Working Papers
Peer Effects on Public School Teachers' Retirement Decisions
(Job Market Paper)
(Job Market Paper)
This paper analyzes peer effects on public school teachers' retirement timing decisions. I use an administrative dataset of the full population of Missouri PSRS (Public School Retirement System) public school teachers with ages 50-75 during academic years 1994-2007 to construct a model incorporating both financial and non-financial incentives when teachers make retirement decisions. This study mainly differs from previous studies in the use of data and controls for financial incentives. I use a much richer dataset that is an administrative dataset of the full population of Missouri PSRS public school teachers, and the schools are more diverse in poverty and urbanicity. For controls of financial incentives, I propose the simulated ``financial benefit of postponing retirement" based on the Stock-Wise (SW) option value model as a new financial incentive proxy. Empirically, when the retirement rate of peers in previous year increases 1%, the probability of retirement increases approximately 0.128 percentage points. In addition, I find evidence of heterogeneous peer effects. For example, teachers with different education levels respond to peers’ retirement differently.
Effects of Pension Rule Changes on Timing of Retirement
This paper uses an economic structural model to fit Missouri 1994-2008 public school teachers' data and study the effects of changing pension rules on the timing of retirement. This study fixes three limitations on applications of the Stock-Wise (SW) option value model. First, the SW option value model can be applied only to retirement data when the pension rules are fixed. Second, there is a selection bias in the sample of senior teachers: among the retirement eligible teachers, only the "stayers" are in the sample, while the "early leavers" are absent. This bias has not been adjusted in the previous studies. Third, the SW model was estimated based on the likelihood of panel data of individuals (teachers or salesmen). The cost of computing the likelihood may be prohibitively high for a larger state, such as Texas, and/or a longer sample period. We propose models of policy expectation and adjust for the sample selection bias in estimation of the option value model. In addition, I group individual teachers into a fixed number of (age, experience) cells to reduce the time of evaluating the likelihood. I estimate the model under different expectation assumptions and compare their performance by simulation. The estimated SW model exhibits good in-sample and out-of-sample fits. Finally, I conduct counterfactual studies to measure the effects of pension rule changes, which suggests that the current pension enhancement cannot retain experienced teachers.
Working in Progress
A Sequential Probit Model for Employees’ Retirement Decisions
with Shawn Ni
This paper focuses on refining and developing econometric tools for empirical studies of job market exiting. Retirement decisions are commonly modeled in static reduced-form settings or by structural models. Both types of models have shortcomings. The static reduced form models do not make use of all sample information (by ignoring the fact that many observations are repeated on the same person), while the structural models tend to be restrictive and costly to estimate. We proposes to use a battery of sequential probit models that include both financial and non-financial factors in a dynamic setting. The financial factor is captured by the Stock-Wise option value model. Hence the latter is a special case of the sequential probit model. The sequential probit model generalizes the Stock-Wise model with appropriate construction of covariates; it is relatively flexible and reasonably simple to estimate; it can use all sample information and allows for serial correlation in preference errors and individual fixed effects. With individual fixed effects, the number of parameters are too large to estimate using MLE. In this study, we use a Bayesian approach.
Statistical Discrimination and Employers’ Learning for Productivity of Young Economics Faculty in Public Universities
with David Kaplan
with David Kaplan
In this study, we collect the data for young economic faculty, including personal, educational, and demographic characteristics and research productivity. Using this dataset, we can examine the statistical discrimination and employer learning for publication. In particular, the fact that it will include detailed measures of research productivity over multiple years allows insights that cannot be obtained from the datasets most commonly used in this field.