The Social Security Earnings Test and Labor Supply of Older Men

Abstract

The social security earnings test reduces a 65-69-year-old's benefits at a 33-percent rate and a 62-64 year-old's benefits at a 50-percent rate once earnings pass a threshold amount--among the highest marginal tax rates in the economy. Previous research dismissed the importance of the earnings test but failed to take advantage of three more recent changes in the earnings-test rules, each applying to some age groups and not others, in order to identify its impact. Data on earnings distributions before and after these rules changes demonstrate that a significant number of older workers are clustered with earnings just at or below the earnings-exempt amount--so the earnings test leads some beneficiaries to hold down their labor supply. Furthermore, the bunching moves when the exempt amount moves and disappears when the earnings test is eliminated. An econometric model of labor supply is also formulated to incorporate the entire range of beneficiaries' responses to the earnings test. The resulting estimates imply substantial deadweight loss from older workers changing their labor supply to avoid taxation. Simulations predict a 5.3-percent boost to aggregate labor supply from eliminating the earnings test, and at a minimal fiscal cost. In contrast, a slight decrease in labor supply is predicted from the recently legislated increase in the exempt amount. It will be important to keep in mind this apparent sensitivity of older workers to tax and transfer rules conditional on working, which will affect the outcome of policies that attempt to induce people to work longer. Another important consideration involves 62-64-year-olds, who face an earnings test as restrictive as it was in the mid-1970s. The tighter earnings test rules will be extended to 65- and 66-year-olds as the normal retirement age is raised.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 57 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.