Every month the Federal government measures the number of -persons who are undergoing spells of unemployment. These statistics are watched closely by policymakers, economists, financial analysts and others for they reflect the health of the Nation's labor market,- and, to a large extent, the economy. The manner in which spells are measured, however, has the result that some spells are not counted. This paper presents an initial attempt to quantify these "missed" spells of unemployment.

The fact that some spells are missed is not a new finding. It was implicit in the work of Kaitz (1970), Salant (1977), Bowers (1980) and others who examined and clarified the meaning of unemployment duration statistics. These data are collected every month through the Current Population Survey (CPS), the source of the Federal government's unemployment statistics and the official unemployment rate. Kaitz and the others drew attention to the fact that what the CPS measured were "ongoing" spells of unemployment and not "completed" spells, and in the process they uncovered other important characteristics of unemployment measurement. One of these was the likelihood that short spells of unemployment occurring-between the CPS monthly surveys were apt to be missed.

Just as the distinction between ongoing and completed spells is important, we believe some attempt at quantifying the spells of unemployment that are missed is important also. Using the Survey of Income and Program Participation (SIPP), a relatively new household survey, we present preliminary estimates of missed spells. We consider them preliminary since they are derived by replicating the CPS methodology in SIPP, and important differences exist between the two surveys. Nevertheless, we find that when these missed spells are quantified in a CPS-type unemployment measure derived from SIPP, their impact is nontrivial.

Our paper begins with a brief discussion of the CPS labor force concepts and its measurement of unemployment. The next section explains the methodology used to develop a CPS-type unemployment measure in SIPP. The fourth section presents our data results. The paper concludes with a summary and a discussion of the implications of our findings for further research.