EMPLOYMENT PROTECTIONS FOR OLDER WORKERS:DO DISABILITY DISCRIMINATION LAWS MATTER?

WENDY A. STOCK and KATHLEEN BEEGLE*

Although age discrimination laws boost employment of older workers, some areexempt from age-law protection and may be better protected by disabilitydiscrimination laws. Disability laws may raise the costs of employing the disabled,however, generating mixed predictions regarding employment effects of protection byboth types of laws. This study finds that employment of older disabled workers with thecoupled laws is lower than those with age-law protection only; employment of theyounger disabled is higher with the coupled laws. For older disabled workers,earnings of those with the coupled laws are lower than their counterparts with onlyage-law protection. (JEL J78, K31, J00)

I. INTRODUCTION

The Age Discrimination in EmploymentAct (ADEA) was enacted in 1967 with the pri-mary goals `̀ to promote the employment ofolder persons based on their ability ratherthan age; to prohibit arbitrary age discrimina-tion in employment; to help employers andworkers find ways of meeting problems arisingfromthe impactofageonemployment'' (ADEA,1967: 29 U.S.C. x 621 (b)). It is the subject ofapproximately 16,000 annual charge receiptswith the Equal Employment Opportunity Com-mission (EEOC), as well as substantial courtlitigation (EEOC, 2001).1 The limited empiricalresearch on the impacts of the ADEAfinds that it has positive effects on the employ-ment of protected workers (Neumark andStock, 1999; Adams, 2000). However, legal

scholars have raised questions about the abilityof the ADEA to protect older workers fromemploymentdiscriminationandpositthatsomeolder workers may instead be better protectedunder the Americans with Disabilities Act(ADA) (Frier, 1993; Posner, 1995).

Frier (1993) notes that gaps in coverage ofthe ADEA based on bona fide occupationalqualifications (BFOQs), and exceptions madefor factors other than age exempt some olderworkers from protection under the ADEA (forexample, an employer does not violate theADEA if he discharges an employee on thebasisofanage-relateddisability).Theseworkersmay be eligible for protection under disabilitydiscrimination laws, however, because someof the characteristics that often accompanyage (e.g., frailty, mild cognitive impairment,or failing hearing or vision) may fall underthe provisions of disability discrimination leg-islation that does not include a BFOQ excep-tion but instead applies to any qualifiedindividual with a disability. In this sense,Frier argues, the ADA makes a BFOQ defenseunder the ADEA more difficult because some

ABBREVIATIONS

ADEA: Age Discrimination in Employment Act

ADA: Americans with Disabilities Act

BFOQ: Bona Fide Occupational Qualifications

EEOC: Equal Employment Opportunity

Commission

SIPP: Survey of Income and Program

Participation

*The authors thank David Neumark, Richard Alston,seminar participants at Montana State University, andanonymous referees for helpful comments and suggestionsand David Rumpel for research assistance. The authorsgratefully acknowledge funding support from NIA grantnumber R03 AG16788-01. These are the views of theauthors, and need not reflect those of the World Bank orany affiliated organization.

Stock: Associate Professor of Economics, Department ofAgricultural Economics and Economics, MontanaState University, Bozeman, MT 59717-2920. Phone1-406-994-7984, Fax 1-406-994-4838, E-mailwstock@montana.edu

Beegle: Economist, World Bank, 1818 H Street NW,Washington, DC 20009. Phone 1-202-458-9894, Fax1-202-522-1153, E-mail kbeegle@worldbank.org

1. Receipts include all charges filed under the ADEA aswell as those filed concurrently under Title VII of the CivilRights Act, the Americans with Disabilities Act, and/or theEqual Pay Act of 1963.

111

Contemporary Economic Policy(ISSN 1074-3529) DOI: 10.1093/cep/byh009Vol. 22, No. 1, January 2004, 111±126 # Western Economic Association International

age-related employment problems could besolved by reasonable accommodations.

Although age in itself is not considered adisability under the ADA, the language ofthe act clearly acknowledges that disabilityand age are correlated, noting, `̀ some43,000,000 Americans have one or morephysical or mental disabilities, and this numberis increasing as the population as a whole isgrowing older'' (ADA, 1990: 42 U.S.C. x 12101(2) (a) (1)). Indeed, as Figure 1 shows, thepositive relationship between age and disabilitystatus is quite striking, hovering below 10% forthose under age 40, but rising fairly dramati-cally starting around age 45, and affectingone-quarter of the population by age 60 andone-third of the population by age 65. As notedby Posner (1995, p. 339), `̀ The interactionbetween [the ADA and ADEA] has yet to beexplored by the courts . . .. The ADA may suc-ceed in helping older workers where the ADEAhas been ineffective.'' The interaction betweenthese two types of laws has yet to be explored byempirical researchers as well. Given Frier'sargument, whether the protective effect ofage discrimination laws is affected when agelaws are coupled with disability discriminationlaws is of particular interest. In addition,because prohibitions against discrimination

in firing and reasonable accommodationrequirements in disability discrimination lawsraise the cost of employing the disabled,whether the coupling of the laws is associatedwith net increases in employment or earnings ofprotected workers is also of interest.

The purpose of this article is to investigate(1) whether there are interaction effectsbetween age discrimination and disability dis-crimination laws in terms of labor market out-comes and (2) whether the coupling of the lawsis associated with net employment or earningschanges, relative to age-law protection alone.An investigation of the ADEA and ADA issomewhat problematic, however, becausethese policies have near-universal coverage fortheir protected groups, making it difficult toidentify a comparison group that can be usedto control for changes in relative outcomes thatare unrelated to the legislation. However, at thetime these federal laws were enacted, similarlegislation already existed in many states. Toapproximate an experimental framework, thisstudy exploits variation in state-level prohibi-tions on age and disability discrimination thatwere passed prior to the federal laws.

Combining data from the U.S. Censuses ofPopulation with information on federal andstate-level discrimination laws by year, this

FIGURE 1

Source: Authors' tabulations based on the 1970±90 U.S. Censuses of Population. For those age 65±70, the percentreporting a disability limiting work is based only on the 1980±90 censuses, as the disability question was not askedof this age group in the 1970 census.

112 CONTEMPORARY ECONOMIC POLICY

study finds mixed results regarding the coup-ling of age and disability discriminationlaws. On one hand, the positive employmenteffect of protection by an age discriminationlaw is larger among individuals in states withdisability discrimination laws. This positiveinteraction effect on employment is concen-trated among nondisabled workers 40±64years old and is negative for older individualswho classify themselves as disabled. On theother hand, however, disability discriminationlaws are associated with slightly lower employ-ment among the nondisabled, so that thecoupling of protection by an age discriminationlaw and the presence of a disability discrimina-tion law in the state is not associated with adifference in the level of employment amongthe nondisabled and is associated with loweremployment for older disabled workers, bothrelative to protection by an age discriminationlaw alone. Only among disabled workers underage 40 is the coupling of the laws associatedwith higher employment levels relative toage-law protection alone.

The earnings regressions indicate that forworkers over 40 years old, the positive earningseffect of protection by an age discrimination lawis largerwhenage-lawprotection iscoupledwitha disability discrimination law. However, formost groups this positive interaction effect isoffset by the negative effect of disability discri-mination laws on earnings in general, so that netearnings are not significantly different betweenthose with the coupled laws and those protectedonly by an age discrimination law. Only forolder disabled workers are earnings levels sig-nificantly different with the coupled laws: olderdisabled workers who are protected by age dis-crimination laws and have a disability discrimi-nation law in their state have lower earningsthan older disabled workers having only age-law protection.

II. AGE DISCRIMINATION AND DISABILITYDISCRIMINATION LAWS

A. Age Discrimination Laws

As noted in the introduction, the first federallegislation protecting workers based on age isthe 1967 ADEA.2 At the time of passage of the

ADEA, however, several states had alreadyenacted their own age discrimination laws. InTable 1, state and federal age discriminationlegislation is summarized starting with lawsin place by 1960 (the earliest year covered bythe data used in the empirical analysis) andincluding laws in place by 1990. The top rowof the table describes the protection offered atthe federal level by the ADEA. Thus, for ex-ample, by 1960, Colorado had an age discrimi-nation law protecting all workers betweenthe ages of 18 and 60 years old, and eight otherstates had similar laws protecting workers ofvarious ages, most commonly those between 40and 65 years old. By 1970, the ADEA coveredall those ages 40±65, and 25 states also hadtheir own legislation, some of which protecteda wider age range than that protected by theADEA (e.g., Colorado, Alaska, and Hawaii).In 1978, the ADEA was amended to coverthose age 40±70 years old, so all those age65±70 also became covered by the federal lawby 1980. Of course, those in the age rangesspecified by broader state laws in place byeach census year are also covered.

Information on the enforcement of age dis-crimination laws is also included in Table 1,letting the variable ADWE denote laws thatwere passed with either weak or no enforce-ment authority (e.g., Georgia's law andColorado's law before 1990). In addition, theADEA in 1970 is denoted as having weakenforcement,becausetheDepartmentofLaborwas initially designated to administer the lawand relatively few resources were devoted toenforcing it. With the 1978 amendment to theADEA, administration of the law was trans-ferred to the EEOC and more resources wereavailable for its enforcement (Stone, 1980).

The theoretical effects of age discriminationlaws on the employment of older individualshave been discussed by Lazear (1979) andNeumark and Stock (1999). The laws prohibitdiscrimination in hiring and discharge ofprotected individuals. However, becauseolder workers have relatively low turnover

2. It is useful to provide a definition of what it means tobe protected by an age discrimination law. The wording ofstate laws varies, but most are similar in language to theADEA, which states, `̀ It shall be unlawful for an employer

to fail or refuse to hire or to discharge any individual orotherwise discriminate against any individual with respectto his compensation, terms, conditions, or privileges ofemployment, because of such individual's age'' (ADEAx623(4)). Thus someone who is in the age group protectedby the law could sue if hiring, firing, or conditions of hisor her employment were based on age. Neumark (2003)includes a discussion of specific case law interpretationsof the employment protections offered by the ADEA.

STOCK & BEEGLE: DISABILITY EMPLOYMENT PROTECTION 113

TABLE 1

State Age Discrimination and Disability Discrimination Laws

Ages Protected by Age Discrimination Laws (PAD = 1) Disability Discrimination Laws (DDL = 1)

State 1960a 1970 1980 1990 1980 1990

US 40±65, ADWEb 40±70 �40

AL

AK >45 >45 >45 x x

AZ 40±70 x, RAc

AR x

CA 40±64 40±64 40±64 x x, RA

CO 18±60, ADWE 18±60, ADWE 18±60, ADWE 40±70 x, RA

CT 40±65 40±65 �40 �40 x x

DE 45±65 40±70 40±70 x, RA

DC 18±65 18±65 x x, RA

FL All ages All ages x x

GA 40±65, ADWE 40±70, ADWE x

HI All ages All ages All ages x x

ID <60 <60 >40 x, RA

IL >45 >45 >40 x x, RA

IN 40±65 x x

IA >18 >18 x x

KS >18 x x

KY 40±65 40±70 x x

LA <50 <50 40±70 x, RA

ME All ages All ages All ages x x, RA

MD 45±65 45±65 45±65 x, RA x, RA

MA 45±65 40±65 40±65 >40 x x

MI 35±60 >18 >18 x x, RA

MN >18 >18 x x, RA

MS

MO 40±70 x, RA

MT All ages All ages x x, RA

NE 40±65 40±70 40±70 x x

NV All ages >40 x x

NH All ages All ages x x

NJ >21 >21 21±70 x x

NM All ages All ages All ages x x, RA

NY 45±65 40±65 18±65 >18 x x

NC All ages All ages x, RA

ND 40±65, ADWE 40±65, ADWE 40±70 x, RA

OH 40±65 40±70 >40 x, RA x, RA

OK 40±70 x

OR 25±65 25±65 18±65 18±70 x x, RA

PA 40±62 40±62 40±62 40±70 x, RA x, RA

RI 45±65 45±65 40±70 40±70 x, RA

SC 40±70 >40 x

SD x, RA

TN >40 x x

TX >40 x x, RA

UT >40 40±70 x x

VT >18 x x, RA

VA x x, RA

WA 40±65 40±65 40±70 x, RA x, RA

continued

114 CONTEMPORARY ECONOMIC POLICY

and job mobility, they are more likely to becurrently employed than at the hiring stage(Jaeger and Stevens, 1999). On net, theprohibition against discrimination in dis-charges implies increased employment of olderworkers because of a reduction in involuntaryterminations.

There has been surprisingly scant researchon the empirical impacts of age discriminationlaws on labor market outcomes. The evid-ence includes studies by Neumark and Stock(1999) using U.S. Census data and Adams(2000) using Current Population Survey data.Both studies find that age discrimination pro-hibitions boost the employment of workers age60 and over. Adams also finds that age discri-mination prohibitions have negative effectson the probability that a worker age 65 orolder is a new hire, which he attributes to areduction in retirement for this age groupbecause of the ADEA. In a study that focuseson the ADEA's elimination of mandatoryretirement, Ashenfelter and Card (2000) findthat the elimination of mandatory retirementfor professors was also associated with reduc-tions in retirement.

In terms of the effects of these laws on olderworkers' earnings, the evidence is scarcer.Neumark and Stock (1999) do not focus onthis question per se, but their estimates indicatea steepening of age-earnings profiles associatedwith age discrimination laws, which is consis-tent with the laws positively impacting theearnings of older workers.

B. Disability Discrimination Laws

As with the ADEA, several states passeddisability discrimination legislation protecting

private-sector workers prior to the passage ofthe ADA. Beegle and Stock (2003) present acomprehensive review of these state-level dis-ability discrimination laws protecting private-sector workers, and in Table 1, their summaryof this legislation is repeated. Because no statehad a disability discrimination law covering theprivate sector by 1970, Table 1 only includesdisability discrimination laws in place by 1980and 1990. In addition, coverage based on theADA is not included in the table because datafrom the 2000 census is not yet available toassess the impact of this federal law usingthis empirical framework. Instead, the focusis on the effects of state disability discrimina-tion laws.

The variable DDL is a dummy variableequal to one for states that had any disabilitydiscrimination law covering workers in theprivate sector as of each census decade. Thevariable RA is a dummy variable equal toone for states for which the discriminationlaw included a clause requiring employers tomake reasonable accommodation of workerdisabilities (for example, altering facilities orchanging the pace of work). Although theselaws varied along other dimensions (e.g., sizeof employers covered), these differences arenot the focus here, in part because censusdata do not provide sufficiently detailed infor-mation with which to explore the nuances of thelaws. Instead, the average effects of the laws areestimated.

Acemoglu and Angrist (2001) discuss thetheoretical impacts of laws designed to protectdisabled workers. Prohibitions against discri-mination in hiring effectively provide a subsidyto hiring the disabled, because not hiring a dis-abled worker increases the risk of a costly

TABLE 1

continued

Ages Protected by Age Discrimination Laws (PAD = 1) Disability Discrimination Laws (DDL = 1)

State 1960a 1970 1980 1990 1980 1990

WV 40±65 >40 x, RA

WI 40±65 40±65 40±65 >40 x x, RA

WY 40±70 x, RA

aFor the law coding, year refers to laws in place as of the last year prior to the census. For example `̀ 1960'' refers to alllaws in place as of 1959, the year to which the 1960 census earnings questions refer.

bADWE indicates an age discrimination law with weak enforcement authority.cRA indicates states with a reasonable accommodation clause included in the disability discrimination legislation.

Sources: Neumark and Stock (1999), Beegle and Stock (2003), and Neumark (2003).

STOCK & BEEGLE: DISABILITY EMPLOYMENT PROTECTION 115

lawsuit for the firm. Alternatively, prohibitionsagainst discrimination in firing raise the costs ofhiring disabled workers through an increasedrisk of a lawsuit associated with firing a disabledworker. Empirically, most lawsuits broughtunder the ADA have been for wrongful termi-nation rather than for discrimination in hiring(DeLeire, 2000; Mudrick, 1997), implying thatthe firm's perceived costs of firing would behigher than for not hiring.3 Coupled withreasonable accommodation requirementsmaking it more costly to hire disabled workers,the net result would be a reduction in hiringof the disabled. Of course, these negativeemployment effects may be somewhat offsetif reasonable accommodation requirements(or other favorable labor market conditionsfor disabled workers) induce an increase inthe supply of disabled workers.

Empirical studies that attempt to evaluatethe impact of the ADA draw varied conclu-sions. Using data from the 1988±97 CurrentPopulation Survey, Acemoglu and Angrist(2001) find a negative impact of the ADA onthe employment of male disabled persons andfemale disabled persons under 40 years oldbut no effect on the wages of disabled workers.DeLeire (2000) uses data from the Survey ofIncome and Program Participation (SIPP)from 1986 to 1995 and concludes that employ-ment rates of men with disabilities fell due to theADA, but he finds no significant effect of theADA on relative wages. DeLeire interpretsthese results as implying that the ADA,through its requirement for reasonable accom-modation, increased the cost of employingdisabled workers. Alternatively, Kruse andSchur (2003) also use the SIPP and find thatdisabled employment rates increased with theADA when an alternative definition of disabil-ity is used. Bound and Waidmann (2002) pre-sent time-series and cross-state evidence thatthe relative decline in employment among peo-ple with disabilities was associated with theexpansion in the Disability Insurance programin the 1990s and not the ADA.

Beegle and Stock (2003) take a differentapproach by investigating the impact of state-level disability discrimination laws. They findthat disability discrimination laws are associ-ated with lower relative earnings for the dis-abled, but not lower employment rates, oncepreexisting employment trends among the dis-abled are taken into consideration.

C. Coupled Age and DisabilityDiscrimination Legislation

In sum, the studies cited on age discrimina-tion laws find that these laws tend to boostemployment of protected workers, while theevidence on the impacts of disability discri-mination laws is less conclusive.4 Althoughresearchers have examined the effects of ageand disability discrimination laws separately,no one has yet investigated whether disabilitydiscrimination legislation coupled with agediscrimination legislation offers additionalprotection or is associated with lower employ-ment or earnings for covered workers.5

As noted in the introduction, Frier (1993,p. 178) argues that the BFOQ exception to theADEA allows a firm to use age as a criteria inits employment decisions if the firm can`̀ show that it has reasonable cause to believe

3. Similarly, a higher proportion of charges broughtunder the ADEA have been for wrongful discharge than fordiscrimination in hiring (Posner, 1995), with the implica-tion that the ADEA would also imply reduced hiring ofolder workers. However, because older individuals aremore likely to be currently employed than at the hiringstage, the prohibition of discrimination in dischargesimplies increased employment via a reduction in involun-tary terminations for them.

4. Although the perception of discrimination moti-vated these laws, it is difficult to identify the extent towhich age or disability discrimination exists (or existed)in the labor market. There are bodies of literature assessingcourt cases filed under the ADEA (e.g., Schuster andMiller, 1984) and measuring the extent of discriminationagainst older workers (e.g., Hutchens, 1988; Hirsch et al.,2000). Regarding disability discrimination, some studiesfind that discrimination accounts for between 30% and50% of the wage differential between disabled and nondis-abled workers (Johnson and Lambrinos, 1985; Baldwinand Johnson, 1994, 1995). On the other hand, DeLeire(2001) finds that only 3.7% of the earnings differential isdue to discrimination against disabled workers.

5. Note that the presence of dual laws may not be arandom event. Beegle and Stock (2003) found noevidence that disability discrimination laws were system-atically related to other state characteristics (e.g., the pre-valence of disabled in the state). Holbrook and Percy (1992)found that variables traditionally successful at explainingother types of state policies were generally much less suc-cessful at explaining the status of disability rights lawsand that an explanation of differences in the laws acrossstates requires other, less orthodox approaches, such asunderstanding the political dynamics in each state. In thepresent study, probit estimates (available from the authors)indicated that state characteristics, such as the age distribu-tion, the percent disabled, and the disabled-nondisabledemployment gap in the state, were not significantlyassociated with the probability that a state had bothtypes of laws.

116 CONTEMPORARY ECONOMIC POLICY

that all or substantially all of a class of appli-cants would be unable to perform a job safelyand efficiently'' and that `̀ the BFOQ is rea-sonably necessary to the normal operation ofthe business.'' Disability discrimination laws,however, make establishing a BFOQ defenseunder the ADEA more difficult because someage-related employment problems could besolved by reasonable accommodation. As aresult, the effectiveness of age discriminationlaws may be altered in the presence of dis-ability discrimination laws. At the same time,because reasonable accommodation provi-sions and prohibitions against firing in dis-ability discrimination laws make employingthe disabled more costly, the coupling oflaws could also generate negative employ-ment changes for covered workers. Thus, itis of interest to investigate the effects of thetwo types of laws together, in part to testwhether the impact of age laws is differentin the presence of disability laws and alsoto estimate net employment and earningschanges associated with the coupled lawsrelative to age discrimination protection inisolation.

III. DATA AND EMPIRICAL FRAMEWORK

The empirical analysis is designed to exam-ine whether the access to disability discrimina-tion legislation offers additional employmentand earnings protections to those availablethrough age discrimination legislation alone.Data from the 1960±90 U.S. Censuses of Popu-lation are merged with the variables coded inTable 1 describing age and disability discrimi-nation laws by state and year. The 1% publicuse microdata samples are used for 1960 and1970 (using the 5% questionnaire sample for1970); for 1980 and 1990, 1% subsamples areextracted from the A samples, which identify allstates.6 The sample contains white males age18±70, excluding nonwhites and females toavoid confounding the effects of age discrimi-nation and disability discrimination lawswith those of race and sex discrimination laws.The sample excludes self-employed, unpaid,agricultural, and public sector workers (includ-

ing those in the armed forces). Also excludedare those with missing data (for example, thosewho reported being employed but did notreport their industry or earnings). The earningsanalysis uses the subsample of full-time work-ers (those working 27 or more weeks per yearand 30 or more hours per week). Individuals areconsidered employed if any earnings areobserved for the previous calendar year.Finally, the earnings variable is the log ofannual earnings, and weeks and hours workedare regressors.7

The simplest approach to assessing theimpact of coupled age and disability discrimi-nation laws is to use the 1960±90 census dataand the coverage of the ADEA starting in1968. Thus, those in age ranges covered bytheir respective state laws are protected by anage discrimination law in 1960. All 40±65-year-olds are protected by the federal agediscrimination law in 1970, and all 40±70-year-olds are protected by the ADEA by1980. Younger individuals in states withlaws that cover younger ages than the federallaw are also protected (e.g., Colorado before1990 and Florida starting in 1980). Thisstrategy allows one to identify two separatetreatment groups: those in states that wentfrom having neither age nor disability discri-mination laws to having only age discrimina-tion laws (e.g., Alaska and California from1960 to 1970, Arizona and Arkansas from1960 to 1980) and those in states that wentfrom having neither age nor disability discri-mination laws to having both typesof laws (e.g., Florida and Iowa from 1970 to1980). The outcomes for these two groups canthen be compared to outcomes for similarindividuals in states where the laws did notchange (e.g., Colorado from 1960 to 1980).This implies an equation of the followingform:

Yist � a� Xitb� �PADist �DDLst�g�1�� PADistq�DDLstj� Ssbs

� Ttbt � �Ait � Tt�bAt � eist,

6. The 1970 census used two different long-form ques-tionnaires, one collected from a 15% sample and the othercollected from a 5% sample of the population. The 1%samples are drawn from those in the 5% questionnairesample.

7. The census weeks of work variable refers to theweeks worked in the previous calendar year, whereashours of work refers to usual weekly hours in the censusyear. These variables are coded in the census data as cate-gorical variables, so the midpoints of the categories areused.

STOCK & BEEGLE: DISABILITY EMPLOYMENT PROTECTION 117

where Y represents an outcome (employmentor log earnings) of individual i in state s andyear t. The vector X is a set of standard demo-graphic controls (age and its square, residencein a standard metropolitan statistical area,marital status, and education status). Thedummy variable PAD is equal to one forthose protected by an age discriminationlaw. Thus, PAD represents the combinationof the presence of either a state or federal agediscrimination law and the individual beingin the age range protected by the law. PAD isequal to one for those in the age ranges pro-tected by their respective state's law in 1960,those ages 40±65 in 1970, those ages 40±70 in1980 and 1990, and younger individualscovered by more broad state laws in 1970±90. DDL is equal to one for those in stateswith a disability discrimination law at time t.S is a vector of state dummies, and T is avector of year dummies, so that bs and bt

identify differences in Y across states (forall years) and across time (for all states) forall individuals. It is also possible to includeinteractions between state and year to allowfor common fixed effects for observations inthe same state and year that are not persistentacross years or states.8 Note, however, thatbecause such interactions are defined by stateand year, including them precludes estimationof j because the interactions generate a com-mon intercept shift for all observations in astate and year (and would thus be perfectlycollinear with DDL). Finally, (A*T ) is a vec-tor of age dummy by year dummy inter-actions, so that bAt captures differences intrends in outcomes by age group.

In this specification, q identifies the effectbeing protected by an age discrimination lawfor those without disability discrimination lawsin their state, and j identifies the effect of hav-ing a disability discrimination law in one's statefor those not also protected by an age discri-mination law. One interpretation of g is that itidentifies the shift in the effect of being pro-tected by an age discrimination law that is

associated with enactment of a disability dis-crimination law. Of course, one could alsointerpret g as the shift in the effect of havinga disability discrimination law in one's statethat is associated with being protected by anage discrimination law. However, given thearguments by Frier (1993) and Posner (1995)highlighted earlier and because age discrimina-tion laws were generally passed prior to disabil-ity discrimination laws, it may be moreappropriate to use the former interpretation.Apart from the question regarding any changein the effect of PAD associated with DDL (g),one could also investigate the more basic ques-tion of whether the passage of state-level dis-ability discrimination laws in addition tocoverage by an age discrimination law is asso-ciated with changes in the levels of employmentor earnings for covered workers, relative tohaving age-law protection only. This netemployment or earnings change would be cap-tured by the sum (g �j), which adds the com-mon effect of DDL across those protected andnot protected by age discrimination laws (j) tothe interaction effect (g) of the two laws.

A. Separate Effects by Age Group

Disability discrimination legislation mayinteract with age discrimination legislation dif-ferently depending on the age of an individualfor several reasons. For example, because therate of job turnover declines as workers age,the expected firing costs associated with thelaws should become increasingly relevantwith age (as compared to the costs of not hir-ing), with the result that the laws may boostemployment more for older than for youngerworkers. Alternatively, however, because dis-abilities become increasingly prevalent andsevere as workers age, the net present valueof costs of reasonable accommodationbecomes larger with age. If accommodationcosts increase relative to expected firing costsas workers age, this implies lower employmentfor older workers. It is therefore of interest toexamine the effects of the laws for older andyounger individuals separately, rather than justan average effect pooled across age groups.

Equation (1) can be augmented to allowseparate effects of the legislation for threeage groups: those under age 40 (denoted by thedummy variable UNDER40); those age 40±64;and those age 65 or older (denoted by thedummy variable OVER65). This specification

8. All specifications were also estimated whileexcluding states in the South, because they may system-atically differ from other states (for example, Alabamaand Mississippi are the only states that have neither typeof law by 1990; Virginia and Arkansas never passedstate-level age discrimination laws). The results are quali-tatively unchanged when these states are excluded from theanalysis.

118 CONTEMPORARY ECONOMIC POLICY

is shown in equation (2).

Yist � a�Xitb� �PADist �DDLst�g�2�� �PADist �DDLst �OVER65it�g 0� �PADist �DDLst �UNDER40it�g 00�PADistq� �PADist �OVER65it�q0� �PADist �UNDER40it�q00�DDLstj�OVER65itb0

�UNDER40itb00 �Ssbs �Ttbt

� �Ait �Tt�bAt � eist

As before, the q coefficients identify theeffects of being protected by only an age dis-crimination law relative to those withoutsuch protection. The g coefficients identifythe shift in this effect resulting from having adisability discrimination law in the state, con-trolling for the impact of disability discrimina-tion laws for those with no age discriminationprotection.9 Because the shift in the effect of theage discrimination laws is identified for thoseunder age 40 and over age 65 (via the g 00 andg 0 coefficients), g estimates the shift in theeffect of age discrimination law protectionfor those 40±64 years old. Finally, the sum(g�j) reveals the net impact of the coupledlaws for those 40±64 years old, relative toPAD alone.

B. Separate Effects for the Disabled

It is also possible to estimate separate effectsof the legislation for workers who classifythemselves as disabled. However, the disabilityquestion is only available for those in the 1970±90 censuses, and in the 1970 census it is onlyavailable for those under age 65, making theidentifying information much more limited.10

For example, the 1960 census can no longer beused, nor can information on 65±70-year-oldsin 1970.11

To see how restricting the sample to the1970±90 censuses affect the identificationstrategy, ignore the age limitation implied bythe disability question for a moment and focuson estimation of equation (1). Using the 1970±90 censuses, g and j would be identified basedonly on two distinct groups of individuals:those over age 65 (the group for whom PADchanges from zero to one with the extension ofthe ADEA in 1978) in the states that passeddisability discrimination laws, and thoseunder 40 years old in states that introduceddisability discrimination laws and changedthe range of their younger protected groupduring the 1970±90 period (e.g., Florida,Iowa, Maine, and Kansas).

An approach that generates additional iden-tifying information is to exploit the variationin coverage of the state-level age discriminationlaws by noting the weak enforcement of theADEA prior to 1978 (when enforcementauthority for the law was transferred fromthe Department of Labor to the EEOC) andimplicitly assuming that the only age discrimi-nation protection prior to the 1980 censuscomes from state-level laws.12 Under thisapproach, g is identified from a broader setof ages because PAD would now change overthe period for those in all age ranges.

Finally, using the disability question in thecensus implies that those age 65 or older mustbe excluded from the analysis when informa-tion on disability status is included. Thus, if theweak age discrimination laws are ignored, theestimated equation is:

Yist�a�Xitb��PADist �DDLst�g�3���PADist �DDLst �UNDER40it�g 00�PADistq��PADist �UNDER40it�q00�DDLstj�DISitd�f1��PADist �DDLst�g��PADist �DDLst �UNDER40it�g 00�DDLstj�Ssbs�Ttbtg�UNDER40itb00�Ssbs�Ttbt

��Ait �Tt�bAt�eist,9. The separate effects of the disability laws by age (e.g.,[DDL * OVER65]) cannot be identified because they wouldbe collinear with DDL and (PAD * OVER65).

10. In 1980 and 1990, the census disability questionstipulated that the disability had to have lasted at leastsix months. The 1970 census did not include this stipulationbut did ask how long the disability had lasted. To make 1970comparable with the other years, persons were classified asdisabled if their disability lasted at least six months (Rugglesand Sobek, 1997).

11. Note that theoretically if the sample is limited to the1980 and 1990 censuses, a separate effect of the laws for

older workers could be identified (i.e., those age 65±70could be included, because that age group was asked thedisability question starting in 1980). However, because theADEA protected all those age 40±70 starting in 1980,empirically the separate effect of the laws for those age65±70 could not be identified because PAD * OVER65would be perfectly collinear with OVER65.

12. Under this approach, protections offered by stateswhere the laws have weak enforcement (e.g., Georgia's law

STOCK & BEEGLE: DISABILITY EMPLOYMENT PROTECTION 119

where DIS is equal to one for individuals whoclassify themselves as having a disability thatlimits work, and d identifies the intercept shiftassociated with having a disability. Two addi-tional sets of regressors, DIS*S and DIS*T,allow for separate state effects and time trendsin Y for the disabled. This specification allowsfor estimation of changes in Y associated withthe coupled laws, estimated separately for thedisabled and by age group.

IV. EMPIRICAL FINDINGS

A. Estimated Effects of the Laws onEmployment

Regression estimates of the combinedeffects of age discrimination and disabilitydiscrimination laws on employment are pre-sented in Table 2. As discussed, in some speci-fications the data must be limited to includeonly the 1970±90 censuses, and identifyinginformation during this period is obtained byignoring weak age discrimination laws. Thuscolumn (1) presents estimates of equation (2)using the 1960±90 data, and column (2) includesthe 1970±90 data but ignores the ADEA prior to1980. Finally, in columns (3) and (4) the esti-mates of the previous two columns are repeatedwhile including state dummy * year dummyinteractions tocontrol for fixedeffects for obser-vations in the same state and year that are notpersistent across years or states. These specifica-tions exclude DDL as it is collinear with thestate * year interactions.

All four sets of estimates indicate that thepositive employment effect of protection byage discrimination laws is larger when theselaws are coupled with disability discriminationlaws. This positive interaction effect is highestamong the 40±64-year-olds, indicating that theeffect of PAD increases by roughly 3 percen-tage points for those in states with disabilitydiscrimination laws (the estimates of g in row1). For the oldest and youngest workers, theinteraction effect is smaller and insignificant, asshown by the coefficients and standard errorson (g � g 0) and (g � g 00) listed in the middle por-tion of the table.

Consistent with other research, the estimatesoftheeffectofagediscriminationlawprotectionfor those with no DDL (q) indicate that olderworkers have higher employment rates whencovered by age discrimination laws whereasyounger workers (under age 40) do not.13

Age discrimination laws have the largest posi-tive effects for those age 40±64. The estimatesof theeffectofdisabilitydiscrimination laws forthose not covered by an age protection law (j)indicate that such laws are associated withlower employment rates in general, althoughthis finding is not robust across specifications.

Finally, the estimates of the sum (g �j)Ðthe net change in employment associated withthe presence of both DDL and PAD, relative toPAD aloneÐindicate that the coupling of thelaws is associated with higher employment forthose age 40±64, but the estimates are onlymarginally significant (t� 1.83). Note how-ever, that the estimates in Table 2 that includej are somewhat misleading in the sense thatthey estimate a pooled effect of the laws acrossthe disabled and nondisabled.

Table 3 reports estimates of equation (3),which allows for separate effects of the lawsfor the disabled and nondisabled. Column(1) includes the interaction between disabilitystatus (DIS) and PAD * DDL, as well as aninteraction between DIS and DDL. Column (2)estimates a differential effect of the laws forthose under 40 years old, and column (3) in-cludes statedummy* yeardummy interactions.

The coefficient on PAD * DDL is positiveand significant in columns (2) and (3), althoughthe magnitude is small, indicating a 2-percen-tage-point positive shift in the effect of PAD onemployment for 40±64-year-old nondisabledpersons when a disability discrimination lawexists in the state. As the net employmentchange (g �j) estimates in the lower portionof the table indicate, however, the positiveshift in the effect of PAD on employment isoffset by a negative shift in employment asso-ciated with DDL laws in general for this group,so that the employment level of the nondisabledis unchanged with the coupled laws, relativeto PAD alone.

For the older disabled, the estimates of thePAD * DDL coefficients (g � dg) are negativeand significant, and the net employment effect

and Colorado's law before 1990) are also ignored.Neumark and Stock (1999) found no statistically signifi-cant difference in the employment effects of age discrimi-nation laws when they allowed the ADEA to have weakerenforcement.

13. Due to space constraints, sums and standard errorsfor (q� q0) and (q� q00) are not shown in the tables but areavailable from the authors.

120 CONTEMPORARY ECONOMIC POLICY

(g � dg �j� dj) is also negative and signifi-cant, indicating that the combined protectionof age and disability discrimination legislationfor 40±64-year-old disabled persons is asso-ciated with reduced employment levels relativeto protection by an age law alone. For youngerdisabled workers, the estimates indicate thatthe coupling of age and disability discrimina-tion laws does not significantly boost the posi-tive effect of age discrimination law protection.Combined with a positive employment effect

associated with disability discrimination lawsfor this group, however, the coupling of PADwith DDL is associated with higher net employ-ment levels among younger (under age 40)disabled workers, relative to protection byage discrimination laws alone. The differingemployment outcomes by age are consistentwith the idea that disability discriminationlaws increase the cost of employing the dis-abled, perhaps more so for the older disabledbecause the perceived or real extent of disability

TABLE 2

Employment Regressions

(1) (2) (3) (4)

Protected by age discrimination law *disability discrimination law (PAD * DDL) (g)

0.032**(0.008)

0.022*(0.009)

0.033**(0.008)

0.025**(0.009)

* age� 65 (PAD * DDL * OVER65) (g 0) ÿ0.026**(0.007)

ÿ0.022**(0.007)

ÿ0.026**(0.007)

ÿ0.021**(0.007)

* age< 40 (PAD * DDL * UNDER40) (g 00) ÿ0.028*(0.012)

ÿ0.019(0.013)

ÿ0.025(0.014)

ÿ0.018(0.014)

Protected by age discrimination law (PAD) (q) 0.016*(0.007)

0.020**(0.007)

0.019**(0.006)

0.025**(0.006)

* age � 65 (PAD * OVER65) (q0) ÿ0.007(0.009)

ÿ0.005(0.010)

ÿ0.011(0.010)

ÿ0.009(0.008)

* age < 40 (PAD * UNDER40) (q00) ÿ0.019*(0.008)

ÿ0.021*(0.009)

ÿ0.022*(0.009)

ÿ0.026**(0.008)

Disability discrimination law (DDL) (j) ÿ0.021**(0.007)

ÿ0.012(0.007)

Ð Ð

PAD * DDL: age� 65 (g � g 0) 0.005(0.009)

0.0004(0.009)

0.006(0.009)

0.004(0.008)

PAD * DDL: age� 40 (g � g 00) 0.004(0.009)

0.002(0.010)

0.008(0.012)

0.007(0.013)

Net employment change: age 40±64 (g �j) 0.011(0.006)

0.010(0.006)

Ð Ð

Net employment change: age �65 (g � g 0 �j) ÿ0.016(0.009)

ÿ0.012(0.008)

Ð Ð

Net employment change: age <40 (g � g 00 �j) ÿ0.017(0.010)

ÿ0.010(0.010)

Ð Ð

Number of observations 1,633,973 1,313,753 1,633,973 1,313,753

Adjusted R2 0.233 0.238 0.234 0.239

State dummy * year dummy interactions No No Yes Yes

Ignore laws with weak enforcement No Yes No Yes

Census years used 60±90 70±90 60±90 70±90

Notes: Estimates result from ordinary least squares. Standard errors are in parentheses and are corrected forclustering of observations by state and year. All specifications include a constant and controls for age and its square anddummies for marital status (one for married and one for widowed, divorced, or separated; never married is the referencegroup), schooling (less than 12th grade; 1±3 years college or associate's degree; 4 years college or bachelor's degree; and5 or more years or advanced degree; 12th grade or high school graduate is the reference group), residence in a metropolitanarea (SMSA), year (1990 is the reference year), and state (New York is the reference state). All specifications also includecontrols for being under 40 years old, for being over 65 years old, and age group dummy * year dummy interactions (agegroup dummies are for ages 25±29, 30±34, 35±39, 40±44, 45±49, 50±54, 55±59, 60±64, and 65±70; age 18±24 is thereference group). Net employment change is computed as the change in employment associated with PAD� 1 andDDL� 1, relative to PAD� 1 and DDL� 0 (i.e., the difference in employment associated with having both types of laws,relative to having age discrimination law protection alone).

*Significant at 5%.

**Significant at 1%.

STOCK & BEEGLE: DISABILITY EMPLOYMENT PROTECTION 121

is higher for that group. It may also be the casethat reasonable accommodation requirementsand prohibitions against nondiscriminationin hiring are more relevant for the youngerdisabled.14

B. Estimated Effects of the Laws onEarnings

Table 4 reports the estimates of the com-bined effect of age discrimination and disability

TABLE 3

Employment Regressions Including Disability Status

(1) (2) (3)

PAD * DDL (g) 0.005(0.006)

0.015*(0.007)

0.017*(0.007)

* age <40 (PAD * DDL * UNDER40) (g 00) Ð ÿ0.011(0.011)

ÿ0.008(0.013)

* disabled (PAD * DDL * DIS) (dg) ÿ0.097**(0.010)

ÿ0.105**(0.010)

ÿ0.105**(0.010)

* age <40 (PAD * DDL * DIS * UNDER40) (dg 00) Ð 0.114**(0.012)

0.113**(0.012)

PAD (q) 0.008(0.005)

0.015*(0.006)

0.016**(0.006)

* age <40 (PAD * UNDER40) (q00) Ð ÿ0.016(0.009)

ÿ0.017*(0.008)

DDL (j) ÿ0.011(0.005)

ÿ0.013*(0.006)

Ð

* disabled (DDL * DIS) (dj) 0.070**(0.010)

0.065**(0.010)

0.064**(0.010)

DIS (d) ÿ0.457**(0.016)

ÿ0.476**(0.010)

ÿ0.475**(0.011)

PAD * DDL: nondisabled age <40 (g � g 00) Ð 0.003(0.009)

0.009(0.011)

PAD * DDL: disabled age 40±64 (g � dg) Ð ÿ0.091**(0.011)

ÿ0.088**(0.018)

PAD * DDL: disabled age <40 (g � g 00 � dg � dg 00) Ð 0.012(0.016)

0.017(0.018)

Net employment change: nondisabled age 40±64 (g �j) Ð 0.002(0.005)

Ð

Net employment change: nondisabled age <40 (g � g 00 �j) Ð ÿ0.009(0.009)

Ð

Net employment change: disabled age 40±64 (g � dg �j� dj) Ð ÿ0.038**(0.008)

Ð

Net employment change: disabled age <40 (g � dg � g 00 � dg 00 �j� dj) Ð 0.064**(0.015)

Ð

Number of observations 1,220,918 1,220,918 1,220,918

Adjusted R2 0.231 0.231 0.232

State dummy * year dummy interactions No No Yes

Year dummy * disabled Yes Yes Yes

State dummy * disabled Yes Yes Yes

Ignore laws with weak enforcement Yes Yes Yes

Notes: See notes to Table 2. Specifications in columns (2) and (3) also include a control for being under 40 years old.Sample includes only individuals age 18±64 and only the 1970±90 censuses.

14. The requirement of reasonable accommodation ofdisabilities by employers is an important aspect of many ofthe disability discrimination laws. To examine the incre-mental effect of this requirement, some specifications alsoincluded controls (and interactions) for the presence of areasonable accommodation clause in the individual'srespective state statute (i.e., the RA variable shown in

Table 1). These estimates indicate that for older disabledworkers, net employment changes were not significantlydifferent, whereas for younger disabled workers netemployment changes were less positive when the combinedlaws include a reasonable accommodation clause. Therewas no differential effect on net earnings when disabilitydiscrimination laws included this clause.

122 CONTEMPORARY ECONOMIC POLICY

discrimination laws on log annual earnings. Asin the employment regressions, the positiveeffect of being protected by an age discrimina-tion law is boosted in the presence of a disabilitydiscrimination law for workers age 40±64, andthe shift in the effect is even larger for workersover age 65. For workers under 40 years old, theshift in the effect of age discrimination lawprotection associated with disability discrimi-nation laws is positive though small and statis-tically insignificant. There is a net gain inearnings levels for older workers with PAD

and DDL, relative to those who are protectedby age laws only. However, as noted, this esti-mate pools the effect of the laws across thedisabled and nondisabled.

Table 5 presents estimates of the earningsequation while allowing for separate effectsof the laws for disabled workers. The estimatesthat are pooled across age groups are small andinsignificant (column [1] ). The estimates incolumns (2) and (3) allow separate effects of thelaws by age and disability status. The presenceof a disability discrimination law does not have

TABLE 4

Earnings Regressions (Dependent Variable: ln[Annual Earnings])

(1) (2) (3) (4)

PAD * DDL (g) 0.023*(0.010)

0.009(0.012)

0.025**(0.008)

0.021*(0.008)

* age � 65 (PAD * DDL * OVER65) (g 0) 0.080**(0.026)

0.071**(0.026)

0.079**(0.026)

0.067*(0.027)

* age <40 (PAD * DDL * UNDER40) (g 00) ÿ0.015(0.017)

0.006(0.019)

ÿ0.015(0.011)

ÿ0.017(0.011)

PAD (q) ÿ0.003(0.013)

0.020*(0.010)

0.006(0.006)

0.016*(0.008)

* age �65 (PAD * OVER65) (q0) 0.014(0.023)

0.030(0.026)

0.011(0.020)

0.036(0.024)

* age <40 (PAD * UNDER40) (q00) 0.003(0.018)

ÿ0.031(0.016)

ÿ0.002(0.007)

ÿ0.007(0.007)

DDL (j) ÿ0.038*(0.016)

ÿ0.027(0.018)

Ð Ð

PAD * DDL if over 65 (g � g 0) 0.103**(0.026)

0.081**(0.028)

0.103**(0.025)

0.088**(0.026)

PAD * DDL if under 40 (g � g 00) 0.008(0.020)

0.016(0.022)

0.010(0.011)

0.004(0.011)

Net earnings change: age 40±64 (g �j) ÿ0.015(0.013)

ÿ0.018(0.015)

Ð Ð

Net earnings change: age �65 (g � g 0 �j) 0.065*(0.029)

0.054(0.030)

Ð Ð

Net earnings change: age <40 (g � g 00 �j) ÿ0.029(0.022)

ÿ0.011(0.024)

Ð Ð

Number of observations 1,010,404 806,964 1,010,404 806,964

Adjusted R2 0.704 0.622 0.706 0.625

State dummy * year dummy interactions No No Yes Yes

Ignore laws with weak enforcement No Yes No Yes

Census years used 1960±90 1970±90 1960±90 1970±90

Notes: See notes to Table 2. Controls for hours and weeks worked (constructed as midpoints of ranges reported inthe census), industry (dummies for construction; transportation, communication, and utilities; wholesale and retail trade;finance, insurance, and real estate; business and repair services; personal services; entertainment and recreation services;professional and related services; and mining; manufacturing is the reference industry), and occupation (dummies fortechnical sales and administrative support occupations; service occupations; craft, repair, and precision productionoccupations; and operators, fabricators, and laborers; managerial, professional, and technical specialty occupations arethe reference) are also included. Those working fewer than 27 weeks or 30 hours per week are excluded. Earnings in the1960±70 censuses are recorded in $100 intervals (and in 1960, earnings above $10,000 are recorded in intervals of $1000).The earnings figures were converted from categorical values to the midpoints of the categories. The earnings in the 1980and 1990 censuses, which are recorded as midpoints of $10 intervals, were not adjusted. The authors also imposed alower limit on earnings, eliminating individuals from the regressions whose earnings would have been less than $1 perhour (1980 dollars) based on half-time, half-year work.

STOCK & BEEGLE: DISABILITY EMPLOYMENT PROTECTION 123

a significant interaction with age discrimina-tion protection for any of the four categoriesof workers. Only for disabled workers age 40±64 are earnings significantly different with thecoupled laws. For this group, the coupling ofthe laws is associated with lower earnings rela-tive to age discrimination law protection alone,suggesting that the negative impact of DDL onearnings in general outweighs any positive shiftin the effect of PAD on earnings associated withdisability discrimination laws.

VI. CONCLUSIONS

Previous researchers have examined sepa-rately the effects of employment protectionsfor older workers available through theADEA and state-level age discriminationlaws and for disabled workers though theADA and state-level disability discriminationlaws. The research on the impact of age discri-mination prohibitions indicates that these lawsare associated with higher employment levels(andperhapshigherearnings)forolderworkers.

TABLE 5Earnings Regressions Including Disability Status

(1) (2) (3)

PAD * DDL (g) ÿ0.006(0.013)

0.009(0.012)

0.020*(0.008)

* age <40 (PAD * DDL * UNDER40) (g 00) Ð 0.008(0.019)

ÿ0.013(0.011)

* disabled (PAD * DDL * DIS) (dg) ÿ0.006(0.011)

ÿ0.007(0.011)

ÿ0.005(0.011)

* age <40 (PAD * DDL * DIS * UNDER40) (dg 00) Ð 0.006(0.018)

0.008(0.018)

PAD (q) 0.006(0.009)

0.020*(0.010)

0.016*(0.008)

* age < 40 (PAD * UNDER40) (q00) Ð ÿ0.030(0.016)

ÿ0.009(0.007)

DDL (j) ÿ0.019(0.018)

ÿ0.025(0.018)

Ð

* disabled (DDL * DIS) (dj) ÿ0.020(0.015)

ÿ0.021(0.015)

ÿ0.027(0.014)

DIS (d) ÿ0.155**(0.016)

ÿ0.156**(0.016)

ÿ0.149**(0.015)

PAD * DDL: nondisabled age <40 (g � g 00) Ð 0.017(0.022)

0.007(0.011)

PAD * DDL: disabled age 40±64 (g � dg) Ð 0.002(0.016)

0.015(0.013)

PAD * DDL: disabled age <40 (g � g 00 � dg � dg 00) Ð 0.015(0.029)

0.010(0.024)

Net earnings change: nondisabled age 40±64 (g �j) Ð ÿ0.016(0.015)

Ð

Net earnings change: nondisabled age <40 (g � g 00 �j) Ð ÿ0.009(0.024)

Ð

Net earnings change: disabled age 40±64 (g � dg �j� dj) Ð ÿ0.044*(0.018)

Ð

Net earnings change: disabled age <40 (g � dg � g 00 � dg 00 �j� dj) Ð ÿ0.031(0.031)

Ð

Number of observations 796,715 796,715 796,715

Adjusted R2 0.626 0.626 0.628

State dummy * year dummy interactions No No Yes

Year dummy * disabled Yes Yes Yes

State dummy * disabled Yes Yes Yes

Ignore laws with weak enforcement Yes Yes Yes

Notes: See notes to Table 4. Specifications in columns (2) and (3) include a control for being under 40 years old.Sample includes only individuals age 18±64 and only the 1970±90 censuses.

124 CONTEMPORARY ECONOMIC POLICY

However, because some older workers may notbe eligible for the protection offered under agediscrimination laws (for example, due toexemptions allowed by the ADEA based ona BFOQ), they may be better protected insteadthrough the ADA or similar disability discri-mination legislation. On the other hand, rea-sonable accommodation clauses andprohibitions against discrimination in firingraise the costs of employing the disabled, inwhich case the coupling of the laws may beassociated with lower employment for pro-tected workers.

The impact on employment and earnings ofhaving both types of laws is investigated. Theestimates indicate that the positive employ-ment effect of protection by age discriminationlaws is higher for those who live in states withdisability discrimination laws. The positiveshift in the employment effect of age discrimi-nation law protection associated with disabi-lity discrimination laws is significant for thenondisabled age 40±64 but negative for the dis-abled age 40±64. Among the nondisabled, thepositive shift in the employment effect of agediscrimination law protection is offset by thenegative overall impact associated with disabil-ity discrimination laws, with the result thatoverall employment is unchanged for thosewith both age-law protection and disability dis-crimination laws in their state, relative to thosewith age-law protection alone. Among the dis-abled, the coupled laws are associated withlower employment for older workers andhigher employment for younger workers, rela-tive to age-law protection alone, which is con-sistent with the idea that prohibitions againstnondiscrimination in hiring associated withdisability discrimination laws may be morerelevant for younger workers. The earningsregression estimates suggest that the only stat-istically significant change in net earnings asso-ciated with the coupled laws is for disabledworkers age 40±64, for whom the coupledlaws are associated with lower earnings thanprotection by age discrimination laws alone.

REFERENCES

Acemoglu, D., and J. Angrist. `̀ Consequences of Employ-ment Protection? The Case of the Americans withDisabilities Act.'' Journal of Political Economy,109(5), 2001, 915±57.

Adams, S. J. `̀ Three Essays on the Economics of Aging.''Ph.D. diss., Department of Economics, MichiganState University, 2000.

Ashenfelter, O., and D. Card. `̀ How Did the Elimination ofMandatory Retirement Affect Faculty Retirement?''Working Paper No. 448, Princeton University, Indus-trial Relations Section, 2000.

Baldwin, M., and W. Johnson `̀ Labor Market Discrimina-tion against Men with Disabilities.'' Journal of HumanResources, 29(1), 1994, 1±19.

ÐÐÐ. `̀ Labor Market Discrimination against Womenwith Disabilities.'' Industrial Relations, 32(4), 1995,555±77.

Beegle, K., and W. A. Stock. `̀ The Labor Market Effects ofDisability Discrimination Laws.'' Journal of HumanResources, 38(4), 2003, 806±859.

Bound, J., and T. Waidmann. `̀ Accounting for RecentDeclines in Employment Rates among Working-Aged Men and Women with Disabilities.'' Journalof Human Resources, 37(2), 2002, 231±50.

DeLeire, T. `̀ The Wage and Employment Effects ofthe Americans with Disabilities Act.'' Journal ofHuman Resources, 35(3), 2000, 693±715.

ÐÐÐ. `̀ Changes in Wage Discrimination against Peoplewith Disabilities: 1984±93.'' Journal of HumanResources, 36(1), 2001, 144±58.

Equal Employment Opportunity Commission. `̀ Age Dis-crimination in Employment Act (ADEA) ChargesFY 1992±FY 2000.'' Online document available atwww.eeoc.gov/stats/adea.html, 2001.

Frier, D. B. `̀ Age Discrimination and the ADEA: How theADA May Be Used to Arm Older Americans againstAge Discrimination by Employers Who WouldOtherwise Escape Liability under the ADEA.''Temple Law Review, 66, 1993, 173±96.

Hirsch, B. T., D. A. Macpherson, and M. A. Hardy. `̀ Occu-pational Age Structure and Access for Older Work-ers.'' Industrial and Labor Relations Review, 53(3),2000, 401±18.

Holbrook, T., and S. Percy. `̀ Exploring Variations in StateLaws Providing Protections for Persons with Disabil-ities.'' Western Political Quarterly, 45(1), 1992,201±20.

Hutchens, R. M. `̀ Do Job Opportunities Decline withAge?'' Industrial and Labor Relations Review, 42(1),1988, 89±99.

Jaeger, D. A., and A. H. Stevens. `̀ Is Job Stability in theUnited States Falling? Reconciling Trends in theCurrent Population Survey and Panel Study ofIncome Dynamics.'' Journal of Labor Economics,17(4 part 2), 1999, S1±S28.

Johnson, W., and J. Lambrinos. `̀ Wage Discriminationagainst Handicapped Men and Women.'' Journal ofHuman Resources, 20(2), 1985, 264±77.

Kruse, D., and L. Schur. `̀ Employment of People withDisabilities Following the ADA.'' IndustrialRelations, 42(1), 2003, 31±66.

Lazear, E. P. `̀ Why Is There Mandatory Retirement?''Journal of Political Economy, 187(6), 1979, 1261±84.

Mudrick, N. `̀ Employment Discrimination Laws forDisability: Utilization and Outcome.'' Annals of theAmerican Academy of Political and Social Science,549, 1997, 53±70.

Neumark, D. `̀ Age Discrimination Legislation in theUnited States.'' Contemporary Economic Policy,21(3), 2003, 297±317.

Neumark, D., and W. A. Stock. `̀ Age Discrimination Lawsand Labor Market Efficiency.'' Journal of PoliticalEconomy, 99(5), 1999, 1081±125.

STOCK & BEEGLE: DISABILITY EMPLOYMENT PROTECTION 125

Posner, R. A. Aging and Old Age. Chicago: University ofChicago Press, 1995.

Ruggles, S., and M. Sobek. Integrated Public Use Micro-data Series: Version 2.0. Minneapolis: HistoricalCensus Projects, University of Minnesota. Availableonline at www.ipums.umn.edu, 1997.

Schuster, M., and C. S. Miller. `̀ An Empirical Assessmentof the Age Discrimination in Employment Act.''Industrial and Labor Relations Review, 38(1), 1984,64±74.

Stone, J. E. `̀ Age Discrimination in Employment Act: AReview of Recent Changes.'' Monthly Labor Review,103(3), 1980, 32±36.

Legislation and Court Cases

Age Discrimination in Employment Act, Volume 29, U.S.Code x 621±634, 1967.

Americans with Disabilities Act, Volume 42, U.S. Code x12101±12213, 1990.

126 CONTEMPORARY ECONOMIC POLICY