A major predictor of becoming unsheltered is a slight worsening of the situation for those who are impoverished rather than a single, large economic jolt, says University of Chicago Economist Bruce Meyer, who lead a major study based on a national survey of homelessness. Meyer, who is an endowed professor in the Harris School of Public Policy, recently spoke with reporter Maureen McKinney as a feature of her ongoing series, Unsheltered. This is an edited, excerpted version of that conversation. The report can be accessed here.
I wanted to know, was there one major overarching theme that came out of the report?
I would say that the main, single point is that those who are homeless are in the midst of a very long period of severe deprivation, not like you sometimes hear reported in a story. Typically people who've had a relatively normal life and then a bad spell where they were homeless, and then maybe return to their previous life. Rather, they're in the midst of not a bad week or month or a year, but a really a bad lifetime with severe deprivation, low-income, high program receipt.
I really thought that some of those factors leading into homelessness were surprising and unexpected. Would you agree with that?
We expected to see more of an obvious decline in income and employment before we see individuals homeless, and we expect to see much more of a pronounced increase in receipt of government benefits. We did see some increase, particularly for SNAP, what we used to call food stamps. But otherwise, most benefit receipt tends to be trending upward in the case of disability benefits from Social Security or SSI and tends to be trending upward for benefits for Medicaid. But the changes over time are fairly slow, not a sudden change around the time of homelessness.
One exception to that we see when looking at people in Illinois, particularly in Chicago —– I should be clear that it's Chicago data — we see that it appears that homeless shelters are very effective in getting people on SNAP. So we see an uptick in receipt of SNAP benefits when people enter a homeless shelter. But that receipt rate, that elevated receipt rate doesn't persist. It looks like people have a hard time staying on benefits.
What also may be the second biggest surprise is that there are fairly high rates of employment among those experiencing homelessness: about half have some formal employment during the year. And we expect formal employment that appears on tax forms — 1040s, W2s — is significantly lower than true employment given the likelihood of informal employment among this group. So, it was not entirely surprising to us that we found that about half of those experiencing homelessness are doing some work.
But it was not a surprise completely because there were some other studies in the past that were not national, were not representative – not we think is reliable as our results or as comprehensive. But those suggested substantial rates of employment. So, I should also will emphasize, though, that these individuals are not earning a lot, it’s intermittent employment at low wages. That leads to median income of those who are unsheltered homeless of only about $5,500. And for those in shelters have about $7,500 over a year. So it's not a lot of income.
I was especially interested in the fact that black people in the survey seem to have higher incomes than white individuals. Can you speak to that?
Yes, that's one of the, I think, most surprising results to people. We see that in other outcomes that, in general, white individuals who are homeless tend to be worse off than black individuals — not only in income being lower. We see less attachment to government benefits and higher mortality among white homeless individuals than black homeless individuals.
We think it's related to higher rates of disability that we see in other research, not the paper that you read but in other work we've done. And we see in some small anecdotal studies that substance use is higher among whites who are homeless. So, we think it's a combination of higher mental and physical disabilities and higher substance use.
We think, on the other hand, that black individuals who end up homeless are more likely to be in that situation through economic problems, just long histories of difficulty finding a job with substantial income that, combined with maybe less support from a smaller or less resourced network of family and friends, means that it's more economic circumstances rather than behavioral health issues that lead black individuals to end up as homeless compared to whites.
You noted that a significant number of homeless people are left out of counts.
We think that the Census Bureau does really a pretty good job of counting people in the census every 10 years, and catching them in an annual survey, the American Community Survey that only surveys the sheltered, but most surveys that are used by the government do not try and count the homeless and get responses from them. So, it's really only these two, the decennial census, and the American Community Survey. And the American Community Survey does not try and interview the unsheltered who are about a third of those who are formally homeless. So most of our statistics ignore them.
But we were pleasantly surprised that when comparing different sources and doing checks on the actual counting of homeless done every 10 years by the Census, we thought they did pretty well. Where more than 90% of those who were in shelters in two cities, Houston and Los Angeles, were recorded in the Census, according to our estimates. So we weren't able to do a comparison nationally. But in those two cities, the Census Bureau seems to have done a pretty good job of at least counting the individuals.
Now, our research emphasizes that counting seems to be done pretty well, just getting a record of someone being homeless and in a particular area. And they're also asked some basic questions like, name and date of birth, and age and gender and race and ethnicity. But when in other surveys, principally the American Community Survey, people are asked about themselves, their income, their program received, their employment, they don't give good answers, even the housed. And that's what the larger body of the research that I've been working on for more than 15 years has pointed to that people no longer want to respond to surveys accurately.
So, what we do is get most of our information from government records from tax records, and payment records from Social Security, from SSI, SNAP, veteran's benefits, housing benefits – in all a dozen different government programs — we use those records to figure out people's income and program receipt because they're not, on average, very accurate in their responses in the surveys. So that's kind of a long winded way of saying, in short, the data we use, we think covers the homeless pretty well.
And we get accurate responses on income and program receipt because we're using government records that we're able to link to the individual records through the reports of name and date of birth of the individuals that were provided to the Census Bureau.
Now, I should emphasize that, for the most part, our statistics ignore the homeless; they're not included in it in most major surveys. So the one that's used to assess poverty does not try and include the homeless. That's also our source of employment and unemployment statistics and health insurance coverage and many other statistics. So, we know very little about the situation of the homeless. Prior, I think, to our work, at least in terms of having national statistics and accurate statistics, because the typical study of the homeless is a small, localized study of those who show up at a particular service location –- a soup kitchen or health clinic. So, it's a non-representative sample, and the responses aren't likely to be accurate because people don't want to take the time to give detailed and accurate responses to interviewers these days. And that's certainly going to be true of the homeless as well as other individuals.
Other surprise findings?
We find that a very high share of the homeless either work or receive some safety-net benefits. So, we find that over 90% of both the sheltered and unsheltered, — even a little bit higher than that 97% for the sheltered and 93% for the unsheltered — are either employed or receive some government benefit during the year. Typically, the benefit is Medicaid or SNAP benefits. So that, I think, is maybe the other key finding, the degree of connectedness to either employment or the government safety net. But despite these connections, there's this long-term deprivation, low income, high program receipt, that we see over a dozen years surrounding when people are homeless.
What would you say are some of the possible solutions to the problems?
Well, we do find that it looks like it doesn't take much of a shock to income to send these individuals into homelessness. On the flip side, it suggests that a small amount of income support when people are at risk of becoming homeless could have a substantial effect. There are some randomized control trials and some other well-designed studies that have recently found that that small amounts of money, $1000 or $2,000, supplied to those who are at risk of losing their apartment can be very effective. And our results are certainly consistent with those findings. Given that it seems that it doesn't take a large loss of income to lead someone who's chronically deprived, to end up as homeless.
