[LIBRARY

Michigan State

Unlverslty

 

 

PLACE IN RETURN BOX
to remove this checkout from your record.

TO AVOID FINES return on or before date due.

 

DATE DUE

DATE DUE

MTE DUE

 

AUG 9 6 2000

W

 

 

was {4025

 

65‘

‘JH
Cf
D

 

 

 

 

 

 

 

 

 

 

 

ma cm.pQ-p.14

 

THE LABOR MARKET EFFECTS OF MATERNITY MANDATES
By

Rosella M. Gardecki

A DISSERTATION

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY

Department of Economics

1 997

ABSTRACT
THE LABOR MARKET EFFECTS OF MATERNITY MANDATES
By

Rosella M. Gardecki

The increased reliance on mandates as a tool for providing benefits has prompted a
renewed interest in the labor market effects that these mandates have on the group they are
intended to benefit. By enacting a group-Specific mandate, legislatures provide a socially
desirable benefit that is not readily available in the mandate’s absence. Although this type of
government intervention can be efficient, it may have an adverse impact on workers either
when they place a low value on receiving the benefit or when wage rigidities exist.

This paper focuses on the labor market effects of maternity mandates enacted during
the 19708 and early 19805. The staggered enactment of these mandates at the state and
federal levels provides an opportunity to study their impact on wages and employment. In
the early to mid-19708, a number of states adopted employer mandates requiring firms to
cover pregnancy in the same manner as other temporary disabilities. However, no national
standard existed until the enactment of the 1978 federal Pregnancy Discrimination Act.

This paper presents estimates of the labor market effects of the state employer
mandates (state study) and the PDA (federal study). In the State study, the wage declines
with no change in net labor input. In contrast, the federal study suggests that strong
disemployment effects resulted from the passage of the PDA, with no corresponding decline
in the wage. A re-estimation of Jonathan Gruber’s federal study (1994), based on state
insurance code amendments instead of the state employer mandates, does not conflict with

this result.

In this paper, I hypothesize that the apparent conflict between the state study and the
federal study results may be due to variations in the expected costs and benefits across states.
States only adopt the level of state maternity legislation considered efficient. Thus, workers
in the states that enacted mandates place a higher value on receiving the mandate’s benefits.

Before deciding whether to implement a group-specific mandate, policy makers need
to recognize both potential long-term gains and possible disemployment effects. Careful

analysis of these factors will provide a framework on which future decisions may be made.

‘It is somewhat conspicuous a truth as the interaction of economics and law should have
waited so long for recognition—a recognition by no means universal. Some of those who
question it maintain the independence and self-sufi‘iciency of law, while others maintain that
of economics. In reality, law and economics are ever and everywhere complementary and
mutually determinative. . .. ’

Fritz Berolzheimer, I 912

iv

ACKNOWLEDGMENTS

Completing my graduate education would not have been possible without the
assistance of many people. David Neumark, the chair of my committee, provided invaluable
guidance throughout the dissertation process. He also instructed me in the finer points of
research during my tenure as his research assistant. Harry Holzer and Robert LaLonde,
members of my committee, offered direction and counsel; their recommendations greatly
enhanced the final product.

I am indebted to the participants in the Labor Lunch Group at Michigan State
University for their suggestions. In particular, I would like to thank Kathleen Beegle and
Jennifer Tracey for their insightful comments on various drafts. I am also grateful to
Jonathan Gruber for generously providing me with his data and programs.

Many friends offered their support throughout this process. Among others too
numerous to mention, I was fortunate to count Janette Canare, Jill Moran, and Jennifer
Wilcox as my closest confidants throughout my graduate school career. I reserve Special
thanks for Amanda McClain, who not only helped with the editing and formatting but
also offered words of encouragement during the final stages.

Finally, I am more grateful to my family than words can express. My brothers,
Joseph and John, provided support at critical points in the process. Most important, I am
fortunate to have parents who encouraged me with their love and patience during the writing

of my dissertation and throughout my entire education. You have my heartfelt gratitude.

TABLE OF CONTENTS

Introduction - ......................

Chapter 1: Historical Motivation

 

1. Action Prior to 1972

 

 

II. Equal Employment Opportunity Commission Guidelines—1972
[11. Relevant Maternity Legislation Enacted During 1972 to 1976

 

IV. Gilbert v. General Electric Company—US. Supreme Court, 1976
V. State Response to the Gilbert Decision—1976 to 1978
VI. Congressional Response to the Gilbert Decision—1978
VII. Conclusions
Bibliography

 

 

 

 

 

Chapter 2: The Effects of Mandated Maternity Benefits .....
I. Theoretical Review
11. Historical Motivation

A. Pregnancy Discrimination Guidelines, 1972 .................................................................................
B. 1976 Court Decisions ....................................................................................................................
C. Federal Pregnancy Discrimination Act, 1978 ................................................................................
III. DDD Estimation Techniques
IV. Previous Empirical Work
V. State Employer Mandates
A. Main Data Source ..........................................................................................................................
B. Methodology .................................................................................................................................
C. Results ...........................................................................................................................................
D. Group-Specific Effects ..................................................................................................................
VI. The Federal PDA

Main Data Source ..........................................................................................................................

Methodology .................................................................................................................................

Results ...........................................................................................................................................

. Group-Specific Effects ..................................................................................................................

Covered/Non-Covered Group

State Study Results ........................................................................................................................

.Federal Study Results ....................................................................................................................

VIII. Disparity between State and Federal Study Results

 

 

 

 

 

 

V

 

weFocwe

 

 

 

IX. Conclusions
Bibliography
Chapter 3: The Incidence of Mandated Maternity Benefits, A Comment ....................

 

I. Types of State Maternity Policy
II. Replication of Gruber’s State Study

 

vi

III. Impact of Employer Mandates on Gruber’s State Study 79

 

 

 

 

 

 

IV. Replication of Gruber’s Federal Study 81
V. Impact of Employer Mandates on Gruber’s Federal Study 82
VI. Conclusions 84
Bibliography 90
Conclusions - - -- ------------------- - -_ - ................ 91
Appendices 96

Bibliography 102

 

vii

LIST OF TABLES

Chapter 1

Table 1: States with Employer Maternity Mandates .............................................................. 15
Chapter 2

Table 1: Dates and Status of the Employment Mandates ....................................................... 57
Table 2: DDD Regression Framework ................................................................................... 59
Table 3: State Employer Mandates, Full Sample ................................................................... 60
Table 4: State Study by Type of Worker: Large Sample ........................................................ 61
Table 5: Federal Pregnancy Discrimination Act, Full Sample ............................................... 62
Table 6: Federal Study by Type of Worker: Large Sample .................................................... 63
Table 7: Labor Market Effects of State Employer Mandates: Small Sample ......................... 64
Table 8: State Study by Type of Worker: Small Sample ........................................................ 65
Table 9: State Study by Type of Worker: Covered Group Only ............................................. 66
Table 10: Labor Market Effects of the Federal PDA: Small Sample ..................................... 67
Table 11: Federal Study by Type of Worker: Small Sample .................................................. 68
Table 12: Federal Study by Type of Worker: Covered Group Only ....................................... 69
Table 13: Legislation Favorable to Females as of 1978 ......................................................... 70
Chapter 3

Table 1: Results from Replicating and Correcting Gruber’s Study.........; .............................. 86
Table 2: States, Enactment Dates and State Status from Gruber’s State Study ..................... 87
Table 3: States, Enactment Dates and Status from the Federal Study .................................... 88
Appendices

Appendix A: Dates of Compliance by Maternity Policy ........................................................ 96
Appendix B: Mandate States and Coverage for State Study .................................................. 98
Appendix C: PDA States and Coverage for Federal Study .................................................... 99
Appendix D: State Study Small Sample, Title VH Workers ................................................ 101

viii

INTRODUCTION

In recent years, federal and state governments have enacted a number of group-
specific mandates. These mandates provide a socially desirable benefit, which most firms do
not voluntarily offer, to a demographically identifiable group of workers. Despite the
increased reliance on using mandates as a tool for providing benefits, however, few studies
attempt to isolate the resulting labor market effects. The following chapters focus on the
legislation leading to, and the wage and employment effects that result from, the enactment
of maternity mandates. These policies require employers to cover the medical costs of
pregnant workers in the same manner as other temporary disabilities.

Ideally, group-specific mandates correct for a perceived failure of the market to
provide a socially desirable benefit. In the absence of a mandate, market failure may occur if
the cost of providing the benefit is higher than the wage. This results because firms that offer
maternity insurance attract those who value this benefit. Since these workers select into the
firm based on the benefit’s availability, they have a higher probability of collecting the
insurance than that of the general population. Higher premiums result, which leads to an
increase in the cost of the maternity insurance for the firm. Over time, these premiums
become prohibitively high for an individual firm.

By enacting a mandate, the government requires all firms to provide the benefit.
Adverse selection does not occur, Since each employer is legally obligated to offer workers at

least the minimum requirements of the mandate. Workers who value the mandate implicitly

1

2
accept a lower wage in order to receive its benefits. When workers fully value the mandate,
the decrease in the wage equals the employer’s cost of providing the benefit. There is no
corresponding decrease in employment. Conversely, when the value that workers place on
the mandate is less than the cost to the employer, disemployment effects accompany a wage

decrease.

Enacting this type of group-specific legislation, however, may adversely affect
individual workers under its protections, regardless of social or worker valuation. For
example, the efficiency of a mandate decreases if employers respond to the increase in the
firm’S cost, which is dependent on the number of mandate-protected workers, by substituting
away from these employees. The same is true if market barriers, such as the minimum wage,
prevent the wages of workers from adjusting downward to reflect the value placed on the
benefit. In the case of group-specific mandates, anti-discrimination legislation and equal pay
laws may also prevent the group’s wages from adjusting, regardless of their valuation of the
benefit. In either scenario, disemployment effects result. These disemployment effects

increase if the value workers assign to the benefit decreases relative to its labor market cost.

Inefficiencies at the legislative and judicial levels may also lessen the effect of a
mandate. These inefficiencies may result from the ambiguous—and sometimes
conflicting—policies on the same issue adopted by different agencies, from a lack of
enforcement, from incomplete information on the coverage of the law, or from varying
policies between states. This paper explores the effects of various maternity policies that
resulted from these inefficiencies. Chapter 1 describes the legislative and judicial activity—
and ambiguity—surrounding the enactment of the state and federal maternity mandates,

which require employers to cover pregnancy disabilities in the same manner as other

3
temporary disabilities. The enactment of and changes in these policies provide a basis for
examining the labor market consequences of group-Specific mandates.

Chapter 2 estimates the labor market impact of federal and state maternity mandates
that require employers to cover pregnancy in the same manner as other temporary
disabilities. Next, it discusses how differences in worker valuation may lead to different
levels of enactment among states. This chapter posits that states will enact a mandate only to
the level that the marginal social benefit of enacting it at the state level is greater than the
benefit’s marginal cost. States that do not enact legislation before a federal mandate perceive
the marginal benefits associated with any state legislation to be below the mandate’s
marginal cost. Thus, unequal passage of a law at the state level may result from a market
failure in the individual state or because these states actively choose not to adopt similar
legislation.

After covering the legal aspects in chapter 1 and the economic aspects in chapter 2,
this paper comments on a previous study of the state and federal maternity mandates. The
study in question, examined in chapter 3, focuses on a select set of insurance code
amendments but ignores other maternity policies in place at that time. In chapter 3, the
coefficients of the original study are re-estimated, taking into account these other policies.

The results found in chapters 2 and 3 are contrasted in the conclusion.

CHAPTER 1: HISTORICAL MOTIVATION

Determining the labor market effects of requiring employers to cover the medical
costs of pregnant workers requires a familiarity with the complicated history behind the
various pieces of legislation. A number of state and federal maternity mandates became
effective during the 19703 and 19803; in many instances, there is a close relationship
between these policies. Maternity guidelines issued by the Equal Employment Opportunities
Commission (EEOC) in 1972, for example, were adopted by various states in the form of
state employer mandates during the period from 1972 to 1976. In addition to these state
maternity policies, which required the medical coverage of pregnant workers, other maternity
policies enacted at that time included maternity amendments to state insurance codes and
various federal agency regulations.

This chapter provides a discussion of the legislative and judicial background to the
state and federal maternity mandates upon which the empirical estimates in chapters 2 and 3
are based. In addition, this chapter reviews other relevant pieces of legislation that may

affect the later estimates.

1. Action Prior to 1972

The original version of Title VH of the Civil Rights Act of 1964 focused on innate
characteristics, and as such, did not define prohibited employer treatment of pregnant
employees. In 1966, the first national statement regarding the rights of pregnant workers

4

5
came as an opinion letter issued by Charles Duncan, the EEOC General Council, on October
17, 1966.1 He writes, “we do not believe that an employer must provide the same fringe

benefits for pregnancy as he provides for illness.”2

This letter reinforced prevailing trends.
On June 9, 1970, the Office of Federal Contract Compliance (OFCC) of the US.
Department of Labor issued the first federal statement regarding the treatment of pregnant
workers by federal contractors. The Federal Register published the following regulations:
“Women shall not be penalized in....employment...on account of childbearing. The
conditions applicable to her leave...shall be in accordance with the employer’s leave

”3 Until 1972, this was the only legal reference at the national level to the treatment

policy.
of pregnant workers.4 As a result, this statement formed the standard for employer maternity
policy in firms covered under a government contract.

In March 1972, congress amended Title VII to include sex discrimination as a
prohibited practice. The text of the amendment, however, did not explicitly mention the

treatment of pregnant workers. As a result, determining whether pregnancy was a form of

sex discrimination became an important legal issue.

11. Equal Employment Opportunity Commission Guidelines—1972
As early as April 5, 1972, the EEOC issued guidelines that interpreted the sex

discrimination clause in Title VII to include discrimination due to pregnancy and childbirth.

 

‘ Many states and federal agencies had compulsory maternity leave policies at that time. These policies required
pregnant employees to take leave of absence at a specified period in their pregnancy, regardless of their ability to
perform their job. Although these laws mandated the leave period, they did not provide any rights (e.g., return to
job) following birth. In 1974, the US. Supreme Court ruled in Cleveland Board of Education v. LaFleur and
Cohen v. Chestfield Countr School Board (414 US 632, 94 S.Ct. 791 [1974]) that the mandated leave period
violated due process.

2 US. Supreme Court 1976/77, excerpt from an EEOC opinion letter dated November 15, 1966.

3 Title 41. Chapter so, Part 60-20 of the Code of Federal Regulations.

4 At that time, a few States had instituted maternity policies other than those relating to compulsory leave. For
example, Rhode Island began covering maternity conditions under its state-run temporary disability insurance
system in 1964. See appendix A for other examples.

6
By broadening the focus to include “the addition of characteristics over which individuals
clearly do exert control to the bases for prohibited employment discrimination,” these
guidelines represented a clean break from previous EEOC policy.S Among the prohibited

conditions, these guidelines stated:

Disabilities caused or contributed to by pregnancy... and recovery therefrom

are, for all job-related purposes, temporary disabilities and should be treated

as such under any health or temporary disability insurance or Sick leave plan

available in connection with employment. Written and unwritten

employment policies and practices involving matters such as... payment

under any health or temporary disability insurance or sick leave plan, formal

or informal, shall be applied to disability due to pregnancy or childbirth on

the same terms and conditions aS they are applied to other temporary

disabilities.6
The EEOC, however, did not seek congressional approval in 1972 for the amendment of
Title VII to include the maternity provisions. Consequently, the maternity guidelines were
not enforceable in the same manner as other EEOC regulations, such as those governing sex
discrimination.

At that time, the EEOC practice was to target employment discrimination by
assigning priority to charges brought against large, nationwide employers or major employers
in a region.7 Federal Communications Commission (FCC) Case No. 19143 was the first test
case settled under the 1972 pregnancy guidelines. In this case, the EEOC joined a number of
other federal agencies in proceedings against American Telephone and Telegraph (AT&T)

and 24 Bell Companies.8 The EEOC investigation focused on charges of sex discrimination,

which included pregnancy discrimination. A January 1973 agreement outlined the terms

 

5 Smith 1980, 499.

" Weyland 1978, 180.

7 American Bar Association 1974. 577-615.

8 In March 1972. the EEOC received the power to subpoena employers. This case was the first settled under the
power of subpoena, as well as under the pregnancy guidelines.

7
under which AT&T and the Bell Companies were in “full compliance with all laws and

9 . . .
” However, these conditions did not

regulations governing equal employment opportunity.
require the company to abolish a policy that denied full insurance benefits to pregnant
employees. Since this was the first case settled under the EEOC’S pregnancy guidelines,
many state agencies and employers subsequently cited the FCC case in defending their
position that the EEOC intended the 1972 guidelines to be recommendations only.lo
Following the AT&T decision, district-level EEOC offices began accepting claims
against employers related to pregnancy discrimination.ll Regardless, the change in the
EEOC’S original intent, combined with the failure to amend Title VII and the lack of
enforcement in a national case, created confusion over whether the guidelines were merely
recommendations or were legally binding. A number of states postponed implementing any

pregnancy discrimination guidelines pending an US. Supreme Court decision or a

congressional amendment to Title VII.

III. Relevant Maternity Legislation Enacted During 1972 to 1976

Although some states chose to wait for a judicial precedent to be established, a few
state Fair Employment Practice (FEP) commissions interpreted state sex discrimination laws
in the same manner as the EEOC in the 1972 guidelines. Other states, such as Alaska, used
the legislature to address the issue of discrimination against pregnant workers. As of

December 1976, seven states had implemented legislation specifically prohibiting pregnancy

 

9 us. Supreme Court 1976177, 67.

"’ US. Supreme Court 1976177, 68-70.

" States that successfully adopted the pregnancy guidelines were generally states with ‘706‘ agencies. In these
States, the EEOC deferred to the State agency decisions on FEP matters. To become a ‘706’ agency, states were
required to adopt and enforce the EEOC regulations and guidelines.

8

discrimination.l 2 Regardless, many of the employers in these states chose to wait for a state
supreme court ruling or a state law prohibiting pregnancy discrimination.l3 For example, the
director of the Connecticut FEP agency stated, ‘A number of large employers with federal
contracts have complied with the statutory mandate... Other large employers appear to be
awaiting the Supreme Court decision...’ ‘4

Aside from the state employer maternity mandates, various states or federal agencies
adopted other pieces of maternity legislation during the period. Five states—Califomia,
Hawaii, New Jersey, New York, and Rhode Island—chose to provide health insurance for

employees under an existing state-run system.’5

The state temporary disability insurance
(TDI) systems provided insurance for non-occupational disabilities when workers were
unable to receive workers’ compensation benefits or unemployment insurance. To provide
funds to cover the costs of temporary disabilities, the state TDI systems collected a payroll
tax either from workers or from both workers and their employers.“5 In these states,
pregnancy was added to the list of temporary disabilities covered under the state TDI systems
(see appendix A for dates of enactment).

In addition, a number of states also chose to provide limited maternity coverage

under an amendment to an existing insurance code. These states mandated that all health

 

'2 These are the following: Alaska, Connecticut, Kansas, Massachusetts, Montana, Pennsylvania, and
Washington (see table 1 for the dates of adoption).

’3 Jan A. Larkin, the director of the anti-discrimination section of the Delaware Department of Labor, writes in a
letter to Ruth Weyland, attorney for Martha Gilbert in Gilbert v. General Electric Company (1976), “Based on
our experience with employers’ practices in Delaware, it is not common for income maintenance plans or other
benefit plans to include payment for pregnancy and pregnancy related conditions.” Letters to Ruth Weyland
from agencies in Wyoming and Maine are similar in content. US. Supreme Court 1976/77, 318-26.

" U.S. Supreme Court 1976177, 318-26.

'5 The exception is Hawaii, which created a state-run system in 1974.

’6 Summers’ (1989) and Gruber’s (1992, 1994) theories predict a different outcome for states that implement a
benefit through a payroll tax versus states that implement a benefit through a mandate. Consequently, I limit the
details of the state TDI legislation to that which is necessary to explain the state systems.

9
insurance contracts sold within the state cover complications that arose due to pregnancy.l7
Although not specific to workers, these amendments, by default, covered pregnant workers in
a firm that purchased health insurance for its employees. Most of these states enacted this
amendment between 1976 and 1978 (see appendix A for dates of adoption). This coincided
with the US. Supreme Court’s ruling on the legality of the EEOC’S maternity guidelines and,
by extension, the legality of the state employer mandates.

At the federal level, confusion over the legal treatment of pregnant workers grew on
December 27, 1973, when the Federal Register published the OFCC’S proposed revised
guidelines. These guidelines extended the previous OFCC policy of ‘justifiable leave for
pregnancy conditions’ to more closely follow the wording of the EEOC’S pregnancy
discrimination guidelines. Enacting the revised guidelines would have affected all firms
covered by a government contract.18 However, the Secretary of Labor noted that the OFCC
acknowledged a continued reliance on pertinent judicial decisions, especially those at the
level of the US. Supreme Court.19 At the time of the US. Supreme Court’s decision in
Gilbert v. General Electric Company, (December 7, 1976) the OFCC guidelines had not
been confirmed.

Additionally, the Department of Health, Education, and Welfare (HEW) issued
departmental guidelines prohibiting pregnancy discrimination in educational institutions
receiving federal funding. The October 1972 regulations abolished the common practice of

mandatory leave for maternity conditions (see footnote 1). Thus, those who worked in such

 

'7 Maryland, which amended the state insurance code in 1975, is the exception. In Maryland, only normal
regnancies—not complications of pregnancy—were covered under the state’s insurance code amendment.
3 38 Fed. Reg. 35336.
'9 US. Supreme Court 1976/77, 46. The issue of concern for the OFCC seems to be that the unequal benefits
implied by the EEOC’S guidelines conflicted with the Equal Pay Act (29 U.S.C. 206 (d)).

10
educational institutions were no longer required to take leave of their position at a pre-set
date.

In 1975, HEW supplemented these regulations prohibiting discrimination against
pregnant workers in employment and required employers to treat pregnancy as a temporary
disability for all job-related purposes—including the provision of health insurance for
pregnancy conditions. Following congressional approval, the Federal Register published the
HEW regulations on July 21, 1975.20 Unlike the OFCC guidelines, these regulations were
legally binding and directly affected employees in educational institutions receiving federal
funding.21

Since not all state FEP guidelines covered pregnancy conditions and enforcement of
the EEOC guidelines was inconsistent at best, there was confusion over which guideline
applied to a specific workforce or firm. A 1973 trade journal summed up the maternity
legislation at the time, stating:

The language and intent of the guidelines appear contradictory and

ambiguous... Other significant contributors to the confusion are the lack of

accountability over the EEOC and OFCC which often results in an abuse of
power; the lack of conformity among EEOC and OFCC officials in giving out

correct information; and the hesitancy of the governmental agencies to

distribute information that would keep businesses up to date with what is
really expected of them.22

23

Regardless, a number of state courts ruled in favor of the state FEP guidelines. Four of

these rulings occurred at the state supreme court level in Wisconsin, Iowa, Pennsylvania, and

 

2" Title 45, Subtitle A, Part 86 of the Code of Federal Regulations.

2' US. Supreme Court 1976/77, 20-1.

22 Pearson 1973, 21.

2" For example see: Ray-O-Vac Div. of E.S.B., Inc. v. Wisconsin Department of Industry, Labor, and Human
Relations.

11
Massachusetts, making these guidelines legally binding for all employers in the state without

any further legislation.24

IV. Gilbert v. General Electric Company—US. Supreme Court, 1976

In 1976, the test of the 1972 EEOC guideline’s legality for which many states,
employers, and regulatory bodies had waited finally occurred. Gilbert v. General Electric
Company (December 7, 1976) brought the question of the legality of the EEOC’S 1972
guidelines before the US. Supreme Court.25 In a 6-3 ruling, the Court reversed the decisions
of Six federal appellate courts, which had stated that women who bear children must receive
the same benefits that employers provide for other non-occupational disabilities.26 The court
ruled that the distinction in this case was between pregnant workers and non-pregnant
workers, not between male and female workers. This effectively ended further judicial
interpretation of the EEOC’S pregnancy discrimination guidelines under the sex
discrimination defense. Any previous court decisions based on state guidelines or the EEOC

1972 guidelines were now subject to re-interpretation by the appropriate lower state courts.27

 

2" Cedar Rapids School Dist. v. Parr, 227 N.W.2d 486 (Iowa Sup. Ct. 1975); Black v. School Dist. of Maiden,
310 N.E.2d 330 (Mass. Sup. Jud. Ct. 1974); Cerra v. East Stroudsberg Area School Dist., 299 A.2d 277 (Sup.
Ct. Pa. 1973); Ray-O- Vac Div. OfESB. Inc. v. Wisconsin Dep’t of lndus., 2376 N.W.2d 209 (Wis. Sup. Ct. 1975)
as found in Weyland, 1978, 182.

25 It should be noted that preceding the Gilbert decision, the US. Supreme Court decided in Geduldig v. Aiello
(417 US. 484 [1974]) that the state of Califomia’s TDI system was not required to fund pregnancy conditions
under the equal protection clause in the 14lh amendment. Further claims of unequal protection in pregnancy
cases were no longer valid following this decision. The Gilbert case differed from Aiello in that the EEOC
guidelines were judged on the basis of sex discrimination, not on the basis of equal protection.

6 The relevant cases at the federal appellate level are: Berg v. Richmond Unified School District, 528 F.2d 1208
(9th Cir. 1975); Gilbert v. General Electric Co., 519 F.3d 661 (4th Cir., 1975); Holthaus v. Compton and Sons,
Inc., 514 F.2d 1024 (2nd Cir. 1975); Wetzel v. Liberty Mutual Ins. Co., 511 F.2d 199 (3rd Cir., 1975); and
Farkas v. South Western City School Dist., 506 F.2d 1400 (6th Cir., 1974).

27 In Nashville Gas Co. v. Satty (434 U.S. 136 [1977]), however, the Court ruled against a requirement that
women forfeit seniority following maternity leave, since it subjected women to a burden that men need not
suffer.

12
V. State Response to the Gilbert Decision—1976 to 1978

The first case the Supreme Court’s decision could potentially influence was Brooklyn
Union Gas Co. v. N)". State Human Rights Appeals Board (1976). In Brooklyn, the New
York State Supreme Court upheld the 1973 New York State Human Rights Commission’s
employer guidelines that mandated maternity coverage.28 This case is instructive Since it set
a precedent for other states and other state courts to interpret the Supreme Court’s ruling as
applying to the 1972 EEOC guidelines only and not to individual state guidelines. This set
precedence for other states with employer mandates already in place.

Because of Brooklyn, existing state employer mandates could still be subject to the
sex discrimination defense if the state courts chose to interpret them in that manner.
Intermediate courts in Pennsylvania and Wisconsin followed the precedent set by the New
York State Supreme Court in Brooklyn.29 States were able to implement new state maternity
legislation or to strengthen existing pregnancy discrimination guidelines and laws through
their legislatures. The human or civil rights commissions in 18 states upheld previous state
guidelines after the decision of the New York court.30 As a result of the state guidelines—in
conjunction with the Supreme Court’s decision—a clear legal division developed between
states that enforced the employer mandates and states that neither enforced nor instituted

employer mandates after that date.

 

2" 316 N .Y.S.Ct. Appellate Division 1973, 6 EPD 1] 8713. This ruling also applied to the TDI system.

2" Weyland 1978, 202.

3" These are: Alaska, Connecticut, Illinois, Indiana, Iowa, Kansas, Maryland, Massachusetts, Michigan,
Minnesota, Missouri, New Jersey, New York, Oregon, Pennsylvania, South Dakota, Washington, and Wisconsin.
Weyland 1978, 204. See table 1 for dates of enactment.

13

VI. Congressional Response to the Gilbert Decision—1978

In March 1977, members of Congress responded to the US. Supreme Court’s ruling
in Gilbert by introducing a bill into both houses that would amend the sex discrimination
prohibition in Title VII to include discrimination against pregnant workers.31 The bill
qualified violations of Title VII to include, “discrimination based on pregnancy, childbirth or
other related medical conditions.”32 Passage of the federal Pregnancy Discrimination Act of
1978 (PDA) required employers to treat women affected by pregnancy and complications
arising from pregnancy in the same manner as workers with other types of non-occupational
disabilities. Employers without temporary insurance coverage for other non-occupational
disabilities were not subject to the maternity provisions in Title VII and were not required to
provide maternity insurance. In addition, the federal PDA stipulated that employers who
provided benefits to a spouse under a health insurance plan must also provide maternity
benefits to the spouse.33 The effective compliance date for employers with benefit programs

was 180 days after the date of enactment, which was October 31, 1978.

VII. Conclusions

Although the PDA ended confusion over the treatment of pregnant workers
concerning health insurance coverage, the labor market outcomes associated with the
maternity mandate are unclear. By examining various state and federal maternity policies,

the wage and employment effects that result may be determined. Chapter 2 presents an

 

3' Simultaneous bills were introduced into both houses of Congress (HR. 5055, S. 995).

32 Pregnancy Discrimination Act of 1978, Pub. L. No. 95.555, 92 stat 2067 (1978) amending 42 US. C § 20000e
(k)-

33 Some state maternity mandates also required spousal coverage. See, for example, guidelines issued by the
Wisconsin Department of Industry, Labor and Human Relations (1975) and the Washington State Human Rights
Commission (1973). The position of equal treatment for spouses under the federal PDA was later upheld in
Newport News Shipbuilding & Dry Dock Co. v. Equal Employment Opportunity Commission, 462 US. 669
(1983).

l4
analysis of the state employer maternity mandates and the federal PDA. In chapter 3, the
results from a study of the State insurance code amendments are re-estimated to take the
employer mandates into account. From these results, the value that workers place on the
benefits afforded by the mandate and by extension, the value of the mandate to workers may

be ascertained.

States:
Alaska

California
Colorado3
Connecticut
Illinois
Iowa

Kansas

Maine3

Maryland2

Massachusetts

Michigan2
Minnesota2

Montana

New York2
Oregon2
Pennsylvania2

South Dakota

Washington

Wisconsin

Notes:

15

Table 1: States with Employer Maternity Mandates

Law, Guideline. or State Court Decision:
Section amending 18.80.220 of Alaska Laws
Against Discrimination

Fair Employment Practice Act

Section 80.8 Fair Employment Guidelines
P.A.31-126 of the Fair Employment Practices Act
Fair Employment Practice Guidelines

Civil Rights Commission Guidelines

Commission on Civil Rights - Guidelines
Legislative Approval of Guidelines

Employment Guidelines of Human Rights Commission

Commission on Human Rights — Guidelines
Section 17 of Fair Employment Practices Act

General Laws

Civil Rights Commission - Guidelines
Section 201-202 Michigan Civil Rights Act

Guidelines
Section 363.03 of Human Rights Act

Maternity Leave Act, Ch320 Laws 1975

Human Rights Commission
New York Supreme Court Decision

Bureau of Labor - Guidelines
Senate Bill 714 amending Fair Employment Act

Human Rights Commission - Guidelines
Human Relations Act

Human Relations Rule §20:03:09: 12

Maternity Leave Policy, Human Rights
Commission

Department of Labor and Human Relations,
Employment Policies Relating to Pregnancy
& Childbirth

Wisconsin Supreme Court

Effective Date:
6/4/75

l/1/79

5/31/72
10/1/73
11/3/71
10/9/72

1/1/74
1975

5/15/76

7/11/72
7/1/77

1 974

9/72
5/22/78

6/22/71
6/ 3/77

7/1/75

9/7 3
1976

6/7 1
10/4/77

12/25/71
5/ 17/75

1972
10/3/73

7/10/72

1975

1. The information in this table is not all-inclusive. Kentucky, Missouri, New Hampshire, and Oklahoma

implemented employer mandates, but did not document the effective date of coverage.

2. These states legally strengthened an earlier state employer mandate as a reaction to the US. Supreme
Court Ruling in Gilbert v. General Electric (1976), and as a result have two dates of compliance.

3. These states, as well as Kentucky, New Hampshire, and Oklahoma did not strengthen their mandates
following the 1976 Supreme Court Decision.

4. The information in this table is primarily from Commerce Clearing House (1978), NELP (1975), US.
Supreme Court (1975f76), and Weyland (1978).

BIBLIOGRAPHY

BIBLIOGRAPHY

American Bar Association, Section of Corporation, Banking and Business Law. 1974.
“Employment Discrimination.” The Business Lawyer (January): 577-614.

Commerce Clearing House. 1978. New 1978 Pregnancy Benefits and Discrimination Rules
with Explanation and State Survey. Chicago: Commerce Clearing House.

Gruber, Jonathan. 1992. “The Efficiency of a Group-Specific Mandated Benefit: Evidence
from Health Insurance Benefits for Maternity.” Working Paper No. 4157.
Cambridge, MA: National Bureau of Economic Research.

Gruber, Jonathan. 1994. “The Incidence of Mandated Maternity Benefits.” American
Economic Review (June): 622—41.

National Employment Law Project, Inc. 1975. Legal Developments in Sex Discrimination
Concerning Maternity Leaves and Benefits. New York: NELP.

Pearson, David W. 1973. “OFCC and EEOC Demands—Guidelines to Frustration.” The
Personnel Administrator 52 (November/December): 21-25.

Smith, Arthur B. 1980. “The Law and Equal Employment Opportunity: What’s Past Should
Not Be Prologue.” Industrial and Labor Relations Review (July): 493-505.

Summers, Lawrence H. 1989. “Some Simple Economics of Mandated Benefits.” American
Economic Review 79 (May): 177-83.

United States Supreme Court. 1975/76. “Martha V. Gilbert, et a]. versus General Electric
Company.” Law Reprints. Labor Series: The Supreme Court of the United States
Petition, Briefs and Opinion 9 (7). New York, Law Reprints.

United States Supreme Court. 1976/77. “Martha V. Gilbert, et a1. versus General Electric
Company.” Law Reprints. Labor Series: The Supreme Court of the United States
Petition, Briefs and Opinion 10(4). New York, Law Reprints.

Weyland, Ruth. 1978. “Pregnancy as a Disability.” In Proceedings of the Third ABA
National Institute on Equal Employment Opportunity Law. New Orleans: ABA
Press.

16

CHAPTER 2: THE EFFECTS OF MANDATED MATERNITY BENEFITS

AS Shown in chapter 1, the legislative history that preceded the passage of the PDA
complicates any attempt to estimate the labor market effects of mandated maternity benefits.
The state employer mandates extended health insurance to pregnant workers employed by
firms that provided coverage for other non-occupational temporary disabilities. This is
Similar to the coverage intended by the EEOC’S 1972 pregnancy guidelines and found in the
PDA. Additionally, the state insurance code amendments provided limited coverage to those
who purchased insurance within a state, regardless of labor force statusfi4 Since states that
adopted both maternity policies adopted the employer mandates before the insurance
amendments, this chapter focuses on estimating the labor market effects of enacting the state
employer mandates and the federal PDA.

To determine the impact of implementing a group-Specific employer mandate, I first
estimate the labor market effects that resulted from enacting the set of state employer
mandates requiring employers to cover the medical costs of pregnant workers. These
mandates comprise the “state study.” Comparing the workers in states with mandates to
those in states without the mandate, I find that the wages of married females of childbearing
age fell by about four percent after the mandate was enacted. This decline is consistent with

the estimated cost to employers of providing the benefit. However, net employment did not

 

3" All who purchased insurance were eligible for the pregnancy coverage provided by the insurance code
amendments. The coverage was limited (in most cases) to complications that arose from pregnancy conditions.
See chapter 1 for details on the legal aspects, and appendix A for a list of states.

17

18
Significantly change in response to the state mandates. This indicates that the value placed
on receiving the benefit by the group that benefits from the state employer mandate is at least
equal to the employer’s cost of providing it.

Next, I exploit the existence of the state mandates to run a “reverse experiment”
using the federal PDA. In this “federal study,” I estimate the effect of the federal PDA in
states without an employer mandate, while using the states with a maternity mandate to
control for other factors. In theory, the reverse experiment should return findings similar to
the state study. The enactment of the PDA, however, resulted in a Significant decrease in the
net labor input of covered workers, which is consistent both with these workers placing a low
valuation on the benefit and with significant barriers to wage adjustment. Regardless, this
translates to a decrease in overall employment, rather than a wage decline.

To reconcile the apparent conflict between the state study and the federal study
results, I suggest that the states in which the marginal benefit of mandate is higher than the
benefit’s marginal cost will enact that mandate. Thus, the value that society assigns the
benefit per firm is larger than the individual firm’s cost of implementing the mandate. States
adopt a matching level of state maternity legislation according to this valuation.
Accordingly, the state study results indicate that covered workers in those states with an
employer mandate placed a higher value the maternity legislation, implicitly accepting a
lower wage to receive its benefits. States that did not enact legislation before the PDA
perceived the marginal benefits of the state legislation to be below its marginal cost. The
disemployment effects found in the federal study are consistent with the case in which the
cost of implementing the PDA in non-mandating states outweighs the mandate’s benefits.

The hypothesis that worker valuation of a benefit differs systematically by State may

have important implications for policy makers. Due to the high value placed on the

19

mandates in the state study, the shifting of the mandate’s cost to worker’s wages Should be
viewed as a best case scenario. This result is not repeated in states that chose not to mandate,
as demonstrated by the federal study. If the PDA had been enacted before any state employer
mandates, the estimated total nationwide effect would fall somewhere between the findings
from the state study and that of the federal study—that is, a moderate decrease in the wage
accompanied by some disemployment effects. This would result despite workers in some
states placing a higher value on the mandate than workers in other states.

The remainder of this chapter is organized in the following manner. I review the
theories of mandated benefits that cover all workers and of group-specific mandated benefits
in Section 1. Section H presents an overview of the state and federal maternity policies that
are relevant to the empirical work in this chapter. In Section III, I introduce the empirical
framework used to estimate the effect of both the state and the federal maternity policies.
Section IV discusses empirical results found in similar studies. I test the effects of enacting
the state employer mandates, adopted during the early to mid-19708, in Section V. Turning
to the federal legislation, Section VI exploits the state employer mandates to estimate the
effect of the federal legislation enacted in 1978. Section VII reconsiders the composition of
the non-covered group used in the empirical framework. Potential explanations for the

differing state and federal study results are discussed in Section VIII. Section IX concludes.

1. Theoretical Review

In a 1989 article, Summers proposes a theoretical framework for evaluating the
effects of a benefit provided to all workers through a government mandate. He concludes
that workers who value the benefit implicitly accept a lower wage by remaining in the labor

force; non-workers who value the mandate enter the labor market to receive it. The resulting

20
increase in labor supply decreases the amount of deadweight loss that would have occurred
from public provision of the benefit funded by a payroll tax. This labor supply response,
derived from worker valuation of the benefit, leads to different labor market outcomes under
a government mandate and a payroll tax system.

If a worker values the mandated benefit at an amount equal to (or above) the cost of
the benefit, he or She is willing to accept a wage decline equivalent to the cost of the
mandate. Since the decreased wage offsets the full cost of the benefit, the level of
employment is not affected. Equivalent to that of a lump sum tax, this is the most efficient
outcome. As the level of worker valuation lessens, the amount of deadweight loss increases,
and the efficiency of the mandate decreases. Summers suggests that when workers do not
value the benefit, the deadweight loss from providing a benefit through a mandate is
equivalent to the deadweight loss from publicly providing the Same benefit through a payroll
tax. Thus, there is no Shift in labor supply, but there are wage and employment effects.

However, Gruber (1992, 1994a, 1994b) shows that Summers’ argument does not
necessarily apply if only a demographically identifiable group receives the mandate’s benefit.
Although the basic theory is the same, Gruber points out that market or legislative barriers
have a larger effect on group-specific mandates. Federal and state barriers that were
designed to limit wage or employment differences (e.g., Equal Pay Act, Title VH,
discrimination legislation) between demographically identifiable workers are more numerous
than barriers that affect all workers (e.g., minimum wage legislation). The group-specific
barriers may prevent the wages of the covered workers from adjusting downward to reflect
the true value that workers place on the benefit. Empirically, the wage and employment
effects in a market containing barriers to wage adjustment are consistent with a freely

adjusting market in which worker valuation is low or non-existent.

21

Another difficulty in using the group-specific mandate as tool for providing benefits
is that the demand for this type of mandate may not be readily apparent before the enactment
of the mandate. Deciding whether to enact any mandate is a summation of two components:
external and private demand.35 External demand reflects the preferences of all voters, while
private demand reveals the preferences of the group that will benefit from the proposed
legislation.36 In the case of a general mandate, the voter’s external demand for the mandate,
in theory, equals to the private demand of those who benefit. This is not necessarily the case
for a group-specific mandate. Since not all workers directly benefit from the adoption of a
group-specific mandate, the external demand for the benefit may differ from the private
demand. Prior to actually enacting a mandate, it is unclear whether a low external demand
reveals a market failure or a low valuation by the group that would be affected by the
mandate. This in an important distinction since a group-specific mandate is more efficient in
the former case.

When worker valuation is low, the efficiency of a group-specific mandate is below
that of an all-encompassing mandate. Gruber (1992) shows that when covered workers do
not value group-specific benefits, the deadweight loss may be larger than if the benefit had
been publicly provided. Employers, in this case, bear the firm-specific cost of providing the
benefit, based on the number of group-specific workers within that firm. Consequently, the
marginal cost of hiring a group-specific worker increases. This may create a hiring wedge

between the covered employees and all other workers, which leads to employer substitution

 

35 See Conway and Butler (1992) and Medoff (1988) for an application of this theory to the demand for abortion
legislation. Wyckoff (1984) presents a similar argument for the demand for primary and secondary education.

36 It should be noted that the majority of states did not vote on whether to adopt the maternity mandate, and as a
result external demand is not determined by the voters in that state. It is instead indirectly determined through
voters by the appointments that elected public officials make. Through this process, public demand in these
states determines whether the legislative, judicial and executive branches are disposed to enacting (and
enforcing) the maternity mandates. Regardless of how public demand determines the state policy, the value the
group receiving the benefit places on it following the enactment of the mandate reveals workers' private demand.

22
of other workers for the demographically identifiable group. With the decrease in wages,
many workers may choose to leave the labor market or to move to firms or industries not
covered by the mandate. This aspect may cause the group-Specific mandate to be less

efficient for the targeted group than public provision through a workplace-wide payroll tax.

II. Historical Motivation

Due to the unique legislative and judicial processes that led to the enactment of the
federal PDA, two opportunities arose for states to review the costs and benefits associated
with maternity policies prior to the enactment of the PDA. The 1972 enactment of the Equal
Employment Opportunity Commission’s (EEOC’S) pregnancy discrimination guidelines
provided the first opportunity. Two 1976 court decisions—the US. Supreme Court’s
decision in Gilbert v. General Electric Company (1976)37 and the New York Supreme
Court’s decision in Brooklyn Union Gas Co. v. NY. State Human Rights Appeals Board
(l976)38—afforded the second.

At each stage, a number of state Fair Employment Practice (FEP) agencies adopted
state guidelines similar to the EEOC’S guidelines. This may indicate that the voters in these
states placed a higher value on receiving the maternity benefits than those in states that did
not do so. In 1978, the federal PDA effectively legislated a minimum standard for pregnancy
conditions in employer-Sponsored health insurance plans. At this point, employers located in
states without a similar mandate were forced to update any existing company health

insurance plans, regardless of the value that either voters or workers placed on the benefit.

 

3’ 205 429 U.S. 125 (1976).
3“ 41 NY2d 84,13 EPD 1111381 (1976).

23
A. Pregnancy Discrimination Guidelines, 1972
The pregnancy discrimination guidelines issued by the EEOC in 1972 represent a
clear break from previous EEOC policy.39 While Title VII of the Civil Rights Act of 1965
originally focused on innate characteristics, the 1972 guidelines broadened the focus to
include “the addition of characteristics over which individuals clearly do exert control to the
bases for prohibited employment discrimination.”40 Congress, however, did not amend Title
VII to incorporate the 1972 EEOC guidelines at this time. Further, in its first decision under
the pregnancy guidelines, the EEOC allowed a large national employer to continue a policy
that discriminated against pregnant workers. The change in the EEOC’s original intent,
combined with the failure to amend Title VII and the lack of enforcement in a national case,
created confusion over whether the guidelines were recommendations or legally binding.
Despite the ambiguous state of the EEOC’s national maternity policy, a number of
state FEP agencies formally adopted the 197 2 pregnancy discrimination guidelines (see table
1 for a list of states and the corresponding dates of enactment).41 When a FEP agency
complement did not exist, the state courts tended to base their decisions on the 1972 EEOC
guidelines. Other States chose to wait until a congressional amendment to Title VII or the
US. Supreme Court reduced the uncertainty surrounding the legality of various maternity

mandates.

B. 1976 Court Decisions
Prior to the US. Supreme Court decision in Gilbert v. General Electric Company

(December 7, 1976), no national policy, other than the ambiguous 1972 EEOC guidelines,

 

39 For an in-depth discussion of the relevant matemity legislation, see chapter 1.
‘0 Smith 1980. 499.
4’ Although many references record the mandate’s adoption, none identify the level at which the state enforced it.

24

was in effect. However, all of the cases decided at the federal appellate level supported the
EEOC’S contention that the sex discrimination amendments to Title VII also prohibited
pregnancy discrimination.42 Reversing these rulings, the Supreme Court in Gilbert stated
that employers who provided an employer health care plan without maternity coverage did
not discriminate based on sex. The distinction, the Court ruled, was between pregnant and
non-pregnant workers. Since this was legal under Title VII, the 1972 EEOC guidelines
effectively became employer recommendations rather than requirements. This ended any
application of the guidelines to court cases at the State level.

Immediately following the Gilbert decision, the case of Brooklyn Union Gas Co. v.
N. Y. State Human Rights Appeals Board (1976), which questioned the validity of the state’s
pregnancy guidelines, came before the New York Supreme Court. In this case, the state
court ruled that the federal decision applied only to the EEOC’S 1972 guidelines, and not to
the individual State employer mandates. The opinion stated, “The determination of the

Supreme Court, while constructive, is not binding on our court.”43

Under this ruling, the
state employer mandate could still be subject to the sex discrimination defense if the state
chose to interpret it in that manner. This set a precedent for other states with employer
mandates already in place. From 1976-78, most states with a FEP guideline converted that
guideline into law (see appendix A for a list of states and the relevant dates). Other employer

mandate States chose to disregard the New York State Supreme Court’s ruling and did not

strengthen the state position against pregnancy discrimination. Regardless, the Gilbert and

 

42 These are: Berg v. Richmond Unified School District., 528 F.2d 1208 (9th Cir. 1975); Gilbert v. General
Electric Co., 519 F.3d 661 (4th Cir., 1975); Holthaus v. Compton and Sons, Inc., 514 F.2d 1024 (2nd Cir. 1975);
Wetzel v. Liberty Mutual Ins. Co., 511 F.2d 199 (3rd Cir., 1975); and Farkas v. South Western City School Dist.,
506 F.2d 1400 (6th Cir., 1974).

43 41 NY2d 84,13 EPD1I11381(1976).

25
Brooklyn decisions created the first clear division between states with and states without an

employer mandate.

C. Federal Pregnancy Discrimination Act, 1978

In 1978, the US. Congress enacted the federal PDA (Public Act 95-555). This
legislation amends the sex discrimination prohibitions in Title VII to include pregnancy
discrimination. Under the PDA, employers are required to treat pregnancy, childbirth, or
related conditions in a manner Similar to any other non-work temporary disability. The PDA
applies only to firms that provide temporary insurance coverage for other non-occupational
disabilities. For employers with benefit programs, the effective compliance date was April
30 1979. This allowed all employers who were not in compliance 180 days after the PDA’S
date of enactment, October 31, 1978, to update their benefit packages to include pregnancy
disabilities. By ending the uncertainty over the EEOC guidelines, the state employer
mandates, and the 1976 court cases, this law helped to standardize employer maternity

policies across all states.

III. DDD Estimation Techniques

To estimate the impact of the maternity mandates, I follow the example of Gruber
(1992, 1994a), who uses a differences-in-differences-in-differences (DDD) analysis. This
technique is typically used to analyze the labor market outcomes that result from a policy
change or a training program when data are not available from a controlled experiment. In
this study, the DDD estimator is used to evaluate the effect of the maternity mandates by

controlling for the status of an individual’s state of residence (with/without a mandate), the

26
year of observation (pre-mandate/post-mandate), and coverage status (benefit/not benefit
from the mandate’s adoption).

To understand the DDD technique fully, it is helpful to understand how the less
complicated differences-in-differences (DD) analysis would be used to estimate the effect of
the maternity mandatesi'4 Since the aim of this study is to estimate the labor market effects
of the maternity mandates on the group that benefited, only females who are age 20 to 40
would be included in the study. In this case, the DD analysis would include controls for
differences in the women’s state of residence and the year of observation.

The DD analysis for the two—state case in which one state adopts the maternity
mandate is the following:

th=01 +131Xj.+BZSj+BgY.+B4Sj*Y.+e,-.
where: j indexes states;
t indexes years;
Sj is equals 1 for an experimental state and 0 otherwise;
Wj. is the natural log of the hourly wage rate;
th is the set of individual demographic controls; and

Y,, is a fixed year effect equals 1 if post-policy, and 0 if pre-policy.

For females in both states, th controls for an individual’s demographic
characteristics (e.g., industry, education, age) in the typical wage regression. A female who
resides in a state affected by a maternity mandate (experimental state) belongs in the

experimental group

 

4" See Ashenfelter (1978) or Ashenfelter and Card (1985) for a more detailed explanation of the DD estimator.

27
(SJ-=1); while a female in a state that is unaffected by a maternity mandate due to her state
status (non-experimental state) is a member of the non-experimental group (SJ-=0). To
control for any changes that are not due to the adoption of the maternity mandate, the non-
experimental state is chosen to be demographically similar to the experimental state.45 The
difference between the labor market outcomes of those in the experimental state and those in
the non-experimental state group is [32.

The DD analysis also controls for differences across time in the two states. For the
group that benefits from the mandate, this difference equals the pre-mandate to post-mandate
change in earnings. The inclusion of the non-experimental group controls for the effect of
any national or regional trends in earnings not related to the maternity mandates ([33). In this
Simple case, the DD estimator compares the post-mandate versus pre-mandate difference
between those who have the ability to bear a child in the experimental state to the same
group in the non-experimental state. This is equivalent to the interaction of the
experimental-state post-mandate effect ([34). Although it seems obvious at this point, it is
important to note that both the females in the non-experimental state and the females in the
experimental state share an identifying characteristic that is necessary for the receipt of the
benefit (e.g., the ability to bear a child).

In the DDD technique, the DD estimator is expanded to include another dimension of
the labor market, a third level of controls, and a second demographically identifiable group
that is not covered by the maternity mandates. The third difference measures the change in a
labor market outcome (e.g., earnings) for a group of workers that is not affected by the policy

change (non-covered group), regardless of state of residence. This non-covered group lacks

 

‘5 In the DD literature, the experimental group corresponds to the treatment group, and the non-experimental
group corresponds to the comparison group.

28
the identifiable characteristic—in this case, the ability to bear a child—necessary to benefit
from the mandate.

Because only the members of the covered group are potentially affected by the
change in the maternity policy, the inclusion of this third difference is essential to correct the
estimation of the labor market effects of the maternity mandates.46 The non-covered group
then comprises any respondent who is not affected by the maternity mandate, regardless of
the state in which he/She resides (e.g., single males and those above the typical age range for
giving birth). Thus, the non-covered group is constructed to control for any policy changes
that are not related to the adoption of a maternity mandate and any legislation at the federal
level, which directly impacts on the females in this age group.

It is important to distinguish the purpose of the covered/non-covered group from that
of the experimental/non-experimental group. Since members of the non-covered group lack
an identifiable characteristic that makes them eligible to receive the mandate’s benefits,
including this group controls for changes in the labor market outcomes of the covered group
that would have occurred in the absence of the mandate. The members of the covered non-
experimental group—in this case, residents of the state without a maternity mandate who
have the ability to give birth—control for changes Specific to the group of women covered by
the maternity mandates. For DDD estimation to be unbiased, Shocks to the experimental
group can not be correlated with the policy change unless the non-covered group is affected

in the same manner as the covered-experimental group.

 

‘6 The group that may be affected by the change in policy may include the following: married females, age 2040;
married males, whose wives may be covered under the policy; and single females, age 20-40. However, I chose
to focus on the group of married females throughout this paper for two reasons. First, there is uneven spousal
coverage in employer mandates across states; some states expressly covered the wives of male employees while
others do not. In addition, the number of states with spousal coverage is too low to test separately. Second, the
incidence of pregnancy among Single females of childbearing age is lower than among married females of
childbearing age. This affects the probability that employers assign to the employee becoming pregnant and the
employer’s actual (insurance) cost of implementing the mandate.

29

In the Simple two-State case in which one state adopts the maternity mandate, the

corresponding DDD regression equation takes the following form:
WUF-a +81Xijt+BZSj+Bth+B4Sj*Y.+BSC,+B.5C,*S,-+B7C,*Y.+BgMANDATE,j, +8”-t
where: i, j, t, Sj, Wm, Xi“, and Yt are defined as before;
i indexes the covered group;
C, is a dummy equal to 1 for the covered group and 0 otherwise; and
MANDATEU-t iS equal to 1 if the respondent is in a covered group
in an experimental state following the mandate and 0 otherwise.

This regression equation identifies the group-Specific effects from the set of eight
base equations found in table 2, Equations I-IV and I’-IV’. For each of the four groups in
this study (covered experimental, non-covered experimental, covered non-experimental, non-
covered non-experimental), this first difference is from the post-mandate year to the pre-
mandate year. Variables that do not change across the years included in the DDD analysis
are controlled for using this first difference. These variables include the following:
individual characteristics, B1; fixed state effects, [32; fixed group effects, [35; and effects
specific to the group covered by the maternity mandate in the experimental State, [35.

As demonstrated in Equations V through VIII, the second difference addressed by the
DDD analysis is between the covered group and the non-covered group. In either the
covered or the non-covered group, any effect Specific to membership is isolated through this
second level of controls. These effects include the following: [33, the fixed year effect, and
[37, the trend effect Specific to the covered group. The third difference is over the State and it
controls for effects Specific to the post-mandate state, B4 in Equations IX and X.

Finally, the mandate coefficient, [is in Equation X1 in table 2, is now equal to one if

the respondent is in an experimental state in the covered group after the enactment of the

30
mandate, and zero otherwise. The relevant assumption of the DDD estimator is that all state-
specific Shocks not generated by the maternity mandates affect the non-covered group in the

same manner as the covered group.

IV. Previous Empirical Work

In a recent paper, Gruber (1992, 1994a) estimates the labor market costs associated
with the state insurance code amendments enacted in the mid- to late-1970s. These
amendments provided all purchasers of insurance with coverage for complications that result
from pregnancy, but do not cover the conditions associated with normal pregnancies.47
Consequently, any health insurance plan purchased by a firm in a state with an insurance
code amendment provided female employees (and all others who purchased insurance in the
state) with the limited pregnancy coverage.

In his study, Gruber uses the DDD technique to measure the post-amendment (versus
pre-amendment) labor market effects on married women (compared to Single males and
older people) in the amendment states (versus non-amendment states). His results Show that
the wages of the covered group significantly declined after the enactment of the state
insurance code amendments and the federal PDA. Further, the probability of employment
among workers in this group declined and the number of hours worked per week increased,
although neither effect is Significant. These findings are consistent with an increase in the
fixed cost of employing married females of childbearing age. Thus, Gruber interprets the full
shift of the mandate’s cost to the wages of this group as an indication that they fully value the

benefits provided by the mandate.

 

‘7 The insurance code amendments are individual-based not employment-based.

31

In practice, two distinct components of the state’s choice of legislation affect
employees: timing and content. States that enacted both maternity policies adopted the
employer mandate before the insurance code amendment (timing). Further, unlike the
insurance code amendments, the state employer mandates and the federal PDA cover normal
pregnancy along with complications that arise from pregnancy (content).48 Therefore, an
increase in an employer’s cost occurs in response to an insurance code amendment only if a
previous state employer mandate does not exist. The converse however, is not true; a prior
insurance code amendment only weakens the effect of a state employer mandate. Since
Gruber’S study classifies states only on their insurance code amendment status, his
comparison is flawed.

In chapter 3, I point out that neither Gruber’s state or his federal studies control for
the impact of the state employer mandates. Using the same data set, I first replicate his state
study and federal study results. I then consider only those states that enacted an employer
mandate before—or instead of—an insurance code amendment and states without either
policy. Estimating the effect of the State employer mandates using this subset allows me to
correct Gruber’s misclassification of states and re—estimate the effects of a maternity mandate
using his set of States. In chapter 3, I find evidence of a wage decline accompanied by some
disemployment effects. This chapter differs in that I focus on the full set of states—instead
of the subset used by Gruber—to determine the wage and employment effects of the

employer mandates on those who receive the maternity benefit.

 

4” California, which enacted an insurance code amendment before an employer mandate, is the exception. See
chapter 1 and appendix A for more information.

32

V. State Employer Mandates

This study tests the labor market effects of maternity legislation by focusing on the
state employer maternity mandates. Because most of the state mandates originated from the
EEOC’S 1972 guidelines, the State employer mandates were enacted earlier and were more
expansive than other state policies. As a result, the mandates are similar in content and
coverage to each other and to the 1978 federal PDA. Thus, using this set of state maternity

mandates allows for a consistent test of state mandates over a large number of states.

A. Main Data Source

This study uses annual data from the May supplement of the Current Population
Survey (CPS). The CPS is a national probability sample in which respondents are in the
survey for 16 consecutive months. Respondents are interviewed each of the first four and the
last four months of the cycle.49 In every supplement, the CPS collects basic demographic
information (e.g., race, sex, age, educational attainment) for each household member
reported by the respondent. The May supplement then asks the respondent about the current
labor market status of each household member above the age of 15. For each person
employed in the week before the survey, data are collected on the characteristics of his or her
main job (e.g., hours per week, hourly wage, industry).

Before 1977, the May CPS supplements suppressed the individual state identifier for
all but the largest states. The non-identified states were then aggregated into a number of
state groupings by region. Complicating matters, the state groupings and the number of

individually identified States changed twice between the survey’s initial year (1968) and

 

4" In this respect, the survey is longitudinal. Because respondents are tracked for only 16 months and the
maternity mandates were adopted over a period of about seven years, the longitudinal nature of this data set

33
1976. From 1968 to 1972, states were identified and grouped according to one formula; from
1973 to 1976, another formula was used. This incompatibility exists between both the

individual states and the state groupings. As a result, this study uses the data from Six May

CPS supplements (1973 to 1978).

B. Methodology

To test the effects of the state maternity legislation, I divide the eligible states into
two groups: those that implemented a state employer maternity mandate before the
enactment of the federal PDA in 1978 (the experimental states) and those that did not (the
non-experimental states). For inclusion in the study, the dates of compliance (month and

50 The non-

year) for each experimental state had to be listed in at least two references.
experimental states were not included in any reference.

Only experimental states that continued to enforce the mandate until the enactment of
the PDA are included in the analysis. The following analysis also excludes any state that
enacted an insurance code amendment but not an employer mandate (see appendix A for a
list of these states). This removes any potential bias that may occur due to the limited

’ Because of the Structure of the

maternity provisions of the insurance code amendments.5
CPS, the sample also excludes state groupings that contain states with different years of or

different types of enactment, including no enactment.

 

cannot be exploited for this study. Consequently, I ignore the longitudinal aspects of the data and treat each
respondent as a cross-sectional observation.

5" For further information, consult the following references: General Accounting Office (1977, 1979), National
Employment Law project, Inc. (1975), United States Supreme Court (1975/76, 1976/77), and Weyland (1978).

5 ' States that enacted both types of maternity policies are included since the majority of states implemented an
employer mandate before an insurance code amendment. The exception is California. However, this state is
excluded due to the state TDI system that began covering maternity conditions in 1974. In later years, California
enacted a state insurance code amendment, and finally an employer mandate. See appendix A for a list of
relevant state maternity policies.

34

Further, states for which either a month or a year of compliance were unavailable are
dropped from the sample. This is the case even when the state FEP agency is listed by two of
the referenced publications as adopting or enforcing an employer mandate prior to the PDA.
For example, I was unable to find the initial year of enactment for Indiana in any source,
although several sources list the state FEP agency as enforcing maternity guidelines before
the Gilbert case (see Weyland 1978, for example).

States that previously enforced the EEOC’S 1972 guidelines but which ended this
practice after the US. Supreme Court ruling in Gilbert (1977) are also not included in the

study.52

Finally, the five states that covered maternity through various payroll tax systems
are dropped from the sample (see chapter 1 and appendix A for details).

The sample also excludes workers who are not covered either under Title VII or the
EEOC’S protections as listed in Title VII (see appendix B for a list of these states and the
unprotected workers). The state FEP agency in a number of experimental states classifies
additional workers as being protected by the State FEP agency. These workers are also
excluded from the Study. Although federal workers are covered under Title VH in theory,
there is no mechanism to enforce it at this level. Consequently, the sample does not include
federal workers or government workers not clearly identifiable as state or local employees.

Using the DDD analysis presented in table 2, I estimate a model that includes the
experimental states and the non-experimental states identified in appendix B. The covered
group is comprised of married females age 20 to 40, who are in theory directly affected by

the maternity mandates. Workers who are not affected by the maternity mandates are

included in the non—covered group (Single males age 20-40 and people over the age of 40).

 

52 See, for example, Colorado.

35

AS described in the data section, all observations from 1973 to the date of the federal PDA
(1978) are used.53

Due to the structure of the May CPS and the dates of passage for the maternity
mandates, a vector of year dummies replaces the pre- and post- mandate variable in the DDD
analysis found in table 2. This iS true for both the first and the second level interactions. The
MANDATEU-l interaction is equal to one for the group of married females, age 20-40, in the
years that an experimental state has a maternity mandate in place. It is zero for all other

states, all other years, and all other observations.

C. Results

The first column in table 3 presents the results of a wage specification that includes
the demographic characteristics; the grouped state, year, and covered group interactions; and
the mandate variable. Overall, the coefficients on the demographic variables are of the
expected Sign and magnitude. More-educated workers earn more, as do workers with more
years of experience (although at a decreasing rate) and those working in an establishment
covered by a union contract. AS expected, married workers earn more than unmarried
workers do. The wages of non-white workers are below those of their white counterparts;
female worker’s wages fall below the wages earned by male workers. In addition, those in
the covered group earn almost 8 percent more than those in the non-covered group. The
coefficient on the mandate variable is positive and significant, indicating an almost 5 percent
increase in wages following the mandate. This implies that workers for whom the benefit

was adopted place a negative value on the mandate.

 

5" Due to the restrictions on the number of years included, states with early dates of enactment (before May
1973) identify differences off other states and the covered/non-covered group.

36

Although the finding of an increased wage for those affected by the mandate is
counterintuitive, the positive coefficient may not reflect the impact of the state mandates, but
instead may reflect the individual state effects not controlled for in the above specification.
Based on the two-state model presented in table 2, this specification measures the
experimental/non-experimental effect through a grouped-state variable that equals one if a
state enacted a mandate at any time before 1978, and zero otherwise. Thus, the coefficient
on MANDATEiJ-t may be measuring individual state-specific effects (e.g., industrial shocks)
not controlled for in the grouped STATEj variable.

The specification in the second column of table 3 controls for state-specific effects by
including a vector of individual state variables that equals 1 if the respondent is in a
particular state and 0 otherwise. These independent state variables replace the
experimental/non-experimental (STATEj) designation in the base regression found in column
2. Including State—Specific variables should control for differences that vary by state but that
are not related to the state maternity policy. Examples include state anti-discrimination
policies or state attitudes towards the female workers.

Using this Specification, the effects of the demographic variables are largely
unchanged. After controlling for individual state effects, the estimated effect of the state
maternity mandates is a 3.5 percent decline in the wages of the covered group. This
translates to a decrease of about $291 per year for a full-time female worker covered by the
mandate, well within the cost range for implementing a maternity mandate ($277/year to

$360/year) estimated by Gruber.54

 

5" I define full-time in this example to be a respondent who works 40 hours per week for 52 weeks. Gruber’s
estimates are for females who hold family coverage and who are between the ages of 20 and 39. All amounts are
in 1978 dollars. See Gruber (1992) for details on how this estimate was calculated.

37
In column three, I include the individual state variables in the second level

. . 5
covered*state luteractlon.5

The decline in the wages of the covered group increases to 4
percent and is significant. Thus, for each full-time worker covered by the mandate, this
equals about a $335-per—year decrease in pay. Again, this is well within Gruber’s cost
estimate.

Further, under this specification overall net labor input falls. The number of hours
worked per week declines by a Significant 3.6 percent, which may indicate that more part-
time work is being undertaken by married females of childbearing age in the affected states.
Although insignificant, the coefficient on the probability of employment is negative.

Overall, the wage and the net employment results are consistent with the members of the

covered group assigning a positive value to the benefit.

D. Group-Specific Efl'ects

Within this covered group, however, the effect of the mandate may vary across
groups of workers. The wages of workers who face existing market barriers, such as anti-
discrimination legislation, for example, may not legally adjust downward to reflect the value
placed on the benefit by females in the covered group.56 On the other hand, women whose
employers did not provide health insurance would not be affected by the legislation. To
determine how the maternity mandates impacted on different types of workers, I re-estimate
the full sample specification across more/less skilled workers and on those in the workforce

before the US. Supreme Court decision in 1976.

 

55 An f-test rejected the inclusion of individual states in the state*year interaction.

56 This situation may also occur when minimum wage barriers are present. However, it is unlikely these barriers
will affect those receiving health insurance benefits since the probability of receiving these benefits increases
with the wage. See Averett and Hotchkiss (1995) for more details.

38
First, I consider the labor market effects of the state maternity mandates on the
workers with different probabilities of receiving medical insurance. Since the May CPS does
not ask respondents about employer health insurance coverage in the 1970S, I estimate the
labor market effects of the maternity mandate on workers holding jobs that require different
levels of skill. As a proxy, I use the level of education attained by the respondent at the time
of the survey. More-educated workers are, in theory, more likely to hold jobs that require

more skills and offer health insurance benefits.57

As a result, there is a higher probability
that the maternity mandates will have an impact on this group. The more-educated group
consists of those who report having completed 12 or more years of regular schooling; the
less-educated group contains workers who report completing less than 12 years of regular
schooling.

These results are in the first two rows of table 4. In the first row, the results for the
more-educated group are reported. The coefficient on the mandate variable indicates a wage
decline of 4.6 percent after the adoption of the maternity mandates. The number of hours per
week worked falls by 4.9 percent. Both effects are Significant, suggesting that
disemployment effects accompanied the wage decrease for more-educated workers. AS in
the larger specification, the probability of employment declines, but is not significant. In the
second row of table 4, the results for less-educated workers are reported. The employment
and wage effects for these workers are insignificant, although the number of hours worked
per week is positive. Thus, it is likely that a large portion of this group did not receive health

insurance in connection with employment and was not affected by the state’s maternity

mandate.

 

’7 Averett and Hotchkiss (1995) find that for each year of education, workers have a 5.6 percent higher
probability of receiving medical insurance.

39

The U.S. Supreme Court’s ruling in Gilbert (1976) may also affect the estimates.
Under this ruling, the lower courts could no longer interpret the 1972 guidelines under Title
VII’S sex discrimination amendment. The Brooklyn (1976) decision reversed this ruling as it
applied to a separate employer mandate enacted by the state. Under this decision, States
could continue to enforce an existing state employer mandate. Although a number of states
chose to enforce their state employer mandates immediately after Gilbert, others allowed a
Significant lag to occur before Strengthening an existing state guideline.58 Still others chose
not to enforce guidelines previously in effect.59 Due to the Court’s decision, the EEOC’S
pregnancy guidelines and the State’s maternity mandates do not apply to employers during
this transitional period. Therefore, the uncertainty concerning the state’s position may result
in a decrease in the amount of employer-offered maternity insurance. If this is case, the
effects from the maternity mandate will be understated.

To determine whether the Gilbert decision and subsequent state or congressional
actions influenced the results, I test a specification that includes only the pre-Gilbert years,
1973-76. The third row of table 4 presents these results. The wage and employment
coefficients are in the same direction as the full year specification. However, only the wage
decline (over 5 percent) in the pre—Gilbert specification is Significant. Using this coefficient,
I estimate that the wages of a full-time female worker (as defined for the full sample)
declined by over $430 per year.60 The decline in the number of hours worked per week is
smaller than that found in the full year specification, indicating that State employer mandates

had a larger impact on the part- or full-time status of workers after the 1976 court decisions.

 

5” No documentation of enforcement or lack of enforcement during this period is available.

59 These include the following: Colorado, Kentucky, Oklahoma, and New Hampshire.

6" This estimate is above Gruber’s estimated range, however, the cost was estimated using a state that was the
midpoint of the locational cost distribution. The range for other states would presumably be higher.

40
VI. The Federal PDA
Due to the confusion created by the 1972 EEOC guidelines, various state maternity
policies, and the 1976 court decisions, the extent to which individual employers adopted the
maternity policies is unclear. It is also impossible to judge the amount of information
workers possessed about their rights concerning pregnancy conditions since the level of

61 .
Because of these lssues, the federal

enforcement from various agencies was inconsistent.
PDA presents a better opportunity to estimate the effects of a mandate than the state
employer mandates. The congressionally approved PDA created a nationally enforceable
standard that applied to all employers under the jurisdiction of Title VII.

By exploiting similarities between the state employer mandates and the federal PDA,
the federal legislation can be used as a ‘reverse experiment.’ The purpose of this federal
study ‘reverse experiment’ is to confirm the direction and magnitude of the state study
results. In this study, workers who were already covered by a state maternity mandate
provide a natural control group for those affected by the federal legislation. The states with
employer mandates are now the non-experimental states and the states that did not enact any
maternity policies prior to the PDA are now the experimental states. The underlying.

assumption is that the value that the covered group places on the maternity benefits does not

systematically differ by experimental/non-experimental state status.

A. Main Data Source
After 1979, the May CPS supplements gather earnings data only for those
respondents in the out-going rotation group (e.g., months 4 and 16). Due to the small sample

Sizes present in the May CPS supplements, I use the March CPS supplements, which include

 

6' See chapter 1 for more details.

41
earnings information for all respondents, to test the wage and employment effects of the
federal legislation. These supplements differ from the May supplements in that respondents
are interviewed about their labor market activity and earnings during the previous calendar
year. All industry, occupation, hours per week worked, and weeks worked questions refer to
the longest held job during that period; the earnings variable contains data on all wage or
salary earnings received for work performed as an employee during the previoils calendar
year.62 Finally, the March CPS differs from the May supplement in that it does not ask the

respondent about his or her union status.

B. Methodology

Aside from the use of the March CPS supplement, the federal study differs from the
state study in two respects: the composition of the experimental and the non-experimental
states and the pre- and post-policy years. I test the effects of the federal PDA using states
with an employer mandate or without any type of mandate policy, such as an insurance code
amendment. States that adopt both maternity policies are included in the study since the
content of the maternity mandates dominates that of the insurance code amendments. In the
federal study, states that enacted and continued to enforce an employer mandate before
January 1, 1977 are in the non-experimental group. The time frame for the federal study is
also different, covering the years from 1978 to 1982.

As in the state study, states that instituted a maternity policy before the Supreme
Court’s ruling in Gilbert (1976) but did not strengthen that policy by October 31, 1978, are

dropped from the sample (see appendix C for a list of states and industries). The federal

 

62 Included in the variable are earnings from wages, salary, armed forces pay, commissions, tips, piece rate pay,
and cash bonuses. If a respondent reports a wage range, the midpoint is used (e.g., respondent-reported earnings

42
sample includes all workers covered by Title VII and the state employer mandates but not
federal government employees. While Title VII theoretically covers federal workers, in
practice it was not enforceable at that time; all self-employed workers are also dropped from
the study. Finally, because earnings data are from all jobs, I drop any respondent who reports

more than one job during the previous calendar year.

C. Results

As in the regression framework used in the state study, states are grouped by
experimental/non-experimental Status in the MANDATED, interaction. The March CPS asks
respondents about the characteristics of the longest job worked in the previous year. As a
result, I am able to calculate the impact of the federal PDA on the number of weeks that the
respondent works per year, as well as wage, hours, and employment effects. Table 5 reports
the results from a specification similar to the one used in the state Study.63

In contrast to the findings in the state study, both the wages and employment of the
covered group affected by the PDA decline, but only the employment effects are significant.
Married females in the post-PDA experimental states work nearly four percent fewer hours
(column 2) than those in the non-experimental states. In shown in column 3, the number of
weeks worked by this group declines by 7 percent. This indicates that those covered by the
mandate are engaging in more part-year and part-time work, which may have a lower
probability of providing health insurance. Finally, the probability of employment is negative
and marginally significant. Those who are covered by the PDA were almost 3 percentage

points less likely to be employed after this legislation was enacted.

 

of between $10,000 and $15,000 are reported as $12,500).
63 The results from an f-test indicate the inclusion of individual state effects in both second level interactions
(state*covered and state*year).

43
These results are consistent with workers in the experimental states placing a very
low—or no—value on receiving the maternity benefits provided under the PDA.64 Although
some firms in the experimental states may have adopted similar health insurance in advance

of the federal act, the strong decrease in net employment implies that the majority of firms

did not.

D. Group-Specific Effects

Next, I examine the effects of the federal legislation across more/less educated
workers. Table 6 presents these results. Similar to the state Study, more educated workers
are affected by the maternity mandate. The number of hours worked per week falls by
almost 4 percent, and is Significant. Less Significant is the decrease in the number of weeks
worked per year for this group. All of the other estimates are insignificant. For the less-
educated group, the PDA does not impact on their wages or the number of hours worked per
week. A large decrease, however, occurs in the number of weeks worked per year, 22.3
percent. In addition, the probability that less-educated workers were employed following the
PDA fell by almost 5 percentage points. Taken together, these effects are consistent with a
Shift to lower hours or part-time employment for more educated workers and part-year work

for less educated workers. This may indicate that covered workers, regardless of their

 

M As an example, this response may occur due to the spousal coverage clauses found in some state mandates and
the federal PDA. These clauses stipulate that employers provide maternity benefits to spouses covered by the
health insurance plan. However, workers with indirect access to the benefit, such as through a spouse, may alter
labor market behavior accordingly. This does not occur if the spouse’s employer provides insurance coverage
below valuation. If the opposite occurs—coverage is equal to or greater than her valuation—the spouse may not
value insurance gained through ‘own’ employment and may view the provision of the mandate in her workplace
as a tax. Therefore, some spouses may find it optimal to leave the labor force and devote more time to home
production, or move to a job, industry, or sector not covered by the maternity legislation. Unfortunately, I cannot
test this hypothesis since the states that chose to include the spousal stipulation either do not appear in the state
experiment, or are grouped with non-spousal stipulation states. In addition, there is no way to gauge whether
states that do not explicitly state the spousal stipulation interpret the guidelines to include spousal coverage.

44
educational background, are increasingly in positions that are less likely to provide fringe

benefits (e.g., low hours, part year), including health insurance after the PDA.

VII. Covered/Non-Covered Group

The Specifications in both the state study and the federal Study rely on three
assumptions. The first is that the state provisions correctly identify each state as either an
experimental state or a non-experimental state. Second, it is assumed that the dates of
enactment for the various state maternity provisions are correct.65 The third assumption is
that the composition of the non-covered group does not bias the results for the non-covered
group.

However, the third assumption may conflict with the theory of mandated benefits,
which predicts that employers may respond to group-specific legislation by substituting
relatively less-expensive non-covered workers for covered workers. If employers substitute
between the covered group and the non-covered group, then one of the identifying
assumptions of the DDD regression framework—that the non-covered group controls for all
labor market Shocks correlated with, but not resulting from the group-specific mandate—is
violated. This may result in MANDATEij-t measuring the total labor market effect of the
legislation, rather than the effect on the group married females. Thus, the inclusion of the
full sample non-covered group would overstate the mandate’s true effect on the covered

group.66 Similarly, the regression estimates will be biased if employers substitute away from

 

‘55 Because the state guidelines were superseded by the federal PDA, a number of states no longer had records of
the guidelines or the exact dates of enactment. For this reason, I only include the states for which the state status
and the dates of enactment are supported by at least two references. For more information, see footnote 13, the
material contained in chapter 1, and the list of states in appendix B.

6" The age-eamings profile of females is not influenced by their age composition, therefore, female workers are
closer substitutes. For more detail, see Freeman (1979). For evidence on the elasticity of substitution between
women and youths (ages 14—24) sec Grant and Hamerrnesh (1980).

45
all types of labor due to the maternity mandates. The wages of all workers fall under this
scenario. In this case, the legislation’s effect on the covered group’s wage would be
understated.

A separate consideration is whether the composition of the group-specific mandate
impacts negatively on either the non-covered group or the non-experimental group. AS an
example, a portion of the non-covered experimental group in the federal study may be
affected by the enactment of the 1978 Age Discrimination in Employment Act (ADEA).
This federal mandate increased the mandatory retirement age to 70. Similar to the maternity
mandates, a number of States enacted laws Similar in content to the ADEA before the federal
mandate. The ADEA, in theory, did not affect workers in these states. As a result of the
state legislation, the extent that the ADEA affected older workers’ behavior depended on
which states were directly affected and how large the relative proportion of older workers
was per affected state.67

If the impact of the ADEA was uniform on workers in all states, the inclusion of the
covered/non-covered groups and the coefficients for years of experience would adequately
control for the affect of the ADEA in the maternity regression. The same result may occur if
the location of states that enacted the state age mandates was not uniform. However, it was
primarily the non—experimental states in the federal study that enacted the state age

legislation during the 197OS—the same period in which these states enacted maternity

 

“7 Neumark and Stock (1997) test the effects of the state and federal laws using the decentennial census. They
find that the age discrimination laws led to steeper age-earnings profiles and a significant increase in the
employment of older workers.

46
mandates.‘58 Thus, inclusion of older workers in the state and federal maternity studies may
affect the results.

In view of the covered/non-covered group problem, I again test the legislative effects
of the state employer mandates, but use a smaller sample containing only married females of
childbearing age. Respondents who report working in industries excluded from Title VII
protections or from any state’s employer mandate now comprise the non-covered group (see
appendix B and appendix C for the eligible states and the affected industries).69 Inclusion of
this non-covered group allows for a test of the effect of the employer maternity mandates on
the group of married females, while avoiding the potential problems presented by the full
sample non-covered group. Employer substitution of the non-covered workers for the
covered workers is no longer an issue. AS with the full sample, the small sample non-
covered group controls for region-specific Shocks that affect both the non-covered group and
the covered group.

Using the law-based non-covered group is not without problems. The industries used
to form the non-covered group are, for the most part, outside the scope of Title VII.
Therefore, workers in these industries may be affected differently by national or regional
shocks than the workers protected by the EEOC. Further, industry-specific Shocks may have
an impact on certain subgroups. The regional grouping of states in both the state study and
the federal study will only exacerbate this problem. Still, the results using this group are

valuable as they provide an estimate of the mandate’s effect for comparison to the full

 

68 Five of the seven non—experimental states implemented state age-discrimination legislation that banned
discrimination of employees age 65-70 before 1978 (the year of the federal law). These include the following:
Alaska, Illinois, Iowa, Michigan, and Minnesota None of the experimental states did the same.

69 I also test the specification that includes only workers exempted under Title VII. The results are unchanged
from those found in table 7. These results are found in appendix D.

47
sample results. More important to policy makers is the extent to which these results are
similar to those of the larger sample.

Another important consideration is that it is possible that the covered sector females
may ‘spill over’ into the non-covered sector as a result of these mandates. In particular, this
occurs if barriers to wage adjustment prevent the wage in the covered sector from adjusting
to reflect the value placed on receiving the maternity benefit. The spillover effect may also
occur if workers do not value the mandate. In either case, the supply of labor increases in the
non-covered sector, causing a decrease in the non-covered sector wage. The composition of
the covered group then would overstate the effect of the mandate. In addition, net
employment in the experimental states would decline as well, since the non-covered sector
cannot absorb the influx of covered-group workers. It is worth noting that the spillover effect
occurs only to the extent that the cost of the mandate is not shifted to the wages of the

covered group. In the case of a full shift of the wage, there is no spillover effect.

A. State Study Results

Using the DDD analysis presented in table 2, I estimate a model that includes the
experimental and the non-experimental states identified in appendix B. The covered group is
comprised of married females age 20 to 40 in industries covered by Title VII or the state
employer mandates. Workers in industries not affected by the maternity mandates compose
the non-covered group. As described in the data section, all observations from 1973 to the

date of the federal PDA (1978) are used.70

 

7" Due to the restrictions on the number of years included, states with early dates of enactment (before May

1973) identify differences off other states and the covered/non-covered group.

48

Table 7 reports the results from the state study using the small sample non-covered
group.71 AS Shown in column 1, there is decline in the covered group’s wages (3.7 percent)
similar to the estimates found in the full-sample study. The disemployment effects in this
sample, however, are not significant, but are in the same direction as those found in the full-
sample specification. Thus, employers did not appear to substitute the full sample non-
covered workers for covered workers following the enactment of the mandate. In addition,
the small-sample full shifting of the wage implies that the Spillover of covered workers into
the non-covered sector did not occur.

Table 8 presents the results from re-estimating the sub-group specifications from
table 4 using the smaller sample. In general, these results are similar to those found in the
full-sample specification. The wages of more-educated workers, found in the first row,
declined by over 4 percent in the experimental states following the enactment of the
mandate. The decrease in the number of hours (4.2 percent) is smaller and less significant
than that found for the more-educated workers in the full-sample results. Further, the
estimates from the second row specification, which examines the effects of the mandates on
the less-educated workers, are insignificant.

In the third row of table 8, I re-estimate the pre-Gilbert specification using the small-
sample non-covered group. The 5.2 percent decline in the covered group’s wage is similar to
that found in the full-sample Specification. This translates to a decline of nearly $440 per

year in the wages of those who benefit from the mandate. The employment results (e.g.,

 

7' Although the previous results are consistent with the predicted effects of group-specific mandates, they may
reflect the effect of the experimental state prohibition of sex discrimination. Most state FEP divisions adopted
the pregnancy guidelines and the sex discrimination mandates concurrently. Although Title VII applies to all
workers, it is possible that the lack of state and local enforcement and the EEOC’s backlog of cases increased
non-compliance with the sex discrimination amendment in the non-experimental states.

49
decreased hours worked per week, increased probability of employment) remain
insignificant.

As an alternative specification, I drop the non-covered group and resort to a DD
estimation. These results, found in table 9, base the mandate’s effect on a state by year
interaction. Therefore, this specification does not control for national or regional effects.
The results from the DD specification are similar in direction and magnitude to those found
in the small sample specifications: the wages of workers affected by the mandate decrease

while net employment remains the same.

B. Federal Study Results

Table 10 reports the estimated wage and employment effects of the federal legislation
in the experimental states small-sample specification. Again, female workers under Title
VII’s protections comprise the covered group. Similar to the full sample results, there is a
significant decline in net labor input due to the federal PDA. The estimated post-PDA
probability of employment among covered workers is nearly the same as in the full sample
specification. The number of yearly weeks worked, found in the third column of table 10,
decreases by just over 13 percent, almost twice the decline found in the full sample results.
AS shown in the second column, the number of hours worked per week falls by 7 percent.
The decline in the wage is negative and larger than in the full sample specification, but it is
still insignificant.

In table 11, the estimated labor market effects of the PDA on the sub-groups are
presented. The first row presents the estimates for more educated workers. As Shown in
column 3, those affected by the PDA work about 13 percent fewer weeks per year after the

mandate is enacted. The 6 percent decrease in the number of hours worked per week by the

50
more educated workers is less significant than that found in the full-sample results. The
specification including less-educated workers is found in the second row of table 11. The
estimates in column 4 indicate that these workers experience a decrease in the probability of
employment of just over 8 percentage points. However, the mandate does not seem to affect
the number of weeks worked per year, the number of hours worked per week, or the wage of
this group.

The first row in table 12 presents the results from the DD specification that does not
include the non—covered group. The insignificant decline in the wage is similar to that found
in both the small- and large-sample specifications. Likewise, those covered by the mandate
experience a Significant decrease in the number of weeks worked per year. The decline in
the number of hours worked per week is much smaller and less significant than the full- or
small-sample specifications. The second and third rows of table 12 present the results for
more- and less-educated workers. Both more and less educated workers experience a decline
in the number of weeks worked per year after the PDA is enacted (7.5 percent and 18.9
percent, respectively). Like the previous specifications, there is a corresponding drop in
employment of about 8 percentage points for the less educated workers.

Although differences do exist, the results from the smaller law-based non-covered
group are similar to those found using the larger non-covered group. The state study results
are nearly identical. Overall these results support the theory that the covered workers place a
positive value on receiving the benefits afforded by the state maternity mandates.

However, the federal study results of the small-sample specification differ from those
found in the full-sample specification. What does not change is the effect of the mandate on
the wage. Although negative, the mandate coefficient is never significant in the wage

equation. Disemployment effects are present in both specifications: the probability of

51
employment and the number of weeks worked per year are lower. Although the effects of
the federal Study are not clear, it seems that females of childbearing age in states that did not
enact a state employer mandate either did not place a high value on receiving the maternity

benefits afforded by the PDA or faced significant barriers to wage adjustment.

VIII. Disparity between State and Federal Study Results

Contrary to the hypothesis for the ‘reverse experiment,’ the labor market effects of
the state employer mandates and the federal PDA differ considerably. The state and the
federal study results are plausible according to the theory of mandated benefits. The wage
and employment differences are surprising given the nearly identical wording of the state
mandates and PDA; both provide female employees with health insurance coverage for
normal pregnancies and complications resulting from pregnancy. To determine the cause of
the disparity between the state and the federal studies, I consider state-specific differences

between the experimental states and the non-experimental states used in the federal study.

In theory, the mandate’s marginal costs and marginal benefits should not differ
systematically between a state with—and a state without—prior legislation. The federal
PDA in this scenario would serve either to quicken the process of adoption in the remaining
states, or to correct for market failures originating in those states’ governing bodies. In this
Situation, the effects of the federal study would confirm the direction and the magnitude of
the state study results. Given the inconsistent findings between the two studies, it seems
clear that the federal PDA did not merely expedite the mandating process. One possible

explanation for these results is that the marginal cost and/or the marginal benefits of the

52

maternity legislation vary by State and, based on these criteria, States actively choose whether
to adopt state maternity policies.72

The actions of states during this period are consistent with this hypothesis. Initially,
a number of States chose to adopt or enforce the EEOC’s 1972 pregnancy guidelines,
Si gnaling that the public demand for the maternity benefit was greater than the marginal cost
of enacting similar state legislation.73 A positive private demand is evident in the full
shifting of the cost of the mandate to the covered group’s wages. Following this, the unique
combination of the Gilbert and Brooklyn decisions forced states that had not formally
adopted a state maternity mandate to choose whether to adopt similar legislation. States in
which external preferences determined that the marginal benefit was greater than the
marginal cost enforced the appropriate level of maternity coverage (e.g., a state employer
mandate, a state insurance code amendment).74 As a result of these divisions and the
requirements of the DDD estimator, the federal study highlights the PDA’s effect on two
types of state. These include the following: states with the highest marginal cost/lowest
marginal benefit (non-mandate states) and states with the lowest marginal cost/highest
marginal benefit (mandate states). Among the non-mandate states, the lack of adoption
indicates a lack of external demand. The large disemployment effects that followed the
federal legislation confirm that low external demand correctly signaled that private demand

was less than full.

 

72 See Rasmusen (1992) for an in-depth explanation of the differences between states that initially enact
legislation and States that do not.

73 States in which the differential between the marginal benefit and the marginal cost was lower may strongly
recommended the mandate instead of adopting it Thus, there may be more disemployment effects that in states
that adopted the mandate. Unfortunately, there are not enough states in the 1970 CPS to test this theory.

7" Some states adopted both an employer mandate and an insurance code amendment. By adopting an employer
mandate first, a state effectively provided the minimum level of maternity coverage for non-employed females.
Female employees had a higher level of coverage due to the mandate. See chapter 1 for more information.

53

Casual evidence supports this hypothesis. Married females of childbearing age
comprised a larger percentage of the population of the federal experimental states than in the
federal non-experimental states (32.7 percent and 30.5 percent, respectively).75 However, the
percent of the employed population in 1979 that was comprised of married females was
lower in the states affected by the PDA (31.5 percent) than in the states with employer
mandates (33.0 percent). Further, 13.3 percent of the women in the experimental states had a
child age three or younger living in the household at the time of the survey. For comparison,
11.6 percent of the population in the non-experimental states are women with children age
three or younger. In addition, a larger percentage of females with children age three or
younger were employed during the previous survey year in the non-experimental States (67.2
percent) than in the experimental states (65.2 percent). Thus, there were more females of
childbearing age in the federal experimental states than the in federal non-experimental
states, although fewer of these females were employed.

Additional evidence of the value that the voters place on legislation favorable to
females can be found in the propensity of states to adopt Similar legislation. Table 13 lists
the status of the federal study states with regard to four separate policies that were favorable
to females during the 19703 and the 19803.76 In the first column, I define each state’s status
with regard to an active FEP agency. The state status is important since a FEP agency had to
enforce all EEOC regulations, including sex discrimination guidelines, to retain its active
status. After 1976, these agencies also tended to quicken the processing of discrimination

claims at the State level. Column 2 lists whether a state adopted a separate Equal Pay act

 

75 The statistics in this paragraph are weighted calculations from the March 1981 CPS.

7" Although I focus on the propensity of a state to enact female-specific policies, it may be the case that these
states should be examined according to proebusiness versus pro-worker policies. Of the nine experimental states
listed in table 13, for example, all but three are right-to-work states. Only three of the 15 non-experimental states
had enacted right-to-work legislation as of 1981.

54
above that found in the state FEP law.77 In the third column, states that ratified the Equal
Rights Amendment (ERA) petition before 1978 are distinguished. The fourth column lists
states that adopted the ERA into their state constitution.

Of the nine federal experimental (non—mandate) states, only West Virginia supported
three of the four female-favorable policies. Four of the remaining states adopted at least one
piece of female-favorable legislation; the remaining four did not enact any. Most of the
federal non-experimental (mandate) states adopted at least three of the four of the pieces of
female-favorable legislation. Preferring not to adopt the ERA into their state constitution,
only three non—experimental states (Iowa, Missouri, and Wisconsin) legislated less than three
pieces of legislation. Thus, the majority of states that adopted the maternity guidelines
before the passage of the federal legislation tended to implement other policies that were
favorable to females during the 19703. This supports the hypothesis that their constituencies
valued similar state policies, such as the state maternity mandates, that were favorable to

females.

IX. Conclusions

This chapter evaluates the labor market impact of state and federal mandates. In
particular, the focus is on group-specific mandates that required employers to cover
pregnancy conditions when providing health insurance benefits. Over the period examined,
three major events had an impact on employees. The 1972 EEOC’S pregnancy guidelines

changed the focus of the EEOC’S policy to include acquired worker characteristics; a limited

 

77 The state equal pay statutes are in effect following the 1963 federal legislation. The following argument
describes why these state statutes are favorable to women: “While a plaintiff in any state may point to federal
laws as well to argue that sex discrimination is contrary to general ‘public policy,’ the existence of a state statute
that explicitly forbids such discrimination further buttresses public policy arguments relevant to State tort and
contract claims...” (Wald, 1982). Thus a number of avenues are created to redress employer discrimination.

55

number of state FEP agencies adopted these EEOC guidelines. Second, the diverging
opinions presented by the Gilbert (1976) decision, which redefined the guidelines to be
‘recommendations,’ and by the Brooklyn (1976) decision, which upheld the individual state
FEP guidelines, created a clear division between the states with and states without employer
mandates. Third, the 1978 federal PDA effectively legislated a minimum standard that all
employer-sponsored health insurance plans must treat pregnancy in the same manner as other
temporary non-occupational disabilities.

Using these legislative and judicial divisions, I test the labor market effects of the
maternity mandates using the state employer mandates and the federal PDA. At the State
level, I find that the enactment of employer mandates resulted in a substantial shifting of the
cost of the benefit to the covered group’s wages. This indicates that those affected by the
state legislation placed a high value on receiving the benefit relative to the employer’s cost of
providing the additional health insurance. The federal legislation, on the other hand,
Significantly affected the employment of females who stood to benefit. The disemployment
effects indicate that those affected by the PDA may not have fully valued the mandate.78

Although neither of these results contradicts the theory of group Specific mandates,
the state study and the federal study effects are disturbing from a policy perspective,
especially given the similarities found in the legislation. The evidence found in the state
study reflects a seemingly efficient choice by the state governments; the federal study results

indicate the opposite. Which results should legislators believe?

 

7" The state study results are similar to Gruber’s (1992, 1994a, 1994b); however, the federal study results are
dissimilar.

56

The disparity between the state study and the federal study results may be explained
by the propensity of the states to adopt similar legislation. Because states implement a
mandate if the perceived social benefit is higher than the benefit’s marginal cost, it is
reasonable that each state adopts the ‘correct’ level of state maternity legislation (an
employer mandate, an insurance code amendment, or neither) prior to the PDA. Covered
workers in states that value the mandate implicitly accept a lower wage to receive the
mandate’s benefits. Therefore, the labor market estimates from the state study should be
viewed as a best case scenario.

By default, the PDA mainly affects workers who place a low—or no—value on the
federally legislated benefits. These workers did not necessarily place such a high value on
receiving the maternity benefit; therefore, the cost of the benefit was not shifted to the
covered group’s wages. True to the hypothesis, a state mandate was not enacted before the
PDA in these states. The total effect of a federal mandate enacted before any similar state
legislation would be somewhere between these results: a moderate decrease in the wage,

accompanied by disemployment effects.

57

Table 1: Dates and Status of the Employment Mandates

 

State Date of Enactment State Study Federal Study
Alabama N/A N E
Alaska 6/4/75 — N
Arizona N/A — —
Arkansas N/A — —
California 9/22/76 — —
Colorado 5/3 1/72 — —
Connecticut 10/ 1 I73 E N
Delaware N/A — —
Florida N/A — —
Georgia N/A — —
Hawaii 5/73 — —
Idaho N/A — —-
Illinois 1/16/73 E N
Indiana Pre-l976" — _
Iowa 10/9/7 2 — N
Kansas l/1/74 — N
Kentucky Pre- 19763 —— —
Louisiana N/A —- E
Maine 5/15/76 — —
Maryland 7/1 1/7 2 — N
Massachusetts 6/ 19/72 N
Michigan 9/72 N
Minnesota 3/71 — N
Mississippi N/A N E
Missouri Pre- 19763 — N
Montana 7/1/75 — N
Nebraska N/A — E
Nevada N/A — —
New Hampshire Pre-l9763 —

New Jersey 1948 — —
New Mexico N/A — E
New York 9/73 — —
North Carolina N/A N E

58

Table 1 (cont’d)

S_ta_t§_ Date of Enactment State Study Federal Study
North Dakota N/A — E
Ohio N/A N —
Oklahoma N/A — —
Oregon 6/71 — —
Pennsylvania 12/7 1 E N
Rhode Island 5/7/75 — —
South Carolina N/A — —
South Dakota 1972 — N
Tennessee N/A — .—
Texas N/A — —
Utah N/A — —
Vermont N/A — E
Virginia N/A — —
Washington 10/3/73 — N
West Virginia N/A — E
Wisconsin 7/10/72 E N
Wyoming N/A — --
NM

1. This information is from Commerce Clearing House (1978), NELP (1975), U.S. Supreme Court
(1975/76), and Weyland (1978).

2. E” denotes an experimental state; “N” denotes a non-experimental state. See chapter 1 and appendix A
for states not in a particular study.

3. The exact dates of compliance in these employer mandates states are unknown.
4. These States have payroll tax systems that fund pregnancy disability.

59

Table 2: DDD Regression Framework

Pre-Mandate

(I) wijt=a +leijt

(II) W,,-.=0t +8 I Xij,+B4COVERi

(III) Wij,=0l +leij,+BZSj

(IV) Wij,=0t +131X,,-.+Bsz+B4COVER,+BgCOVER,*Sj

Post-Mandate

(1’) Wijr=a +leijt+B3Yt

(II’) Wij,=0t +131Xij,+B3Y,+B4COVERi+ByCOVER,*Y,

(1H,) Wijtza +13lxijt+stj+B3Yt+Bssj*Yt

(IV ’) W,,-,=or +151Xij,+BZS,-+[33Y,+B..COVER,+BSS,'*Y,+[3,,COVER,"'SJ+B7COVER,*Yt
+B8MANDATEJ-u

First-Difference

(V) [I'I’li Awijt:B3Yt

(VI) [II-II’]: AWfi,=B3Y,+B7COVERi*Y,

(VII) [III-III’]: AW,,-.=B3Y.+BSSJ*Y,

(VIII) [IV-IV’]: AWUFBgY, +135Sj’l‘Yl +B7COVER,*Y, +BgMANDATEJ-u-

Second-Difference
(IX) [(III-III’)-(I-I’)l AAwijr—‘BSSj*Yt
(X) [(IV-IV,)-(II-II,)] AAWiJ-tzfissz, +B3MANDATEN

Third-Difference
(XI) [[(IV-IV’)-(II-II’)]-[(III-III’)-(I-I’)]] AAAWfiFBgMANDATEJ-u

m;

1. In the set of equations:
Xij, refers to the set of demographic variables.
S,- is a dummy variable which equals 1 if the respondent is in the experimental state, and zero otherwise.
Yt is a dummy variable which equals 1 following the mandate and zero before the mandate.

COVER, is a dummy variable which equals 1 for respondents covered by the mandate, and zero
otherwise.

MANDATEijt equals one if the respondent is in an experimental in the post-mandate, and is covered by
the law.

Equations I and I’ refer to respondents in the non-experiment state and the non—covered group.
Equations 11 and II’ refer to respondents in the non-experiment state and the covered group.
Equations III and III’ refer to respondents in the experiment state and the non-covered group.
Equations IV and IV’ refer to respondent in the experiment state and the covered group.

.U‘PP’E"

Education:

Experience:

Experience

Squared/ 1000:

Non-White:

Female:

Female*Marn'ed:

Union
Coverage:

Covered
Group:

Mandate:

Status of Variables:3

State:
Covered*State:
State*Year".

Adjusted R2 :
N:

Notes:

1. The numbers in parenthesis are standard errors.

60

Table 3: State Employer Mandates, Full Sample

(1)
Log
Hourly
Wage
.057
(.001)

.020
(.001)

-.339
(.012)

-.101
(.007)

-.258
(.010)

.168
(.008)

-.205
(.011)

.129
(.005)

.077
(.015)

.047
(.016)

Group
Group
Group

.481
33277

(2)
Log
Hourly
was
.055
(.001)

.020
(.001)

-.338
(.012)

-.O89
(.007)

-.261
(.010)

.167
(.007)

-.203
(.011)

.1 13
(.005)

.078
(.015)

-.035
(.016)

Individual

Group
Group

.494
33277

(3)
Log
Hourly
me.
.054
(.001)

.020
(.001)

-.337
(.012)

-.089
(.007)

-.261
(.010)

.167
(.007)

-.202
(.011)

.1 12
(.005)

.102
(.024)

-.040
(.020)

Individual
Individual

Group

.494
33277

2. Regression also includes 15 industry dummy variables.
3. ‘Individual’ indicates that the separate state and year dummy variables are used. ‘Group’ indicates that

the variables are entered as a group. ‘N/A’ indicates that the variables are not included.
The variables included in the probit are age and age squared, not experience and its square.

5‘

(4)
log

Weekly

Hours
-.001
(.001)

.002
(.001)

-.091
(.012)

-.001
(.007)

-.040
(.010)

.065
(.007)

-.I89
(.011)

.007
(.005)

-.136
(.024)

-.036
(.019)

Individual
Individual

Group

.152
33277

(5)
Employ-
ment
Bro—bit
.016
(.001 )

.038
(.001)

-.543
(.012)

-034
(.006)

-.057
(.009)

.303
(.008)

-.485
(009)

-. 192
(.020)

-.009
(.016)

Individual
Individual
Group

.163
83780

5. Self-employed respondents, government employees (except state and local administration), and those
who report a wage less than $1 per hour or greater than $100 per hour are dropped from the sample.
Workers who report an industry not covered by Title VII or the FEP acts are dropped from the sample.

See text for details.

12th Grade Education
or Greater:

Less than a 12"1
Grade Education:

Pre-Gilbert:

(1 )
Log
Hourly
flags

-.046
(.023)
[22133]

-.009
(.041)
[11144]

-.051
(.024)
[22166]

61

Table 4: State Study by Type of Worker: Large Sample

(2)
Log
Weekly
Hourly

-.049
(.023)
[22133]

.042
(.040)
[1 1 144]

-.023
(.025)
[22166]

(3)
Employ
ment
Probit

-.005
(.018)
[53940]

-.027
(.036)
[29840]

.010
(.020)
[56131]

Education:

Experience:

Experience

Squared/1000:

Non-White:

Female:

Married:

Female*Married:

Covered
Group:

Mandate:

Status of Variables:3

State:
Covered*State:
State*Year:

Adjusted R2:
N :

Notes:

H

PPP

(1 )
Log
Hourly
Mes
.054
(.001)

.021
(.001)

-.322
(.012)

-.103
(.007)

—.279
(.010)

.211
(.008)

-.259
(.012)

.158
(.033)

-.007
(.021)

Individual
Individual
Individual

.373
49152

62

(2)

Log
Weekly
HM
-.003
(.001 )

.003
(.001)

-.098
(.000)

.01 l
(.006)

-.049
(.009)

.063
(.007)

-. 199
(.010)

-.073
(.027)

-.037
(.017)

Individual
Individual
Individual

.142
49152

The numbers in parenthesis are standard errors.
Regression also includes 15 industry dummy variables.

The variables included in the probit are age and age squared, not experience and its square.
Self-employed respondents, government employees (except state and local administration), and those

Table 5: Federal Pregnancy Discrimination Act, Full Sample

(3)
Log
Yearly
Weeks
.009
(.001 )

.022
(.001)

-392
(.017)

-.036
(.010)

-.037
(.015)

.061
(.011)

-.182
(.017)

-.094
(.045)

-.O7O
(.029)

Individual
Individual
Individual

.088
49152

(4)
Employ-
ment
ms
.020
(.001 )

.023
(.001)

-.388
(.012)

-.075
(.006)

-.064
(.008)

.251
(.008)

-.378
(.009)

-.027
(.028)

-.027
(.017)

Individual
Individual
Individual

.149
83758

who report a wage less than $1 per hour or greater than $100 per hour are dropped from the sample.
Workers who report an industry not covered by Title VII or the state FEP acts are dropped from the

sample. See text for details.

Table 6: Federal Study by Type of Worker: Large Sample

(1 )
Log
Hourly
flags

12‘h Grade Education
or Greater: -.005
(.024)
[36868]

Less than a 12th
Grade Education: .002
(.049)
[12284]

63

(2)
Log
Weekly
Hourly

-.038
(.020)
[36868]

-.007
(.036)
[12284]

(3)
Log
Yearly
Weeks

-.041
(.033)
[36868]

-.223
(.070)
[12284]

(4)
Employ
ment
13m

-.014
(.019)
[59234]

4047
(021)
[24524]

64

Table 7: Labor Market Effects of State Employer Mandates: Small Sample
(1) (2) (3)
Log Log Employ-

Hourly Weekly ment
Wage Hours Probit

Education: .059 -.012 .014
(.002) (.003) (.002)

Experience: .017 -.036 -.083
(.002) (.003) (.006)

Experience -.480 l .300 l .323

Squared: (.096) (.133) (.106)

Non-White: -.033 .090 .066
(.013) (.018) (.012)

Union .172 .087

Coverage: (.012) (.018)

Covered Group: .166 .602
(-089) (.122)

Mandate Effect: -.037 -.024 .005
(.020) (.027) (.015)

Status of Variables:3

State: Individual Individual Individual

Covered*State: Individual Individual N/A

State*Year: Group Group Group

Adjusted R2: .304 .184 .018

N: 7664 7664 22778

Notes:

1. The numbers in parenthesis are standard errors.
Regression also includes 15 industry dummy variables.

9"!"

9'.“

The term ‘Individual’ indicates that the state, year, and covered group variables are entered individually.
The term ‘N/A’ indicates that the variables are not used in this specification.

The variables included in the probit are age and age squared, not experience and its square.

Respondents who report self-employment, government employees (except state and local
administration), and those with a wage less than $1 per hour or greater than $100 per hour are dropped
from the sample.

All workers who report working in industries exempt from Title VII regulations or from the state
employer mandates are in the non-covered group.

Table 8: State Study by Type of Worker: Small Sample

(1 )
L98
Hourly
flags

12th Grade Education

or Greater: -.041
(.022)
[6199]

Less than a 12m

Grade Education: -.002
(.042)
[1465]

Pre-Gilbert: -.052
(.024)
[4962]

65

(2)
Log
Weekly
Hourly

-.042
(.031)
[6199]

.056
(.055)
[1465]

-.012
(.034)
[4962]

(3)
Employ
ment
EM

.008
(.017)
[17921]

-.012
(.033)
[4857]

.016
(.019)
[15424]

Table 9: State Study by Type of Worker: Covered Group Only

Main Sample

12"1 Grade Education
or Greater:

Less than a 12'h
Grade Education:

Pre-Gilbert:

(1 )
L08
Hourly
Edge

-.044
(.018)
[7257]

-.047
(.020)
[5 845 ]

-.004
(.038)
[1412]

-.062
(.022)
[7257]

(2)
Log
Weekly
Hourly

-.017
(.024)
[7257]

-.033
(.028)
[5845]

.042
(.048)
[1412]

-.006
(.031)
[7257]

(3)
Employ
ment
Probit

.005
(.015)
[22778]

.008
(.017)
[17921]

-.012
(.033)
[4857]

.016
(.019)
[22778]

Table 10: Labor Market Effects of the Federal PDA: Small Sample

67

(1) (2) (3) (4)
Log Log Log Employ
Hourly Weekly Annual -ment
Wage Hours Weeks Probit
Education: .073 .006 .048 .035
(.002) (.002) (.003) (.001)
Experience: .019 -.027 -.012 -.050
(.002) (.002) (.004) (.006)
Experience -.463 .897 1 .028 .871
Squared/1000: (.099) (.1 10) (.167) (.103)
Non-White: -.040 .109 .014 .044
(.013) (.015) (.023) (.011)
Control Group: .262 .079 .094
(.063) (.070) (. 108)
Mandate Effect: -.040 -.070 -. 131 -.026
(.041 ) (.046) (.071) (.014)
Status of Variables:3
State: Individual Individual Individual Individual
State*Year: Individual Individual Individual N/A
State*Covered: Individual Individual Individual N/A
Adjusted R2: .206 .099 .064 .036
N: 1.4791 14791 14791 25189
Notes:

1. The numbers in parenthesis are standard errors.

991°

Regression includes 15 industry dummy variables.
The term ‘Individual’ indicates that the variables are entered individually. The term ‘N/A’ indicates that

the variables are not used in this specification.

P

The variables included in the probit are age and age squared, not experience and its square.

5. Respondents who report self-employment, government employees (except state and local
administration), and those with a wage less than $1 per hour or greater than $100 per hour are dropped
from the sample.

6. All workers who report working in industries that are exempt from Title VII regulations or from the
state employer mandates are in the non-covered group.

Table 11: Federal Study by Type of Worker: Small Sample

(1)

Log
Hourly
figs
12th Grade Education
or Greater: -.032
(.043)
[13100]
Less than a 12th
Grade Education: -.l3l
(.174)

[1691]

68

(2)
Log
Weekly
Hourly

-.062
(.048)
[13100]

-.122
(.177)
[1691]

(3)
Log
Yearly
Weeks

-.131
(.072)
[13100]

-.084
(.345)
[1691]

(4)
Employ
ment
Probit

-.010
(.015)
[21276]

-.082
(.033)
[3913]

Table 12: Federal Study by Type of Worker:

( 1 )
L98
Hourly
Wage

Main Study .006
(.018)
[12229]

12th Grade Education
or Greater: .002
(.019)
[10684]

Less than a 12'h

Grade Education: .021
(.045)
[1514]

69

(2)
Log
Weekly
Hourly

-.026
(.018)
[12229]

-.028
(.020)
[10684]

.028
(.042)
[1514]

Covered Group Only

(3)
Log
Yearly
Weeks

-.086
(.030)
[12229]

-.075
(.033)
[10684]

-.189
(.089)
[1514]

(4)
Employ
ment
Pr_0bi_t

-.026
(.014)
[25189]

-.010
(.015)
[21276]

-.082
(.033)
[3913]

70

Table 13: Legislation Favorable to Females as of 1978

Federal Experimental Maternity Mandate States

FEP Status2 @ual Pay Status‘ Signed ERA4 Adopted ERA
Alabama N N N N
Louisiana N N N N
Mississippi N N N N
Nebraska Y Y N5 N
New Mexico N N Y Y
North Carolina N N N N
North Dakota N Y Y N
Vermont N Y N N
West Virginia Y Y Y N

Federal Non-Experimental Maternity Mandate States
FEP Status2 Equal Pay Status3 Signed ERA4 Adopted ERA

<0

Alaska
Connecticut
Illinois

Iowa

Kansas
Maryland
Massachusetts
Michigan
Minnesota
Missouri
Montana
Pennsylvania
South Dakota
Washington
Wisconsin

~<~<~<~<~<~<~<~<~<~<~<~<~<~<~<
z~<~<~<~<~<~<~<~<-<~<z~<-<~<

~<~<<¢<~<2~<~<~<~<~<~<z~<
z<z~<zzzz~<~<zz~<z~<

Notes:

1. This information is from the Federal Register (1977), Lee (1980), and the U.S. Department of Labor
Women’s Bureau (1975, 1983).

2. States that have a FEP agency were designated ‘706’ agencies by the EEOC before September 1977.

3. The states listed as having an equal pay statute are those that adopted a separate statute as of January
1982. However, states that include equal pay as part of their FEP laws are not listed.

4. The states that ratified the ERA did so by 1980.
5. Nebraska signed the petition for the ERA in 1972, but rescinded support for it in 1973.

6. The 1975 Handbook of Women Workers lists Alaska as having a separate ERA statute; however the
1983 edition does not.

7. South Dakota signed the petition for the ERA in 1973, but rescinded support for it in 1979.

BIBLIOGRAPHY

BIBLIOGRAPHY

Ashenfelter, Orley, and David Card. 1985. “Using the Longitudinal Structure of Earnings to
Estimate the Effect of Training Programs.” The Review of Economics and Statistics
47 (February): 648-60.

Ashenfelter, Orley. 1978. “Estimation the Effect of Training Programs on Earnings.” The
Review of Economics and Statistics 40 (1) (February): 47-57.

Averett, Susan L., and Julie L. Hotchkiss. 1995. “The Probability of Receiving Benefits at
Different Hours of Work.” The American Economic Review, 85 (May): 276-80.

Commerce Clearing House. 1978. New 1978 Pregnancy Benefits and Discrimination Rules
with Explanation and State Survey. Chicago: Commerce Clearing House.

Conway, Karen Smith, and Michael R. Butler. 1992. “State Abortion Legislation as a Public
Good—Before and After Roe V. Wade.” Economic Inquiry XXX (October): 609-
26.

Freeman, Richard B. 1979. “The Effect of Demographic Factors on Age-Earnings Profiles.”
The Journal of Human Resources XIV (Summer): 289-318.

General Accounting Office. 1977. “Discrimination on the Basis of Pregnancy.” Hearings
Before the Subcommittee on Labor of the Committee on Human Resources. 95th

Cong, lst sess., S. 995.

General Accounting Office. 1979. “Legislation to Prohibit Sex Discrimination on the Basis

of Pregnancy.” Hearings Before the Subcommittee on Employment Opportunities of
the Committee on Education and Labor. 95th Cong., lst sess., H.R.5055 and HR.
6075.

Grant, James H. and Daniel S. Hamermesh. 1981. “Labor Market Competition Among
Youths, White Women and Others.” The Review of Economics and Statistics
(August): 354-60.

Gruber, Jonathan. 1992. “The Efficiency of a Group-Specific Mandated Benefit: Evidence

from Health Insurance Benefits for Maternity.” Working Paper No. 4157.
Cambridge, MA: National Bureau of Economic Research.

71

72

Gruber, Jonathan. 1994a. “The Incidence of Mandated Maternity Benefits.” American
Economic Review (June): 622-41.

Gruber, Jonathan. 1994b. “Payroll Taxation, Employer Mandates, and the Labor Market:
Theory, Evidence, and Unanswered Questions.” Prepared for the W.E. Upjohn
Institute Conference an Employee Benefits, Labor Costs and Labor Markets in
Canada and the U.S.

Lee, Rex E. 1980. A Lawyer Looks at the Equal Right Amendment. Provo: Brigham Young
Press.

Medoff, Marshall H. 1988. “An Economic Analysis of the Demand for Abortions.”
Economic Inquiry XXVI (April): 331-51.

National Employment Law Project, Inc. 1975. Legal Developments in Sex Discrimination
Concerning Maternity Leaves and Benefits. New York: NELP.

Neumark, David and Wendy Stock. 1997. “Age Discrimination Laws and labor Market
Efficiency.” Working Paper No. 6088. Cambridge, MA: National Bureau of
Economic Research.

Rasmusen, Eric. 1992. “Observed Choice and Optimism in Estimating The Effects of
Government Policies.” Working Paper, Indiana University Department of
Economics, December 1992.

Smith, Authur B. 1980. “The Law and Equal Employment Opportunity: What’s Past Should
Not Be Prologue.” Industrial and Labor Relations Review (July): 493-505.

Summers, Lawrence H. 1989. “Some Simple Economics of Mandated Benefits.” American
Economic Review 79 (May): 177-83.

United States Department of Labor, Women’s Bureau. 1975. “Chapter 7, State Labor Laws

of Special Interest to Women.” In 1975 Handbook on Women Workers. Washington
DC: U.S. General Printing Office.

United States Department of Labor, Women’s Bureau. 1983. “Chapter VH, State Labor
Laws of Special Interest to Women.” In Time of Change: 1983 Handbook on Women
Workers. Washington DC: U.S. General Printing Office.

United States Supreme Court. 1975/76. “Martha V. Gilbert, et a1. versus General Electric
Company.” Law Reprints. Labor Series: the Supreme Court of the United States
Petition, Briefs and Opinion 9 (7). New York, Law Reprints.

United States Supreme Court. 1976/77. “Martha V. Gilbert, et al. versus General Electric
Company.” Law Reprints. Labor Series: the Supreme Court of the United States
Petition, Briefs and Opinion 10(4). New York, Law Reprints.

73

Wald, Sarah E. 1982. “Alternatives to Title VII: State Statutory and Common-Law

Remedies for Employment Discrimination.” Harvard Women ’3 Law Journal 5
(May): 35-72.

Weyland, Ruth. 1978. “Pregnancy as a Disability.” In Proceedings of the Third ABA
National Institute on Equal Employment Opportunity Law. New Orleans: ABA
Press.

Wyckoff, James H. 1984. “The Nonexcludable Publicness of Primary and Secondary Public
Education.” Journal of Public Economics (August): 331-51.

CHAPTER 3: THE INCIDENCE OF MANDATED MATERNITY BENEFITS, A
COMMENT

In a recent AER paper, Gruber (1994a) examines the labor market effects of
amending state insurance codes such that complications of pregnancy are covered. His
estimates suggest that the wages of those who were eligible to receive the benefit fell at rate
equal to the cost associated with the benefit. Because he does not find a decline in net labor
input, he concludes that the decrease in wages is approximately equal to the employer’s cost
of providing it. This indicates that the group receiving the benefit places a monetary value
on the maternity mandate at or above its actual cost. To confirm these results, Gruber
performs a ‘reverse experiment’ which estimates the effect of the federal Pregnancy
Discrimination Act of 1978 (PDA) on the states that did not enact an insurance code
amendment before 1978. The results from this study support his earlier findings.79

Abstracting from these results, Gruber asserts that employers are able to shift
mandated costs to the wages of demographically identifiable groups within the workplace
(Gruber, 1994b). This conclusion is relevant in light of the ongoing debate over the
effectiveness of employer mandates. Since Gruber’s findings are central to this debate, it is
important to confirm his results capture the effects of only the insurance code amendments

and not other maternity policies in place at the time.

 

79 Gruber bases his claim for the relative equivalence of the PDA and the insurance code amendments on the
congressional testimony of health insurance representatives. This group estimated that the existing insurance
code amendments represent two-thirds of the cost of implementing the PDA. However, as described in section
H, the federal legislation still affects employers in a non-experimental state who have met the minimum
requirements of an insurance code amendment.

74

75

Reviewing the legislative history described in chapter 1, it quickly becomes apparent
that the state insurance code amendments were not the sole determinants of state maternity
policies during the 19703. This observation calls into question Gruber’s results on the
grounds that he does not control for a second set of closely related maternity policies, state
employer maternity mandates. The state employer mandates and the state insurance code
amendments are Similar in scope: both require complications of pregnancy to be covered
under an insurance policy. In states that adopted both maternity policies, however, the state
employer mandates preceded the insurance code amendments.80 The dates at which states
adopted either——or both—matemity policies are critical to evaluating the labor market
effects of a maternity mandate. Since states enacted the employer mandates first, estimates
of the effect of an insurance code amendment should control for the presence of an employer
mandate. Otherwise, this estimate does not represent the true effect of a group-specific
mandate. Conversely, some states that Gruber includes as non-amendment states do have an
employer mandate; thus, testing the effects of the insurance code amendment against these
states may affect the results he derives.

The remainder of this comment is organized as follows. Section I begins with an
abbreviated history of maternity mandates; then, differences between the relevant pieces of
state legislation are discussed. Section 11 presents an identical replication of the results from
Gruber’s state study. The potential impact of the state employer mandates on Gruber’s state
study results is examined in Section II]. Section IV replicates Gruber’s federal study. In
Section V, I use the subset Gruber’s federal states that adopted an employer mandate to
estimate more precisely the labor market effects of the 1978 federal Pregnancy

Discrimination Act (PDA). Section VI concludes.

 

8" California is the exception, see appendix A.

76
1. Types of State Maternity Policy

A review of the legislation relating to maternity mandates reveals three distinct
periods of enforcement.81 Before 1972, most state legislatures did not address the issue of
mandating insurance for maternity conditions.82 The Equal Employment Opportunity
Commission changed its stance in 1972 to support the position that employers should treat
all conditions relating to pregnancy as “...temporary disabilities...under any
health...insurance plan available in connection with employment.”83 Although the EEOC
did not seek congressional approval for the pregnancy guidelines at that time, it was the
beginning of a transitional period during which many states adopted some type of maternity
policy. The end of this period occurred in 1978 when President Carter, with congressional
approval, signed into law the federal Pregnancy Discrimination Act of 1978. The PDA
closely follows the wording of the EEOC’S April 1972 position.

During the transitional period (April 1972 to October 1978), a number of states
adopted a state insurance code amendment and/or a state employer mandate. States with
both types of maternity policies adopted the employer mandates first. In addition to the
timing of the adoption, these maternity policies differ on two key points: the group covered
under the legislation, and the extent to which the legislation covers that group. The employer
mandates, which followed the wording of the EEOC’S stated position as of April 1972,

ensured that all employers include any pregnancy condition when providing employee

 

8' An in-depth review of these maternity policies enacted during the 19703 can be found in chapter 1.
32 A few exceptions do exist. The state of New Jersey did provide insurance benefits for some maternity
conditions at an earlier date. See appendix A for a full listing of the dates at which states adopted maternity
olicies.

3 Weyland I978, 180.

77
disability insurance.84 In contrast, the insurance code amendments generally provided all
who purchase insurance with coverage for complications of pregnancy.

By adopting both policies, a state effectively provided the minimum level of
maternity coverage for non-employed females, as well as a higher level of coverage to
female employees whose employer offered health insurance coverage for temporary
disabilities.85 Other states chose between providing coverage under the terms of the
insurance code amendments or the employer mandates. The remaining states adopted
neither.

Regardless, it is the differences in the dates of enactment that may affect Gruber’s
results. His state study includes states with employer mandates that are rnisclassified as non-
amendment states. Conversely, he correctly classifies states that enacted both maternity
policies as amendment states, but misspecifies the date of adoption. He uses the date of
enactment for the insurance code amendments rather than the earlier dates of adoption for the
employer mandates. Because the labor market estimates found in his state study derive from
the pre- to post-legislative change in states with an insurance code amendment, his state
study may underestimate the effect of the insurance code amendments.

In his federal Study, Gruber classifies amendment and non-amendment states based
on whether the state enacted an insurance code amendment before the passage of the federal
PDA. Factoring in the employer mandates changes the classification of some states. The

effect of the insurance code amendments is then underestimated due to Gruber’s

 

V

8" Most employer mandates only required firms with a minimum number of employees—under EEOC guidelines
at that time the minimum number was 25—to provided maternity coverage if the employer provided insurance
coverage for other temporary non-occupational disabilities. Some states legislated exceptions to the EEOC’S
rule. Connecticut, for example, required all employers to provide insurance coverage for all pregnant workers,
regardless of whether the employer provided coverage for any other type of temporary worker disability.

85 It should be noted that in a number of states, non-employed married females benefited through their husband’s
insurance policy.

78
misclassification of employer mandates states as states without pregnancy coverage, or “non-
amendment states.” Thus, the estimates from both Gruber’s federal study and his state study

may be sensitive to the inclusion of states with employer mandates.

II. Replication of Gruber’s State Study

To test the effects of the state insurance code amendments, Gruber divides the
eligible states into 2 groups: (1) those that enacted this maternity policy after July 1, 1976 but
prior to the PDA, (experimental states) and (2) those that did not (non-experimental states).86
States ineligible for this study include those that enacted an insurance code amendment after
January 1, 1977. Defining the legislative period as 1976-77, Gruber examines the pre-policy
labor force Situation for the years 1974-75, and the post-policy situation for the years 1977-
78. His primary focus is on the labor market effects of the amendments on married females
age 20-40.87 The composition of the non-covered group, which serves as the control group in
the DDD estimation, is the following: single males age 20-40 and people over the age of 40.

The first row of table 1 presents the results of an identical replication of Gruber’s
state study. The post-amendment wages of married females fell by 4.3 percent in the
experimental states. Two labor input effects are of note: employment decreases, although

insignificantly, and the number of hours worked per week increases by a significant 4.9

 

8" The experimental states are Illinois, New Jersey and New York, while the non-experimental states are Ohio,
Indiana, Connecticut, Massachusetts and North Carolina (see appendix A for the adoption dates for all states).
Gmber states that he chooses North Carolina as a control state for New Jersey and New York to avoid comparing
the Mid-Atlantic experimental states to the New England non-experimental states. He tests this specification
without the state of North Carolina, and finds that the results are similar to those that include the state (Gruber
1992).

“7 Gruber also tests the labor market effects of the insurance amendments separately on two other groups:
married males age 20—40 whose wives are covered by the insurance amendments; and single females age 20—40.
In this paper, I chose to focus on the group of married females for two reasons. First, there is uneven spousal
coverage in employer mandates across states; some states expressly covered the wives of male employees while
others do not. Second, the incidence of pregnancy among single females of childbearing age is lower than
among married females of childbearing age.

79
percent. The change in worker composition to low-hours employees is consistent with the
hypothesis that the insurance code amendments increase the fixed cost of group-specific
labor. For the group that receives the maternity benefit, the net result is a shift of the
amendment’s cost to their wages. These results indicate that those who receive the benefit

under a state insurance code amendment place a positive value on it.88

III. Impact of Employer Mandates on Gruber’s State Study

In the absence of the state employer mandates, Gruber’s results reveal a great deal
about the effect of maternity mandates on wages and employment. His conclusions may be
incorrect, however, if the states are reclassified from non-amendment states to amendment
states—using the date of adoption for the employer mandates—when the employer mandates
are incorporated in the analysis. Table 2, which lists the states Gruber uses in his state study,
presents a comparison of the states with insurance code amendments and states with
employer mandates.89 In column 1, the date (if any) that a state enacted an insurance code
amendment is listed. The state status that Gruber assigned (e.g., amendment, non-
amendment) in the state Study is found in column 2. For comparison, column 3 presents the
date (if any) that a state adopted an employer mandate; column 4 lists the state’s employer
mandate status after the employer mandates are included.

AS Shown in column 3, employer mandates are present in most of the states in
Gruber’s state study. Indiana, for example, did not enact an insurance code amendment
before 1976 but it did adopt an employer mandate. Thus, classifying Indiana as a non-

amendment state in light of existence of the employer mandate is incorrect. A Similar case is

 

8“ Gruber also tests a cost specification in which he imputes the cost of increasing the level of health insurance
instead of the mandate variable. His basic result is unchanged.
“9 A full comparison can be found in appendix A.

80
found in Connecticut and Massachusetts, both considered non-amendment states by Gruber.
Illinois, on the other hand, enacted an employer mandate more than three years before the
state’s insurance code amendment took effect.90 Thus, the amendment status of Illinois is
correct; however, Gruber uses the incorrect date—that of the insurance code amendment
rather than the employer mandate—to estimate the full effect of implementing a maternity
policy.

Unlike the other states, New Jersey and New York provided insurance for pregnancy
conditions through a state-run temporary disability insurance (T DI) system.91 This system
required both males and females to contribute to the state plan, which covers temporary
disabilities. In 1975 and 1976 respectively, New Jersey and New York considered pregnancy
to be a temporary disability under the TDI system.92 This system is separate from both the
state insurance code amendments, which affected only those who purchased insurance, and
the state employer mandates, which did not require any type of employee contribution to a
state-run system. Therefore, analyses of both the insurance code amendments and the
employer mandates should exclude states with a TDI system.

Ideally, the correct test of the effect of the insurance code amendments in Gruber’s
State study should be conducted using the states that have only an insurance code amendment
or neither policy. A state that uses a TDI system for implementing a maternity mandate (e.g.,
New Jersey, New York) would not be included. Also excluded would be any state (e.g.,
Connecticut, Illinois, Massachusetts) in which an employer mandate was adopted during the

pre- or post-policy years. The remaining states would then be divided according to their

 

9" Since most workers were covered by the employer mandate, the impact of the insurance code amendment falls
only on workers in firms with less than 25 employees that provided employee health insurance. Thus, Illinois’
insurance code amendment may have a small effect, although the magnitude would certainly be underestimated.
9' See chapter 1 for more details.

92 The amendment of the insurance codes and the change in the New York TDI system were independent events.

81
insurance code amendment classification. In this case, however, only the status of North
Carolina and Ohio—both non-amendment states—can be determined. Thus, the effect of the
insurance code amendments, after removing the employer mandate states, cannot be
measured using this group of states in conjunction with the constraints imposed by Gruber’s

state study. The same is true of estimating the effects the employer mandates.93

IV. Replication of Gruber’s Federal Study

To test the effect of the federal PDA on states without a maternity mandate, Gruber
classifies states with an insurance code amendment as non-experimental states. The purpose
of this federal ‘reverse experiment’ is to confirm the direction and magnitude of his state
study results. Gruber’s state and federal studies differ in only three respects: the composition
of the experimental and non-experimental states; the pre- and post-policy years; and the
union variable, which is present only in the state study.94 In the federal study, states that
enacted an insurance code amendment before January 1 1977, are in the non-experimental
group; states without an amendment are in the experimental group. Gruber drops states that
enacted an insurance code amendment after January 1977, but before the federal legislation.
Finally, the pre-PDA years in this analysis are 1978-79 and the post-PDA years are 1981-82.

The second row of table 1 presents the results of replicating Gruber’s federal
specification. In the experimental states, the decrease in the wages of married women is
almost one-half the decrease (2.1 percent) reported in his state study.95 It is marginally

significant. The coefficients on the number of hours worked per week and probability of

 

93 See chapter 2 for an empirical test of the employer mandates using a larger subset of states.

9" Starting in 1979, the May CPS supplement began reporting the earnings variables only for the outgoing
rotation groups. As a result, Gruber tests the labor market effects of the federal PDA using the March CPS. See
section VI in chapter 2 for a detailed explanation of the differences between these data sets.

95 The coefficients are identical to Gruber’s (1994a), but, this replication contains an additional observation.

82
employment are small and not significant, but they are in the same direction as the results
found in the state study. The net effect of the maternity legislation on this group is a full
shifting of the PDA’s cost to the covered group’s wages in the experimental states. There is
no Significant change in net labor input. These results confirm the findings of the state study
and reinforce the hypothesis that workers receiving the maternity benefits assign a positive

value to receiving them.

V. Impact of Employer Mandates on Gruber’s Federal Study

AS in Gruber’s state study, the presence of the employer mandates affects a number
of states in his federal study. Unlike the state study, the enactment status of an employer
mandate as of January 1, 1977, and not the actual date of enactment, matters. This is because
the federal study is structured such that it depends only on the status of a state (e.g.,
amendment, non-amendment). In conjunction with the larger number of states included in
the federal Study, this allows for a better test of the employer mandates than the State study.
Table 3 is the federal compliment to the table 2. The dates of enactment for states with an
insurance code amendment and/or an employer mandate are listed in column 1 and column
3, respectively. Column 2 lists the amendment status used by Gruber to determine the
impact of the PDA.

Since few of the states with an insurance code amendment are left, this study
attempts to estimate the labor market impact of the state employer mandates. Eighteen states
remain (see table 3, column 4 for a list of these states) following the imposition of the
following restrictions. To determine the impact of the federal PDA on states without either
maternity policy, I first isolate the states that clearly enacted and enforced a state employer

mandate and the states that adopted neither policy. States that only enacted an insurance

83

code amendment are not included in this estimation. I delete any respondent in a state that
enacted a maternity policy during the early 19703 and did not continue to enforce that
legislation.96 Finally, for the reasons given earlier, I do not consider any of the states with a
State-run TDI system

Next, I isolate any effects that may result from dropping the ineligible states and the
TDI states. To do this, I re-estimate Gruber’s federal study according to his classifications
(e.g., Alaska is included as an experimental state, while Colorado is dropped), but using the
smaller sample of 18 states. The third row of table 1 presents the results from the amended
version of Gruber’s federal study. In the remaining states, the decrease in the wages of
married women is of a larger magnitude than in Gruber’s federal study—3.4 percent—but
less Significant than the full-sample results. As in the full-sample federal study, the
coefficients on the number of hours worked per week and the probability of employment are
insignificant. Thus, the decline in women’s wages is greater using the smaller sample, which
excludes states with an unclear employer mandate status and states that insure maternity
through a state-run TDI plan, than the larger sample.

Using the smaller sample, I next change the status of these states to reflect whether a
State has an employer mandate or neither maternity policy (see table 3, column 4). States are
grouped according to their employer mandate status. In every other way, this test is identical
to the amended version of Gruber’s federal study described in the preceding paragraph. The
results from this study are found in table 1, row 4. The post-mandate wages of married
females fell by 2.4 percent in the experimental states. The number of weekly hours worked

decreased, although insignificantly. The decline in the probability of employment is weakly

 

9" See chapter 1 for a complete history of the maternity legislation in place at that time.

84
Significant—below the ten percent level—suggesting that disemployment effects may

accompany the wage decline.97

VI. Conclusions

The inclusion of employer mandates in both Gruber’s studies results in the
reclassification of a number of states. Several of these changes occurred because Gruber’s
studies did not account for a state-enacted employer mandate. Others reflected a situation in
which a state adopted an employer mandate at an earlier date than an insurance code
amendment. The purpose of this comment was to re-estimate the effect of the maternity
mandates after reclassifying states to reflect all maternity mandates in place at that time.
However, an examination of Gruber’s state study in light of the employer mandates was not
possible due to the small number of states remaining.

Conversely, a sufficient number of states remained in the federal study to estimate
the effect of the employer maternity mandates. Although these results do not differ
drastically from Gruber’s, they do indicate that disemployment effects may result from the
adoption of the employer mandate. The decline in the wage is nearly the same as Gruber’s
estimate, but less significant. This decline indicates that the workers receiving the benefit
value the mandate. Apart from the wage effects, however, the coefficient on the probability
of employment indicates the PDA has a negative effect, although this decline is weakly
significant. Thus, changing the state status to reflect the employer mandate classifications
suggests evidence of disemployment effects in addition to the wage decreases. Since these

effects indicate either a low valuation by the covered group or significant barriers to wage

 

97 Chapter 2 considers the labor market effects of a group-specific mandate using the full set of eligible states.
The results indicate that disemployment effects occur following the PDAS enactment. See chapter 2 for details.

85
adjustment, the potential for disemployment effects in this case is especially important to

policy makers who are considering enacting similar group-specific mandates.

Table 1: Results from Replicating and Correcting Gruber’s Study

Log Log Employ-
Hourly Weekly ment
Wage Hours Probit
State Study:
Insurance Code Amendment
Status—Replication: -.043 .049 -.018
(.023) (.022) (.018)
[27033] [27033] [62575]
Federal Study:
Insurance Code Amendment
Status—Replication: -.021 .001 -.006
(.012) (.010) (.009)
[131513] [131513] [190038]
Insurance Code Amendment
Status—using Smaller Group
of States: -.034 -.002 -.012
(.022) (.018) (.017)
[49984] [49984] [72270]
State Employer Mandate
Status—using Smaller Group
of States: -.024 -.010 -.023
(.020) (.016) (.015)
[49984] [49984] [72270]
Notes:

1. This is an exact replication of Gruber’s State Study results.

2. This replication contains one more observation than Gruber’s Federal Study; however the results found
in this replication are the same as Gruber finds.

3. The smaller group of states is comprised of states with an employer mandate, an employer mandate and
an insurance amendment, or neither type of legislation. States with only an insurance amendment are
dropped from the sample. The smaller sample of states allows for a direct comparison the results from
of Gruber’s insurance amendment classification with the employer mandate classification found in this
study.

87

.38: 2263 as .653: :an seesaw .3 .ES: .52 .88: <5: .38: use: season: 68258 See a 8:252 25. .N
datum £5 warn: @885 on 2 SEEM...— DEBaE 05 8:32 aux—EU .—

085 RE.
28m RE.

05: x — 0.5.«00 wean—5 Sufi—um:
Coo—25 E2585 Museums. Co 8.5
2.2: 283 3895 03:52
Oh _ \ _ Ouch—OD anowco ENE—m2

 

 

:22:me .8 238%

 

83:52 ego—oEm

RES can :25.
5253 336

23

m 63 _ roam

aha—B
©3795
mtg:

mt S:
vacuum

2mm

LIJZZZLIJLIJZ

Z

 

.55 see 2::
:833 £55m

:2:
2.23

025.

362m 630

 

 

£=oEe=oE< 0.50 8:83.:—

mcaam 88m mien—EU EP: 33% 83m 93 8:5 «noguacm .835 ”N 033—.

find

250

CE; 302
Each 302
SEE—«U 582
Bows—.0882
«SEE—
£65.:

36:02:50

88

.:0E::0:5 0:00 005.85 5 3:0 I

2.: 2 :05 :00:8:0 .0: 05:52 It

E: 2 :05 :00.8:0 :0: 05:52 I
:2: :05 :00.8:0 .0: 05:52 11
2.2: :05 :00:8:0 :0: 05:52 1

50:65.5 0:00 00508:: 5 2:0 It
055 RE. 1

520:: 2.55 05:52 I

E: 2 :05 :00:8:0 :0: 05:52 1
0.8m RE. 1

50:65.5 0:00 005.55 5 3:0 I

2
m

 

5.8.8.”: :8 553: 1.2 S
c0030: 0:55

 

85::102 5.5.35

55 .2006... a... :8: e350 .5 85: 5255:: .850. an 0.5...

 

wt 5.
0370.5
Ntatc
NC. :5
02m Cm
0370.8
NEQE
03—05
mg:—

NQSHR
2.2:

2&3:

:0L5:m 05G

mzzzmzzzmzmmmzmmmmmm

m

 

ES:
0.80: 2.5%

PC:—

ctm ::
2.22.
:2?

VEOQS
0Q: _
2.2:.
Shh

:055:m 059

 

3:0E::0E< 0:00 ogflms

0.2353 302
87.8002
33:02
8.08.2
£38552
0380:0852
5.52
0:52
«53:3
0.550.
«38

5......
ace...

2......
5%..
5.5.09
0:80—00
25250
50.50:...
382
seeps?

89

$3: :5303 :5 .6230: .500 0505.6 .m.D 22.0: mqmz .800: <5: .83: 00:0: 3505 00:08:80 E08 fl :0_5E:8:_ 0:2. .m
:05: 0:: m:_:0: :0:05:0 00 0: 5.5.050. 2805:. 0.: 00:370.. :0080 .—

5205. 0:55 05:52

520:: 0:55 05:52
::0.:::0:5 0:00 00:83:: 5 2:0
::0E::0.5 0:00 00550:: 5 2:0

50.0.5 355 05:52

05% me.
520:: 0:50.: 05:52

<0m 0:800 :5
2. b2 2 :05 05:55 5.83:5 05.0 :8 5.57304

05% 2e

2% .0:

 

:0_m.._0xm :8 :0508

 

mug—202m: uOMO-QEM

Z

LIJZLIJ

m

 

1.2 _ 2
:00350 35%

 

«to S. 2

I m

I m

mtg: m

I m

I m

I m

I Z

~3— m

I m

I Z

I m

I m

I m

I 2

2.3 2

I m

m .20. -05 Z

:0005:m 050 .E22

0:800 0055

2.2.5—

E22

0523

:0:05:m 050

 

8:0E::0E< 0:00 005:3:—

A:.::08 n 0308

 

”00:02

50:00.05»
32:8»?
255:; .83
53:25?
::0E:0>

55

50.0.:
080050.:
8850 5=0m
5:20.50 500m
:53 0:05—
580.530

2.5

59...: 582
2:26 552
3:0? 302
003.02 302

>050. 302

BIBLIOGRAPHY

BIBLIOGRAPHY

Commerce Clearing House. 1978. New 1978 Pregnancy Benefits and Discrimination Rules
with Explanation and State Survey. Chicago: Commerce Clearing House.

Gruber, Jonathan. 1992. “The Efficiency of a Group-Specific Mandated Benefit: Evidence
from Health Insurance Benefits for Maternity.” Working Paper No. 4157.
Cambridge, MA: National Bureau of Economic Research.

Gruber, Jonathan. 1994a. “The Incidence of Mandated Maternity Benefits.” American
Economic Review (June): 622-41.

Gruber, Jonathan. 1994b. “Payroll Taxation, Employer Mandates, and the Labor Market:
Theory, Evidence, and Unanswered Questions.” Prepared for the W.E. Upjohn
Institute Conference on Employee Benefits, Labor Costs and Labor Markets in
Canada and the U. S.

Health Insurance Association of America. New Group Health Insurance. Washington, DC:
Health Insurance Association of America, 1978.

National Employment Law Project, Inc. 1975. legal Developments in Sex Discrimination
Concerning Maternity Leaves and Benefits. New York: NELP.

United States Supreme Court. 1975/76. “Martha V. Gilbert, et al. versus General Electric
Company.” Law Reprints. Labor Series: the Supreme Court of the United States
Petition, Briefs and Opinion 9 (7). New York, Law Reprints.

Weyland, Ruth. 1978. “Pregnancy as a Disability.” in Proceedings of the Third ABA

National Institute on Equal Employment Opportunity Law. New Orleans: ABA
Press.

90

CONCLUSIONS

The increased reliance on mandates as a tool for providing benefits has prompted a
renewed interest in the labor market effects that these mandates have on the group that
receives them. By enacting a group-specific mandate, the government provides workers with
a socially desirable benefit that is not readily available in the absence of the mandate. This
type of government intervention is efficient when the workers covered by the mandate value
its benefits. Enacting this type of legislation, however, may have an adverse impact on
individual workers either when wage rigidities exist or when workers place a low value on
receiving the benefit. In this case, the outcomes are less efficient when compared to
provision of the same benefit though a lump sum tax.

Despite the government’s growing tendency to enact group-specific mandates, few
studies have attempted to isolate the labor market effects of these mandates. In a 1994 paper,
Gruber focuses on two group-specific maternity policies, the state insurance code
amendments and the federal PDA. These maternity policies required insurance contracts to
cover the medical costs of pregnant workers. His results indicate a substantial shifting of the
‘ cost of the mandate to the wages of those who benefit from it. In both of his studies, he finds
small disemployment effects. Together, these results indicate that state and federal maternity
policies were efficient—workers placed a high value on receiving the benefits of both.

When reviewing the history of pregnancy discrimination legislation, however, it
quickly becomes apparent that the insurance code amendment was not the sole determinant

91

92

of a state’s maternity policy during this period. During the 19708 and early 19803, three
major events had the potential to impact on the maternity policies of firms. The first was the
EEOC’s 1972 pregnancy guidelines, which changed the focus of the EEOC’s discrimination
policy to include acquired worker characteristics, in addition to innate qualities such as race
or sex. Following the EEOC’s lead, a limited number of state FEP agencies adopted these
guidelines into their own state employer mandates. These employer mandates required firms
to cover pregnancy in the same manner as other temporary disabilities. At that time, a
number of states also amended their insurance codes to require these policies to cover
complications that arise from pregnancy.

In 1976, two court decisions clearly separated the states with and the states without
an active employer mandate. The Gilbert (1976) decision redefined the EEOC’s guidelines
to be ‘recommendations,’ while the Brooklyn (1976) decision upheld the individual state
guidelines. In 1978, congressional enactment of the federal PDA ended this division by
creating a minimum standard—treating pregnancy in the same manner as other temporary
disabilities—to which all employer-sponsored health insurance plans must adhere. This
division among state policies and the subsequent federal law created a unique opportunity to
investigate the labor market effects of the group-specific mandate.

Using the full set of states eligible to be included, chapter 2 presents the estimates of
the labor market effects of the state employer mandates and the federal PDA. In the state
study, the wage declines by about 4 percent with no change in net labor input. Overall, the
costs of the mandates in the state study are substantially shifted to the wages of those who
benefit, indicating an efficient outcome. In contrast to the findings presented in the state
study, the evidence in the federal study indicates that strong disemployment effects resulted

from the passage of the PDA, with no corresponding decline in the wage. This is consistent

93
either with workers placing a low valuation on the benefit or with significant barriers to wage
adjustment.

Variations in the expected cost and the expected benefit of the maternity mandate
across states may cause the apparent conflict between the state study and the federal study
results. As a result of these variations, states adopt only the level of state maternity
legislation that is considered efficient. In states that adopted a mandate before the federal
PDA, the perceived value assigned by society is larger than the cost to the individual firm of
implementing the mandate. Since it was inefficient, the experimental states in the federal
study did not implement a state mandate before the enactment of the federal legislation.

In addition, chapter 3 reconsiders the results of Gruber’s federal study on insurance
code amendments using states that enacted employer mandates or states that enacted neither
maternity policy. The decline in the wage (2.4 percent) is nearly the same as Gruber’s
estimate (2.1 percent), but it is less significant in the new study. Apart from the decline in
the wage, however, the coefficient on the probability of employment suggests a decline of
about 2 percentage points. This estimate is significant, although at less than the 10 percent
level. Thus, changing the state status to reflect the employer mandate classifications reveals
weak evidence of both wage and employment effects. These results support the chapter 2
hypothesis that the effects of the PDA may not be limited to a decline in the wage.
Unfortunately, I was unable to re-estimate Gruber's state study results using the employer
mandate states because too few states remained.

Although the federal study results that are found in chapter 2 suggest that enacting
the PDA was inefficient in states without prior maternity legislation, these findings are silent
on the issue of equity. Long-term benefits may include an increased participation in and an

attachment to the labor force by females who previously left the labor force to have a child.

94

Future gains would be realized if females who planned to have children attained additional
education or training with the intention of beginning a career in a more-skilled field that
provided maternity benefits. Aside from the gains to female workers, however, society
would benefit if healthier babies resulted from the additional health insurance coverage.
This may occur through better pre-natal care, longer hospital stays, or the use of increased
technology. The extent to which states that did not adopt a mandate before the PDA
considered these long-term benefits is unclear and requires further analysis.

As well as increasing the equity between demographic groups, the PDA may have
decreased any costs associated with legislative inconsistency between states and government
agencies. The high information costs associated with multiple legal statements and
guidelines decrease with the federal mandate, which sets a minimum standard of compliance.
Although EEOC complaints may rise due to the increased information on the part of the
workers, related court costs may decrease if more disputes are settled without legal
intervention. While these issues are not addressed in this paper, they have further
implications for the cost/benefit analysis that states may perform when deciding whether to a
mandate particular benefits.

Before deciding whether to implement a group-specific mandate, policy makers
should recognize these potential long-term gains, as well as the possibility of disemployment
effects. This is particularly true at the federal level, where the total effects of the mandate
may be distributed unequally over various states. In part, the failure of a state to adopt a
mandate before the enactment of a federal policy may be due to a low valuation by those who
benefit from the group-specific mandate rather than a market failure. Depending on the
value placed on the mandate by the group to benefit, the state-specific effects of that mandate

may range from a decrease in the wage equal to the cost of the benefit to significant

95
disemployment effects. Careful analysis of the labor market effects of group-specific
mandates, as well as other potential long-term benefits not immediately present in the labor

market, will provide a framework for policy makers to base such decisions in the future.

APPENDIX A

Appendix A: Dates of Compliance by Maternity Policy

Dates of Compliance1

Insurance Employer Temporary Disability
Code Amendment Mandate Insurance

Alabama — — —
Alaska — 6/4/75 —
Arizona 6/1 3/77 — —
Arkansas 2/1/76 — —
California 7/1/76 9/22/76 1979
Colorado 1/ 1/76 5/31/72 —
Connecticut — 10/1/73 —
Delaware — — —
Florida 1/1/78 — —
Georgia 7/1/77 — —
Hawaii 12/30/74 — 5/7 3
Idaho 7/1/77 — ——
Illinois 7/1/76 1/ 16/73 —
Indiana — Pre- l 9762 —
Iowa 4/13/76 10/9/72 —
Kansas 2/15/77 1/1/74 —
Kentucky — Pre- 1 9762 —
Louisiana —— — —
Maine — 5/15/76 —
Maryland 7/1/75 7/1 1/7 2 —
Massachusetts — 6/19/72 —
Michigan — 9/72 —
Minnesota 6/3/77 3/7 1 —
Mississippi — — —
Missouri — Pre-l9762 —
Montana — 7/1/75 —
Nebraska — — —
Nevada 9/ 1 1/7 7 — —
New Hampshire 1/1/77 Pie-19762 —
New Jersey 9/ 1/7 6 — 1948
New Mexico — — —
New York 1/ 1/77 9/73 7/76
North Carolina — — —
North Dakota — — —
Ohio — — —
Oklahoma — —— —
Oregon 10/4/77 6/71 —
Pennsylvania 10/26/‘77 12/71 —
Rhode Island 10/4/78 — 5/7/75
South Carolina — — —
South Dakota — 1972 —
Tennessee 6/ 1 /7 6 — —

96

97

Appendix A (cont.)

Dates of Compliance1

 

Insurance Employer Temporary Disability

Code Amendment Mandate Insurance
Texas 1/26/77 —- —
Utah — — —
Vermont — — —
Virginia 7/1/78 — —
Washington 3/1/77 10/3/73 —
West Virginia — — —
Wisconsin 6/1/76 7/10/72 —
Wyoming — — —
PM;

1. This information is from Commerce Clearing House (1978), HIAA (1990), NELP (1975), U.S. Supreme
Court (1975/76), and Weyland (1978).

2. The exact dates of compliance in these employer mandates states are unknown.

APPENDD( B

Appendix B: Mandate States and Coverage for State Study

State Groupings

New England
Connecticut

Group: Maine, New Hampshire,
& Vermont

Massachusetts
Rhode Island

Mid Atlantic
New Jersey
New York
Pennsylvania

East North Central

Illinois

Indiana

Ohio

Group: Michigan & Wisconsin
West North

Group: Minnesota, Iowa, Missouri,

North Dakota, South Dakota,
Nebraska & Kansas

South Atlantic
Group: Delaware, Maryland,
Virginia & West Virginia

District of Columbia

Florida

Group: South Carolina & Georgia
North Carolina

East South Central

Group: Kentucky & Tennessee
Group: Mississippi & Alabama
West South Central

Group: Arkansas, Louisiana,
& Oklahoma

Texas

Mountain

Group: Arizona, Colorado, Idaho,
Montana, Nevada, New Mexico,
Utah & Wyoming

Pacific

California

Hawaii

Group: Alaska, Oregon, & Washington

Notes:

Included in Experiment

Yes

No
Yes
No - TDI State

No - TDI State
No - 'IDI State
Yes

Yes
N 0- Law Unclear
Yes
Yes

No

No

No - Laws Unclear

No - Insurance code amendment
No - Insurance code amendment
Yes

No
Yes

No

No - Insurance code amendment

No - TDI
No - TDI
No

Mandate Coverage

All

No Religious, Education

All

N 0 Religious, Education

None

Not Religious

None
None

None

None

None

1. The information in this table is primarily from Commerce Clearing House (1978), NELP (1975), U.S.
Supreme Court (1975/76), and Weyland (1978).

98

APPENDIX C

Alabama
Alaska
Arizona
Arkansas
California
Colorado
Connecticut
Delaware
Florida
Georgia
Hawaii

Idaho

Illinois
Indiana

Iowa

Kansas
Kentucky
Louisiana
Maine
Maryland
Massachusetts
Michigan
Minnesota
Mississippi
Missouri
Montana
Nebraska
Nevada

New Hampshire
New Jersey
New Mexico
New York
North Carolina
North Dakota
Ohio
Oklahoma
Oregon
Pennsylvania
Rhode Island
South Carolina

Appendix C: PDA States and Coverage for Federal Study

Included in Exxriment
Yes

Yes

No - Insurance code amendment
No - Insurance code amendment
No - TDI State

N o - Change in Mandate Statusl
Yes

No - Letter to Ruth Weyland2
No - Insurance code amendment
No - Insurance code amendment
N o - TDI State

No - Insurance code amendment
Yes

No - Extent of Mandate Unclear
Yes

Yes

No - Change in Mandate Statusl
Yes

No - Change in Mandate Status1
Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

N o - Insurance code amendment
No - Change in Mandate Status1
N o - TDI State

Yes

No - TDI State

Yes

Yes

N o - Legislation Enacted

No - Change in Mandate Status1
No - Change in Mandate Statusl
Yes

No - TDI State

No - Evidence of Previous Mandate3

99

Affected by PDA
All

None

None

Religious & Education

None
Religious

All

All
Religious
All

All

All
Religious
None

All

Not Education

All
All

None

South Dakota
Tennessee
Texas

Utah
Vermont
Virginia
Washington
West Virginia
Wisconsin
Wyoming

Notes:

1. These states had employer mandates in place previous to the 1976 Supreme Court Ruling in Gilbert.
However, there is no record that these states either enforced or strengthen the existing employer

100

Appendix C. (cont’d)

Included in Experiment
Yes

No - Insurance code amendment
No - Insurance code amendment
No - Letter to Ruth Weyland2
Yes

No - Insurance code amendment
Yes

Yes

Yes

No - Letter to Ruth Weyland2

Affected by PDA
None

All
All

Religious & Education
All
Religious

guidelines after 1976. Therefore, these states were dropped from the sample.

2. Commissioners in these states informed Ruth Weyland, that their intent was to uphold the EEOC 1972

guidelines.

3. One source lists South Carolina as having the maternity mandates in 1976. No other source confirms

this, so I do not include South Carolina in the sample.

4. The information in this table is primarily from Commerce Clearing House (1978), NELP (1975), U.S.

Supreme Court (1975/76), and Weyland (1978).

APPENDD( D

Appendix D: State Study Small Sample, Title VII Workers

(1) (2) (3)
Log Log Employ-
Hourly Weekly ment
Wage Hours Probit
Education: .059 -.012 .014
(.002) (.003) (.002)
Experience: .017 -.036 -.083
(.002) (.003) (.006)
Experience -.480 l .275 l .323
Squared: (.097) (.133) (. 106)
Non-White: -.O39 .089 .066
(.013) (.018) (.012)
Union .172 .085
Coverage: (.012) (.016)
Covered Group: .272 .738
(.159) (.219)
Mandate Effect: -.038 -.027 .005
(.020) (.027) (.015)
Status of Variables:3
State: Individual Individual Individual
Covered*State: Individual Individual N/A
State*Year: Group Group Group
Adjusted R2: .303 .188 .018
N: 75 84 75 84 22778
Notes:

1. The numbers in parenthesis are standard errors.
Regression also includes 15 industry dummy variables.

The term ‘Individual’ indicates that the state, year, and covered group variables are entered individually.
The term ‘N/A’ indicates that the variables are not used in this specification.

The variables included in the probit are age and age squared, not experience and its square.

Respondents who report self-employment, government employees (except state and local
administration), and those with a wage less than $1 per hour or greater than $100 per hour are dropped
from the sample.

6. All workers who report working in industries exempt from Title VII regulations or from the state
employer mandates are in the non-covered group.

101

9’!"

5"?

BIBLIOGRAPHY

BIBLIOGRAPHY

Commerce Clearing House. 1978. New 1978 Pregnancy Benefits and Discrimination Rules
with Explanation and State Survey. Chicago: Commerce Clearing House.

Health Insurance Association of America. New Group Health Insurance. Washington, DC:
Health Insurance Association of America, 1978.

National Employment Law Project, Inc. 1975. Legal Developments in Sex Discrimination
Concerning Maternity Leaves and Benefits. New York: NELP.

United States Supreme Court. 1975/76. “Martha V. Gilbert, et a1. versus General Electric
Company.” Law Reprints. Labor Series: The Supreme Court of the United States
Petition, Briefs, and Opinion 9 (7). New York, Law Reprints.

Weyland, Ruth. 1978. “Pregnancy as a Disability.” In Proceedings of the Third ABA

National Institute an Equal Employment Opportunity Law. New Orleans: ABA
Press.

102

MICHIGAN STATE UNIV. LIBRARIES
ll1WIHIIII!NIHUHIHI441111!erlIIWHlHrIIHHHI
31293017129069