Session 3. An In Depth Look at Current Statistics: Economic Outcomes
Trends in the Income of Working-Age People with Disabilities: 1980-2004
Robert Weathers, Mathematica Policy Research, Inc.
What you will hear in my presentation will correlate to what Dave talked about in employment. This presentation is based upon the User Guide series that Andrew mentioned earlier, as well as work that Rich Burkhauser and Andrew Houtenville have done at Cornell.
Next slide.
To emphasize the importance of measuring trends and income over time I thought I would start with a quote from Franklin Roosevelt, "The test of our progress is not whether we add more to the abundance of those who have much, it is whether we provide enough for those who have too little.” I think that captures what I will talk about today in terms of trend in income for people with and without disabilities.
What I will do is do four things with this presentation. First I'm going to talk briefly about disability and income measurement, and I will describe the data sources and provide cross sectional statistics on household income for persons with disabilities and by the disability types we have been using throughout the day.
I'm going to make comparisons across datasets like Dave did with the employment rate and I will finally talk a little bit about tracking progress over time and income using data from the Current Population Survey that uses a rather limited definition of disability but yet provides us with good information about how the economic well-being of persons with disabilities is going over time and comparing that to the economic well-being of persons without disabilities over time.
Next slide.
In our User Guide series we identified disability as having one of the following six concepts. A sensory impairment, which is difficulty hearing or seeing, a physical impairment, which is difficulty moving, climbing, walking, etc., mental impairment, which is difficulty learning remembering, concentrating, activities of daily living which are difficulties doing things like taking care of yourself every day, bathing getting out of bed, etc, and instrumental activities of daily living and work limitations.
However, within each of these concepts there are multiple dimensions of how to measure them. One is the environment, which we heard about earlier today from Andrew and Gale. The other is the duration of the disability. Some surveys ask whether the person has had a condition lasting six months or more; some surveys don't ask the qualifier.
The third is severity. Some surveys will ask, “Do you have difficulty doing this activity,” and others will ask “Do you need help from somebody to do the activity,” and those are two different measures of severity. The point is that the national surveys differ in the way they ask these questions and as a result you get differences in disability measurement.
Next slide.
Now for income measurement -- and unfortunately, income measurement is not clear cut either. There are a number of concepts you need to consider, one is the sharing unit, do you care about the person, the family or the household. Family income is used in our poverty measure that the Census Bureau uses every year. It is also used to administer a lot of the means-tested programs that the government does.
Household income is used by the Census Bureau every year to report economic well-being in their report in August. But then there is another concept which is household size-adjusted income, which takes into account that there are differences across households in the number of people living in the household, adjusts for that, and makes comparisons across households of different household income sizes.
For example, you could have a family of four earning $40,000 and you generally do not consider them as well off as a person living alone with $40,000 because that person has to only take care of themselves whereas the family of four has four people.
But at the same time you don’t want to say that each member of that household is just as well off as a person who only makes $10,000 a year because there is some sharing going on, they share a residence, can take advantage of economies of scale at dinnertime by shopping at Sam's Club. We need to take into account economic well-being adjustment of household size.
Today I will use a common household size adjusted measure that has been put forward by Patricia Ruggles. The point is that when you see these numbers you are going to see that when I adjust for household size it is going to be a lower level than when you see household income as a whole because that is not taking into account household size whatsoever.
The other key point is that the surveys use different methods to measure income. Some surveys will ask individual components of income -- for example, how much did you earn last year, how much do you receive in disability transfer programs, how much do other members of your household make. And they will ask all these individual questions and aggregate it and come up with a household income measure.
But there are other surveys that just use one measure -- how much was your family income last year. These differences can lead to differences in income measurement and it is important to take these into account when you are looking at income measures.
The datasets I'm going to talk about today are the American Community Survey, which contains six questions used to identify disability and eight questions that ask each member of the household about the sources of income they receive. I will also talk a little bit about the Current Population Survey, which has the most detailed information on income. It has 18 individual questions and they are aggregated to the individual and up to each member of the household to get a household income measure. The limitation is that the Current Population Survey only has one question on disability and that is the work limitation question that captures a narrow population of persons with disabilities.
The National Health Interview Survey has great information on health, which I heard about in Gerry Hendershot’s presentation, but contains little information about income. It has cues which ask yes or no to sources of income so a person can identify what kind of income they are receiving and then they ask a global question, which is how much income has your family received during the past year. It does also ask a question about earning over the years as well but those are two different concepts and you can't take the earnings and aggregate it up to income.
And the final one I will talk briefly about, because unfortunately it is going away, is the Survey of Income and Program Participation. It captures income on a monthly basis and you can aggregate it and it captures it from a number of different sources similar to the Current Population Survey. You can aggregate it up to get a monthly earned income level for any individual and aggregate it over months to get annual measure.
But the bottom line is that I'm going to start off with the American Community Survey because it strikes a nice balance between measurement of disability and measurement of income. It is not the best survey for measuring disability; it is not the best survey for measuring income; but it is the best that captures both those concepts.
I just want to briefly show you what sources of income it collects. It asks eight different questions about each member of the household, about the wage or salary income, about net self-employment income, interest income, Social Security income, supplemental security income, public assistance and welfare payments, retirement survey and disability payments. It has an all other income sources category and it gives some examples -- Unemployment Insurance, Workers' Compensation, etc. It also allows you to measure income at the person level, the family level and the household level.
Next slide.
This slide does two things. First it shows measures of household income not adjusted for household size and those are in blue. What we do is we look at that between first those without a disability, which is a little over 60,000 per year and this is in 2005. And it compares that to the blue line for the bar that says disability, which is 34,800. So you see a big difference between those with and without disabilities for household income. When we adjust for those in the red bar so we have household size-adjusted income, you see the values go down but there is still a significant gap.
The other thing to take a look at here is as you saw with the employment presentation, those people with a sensory disability or physical disability tend to have higher incomes and do better and have higher employment rates than those with mental impairment, activity of daily living impairments, IADLs, or works limitations.
Next I want to put up a slide that showed another nice feature of the ACS -- you can get state-level estimates. This shows the absolute values of household size-adjusted income for the ten lowest states on the left and the ten highest states on the right. The key here is that it is very low in that area of the country that Andrew identified earlier as the disability belt and then you have other areas where it is pretty high -- over 30,000 dollars. And I think that is in Maryland. There is a lot of variation across states. As Andrew mentioned earlier, you have a large enough sample size in the ACS so these are statistically different across measures and household size adjusted income.
You might not be interested in the absolute values. You might really care about how persons with disabilities are doing in terms of household size-adjusted income compared to persons without disabilities and this is what this slide shows. For example, in the District of Columbia on the far left of the graph, you see the lowest percentage of persons with disabilities income compared to persons without disabilities. Persons with disabilities have a household size-adjusted income that is 38 percent of persons without disabilities. You see a lot of a grouping right around 50 percent and that is about what the national average is in the United States. But then you see also some states on the far right that have very high, relatively speaking, incomes as a percentage of those without disabilities.
Next slide.
Briefly, I would like to talk about comparisons across datasets. If you go to the next slide, this is just people without disabilities and people with disabilities across the American Community Survey in 2003, the CPS March 2004, which captures 2003 income, so it is comparable to ACS in terms of income year, the NHIS and SIPP 2002. For the SIPP we will stick to 2002 because of disabilities measures we were using. Not a comparable measure for later years.
But what you see are differences across income both for those without a disability and those with a disability. Part of that is due to the fact that we have differences in the concept of what a household is for the ACS and the CPS. For the NHIS and SIPP measures you are getting a family as a household. You have a different concept in terms of the household partly due to differences in the definition of disability and partly due to differences in the definition of income.
So the next slide, one of the nice things about going into these individual concepts is that you can cut down the variation in disability measure by going into these different definitions of disability. What you see when you go to these is there is less variation within the employment disability type for the CPS and the ACS. And it also shows why the CPS number for overall disability is low because you don't capture disability for many of -- you don't capture those with other conditions.
Next slide.
I’ll skip this and jump right to the trends so if we go one more slide.
Briefly I want to talk about the Current Population Survey and the importance of having time series in order to track progress of persons with disabilities and without disabilities.
Here we are really limited in our definition of disability to the work limitation, but what we have is a long time series from 1979 to 2004. And in this graph the red line is persons without disabilities and the blue line is persons with disabilities. What you see are rises and falls in income of people without disabilities that follow the business cycle. But what you see is a relatively flat pattern for people with disabilities. The bottom line is that if we compared it from trough to trough in the cycle, you see persons without disabilities having this increasing trend over time, persons with disabilities remain flat. That doesn't show very much progress in terms of the economic well-being of persons with disabilities.
If you go to the next slide this shows the three different points in time the composition of household income for people with disabilities in the CPS or people with work limitations, I should be very clear about that. What you see is something that reflects what Dave mentioned with the trends of employment. There is declining employment and earnings for persons with a work limitation from trough to trough to trough and it almost is cut in half.
And the other interesting thing is if you go up to the yellow bar, which captures disability payments in SSDI or SSI, you capture another thing that Dave said -- increasing trend toward enrollment in the SSI program that makes up a larger share of income over time for household of people with work limitations.
To summarize, there are differences across surveys due to definition differences, but also related to the differences to the measurement of income in surveys.
We think the ACS strikes a pretty good balance between definition of disability and income measure, but it is relatively new and we don't have a good manner of tracking changes in household income over time, which is unfortunate. As we move forward in time we will be able to use that as really effective tool.
The Current Population Survey is the only one that is able to capture the long-term trends in income but it is really limited to the work limitation measure of disability and that somewhat limits it in terms of what it can say about the broader population of persons with disabilities.
In the handouts before I leave, I also have a couple of additional slides that talk about areas of future research. I'm not going to go into them now for time, but they are areas that we should be taking a look at in order to improve our income measure and see how that affects the income measurement for persons with disabilities. Thank you for your time and I look forward to questions.
CornellUniversity
Rehabilitation Research and TrainingCenter on Disability Demographics and Statistics (StatsRRTC)
Page1