A Young Person’s Guide to Larry Katz, Part 2
What 35 years of groundbreaking research on education, neighborhoods, and inequality has taught us
Welcome back to Forked Lightning! Last week was part one of my tribute / explainer of the contributions of Larry Katz. Part One discussed Larry’s contribution to our understanding of how education and skills affect wage inequality. Part Two tells you all about Larry’s work on neighborhoods, with a particular focus on the Moving to Opportunity (MTO) demonstration. Both parts are derived from a chapter that David Autor and I just completed for the Palgrave Companion to Harvard Economics, entitled “A Young Person’s Guide to Lawrence F. Katz”. You should read the whole thing!
Moving to Opportunity
Larry Katz was a principal architect of one the most important social policy experiments in U.S. history – the Moving to Opportunity (MTO) demonstration. In the early 1990s many Americans were deeply concerned about the problems of concentrated urban poverty, and MTO gave us an answer to a very important policy question – what would happen if the residents of some of the poorest neighborhoods in the country were given the opportunity to move elsewhere? MTO was funded by a Department of Housing and Urban Development (HUD) appropriations bill, but the creation of the experiment itself was a multiagency process that also included the Department of Labor, where Larry was serving as the Chief Economist.
The MTO demonstration began in 1994 and ran in five cities: Baltimore, Boston, Chicago, Los Angeles, and New York. Nearly five thousand families living in public housing in the poorest Census tracts in the country were randomly assigned to one of three conditions 1) a “control” group who stayed where they were; 2) a regular Section 8 housing voucher group, received a subsidy to move to any neighborhood of their choice; and 3) a group receiving a voucher that could only be used in neighborhoods that were low poverty (this group also received housing search assistance).
The theory of change for MTO is embedded in the phrase itself - “Moving to Opportunity”.1 The idea was that moving to a better neighborhood would allow the (mostly female) heads of household to obtain a better job, leading to more success and opportunity for them and for their children. That did eventually happen, but not in a way that anybody expected.
Larry and his coauthors published the “final” experimental analysis in 2007. Voucher recipients – especially those in the “low poverty” group – still lived in safer, lower-poverty neighborhoods than the control group 4-7 years after random assignment. But they found no impact at all on economic outcomes ranging from employment to earnings to welfare receipt (and boy did they look).
Families moved, but not to opportunity. Instead, they moved to tranquility. The MTO team collected survey data and conducted lots of interviews, and they found that families who moved were motivated primarily by fear of violence. Voucher recipients felt safer in their new neighborhoods, and they reported fewer sightings of crime near their houses. Children were less likely to be victimized by crime or to visit the hospital for injury or ill health.
The final experimental analysis reported large and long-lasting gains in mental health and well-being. In another follow-up paper, they obtained physical health data and found lower prevalence of extreme obesity and diabetes among voucher recipients. MTO had made families better off, just not in a way that translated into economic opportunity.
This set of findings elicited a mixed, perhaps even confused, reaction from the social science community. Scholars who had spent their careers studying the importance of neighborhoods were disappointed in the lack of economic impacts. Some argued that MTO was a weak treatment because the new neighborhoods weren’t dramatically different from the old ones – in particular, voucher recipients often moved to neighborhoods that were less poor but still racially segregated.
Everyone wanted MTO to work. As a lead author and a scholar who was personally committed to understanding and alleviating poverty, Larry had every incentive to torture the data until it confessed, or to undermine the experiment because it did not turn out how he had hoped. He chose instead to focus on understanding why MTO didn’t increase economic opportunity.
And it’s a good thing he did. Nearly two decades after the study began, Larry and a new team of coauthors overturned the consensus about MTO by bringing new data to bear. In 2016, Larry (along with Raj Chetty and Nathan Hendren) published a study that linked the original MTO data to federal income tax records through 2012, 16 years after random assignment.
They first replicated the original results – there was no impact of MTO on the economic self-sufficiency of adult voucher recipients. The children were a different story. Like the adults, children who were already teenagers when their families received a housing voucher did not benefit from moving to a lower-poverty neighborhood.
However, children younger than 13 who moved because of MTO had 30% higher earnings as adults. They were more likely to attend college, and the colleges they attended were more selective. They lived in lower-poverty neighborhoods as young adults, and they were less likely to be single parents. Neighborhoods really do affect economic outcomes – but mostly for children, not adults.2
I came to Harvard as a PhD student in 2005 intending to study poverty and to work with one of the study’s main authors (Jeff Liebman), and I remember how the discussion of MTO played out in real time. I learned so much from the way that Larry handled this episode. A commitment to empirical social science means that you must follow the truth wherever it leads. And if you are never surprised by the results of one of your own studies, you are either cooking the books or asking the wrong questions.
Commitment to empiricism is more than just good scientific practice: it’s a moral imperative. Social programs like MTO – and Section 8 housing vouchers more broadly – have extremely high stakes. There is only so much money available to help people in need. As a researcher I might advocate for programs because they comport with my prior beliefs about what works or because doing so enhances my scholarly reputation. But if the program is ineffective, I’m actively harming people who need help by misdirecting resources. This is a very easy trap to fall into, and Larry has shown us all how to avoid it: follow the data wherever it leads, even if it seems to prove you wrong.
On a personal note, I am deeply grateful to Larry for taking me on as a student after my first two advisors left Harvard early in my career as a doctoral student. He read my papers carefully and always made time to meet with me despite his prior commitments to many other students. We bonded over our shared childhood love of baseball statistics.
I don’t think my story is unusual at all – as a matter of fact, I know several other contemporaries who had similar experiences. Larry is THE reason that the Harvard economics department has churned out so many successful graduates in applied microeconomics over the last few decades. His contributions to the field of economics are staggering. And most importantly, he’s one of the kindest people you’ll ever meet.
David Autor and I both felt as though we were writing on behalf of the hundreds of other beneficiaries of Larry’s wisdom and generosity. If you are one of the lucky ones, we hope that the essay did him justice.
Larry Katz, the world is better because of the research you’ve done and the lives you’ve touched along the way.
The motivation for MTO came out of the work of John Kain, William Julius Wilson, and others who argued that there was a “spatial mismatch” between public housing residents’ physical location and the location of good jobs that would lead to economic opportunity.
The pattern of impacts by age can be explained by a simple process where better neighborhoods have a positive “exposure” effect that is linear in time, but the act of moving itself imposes a one-time “disruption” cost. Exposure outweighs disruption for younger kids, but not for teenagers. Later, Chetty and Hendren validated this idea by using the tax data to study childhood mobility patterns of seven million families all across the U.S.