Patterns of Upward Mobility For Children of Poor Parents
Slide 70 From Slides For The Effects of Neighborhoods on Intergenerational Mobility II: County-Level Estimates
Raj Chetty and Nathaniel Hendren National Bureau of Economic Research Working Paper No. 23002, 2016
Part II of a two part series looking at intergenerational mobility. Part I focused on national and global trends. Part II focuses on implications of the variations in intergenerational mobility within the US.
Summary:
- Relative upward mobility for children whose parents are in the bottom income quintile is lower than would be expected in a world of "neutral advantage." How can it be improved? Insights are gained from reading eight research papers available on the Equality of Opportunity website which examine variations in upward mobility within the US. Those variations include differences by geography, teacher quality, and gender. In some cases, probable causality for these variations could be suggested by comparing families with children that moved from one area to another.
- The primary correlations with increased upward mobility for children of parents in the bottom 25th percentile of income by commuting zone were: 1) less segregation (racial, economic, and commute time), 2) lower income inequality, 3) better quality school systems, 4) stronger social capital, 5) stronger family structure and 6) better teenage labor force participation.
- Some of these correlates suggest that improvement in upward mobility could be encouraged by local and/or Federal government policies.
Geographical Variations Within The US
Differences in relative mobility across the US. allow researchers to try to isolate characteristics that seem to foster more upward mobility. Table 1 from the 2016 paper The Effects of Neighborhoods on Intergenerational Mobility II: County-Level Estimates by Chetty and Hendren illustrates both factors that were, and were not, highly correlated with differences in upward mobility for children of parents in the bottom 25th percentile of income.
These kinds of correlations are fraught with interpretational pitfalls as each factor often shows strong correlations with other factors and it is difficult to establish whether they are truly causal. An alternative to being causal is the possibility they may simply reflect “selection bias” or “sorting” (when people with similar socioeconomic characteristics live together).
Table 1: US Relative Mobility Geographic Correlates For Children Of Parents At 25th Percentile Of Income
Raj Chetty and Nathaniel Hendren National Bureau of Economic Research Working Paper No. 23002, 2016
Chetty and Hendren are able to increase the likelihood that factors they identify could be causal by looking at children who move from one area to another at different ages. By looking at movers, they can isolate the differential effect of how long a child was exposed to a positive (or negative) area. They also take into account the income of the child’s parents, the parents’ marital status and the area from which they moved. Following this procedure, they can estimate how much of an observed correlation is based on selection or sorting effects (see Column 5 in Table 1) versus how much seems to reflect causal exposure (see Column 4 in Table 1). Even so they are careful not to claim they have proved causality.
For the balance of this essay, when I refer to something as being “positively” or “negatively” correlated, I am referring to Column 4 in Table 1 that attempts to isolate correlations not associated with sorting.
Residential Segregation And Income Inequality
The strongest single correlation between upward mobility for poor children is living in an area with commute times of 15 minutes or less. Note that this is not their commute time, but their parents’. Lower population density, lower levels of segregation by income or race, and a lower concentration of African Americans are also associated with upward mobility.
While income inequality as measured by the Gini coefficient is quite negatively associated with upward mobility, the poverty rate was not especially strongly correlated with upward mobility. In fact, the share of top 1% income families within a commuting zone is negatively correlated with upward mobility for poor children. High rent and home prices are also negatively correlated with mobility. This is a recurring result: poor children do not need to live in wealthy areas or areas with high government expenditures/capita to do well. Indeed poor children may be at a disadvantage in high cost-of-living wealthy neighborhoods in which wealthy parents have little need for public education.
Upward mobility is explicitly correlated with having a high percentage of middle class parents. In short, it appears the best neighborhoods for upward mobility are relatively diverse suburbs which reflect average American compositions of income and race in which parents have enough time before and after work to spend time with their children.
Education
Living in an area with schools that produce high test scores and a low high school dropout rate (both controlled for parent income) is highly correlated with improved upward mobility. Conversely, a high ratio of students to teachers is negatively correlated with improved outcomes for poor children. But school expenditures per student is not correlated with improved upward mobility.
In a separate paper, Chetty and co-researchers document that educational environment makes a measurable difference beginning in kindergarten.[1] In kindergarten, small class size, capable peers and an experienced teacher produce better outcomes in areas like test scores, future income, college attendance and other outcomes. They are careful to explain that teacher experience per se is not likely to be causal (i.e., tenure) and that their study provides no basis for believing paying teachers more would improve outcomes.
In two later papers, Chetty and co-researchers analyze whether teachers that exhibit more skill in improving their students’ test scores (from their incoming baseline) have any measurable impact on children. “Value Added” or “VA” is the term used for this approach of measuring teacher skill.[2] They find that skilled teachers measured in this way have positive effects on their students future salaries, college attendance, and diminished likelihood of having children as teenagers. The impacts are large throughout grades 4-8. In a policy context, they observe in the second of the two papers (use of bold is mine):
“We also evaluate the expected gains from policies that pay bonuses to high-VA teachers to increase retention rates. The gains from such policies are only slightly larger than their costs because most bonus payments end up going to high-VA teachers who would have stayed even without the additional payment. Replacing low VA teachers may therefore be a more cost effective strategy to increase teacher quality in the short run than paying bonuses to retain high-VA teachers. In the long run, higher salaries could attract more high VA teachers to the teaching profession, a potentially important benefit that we do not measure here.”
While “teaching to the test” has become controversial, their research unequivocally indicates that children would benefit from any mechanism which increases the number of teachers who produce better relative outcomes that can be measured, while removing those that produce below average outcomes.
Experienced elementary and middle school teachers I know don’t like the fact that curriculums are now tightly controlled to optimize test scores, often to the detriment of their ability to teach more effectively. This is not a good outcome, but dictating curriculums can also be seen as a mechanism that tries to compensate for the inability to remove ineffective teachers due to union rules. Other controversial mechanisms to improve student outcomes include increasing parental choice through charter schools that compete for good teachers and parent funding based on their results.
Social Capital, Family Structure and Teenage Labor Force Participation
Table 1 shows that upward mobility is strongly positively correlated with social capital (e.g., strength of clubs, community associations, civic involvement etc.) and negatively correlated with violent crime. It is also very positively correlated with the percentage of married adults and negatively correlated with the fraction of children with single mothers. Finally, increased teenage labor force participation (14 - 16 year olds) is positively correlated with upward mobility for poor children.
Crime is an good example of how these factors are confounding. In a paper which looked at gender gaps in employment, earnings, and college attendance, Chetty and co-researchers found that males growing up in poor single parent families in high-poverty, high-minority areas were less likely to be employed than females. They speculated that “These areas also have higher rates of crime, suggesting that boys growing up in concentrated poverty substitute from formal employment to crime.”[3]
While it is not particularly clear to me how policies could strengthen voluntary associations like marriage and social clubs, they may be able to help suppress crime and increase employment opportunities for 14 to 16 year olds.[4]
Moving As A Solution
Many of the results summarized above were strengthened by studying children whose parents had moved. There are interesting nuances there. In their paper on neighborhoods, Chetty and Hendren found continuous linear benefit by years of exposure for children moving from a worse area to a better area (i.e., one year of exposure produced a constant benefit whether the child moved at age 8 or at age 15).[5] But, as they explained, that result was conditioned on parents who had decided to move.
An earlier paper by Chetty, Hendren and Katz analyzed a US Department of Housing and Urban Development program in the mid-1990s called “Moving to Opportunity” that offered vouchers to those in subsidized housing in high-poverty areas to allow them to move to less impoverished areas.[6] In this case they found there were benefits for children who were moved when they were 13 years or younger, but some net harm if children older than 13 were moved (perhaps due to the disruption of moving during high school). This was a direct comparison of children who had moved relative to those who had not.
Moving to the right area as defined by the characteristics we have discussed above appears to be a reasonable strategy. As Chetty, Hendren and Katz observe, expenses related to moving vouchers can be offset by higher future tax revenues from the increased earnings of those who have been positively impacted. However, the map shown at the top of this essay suggests it may be a limited strategy. Notably, a large contiguous section of the Southeast has poor upward mobility exposures -- it is certainly not feasible to move every child from that region or many similar large regions throughout the country. In some cases there may be county or borough level granularity that can help. But ultimately, improvements are needed at the local level throughout the country.
Improving outcomes by policy is challenging. Policies that attempt to improve outcomes must be measurable and not introduce new problems. In a world of finite resources, they must also be cost effective.
New Data, An Old Problem
The advent of “big data” can create an impression that we are discovering new problems, and hopefully new ways of solving them. When facing an issue like the inadequate upward mobility for children of poor parents, it is perhaps helpful to remember that it has a long pedigree. The purpose of remembering persistent history is most certainly not to give up on the problem, but to realize progress requires sustained effort and is measured in generations.
So long as there shall exist, by virtue of law and custom, decrees of damnation pronounced by society, artificially creating hells amid the civilization of earth, and adding the element of human fate to divine destiny; so long as the three great problems of the century — the degradation of man through pauperism, the corruption of woman through hunger, the crippling of children through lack of light — are unsolved; so long as social asphyxia is possible in any part of the world; — in other words, and with a still wider significance, so long as ignorance and poverty exist on earth, books of the nature of Les Miserables cannot fail to be of use.
[1] How Does Your Kindergarten Classroom Affect Your Earnings? Evidence from Project STAR Raj Chetty, John Friedman, Nathaniel Hilger, Emmanuel Saez, Diane Schanzenbach, and Danny Yagan Quarterly Journal of Economics 126(4): 1593-1660, 2011
[2] Measuring the Impacts of Teachers I: Evaluating Bias in Teacher Value-Added Estimates Raj Chetty, John Friedman, and Jonah Rockoff American Economic Review 104(9): 2593-2632, 2014 and Measuring the Impacts of Teachers II: Teacher Value-Added and Student Outcomes in Adulthood Raj Chetty, John Friedman, and Jonah Rockoff American Economic Review 104(9): 2633-2679, 2014
[3] Childhood Environment and Gender Gaps in Adulthood Raj Chetty, Nathaniel Hendren, Frina Lin, Jeremy Majerovitz, and Benjamin Scuderi American Economic Review Papers and Proceedings 106(5): 282-288, 2016
[4] There certainly should not be any penalties for marriage (e.g., in the tax code or in eligibility for benefits).
[5] The Effects of Neighborhoods on Intergenerational Mobility I: Childhood Exposure Effects Raj Chetty and Nathaniel Hendren National Bureau of Economic Research Working Paper No. 23001, 2016
Raj Chetty, Nathaniel Hendren, and Lawrence Katz American Economic Review 106(4): 855-902, 2016
Comments