Frequently Asked Questions
  1. Why are the City Rankings of Upward Mobility in Phase 2 of the project different from those in Phase 1?

    Phase 1 of the project reported raw statistics on upward mobility for each commuting zone (CZ) in the United States. The differences are driven by a combination of differences in the types of families living in each area (e.g., differences in demographic makup and wealth) as well as causal effects of each area on a given child. In phase 2, we isolate the causal effects of each county and CZ in the U.S. by studying families that move across areas. The causal effect estimates can be interpreted as the answer to the question, “How would my child do on average if he/she were to grow up in a different county?”

    Although the causal effects estimated in phase 2 are highly correlated with the raw statistics on upward mobility reported in phase 1, there are several differences driven by variation in the types of people who live in each area. For example, the New York area ranked highly on rates of upward mobility in phase 1, but its causal effect on upward mobility for a given child is not much better than other cities such as Washington D.C. New York has high levels of mobility on average because it has many immigrants, whose children tend to have high rates of mobility irrespective of where they grow up. But this does not mean that moving to New York will improve a given child’s chances of moving up.


  2. How do you identify the causal effect of each county? How can you predict how a child would do growing up elsewhere when you do not see that hypothetical?

    We estimate the causal effect of each county by comparing the experiences of families that move across areas. For example, we find that children in low-income families who moved from Manhattan to Queens at a younger age tend to earn more as adults than those who made the same move at older ages. This information allows us to infer that Queens has beneficial effects for children relative to Manhattan. We use a statistical model to combine information from such comparisons across all counties in America to estimate the effect of each county on children’s mean earnings.


  3. Why do you use data starting only at age 9 to estimate childhood exposure effects?

    Our data begin in 1996 and currently go through 2012. The earliest age at which we observe children who move in our data is currently nine. In future years, we will be able to estimate childhood exposure effects at earlier ages.


  4. Why do children who move to low-poverty areas as teenagers in the Moving to Opportunity (MTO) data seem to do worse as adults, while the national quasi-experimental study finds positive effects of moving to better areas until age 23?

    MTO compares movers to non-movers, so the impacts include potential disruption effects of moving. In contrast, the national quasi-experimental study compares the effects of moving to a better vs. worse place within a set of families that move. The national study says, “If you have to move as a teenager, it’s better to move to a good neighborhood.” MTO says “you probably should not move as a teenager to begin with.” The findings of the two studies are therefore consistent with each other, and both studies show that the childhood environment plays a key role in upward mobility.


  5. Areas with larger African-American populations have lower levels of upward mobility. Why is this, and what is the role of race in upward mobility?

    We do not observe race in our data, so we cannot directly measure mobility patterns by race. However, we do find that moving to an area with a larger African-American population reduces the prospects of upward mobility for a given child. In particular, we see that when a family with two kids moves to such an area, the younger sibling earns less in adulthood on average. Since race doesn’t vary within families, this shows that cities with large African-American populations reduce children’s odds of reaching the middle class – regardless of whether they are black or white.

    Cities with large African-American populations tend to be more segregated and have lower levels of investment in public schools and other public goods. This may explain why these cities generate poorer outcomes for children of all races. Irrespective of the root cause, it is clear that disparities across neighborhood amplify racial inequalities in the United States. We estimate that 20% of the earnings gap between black and white adults is explained simply by the county in which they grew up.


  6. What is Absolute Upward Mobility?

    Absolute Upward Mobility is a measure of the average economic outcome of a child from a below-median income family. Statistically, we define absolute upward mobility as the average percentile in the national income distribution of a child who is born to parents at the 25th percentile in the national income distribution. In areas with higher absolute upward mobility, children from low-income parents earn higher incomes on average as adults.


  7. What is Relative Mobility?

    Relative Mobility measures the difference in incomes between a child from a low income family vs. a high income family in a given area. Statistically, we define relative mobility as the average percentile in the national income distribution of a child who is born to the richest parents (top 1 percent of the national income distribution) minus the average percentile of a child born to poorest parents (bottom 1 percent). Smaller values of this statistic correspond to greater relative mobility, i.e. a smaller difference in outcomes between children from low vs. high income families.


  8. How do you define the Odds of Reaching Top Fifth Starting from Bottom Fifth?

    We start by taking the set of children whose parents are in the bottom 20% of the national income distribution. We then calculate the fraction of this group who reach the top 20% of the national income distribution in each area.


  9. What is a Commuting Zone?

    We divide the U.S. into 741 Commuting Zones. Commuting Zones are groups of counties that are defined based on commuting patterns. For example, if people in neighboring counties work in the same city, then those counties are likely to belong to the same Commuting Zone. Commuting zones are similar to metro areas, but have the advantage of covering rural areas as well. Note that each Commuting Zone is typically named after the biggest city in that zone. Hence, our statistics reflect average outcomes in a broad area around that city and not just that one city itself.


  10. Whom can I contact if I have questions about the study or the data?

    Please email us at This email address is being protected from spambots. You need JavaScript enabled to view it.