Association for Public Policy Analysis & Management

Public old-age pensions are an important component of the social safety net in many countries. The objective of this paper is to estimate the causal effect of pension receipt on employment and health by exploiting exogenous changes in the full retirement age under South Korea's public pension system. South Korea is experiencing one of the most rapid and severe aging society problems, which makes it a good example for studying this issue. When I account for the endogeneity of employment and health with respect to pension receipt, I find that receiving a pension decreases employment and improves self-reported health.

Disentangling the effects of pension receipt on employment and health is not straightforward, as employment status may directly affect the timing of pension claiming, and unobserved factors such as health might be correlated with both employment and pension claiming. Considering these possible issues, I estimate the causal effect of pension receipt on both employment and health by using exogenous variation in the full retirement age as an instrument for pension receipt. I use an instrumental variable strategy to address the endogeneity of employment and health with respect to pension receipt.

The first-stage estimates of the baseline specification with additional controls and individual and year fixed effects suggest that approximately 22 percent of people in the sample are compliers (people who begin claiming pension when they reach full retirement age). In the second-stage estimation, I find that pension receipt reduces the employment rate by 5 percentage points. According to two different employment indicators, the aggregate employment rate was approximately 60 percent for two different employment indicators in the period of study. Although the statistical significance was not particularly high, the size of the effect was 8.3 percent of the aggregate level ($5/60=0.083$). I also find that pension receipt improves health. Both of the health indicators of interest improved 8 to 9 percentage points, which indicated at least a 20 percent increase relative to the average values.