Earnings Adjustment Frictions: Evidence from the Social Security Earnings Test

Alexander M. Gelber, Damon Jones, Daniel W. Sacks

NBER Retirement Research Center Paper No. NB 13-07
Issued in September 2013

Using a panel of Social Security Administration microdata on 1 percent of the U.S. population from 1961 to 2006, we study frictions in adjusting earnings to changes in the Social Security Annual Earnings Test (AET). Individuals continue to "bunch" at the convex kink the AET creates even when they are no longer subject to the AET, consistent with the existence of earnings adjustment frictions in the U.S.. We specify a model consistent with the evidence that allows us to estimate the elasticity and adjustment cost. Intuitively, inertia due to an adjustment cost leads to residual bunching after a kink in the budget set becomes less pronounced (or disappears altogether). Our primary estimation method uses the degree of such inertia (in combination with the initial amount of bunching

at the kink) in estimating the size of the adjustment cost (and elasticity). We estimate in a baseline case that the elasticity of earnings with respect to the implicit net-of-tax share is 0.23, and the fixed cost of adjustment is $152.08. Our results imply that eliminating the AET would cause a substantial rise in the earnings of many elderly and near-elderly individuals.

PDF (4742 K)Executive Summary (61 K)