Simulation Estimation of the Convergence of the Black–White Earnings Gap and Income Dynamics*
| DOI | http://doi.org/10.1111/j.1468-0084.2011.00647.x |
| Author | Sheng‐Kai Chang |
| Published date | 01 June 2012 |
| Date | 01 June 2012 |
363
©Blackwell Publishing Ltd and the Department of Economics, University of Oxford, 2011. Published by Blackwell Publishing Ltd,
9600 Garsington Road, Oxford OX4 2DQ, UK and 350 Main Street, Malden, MA 02148, USA.
OXFORD BULLETIN OF ECONOMICS AND STATISTICS, 74, 3 (2012) 0305-9049
doi: 10.1111/j.1468-0084.2011.00647.x
Simulation Estimation of the Convergence of the
Black–White Earnings Gap and Income Dynamics*
Sheng-Kai Chang
Department of Economics, National Taiwan University, 21 Hsu-Chow Road,
Taipei 100, Taiwan (e-mail: schang@ntu.edu.tw)
Abstract
In this article, the effect of Title VII of the 1964 Civil Rights Act on the convergence of
the black–white earnings gap and income dynamics is studied through the dynamic panel
Tobit models implemented using the simulation estimators. It is found that the black–white
earnings gap declined moderately after the implementation of Title VII of the 1964 Civil
Rights Act in 1965. Based on the simulation studies of wage trajectories, it is also found
that the positive impact of the Civil Rights Act on the convergence of the black–white
earnings gap is especially significant for the group of middle-aged and highly educated
workers. Moreover, the rich dynamic structure of the earnings process is identified from
the Current Population Survey-Social Security Administration data set. It is shown that
the various sources of dynamics in the earnings process are dominated by spurious state
dependence for both blacks and whites.
1. Introduction
To raise relative minority earnings through the federal equal employment opportunity
regulations, TitleVII of the Civil Rights Act of 1964 outlawed discrimination against black
and female workers. The Equal Employment Opportunity Commission was also established
based on the law to monitor compliance with Title VII and enforce the related statute. Title
VII of the Civil Rights Act of 1964 took effect on 2 July, 1965, and subsequently a second
piece of federal legislation, Executive Order 11246, was also signed by President Johnson
on 24 September, 1965 to create the Office of Federal Contract Compliance to enforce the
law which prohibited discrimination again black and female workers by federal contractors.
The question of whether the black–white income gap converges following the imple-
mentation of Title VII of the Civil Rights Act of 1964 and Executive Order 11246 is an
important topic in labour economics. Among other things, empirically assessing the impact
ÅI gratefully acknowledge research support from the National Science Council of Taiwan (NSC 98-2410-H-
002-223). I would like to thank John Kennan, Yuichi Kitamura and James Walker for their helpful discussions and
suggestions. I am especially grateful to Kenneth Chay for providing the combined CPS-SSA data set. I would also
like to thank Victoria Prowse and the referee for their useful comments and suggestions.All errors which remain are
mine.
JEL Classification numbers: C15, C23, C24, J15, J18.
364 Bulletin
of federal civil rights policy on the earnings process will help to evaluate the efficacy of
affirmative action policy. There is still disagreement, however, on the contribution ofTitle
VII in terms of narrowing the black–white income gap.
By using aggregate time series data, Freeman (1973) and Brown (1984), for instance,
show that the black–white earnings ratio increases after 1964. With aggregate data, how-
ever, one cannot exclude from the analysis the explicit examination of factors that are
related to the income-generating process but are uncorrelated with Title VII, something
which was not done by Freeman (1973) and Brown (1984). For example, Smith and Welch
(1989) argue that education plays an important role in closing the wage gap after 1960.
This type of difficulty can be easily disentangled using a panel data methodology.Chay
and Honor´e (1998) analyse the impact of Title VII on the black–white earnings gap by
using individual-level panel data. They find that there is a significant gain in earnings for
black men living in the South following the implementation of Title VII of the 1964 Civil
Rights Act.
The data set used by Chay and Honor´e is a matched data set that links the 1973 and 1978
March Current Population Survey (CPS) to social security payroll tax records in the Social
Security Administration (SSA) by means of social security numbers. This combined CPS-
SSA data set includes men who were born between 1910 and 1939 and lived in the South in
1963, 1964, 1970 and 1971. Their motivation for using data exclusively from the South is
that TitleVII of the 1964 Civil Rights Act would have had the largest impact in this region.1
Although this combined CPS-SSA data set is ideal for studying the influence of Title
VII of the 1964 Civil Rights Act, a high percentage of the data is top-coded because of the
maximum social security tax ceiling. In addition, the top-coded income level has changed
over time. For example, the maximum social security taxable income level was $15,686 in
1963, $19,760 in 1967 and $19,259 in 1971.2All these factors make traditional linear panel
data techniques inappropriate but pose no difficulty for the simulation estimator developed
by Hajivassiliou and McFadden (1998) and Chang (2011b).
The impact of the 1964 Civil Rights Act on the convergence of the black–white
earnings gap is investigated using the simulation estimator proposed by Chang (2011b).
The censoring problems embedded in the CPS-SSA data set can be solved with this
simulation estimator. The data set used in Chay and Honor´e’s paper is applied to study the
impact of Title VII in the current article. However, instead of using a pairwise panel data
set over a 4-year period as in Chay and Honor´e’s paper, 9 years of panel data from 1963 to
1971 are used in this article. This relatively long history of earnings information is helpful
for identifying the complicated dynamic structures embedded in the income process. Three
different types of dynamic structures are expected to be identified from this matched data
set including lagged earnings, random effects and serially correlated errors.
A lagged earnings variable, not included in Chay and Honor´e’s(1998) model, is incor-
porated in the empirical model of this article with random effects plus AR(1) errors. It
has become quite normal to allow rich dynamics in economic models since Heckman
(1981) pointed out the importance of distinguishing true state dependence from spurious
state dependence. For instance, Hu (2002) applies this CPS-SSA data set to study earnings
1Furthermore, some northern states had fair employment practice laws before 1964.
2All prices are in 1982–84 dollars.
©Blackwell Publishing Ltd and the Department of Economics, University of Oxford 2011
To continue reading
Request your trial