This paper considers estimation and testing of vector autoregression coefficients in panel data, and uses the techniques to analyze the dynamic
properties of wages and hours among American males. The model allows for non-
stationary individual effects, and is estimated by applying instrumental
variables to the quasi—differenced autoregressive equations. Particular
attention is paid to specifying lag lengths and forming convenient test
statistics. The empirical results suggest that the wage equation contains at
most a single lag of hours and wages, and that one cannot reject the hypothesis that lagged hours may be excluded from the wage equation. Our results
also show that lagged hours is important in the hours equation, which is
consistent with alternatives to the simple labor supply model that allow for
costly hours adjustment or preferences that are not time separable.
labor supply
Tightly knit extended families, in which people often give money to and get money from
relatives, characterize many developing countries. These intra-family flows may mean that
public policies may affect a very different group of people than the one they targeted. To assess
the empirical importance of these effects, we study a cash pension program in South Africa that
targeted the elderly. We use the variation in pension receipt in three-generation households that
comes from differences in the age of the elder(s) in the households. We find sharp drops in the
labor force participation of prime-age men in these households when the elder women reach 60
or elder men reach 65, the respective ages for pension eligibility. We also find that the drop in
labor supply diminishes with family size, as the pension money is split over more people, and
with educational attainment, as the pension money becomes less significant relative to outside
earnings. Other findings suggest that power within the family might play an important role:
(1) labor supply drops less when the pension is received by a man rather than by a woman; (2)
middle aged men (those more likely to have control in the family) reduce labor supply more than
younger men in the family; and (3) female labor supply is unaffected. These last two findings
also respectively suggest that the results are unlikely to be driven by increased human capital
investment or by a need to stay home to care for the elderly. As a whole, the program seems to
have had large effects on a group—prime age men living with the old—quite different from the
ones it targeted—elderly men and women.
The purpose of this paper is to summarize a certain line of work on
the interpretation of unemployment in the analysis of male labour supply
behavior. Specifically, this work investigates whether the data support
the null hypothesis that individuals experiencing unemployment are on a
labour supply function, and if the data do not support this hypothesis,
how might a researcher proceed in empirical work. The motivation for
doing this is two fold. First, what unemployment represents is an
intrinsically interesting question, and may have implications beyond
labour supply analysis in terms of macroeconomic theory. Second, if
unemployed workers are constrained in the sense that they are off their
individual labour supply functions, standard labour supply estimation
may involve a fundamental misspecification of the equation. However, it
should be emphasized that the purpose of this paper is to survey one
possible approach to this problem; the paper does not attempt to provide
a general survey on labour supply estimation or on constraints in the
labour market.
This paper uses longitudinal survey data to analyse the relationship between recent increases in individual
wage inequality, and individual and family earnings inequality. The analysis compares the implications
of a model of intertemporal family labor supply, to those of a model with no behavioral content. The
results imply that the fraction of family earnings variance attributable to permanent wage differences is
about 70 percent, and remained constant as the total variance of earnings increased during the 1980's.
The parameter estimates imply the intertemporal labor supply for males is zero, and for females is about
0.4. The net contribution of behavioral responses to increasing wage inequality reduced the level of
family earnings inequality by 14 percent.
Two approaches to estimation and testing of fixed effects models are
commonly found in the econometrics literature. The first involves variations on
instrumental variables. The second, a Minimum Chi-Square (MCS) procedure
introduced by Chamberlain, minimizes a quadratic form in the difference between
unrestricted regression coefficients and the restrictions implied by the fixed
effects model. This paper is concerned with the relationship between Three-Stage
Least Squares (3SLS) and MCS. A 3SLS equivalent of the MCS estimator is
presented and, in the usual case wherein the time varying error component has a
scalar covariance matrix, 3SLS is shown to simplify to the conventional
deviations from means estimator. Furthermore, the corresponding over-
identification test statistic is the degrees of freedom times the R2 from a
regression of residuals on all leads and lags of right hand side variables. The
relationship between MCS and some recently introduced efficient instrumental
variables procedures is also considered.
An empirical example from the literature on life-cycle labor supply is used
to illustrate properties of 3SLS procedures for panel data under alternative
assumptions regarding residual covariance. Estimated labor supply elasticities
and standard errors appear to be insensitive to these assumptions. In contrast,
the over-identification test statistics are found to be substantially smaller
when residuals are allowed to be intertemporally correlated and heteroscedastic.
At conventional levels of significance, however, even the smallest of the test
statistics leads to rejection of the over-identifying restrictions implicit in
the labor supply models.
Labor supply research has not yet produced a clear statement of the size of
the labor supply elasticity nor how it should be measured. Measurement error in
hourly wage data and the use of inappropriate identifying assumptions can account
for the poor performance of some empirical labor supply models. I propose here a
generalization of Wald's method of fitting straight lines that is robust to
measurement error, imposes mild testable identifying assumptions, and is useful
for the estimation of life-cycle labor supply models with panel data. A
convenient Two-Stage Least Squares (TSLS) equivalent of the generalized Wald
estimator is presented and a TSLS over-identification test statistic is shown to
be the test statistic for equality of alternative Wald estimates of the same
parameter. These results are applied to labor supply models using a sample of
continuously employed prime-age males. Labor supply elasticities from the two
best-fitting models that pass tests of over-identifying restrictions range from
0.6 to 0.8 . A test for measurement error based on the difference between
generalized Wald and Analysis of Covariance estimators is also proposed.
Application of the test indicates that measurement error can account for low or
negative Analysis of Covariance estimates of labor supply elasticities.
This paper investigates women's and men's labor supply to the firm
within a structural approach based on a dynamic model of new
monopsony. Using methods of survival analysis and a large linked
employer-employee dataset for Germany, we find that labor supply
elasticities are small (1.9-3.7) and that women's labor supply to the firm
is less elas~ic than men's (which is the reverse of gender differences in
labor supply usually found at the level of the market). Our results imply
that about one third of the gender pay gap might be wage discrimination
by profit-maximizing monopsonistic employers.
This study proposes and implements a specification test for the
hypothesis that unemployment represents intertemporal labour supply
behavior. The test allows for uncertainty and endogenous unemployment.
Given standard specifications of the intertemporal labour supply model, I
find strong evidence against this interpretation of unemployment. These
results indicate the need to turn to either 1) alternative models where
unemployed workers are off a supply function or ii) more complex models
of intertemporal substitution.
Throughout the late 1980s unemployment rates remained 2-3 percentage
points higher in Canada than the U.S. We use individual microdata from the
U.S. Current Population Survey and the Canadian Survey of Consumer Finances
to study the emerging unemployment gap between the two countries. For
women, we find that the relative rise in Canadian unemployment occurred
with relative increases in per capita weeks of work. The unemployment gap
for Canadian women was driven by a rise in the probability that nonworkers
are classified as "unemployed" as opposed to "out of the labor force". For
men, the increase in unemployment was accompanied by a relative decrease in
Canadian employment rates, and an increase in the probability that men with
no weeks of work are classified as "in the labor force". A comparison of
annual work patterns and income recipiency in the two countries suggests
that Canadians of both sexes have increasingly adjusted their labor supply
to the parameters of the Canadian Unemployment Insurance system.
This paper uses aggregate birth year/calendar year level data derived
from the Current Population Survey (CPS) to estimate the effect of Social
Security wealth on the labor supply of older men in the l970s and 1980s.
The analysis focuses on the 1977 amendments to the Social Security Act
which lead to a substantial, unanticipated differential in benefits for
otherwise identical individuals depending on whether they were born before
or after 1917. This differential has become known as the benefit notch.
There are two principal differences between the present analysis and the
previous literature. First, this paper uses time-series variations in
benefit levels to estimate the relationship between benefits and labor
supply in an era when real benefits were falling for new recipients.
Second, variation in benefit levels across cohorts is used to estimate the
relationship between benefits and labor supply. The results support a
conclusion that labor supply continued to decline for the "notch babies"
who received lower Social Security benefits than earlier cohorts.