آیا قضات متعصب از شرایط بازار کار هستند؟

When a firing litigation is taken to court, only the characteristics of the employee's misconduct should be relevant for the judge's decision. Using detailed data from an Italian bank and aggregate macro data, this paper shows that, instead, local labor market conditions influence the court's decision: The same misconduct episode may be considered sufficient for firing in a tight labor market but insufficient otherwise. We reach this conclusion after taking carefully into consideration the non-random selection of firing litigations for trial. Although these results refer to the specific situation considered, they raise more general issues. For macroeconomists they suggest that higher unemployment rates may increase firing costs via the effect on courts’ decision criteria; thus, the real extent of firing rigidities cannot be assessed without considering the role of courts. For labor law scholars, these findings are important because, following traditional principles, the law should be applied in the same way for all citizens and over the entire national territory.

The nature and relevance of firing regulations in the labor market differ sharply across countries, as recent surveys show (OECD, 1999). The first ground to evaluate these rigidities lies in the laws which specify the rights, duties and constraints that the employer and the employee must respect when dealing with a firing decision. However, the law is only one important component in the whole story, because, when a fired worker files the case in a court, the effective judicial enforcement of such rules plays a crucial role as well. From this viewpoint, it is important whether the statements of the law define a narrow grid of prescriptions which call for an almost automatic and mechanical decision by the judge, or if they are very general, leaving room for a wide range of possible interpretations. The role of judicial enforcement is much more relevant in this latter case, since the headings under which a firing decision can be taken must be filled with the interpretation prevailing in the jurisprudence and with the discretion left to the judge, who can confirm or overrule the firm's decision. The degree of effective labor market rigidity can therefore be assessed only when enforcement is considered together with the legal framework. Assessing judges’ decision criteria seems particularly relevant, from a macroeconomic point of view, if the legal standard applied is in turn influenced by labor market conditions. The fact that higher unemployment rates may induce judges to be more favorable to workers in case of a firing litigation creates the conditions for a potential reversed channel of causation between unemployment and firing costs. It is well known that higher firing costs generate longer unemployment spells and may also increase unemployment levels if they reduce hiring more than they prevent firing, given wage rigidity.1 But if at the same time higher unemployment rates increase firing costs via the effect on courts’ decisions, multiple equilibria may arise: On the one hand, equilibria with low unemployment and low firing costs, because courts tend to decide in favor of firms; on the other hand, equilibria with high unemployment and high firing costs, because courts tend to decide in favor of employees. In this paper we investigate whether judges are biased by labor market conditions and we reach the conclusion that such a bias exists. Using detailed micro data from a large Italian bank, we show that in regions where unemployment is high the gravity of the cases of misconduct going to trial tends to be higher than in the other regions. Therefore, if judges decided only on the basis of the gravity of misconduct, the probability of a pro-firm decision should be larger where the unemployment rate is higher. We find, instead, the opposite result: judges are less likely to decide in favor of firms where unemployment is high, despite the fact that more serious misconduct cases are brought in front of them. Therefore, we conclude that worse labor market conditions induce judges to be more favorable to workers. Our empirical analysis is a good case study to address the question we are interested in because Italy is a perfect example of a very rigid legislation that leaves wide scope for judicial discretion in the enforcement phase. Moreover, our company data allows us to reconstruct the entire selection process that generates a trial from a misconduct case and this is of crucial importance to derive conclusions on the criteria followed by judges in deciding the cases brought to court. If a worker files a firing decision before the judge, the latter will compare the specific nature of the case with some legal standard that summarizes his interpretation of the law, and will decide in favor of the firm or the worker. Hence, by considering the decisions of the judges, it may seem possible to reconstruct empirically the criteria followed to enforce the firing regulation. However, as Priest and Klein (1984), Waldfogel (1995) and Eisenberg and Farber (1997) have stressed in different contexts, such inference must take into account that the cases reaching the judicial stage are not representative of the entire population of cases. In our situation, this selection bias derives from the fact that a case filed before the judge originates from a decision of the firm to fire a worker (instead of keeping him or her on payroll), which is followed by the choice of the worker to go to trial (instead of accepting the firing and searching for a new job): These two previous steps are taken by the firm and the worker considering the likely outcome of the entire process, which depends on the nature of the case itself, on the judge's preferences, and on the alternative options that at each step are feasible. Hence, an empirical evaluation of judges’ preferences must take into account this source of selection bias, and possibly try to reconstruct the whole process instead of considering just the final judicial stage. For this reason, aggregate evidence on the final decisions of judges based on macro data would not be sufficient alone to draw conclusions on the existence of a judicial bias. This paper contributes to the literature also by showing how to use micro and macro data jointly to investigate the existence of a judges’ bias. Related to our results are the findings of Donohue and Siegelman (1991) who show that the number of anti-discrimination cases filed to court in the US increases with the worsening of labor market conditions, suggesting that anti-discrimination laws are becoming a tool to protect the existing position of incumbent workers. This is an example similar to ours of the endogeneity of firing costs induced by the judicial system because of the interference of labor market conditions. An example of effects going in the opposite direction is instead offered by Autor (2003) who shows that the recent growth of Temporary Help Agencies in the US is a causal effect of the decline of the “Employment at Will” doctrine in the legislation and jurisprudence of US states. In this case, it is the judicial attitude towards firing litigations that modifies labor market equilibria. This paper is organized as follows. Section 2 describes the data and gives an introductory statistical and institutional overview of the selection process and of the outcomes at trial. Section 3 presents a theoretical model of the selection process while Section 4 presents the econometric evidence and discusses why it suggests that courts are biased by local labor market conditions. Concluding remarks follow in Section 5.

When a firing litigation is taken to court, only the characteristics of the employee's misconduct episode should be relevant for the judge's decision. This paper shows that, instead, local labor market conditions are also likely to influence the court's decision: The same misconduct episode may be considered sufficient for a legitimate firing in a tight labor market but insufficient otherwise. This conclusion has been reached on the basis of the following argument. We observe that where unemployment is higher misconduct episodes brought to trial are no less serious than elsewhere, and if anything more serious. As a consequence, if the legal standard of judges were independent of unemployment, we should also observe the same number of firms’ victories at trial in all regions. We observe, instead, that higher unemployment is associated with more workers’ victories at trial and this suggests that worse labor market conditions induce judges to be more favorable to workers. This effect is not only statistically significant, but also large in size. A 1 percentage point increase of the unemployment rate decreases the probability of a firm victory by 2.5 percentage points. This result suggests, for example, that given the 14 percentage point difference between the regional unemployment rates of Milan and Naples in 1993, the probability of a firm victory would have been approximately 35 percentage points lower in the southern of these two cities for the same misconduct episode. Interestingly, the combined effect of other regional differences, like for example the political orientation of judges, dampens the consequences of the unemployment differential that would be observed if regions had the same characteristics. This finding is important for at least two research fields. From a macroeconomic point of view, in addition to the traditional channel of causation according to which higher firing costs may increase unemployment rates if they reduce hiring more than they prevent firing, the evidence offered by this paper suggests the existence of a reversed channel of causation: higher unemployment rates increase firing costs via the effect on judges’ decisions. More generally, the real extent of firing rigidities, that has been considered so crucial in the debate on European unemployment, cannot be really assessed without taking into consideration the role of courts. The analysis of the relation between labor market conditions and the decision criteria of judges is also important for labor law scholars because following traditional Italian principles, the law should be applied in the same way for all citizens and over the entire national territory: judges should not be influenced by local labor market conditions when deciding on firing litigations. The evidence presented in this paper is mainly based on one, albeit large, Italian firm from which we were able to gather high quality information on misconduct episodes and on the selection of litigations for trial. These firm data, in spite of their high quality and nationwide extension, do not allow us to draw general conclusions. Nevertheless, they suggest that more research on the role of courts in raising labor market rigidities is useful and promising.