ارتباط میان حقوق بازنشستگی و زمان بازنشستگی: درسهایی از آموزگاران کالیفرنیا

I exploit a major, unanticipated reform of the California teachers' pension to provide quasi-experimental evidence on the link between pension features and retirement timing. Using two large administrative data sets, I conduct a reduced-form analysis that leverages the nonlinearities in the return to work generated by the pension features and the reform-induced shifts of these nonlinearities for identification. The implied estimates of the elasticity of lifetime labor supply with respect to the return to work are centered around 0.04 in the medium-run and are less than 0.1 in the long-run

With the baby boomers reaching retirement age, public officials and private pension managers are scrambling to design policy that will reduce the burden of pension obligations on younger workers and shareholders, while still fulfilling the promises made to those nearing retirement. The proposed reforms will inevitably alter key pension financial incentives faced by members, such as the financial gain for an additional year of work, making the degree to which these incentives affect retirement timing central to the policy debate. Although there is an extensive literature that addresses the relationship between pensions and retirement, there is no firm consensus on the magnitude of the behavioral response to pension incentives. Recent work has emphasized the importance of forward-looking pension financial incentives to individual retirement decisions and has utilized both structural and reduced-form approaches to estimate the behavioral response to these incentives.1 These estimation techniques, which assume that retirees facing diverse pension incentives are otherwise identical after controlling for other observable characteristics, prove unsatisfactory as the potential for endogenous sorting makes it difficult to infer the true causal effects of the pension features. In this paper, I address this concern by using a quasi-experimental approach to estimate the price elasticity of lifetime labor supply, a key parameter for predicting the response of individuals to pension reforms and for measuring the deadweight loss associated with retirement programs. I use two unique administrative data sets to exploit the exogenous variation in the return to work that is generated by the nonlinear features of the California teachers' pension benefits and by the reforms of these pension benefits. The distribution of retirements about the budget constraint nonlinearities reveals how much labor supply responds to changes in the return to work and is the basis for the estimates of the elasticity of lifetime labor supply. The results imply that California teachers' lifetime labor supply is relatively insensitive to the financial return to work. This paper builds on both a growing literature that uses budget constraint nonlinearities to identify the causal effect of price changes on individual choice and a small literature that uses policy-reform based variation in pension financial incentives to address potential omitted variable bias. Saez (2010), which demonstrates that the compensated price elasticity is proportional to bunching at price schedule kink points in the context of income taxation, is a foundational work of this first literature.2 More closely related to this paper, Manoli and Weber (2011) find that the extensive margin labor supply is inelastic by exploiting discontinuities in the level of benefits in the Austrian pension system. I contribute to this literature by using discontinuities in the growth rate of pension benefits to estimate the lifetime labor supply elasticity. I also extend the estimation method used in this literature by using policy reforms that shift budget constraint nonlinearities to provide information about the counterfactual labor supply on a linear budget constraint. The policy-reform retirement literature generally finds a smaller role for financial incentives than non-reform studies. For example, Burtless (1986) and Krueger and Pischke (1992) examine individuals' responses to changes in the level of Social Security benefits and find a small role for Social Security in the trend toward later retirement. More recently, Mastrobuoni (2009) examined the effect of the expected rise in the Social Security normal retirement age on retirement behavior and found somewhat larger effects.3 This paper adds to this literature by focusing on distortions to the return to work rather than changes in the level of benefits. The California public school system is an advantageous setting in which to estimate the impact of pension price incentives on retirement timing. California teachers are required to participate in a state pension system with a simple benefit formula, do not participate in Social Security, have tenure, and face a rigid collectively-bargained wage schedule, so there is little uncertainty in the financial return to work and it is both salient to the teachers and easily calculated with administrative data. Importantly, in contrast to the Social Security reforms addressed in the literature that primarily changed benefit levels and the focal retirement age, the California pension reform explicitly altered the financial return to an additional year of work. A further advantage of this study is that a large portion of the sample is women, a group which has arguably been understudied. Given the minimal employment-related uncertainty faced by the teachers, I use a nonstochastic lifetime budget constraint framework to model their retirement decisions. One salient theoretical prediction of this model is that a bunching of retirements will be observed at budget constraint kinks and discontinuities. In the California teachers' case these nonlinearities are a product of the pension program. I first examine the response of individuals to their pension features and to the pension reform in a flexible way. I construct the prereform and postreform distributions of retirees over age and show that there is a spike in the distribution at the universal prereform budget constraint kink and that this spike shifts to the new kink following the reform. The distributions over service are also consistent with the discontinuity in the level of benefits that is introduced by the pension reform. The reform provides evidence that the distinct retirement pattern is shaped by pension financial incentives rather than other coincident factors. Next, I incorporate the pension reform into the estimation method introduced by Saez (2010) to quantify the excess retirements at the budget constraint kinks and to estimate the elasticity of lifetime labor supply. I determine the excess retirements as the difference between the pre- and post- reform retirement distributions at points where the kinks are removed or introduced. The estimates of the labor supply elasticity with respect the financial return to work are relatively small with the preferred estimates centered at 0.04. The results imply that teachers are willing to adjust their retirement dates by less than two months in response to a 10% increase in compensation. I investigate the impact of potential extensive-margin frictions, specific to this setting, that may cause the elasticity estimates to be downward biased. These include a high implicit cost to retiring during the school year, a cost to adjusting retirement plans in response to the pension reform, and the cost of health insurance coverage. I find that these factors have little effect on the overall results. Finally, I use an instrumental variable strategy to estimate an alternative measure of labor supply — the effect of the financial return to working on the probability of working an additional year. This alternative measure of labor supply allows me to compare the behavior of California teachers with findings in the literature. I find that California teachers behave similarly to the Social Security population in the U.S. and the estimated elasticity is similar to the findings of Manoli and Weber (2011) for the Austrian population. The remainder of this paper is organized as follows. In Section 2, I provide an overview of the California teachers' defined benefit program, the reforms of the program, and the data used in this study. Section 3 introduces the empirical strategy which is based on a simple lifetime budget constraint model that captures the teacher retirement decision. Section 4 presents the main labor supply elasticity estimates based on retirement behavior at budget constraint nonlinearities. Section 5 includes robustness checks for the main results and Section 6 presents alternative labor supply estimates. Section 7 concludes.

I use the nonlinearities of the California teachers' pension coupled with a pension reform to estimate the elasticity of lifetime labor supply with respect to the financial return to work. The results suggest that retirement-eligible individuals will work less than an additional month in the short-run and less than an additional half year in the long-run in response to a 10% increase in the financial return to work. Due to the large size of the budget constraint kink for California teachers it is possible that the lifetime labor supply elasticity estimate includes both income and substitution effects. However focusing on behavior at the kink minimizes the contamination from income effects, while the discontinuity estimate further supports a small substitution elasticity. The small elasticity estimates imply that defined benefit retirement programs do not greatly distort retirement timing and that the deadweight burden of the programs is smaller than suspected. This paper also highlights that identification and estimation strategies based on policy reforms and reduced-form nonlinear budget constraint methods are complementary. Attention to changes in the nonlinearities of a pricing schedule provides more precise predictions for individual behavior following a reform, while reforms provide more information regarding individual behavior in the absence of price nonlinearities. In the California teacher context, the precise predictions for changes in the retirement distribution also provided an opportunity to verify that the pension financial incentives were salient and well-understood. This is beneficial for addressing concerns about financial literacy, tax salience, or norms, that may confound results in such contexts. The findings of this research have timely and direct policy implications. Many teachers' retirement programs are facing large expected shortfalls. The management of these deficits will not only affect teachers' retirement security but also education security. Understanding how teachers will respond to the inevitable reforms of their retirement programs is essential to safeguard the financial and human resources needed to produce a high quality education environment. Further, although California teachers are a select group, the features of their pension are similar to those of Social Security, a program that covers a broader population. Social Security serves as the only defined benefit plan for a growing number of workers in the United States and it is both important and salient to this population as coverage is retained when individuals change jobs. Also like the California teachers' pension, Social Security has key program ages at which covered workers exhibit a propensity to retire. Given these strong parallels, the finding that California teachers' retirement behavior was little affected by a large pension reform raises concerns about how much of an impact the recent increase in the Social Security “full retirement age” will have on the labor supply of older workers.