This paper presents a new way to look at
monthly changes in unemployment rates using CPS
data on the labor force. The percentage monthly
change in the unemployment rate is expressed as the
sum of three terms: (1) the amount due to growth in
the population, which must be offset by growth in
employment and nonparticipation if the unemployment
rate is not to increase; (2) the amount due to
changes in the size of the labor force; and (3) the
amount due to changes in employment.
Average values for these amounts are computed
for recession and non-recession months from Febru-
ary 1950 to December 1983. The results for teen-
agers are consistent with the "baby boom" hypothe-
sis about sources of trends in unemployment rates.
However, because of their highly unstable patterns
of employment and labor force participation, demo-
graphic forces on unemployment rates are weaker
for teenagers than for those 20 and older.

By combining features of rationing models
and hedonic models in a novel way, this paper
develops a structural model of categorical
labor force behavior to help explain several
puzzles in data on unemployment and discouraged
workers. It traces the links among minimum
wages or other rigidities, hiring and firing
decisions by firms, and labor force participa-
tion decisions by individuals of differing skill
levels. A key comparative static result is that
a rise in an effective minimum wage increases the
labor force participation of more skilled marginal
workers but reduces the participation of less
skilled marginal workers.

This paper includes a brief review of the economic litera-
ture on unemployment, with implications for empirical work
on youth unemployment and participation; original empirical
work using linear probability and logit models on 1980 Cen-
sus microdata; and suggestions for structural models to be
used in future work with microdata on youth labor force be-
havior.
On Census Day in 1980, 41.4% of black male teenage school-
leavers, but only 15.9% of white male teenage school—leav-
ers, were out of the labor force as well as out of school.
Among the labor force participants, the white male unemploy-
ment rate was 18.5%, while the black male rate was 30.5%.
The behavior of young women was quite similar, except for
worse labor force participation among white school leavers.
For student male teenagers, the racial differential in la-
bor force nonparticipation was only half as bad as for non-
students, but the unemployment differential was 22% more
unfavorable to blacks. The female student differential in
unemployment rates was 14% more unfavorable to blacks.
Among both students and nonstudents, and among both young
men and young women, there was no significant or large
racial differential in the ratio of unemployment to popu-
lation. The large racial differential in unemployment is
counterbalanced by a large racial differential in labor
force participation. There are two very different struc-
tural interpretations of these findings: higher black reser-
vation wages and discouraged worker effects.
As previous researchers have found using other data, sim-
ple statistical models do not explain much of the individ-
ual variation in youth labor force behavior. Structural
models of youth unemployment are proposed for estimation
with microdata. These models are designed to ameliorate
biases in the simple models from ignoring simultaneity and
ecological correlation. These models for microdata have
added worker and discouraged worker effects exactly anal-
ogous to those in macroeconomic models.

This paper presents some extremely flexible
identities useful in analyzing changes in unemploy-
ment rates from month to month, from year to year,
and over longer periods. An aggregate unemployment
rate change is expressed as a polynomial in labor
force stocks and first differences in labor force
stocks. Terms of this polynomial are interpreted
as the effects of (1) changes in the distribution
of the labor force among demographic groups; (2)
unexpected changes in labor demand within a demo-
graphic group; and (3) unexpected changes in labor
supply within a demographic group. A simple exten-
sion of the framework shows its relationship to
recent work with labor force gross flow data. The
framework is applied to the increase in black youth
unemployment between 1950 and 1970. Most of it may
be attributed to a decline in employment among male
in the South.