Earnings Adjustment Frictions: Evidence from the Social Security Earnings Test

Recent literature has suggested that individuals face frictions in adjusting their earnings in response to policy, such as the cost of gathering information about policy or of finding a job that pays their desired earnings level. Using a panel of administrative data from the Social Security Administration on 1% of the U.S. population from 1961 to 2006, we document descriptive evidence about such barriers and use this evidence in estimating the parameters of the adjustment process. We analyze “bunching” at the convex kink in individuals’ budget sets created by the Social Security Annual Earnings Test (AET). When individuals transition out of being subject to the AET, the vast majority of earnings adjustment takes place within three years, suggesting rapid adjustment. We specify a model consistent with the descriptive patterns and estimate that the mean of the elasticity of earnings with respect to the implicit net-of-tax share is XX and the mean fixed cost of adjustment is $XX. Using data from 1986-1988 and 1992-1994, during which period the AET became less stringent for those at Normal Retirement Age and above. We estimate a smaller elasticity of only 0.XX when we constrain adjustment costs to be zero, though the difference is not statistically signifi cant.