The Effects of Social Security Wealth on the Resolution of Unemployment Spells of Older Men.

This study examines the effect of non-labor income on the resolution of unemployment spells of older men. A theoretical model of job search behavior is empirically tested using a sample of men ages 56-71 who had experienced unemployment spells that were resolved either by retirement or by re-employment. The model indicates that unemployed individuals with low potential wages and /or access to income from assets, social security benefits, or pensions are likely to view the expected returns from job search as being unattractive relative to retirement. The Retirement History Study, a ten year longitudinal survey of 11,000 older men and women, is the source of the data used to test the hypotheses generated by the job search model. A key feature of this data set is the matching of respondents with their lifetime summary earnings records in Social Security Administration files. This permits the determination of the social security primary insurance amounts that respondents qualified for when they lost their jobs even if they were not then receiving social security benefits. A recursive stochastic model consisting of a potential wage equation and a non-linear probability of labor force participation equation is estimated following a procedure analogous to two stage least squares. Unbiased estimates of the potential wage equation are obtained using a maximum likelihood technique. Sample self-selection is a possible source of bias in this equation because the potential wages of recently unemployed individuals are observable only for those who obtain new jobs. The probability of labor force participation equation is estimated by probit analysis, using imputed potential wages. Estimates of the probability of labor force participation equation support the predictions of the job search model. The potential wage has a strong positive effect upon the probability of labor force participation. Social security wealth, assets, and pension coverage are found to be significantly negatively associated with the probability that a recently unemployed older man will participate in the labor force. No significant difference is observed in the effects of social security wealth on the labor force responses of men ages 56-61 and men ages 62 and older to unemployment. This, along with an insignificant difference in the intercept coefficients for men ages 56-61 and men ages 62-64, implies that it is the actuarial present value of expected social security benefits rather than the current eligibility for them that affects the labor force responses of older men to unemployment. Therefore, actuarially reduced social security benefits could be made available to unemployed men in their late fifties and early sixties without increasing their likelihood of leaving the labor force. Such provisions currently exist in the social security laws of several European countries.Ph.D.EconomicsUniversity of Michiganhttp://deepblue.lib.umich.edu/bitstream/2027.42/158250/1/8116234.pd