Session 3. An In Depth Look at Current Statistics: Economic Outcomes
Trends in the Employment of Working-Age People with Disabilities: 1980-2004
David Wittenburg, Mathematica Policy Research, Inc.
I have done two things. First of all, I changed the title of my presentation, looking at the other presentations on the panel, made it more consumer based and it is really about measuring employment statistics for people with disabilities. One of the reasons why I did that is because while trends are extremely important, there are actually only a few datasets that provide us enough information that allow us to track trends over a long period of time. The second thing I did was, my son got me up around 4:00 this morning so I had time to further reedit my presentation. These are revised slides that you are looking at so I apologize for that.
My presentation has four sections. First I'm going to provide some background information on why we care about employment rates of people with disabilities and the different employment rates that are out there.
Next I'm going to talk about cross-sectional estimates of employment rates and specifically I'm going to talk about why those estimates are different and why we see so many different estimates out there.
You heard in Andy’s lunchtime talk as well questions from consumers specifically saying, "Which numbers should we use?" and this is a big issue because there are a number of different prevalence rates and a number of different disability employment rates.
Third I will talk about analyses of trends and state differences in employment rate -- really the topic of the panel. And finally I will conclude with a summary of results. There are three things that are important for my presentation. First, measuring employment statistics is really, really important. There are a number of policies that affect people with disabilities -- the Americans with Disabilities Act, the New Freedom Initiative and Social Security initiatives. These policies are going to get more important in future years so getting good measures of employment rates with people with disability is critical.
The second point is there are a lot of measures, and you can paint different pictures of the employment rate depending on the measure you use. You can get different pictures depending on the disability measure you use. If you heard Andrew's talk this morning, there are several different disability conceptualizations out there that range from severe to fairly broad categories of impairment. You can get very different employment rates depending on the employment measure you use, whether you use an annual measure or whether you use a very restrictive full-time during the week measure. And finally you can get a different measure depending on the data source that you use.
The last thing I want you to take away from the presentation is that one of the problems in reporting these employment rates is that people aren't very specific when they refer to people with disabilities -- they tend to lump them in one category. And the reality is we have several different disability programs out there that serve different pockets of the population with disabilities. We have Social Security Disability programs that serve a very severe population. We have the Americans with Disabilities Act and New Freedom Initiatives that serve a fairly broad population. Maybe one thing we need to do in reporting disability employment and prevalence rates is not use the term "disability" but rather use the term "work limitation, functional limitations and people with impairments" to define what we're doing.
Why measure employment of people with disabilities? The first bullet here is the gaps in service. This is the most common reason that people tend to use employment rates of people with disabilities. How do people with disabilities fare relative to other populations?
The other reason used in SSA demonstration projects across a variety of agencies -- what are the successful practices and policies for further studies? For example are there differences around state and localities that inform policy.
The third is the policy relevance issue. There are an increasing number of policies and programs interested in promoting employment of people with disabilities. The ADA is a landmark legislation that has led to numerous initiatives that focus on promoting employment of people with disabilities. This is steeped in Social Security Disability programs for people who got on the program with an inability to work and now there is a focus on returning people with an inability to work to work.
And finally general trends. Are policies and programs going in the right direction?
So what is the employment rate of people with disabilities? I went to ASK.com and asked that question. These are the answers. The employment rate of people with a disability is still a deplorable 60 to 70 percent. There are a number of articles by The Enquirer on October 29, 2000. That is the first hit. I like the deplorable 60 to 70 percent because we will shoot even lower.
Well, the employment rate of people with disabilities is only 56 percent. This comes from a Department of Labor ODEP newsletter. That is closer to the Enquirer estimate. Then the employment rate of people with disabilities hovered around 35 percent. Consumers are probably sitting out in the audience saying this is the exact sort of thing that frustrates the heck out of me. Why do we have all the different employment rates and what do we make of them. Well, let's talk about them, let's talk about the cross-sectional employment rates.
The employment rates measure we use are sensitive to several factors. The first factor is the disability conceptualization. We have a range of disability conceptualizations out there. There is not a single disability definition out there.
The second reason is we have got a variety of employment conceptualizations. People with disabilities, relative to almost every other demographic group, have a much higher incidence of part time or part year work. Depending on how you measure employment rates -- whether you use annual definition or weekly definition -- you will get a very different result. Finally there are the data sources. Some contain good health information, but weak employment information and vice versa.
Next slide.
So here is a slide from the SIPP Users Guide that I wrote with Sandy Nelson that looks at the annual employment rates for people with disabilities across a variety of disability conceptualizations, some of which you heard about this morning. The first bar shows people without disabilities. These are people who report no impairment, no functional limitations. It is a broad population. Over 90 percent are employed throughout the course of the year.
Next we move to the “any disability” category and this is similar to the SIPP number cited in the ADA. These are people who report any type of impairment, ADL, IADL -- so a hodgepodge of disability definitions.
The next three bars, which have sensory, physical and mental disabilities, refer to impairment status. These are more general measures of disability and they are not as close to the typical measure of disability that we use in programs because they don't measure the interaction of animpairment with a social activity such as work.
The final three measures, which are work limitations, IADLs, and ADLs are measures of activity limitations -- the interaction of an impairment with a social activity.
The thing that jumps out immediately when you use a fairly broad definition for any disability is that approximately 60 percent are employed at anytime during the course of the year. Similar to the Enquirer quote. When we use impairment definition, sensory, physical and mental impairments, you get a number between 45 percent and just over 60 percent, broad measures of health and impairment categories.
And the final categories -- and these are the categories that we think about in Social Security Disability programs as a measure of an impairment with a social activity -- you see much lower employment rates, between the ranges of 35 to 41 percent. So already we have a range just based on disability definition between 35 and 61 percent within the Survey of Income and Program Participation.
Now I take a look at employment rates for a single disability conceptualization, which is work limitation. This slide shows how the employment rate varies depending on the employment rate we are measuring. The first bar chooses any annual employment rate. What we see with people with work limitations, this is population with fairly severe limitations, 41 percent are working throughout the course of the year.
Now in the SIPP you can go down from any annual level changing the time frame to a single month. What you see is immediately the employment rate drops to about 27 percent. And we can further take down that employment rate to 17 percent by saying, okay, we only want people who are working full time within the course of the year.
Annual employment rates will also vary across data sources. Here the differences are important but not as large. The SIPP and the NHIS are two data sources talked about earlier by Gerry Hendershot. These produce employment rates of 41 percent. The ACS has about 28 percent. Does this make these measures bad? No. The work limitation measure appears in different parts of the survey. There are fewer health questions and as a result it picks up a smaller portion of the population.
Next slide.
Summary of the employment findings. Severity really matters. The employment time frame matters showing large differences -- lower rates for full time employment, here the rates range from 18 to 40 percent. Finally the data source. This has more moderate differences with lower rates and data that capture lower prevalence rates. This includes the CPS and ACS in lower employment rates relative to the SIPP.
What are the trends in state differences in measuring employment? There are several challenges that exist to measuring employment trends and this is frustrating to consumers. First of all, there are a limited number of data sources consistently collected with repeated information, this includes the CPS, the NHIS and the SIPP up to 2004 (which looks like it will be cancelled). The PSID, with a limited sample, and ACS that started in 2003. There is a trade off of health versus employment information. If you look at the NHIS you have more ability to measure employment trends but you have limited employment information and vice versa for the CPS.
And finally the surveys change their questions. For example, a lot of times people wanted to use the SIPP to measure trends. The problem is, a small change in question can lead to large changes in disability prevalence. Employment rates based primarily on activity limitations are the ones we can use. These are the ones most consistently repeated throughout each of the data sources. The following data sources have the most promising information on trends -- they include the CPS, the NHIS and hopefully the ACS, although we found problems with the early tables Cornell has been working on. In each of the data sources we found a significant decline in employment among men with disabilities. This decline started in the 1990s around the recession and it is consistent with a large increase in Social Security Disability program participation, which has increased by approximately 70 percent during this period.
In terms of state differences and measuring state employment rates, this requires the collection of representative samples at the state level. Few data sources actually do this. Ones that do include the CPS and ACS. I put up SSA administrative data but there are various administrative data sources that do that. We don't get detailed information at the state level which is frustrating because you want more detailed health information like the SIPP or NHIS. The information from the Cornell website, DisabilityStatistics.org, reveals large differences in the regional employment rate. Wyoming had a 51 percent employment rate whereas West Virginia had a 24 percent employment rate.
So which measure is best? The choice of measure depends on the policy question. For cross-sectional analysis of broad policies you have numerous options out there. You have to choose the best for that situation depending on what employment rate you want to measure. Do you want to look at people over the entire course of the year? Do you care about part-time employment -- if so you use a broad measure. What measure of disability do you want to use? For more targeted populations more severe definitions are better. These are, for example, Social Security disability policies. You want to use definitions like ADLs and IADLs which require more health focus specialized surveys.
Finally, which measure is best for state analysis? We are sort of stuck here. There are a few data sources that provide information. We need large samples at the state level. The problem there is a trade off in term of the information we can collect. And finally for trends we need a consistent measure over time. So I will end there.
CornellUniversity
Rehabilitation Research and TrainingCenter on Disability Demographics and Statistics (StatsRRTC)
Page1