تجزیه و تحلیل اقتصادی از قانون اخراج: تعیین حق سنوات در غرب آلمان

Severance pay is a vital part of employment protection legislation (EPL). We investigate the incidence and level of severance pay for dismissed employees. Our theoretical model predicts that not only the law and its interpretation by labour courts but also the costs of a suit have an impact. Using West German panel data for 1991–2006, we find that the employees’ costs resulting from a suit and the legal determinants of such transfers affect the incidence of severance payments. In contrast, their level only varies with legal regulations. Our results imply that the strictness of EPL in Germany varies with extra-legal factors like employees’ financial constraints.

Despite recent reforms, employment protection legislation (EPL) in Germany is still regarded as comparatively strict (see OECD, 2004 and World Bank, 2006). However, employers can circumvent restrictive regulations by making a sufficiently high severance payment. Accordingly, the majority of such payments in Germany result from private agreements between firms and employees. Yet even if severance pay arises from a negotiation, EPL and its interpretation by labour courts can have a strong impact. This is because employees may initiate a court procedure to enforce the restrictions on dismissals which EPL constitutes. In this paper we investigate the impact of legal criteria as well as of the costs of enforcing EPL on severance payments (see, e.g. Bertola, Tito, & Cazes, 1999). The analysis can provide insights as to whether the legal rules – underlying, for example, the OECD's and World Bank's evaluations – are an adequate proxy for the actual extent of employment protection. In consequence, we also contribute to the debate on the distinction between the law in action and law on the books (Jolls, 2004 and Jolls, 2007). Finally, our estimates enable us to calculate a lower bound of the expected costs of a dismissal. Knowing such costs can help in evaluating the strictness of EPL in Germany. Subsequently, in Section 2 we describe the legal situation in Germany and survey the relevant empirical studies. In Section 3, we present a theoretical model of severance pay determination which allows for all major observable consequences: namely, a dismissal without a severance payment, an agreement between firm and employee including a payment, and outcomes involving a labour court. We inquire how the prevalence and magnitude of severance payments are affected by variations in parameters which, first, are defined by law as determinants of EPL and, second, affect the costs of a legal dispute. In Section 4 we describe the dataset, the German Socio-Economic Panel (SOEP), and the empirical specifications used. In Section 5 we put our theoretical hypotheses to an empirical test. We find that the probability of obtaining severance pay and its level are indeed affected by legal determinants. In addition, the probability increases with the employee's costs of losing a job and declines with those of a legal dispute. Our results indicate that the expected costs of a dismissal and, more generally, the strictness of EPL as a proxy for the law in action differ substantially from the law on the books. Therefore, indices of EPL based on the latter, such as that used by the OECD, 1999 and OECD, 2004, may misrepresent the actual extent and severity of EPL in Germany. To illustrate our results, we focus on a selection of “typical” employees. Our most common employee obtains a payment with a probability of 14% and its real expected level is about € 900. If this employee is a member of a trade union, the probability of obtaining a payment more than doubles, while the non-applicability of the central law regulating dismissals reduces it to less than 1%. In Section 6 we conclude. The Appendix contains some of the derivations and additional information on the data. Our paper is related to analyses of EPL which explicitly incorporate the legal process and allow for an interaction between firms and workers, on the one hand, and courts on the other. Ichino, Polo, and Rettore (2003), for example, present a model of the litigation process against the backdrop of the Italian legal situation. They investigate theoretically whether labour market conditions are reflected in court outcomes and find empirical evidence for such a relationship. Malo (2000) views bargaining about payments in the case of individual dismissals in the context of the Spanish legal situation as a game of incomplete information. He shows, inter alia, that the amount demanded increases with the expected award by the court and declines with the employee's costs of filing a suit. Malo and Pérez (2003) extend the model to enhance its applicability beyond the Spanish context. None of the approaches outlined above focuses on a distinctive feature of German EPL: severance pay can result from offers by firms and can also be court-induced, but there is no universal entitlement. Hence, the probability of obtaining a payment is determined endogenously and affected by employee- and match-specific features. While we investigate the impact of income taxes in a companion paper (Goerke & Pannenberg, 2009), in the present contribution we analyse an extended model which explicitly allows for court verdicts and focuses on these employee- and match-specific effects, as well as the costs of enforcing EPL. Our contribution is also associated with analyses assuming that a labour court may erroneously evaluate the cause of a dismissal (Besancenot and Vranceanu, 2009, Galdón-Sánchez and Güell, 2003, Huang et al., 2009 and Stähler, 2008), and that a court can affect the incentives to undertake match-specific, productivity-enhancing investments (Deffains, Gabuthy, & Lambert, in press). From a wider perspective, we touch upon the literature on litigation and settlement as recently surveyed, for example, by Spier (2007) and Daughety and Reinganum (in press).

There is no universal entitlement to severance pay in Germany. We have developed a model which allows severance payments to be rationalised as the outcome of a labour court procedure and the employer's desire to prevent such a conflict. Our model predicts that severance payments offered by firms in order to avoid a verdict rise with the level of expected court-awarded payments. As a consequence, the incidence and expected level of severance pay increase with the determinants of such payments (implicitly) laid down by employment protection legislation. More importantly, our model predicts that the incidence of severance pay declines with the employee's cost of a court procedure and that the average payment is also affected by such extra-legal characteristics. This suggests that the extent of employment protection – as captured by the expected amount of severance pay – varies with personal characteristics of employees, such as the ability to afford a dispute, which employment protection legislation in Germany deems irrelevant. In our empirical exercise we use data from the German Socio-Economic Panel (SOEP) for 1991–2006 and West Germany. We find descriptive evidence for the impact of rules, either explicitly mentioned in employment protection legislation or applied by labour courts, on the incidence and level of severance pay. Our regression analyses confirm these findings. For the incidence we also find substantial effects of the costs of a court procedure and of a dismissal without a severance pay offer. As a consequence, employment protection legislation does indeed affect who obtains severance pay in (West) Germany and how much a dismissed employee receives. Perhaps more importantly, our findings suggest that extra-legal factors capturing the costs of a dismissal and of fighting a dismissal in a labour court have a strong impact on the probability of obtaining severance pay and, hence, the strictness of employment protection legislation. In terms of magnitudes, we calculate that a typical employee without tenure and age restrictions receives severance payments in only 14% of all dismissals, but that this probability increases by about a factor of five for older employees with substantial tenure (cf. Table 5, column 1). The payment a typical employee receives, if he obtains a transfer, is about € 6500 (in 2000 prices) and increases sixfold for older employees with very high tenure. It is plausible to assume that observed severance payments represent a lower bound for the costs of a dismissal because, for example, firms face notice periods and have to bear administrative and further legal costs of dismissals (World Bank, 2007). Focussing on a typical employee without age and tenure limits and assuming that these additional costs amount to twice the expected severance pay indicates that dismissing a typical employee will cost a firm about 85% (3 × € 910 = € 2730) of this employee's previous monthly gross wage of about € 3200. Our calculations also show that the costs of a dismissal are substantially higher for older employees with higher tenure, union members and those particularly protected by employment protection legislation. In summary, our investigation of employment protection legislation in Germany indicates that the law on the books is indeed an important determinant of the law in action, but simultaneously reveals a substantial discrepancy between the two. As a consequence, intertemporal or international comparisons of the impact of employment protection legislation should aim to use consistent information on the law in action, whereas the use of data on the law on the books may seriously distort policy conclusions.