Employment protection legislation, labor courts, and effective firing costs

Abstract In many countries, labor courts play a central role in the determination of firing costs by monitoring and supervising the procedures for dismissals, and, eventually, deciding severance payments mandated by the employment protection legislation (EPL). To get some insights about the impact of labor courts on effective firing costs, we explore a new database that contains information on labor courts’ intervention in firings before and after the implementation of significant EPL reforms modifying severance payments and procedures for dismissals. Our results suggest that labor court rulings on economic dismissals did not fully translate the reduction of firing costs mandated by the new EPL to effective firing costs.


Introduction
Labor market institutions are considered key determinants of the functioning of the labor market and, hence, of economic performance (Blanchard and Wolfers, 2000, Botero et al., 2005, Turrini et al., 2014, OECD, 2019. A less studied issue is, however, how enforcement of regulation by the judicial system affects labor market outcomes. In the case of employment protection legislation (EPL), labor courts play a central role.
EPL mandates procedural rules for the resolution of firings. These rules create some room for strategic behavior by employers and fired employees, which is even more pronounced when, as happens in many countries, there is the possibility of settlement before labor courts intervene. Hence, the intervention of labor courts determines effective firing costs, both for cases settled out-of-the courts and those ruled by judges (see Daughety and Reinganum, 2012).
In sum, effective firing costs depend both on severance payments established by EPL and costs associated with litigation.
Judges are neither neutral nor unbiased agents, but rather they seem to decide on labor conflicts with some "social motivation" (Bornstein and Miller, 2009;Posner, 2010;Feld and Voigt, 2003;Muñoz Aranguren, 2011;Danziger et al., 2011), which widens the scope for strategic behavior by employers and dismissed employees. For instance, in Italy, there is some association between local labor market conditions, such as the unemployment rate, and labor courts' decisions (Macis, 2001, Ichino et al., 2003; in Germany, even after controlling for the fact that court activity varies systematically with the political leaning of the government that appoints judges, there is a significant positive relation between court activity and unemployment (Berger and Neugart, 2011); in the UK, unemployment and firms' bankruptcy rates seem also to be statistically associated with the probability of judges deciding in favor of dismissed employees (Marinescu, 2011); and in Spain, labor courts' rulings in dismissal conflicts are similar across neighboring provinces suggesting that judges are subject to some "peer effects" (Martín-Román et al., 2015).
With the goal of reducing firing costs in Spain, two EPL reforms in 2010 and 2012 changed procedural rules for economic dismissals (mainly by widening the definition of "fair economic dismissals") and reduced severance payments for unfair dismissals. 1 The aim of this paper was to gauge to what extent changes in firing procedures mandated by these EPL reforms translated into changes in the effective costs of economic dismissals.
More broadly, the paper provides some empirical evidence to the view that the effects of EPL reforms aimed at reducing firing costs may depend not only on mandated severance payments but also on labor courts' rulings on firing conflicts and expectations of the parties on these rulings. We do so by performing comparisons of the likelihood of settlements and labor courts' rulings before and after the EPL reforms. Both differences are estimated conditioning on a set of covariates, mostly related to local labor market conditions, which control for the incidence and the selection of dismissal cases brought to labor courts' resolutions. We find that (i) settlements increased after the reforms (in particular, after 2012) and (ii) despite the widening in the definition of fair economic dismissals, the probability of a labor court ruling a firing as fair did not significantly increase. Therefore, the main effects of the Spanish EPL reforms on effective firing costs took place through the 1 Labour Reform Law of 2010, Royal-Decree-Law of 2011, and Labour Reform Law 2012. reduction of severance payments, not as a consequence of the changes in procedural rules for economic dismissals.
First, we describe the particular features of Spanish EPL and its reforms in 2010 and 2012, and advance some propositions regarding the implications of the EPL reforms for settlements and labor court decisions (Section 2). Empirical analysis of these implications is given in Section 3. In the Appendix, we lay off a theoretical model of the determinants of effective firing costs that illustrates the main mechanisms we have in mind to rationalize propositions and to interpret empirical results. Finally, Section 4 concludes.
2 EPL: severance payments and procedural rules

Institutional framework
In Spain, terminations of regular employment contracts have to be justified either by the lack of performance of the employee (disciplinary layoffs) or by economic, technological, or organizational reasons (economic dismissals). Before 2010, economic dismissals were considered to be justified only in very restrictive cases and exclusively as a measure of last resource. A fair economic dismissal required the existence of negative economic conditions, but the law did not clarify what "negative conditions" meant, so the ambiguity of the definition gave judges a great deal of discretion. In 2010, there was a substantial extension of the conditions under which economic dismissals were justified, such as the incidence of current or anticipated losses, and a persistent decline in revenues that could jeopardize either firm's economic viability or employment. The labor market reform of 2012 made further progress on the clarification of what negative economic causes meant referring to situations in which "for three consecutive quarters the level of revenues or sales of the company was lower than in the same quarters of the previous year".
On the other hand, disciplinary layoffs are considered to be fair only in the cases of misconduct or lack of adaptation of the employee to the job tasks. Spanish EPL reforms in 2010 and 2012 did not substantially change the definition of fair disciplinary layoffs.
Fired employees may appeal to a labor court. In contrast with other countries, Spanish judges are not entitled to establish severance payments according to the characteristics of each individual case but only to declare the dismissal fair, unfair, or null. An out-of-court settlement stage prior to the file of the claim at the labor court is compulsory.
Employees dismissed under fair economic reasons are entitled to a severance pay of 20 days' wages per year of seniority, with a maximum of 12 months' wages. Employees laid off for fair disciplinary reasons do not receive severance pay. EPL reforms in 2010 and 2012 did not change the amount of severance pay for fair dismissals. If either the economic dismissal or the disciplinary layoff is ruled out unfair, firms either had to pay 45 days' wages per year of seniority with a maximum of 42 months' wages or had to reinstate the worker. After 2012, severance pay for all unfair dismissals/layoffs was reduced to 33 days' wages per year of seniority with a maximum of 24 months' wages.
Some employees (i.e., pregnant employees, employees enjoying reduced working time in order to take care of a child, trade union officials, employees who have filed a claim against the company, among others) are further protected against unfair dismissals/layoffs, so that firing them could be declared as null/void, and the employees are entitled to reinstatement and interim wages (those corresponding to the period between the dates of dismissal and reinstatement). Moreover firings may be ruled as null if there is discrimination (a violation of the fundamental rights of the employee) or breach of union rights. Hence, the termination of employment contracts under these circumstances implies specific negotiations, which can result in the payment of very high compensations. 2 Nevertheless, there are few cases of this nature brought to labor courts (see Palomo Balda, 2013).
Before 2012, the employer was entitled to dismiss an employee recognizing unilaterally its "unfairness" (termination without cause). Thus, termination was effective on the same date the dismissal was initiated and after the payment of the full severance for unfair dismissal. Hence, labor authorities (either judicial or administrative) did not intervene (this was usually called "express dismissal", despido exprés in Spanish). 3 The cost advantages of the so-called express dismissal for the employer were twofold. First, it avoided the red tape costs of legal proceedings. Second, it eliminated the payment of interim wages (salarios de tramitación) when the labor court ruled the dismissal unfair. After 2012, the so-called express dismissal was eliminated. Thus, the labor court intervention could no longer be avoided (in case of no out-of-court settlement), and interim wages for unfair dismissals and layoffs were reintroduced. 4 Figure 1 sketches these firing regulations. Since the legal procedures for disciplinary layoffs were simpler and severance pay in case of unfair dismissal was the same than under economic dismissals, employers most frequently initiate firings alleging disciplinary causes.
During 1984-2010, about 70% of dismissal cases resolved by labor judges' rulings were declared unfair, with only a few of them being declared null.
It is also important to bear in mind that there are two alternative ways to terminate an employment contract besides individual firings. Since 1984, Spanish policy-makers, facing strong opposition to change EPL under regular employment contracts, introduced employment flexibility at the margin by creating a wide array of "atypical" contracts. 5 Regulation of these types of contracts changed several times and in fundamental ways, but segmentation between permanent and temporary employees, which began in the late 1980s, has prevailed since then. While regular employees are entitled the right to go to court to appealing the cause of the dismissal and may get higher severance payments in the case of unfair dismissals, temporary employees did not have the right to appeal the termination of their contracts. Hence, employers use fixed-term contract and other kinds of temporary contracts (nowadays amounting to more than 25% of employment) to buffer against negative shocks leading to downsizing of their labor force (Costain et al., 2010). Additionally, economic dismissals may be implemented collectively, and it is obliged to do so when they affect to more than 10% of the firm's labor force in a given quarter. Firing costs under collective dismissals are typically higher than for individual dismissals/layoffs.
2 In order to avoid legal proceedings, it is common to reach agreements of severance compensations between 50 and 60 days of salary per year of service, which are much higher than the ones stated by EPL. 3 Nevertheless, the dismissed employee could still challenge the employer's decision before labor courts by claiming that the dismissal was null/void. 4 If the court ruling is notified exceeding 90 days since the lawsuit has been filed, the employer may claim from the Spanish Government the reimbursement of interim wages corresponding to the excess of that period (

Some hypotheses on the effects of EPL reforms on litigation
Given the Spanish institutional framework, employers take three decisions when considering firings: (i) when to initiate a firing, (ii) whether to justify the firing as an economic dismissal or as a disciplinary layoff (notice that the employer could initiate a firing as a disciplinary layoff even if the true cause is economic and vice versa; we will refer to these cases as disguised dismissals), and (iii) under what circumstances to reach a settlement before the labor court ruling. Similarly, the dismissed employee also has to decide whether to reach a settlement or to litigate. 6 Finally, judges rule those cases that are not settled following the EPL mandate. For employer and employee decisions, expectations about the sign of labor court rulings (fair or unfair) play a crucial role. For the employer, these expectations determine when a settlement is less costly, the relative cost of initiating the firing as a economic dismissal versus as a disciplinary layoff, and, hence, effective firing costs, which, themselves, determine when to initiate a firing. Similarly, the expectations of the fired employee on the labor court ruling determine his or her acceptance of a settlement.
There is a large literature on settlements and litigation, developed after the seminal work by Priest and Klein (1984) who argued that, because of selection effects, the percentage of litigated cases won by plaintiffs will not vary with legal standards. Thus, EPL reforms would not have any effect on the proportion of labor court ruling in dismissals/ layoff conflicts. However, a more formal analysis rejects the so-called "No Inference Hypothesis": Klerman and Lee (2014) concluded that "even taking selection effects into account, one may be able to make valid inferences from the percentage of plaintiff trial victories, because selection effects are partial". They also proved that, under plausible conditions, a change in the law ought to increase labor court rulings in favor of the party that wins more from it. Hence, changes in legal standards affect both the incentives to litigate and the expectations of the agents of outcomes of litigation, but not the extent to eliminate any effect on labor court rulings.
In the Appendix, we formally lay out a simple model of firing conflicts, similar to Klerman et al. (2018)  In what follows, we turn to the data available to provide either confirmation or rejection of propositions above. Source: Authors' own elaboration. Notes: a. Out-of-court settlements are resolved in Spain by the "MAC" units ("Mediation, Arbitration and Conciliation Units"). The majority of out-of-court settlements resolved with an agreement between the employer and the employee end up with the effective firing of the employee. Settlements ended without an agreement are the main group of dismissal conflicts which arrive to the labor courts. Following the data of the Ministry of Employment and Social Security, there was a total of 220,095 out-of-court settlements in 2014, of which 101,426 ended with agreement between the employer and the employee. b. In 2014, the number of dismissals resolved at the labor court was 118,225. This amount is calculated by adding the number of pre-trial settlements with agreement, the dismissals finally ruled by a labor court, and the number of cases withdrawn (including tacit withdrawals and voluntary dismissal of action by the parties). c. The number of pre-trial layoff settlements in 2014 was 48,508. d. In 2014, the number of dismissals resolved at the trial level in the labor courts was 42,992, of which a 78% were dismissals ruled as "unfair" (in favor of the employee). e. In 2014, 26,725 dismissals were withdrawn (thus, they were not resolved by a judge in a trial) as a result of formal failures, tacit withdrawals, and voluntary dismissal of action by the parties.

EPL reforms, settlements, and labor court rulings
dismissal or a disciplinary layoff. Information on employer/employee's characteristics is not available. As for labor courts' characteristics, we observe the type of judge ruling on the dismissal conflict, that is, whether he or she is assigned to a particular court or appointed as a temporary replacement, reserve or substitute of the former. Additionally, we also compute a measure of congestion at labor courts. 11 As for settlements, they take place at two stages. First, there is an out-of-court settlement stage that is compulsory at the so-called mediation, arbitration and conciliation units before the file of the claim in the labor court. Second, settlements may also occur at the labor courts before the judge's ruling. For each labor court, we observe a settlement ratio computed as the proportion of settlements within all the firing cases brought at labor courts.
Since local labor conditions affect the number of firing conflicts and labor court rulings (see the Appendix), we also collect information on the provincial unemployment rate  in 2010-2012 and 22.7% after 2012, with noticeable differences across provinces (see Figure 3 and Table 3). Also interestingly, this proportion is negatively correlated with the local unemployment rate (see column 3 in Table 5). We interpret this correlation as an indication that the local unemployment rate affects directly labor court decisions (see Appendix).

Empirical approach
Variation across time and labor courts with controls by labor court and provincial characteristics allows us to make inferences on the effects of EPL reforms on labor court rulings and on the incidence of settlements.
As discussed above, labor market reforms of 2010 and 2012 affected red tape costs and procedures of economic dismissals and disciplinary layoffs differently. Hence, the incidence of economic dismissals and disciplinary layoffs and the selection of both conflicts into settlements and litigation are likely to have changed as a result of the reforms. Given that we do not observe economic dismissals and disciplinary layoffs separately, we perform an event study (comparisons before and after) with a set of controls that proxy the incidence and composition (economic dismissals versus disciplinary layoffs) of firings. Thus, we regress the ratio of labor court rulings stating that the dismissal/layoff is fair and the proportion of settlements on time dummies that capture the entry into force of the EPL reforms and a group of relevant controls that vary by time (at the quarterly frequency), by labor court, and by province (also including fixed effects by province). All models are estimated both fitting a linear specification and odds-ratios (Bishop et al., 1975;Williamson et al., 1995). 14 Errors in all models are clustered at the provincial level, and standard errors are robust to heteroskedasticity and autocorrelation.
Among the covariates, we specifically focus on the local (provincial) unemployment rate and the local (provincial) profitability of firms (the two variables that may directly influence judges' decisions on economic dismissals according to our model), and interact both with the time dummies indicating the timing of the reforms. These interactions should capture by how much judges' discretion on rulings changes with the EPL reforms.   Less than More than 31% 23% -31%

Less than
More than 31% Source: Authors' own elaboration based on data provided by the CGPJ.

Notes:
a. Unemployment rate (national value) shows the average rate of unemployment for the full period. Provincial values represent differences from that national average.
where Y ist is, alternatively, the settlement ratio and the proportion of labor court rulings declaring the dismissal/conflict fair at labor court i province s, and time t; X kist is the set of controls that includes the proportion of employees with temporary contracts, the sectoral distribution of employment (agriculture, manufacturing, construction, and services), the proportion of establishments with more than 200 employees, the number of employees affected by collective dismissals (in logs), the proportion of dismissal conflicts ruled by professional judges, days covered by temporary judges at the labor court per year, and the judicial congestion rate at labor court i, province s, and time t; and U st and ∏ st are, respectively, the unemployment rate and the lowest decile of the distribution of firm profit rates 15 at province s and time t. D 1 and D 2 are time dummies that take value one after the second quarter of 2010 and the first quarter of 2012, respectively, the dates at which reforms came into effect. Regressions also include fixed province effects, a s . Controlling by characteristics of labor courts (status of the judge, interim days at the labor court) should take care of changes that might have affected the rulings other than the EPL reforms.

Results
The main results are displayed in Tables 4 (settlements) and 5 (likelihood of a labor court ruling declaring the dismissal/layoff as fair).
While the 2010 reform led to some significant decrease in the likelihood of settlements, the 2012 reform had the opposite effect. Overall, the proportion of settlements is about 7% to 10% points higher after 2012 (columns 5 and 6). Settlements are less likely the higher the local unemployment rate and the lower firm profitability are. Interestingly, the association between the incidence of settlements and local unemployment became positive after the 2010 reform and negative again after the 2012 reform. Under our interpretation of the coefficients of these variables as the divergence between employer and employees expectations on labor court rulings, these results suggest that the reforms reduced this divergence both overall and, particularly, when local labor market conditions and firm profitability were worse. 16 Another conclusion from the estimated changes in the likelihood of settlements is that dismissal conflicts being solved by labor court trials after the reforms are those in which employers' expectations on the probability of a fair ruling increased by more and above dismissed workers' expectations. We cannot see why this should happen in the case of disciplinary layoffs (whose regulation was unchanged). Thus, by joining this to the observation that the overall proportion of firings initiated as economic dismissals was higher after the reform, we conclude that, if anything, the proportion of economic dismissals out of all firing conflicts solved by labor courts ought to have increased. An increase in the weight of economic dismissals being solved by trial in the labor courts and the broadening of the definition of fair causes of economic dismissals should weight positively in the likelihood of firings being ruled as fair by the labor courts.
However, Table 5 shows that the proportion of dismissals/layoffs being ruled as fair increased immediately after the 2010 reform but was not very much changed with the 2012 15 ROA according to the Bank of Spain database. 16 It is also noteworthy that the judicial congestion rate increased the likelihood of settlements (not shown in Table 4).  reform. In fact, under the most complete specifications (columns 5 and 6), neither of the two reforms seem to have a significant effect on the rulings. An increase in the unemployment rate of 10% points is associated with a decrease in the proportion of fair rulings of approximately 3% to 5% points. As for firm profitability, there is no statistically significant association with judges' rulings, once that other covariates controlling for incidence and composition of firings are included Nevertheless, after the 2010 reform, it seems that judges' decisions took more into account the economic situation of the firm, so that lower profitability led to a higher probability of a fair ruling (effect that is not observed after the 2012 reform, somehow surprisingly).
To better gage the impact of the EPL reforms on labor court rulings, in Figure 4, we plot the observed probability of a dismissal/layoff being ruled as fair by the labor court and the change in that probability due to EPL reforms and their effects through interactions with the local unemployment and firm profitability (using estimates from column 5 in Table 5

Concluding remarks
Labor courts' intervention on dismissal cases is key for the determination of effective firing costs. Since judges often behave as socially motivated agents and have some discretion in the application of EPL, the parties (employers and dismissed employees) act strategically taking into account the procedural rules for the initiation and resolution of dismissal conflicts. As a result, there are several channels by which EPL affects effective firing costs and the consequences of EPL reforms may be different than intended.
We analyze two significant EPL reforms in 2010 and 2012 than changed both severance payments and procedural rules in Spain to make economic dismissals less costly. Even though the proportion of economic dismissals over all firings increased, the average probability that a dismissal was declared fair by a labor court did not increase significantly, despite the widening of the fair causes of economic dismissals. By controlling for local labor market conditions, diminishing firm profitability, and reduction of severance payments for unfair dismissals, we identify the effects of EPL reforms on labor courts' ruling on firing conflicts. We conclude that the reduction of effective firing costs in Spain after 2010 took place mainly because of the lower severance payments for unfair dismissals and less so due to the extension of the fair causes of economic dismissals.
This conclusion has three implications for the policy debate on the need of introducing further labor market reforms. One is that the reduction in effective firing costs has been lower than the one intended by the legislated EPL reforms. Second, and similarly, the changes in the indicators about the stringency of EPL for regular contracts usually discussed in the debate (for instance the OECD indicators) overestimate the impact of the EPL reforms, since they are based on changes in legal costs and neglect the costs from enforcement (i.e., labor courts'  intervention). Finally, the reduction of effective firing costs for economic dismissals under the regular employment contract has not been as large as envisioned by the policy-makers. Given that the difference between these firing costs and termination costs of temporary contracts, which determines the proportion of employees with fixed-term contracts, is still very large, the high incidence of temporary employment observed in Spain would not be very much reduced by these reforms.

Availability of data ad materials
The data set is partially available from the corresponding author under reasonable requests.

Figure 4
Marginal effect of reforms, unemployment, and profitability on labor court rulings declaring dismissals/layoffs as fair.
Note: Proportion of fair rulings by labor courts (observed values) is measured in left axis and marginal effect (from estimates in column 5 in Table 5) on the right axis.
layoff being ruled as fair by the labor court does not depend on firm profitability but only on local labor market conditions: For untruthful dismissals, x d = x r = 0. There are red tape costs (court costs to be paid only by employers) that are higher for economic dismissals than for disciplinary layoffs (t j , j = {r,d}, with τ τ > r d ).
Severance payments are as follows:

A.2 Settlements
Settlements may arise after several rounds of proposals by the employers about severance payments and workers' counterproposals. 23 Notice that in the case of economic dismissals, the worker's expectations of a fair ruling is also affected by the signal sent by the employer about 21 For a model of dismissal conflicts with imperfect monitoring, see Galdón-Sánchez and Güell (2003). 22 That dismissed workers do not observe firm profitability is the source of asymmetric information in the sequential settlement game. Employers' uncertainty on the probability of a fair ruling may arise from judges being of different types regarding their leaning toward social motivations. 23 Alternative models of settlements are presented in Daughety and Reinganum (2012).

Table A1
Severance payments in the theoretical model

A.4 Some information on the social motivation of judges acting in the labor jurisdiction
Information on the social motivation of judges is limited. We proxy it by gathering information on memberships of the associations of judges. While Spanish law prohibits a judge to join a political party or a trade union while he or she is on active duty, it allows association in professional groups, which happen to have some "ideological" orientations. Currently, the major associations in Spain are the "Asociación Profesional de la Magistratura" (APM), "Jueces para la Democracia" (JpD), "Asociación Francisco de Vitoria" (FV), "Foro Judicial Independiente" (FJI), and the "Asociación Nacional de Jueces". APM is perceived as conservative. JpD is considered as leftist, and FV and FJI are considered as "moderate".
Since we have not obtained information about the association of lower level judges to professional associations (and, therefore, no control has been included in the regressions), we have built this measure for Supreme Court justices. In this regard, we observed the relative weight of justices of different judicial associations at the fourth chamber of the Supreme Court (the one that settles employment and social security conflicts). This may be representative of the ideological orientation of first-instance labor courts for two reasons. First, the presence of the various associations in the Supreme Court may be the direct consequence of the presence of these associations at lower levels. Second, Supreme Court's decisions (jurisprudence) are compulsorily followed by judges at lower levels for the interpretation of the law, so that the survival of lower court decisions depends partially on their consistency with the Supreme Court's decisions. As shown in Figure A1, the relative weight of magistrates pertaining to leftist associations gradually increased at the expense of conservative and moderate associations in the central years of our research (from 2005 until 2011) and maintained their supremacy in the chamber until 2014. It is conceivable that this trend was associated with an increase in the propensity to rule dismissal cases in favor of the employee, counteracting therefore the impact of the widening of fair causes for economic dismissals. If this were the case, the impact of the EPL reforms on labor court rulings should be upgraded by the change in the social motivation of judges taking place through the higher weight of leftist judges.

Figure A1
Proportion of magistrates belonging to professional associations.
Note: Magistrates of the fourth chamber of the Supreme Court. Source: Authors' own elaboration using Memorias del Tribunal Supremo.