[PEN-L:5039] Re: Unemployment Duration

["Eric Nilsson" <enilsson@xxxxxxxxxxxxxxx>]

Rich P wrote:
> How do you correct for this (BLS unemployment duration
> data measuring _uncompleted_ spell and not for
> _completed_ spell).

Here's a longer answer than anyone likely wants.

The BLS does not publish data on completed spell of
unemployment. However, there is a literature on using
existing BLS data to generate estimated for the length
of a completed spell. A good recent example surveying
different techniques is Michael Horrigan, "Time Spent
Unemployed" in Montly Labor Review, July 1987.

An accurate estimation of completed spells of unemployment
is a major undertaking (Horrigan tells me he might
undertake this project in a year or so). However, a simple
technique developed by Kaitz gives pretty good answers with
little data. In essence, you assume a steady state (unchanged
temporal structure of unemployment) which permits one
to simply: divide the total number of unemployed by the number
unemployed for a month or less. The answer, it turns out, is
a fairly good estimate of the average expected length of a
completed spell of unemployment (measured in months).

If anyone actually wants more information or citations,
e-mail me and I'll give you more details. There
are many other things I had to do to estimate the completed
spell of unemployment for permanently laidoff/fired
workers. And, I'm sure over the next few weeks I'll
slightly modify my estimates. (But will spare those
on pen-l from the updates.)


Eric Nilsson
Department of Economics
California State University
San Bernardino, CA 92407
enilsson@xxxxxxxxxxxxxxx