The Effect of Unemployment Insurance on Temporary Layoff Unemployment

Abstract

Economists are now beginning to recognize that an understanding of layoffs is crucial for a proper analysis of unemployment. In manufacturing, about 75 percent of those who are laid off return to their original More generally, among all persons classified as losers, layoffs account for about 50 percent of all unemployment spells. Temporary layoffs are an even larger fraction of cyclical changes in the number of losers. While this group includes some seasonally unemployed, most layoffs are induced by short random or cyclical fluctuations in demand. The conventional model of search unemployment is inappropriate for those and the modern theory of the Phillips curve requires substantial modification because of the size and cyclical variation of unemployment. I In a previous paper (1976), I showed analytically that our current system of unemployment insurance (UI) provides a substantial incentive for increased unemployment.2 The present paper provides micro-economic evidence that UI actually such a powerful effect. The estimates imply that the incentive provided by the current average level of UI benefits is responsible for approximately one-half of unemployment. It is important to note that the current study shows that UI increases the amount of unemployment, but does not deal with the mean per spell. This distinction deserves emphasis because nearly all previous empirical work focused the potential effect of UI duration. This focus is both unfortunate and surprising since UI can actually increase total unemployment while decreasing the mean per spell. While UI increases the of any given spell of unemployment, it may also induce more very short spells of unemployment. This possibility of reduced mean is clear in my 1976 theoretical analysis. An additional practical *Professor of economics, Harvard University. I am grateful to the National Science Foundation for support of this research, to David Ellwood and Joseph Kahan for assistance with the statistical calculations, and to Richard Freeman, Zvi Griliches, Daniel Hamermesh, James Medoff, Melvin Reder, and Jeffrey Sachs for discussions and comments. Earlier versions of this paper were presented at seminars at Chicago, Harvard, and Yale universities. IIn my 1975 paper, pp. 737-42, I discuss the implications of layoffs for the theory of search unemployment, the Phillips curve, and wage inflexibility. Although the standard criterion of unemployment is active seeking within the past four weeks, individuals are officially classified as unemployed without any inquiry about recent job-seeking activity if they state that they are on awaiting recall by their employers. Some of those look for jobs or alternative permanent employment, but the vast majority do return to their original Readers should not be confused by the two quite separate meanings of the term layoff in the Department of Labor's lexicon. In manufacturing establishment data, a is a separation initiated by the employer (not a quit) and may be permanent or In the Current Population Survey (CPS), an individual is if he is not working but has a job to which he is expecting to be recalled by his employer. To emphasize that I am dealing with those layoffs expected to terminate in recall, I use the adjective temporary. Unfortunately, the CPS uses the word in a different and quite confusing way: persons are divided into an indefinite duration group (in which the individual does not have an expected date of recall within thirty days) and a temporary group (when such a date is known). When it is useful to distinguish these groups, I use the terms indefinite duration and fixed duration; in my usage, the term includes both groups. 2My 1976 paper is really an explicit proof of arguments made more informally in my earlier study for the Joint Economic Committee (1973). For a similar development, see Martin Baily.

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 — 248 citations exist upstream.

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