For married men, we find the conventional view of retirement trends — that the long term trend to early retirement has been reversed — is partially contradicted by recent data. Specifically, descriptive data collected from both the Census and the Health and Retirement Study (HRS) suggest that for those in their fifties, over the periods 1992 to 1998 and 1998 to 2004, the trend to early retirement reasserted itself and labor force participation fell. In contrast, for those in their sixties, there was an increase in work. Similarly, for those 65 and over, the amount of work increased.

Simulations with a structural retirement model suggest that the recent acceleration of the trend to early retirement for those in their fifties is not the result of the change in Social Security rules. According to our model, changes in Social Security rules are expected to reduce the number of those in their early sixties who are working. This suggests that forces other than changing Social Security rules account for the observed increase in work by those in their early sixties, and that the effects of these forces are stronger than those suggested by the trends in descriptive data. Lastly, the analysis suggests that changing Social Security rules do help to explain the increase in work by those age 65 and older. The effects of these rule changes encourage workers to remain in their long term jobs for a longer time, encourage some to return from retirement to full time work, and encourage more partial retirement. Nevertheless, the changes in retirement induced by Social Security changes have been modest. Due to Social Security changes, the number of 65 year old married men at work increases by about two percentage points at ages 65 and 66, with slightly smaller changes at 67 to 69. Given the low basic labor force participation at 65 and 66, with 20 to 25 percent at full time work, and another 17 percent at part time work, the percentage increases in work due to Social Security changes are three or four times higher.