Measuring Functional Capacity and Work Requirements: Summary of a Workshop (1999)

The Social Security Disability Insurance (SSDI) program was established in 1956 and was fully operational in 1960, nearly four decades ago (Berkowitz, 1987; Derthick, 1990; Mashaw and Reno, 1996a; Stone, 1984). Many of the problems in disability determination that bedevil the SSDI program were evident prior to its passage because of the experience gained from private disability insurance programs and workers' compensation (Starr, 1982). However, many were not, because the economy and society had changed. The procedures that were implemented to make disability determinations in 1960 reflect an economy dominated by goods production, physical labor, hierarchical organization, and long job tenures (Yelin, 1992); a population thought to be at risk for work loss primarily because of the chronic diseases of aging (Chirikos, 1995; Stapleton et al., 1995); and the view that most such conditions would lead, inexorably, to functional decline without any prospect for improvement.

The procedures which the Social Security Administration (SSA) will soon put in place to assess functional capacity for work in the contemporary economy may still be in use in 2040, when the youngest of the baby boomers will be 80 years old and their children will be within a decade of retirement. Thus, when we evaluate procedures to assess functional capacity for work now, it is necessary to keep in mind that they must prove relevant to the economy four decades in the future.

This paper describes some of the changes in the labor market that have occurred since 1960 and shows the extent to which the labor market experience of people with disabilities reflects these trends. It then describes briefly the Department of Labor's (DOL) new Occupational Information Network (O*NET)

system, which is designed to capture the changes in the labor market, and with which SSA hopes to assess the demands of contemporary jobs.

Although it would be hazardous to predict what the labor market will be like in the distant future, several of the most important trends have been unfolding for several decades and can be expected to continue in the years to come (Bell, 1983; Hirschhorn, 1988; Levy, 1987; Piore and Sabel, 1984; Wilson, 1997). These trends include a relative shift from goods-producing occupations and industries to the distribution of services; the increasing demand for highly skilled and highly trained labor and the erosion of demand for those with less skill and training; the emergence of new ways of accomplishing work within the firm; and the emergence of alternative work arrangements throughout the economy.

Some of these trends are relatively easy to quantify, for example, the growth of jobs in services. Some are more difficult to measure and evaluate, for example, the growth of contingent employment arrangements (Belous, 1989; Polivka, 1996), the putative erosion of job security (Nardone et al., 1997), and the flattening of workplace hierarchies (Osterman, 1988). And many of the changes are not quite as dramatic as some analysts claim: much of service work is physically demanding and much of it, regardless of the physical demand, is repetitious. All, however, are difficult to translate into a simple set of instructions for assessing functional capacity for work. Indeed, if there is a message that emerges from an analysis of the trends in the labor market, it is that in the contemporary economy, the division of tasks within and among jobs is growing increasingly complex.

As work demands change, the most important characteristic of those capable of thriving may be the ability to do multiple tasks in an overlapping and constantly evolving series of relationships and to adapt to new responsibilities. The problem facing the SSA is a daunting one: how to assess an individual's capacity to do a complex mix of tasks now and to learn a new mix later.

Dynamics in Labor Force Participation. The 1950s and 1960s are viewed by some as the halcyon era in the U.S. economy, with high growth rates sustaining unprecedented increases in the standard of living, allowing most families to survive on one income, and in turn, reinforcing the social ethic of the time that women should not work outside the home (Levy, 1987). In 1960, just under 60 percent of the working age population was in the labor force (Table 2-1). The overall labor force participation rate has increased by more than 12 percent in the interim, having reached almost two-thirds as of 1996.

Gender. This overall increase in labor force participation rates masks substantial differences by gender and age. Among all working age men, labor force participation rates declined by more than 10 percent, but men 55 to 64 years old experienced an even steeper decline, 22.8 percent (Table 2-1). Conversely, among all

working age women, labor force participation rates rose by 57.3 percent, from 37.7 percent in 1960 to 59.3 percent in 1996. Among women 25 to 34 years old, labor force participation rates more than doubled, from 36.0 percent in 1960 to 75.2 percent in 1996. Thus, the overall increase in labor force participation rates represents the net effect of a decline among men, particularly older men, and an increase among women, particularly younger women.

Race. Race plays a part in labor market dynamics and would appear to interact with gender.³ In the last quarter century, labor force participation rates increased among all working age white persons by 11.5 percent, but the increase among all working age black persons was only 3.7 percent (Table 2-2). The decline in labor force participation rates among all working age white men was only about half that experienced by black men (5.3 and 10.2 percent, respectively), while the increase among white women was far larger than that among black women (38.7 and 22.0 percent, respectively). In 1970, black men were almost as likely as white men to be in the labor force, but this was no longer the case in 1996. In 1970, black women had substantially higher labor force participation rates than white women. The larger increase in labor force participation rates among white women since 1970 has resulted in the two groups of women having nearly identical participation rates.

Age. Another factor affecting the labor market over the last several decades—one likely to have an even more profound impact on the proportion of the working age population at risk for work disability in the years to come—has been the dramatic change in the age structure of society as the baby boomers age (Table 2-3). The proportion of the population 18 to 34 years of age rose substantially between 1960 and 1980, but has since fallen, while the proportion 34 to 44 rose between 1980 and 1996, and the proportion 45 to 54 has just now begun a precipitous increase, to be followed in the decade to come by a substantial rise in the proportion of workers 55 and over.

The importance of the aging of population for the labor market can be seen in Table 2-4. In 1996, more than 80 percent of people 20 to 34, 35 to 44, and 45 to 54 years of age, respectively, were in the labor force. In each case, these percentages had risen over time, as the labor market accommodated the substantial increases in labor force participation rates among women. The increases in labor force participation rates were all the more remarkable, given that the absolute number of young and middle-age workers was increasing because of the baby boom generation. Thus, the labor market accommodated an increasing percentage of a substantially larger number of persons.

However, labor force participation rates are much lower among people 55 to 64 than among those 45 to 54, and they declined among persons in the former age group throughout most of the last two decades. The decrease in labor force participation rates among persons 55 to 64 before 1990 occurred because more people these ages chose to leave work prior to the ages when Social Security eligibility begins (at 62) and reaches its maximum (currently at 65). Labor force participation rates are lower among persons 55 to 64 at any one point, because persons in this age group face higher rates of displacement from their jobs and

because the prevalence of health problems associated with aging begins to affect substantial number of people at these ages. As a result of the increasing number of persons 55 to 64 years of age, in the future, a higher proportion of the working age population will be at risk for the onset of the chronic diseases of aging, putting increased pressure on the SSDI program.

Education. As was seen in Table 2-1, the proportion of working age adults in the labor force rose substantially between 1970 and 1996. The increase in labor force participation rates affected all but those individuals who had not finished high school (Table 2-5). Moreover, the magnitude of the increase was larger with each increment in educational attainment. Thus, labor force participation rates increased among high school graduates by 11.0 percent, among those with

some college by 13.4 percent, and among those with a college degree or more, by 19.0 percent. As a result, by 1996, labor force participation rates among college graduates were almost 50 percent higher than among persons with less than a high school education.

Since 1960, the proportion of the adult population with at least a high school diploma has almost doubled (from 41.1 to 81.7 percent), and the proportion with four or more years of college has more than tripled (from 7.7 to 23.6 percent) (Bureau of the Census, 1997, p.159). Nevertheless, a substantial fraction of the cohorts entering the ages of highest risk for work disability have less than a high school education, including more than 12 percent of those now 35 to 44, more than 13 percent of those now 45 to 54, and more than 22 percent of those now 55 to 64 (Bureau of the Census, 1997, p.160). These individuals may face a difficult time maintaining a toehold in the labor market. In addition, about a third of these cohorts have no more than a high school degree. Although the labor force participation rate for high school graduates increased by 11.0 percent overall after 1970, it decreased slightly between 1990 and 1996. If the latter trend continues or accelerates, more high school graduates will fail to enter the labor market.

Dynamics in Employment Characteristics. There is little doubt that there has been a fundamental shift in the kind of work done, as reflected in the change in the distribution of occupations and industries. However, analysts disagree on the degree to which there has been a corresponding shift in how work is done. Osterman (1988) noted that throughout much of this century, firms had two kinds of employees: a salaried workforce paid to design and monitor work processes, who were given relative autonomy to carry out their work, and had security of employment (''white

collar'' workers), and an hourly wage workforce paid to implement these work processes, with little discretion over how the work was done, and who were retained only when the demand for products justified continued employment ("blue collar" workers). Osterman observed that more recently, many firms were melding the two kinds of jobs: bringing the expertise of those involved in production of goods and services into the design of work processes, while reducing the security of employment among the white collar workforce.

The signposts for the changes described by Osterman include flattened workplace hierarchies, broadened and variable work tasks for each job, reduced job tenure, increased use of part-time and temporary workers, alternative work arrangements, and higher rates of job displacement. There is strong evidence in the work disability literature that providing flexible working conditions and job autonomy reduces the probability that an individual with an impairment will stop working (Yelin et al., 1980). Indeed, the Americans with Disabilities Act of 1990 (ADA) mandates the provision of such accommodations to help sustain employment (West, 1991). The model underlying the research on the effect of accommodation on employment as well as the reasonable accommodation provisions of the ADA, is that increased autonomy to perform an existing mix of job demands in the context of a long-term relationship with an employer will improve job prospects. However, it is not known how well persons with disabilities can function when asked to flexibly shift among job tasks and work groups, especially with decreased levels of job security.

Ongoing data collection efforts at the DOL's Bureau of Labor Statistics (BLS) measure some of the shifts in working conditions—job tenure, frequency of part-time and temporary employment, alternative work arrangements, and rates of job displacement. They do not capture changes in the nature of work-place hierarchies and in the mix of work tasks for each job. Obtaining such information will be critical in assessing the functional demands of work and, therefore, in assessing the capacity of persons with disabilities to function on the job.

Industries. Table 2-6 shows the change in the number of employees and share of nonagricultural employment among industries since 1960. It provides information on the most tangible signpost of the change in the nature of work. In 1960, the goods-producing sectors of the economy (mining and construction, and manufacturing) accounted for 6.7 and 31.0 percent of employment, respectively. Since then, the share of employment accounted for by mining and construction has decreased by about a quarter, and the share accounted for by manufacturing decreased by slightly more than half. Indeed, at a time when total employment more than doubled (datum not in table), the absolute number of manufacturing workers increased by only 8 percent, from 16.8 million in 1960 to 18.2 million in 1996. Thus, as of 1996, the goods-producing sectors of the economy accounted for only a fifth of total employment.

Concurrently, there was substantial growth in the share of employment in the finance, insurance, and real estate sectors (18.4 percent net decline from

1990 to 1996) and in the service industry (111.0 percent). Primarily because of the growth occurring prior to 1980, the share of total employment in the public administration sector increased by 5.8 percent since 1980; however, its share has declined by 9.4 percent.

Because the service sector is heterogeneous, encompassing, for example, those who work in private households, physicians' offices, engineering firms, and home cleaning services, it is far more informative to study the employment dynamics within the components of the overall services category. The share of employment in all but the personal services component expanded between 1970 and 1996, with business and repair, entertainment and recreation, and professional services growing by 247.4, 90.0, and 41.9 percent, respectively (Table 2-7). By 1996, the absolute number of workers in professional services exceeded 30 million, almost a quarter of all nonfarm employment. Within the business and repair services component, the absolute number of workers in personnel supply firms (including temporary employment agencies) increased more than fivefold during this time between 1970 and 1996, while the number in the computer and data processing fields increased more than fourfold (data on absolute number of workers in these specific industries not in the table).

Occupations. The change in the share of employment among occupations reflects the shift in the overall economy from the production of goods to the production and distribution of services (Table 2-8). Thus, the share of employment in professional, specialty, and managerial occupations; technical, sales, and administrative workers; and service workers increased by 30.3, 39.4, and 11.5 percent, respectively, while the share in precision production and craft occupations; operatives, fabricators, and nonfarm laborers; and in farming and fishing occupations decreased by 17.7, 39.0, and 64.0 percent, respectively.

The shift from manufacturing to service occupations does not necessarily mean an absolute reduction in the former. Indeed, in absolute terms, the number of precision production and craft workers and operatives, fabricators, and nonfarm laborers is substantially greater now than in 1960 and has been relatively stable since 1980. Among major occupational classifications, only farming and fishing have declined in absolute terms throughout the period covered. In contrast, the absolute number of persons in professional and managerial and technical, sales, and administrative occupations has more than doubled since 1960 (from under 14.6 to 36.5 million and from 14.0 to 37.7 million, respectively). The number of service workers also has increased twofold (from 8.0 to 17.2 million). Growth in the number of professional and managerial workers has continued apace but at a somewhat slower pace since 1980. Growth is slow among technical, sales, and administrative and service workers, even more so since 1990 (Table 2-8). The continued growth in professional and managerial occupations, with relative stasis among technical, sales, and administrative and service workers, belies the prediction that the American economy would be producing few good jobs and many bad ones (Braverman, 1974; Wright and Singleman, 1982).

Part-Time Employment. The proportion of the employed population working part-time has increased steadily since 1970 from 13.2 to 17.4 percent (Table 2-9). BLS divides part-time employment into voluntary and involuntary components (labeled "noneconomic" and "economic" reasons, respectively). Overall, the proportion of all employment which is part-time due to economic reasons increased from 2.8 to 3.4 percent between 1970 and 1996, more than 21 percent in relative terms. However, the proportion of the total employed population working part-time for economic reasons has actually decreased recently from the 4.3 percent level in 1990 due to the improvement in the labor market. In contrast, the proportion of the total employed population working part-time for noneconomic reasons continues to increase, having grown by more than a third from 1970 to 1996, from 10.4 to 14.0 percent of the employed population.

Terms of Employment. It is frequently claimed that an increasing fraction of all work is not in the traditional mode of being permanent, reasonably secure, in the direct employ of the firm in which the work is done, and at a work site maintained by the firm. BLS has kept abreast of many of the changes in the terms of employment in its data collection efforts, but trend data are not available for many of them.

Job security is measured by length of time on the job (tenure) and the expectation of staying on the same job for an additional year (contingency) (Nardone et

al., 1997). Among men, the overall median length of time at a job has not changed much since the early 1980s because the male workforce has aged and older workers have longer tenures. Within each age range, job tenure among men has decreased. Among women, job tenure has increased, both because the fraction in older age groups has increased and because tenure for women 35 to 44 and 45 to 55 has increased (BLS, 1997a). Thus, the picture for job tenure is mixed, with women having unambiguously longer tenures and men having shorter tenures at each age, but more men being in the ages with longer tenures.

BLS defines contingent employment three ways: (1) as the proportion of wage and salary workers whose jobs have lasted a year or more but who do not expect them to last another year; (2) the proportion of such workers as well as the self-employed and independent contractors in this situation; and (3) the proportion of both who do not expect their jobs to last another year, regardless of how long they have been in those jobs. The proportion meeting each definition declined slightly between 1995 and 1997. For the first definition, the decrease was from 2.2 to 1.9 percent of all workers; the second was from 2.8 to 2.4 percent; and the third was from 4.9 to 4.4 percent (BLS, 1997b). Thus, contingency is reasonably common, but has definitely not increased in the last few years. However, the recent decline may be due to the strength of the labor market in the last few years and may not reflect a long-term trend in security of employment.

Alternative work arrangements involve the shift from the direct hiring of workers to perform certain functions to the purchase of the services of other firms for those functions. These include the use of independent contractors, on-call workers, workers provided by temporary help agencies, and workers provided by contract firms. BLS has only collected information on such arrangements twice, in 1995 and 1997. The proportion of the employed with alternative work arrangements did not change substantially during this two-year period. As of 1997, 6.7 percent of all workers were independent contractors, 1.6 percent were on-call workers, 1.0 percent worked for temporary help agencies, and 0.6 percent worked for contract firms.

Procurement of services outside the firm does not necessarily reduce the number of employees in the firm because outside services may be new or firm employees may be shifted to new functions as their old functions are outsourced. BLS collects information on proxy measures of the magnitude of employment in industries and occupations that represent services that could be done outside a firm (Clinton, 1997). Data on such measures suggest substantial growth in procurement of services outside of firms. The share of total employment in the business services sector has increased threefold since 1972, and one component of this industry, personnel supply, has increased more than sevenfold during this time. In addition, there has been substantial growth in the engineering and management consulting sectors. Also, firms in a majority of industries have reduced their direct employment of business support occupations, those occupations that are most likely to be performed by outside contractors.

Change in the Location of Work. BLS collected information on the number of persons who do at least part of their jobs from home in 1991 and 1997 (BLS, 1998b). The number of persons who do some work at home was slightly more than 21 million (17.8 percent of the workforce) in 1997 and had not increased substantially since 1991. However, an increasing fraction of persons who work at home are paid to do so. Almost two-thirds of persons who work at home are in managerial and professional specialty occupations.

Change in the Internal Structure of Work. Workplace literature suggests a trend to diffuse authority over decisions about the way work is done throughout the hierarchy, to increase use of flexible work groups that coalesce only for the duration of specific projects, and to increase the mix of tasks done by the individual (Cornfield, 1987; Hirschhorn, 1991; Kelley, 1990; Osterman, 1988). The evidence for this kind of shift derives from qualitative studies of work settings (such as the shop floor and office) and from interviews and case studies of managers and line workers. However, without statistical evidence that such changes are widespread, it is difficult to ascertain what proportion of the workforce has experienced them. In the 1970s, the DOL collected this kind of data in Quality of Employment surveys; it has not been collected since (Quinn and Staines, 1979; Schwartz et al., 1988).

The potential importance of changes in the internal organization of work for people with disabilities is profound. Flexibility in the pace and schedule of work and autonomy in how work is done are strongly correlated with whether or not someone is able to maintain employment (Yelin et al., 1980). Thus, if the observation that these conditions are more prevalent in work now than in the past were to be true, it might augur an improvement in the employment picture for persons with disabilities. On the other hand, for persons with cognitive, communications, and psychological disabilities, the need to interact with a constantly changing

array of work groups and the impermanent working conditions may make it more difficult to work. Although it would be hard to capture these qualitative changes in working conditions in large-scale labor market surveys, they may be more important in determining the employment prospects of persons with disabilities than the more objective changes in employment described above.

Rates of Displacement. BLS defines job displacement as the loss of a job held on a long-term basis (three or more years). BLS has tracked job displacement since the early 1980s (Hipple, 1997). The overall rate of job displacement seems tied to the economic cycle. It rose with the recession in the early 1980s, fell with the recovery late in that decade, rose once again with the recession early in this decade, and has since fallen. However, the composition of the population of displaced workers has changed considerably. In the early years of the BLS data collection efforts, the rate of displacement was greater in manufacturing industries and in occupations, such as craft workers and operatives, that were concentrated in those industries. In the interim, the rate of displacement has grown faster in white collar occupations and is now almost as large as in blue collar occupations. It has also begun to spread to rapidly expanding industries, such as the finance, insurance, and real estate sectors. Thus, although a large proportion of displacement is due to cyclical changes in the economy, a portion of job displacement also occurs in successful and expanding sectors. Job displacement is becoming a more generalized strategy of accommodating change in the labor force, and is not limited to select occupations and to industries facing difficult times.

People with disabilities have experienced most of the major trends in the labor market over the last several decades, albeit in exaggerated form. This section reviews the evidence to support this statement. Trend data on persons with disabilities, however, do not cover the same time periods as the general labor market data reviewed in the previous section, because most federal data series do not collect information on disability status with the same regularity as on such characteristics as gender, race, and age.

Labor Force Participation Rates. Between 1983 and 1994, labor force participation rates among all working age persons increased by 4.8 percent (Table 2-10). Although persons with disabilities continue to have lower labor force participation rates than persons without disabilities (51.8 percent and 83.0 percent, respectively), such persons experienced a larger relative increase (6.6 percent) than those without (4.9 percent). Thus, persons with disabilities more than

shared in the overall increase in the proportion of working age adults actually in the labor force.⁴

Gender, Age, and Race. Trends in labor force participation are exaggerated for both men and women with disabilities. While labor force participation rates were increasing 10.8 percent among women without disabilities between 1983 and 1994, women with disabilities experienced an increase of almost twice the magnitude during this time (20.0 percent). Concurrently, men with disabilities experienced a much larger decline in labor force participation rates than men without (2.0 and 0. 1 percent, respectively).

Recall from Tables 2-2 and 2-3 that the decline in labor force participation rates among men was concentrated in the 55 to 64 age group, especially among black men in that age group. The increase in labor force participation rates among women was concentrated among women 25 to 34, especially white women in this age range. Labor force participation rates among men with disabilities 55 to 64 years old declined to a greater degree than among those without disabilities, and black men with disabilities in this age range experienced the largest relative decline in labor force participation of any single group defined by gender, age, race, and disability status. By contrast, young women with disabili-

ties, particularly young white women with disabilities, experienced the largest increase of any single group defined by these four characteristics (Yelin, 1994).

Education. People with disabilities are overrepresented among persons with a high school education or less and underrepresented among those with some college or more (data from author's analysis of 1995 Current Population Survey Public Use Tapes). However, at every level of education they have lower labor force participation rates than persons without disabilities, even after statistical adjustment for differences in health and functional status and demographic characteristics (Table 2-11). The difference in labor force participation rates is greater at lower levels of education. For example, the labor force participation rate among persons with disabilities with less than a high school education is about a fifth as great as among such persons without disabilities (12.4 and 67.7 percent, respectively), but persons with disabilities who have some graduate school or more have a labor force participation rate more than half that of persons without disabilities (47.5 and 88.0 percent, respectively). Attaining higher levels of education improves the employment prospects of persons with disabilities to a greater degree than persons without disabilities. However, even when persons with disabilities have gone to graduate school, they still do not achieve as large a return on education as persons without disabilities. Overall, persons with disabilities experience low labor force participation rates because they have low levels of education and lower returns from every level of education.

Employment Characteristics and Persons with Disabilities. Do persons with disabilities have access to the same mix of jobs and to the same working conditions as those without disabilities?

Industries. Recall from Table 2-6 that three industrial sectors have had a declining share of employment (mining and construction, manufacturing, and transportation, utilities, and communications), three have had a substantially increasing share (wholesale/retail trade, finance, insurance, and real estate, and services), and one (public administration) has had little change, as a result of an increase prior to 1980 and a decline since then. Table 2-12 shows the mix of industries in 1995 among persons with and without disabilities who were employed. There are no clear patterns. Persons with disabilities are underrepresented among two sectors with a declining share of employment (manufacturing and transportation, utilities, and communications) and one with an increasing share (finance, insurance, and real estate). They also have a larger share of overall employment in the service industry and in two components of this sector, business and repair and personal services. Persons with disabilities have a smaller share of employment in professional services, the largest service industry component, than persons without disabilities.

Occupations. The occupations with an increased share of employment over the last several decades include professional and managerial occupations, technical, sales, and administrative workers, and service occupations, while craft workers, operatives, fabricators, and nonfarm laborers, and farming and fishing occupations have had declining shares of employment. With respect to the occupations with an increased share of employment, persons with disabilities are much less likely than those without to be in professional and managerial occupations, about as likely to be in technical, sales, and administrative occupations, and more likely to be service workers (Table 2-13). With respect to the occupations with a declining share of employment, people with and without disabilities are equally likely to be craft workers, but people with disabilities are much more likely to be operatives, fabricators, and nonfarm laborers, and to be in farming and fishing.

Part-Time Employment. Persons with disabilities have experienced a disproportionate amount of the increase in part-time employment (Table 2-14). As of 1995, persons with disabilities reported that 36.8 percent of their employment was part-time, an increase of 31.9 percent since 1981. The increase in all forms of part-time employment among persons without disabilities was far smaller (1.8 percent). Among persons with disabilities, the prevalence of part-time work for economic reasons rose at least until the early 1990s, but has since fallen. Among persons without disabilities, it has fallen steadily since the mid-1980s. Over the entire period under study, persons with disabilities have experienced a much smaller decline in part-time employment for economic reasons than persons without disabilities—1.6 versus 11.6 percent.

People with disabilities have experienced a substantial increase in part-time employment for noneconomic reasons during this decade, leading to an overall increase of 41.7 percent in this measure over the entire period under study. By contrast, the rate of part-time employment for noneconomic reasons has not

changed much among those without disabilities in this decade and has risen by only 6.5 percent since 1981.

Terms of Employment. Of the measures of the terms of employment reviewed with respect to the entire labor force, none is available on an ongoing basis from the monthly Current Population Survey (CPS). Instead, the measures—tenure, contingency, flexibility, alternative work arrangements, and work at home—are not collected routinely, and when collected, they are part of infrequent surveys in which respondents are not asked to report disability status.

Because of the lack of consistent data on terms of employment among persons with and without disabilities from the BLS surveys, the results of less comprehensive surveys must be used. In one such survey, a random sample of California working age adults was interviewed in 1996 about working conditions and current employment status. The results indicate that people with disabilities were more likely to have temporary employment. Paradoxically, they reported longer job tenure, even after adjustment for age and gender. This suggests that they may be locked into jobs because of their disability and the attendant need to maintain benefits, especially employer-provided health insurance. People with disabilities were no more likely to work at home, the only measure of work arrangement available in the survey. Finally, compared to people without disabilities, persons with disabilities were less likely to report high levels of job autonomy and sufficient time to get their jobs done.

Job Displacement and Accession. The biannual BLS survey used to establish the rate of job displacement does not include a measure of disability status. The March Supplement to the CPS, in which respondents report their employment status for the year prior to the survey as well as for the prior week, is analyzed here to proxy such a measure (Yelin, 1996). Among those who were employed in the year prior to the survey, people with disabilities are three times as likely as those without disabilities to report not being employed as of the week before the survey (39.8 and 13.2 percent, respectively). Even after adjustment for health and functional status, demographic characteristics, and the nature of employment in the prior year, people with disabilities who worked in the year prior to the survey are more than twice as likely as those without disabilities to report not being employed as of the prior week (31.9 and 13.7 percent, respectively).

Among people who reported no employment in the year prior to the interview, those persons with disabilities were only one fifth as likely to be employed as of the week prior to the interview as persons without disabilities (2.0 and 10.0 percent, respectively). Adjustment for health and functional status, demographic characteristics, and work history did little to change this result (after adjustment, 2.1 and 9.4 percent of people with and without disabilities who did not work in the year prior to the survey, respectively, reported that they were employed as of the week before the interview).

A second set of analyses correlates the proportion of persons with disabilities employed in an industry in each year with that industry's total share of employment in that year. The results suggest that persons with disabilities are more likely than those without disabilities to be displaced from industries with a declining share of employment and more likely to obtain jobs in industries gaining employment (Yelin, 1992).

Finally, in the 1996 California survey described above, people with disabilities did not report higher rates of job displacement, but they did report that when displacement occurred, it was more likely to result in a major problem in their lives.

This review of overall trends in the labor market and of trends affecting persons with disabilities has yielded a partial description of how things are, not how they might be in the years to come. However, the major trends in employment—the decline in labor force participation among older men, the increase among younger women, the shift from manufacturing to service industries and occupations, and the emergence of new terms of employment—have been unfolding for several decades, and there are no major disjunctures forecast for these trends in the years to come (Bowman, 1997).

More importantly, this review is a description of whether persons with disabilities do work and, if so, how and where, not of whether they can work. However, the evidence presented in this paper is consistent with the notion that, given the appropriate economic climate, a substantial number of persons with disabilities will enter the labor market and then maintain employment.

What is preventing them from doing so? Yelin and Trupin (1997) recently completed an analysis of the factors affecting transitions into and out of employment among persons with and without disabilities. For persons with disabilities, demographic characteristics were the principal factors affecting the probability of entering employment, those 18 to 24 years of age were six times more likely to do so than those 55 to 64 years of age, and white persons with disabilities were 40 percent more likely to enter jobs than black persons. Interestingly, the principal factor affecting whether persons with disabilities maintained employment was the industry in which they worked, while the principal factor affecting whether persons without disabilities did so was their occupation. This suggests that the probability that persons with disabilities will be able to keep working after onset of impairment is determined to a large extent by the welfare of the sectors in which they work, rather than their own characteristics. The welfare of persons without disabilities, in contrast, is tied to a greater extent to their personal background. Expanding industries will find a way to accommodate the needs of their workers with disabilities, level of impairment notwithstanding.

Thus, the question of how to assess functional capacity for work cannot be asked abstractly. Instead, it must be asked assuming a strong demand for labor and the presence of reasonable accommodation, as mandated by the Americans with Disabilities Act of 1990 (West, 1991). Nevertheless, even when these conditions are met, many individuals will not be working, suggesting that it may be possible to describe a core set of functional requirements that apply even when the demand for labor is strong. Although the capacity to tote the barge and lift the bale still applies to some jobs, increasingly the core competencies would appear to revolve around the ability to communicate, concentrate, interact with others, learn new tasks, and be flexible in how and with whom work gets done (Osterman, 1988). This is true even when a job demands the capacity for toting and lifting, but it is especially true in the growth sectors of the economy in which the physical demands of work may be minimal.

O*NET⁵ has been developed under a contract from the Department of Labor to replace the Dictionary of Occupational Titles (DOT) (for a detailed description see Peterson et al., 1996). The purpose of O*NET was twofold: (1) to create

an online database of work requirements in order to provide job information in an accessible format that can be readily updated, and (2) to provide a listing of job characteristics that reflect the contemporary economy. The DOT characterized jobs on the basis of the complexity of dealing with data, people, and things. The O*NET characterizes both the attributes of occupations and the characteristics of the people who fill each job. Data are collected on six separate dimensions: experience requirements (training, experience, licensing); worker requirements (functional skills, general knowledge, and education); worker characteristics (abilities, interests, and work styles); occupational characteristics (labor market information, occupational outlook, and wages); occupational requirements (work activities, work context, and organizational context); and occupation-specific information (the knowledge required for an occupation, occupational skills, and the specific tasks on the job). The data for O*NET derive from a survey of job analysts and from interviews with persons in each occupation (the latter source will include a greater number of characteristics than the former one, but the data will not be available for some time). In both instances, respondents will be asked to report the level of each characteristic on a scale; the average level among all respondents for each characteristic will be disseminated.

A thorough description of O*NET and of how it may be used is beyond the scope of this paper, as is a listing of its shortcomings with respect to the assessment of the functional capacity of Social Security disability applicants. For the former, suffice it to state that O*NET has the capacity to capture the complexity of each job through the diversity of the dimensions measured and the rapid pace of change in the nature of each job. For the latter, O*NET's principal limitation is its reliance on the average level among respondents for each job characteristic; SSA needs to assess minimal requirements on each such characteristic. However, in capturing the complexity of the modern job, O*NET solves one problem for SSA (providing a contemporary model of work), while raising another (providing no easy method to assess which among six dimensions and 300 specific characteristics are the essential functions of a job and, thus, are central to an assessment of functional capacity).

Indeed, this conundrum is not unique to the situation SSA faces. In assessing whether employers are in compliance with the employment requirements of the ADA, the Equal Employment Opportunity Commission is asked to assess whether an individual can perform a job's essential function, but the law provides little guidance in how to determine what such a function is (Jones, 1991). If it is true that an increasing proportion of jobs involve complexity and dynamism in tasks, competencies, and relationships with colleagues, then it necessarily follows that a system to assess functional capacity must take this complexity into account today and incorporate the ability to measure, if not predict, change in these characteristics in the years to come. The jobs that can be reduced to one unvarying essential function may be those that few people want and, paradoxically, those that because of their high physical demand, few persons with disabilities can perform.

Unless the pace of legislative change quickens, the Social Security Administration may use the techniques put in place in the next few years to assess functional capacity for work in the year 2040. If this is so, a workshop of the Institute of Medicine on assessing functional capacity held in 2040 may review the deliberations of this workshop just as this committee is looking back upon the deliberations prior to the passage and subsequent implementation of the SSDI program. It would behoove us to be humble in predicting the future, for many of the predictions of the late 1950s and 1960s proved unfounded. At the time the SSDI program was initiated, many analysts saw automation as the principal threat to the labor market, with rising unemployment and deskilling of jobs the necessary result of this trend.

Today, we are concerned about the erosion of job security, and wonder how many of us can cope with the demands of the service economy (and even the manufacturing sector) for a flexible response to a varying set of tasks. However, recent projections concerning the nature of the labor market call into question some of our predictions about even the near future (Bowman, 1997). In the last several decades, with the entrance of women into employment, the labor force has grown and the service sector has expanded. Attenuation of the former trend necessarily will occur: most of the women who could enter work have already done so. While the latter trend is expected to continue overall, some parts of the manufacturing sector also are projected to expand, particularly industries related to exports and the manufacture of items requiring high levels of capital investment. Nevertheless, all projections for the future suggest that the premium paid to those with high levels of education will continue, and that flexibility on the part of the worker will be of paramount importance.

The fears of 40 years ago proved unfounded, because the only model we had to work with was a mechanistic model of the production of goods. In that model, we believed it would be relatively easy to assess capacity for work. Most of the people who would apply for SSDI benefits were blue collar workers in the manufacturing sector with degenerative, largely physical conditions of aging. The fears of today may be unfounded, because the majority of tomorrow's workers may function much better than our own generation in jobs with a complex and varying set of tasks and because we may learn to accommodate the needs of workers with cognitive and behavioral impairments better than we do today.

Just as the past generation was unable to predict what the world of work would be like in year 2000, we cannot know with certainty what jobs will demand of us in the future. However, we have learned something: any system put into place must accommodate rapidly changing conditions. The visionary and all-encompassing criteria of today necessarily become the mechanistic ones of tomorrow, unless we build in the capacity to change the criteria as quickly as the economy evolves. This, in turn, requires us to gauge the changes through statistical measurement. As users of the tools developed by BLS and as potential users

of O*NET, we know that we have the capacity to measure the changes taking place in employment. As evidenced by the lack of investment in statistical agencies such as the BLS and the National Center for Health Statistics (NCHS) over the past two decades, what is lacking is the will to take that pulse.

Research on the use of functional capacity and work requirements must start with a thorough examination of the labor market in which this capacity must be used and the conditions that are likely to affect its determination in the future. As Dr. Yelin pointed out, the economic engine that will move this country in the twenty-first century will be spearheaded by an industrial composition that is likely to be much more service producing than goods producing, a labor force that will be growing more slowly than in the past, a labor force that will be on the average somewhat older than in the past, and which will have a much heavier representation of minorities, especially Hispanics, than in the past.

Even more important, for purposes of the current discussions, is the fact that the economic and industrial shifts reviewed in Dr. Yelin's paper are expected to continue to bring a significant change in occupational requirements. The fastest growing occupations can certainly be expected to place increasing demands on the technical and cognitive skills of the workers seeking jobs as the country moves forward into the next century and beyond. Clearly, employment in the future, more than in the past, will require improved educational attainment on the part of all workers. Employers can be expected increasingly to demand workers who are technologically literate and learn new skills easily, who can think critically and solve problems, and who have the skills to communicate with others and to work in teams. In addition, much of the labor market data suggest worker relationships to employers and companies in the future may be less stable than in the past, requiring each worker to be more flexible in his or her search for a job and in the use of worker skills. In the future, all workers will be forced to upgrade their education and skills throughout their working lives to be able to cope with the challenges of new technologies and greater global competition.

These trends must be kept in focus as SSA assesses new approaches to evaluating disability and the capacity of those with disabilities to hold jobs. Some of the available data, displayed in Dr. Yelin's paper, on persons with disabilities who are not employed suggest that young people with disabilities who are 18–24 years old are six times more likely to work than those with disabilities who are 55–64 years old. Further, he points out that white people with disabilities are 40 percent more likely to enter jobs than nonwhite people who have disabilities. The implications of these data become obvious when we consider that the BLS projects a median labor force age of 41 in year 2006 and a workforce that will have a higher representation of minorities than in the past.

It is indeed unfortunate that adequate information is not available on the population with disabilities and especially on those people with disabilities who work. Obtaining such information through surveys is not easy. Before data can be collected, the issue under study must be clearly defined, and questions must be developed that respondents can answer and that will produce objective information that is factual and reliable. Many efforts have been made in the past, but the results have been quite limited. Survey questions have focused either on counting the particular kinds of disabilities that exist or on the functional activities required for a person to live (i.e., to eat, travel, and take care of oneself). Sometimes disability has been defined in terms of the ability to perform, or inability to perform, the functional activities to live in our society. Insufficient attention has been given to the difficulties involved in relating disability to the capacity of a respondent to work. Also, sufficient thought has not been given to the two sides of the issue that must be involved in the employment contract—attitudes toward work and the capability of workers to perform in the workplace, and the conditions in the workplace and the flexibility of employer attitudes toward accommodating workers with disabilities. Needless to say, these are not easy questions and much more needs to be done in this area. This is the very issue with which the SSA is now struggling.

It is important to note the point made toward the end of Dr. Yelin's paper that ''. . . the question of how to assess functional capacity for work cannot be asked abstractly . . . it must be asked assuming a strong demand for labor and the presence of reasonable accommodation, as mandated by the ADA of 1990.'' Even then many will not be working—and that is the issue on the table at this workshop.

Dr. Yelin suggested that ". . . it may be possible to describe a core set of functional requirements that apply . . ." and then discussed the application of O*NET to the problem. O*NET is being developed as a replacement for DOT for the DOL. The DOL used the DOT and apparently expects to use its successor as a comprehensive database of work requirements for use in job training, job counseling, and job placement for the department's Employment and Training programs and for use by individual State Employment Security Agencies in the extensive work that they do with workers who need jobs or who have recently become unemployed.

Although O*NET is extremely useful for DOL's purposes, SSA's purpose in defining the functional capacity to work for purposes of the disability legislation is very different from the purposes of the DOL in creating O*NET. SSA's purpose is much more difficult. Moreover, the labor market and occupational literature indicate that there are many difficult measurement problems related to occupation and job characteristics. Information developed by job incumbents is not always consistent with the information developed by job analysts, and the information developed by job analysts is not always consistent with the views of worker supervisors. The BLS conducts employer surveys that try to define the characteristics of a job that affect its pay levels, but even there measurement

difficulties sometimes exist. In addition, from the perspective of the worker—as with a disabled individual—it is often a bundle of capabilities that the worker brings to the job that makes the work experience a success or a failure.

One of the major issues in income inequality today is the within-group occupational differences of people who are performing the same occupations, with the same educational backgrounds, and the same sort of capabilities, but who are being paid very different salaries because of how the workplace is operating. Experience has shown that workers with the same educational backgrounds have different skills, that changing work ethics and different work psychologies bring a different bundle of capabilities to a job, and that their performance is affected by those capabilities. In addition, the task of developing a set of factors for each occupation that makes practical sense is complex and difficult. Clearly, a great deal more careful research and experimentation is required to evaluate what functional capacity to work really means and exactly how it would be applied to persons with disabilities.

In conclusion, the issues discussed are important, but also complex and difficult. Constructive discussion of them could be helpful. It is useful, however, to apply three standards to most definitional and measurement issues:

1. Measurement can only take place when concepts are carefully defined in very specific terms and field tested.

2. We must always be sure that what we want to measure can be applied objectively without subjective determination.

3. The information must be reliable and reproducible, that is, persons with different assets and capabilities can effectively be classified in a reliable manner and that different, in this case, SSA assessors will reach the same classification decision.

The application of these standards requires experimentation and testing to ensure that the results will be accurate across different kinds of people. Occupations are much like the commodities and services included in a price index; each has a band of characteristics that result in quality determinants, and each quality determinant affects the price, which therefore, must be taken into account in producing the final index. Likewise, each person brings a different set of quality characteristics to the workplace.

Some of the key issues that surfaced during the general discussion are:

• A fourth standard for the definitional and measurement issues could be added to the three identified by Dr. Norwood. The disability commu-

nity, and particulary potential respondents, needs to be included in the development of survey instruments.

• Lack of comprehensive data on workers with disabilities is a serious concern. However, research and development work is needed on formulating the questions and survey design. Often, limited questions are found in general surveys conducted for purposes other than measuring disability and workers with disabilities. For instance, the March supplement of the CPS includes a question on disability, but its purpose is less about measuring disability and more about helping people who use the survey to determine who is in and who is out of the labor force.

• In recent years cognition has begun to play an important role in survey design. The movement to consider cognition in survey design, triggered by the National Science Foundation and the National Academy of Sciences, has taken hold with some agencies developing successful cognitive laboratories to investigate cognitive aspects of survey methods. Identifying the underlying cognitive difficulties that respondents experience in dealing with difficult tasks implicit in some survey questions, helps in improving the questions or procedures. BLS, the Bureau of the Census, and the NCHS are collaborating on cognitive work. The second interim report of this committee has recommended that SSA establish a cognitive laboratory to study questions that are asked in their survey and research activities in order to elicit improved responses and for other purposes of the agency (Wunderlich and Rice, 1998).

• Work history is one of the strongest determinants of current work status and future prospects. Some information is available on the effect of work experience prior to onset of disability on current unemployment status. The CPS supplement has a work history question and the Health and Retirement Survey obtains more systematic work history information.

• Given the large differences within the same occupation title, to what extent are the environment and demands of work capable of being generalized in categories? No data sets exist that provide information on accommodations that employers provide. In addition to looking at changes in the macro structure of employment, the micro structure of employment also needs to be studied. The HRS comes closest to doing that, but the sample for that survey is people 51–61 years of age in the baseline year. Therefore, it is not applicable to the bulk of the people of working age, who are under 51 years of age.

• In O*NET 1,200 occupations with a matrix of about 300 different characteristics are being developed. Yet a person brings to the job more qualities and characteristics than those of the occupation itself. Jobs can be modified so people with disabilities could do those jobs. It takes both an employer as well as a worker to construct the kind of situation that will take advantage of the particular capabilities of the worker. Every person has capabilities to offer.

• The need for clear definitions of concepts to be measured before attempting to measure them was underscored. It is imperative that SSA, in its redesign work, clearly define what is being measured to prevent continued comparisons of apples and oranges. Often similar terms are used that mean different things.