1 The Impact of Adoption of Wrongful Discharge Laws on Wages: 1979-2014 Dissertation Paper 2 15 March 2018 Eric Hoyt, Tufts University Visiting Scholar, University of Massachusetts Amherst Economics Doctoral Candidate Abstract: While economic theory offers predictions about the effect of employment protection on wages, the empirical literature has focused primarily on employment effects. This paper uses individual-level panel data from the National Longitudinal Survey of Youth 1979 and weighted ordinary least squares and unconditional quantile difference-in-differences regressions to estimate the effect of wrongful discharge laws, a court-based form of employment protection in the United States, on average, and across the marginal distribution of, real wages. I find that one wrongful discharge law, the good faith doctrine, increases average real wages by 18.59 percent for all private-sector non-self-employed workers, 17.18 percent for nonmanufacturing workers, 13.75 percent for men, 21.68 percent for women, 21.04 percent for white workers, and 12.97 percent for nonwhite workers. My results show that wrongful discharge laws, whether by increasing workers’ bargaining power or by increasing their labor productivity, can boost workers’ wages, particularly for women, minorities, and workers in nonmanufacturing industries who have been hardest hit by recent economic changes. Keywords: wages, unconditional quantile regression, wrongful discharge laws, employment protection Acknowledgements: I am grateful for the guidance of Professors Fidan Kurtulus, Gerald Friedman, and Eve Weinbaum as well as for the numerous helpful comments of professors and fellow graduate students at the University of Massachusetts Amherst. Author Contacts: [email protected], Room 101 Dearborn House, 72 Professors Way, Tufts University, Medford, MA 02155, (413)-362-4517
39
Embed
The Impact of Adoption of Wrongful Discharge Laws on Wages ...
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
1
The Impact of Adoption of Wrongful Discharge Laws on Wages: 1979-2014
Dissertation Paper 2
15 March 2018
Eric Hoyt, Tufts University Visiting Scholar,
University of Massachusetts Amherst Economics Doctoral Candidate
Abstract:
While economic theory offers predictions about the effect of employment protection on wages,
the empirical literature has focused primarily on employment effects. This paper uses
individual-level panel data from the National Longitudinal Survey of Youth 1979 and weighted
ordinary least squares and unconditional quantile difference-in-differences regressions to
estimate the effect of wrongful discharge laws, a court-based form of employment protection in
the United States, on average, and across the marginal distribution of, real wages. I find that one
wrongful discharge law, the good faith doctrine, increases average real wages by 18.59 percent
for all private-sector non-self-employed workers, 17.18 percent for nonmanufacturing workers,
13.75 percent for men, 21.68 percent for women, 21.04 percent for white workers, and 12.97
percent for nonwhite workers. My results show that wrongful discharge laws, whether by
increasing workers’ bargaining power or by increasing their labor productivity, can boost
workers’ wages, particularly for women, minorities, and workers in nonmanufacturing industries
who have been hardest hit by recent economic changes.
N 54,185 54,185 54,185 54,185 54,185 54,185 54,185 54,185 54,185 54,185 Note: Each entry in column 1 shows output from analysis using equation 1, which is a weighted OLS fixed-effects difference and differences regression in which the dependent variable is 100 times the natural logarthim of real U.S. year 2000 hourly wages of private sector non-self-employed wage and salary employees economy-wide in 50 U.S. states. Each entry in columns 2 through 9 shows output
from analysis using equation 2, which is a a difference in differences unconditional quantile regression in which the dependent variable is a recentered influence function of 100 times the natural
logarithm of real U.S. year 2000 hourly wages of private sector non-self-employed wage and salary employees economy-wide in 50 U.S. states. The real U.S. year 2000 hourly wages are taken from
National Longitudinal Survey of Youth 1979 panel data set for the years 1979 to 2014. All models include state main effects and indicators for each year in the sample, indicators for individuals’ industry, demographics, the size of the firm they work in, if they ever change state of residence, as well as interactions between four Census-region dummies and calendar year dummies. Models are
weighted using NLSY79 sample weights. Huber-White robust standard errors are used to allow for unrestricted error correlations across observations within states.
30
TABLE 2. --- Difference in Differences Estimates of the Impact of Wrongful Discharge Laws on the Distribution of State Real Hourly
Earnings for Female Workers: Contrasting Outcomes in Years 2 and 3 Following Adoption with Years 1 and 2 Preceding Adoption,
N 22,607 22,607 22,607 22,607 22,607 22,607 22,607 22,607 22,607 22,607 Note: Each entry in column 1 shows output from analysis using equation 1, which is a weighted OLS fixed-effects difference and differences regression in which the dependent variable is 100 times the natural logarthim of real U.S. year 2000 hourly wages of female private sector non-self-employed wage and salary employees economy-wide in 50 U.S. states. Each entry in columns 2 through 9 shows
output from analysis using equation 2, which is a a difference in differences unconditional quantile regression in which the dependent variable is a recentered influence function of 100 times the natural
logarithm of real U.S. year 2000 hourly wages of female private sector non-self-employed wage and salary employees economy-wide in 50 U.S. states. The real U.S. year 2000 hourly wages are taken
from National Longitudinal Survey of Youth 1979 panel data set for the years 1979 to 2014. All models include state main effects and indicators for each year in the sample, indicators for individuals’ industry, the size of the firm they work in, if they ever change state of residence, as well as interactions between four Census-region dummies and calendar year dummies. Models are weighted using
NLSY79 sample weights. Huber-White robust standard errors are used to allow for unrestricted error correlations across observations within states.
31
TABLE 3. --- Difference in Differences Estimates of the Impact of Wrongful Discharge Laws on the Distribution of State Real Hourly
Earnings for Male Workers: Contrasting Outcomes in Years 2 and 3 Following Adoption with Years 1 and 2 Preceding Adoption,
N 29,384 29,384 29,384 29,384 29,384 29,384 29,384 29,384 29,384 29,384 Note: Each entry in column 1 shows output from analysis using equation 1, which is a weighted OLS fixed-effects difference and differences regression in which the dependent variable is 100 times the
natural logarthim of real U.S. year 2000 hourly wages of male private sector non-self-employed wage and salary employees economy-wide in 50 U.S. states. Each entry in columns 2 through 9 shows
output from analysis using equation 2, which is a a difference in differences unconditional quantile regression in which the dependent variable is a recentered influence function of 100 times the natural
logarithm of real U.S. year 2000 hourly wages of male private sector non-self-employed wage and salary employees economy-wide in 50 U.S. states. The real U.S. year 2000 hourly wages are taken
from National Longitudinal Survey of Youth 1979 panel data set for the years 1979 to 2014. All models include state main effects and indicators for each year in the sample, indicators for individuals’ industry, the size of the firm they work in, if they ever change state of residence, as well as interactions between four Census-region dummies and calendar year dummies. Models are weighted using
NLSY79 sample weights. Huber-White robust standard errors are used to allow for unrestricted error correlations across observations within states.
32
TABLE 4. --- Difference in Differences Estimates of the Impact of Wrongful Discharge Laws on the Distribution of State Real Hourly
Earnings for White Workers: Contrasting Outcomes in Years 2 and 3 Following Adoption with Years 1 and 2 Preceding Adoption,
N 30,533 30,533 30,533 30,533 30,533 30,533 30,533 30,533 30,533 30,533 Note: Each entry in column 1 shows output from analysis using equation 1, which is a weighted OLS fixed-effects difference and differences regression in which the dependent variable is 100 times the
natural logarthim of real U.S. year 2000 hourly wages of white private sector non-self-employed wage and salary employees economy-wide in 50 U.S. states. Each entry in columns 2 through 9 shows
output from analysis using equation 2, which is a a difference in differences unconditional quantile regression in which the dependent variable is a recentered influence function of 100 times the natural
logarithm of real U.S. year 2000 hourly wages of white private sector non-self-employed wage and salary employees economy-wide in 50 U.S. states. The real U.S. year 2000 hourly wages are taken from National Longitudinal Survey of Youth 1979 panel data set for the years 1979 to 2014. All models include state main effects and indicators for each year in the sample, indicators for individuals’
industry, the size of the firm they work in, if they ever change state of residence, as well as interactions between four Census-region dummies and calendar year dummies. Models are weighted using
NLSY79 sample weights. Huber-White robust standard errors are used to allow for unrestricted error correlations across observations within states.
33
TABLE 5. --- Difference in Differences Estimates of the Impact of Wrongful Discharge Laws on the Distribution of State Real Hourly
Earnings for Nonwhite Workers: Contrasting Outcomes in Years 2 and 3 Following Adoption with Years 1 and 2 Preceding
N 21,372 21,372 21,372 21,372 21,372 21,372 21,372 21,372 21,372 21,372 Note: Each entry in column 1 shows output from analysis using equation 1, which is a weighted OLS fixed-effects difference and differences regression in which the dependent variable is 100 times the natural logarthim of real U.S. year 2000 hourly wages of nonwhite private sector non-self-employed wage and salary employees economy-wide in 50 U.S. states. Each entry in columns 2 through 9
shows output from analysis using equation 2, which is a a difference in differences unconditional quantile regression in which the dependent variable is a recentered influence function of 100 times the
natural logarithm of real U.S. year 2000 hourly wages of nonwhite private sector non-self-employed wage and salary employees economy-wide in 50 U.S. states. The real U.S. year 2000 hourly wages
are taken from National Longitudinal Survey of Youth 1979 panel data set for the years 1979 to 2014. All models include state main effects and indicators for each year in the sample, indicators for individuals’ industry, the size of the firm they work in, if they ever change state of residence, as well as interactions between four Census-region dummies and calendar year dummies. Models are
weighted using NLSY79 sample weights. Huber-White robust standard errors are used to allow for unrestricted error correlations across observations within states.
34
TABLE 6. --- Difference in Differences Estimates of the Impact of Wrongful Discharge Laws on the Distribution of State Real Hourly
Earnings for Nonmanufacturing Workers: Contrasting Outcomes in Years 2 and 3 Following Adoption with Years 1 and 2 Preceding
N 42,059 42,059 42,059 42,059 42,059 42,059 42,059 42,059 42,059 42,059 Note: Each entry in column 1 shows output from analysis using equation 1, which is a weighted OLS fixed-effects difference and differences regression in which the dependent variable is 100 times the natural logarthim of real U.S. year 2000 hourly wages of private non-self-employed wage and salary employees in nonmanufacturing industries in 50 U.S. states. Each entry in columns 2 through 9
shows output from analysis using equation 2, which is a a difference in differences unconditional quantile regression in which the dependent variable is a recentered influence function of 100 times the
natural logarithm of real U.S. year 2000 hourly wages of private sector non-self-employed wage and salary employees in nonmanufacturing industries in 50 U.S. states. The real U.S. year 2000 hourly
wages are taken from National Longitudinal Survey of Youth 1979 panel data set for the years 1979 to 2014. All models include state main effects and indicators for each year in the sample, indicators for individuals’ demographics, the size of the firm they work in, if they ever change state of residence, as well as interactions between four Census-region dummies and calendar year dummies. Models
are weighted using NLSY79 sample weights. Huber-White robust standard errors are used to allow for unrestricted error correlations across observations within states.
35
TABLE 7. --- Difference in Differences Estimates of the Impact of Wrongful Discharge Laws on the Distribution of State Real Hourly
Earnings for Manufacturing Workers: Contrasting Outcomes in Years 2 and 3 Following Adoption with Years 1 and 2 Preceding
N 9,438 9,438 9,438 9,438 9,438 9,438 9,438 9,438 9,438 9,438 Note: Each entry in column 1 shows output from analysis using equation 1, which is a weighted OLS fixed-effects difference and differences regression in which the dependent variable is 100 times the
natural logarthim of real U.S. year 2000 hourly wages of private sector non-self-employed wage and salary employees in manufacturing industry in 50 U.S. states. Each entry in columns 2 through 9
shows output from analysis using equation 2, which is a a difference in differences unconditional quantile regression in which the dependent variable is a recentered influence function of 100 times the
natural logarithm of real U.S. year 2000 hourly wages of private sector non-self-employed wage and salary employees in manufacturing industry in 50 U.S. states. The real U.S. year 2000 hourly wages
are taken from National Longitudinal Survey of Youth 1979 panel data set for the years 1979 to 2014. All models include state main effects and indicators for each year in the sample, indicators for individuals’ demographics, the size of the firm they work in, if they ever change state of residence, as well as interactions between four Census-region dummies and calendar year dummies. Models are
weighted using NLSY79 sample weights. Huber-White robust standard errors are used to allow for unrestricted error correlations across observations within states.
36
Figure 1. --- Real Hourly Earnings for Manufacturing Workers Before and After Adoption of the Implied Contract Doctrine: Yearly
Leads From 2 Years Before to 3 Years After Adoption, 1979-2014
100 x ln(Real Hourly Wage)
Note: The figure shows dynamic effect of the good faith doctrine on the fourth decile of wages using equation 3, which is a difference and differences unconditional quantile regression with two years
of leads and lags in which the dependent variable is 100 times the natural logarithm of real U.S. year 2000 hourly wages of private sector non-self-employed wage and salary employees in manufacturing industries in 50 U.S. states. The real U.S. year 2000 hourly wages are taken from National Longitudinal Survey of Youth 1979 panel data set for the years 1979 to 2014. All models
include state main effects and indicators for each year in the sample, indicators for individuals’ demographics, the size of the firm they work in, if they ever change state of residence, as well as
interactions between four Census-region dummies and calendar year dummies. Models are weighted using NLSY79 sample weights. Huber-White robust standard errors are used to allow for
unrestricted error correlations across observations within states.
-8
-4
0
4
8
12
-2 -1 0 1 2
Point Estimate Robust 90 Percent Confidence Interval
Q(0.40)
Year Relative to Adoption
37
TABLE 8. --- Difference in Differences Estimates of the Impact of Wrongful Discharge Laws on the Distribution of State Real Hourly
Earnings for Communication Workers: Contrasting Outcomes in Years 2 and 3 Following Adoption with Years 1 and 2 Preceding
N 26,179 26,179 26,179 26,179 26,179 26,179 26,179 26,179 26,179 26,179 Note: Each entry in column 1 shows output from analysis using equation 1, which is a weighted OLS fixed-effects difference and differences regression in which the dependent variable is 100 times the
natural logarthim of real U.S. year 2000 hourly wages of private sector non-self-employed wage and salary employees in the communication industry in 50 U.S. states. Each entry in columns 2 through
9 shows output from analysis using equation 2, which is a a difference in differences unconditional quantile regression in which the dependent variable is a recentered influence function of 100 times the
natural logarithm of real U.S. year 2000 hourly wages of private sector non-self-employed wage and salary employees in the communication industry in 50 U.S. states. The real U.S. year 2000 hourly
wages are taken from Current Population Survey Monthly Outgoing Rotation Group files for the years 1979 to 2013. All models include state main effects and indicators for each year in the sample, indicators for individuals’ demographics, as well as interactions between four Census-region dummies and calendar year dummies. Models are weighted using CPS earnings weights. Huber-White
robust standard errors are used to allow for unrestricted error correlations across observations within states.
38
TABLE 9. --- Difference in Differences Estimates of the Impact of Wrongful Discharge Laws on the Distribution of State Real Hourly
Earnings for Transportation Workers: Contrasting Outcomes in Years 2 and 3 Following Adoption with Years 1 and 2 Preceding
N 65,756 65,756 65,756 65,756 65,756 65,756 65,756 65,756 65,756 65,756 Note: Each entry in column 1 shows output from analysis using equation 1, which is a weighted OLS fixed-effects difference and differences regression in which the dependent variable is 100 times the
natural logarthim of real U.S. year 2000 hourly wages of private sector non-self-employed wage and salary employees in the transportation industry in 50 U.S. states. Each entry in columns 2 through 9
shows output from analysis using equation 2, which is a a difference in differences unconditional quantile regression in which the dependent variable is a recentered influence function of 100 times the
natural logarithm of real U.S. year 2000 hourly wages of private sector non-self-employed wage and salary employees in the transportation industry in 50 U.S. states. The real U.S. year 2000 hourly
wages are taken from Current Population Survey Monthly Outgoing Rotation Group files for the years 1979 to 2013. All models include state main effects and indicators for each year in the sample, indicators for individuals’ demographics, as well as interactions between four Census-region dummies and calendar year dummies. Models are weighted using CPS earnings weights. Huber-White
robust standard errors are used to allow for unrestricted error correlations across observations within states.
39
TABLE 10. --- Estimates of the Impact of Wrongful Discharge Laws on the Probability Women Are Low-Wage Workers, 1979-2014
Family of Two,
100% Federal Poverty
Level
Family of Three,
125% Federal Poverty
Level
Implied Contract 0.00 -0.01
(0.01) (0.02)
Public Policy -0.04** -0.04*
(0.01) (0.02)
Good Faith -0.05* -0.07*
(0.02) (0.03)
Pseudo R-Sq 0.16 0.16
N 49,263 49,263 Note: Each entry in columns 1 and 2 shows output from analysis using equation 4, which is a difference and differences probit model, in which the dependent variable is, respectively, a dummy that
indicates if female private sector non-self-employed wage and salary employees in 50 U.S. states are paid, respectively, a wage less than or equal to an amount sufficient to bring a family of two (i.e. one
parent and one child) a yearly income equivalent to the fedral poverty level, or a wage less than or equal to an amount sufficient to bring a family of three (i.e one parent and two children) a yearly
income equivalent to 125% of the federal poverty level. These dichotomous dependent variables are constructed from the data in National Longitudinal Survey of Youth 1979 on family size, number of children, as well as the real U.S. year 2000 hourly wages described previously, for the years 1979 through 2014. All models include state main effects and indicators for each year in the sample,
indicators for individuals’ industry, the size of the firm they work in, if they ever change state of residence, as well as interactions between four Census-region dummies and calendar year dummies..
Models are weighted using NLSY79 sample weights. Huber-White robust standard errors are used to allow for unrestricted error correlations across observations within states.