مالیات از پنجره های بازنشستگی زود هنگام و به تاخیر انداختن بازنشستگی: تجربه ای در فرانسه

This paper investigates the effect of the 2003 French pension reform on hiring, firing and employment rates among older workers. This reform increased the mandatory retirement age and simultaneously it set a tax levied on early retirement windows paid by firms to their older workers, to encourage them to leave their job early. We use a matching model with endogenous job destruction extended to account for a mandatory retirement age and we calibrate the model with data drawn from the French Labor Force Surveys for the years 2002 and 2003. We show that in the case of a high tax rate, delaying retirement raises job separation rates, which partially offsets its positive effect on job finding rates. Consequently, the combination of an increase in the retirement age and a taxation on early retirement windows may have negative effects on the employment rate among older workers.

Effective retirement ages had fallen significantly over the last 30 years in most OECD countries. This pattern may lead to fiscal sustainability problems, especially for countries with Pay-As-You Go pension schemes, in which pensions of retired individuals are financed through contributions of workers. The creation of early retirement schemes in the 1970s may be an explanation of the decline in employment among elderly people observed in Europe during the same period. Zaidmann et al. (2000) has shown that in France these schemes led to a consensus between older workers, firms and government. This phenomenon may be due to two main reasons. First, early retirement schemes were partly financed by the government so they could be treated by firms as a layoff subsidy (Hakola and Uusitalo, 2005 and Hutchens, 1999). Second, the government encouraged early retirement to make more room for young workers in the labor market in a setting of high youth unemployment (Zaidmann et al., 2000). Since the employment rate among workers aged more than 55 in France was in the early 2000s one of the lowest in the European Union (29.9% in France with respect to a European average of 37.8%1), the French government implemented important changes to constrain early retirement. These changes aimed simultaneously at restricting the access conditions to publicly subsidized early retirement schemes for workers and firms and at increasing the share of early retirement expenses charged to employers. However, in spite of the increase of early retirement costs for firms, employers continue to encourage their older workers to leave early their job, offering them generous financial incentives called “early retirement windows”. We can discuss the motivations that lead employers to offer early retirement windows to their older workers rather than firing them. One potential explanation suggested by Amauger-Lattes and Desbarats (2006) is that the legislation regarding separations among older workers implies a cumbersome and often costly procedure for employers. Studying more than 300 court rulings of the Court of Cassation during the period 1994–2004 regarding job separations for older workers, they highlight that in most cases, employers strike a mutual agreement with their older workers, offering them generous early retirement windows and reporting a “dismissal for serious misconduct”, although it is not the case. In the face of this widespread phenomenon, especially in the case of big firms, the French government set in 2003 a tax levied on the amount of early retirement windows paid by firms. The tax rate amounted to 23.85% in August 2003 and in August 2007 it raised to 50%. In addition, to deter firms from pushing their older workers into retirement too early, the 2003 reform also led to an increase in the mandatory retirement age. Initially, this age was 60, which means that when a worker reached 60 and if her insurance period was sufficient to allow her to draw a full pension, an employer could push her into retirement paying her a low retirement allowance. In 2003, this age has been increased to 65 and since 2008 it is 70. The goal of this paper is to investigate the effect of the combination of these two reforms on the hiring rate, the job destruction rate and the employment rate among older workers. Let us first define what job destruction means in this paper. As we allow employers to offer early retirement windows to their older workers to encourage them to leave their job and to avoid a layoff, the definition of a separation (or job destruction) differs across the age group of workers. On the one hand, for middle-age workers, a separation results from a layoff and the employer has to pay to the worker a severance pay, as defined in the French Employment Protection Legislation. On the other hand, for older workers close to retirement, a separation has not the same nature. Indeed, for this age group of workers, job destruction means that the employer encourages a worker to leave her job, before she reaches the mandatory retirement age, offering her early retirement windows. In this setting, the tax set in 2003 and levied on the amount of early retirement windows paid by firms is a sort of age-dependent employment protection, given that it concerns separations for one specific age group of workers. In this paper, we put forward the idea that in the case of a high tax rate on early retirement windows, delaying retirement may raise the separation rate among the group of workers concerned by the tax. We refer to this effect as the “impatience effect”. The idea is the following: as shown in the previous literature on age-dependent employment protection, especially on the Contribution Delalande in the French case2 (Behaghel, 2007, Behaghel et al., 2008, Chéron et al., 2007 and Chéron et al., 2008), a high tax rate deters employers from laying their older workers off after a shock, even though their filled jobs imply negative profits for firms. Indeed, as long as their firing cost exceeds the expected losses, employers prefer waiting until their older workers reach the mandatory retirement age. This is the well-known labor-hoarding effect of the tax. In this paper, we point out that in this setting, delaying retirement leads to an increase in expected losses incurred by firms, making them more impatient to get rid of their older workers. Consequently, an increase in the retirement age may encourage firms to offer generous early retirement windows to their older workers to force them to leave, rather than waiting until they reach the new mandatory retirement age. Highlighting this so-called impatience effect, our paper gains new insights into the effect of an increase in the retirement age on job creation, job destruction and employment. The effect of an increase in the legal retirement age on employment among older workers has been already considered in the literature. Previous studies pointed out that delaying retirement may have a positive “horizon effect” on job creation, when labor is treated as a quasi-fixed factor that implies fixed costs (Oi, 1962). These costs result from either a bilateral monopoly problem (Hutchens, 1986) or from an accumulation of specific human capital through training (cf Hashimoto, 1981) and imply that firms are more reluctant to hire a worker close to the retirement age. Furthermore focussing on job destruction, Aubert et al. (2006) showed that there is an age-bias technological change, so employers are less likely to retain an older worker in the case of a shock on her job, if her employment duration is too short. Extending the model of Mortensen and Pissarides, (1994) in order to account for the life cycle of the worker with a bounded retirement age, Chéron et al. (2007) have drawn similar conclusions and argue that an increase in the retirement age may have a positive effect on hiring rates and may reduce firing rates by lengthening the employment duration of older workers. In this respect, the impatience effect that we highlight in this paper may offset the horizon effect in the case of high age-dependent employment protection, leading therefore to a rise in separation rates among the age group concerned by the tax. While previous studies on age-dependent employment protection (Behaghel, 2007, Behaghel et al., 2008, Chéron et al., 2007 and Chéron et al., 2008) showed theoretically and empirically that an age-specific firing tax may have a strong negative effect on job creation among the protected age group and also that it may raise separation rates among the previous cohort of workers not concerned by the tax, these works did not pay any attention to the effect of delaying retirement on job creation, job destruction and employment in the case of a high age-specific separation cost. In this paper, we determine a critical value of the rate of the tax on early retirement windows, above which the impatience effect is higher than the horizon effect. If the tax rate is higher than this critical value, an increase in the mandatory retirement age raises job destruction among older workers. In that case, the positive horizon effect of delaying retirement on job finding rates among the age group of workers concerned by the tax is attenuated. In addition, as an increase in the retirement age reduces transitions from unemployment to retirement, it exerts a negative effect on employment among older workers, if the horizon effect is not sufficiently strong. Consequently, the higher the tax rate on early retirement windows, the lower the horizon effect with respect to the impatience effect and the stronger the negative effect of postponing retirement on employment among the protected age group. Calibrating our model using data drawn from the French Labor Force Survey for the years 2002–2003, we provide a numerical illustration of these findings. We show that the change in the job separation rate for older workers after an increase in the retirement age strongly depends on the tax rate. Consequently, the effect of delaying retirement on employment among the protected age group is sensitive to the level of taxation of early retirement windows. The remainder of the paper is structured as follows. In the next section, using the data drawn from the French Labor Force Survey for the period 2001–2009, we provide a brief presentation of the patterns in job destruction rates among older workers between 2001 and 2009, investigating to what extent separation rates have been affected by the 2003 and 2008 reforms. Then, using data from the Survey on Health Ageing and Retirement in Europe (SHARE), we show the incidence of early retirement windows in the job destruction among older workers in France. In Section 3, we describe in detail the theoretical model, following the specification of Behaghel, (2007). In Section 4 we present our main theoretical findings on the effect of an increase in the mandatory retirement age on hiring rates and firing rates of middle-age and older workers in a setting of partial employment protection. In Section 5, we describe our quantitative analysis based on the French Labor Force Surveys for the years 2002 and 2003 and we present our results. Section 6 concludes.

In this paper, we studied the effects of postponing retirement in a setting of an age-specific employment protection on the hiring and separation rates of older workers and also on employment of the elderly. Reproducing the 2003 French pension reform, we set a tax levied on early retirement windows paid by firms to their older workers to dismiss them. We provided some theoretical findings considering a matching model with endogenous destruction extended to account for a mandatory retirement age and in which we introduce age-dependent separation costs. We highlighted that in the case of a high tax rate, delaying retirement may raise separations among the targeted age group of workers through an impatience effect. Indeed, a high tax rate discourages firms from dismissing older workers paying them financial incentives, so employers prefer waiting until their workers reach the mandatory retirement age. In this setting, delaying retirement forces employers to retain their workers for a longer time, and they could have interest to dismiss them before they reach the retirement age in spite of the cost induced by the tax. We pointed out that there exists a critical value of the tax rate above which the impatience effect offsets the labor-hoarding effect of postponing retirement, leading to a rise of the separation rate of older workers after an increase in the mandatory retirement age. Calibrating our data using the French Labor Force Surveys for the years 2002 and 2003, we showed that this critical value is negatively correlated to the amount of early retirement windows but it is positively correlated to the Poisson arrival rate of idiosyncratic shocks. Nevertheless, for reasonable value of parameters, we have shown that the impatience effect may have a strong incidence on the predicted effects of delaying retirement on the separation rate among older workers. In addition, the impatience effect that we highlighted in this paper may partially offset the positive impact of an increase in the retirement age on the hiring rate of these workers. Theoretically, we determined a second critical value of the tax rate above which delaying retirement reduces the hiring rate among the group of workers targeted by the tax. However, we remarked in our numerical illustration that the extent to which the tax rate influences the effect of postponing retirement on the job finding rate is negligible for both age groups of workers. However, the relative variation in the employment rate of the elderly after an increase in the mandatory retirement age strongly depends on the taxation level. Indeed, in the case of a high tax rate, delaying retirement may lead to more separations, which exerts a negative effect on the employment rate. Similarly, regarding the previous cohort of workers, we have shown theoretically that the impatience effect may affect the impact of delaying retirement on the firing and the hiring rates but we found in our numerical illustration that the impatience effect has a lower incidence on the predicted effects of an increase in the retirement age on either job creation or job destruction. Consequently, the relative variation in the employment rate among this cohort of workers after an increase in the mandatory retirement age is not altered in a significant way by a taxation of early retirement windows. However, these results have to be considered with caution. Indeed, in our model we consider an exogenous and constant amount of early retirement windows, while we can expect that the amount offered by employers to their workers may depend on several factors as the level of the tax, implying therefore a substitution between early retirement windows and the tax, or the characteristics of this worker. Too few information are still available on early retirement windows in France, however since the 2008 reform, employers who offer such financial incentives to their older workers have to indicate their names, their ages and the amount that they received prior to exit. Using this data, we could carry out an empirical study aiming at better understanding the factors that lead employers to offer such financial incentives and the determinants of the amount offered. It could also allow us to better grasp the way separation costs are negotiated between employers and older employees, through a mutual agreement. We leave this issue for further investigation.