Editor’s Note: In a speech to the French Ministry of Labor, Social Affairs and Solidarity, Jeffrey Kling discussed the use of housing vouchers in the US.
Thanks for inviting me. It’s delightful to be here. I’m going to talk primarily about a particular project that I’ve worked on with a large team in the US. And just to give a little bit of context to this, there is a long tradition of having pretty large scale randomized experiments, particularly on US labor market policy and this study comes in that tradition, being a bit larger in scale and more closely affiliated with government policy than some of the things that Abhijit and Joshua Angrist talked about this morning.
Over the last forty years in the US there have been a long series of randomized social experiments, particularly in employment and training, and in welfare to work policy, and there is a nice catalog of all these in the Digest of Social Experiments. There are literally hundreds of trials that have been undertaken, and there’s an evaluation industry in the US. There are three major firms and hundreds of millions of dollars of business that they do in evaluating programs for the states and the federal government that have been undertaken. In Colloque Expérimentations pour les politiques publiques de l’emploi et de la formation – mai 2008 – Paris 42 the last six years or so, Abhijit was emphasizing the major developments in social experiments in the area of development economics. In the US, that has also been true in the area of education, where the Institute for Education Sciences has sponsored many new randomized studies.
The one I am going to be talking about today is in the housing policy area, in particular about the use of public housing vouchers. The idea is that in the US there are a large number of apartment buildings owned by the government where people can live for very low rent and you have very high concentrations of very low income people all living in these governmentowned apartment buildings. So people are very interested in the question of what happens if you deconcentrate poverty and if you offer people housing vouchers that they can use to move out of these government-owned units into private market apartments where there is a lower concentration of poverty around them.
You might think that this would be beneficial for people in families, especially some of the children who would be living in safer neighborhoods; they would tend to have role models who had more education or who were more attached to the labor force. On the other hand you might be worried that they would experience some adverse effects. For instance, if you used to be in the middle of your class then you move to a new neighborhood where you’re now at the bottom of the rung of academic achievement then how will that affect kids?
These are some of the questions that we were trying to address by looking at some data. What are the impacts of having families move to different kinds of neighborhoods? The way this experiment worked was that in 1994 to 1998 in five different cities in the US, families in public housing were eligible to participate and there were forty-six hundred families in this demonstration. One thing that could happen to you if you were one of the forty-six hundred families who were interested in participating was that you participated in a lottery and eighteen hundred people received what were called low poverty vouchers. This is something where you could move to a geographical area that had an area-wide average of poverty rate of less than ten percent, and then you could use this voucher to help pay your rent in that area and you also received some counseling to help you move into one of these areas. And so about half of the people that were offered these low poverty vouchers actually used them to move to a new place.
Then there is another group that received a traditional voucher which in the US is known as a Section Eight voucher, closer to two-thirds of the people who received one of these traditional vouchers that you could use to move wherever you wanted—it wasn’t geographically restricted to people being in a low poverty area—and a larger fraction of them used this one. Then there was a control group of people who continued living in US public housing who didn’t receive any new assistance through this housing voucher lottery. The families who are participating in this demonstration are primarily mothers with children who don’t have male adults in the household. Less than a quarter of them were employed at the time when this started, although over time many more people became employed. The results that I’m going to talk about really came from a large amount of data that we collected about five years after people had been offered these vouchers, and so this graph shows you, in the control group, Colloque Expérimentations pour les politiques publiques de l’emploi et de la formation – mai 2008 – Paris 43 some of the conditions in the neighborhoods, where you could see what the average poverty rate was, how many people reported being victimized by crime, or other statistics. And then the red bar shows you, for the low poverty voucher group, that these were much lower, so the average poverty rates are lower, the percentage victimized by crime falls from twenty-one to thirteen and so on, so that the local conditions were substantially improved.
For the adults, one of the things they were interested in testing was whether if you moved to a lower poverty area, people would find it easier to obtain employment, perhaps because the labor market was better there, or because they would have more connections to people who were working who would help them find jobs. That turned out not to be true. There is not an appreciable impact on labor market outcomes of the adults. They are a bit more healthy. Particularly in terms of mental health, the adults are doing better. Similarly for youth, people were interested in whether being in potentially different schools or having different peer groups was going to have an impact on educational achievement. That also does not seem to have borne out in that the test scores are not particularly different in math and reading between the control group and the low poverty voucher group.
There were much more substantial differences in some things. The teen girls were much less distressed, much less likely to use marijuana, had fewer behavior problems, and were less likely to have been arrested for crime. The opposite seems to have been the case for the teen boys. The red bars are the control group, and the yellow bars are showing that they are more likely to use marijuana, more likely to have behavior problems and more likely to be arrested for crime. So there was this particularly unexpected gender difference. Joshua Angrist talked a bit how in the educational incentives it seemed that the girls were the most responsive, here not only are the girls more responsive but there actually seems to have been some adverse effects on the boys.
To summarize, there’s improved housing, increased safety, and lowered adult depression. There are even lower rates of adult obesity that were accompanied by slightly higher rates of exercise or they were more likely to eat a healthy diet. It seems to have been good on most dimensions for the teen girls, not so good for the teen boys and had little effect on employment or on kids’ achievement.
In terms of thinking of the cost benefit analysis for something like this, from the government’s point of view, they are paying about the same for having people in public housing units or paying for the vouchers that people used to move to different areas. If you think of the positive effects for teen girls, then the negative effects for teen boys are basically a wash. And since the higher mental health of the adults is a benefit, then you might say this is beneficial overall, although it’s hard to make this characterization, because it’s hard to say exactly how to weigh the fact that the boys seem to be somewhat worse off.
What we are doing now is to look at much longer term impacts. We are just starting a survey now, where we will be looking ten to twelve years after people have received their housing vouchers, and collecting some data. So we are interested in what the long term effects are, how they evolved over time, and especially, what the impacts on kids were, who were very young when they received a housing voucher. So if you were one or two years old and your Colloque Expérimentations pour les politiques publiques de l’emploi et de la formation – mai 2008 – Paris 44 family moved, then we will look at what’s like for you to have spent your whole life in one of these areas relative to other kids in families who were not offered a voucher and didn’t have this opportunity to live in a different kind of place.
Some of the mechanisms we are trying to look at more are the social ties and whether there was an important difference for boys versus girls, and at their connections to adult role models. We have some preliminary evidence that it was more disruptive to teen boys to have moved away from, say, their uncles or their mother’s boyfriends, or other adult male role models; they didn’t seem to have these role models after they moved to low poverty areas. We’re trying to dig into that some more to understand what was happening to them. We’ll be using a variety of different types of data so this gets to some of the innovations of measurement Abhijit talked about this morning. We will do some administrative data, looking at some public assistance receipts, employment and earnings, some survey data where we can craft specific questions so that we can get at things like the adult role model questions I mentioned a second ago. Some things that are more specifically about physical health, that is, height, weight, and waist measurements, blood pressure, trying to look at some of the early precursors in blood samples for cardiovascular disease, having kids do some math and reading achievement tests for us, and looking at what is happening in the neighborhood. So we will look again at the same categories for which I showed you graphs, in terms of education, employment health and risky behavior measures and try to learn what we can in the long term about what the impact of moving to new areas has been. Thanks.