was irrelevant in terms of explaining performance increases during the bankruptcy procedure. On the other hand, cost retrenchment (Hypothesis 2a) proved significant and positive when comparing successful firms with both liquidating ones and marginal survivors, and it was also a critical factor in contributing to ROA increases. Hence, deep cost retrenchment is required for a successful turnaround during a bankruptcy procedure. Employee reduction (Hypothesis 2b) increased the likelihood of being liquidated with respect to the other two outcomes, and also decreased performance. Asset retrenchment (Hypothesis 2c) was a non-significant variable in bankruptcy outcomes, but negatively impacted performance. Thus, it can be stated that having stakeholder support is necessary to survive a bankruptcy procedure, but an intense cost reduction contributes the most to improving performance. Deep employee retrenchment measures push firms towards liquidation and diminish their performance, while substantial asset retrenchment does not affect the probability if survival, though diminishes performance. Structural variables that consistently affected the outcomes included size (positive) and debt (negative). Overall, the results revealed that in turnaround studies in the case of bankruptcy, it is worth distinguishing between at least three outcomes, given the differences that aroused in the analyses.

Discussion

Implication for theory and practice

This research aimed to explain the effectiveness of retrenchment strategies in a regime with a low survival rate, such as that shown by the Spanish bankruptcy procedure, and seeks to offer guidance to legislators, academics, and practitioners in relation to the issues that threaten the future survival of firms. In light of this aim, a sample of 868 bankrupt Spanish firms were observed throughout the bankruptcy procedure until an outcome was reached. This study examined the intensity of retrenchment responses and the impact of stakeholders, while transcending the traditional turnaround dichotomy by evaluating three proposed outcomes. These three outcomes facilitated a more fine-grained analysis which allows the introduction of more nuanced conclusions.

In particular, the results show that the stakeholder support variable offers a better explanation of the survival and success of bankrupt firms. The influence of stakeholders becomes so significant that obtaining their support can be enough to ensure the survival of a firm. However, stakeholder support is not enough to yield superior results.

In respect to the responses to decline, the results show that only cost retrenchment represents an advisable measure to encourage survival and improve performance in a bankruptcy procedure. Stabilization of the crisis is key to surviving such severe conditions and cost retrenchment, despite being a harsh measure, is a decisive contributor to achieving this aim. This confirms the findings of other turnaround scholars who studied mature and declining firms, but also elaborates upon their findings by examining a neglected context (bankruptcy). Conversely, excessive employee redundancies have detrimental effects in terms of both the probability of survival and subsequent performance. While reducing the number of employees is very often adopted as a retrenchment measure, it is not recommended during bankruptcy. This may indicate that employees of bankrupt firms suffer from additional tensions that may become more serious when the threat of redundancy is near. As hypothesized, the immediate cost savings are not enough to outweigh the detrimental impact that results from massive job losses. This contributes to existing findings in the turnaround literature regarding reductions in human resources, which generally pointed to an overall negative impact. Finally, the results revealed that asset retrenchment measures were irrelevant during bankruptcy, but had a clear negative effect on performance increases. While previous scholars signalled that the selling of assets was largely necessary in extremely severe situations, the bankruptcy context showed that it might be worth retaining rather than disposing of them. This could be due to particularities of the bankruptcy context, for which one of the key principles of survival is the protection of assets in order to continue running the business.

Thus, the low survival rates of the Spanish bankruptcy procedure are due to the adoption of inappropriate retrenchment strategies, such as massive employee layoffs, hastily selling valuable assets, and ignoring critical operational costs. Moreover, bankrupt Spanish firms rarely attempt to realign their stakeholders’ expectations. Consequently, this implies that survival and success are extremely rare events in the Spanish bankruptcy procedure.

It can be concluded that successful firms that entered the bankruptcy procedure included firms that had obtained the support of their stakeholders, implemented intense cost retrenchment measures, and safeguarded their employees and most valuable assets. As a general recommendation, we can assert that the success of a bankrupt firm depends upon the implemented strategies, not upon other structural or uncontrollable factors.

This research outlines relevant contributions and implications for academics, legislators, and practitioners. First of all, from the perspective of the turnaround literature, this study contributes to the contingency theory by drawing a link between bankruptcy and turnaround. Considering that bankruptcy is the most extreme and severe crisis that a firm may face, we challenged the validity of retrenchment as a cure-all for firms suffering such a crisis (Robbins and Pearce, 1992). The study also contributes to the discussion of retrenchment efficacy, which has not proved homogeneous depending on the strategic content (Schoenberg et al., 2013; Tangpong et al., 2015) or the context (Morrow et al., 2004; Ndofor et al., 2013). While cost retrenchment was associated with both survival and turnaround, as proposed by previous authors (Hofer, 1980; Robbins and Pearce, 1992), layoffs even reduced the likelihood of survival and success (Tangpong et al., 2015). Asset retrenchment was not associated with any outcome, contrary to the suggestions of Bruton et al. (2003). These findings propose that bankruptcy is substantially different from the common contexts in which turnarounds are studied, and further contributions can be expected from this field given the challenges that bankrupt firms face when attempting a turnaround.

The findings also provide legislators with greater insight into how bankruptcy regimes may be regulated in order to increase their efficiency by encouraging the survival of firms. This research has developed profiles of liquidated firms, marginally survivors, and successful firms. What is the difference between them? In addition to a rapid response, such as an improvement in the stakeholder relationship and resolute cost-cutting, legislators should ensure that bankrupt firms raise enough cash, and safeguard their employees and most valuable assets. Despite the fact that the turnaround literature advocates that retrenchment should be adopted in its most diverse forms, an excessive reduction in employees proves detrimental to the firm's survival, and indiscriminate asset disposals decisively reduce performance.

At the managerial level, the study highlights several valuable lessons. On the one hand, the research proposes that recognizing and diagnosing the crisis is critical for subsequent success. An erroneous diagnosis delays the response and, thus, adversely affects the probability of initiating the turnaround process. On the other hand, the intensity of response has been identified as a relevant variable that should be taken into account. The success of a bankruptcy procedure is dependent upon an adequate adjustment of the intensity, content, and stakeholder support. In order to increase the probability of survival for a bankrupt firm, this interdependence must be reflected in the strategic decision that is adopted. How can managers gain the support of stakeholders? A firm's management should focus on reducing redundant costs in order to generate the necessary cash flow to ensure the firm's immediate stabilization and viability. A reduction in superfluous expenses sends a clear signal to stakeholders that the firm is committed to meeting their expectations.

Limitations and future research

Despite providing several contributions, this research has some limitations. First, it is limited to the Spanish bankruptcy context, which is relatively new and has relevant particularities. As time passes, a larger number of bankrupt firms will integrate the Bankruptcy Public Register and such firms could be included in a similar research study in order to evaluate the validity of the findings. As such, this research may offer the basis for a wider longitudinal study examining firm failure and turnaround in Spain. Moreover, the sample employed in this research could be observed in the future in order to assess the evolution of the firms. Thus, far more scenarios could be researched. For example, future research could examine surviving firms that cannot continue after having survived the procedure and instead liquidate, or marginal survivors that improve their performance in subsequent years and manage to safeguard their survival and repay their debts. Firms that show a decline prior to becoming bankrupt or firms that cease trading without even attempting a turnaround are of interest, and some of the principles identified in this research could be applied to these cases. Therefore, as a result of these contributions, the fields of turnaround and bankruptcy are definitively united.