by Arthur-Damon Jones, University of Chicago
A large body of literature has examined the effects of employer-provided benefits on employee mobility, and, in particular, the effect of retirement plans. This literature has generally documented a negative correlation between the provision of employer-provided retirement plans and employee mobility. Possible channels for this mechanism include the incentives of the retirement plans, which may include backloaded benefits and vesting requirements. Interestingly, a comparable effect has been found among both defined benefit (DB) plans and defined contribution (DC) plans. In the latter case, the relative portability of DC plans appears at odds with the notion that the incentives of a retirement plan affect mobility. An alternative hypothesis is that workers are sorting across firms due to their preferences for deferred compensation, and thus, the observed relationship is confounded by unobserved factors. The literature to date has not conclusively disentangled the respective hypothesis, in part due to a lack of exogenous variation in employer-provided benefits and also due to coarse information regarding retirement plan features. In this study, we aim to fill in these gaps in the literature by using natural and field experimental variation in retirement plan type and participation. Using administrative data from two firms that enacted changes in their retirement plan offerings, we can estimate the effect of employer-provided retirement benefits on employee turnover. The quasi-random and random nature of plan variation allows us to purge our estimates of employee sorting effects. Furthermore, the detailed administrative data affords us precise information regarding the features of each retirement. As a result, we can use our results to distinguish between the competing hypotheses regarding employee mobility and employer-provided retirement plans.