Abstract

The purpose of this paper is to critique economic analysis of the Americans with Disabilities Act. Contrary to recent research, we conclude that the law did not create significant amounts of involuntary unemployment for workers with disabilities during the 1990s. Observed decreases in employment for this group can be explained, instead, as voluntary exits from the labor force. Our suggestions for more balanced economic analysis of the ADA conclude the paper.

Keywords: ADA, economic analysis, employment

INTRODUCTION

As an economist and as a provider of services for students with disabilities we share a common interest in the success of the Americans with Disabilities Act (ADA). We view passage and enforcement of the ADA as part of the ongoing struggle to secure a fairer society, one in which the quality of life is enhanced for all people.

Our strong bias for the ADA is challenged by recent economic analyses highlighting failures of the law. Notably, DeLeire (2000) concludes, "The employment provisions of the ADA clearly have failed to increase the employment of people with disabilities and indeed very likely have led to lower employment for this disadvantaged group." (DeLeire a, 11) Acemoglu and Angrist (2001), while more restrained than DeLeire in their final conclusions, nevertheless estimate statistically significant decreases in weeks worked for younger workers with disabilities following enforcement of the law.

The purpose of this paper is to analyze the DeLeire and Acemoglu and Angrist papers and to reframe the discussion in a way that supports our bias for the ADA. The choice to acknowledge our bias from the beginning is deliberate. We want to expose the implicit biases of the authors' economic analysis, a priori biases that reject government intervention as unnecessary and counterproductive. In particular we reject DeLeire's conclusion, in a recent CATO Institute publication, that the ADA is yet another example of the "law" of unintended consequences.

The paper is organized as follows: We first summarize the employment provisions of the ADA and the DeLeire and Acemoglu and Angrist papers. We then develop alternative explanations for the authors' findings, drawing from survey data and insurance company data. We are heavily indebted to the existing literature, including the work of Hale (2002), Schwochau and Blanck (2000), Stein (2000), Conti and Burton (1999), McNeil (2000), Smith and Rooney (1999), and Yelin (1997).

We conclude that the ADA did not create significant amounts of involuntary unemployment for workers with disabilities during the 1990s. Observed decreases in employment for this group can be explained, instead, as voluntary exits from the labor force. Our suggestions for more balanced economic analysis of the ADA conclude the paper.

1 BACKGROUND

1.1 Employment Provisions of the ADA The ADA defines a person as disabled if he/she has "a physical or mental impairment that substantially limits one or more of the major life activities...; a record of such impairment; or (is) regarded as having such an impairment." Major life activities include walking, lifting, seeing, hearing, breathing, and working. (ADA, 1990) As of 1994, the ADA requires firms with more than 15 workers not to discriminate in employment or wages against qualified individuals with disabilities.1 To be "qualified," a worker "...with or without reasonable accommodation, can perform the essential functions of the employment position." (ADA, 1990)

In addition to prohibiting discrimination in hiring and wages, the law also requires businesses to provide reasonable accommodations to individuals with disabilities. For example, an employer may be required to change existing facilities, restructure jobs, modify work schedules, provide special equipment or assistance, provide training or other forms of support, or eliminate nonessential job functions for a disabled worker. A business can legally avoid providing an accommodation only if it would cause "undue hardship" to its nature or operation. (DeLeire, 696)2

Recent Supreme Court decisions have limited the scope of the law, especially with respect to how disabilities are defined,3 but the DeLeire and Acemoglu and Angrist papers analyze shorter periods just after the law was passed and before employers knew how the Court would rule.

1.2 To Work or Not To Work? (DeLeire, JHR, 2000) Both papers use neoclassical economic theory to model hiring decisions made by firms. Businesses are assumed to maximize profits. If the ADA increases the costs of hiring workers with disabilities, then the "rational" firm will decrease the quantity of labor demanded for this group of workers at any given wage rate.4

DeLeire argues the ADA could increase the cost of hiring a worker with disabilities for two reasons. First, there could be more expensive job accommodations. Second, with the ADA, workers can initiate lawsuits to secure accommodations or to prevent terminations.5 To avoid potential litigation costs, firms may decide not to hire workers with disabilities. The least productive or highest marginal cost workers are at greatest risk. DeLeire's study relies on ten panels of data from the Survey of Income and Program Participation (SIPP) from 1984-1995. Respondents include only men aged 18-64. Within each panel of data each respondent is interviewed six to nine times. Questions about disabilities are asked only once of each respondent in each panel.

Parameters are estimated for two models of the hiring choice. The first model estimates year-specific effects from 1986 to 1995. DeLeire concludes that there was a 7.2 % decrease in the employment rates of men with disabilities relative to that of men without disabilities following passage of the ADA. According to his estimates, "these employment declines first began in 1990 and continued each year through 1995." (DeLeire a, 705)

The second model isolates disability specific effects and job specific effects on employment. DeLeire finds, "Men with physical and mental disabilities experienced large employment declines of 8.9 and 8.5 percentage points respectively, while men with other disabilities (including heart disease, asthma, high blood pressure, and others) were unaffected by the ADA." Government workers also were unaffected by the ADA. (DeLeire a, 705)

1.3 How Many Weeks Shall We Work? (Acemoglu and Angrist, JPE, 2001) Acemoglu and Angrist develop a model in which the costs of hiring and firing decisions are separated. Accommodation and firing costs are basically the same as those described by DeLeire. Hiring costs in Acemoglu and Angrist's model are the litigation costs firms can expect if they choose not to hire workers with disabilities. "Accommodation and firing costs are likely to reduce employment, whereas hiring costs have the opposite effect." (Acemoglu and Angrist, 10)

The dependent variable in the analysis is the number of weeks worked annually. This variable is regressed on various combinations of independent variables to isolate the effects of the ADA on weeks worked and to check parameter estimates for robustness. Acemoglu and Angrist use data from the Current Population Survey (CPS), spanning 1988-1997. Men and women are included in the study, and the sample is split into two age groups: 21-39 and 40-58.

The following is excerpted from the conclusion of the paper:

In 1993, the year after the ADA came into effect, there were marked drops in employment of disabled men aged 21-39, both in absolute terms and relative to the nondisabled. A similar drop is observed in 1992 for disabled women aged 21-39. ...In contrast to the results for younger groups, we find no decline in the employment of disabled women aged 40-58. Moreover, in some specifications the decline in the employment of disabled men aged 40-58 can be accounted for by increased transfers. (Acemoglu and Angrist, 36)

2 ALTERNATIVE EXPLANATIONS

2.1 The Samples are Poorly Defined: People without Disabilities Are Defined as Disabled. Theory is elegant. Statistics are messy. In both papers, people who do not have a disability are incorrectly recorded as respondents with disabilities. Hale reports that the CPS question which Acemoglu and Angrist use to define disability "...also may identify people with the flu, colds, broken legs, and other temporary illnesses or conditions." (Hale, 3 of 5) He concludes:

... given that questions in the CPS are not designed to measure a specific definition of disability, the burden of proof is on those who use the data to infer the labor force status of people with disabilities. To proceed as though the data are valid measures of disabilities turns a data issue into a policy issue. (5 of 5)

In DeLeire's SIPP sample,6 one-third of the respondents list heart problems, asthma, diabetes, cancer, high blood pressure, kidney or stomach ailments, HIV or AIDS, or "other" impairments as their type of disability. 27% of the respondents report back impairments as their disability, and 10% do not have a specific record for type of impairment. It is unclear how many people in this combined group, 70% of the "disabled" sample, are disabled as defined by the ADA. DeLeire observes, "The group commonly thought of as disabled - those with mobility, vision, or hearing impairments - represent merely 17 percent of the (sample) of men with disabilities." (DeLeire a, 700)

McNeil describes other problems with the SIPP data. He concludes, "...regarding the measurement over time of the employment status of individuals with disabilities... there is currently no satisfactory vehicle for producing such a measure." (McNeil, 17-18)

It is a poor beginning. The key variable of interest is crudely defined in both papers, so the samples may be badly drawn. It is impossible to know if the papers' conclusions are affected, but the use of poorly constructed samples raises doubt. It is possible that the statistical estimates are biased by such errors.

2.2 Fewer People with Disabilities Were Able to Work as the Decade Progressed. The premise in both papers is that decreased hiring caused the observed decreases in employment among people with disabilities during the 1990s. Surveys of individuals with disabilities provide another explanation. The drop in employment seems to have been caused by people with disabilities leaving the labor force, not to less hiring.

According to the Harris Survey commissioned by the National Organization of Disabilities (NOD), the number of individuals with disabilities who were unable to work increased throughout the decade, and for those who remained employable, employment actually increased. From the 2000 NOD/Harris Survey:

Over the past fourteen years, the percentage of people (with disabilities age 18-64) who say they are unable to work has risen steadily from 29% to 43%... Most importantly, while the percentage of all people with disabilities who are working has remained relatively stable over the past fourteen years, there has been a significant increase in the percentage of people with disabilities who are able to work and are working from 46% in 1986 to 56% today. (Tables 3D & 3F) (27-28)

Including respondents with disabilities who can not work will overstate the effect of the ADA on employment because these people are not protected by the ADA. As Schwochau and Blanck explain,

Because the studies rely on comparisons of the disabled and the nondisabled, inclusion in the category of the disabled individuals who cannot work at all would depress coefficients associated with disability and make differences more likely to be found. (Schwochau and Blanck, 301)

The drop in employment during the 1990s need not be explained by "fewer hires" as marginal costs increased; instead, there seems to have been "more quits" as disabilities worsened. An increase in the number of individuals exiting the labor force is not evidence of any failure of the ADA. On the contrary, given that individuals with disabilities who are able to work actually report that they are working more, the ADA seems to have had some success realizing its employment goals.

2.3 Mergers and Layoffs Triggered More Disabilities. We now shift the discussion to the issue of cause and effect. Did the ADA increase the cost of disability accommodations and cause a decrease in employment, or did decreases in employment cause more disabilities? Mergers and layoffs caused stress-related disabilities to increase sharply during the time period following passage of the ADA. Conti and Burton report insurance company data from the early 1990s:

UNUM Life Insurance Company of America ranks mental and nervous disorders as their fourth-fastest-growing cause of workplace disabilities, with a growth rate of 335% between 1989 and 1995... A preliminary report of a survey sponsored by UNUM... shows a clear correlation between layoffs and mergers and high mental and nervous disorders claims experience. For example, surveyed companies that recently experienced mergers that led to layoffs reported a 33% higher mental and nervous disorders claims incidence. (319-20)

They continue:

A survey released by the American Management Association and CIGNA Corporation in 1996 showed that companies that eliminated jobs between 1990 and 1995 were more likely to report increases in seven of eight disability categories listed in the survey questionnaire than in companies that did not eliminate jobs. The differential was greatest among psychiatric and substance abuse disorders. (320)

The issue of cause and effect is a simple one: Which came first, the disability or the unemployment? Neither DeLeire nor Acemoglu and Angrist tell us. Including respondents who became disabled because they lost an existing job will overstate the case against the ADA. No effort is made by these authors to distinguish between "disability-induced unemployment" and "unemployment-induced disability."

2.4 More People with Disabilities Went to College during the 1990s. The ADA lowers barriers to education by requiring reasonable academic accommodations, making education more attractive (i.e., less costly) to people with disabilities. This is consistent with the neoclassical "rational agent" story in both papers, but neither study controls for this possibility. Data are readily available indicating that school enrollments for people with disabilities increased during the past decade. After conducting four national surveys of Americans with disabilities, Harris Interactive reported in 2000:

... over the past fourteen years, there has been marked progress in the area of education. In fact, almost 8 out of 10 people with disabilities (77%) have graduated from high school today, compared to 6 out of 10 (61%) in 1986. (Harris, 8)

The National Center for Education Statistics (NCES) attributes increases in college attendance during the early 1990s to the ADA and other legislation:

The passage of the Individuals with Disabilities Education Act (IDEA) and other laws such as the Rehabilitation Act in 1973 and the ADA in 1990, which ensure equal access to education for individuals with disabilities, have catalyzed an increase in postsecondary enrollment among students with disabilities. In 1994, approximately 45 percent of persons 16 or older who reported having a disability had either attended some college or had a bachelor's degree or higher. In contrast, 29 percent had reported doing so in 1986.... (NCES, 1)7

The increase in school attendance may explain why employment decreased for individuals with disabilities during the 1990s. This finding also may explain why Acemoglu and Angrist's negative employment effects are more pronounced for the 21-39 age group. Younger people are more likely to attend school, so after the ADA was enforced one would expect to see employment for this group decrease. But this is a success of the ADA, not failure, and a person can only wonder why this possibility is ignored in the studies being considered.

2.5 Transfer Payments Increased. Another explanation for the decrease in employment is that increases in disability insurance (DI) and Supplemental Security Income (SSI) payments made it possible for people with disabilities to work less. DeLeire discusses this possibility and rejects it, but he admits, "...the empirical approach used in this paper cannot separately identify the effects of the ADA from any effects of other programs...." (DeLeire a, 709) Acemoglu and Angrist control for changes in DI payments and SSI benefits to test the robustness of their initial results, and they conclude that increased transfers have a small, statistically significant negative effect on employment.

This hardly seems to be evidence that increases in transfers explain the reductions in employment among workers with disabilities, until one considers the dependent variables in question. Acemoglu and Angrist are examining variations in total weeks worked. DeLeire is studying the work, not work event. It is quite possible that DI and SSI have a more pronounced effect on the decision to work rather than on the number of weeks worked. This argument is supported by the literature about wage effects on female labor supply decisions. Borjas explains:

... female labor force participation rates are very responsive to changes in the wage. Among working women, however, there is growing evidence that hours of work, like those of men, are not very responsive to changes in the wage. Put differently, female labor supply mainly responds to economic factors at the "extensive" margin (that is, the decision of whether to work or not), rather than at the intensive margin (that is, the decision of how many hours to work). (Bjoras, 54-55)

To place the argument in the context of our discussion, Acemoglu and Angrist offer evidence that changes in disability insurance and SSI had weak but statistically significant effects at the "intensive margin." DeLeire has no data to test the possible effects of such changes at the "extensive margin." Failing to control for such changes means that DeLeire's estimates are contaminated by omitted variable bias, and it is quite possible that his conclusions about the negative employment effects of the ADA are exaggerated as well.

2.6 Part-time Employment Increased. One argument that is specific to Acemoglu and Angrist's research is that they do not control for the possibility that people with disabilities worked part-time more after passage of the ADA. Part-time work could be part of a strategy to offer more flexible hours, which would be consistent with the mandate of the law to provide reasonable accommodations. Yelin reports

Persons without disabilities experienced no change in the proportion working full-time over the period 1981-93. In contrast, the proportion of persons with disabilities working full-time declined from 72 to 66 percent, or by 8 percent in relative terms." (Yelin, 125)

The finding by Acemoglu and Angrist that the number of weeks worked by people with disabilities declined after passage of the ADA could be explained by a continuation of the trend to more part-time employment by this group. Again, this would be evidence of the success of the ADA.

3 DISCUSSION

To this point we have developed several alternative explanations for the DeLeire and Acemoglu and Angrist results. The use of inaccurate data, problems with cause and effect, and voluntary exits from the labor force may be some of the reasons these authors attribute negative employment effects to the ADA. In this section, we offer some general comments to challenge both the veracity and the relevance of these papers.

3.1 Many People with Disabilities Report that the ADA Made Life Better.

If the DeLeire and Acemoglu and Angrist theories are correct, then people with disabilities experienced significant increases in involuntary unemployment following passage of the ADA. But in the 2000 NOD/Harris Survey, in response to the question, "Do you think that the Americans with Disabilities Act made your life better, worse, or made no difference?" 28% answered that the ADA made life better, while only 2% reported that the ADA made life worse. (Harris, 98)

Why would nearly 30% of people with disabilities credit a law with improving their quality of life if that law was the source of reduced employment opportunities and increased unemployment? Why would only two percent complain about the ADA if a significant number of workers with disabilities lost their jobs following passage of the law?

3.2 The Complete Employment Effects of the ADA May Require More Time to be Observed. There are at least two reasons why employment gains of the ADA may not be able to be measured accurately for many years. First, improving access to education is intended to help people with disabilities secure better jobs, but such employment effects will be lagged, appearing several years (at the earliest) after enactment of the law. DeLeire's data extends only to 1995, and Acemoglu and Angrist have data only through 1997. Second, given that the law was written to combat discrimination against individuals with disabilities, the long run success of the ADA depends on overcoming prejudice and ignorance. Initially, one would expect some firms to implement the law more willingly than other firms. Eventually, a record of best management practices toward specific disabilities should accumulate that will serve as models for other, more reticent firms.

The case of L.L. Bean is worth noting. The catalogue retailer began instituting disability management programs in 1990 and implemented a formal Employee Assistance Plan (EAP) in 1995. Smith and Rooney (1999) summarize the results:

From 1990 to 1997, L.L. Bean experienced a 47% reduction in work-related lost-time injuries, and an 85% reduction in injuries for which three or more weeks of work are lost. This latter measure is especially significant because the injury management system is targeted at minimizing disability once employees experience a health-related problem... L.L. Bean sees the contribution of the EAP, with its focus on behavioral health care, as a major contributor to these reductions. (Smith and Rooney, 355-6)

As firms such as L.L. Bean develop successful strategies for workers with disabilities, other businesses undoubtedly will follow suit, and as more managers become more aware of the ADA they may discover that there are hidden benefits to accommodating people with disabilities. Stein suggests,

Among such desirable consequences are higher productivity, greater dedication, better identification of qualified candidates for promotion, fewer insurance claims, reduced post-injury rehabilitation costs, improved corporate culture, and more widespread use of available technologies. (Stein, 326)

With regard to the DeLeire and Acemoglu and Angrist papers, the assumption that the ADA must raise costs is too simple. As businesses learn how to better accommodate workers with disabilities, experience should lead to lower costs and greater rewards attributed to these workers. Given enough time, the ADA may significantly increase the employment of workers with disabilities.

3.3 Other Social Benefits Flow from the ADA To assess whether or not the ADA is an effective law, one can not ignore social benefits that flow from the legislation. Stein lists some of these intangibles:

These benefits include placing people with disabilities in a position to exercise all the responsibilities of citizenship, acknowledging that capable individuals have a "right" to work, permitting the disabled to achieve dignity through labor and productivity, and realizing the value of a diverse society. The value of these gains, as well as what any of them is worth to individual employers, is not necessarily negligible even if it is unclear. (Stein, 327)

While DeLeire discusses difficult-to-measure costs of the ADA,8 he ignores the possibility of external benefits, and external benefits must be included in any balanced assessment of the ADA.

4 CONCLUSION

DeLeire and Acemoglu and Angrist conclude that significant decreases in employment among people with disabilities during the 1990s can be attributed to the ADA. DeLeire abandons the careful language of statistical inference in a CATO Institute publication and claims that the "ADA caused a decrease of about 8 percentage points in the employment rates of men with disabilities," and that it "is a striking example of the law of unintended consequences." (DeLeire b, 23)

This claim is an overstatement, more representative of DeLeire's bias than of good science. Both studies make use of inaccurate measures of disability. People who are not disabled and people who are unable to work are included as respondents in the samples. Neither group is protected by the ADA, and the latter group increased in numbers as the decade progressed. The studies also do not differentiate between unemployment "caused" by the ADA and disabilities caused by unemployment. Insurance company data from the early 1990s indicate that stress- related disabilities spiked during the period as corporate America "downsized." This fact may explain why both studies find a negative correlation between the ADA and employment. Most important, surveys of people with disabilities indicate that employment actually increased throughout the 1990s for those who remained able to work. We conclude that decreases in employment can be attributed to voluntary exits from the labor force. The decrease in labor force participation may be explained by increases in the number of people who were unable to work, or who returned to school, or who chose to rely on transfer payments rather than work. An increase in part-time work also may explain decreases in weeks worked.

Rather than being a failure, the ADA has enjoyed success. The fact that employment increased among those who remained employable is evidence of success, and people with disabilities are more likely to credit the law with improving their lives than not. Also, the complete effects of the ADA may take time to materialize as businesses gain more experience accommodating workers with disabilities. The external social benefits of the policy also need to be assessed.

Our positive conclusions about the ADA reflect our bias for the law, but at least we acknowledge this bias, and we have framed our analysis in terms that are more speculative than DeLeire's. It may be that some firms did hire fewer workers with disabilities for reasons associated with costly accommodations, but this hardly constitutes a complete story of the employment experiences of workers with disabilities during the 1990s. A reading of the larger literature indicates a much more complex, richer story than either DeLeire or Acemoglu and Angrist suggest. Many questions remain that have not been answered. How should disability be defined for the purposes of such studies? What is the appropriate time frame for analysis? Can a model be developed that includes all competing hypotheses? And how will social scientists measure external benefits such as dignity and enhanced self-worth?

Perhaps the Supreme Court will limit the scope of the ADA so much that the law will have little relevance to firms and to workers. Still, economists have an obligation to consider such questions. To ignore difficulties because the data are not available or because the questions are too difficult is to represent bias as science. The public deserves better analysis. People with disabilities deserve better treatment.

ENDNOTES

1. The law was passed in 1990, initially enforced for firms with 25 or more employees in 1992, and extended to firms with 15 or more workers in 1994.

2. Schwochau and Blanck (2000) provide a thorough analysis of the employment provisions of the ADA and the predicted employment effects of the legislation in neoclassical economic theory.

3. Sutton v. United Air Lines, 527 U.S. 471 (1999); Murphy v. United Parcel Service, 527 U.S. 516 (1999); Albertson's, Inc. v. Kirkingburg, 527 U.S. 555 (1999).

4. Firms could offset the increased labor costs of the ADA by reducing wages paid to workers with disabilities at given employment levels. Both papers test for this possibility, and neither finds that the ADA significantly affected relative wages.

5. DeLeire reports that the Equal Employment Opportunity Commission (EEOC) resolved 106,988 cases involving the ADA between 1992 and 1998. In 14% of these cases discrimination was found and a settlement awarded. The average payment was $14,325.

6. DeLeire also uses data from the Panel Survey of Income Dynamics (PSID) in unpublished work. The CPS question used to identify disabilities is similar to the ones used in the PSID and SIPP questionnaires. It follows that the PSID data would be subject to the same data problems.

7. Daniel Gardner, information specialist for HEATH Resource Center, writes that according to the American Freshman: National Norms survey, "... in 1978... 2.6 percent of first-time, full- time freshmen reported having at least one disability. Since 1991, a steady average of more than 9 percent of first-time, full-time freshmen have reported having at least one disability." (Gardner, 2 of 3)

8. Regulation, 22.

REFERENCES

Acemoglu, Daron and J. Angrist. 2001. "Consequences of Employment Protection: The Case of the Americans with Disabilities Act." Journal of Political Economy. 109: 5, 1-43.

Bjoras, George. 2000. Labor Economics, Boston: McGraw Hill.

Conti, Daniel and W. Burton. 1999. "Behavioral Health Disability Management" in The Employee Assistance Handbook, ed. J. M. Oher. New York: Wiley.

DeLeire, Thomas a. 2000. "The Wage and Employment Effects of the Americans with Disabilities Act." The Journal of Human Resources. 35:4, 693-715.

DeLeire, Thomas b. 2000. "The Unintended Consequences of the Americans with Disabilities Act." Regulation. 23:1, 21-24.

Gardner, Daniel. 1999. "Information from HEATH." American Council on Education (available at <http://www.acenet.edu> as of 10/25/99).

Hale, Thomas. 2002. "The Lack of a Disability Measure in Today's Current Population Survey." (available at <www.nod.org> as of 5/29/02).

Harris Interactive, Inc. 2000. 2000 N.O.D./Harris Survey of Americans with Disabilities. New York: Harris Interactive, Inc.

McNeil, James. 2000. "Employment, Earnings, and Disability." Unpublished Manuscript presented at the Annual Conference of the Western Economic Association, June 29 - July 3, 2000.

NCES. 1999. "Students With Disabilities in Postsecondary Education: A Profile of Preparation, Participation, and Outcomes." U. S. Department of Education, Office of Educational Research and Improvement. NCES 1999-187.

Schwochau, Susan and P. D. Blanck. 2000. "Economics of the ADA." Berkeley Journal of Employment and Labor Law. 21, 271-313.

Smith, Geoffry and T. Rooney. 1999. "EAP Intervention with Workers' Compensation and Disability Management" in The Employee Assistance Handbook, ed. J. M. Oher. New York: Wiley.

Stein, Michael. 2000. "Labor Markets, Rationality, and Workers with Disabilities." Berkeley Journal of Employment and Labor Law. 21, 314-334.

Yelin, Edward. 1997. "The Employment of People With and Without Disabilities in an Age of Insecurity." The Annals of the American Academy. AAPSS, 549, 117-127.