Research Projects

The following is a sampling of CPI’s research initiatives. More information about our research activities is available on individual research group pages.

The “absolute mobility” of the poor: In a project undertaken by Raj Chetty, David Grusky, Max Hell, Nathan Hendren, Robert Manduca, and Jimmy Narang, new evidence on trends in the amount of absolute mobility in the U.S. is provided by combining de-identified tax returns, decennial Census data, and CPS data. Among poor parents, there is much ethnographic evidence suggesting that they care deeply whether their children will exceed their own (very low) standard of living, but surprisingly we don’t know whether these aspirations for their children are frequently realized. What fraction of poor children grow up to earn more than their parents? Have rates of absolute upward mobility changed over time? This project, which develops a new method of estimating rates of upward mobility for the 1940-1984 birth cohorts, will tell us whether rates of mobility out of poverty are mainly affected by policies that change rates of relative mobility, policies that increase aggregate growth rates, or policies that reduce inequality.

The American Opportunity Study: This Social Mobility research group is developing a new administrative infrastructure for monitoring mobility and evaluating programs and policies. The American Opportunity Study (AOS) is an initiative that entails (a) linking individual records across the 1960-2010 decennial censuses and the 2008-2015 American Community Surveys (ACS), (b) appending further information to these linked records from other administrative sources, (c) making intergenerational matches between parents and children within the resulting AOS, and (d) linking to other surveys that are large enough, have individual identifiers, and for which consent to link has been obtained. This approach, which is being led by the U.S. Census Bureau, will yield a low-cost intergenerational panel based entirely on existing data. It will provide the U.S. with the capacity to monitor longitudinal processes and allow for evidence-informed policy on mobility, opportunity, and other labor market outcomes. 

Arrests, race, and poverty: It is well established that a criminal record reduces the chances of employment and that blacks are more likely than whites to have a criminal record. These effects are very large: That is, there is a large effect of a criminal record on employment, and there is likewise a large effect of race on the chances of having a record. It follows that the race gap in employment could be reduced substantially by eliminating the race gap in the likelihood of having a record. How, then, might that latter gap be reduced? In a project led by Stephen Raphael, it’s hypothesized that the gap is partly attributable to racial differences in the process of converting an arrest into a record. This hypothesis can be assessed by linking the Multiple Arrest and Citations Registry (i.e., microlevel records of arrests) with the Automated Criminal History System (i.e., microlevel records of criminal booking). The former data reveal many more interactions with law enforcement than are ultimately converted into an official criminal record. For example, some arrests result in a citation or are resolved informally, with the implication that they don’t generate a criminal record. Are blacks more likely than whites to be booked after an arrest (even after taking into account the possibly different circumstances behind the arrests)? If so, it suggests that “booking reform” could reduce the race gap in criminal records (and hence the race gap in poverty), all at a very low cost. Because there are many domains, in addition to employment, in which criminal histories may be relevant (e.g., access to housing), the effects of this reform could be substantial.

The biological mechanisms of disadvantage: The disadvantages that arise from prenatal and postnatal exposure to poverty partly arise from biological processes. When children are raised in poverty, they are exposed to stress that then alters their biological development and influences their health, learning, and behavior over the life course. This biological embedding may arise, in part, because children exposed to poverty-induced stress experience epigenetic changes. Do these epigenetic changes affect health? And is DNA methylation, a stable yet sensitive epigenetic mechanism, an important source of elevated risks of chronic disease for those exposed to poverty and its associated stresses? Using the National Health and Nutrition Examination Survey (NHANES), David Rehkopf is addressing these two questions as an entry point into the growing debate over just how important such epigenetic processes are. If indeed Rehkopf finds that DNA methylation is playing a large role in explaining the effects of poverty-induced stress, it will be an important development in identifying how poverty “gets under the skin” and creates lasting disadvantage.

California Poverty Project: Although both the SPM and OPM are regularly used to characterize poverty at the national level, neither is well suited for the purpose of assessing (a) how poverty is changing at the state or local levels, (b) whether existing state or local policies are making much headway in reducing poverty, and (c) the extent to which new state or local proposals and initiatives would reduce poverty. The U.S. is accordingly in the untenable position of having committed itself to a decentralized policy regime without also having the capacity to measure child and adult poverty at the level at which policy is being developed and policy “experiments" are being undertaken. We have therefore built an SPM-style poverty measure, dubbed the California Poverty Measure (CPM), in California. In collaboration with the Public Policy Institute of California (PPIC), we have so far issued two years of CPM data, and we are currently exploring ways to upgrade the CPM by taking advantage of the California Longitudinal Administrative Database (CLAD). The new CPM will make it possible to better assist California legislators and other policymakers in designing poverty policy. 

California Welfare Laboratory: The Safety Net research group, in collaboration with Charles Varner and Pablo Mitnik, is responsible for building and maintaining the California Welfare Laboratory. The threefold purpose of the California Welfare Laboratory is (1) to inform the public, policymakers, and practitioners about research on poverty and the safety net in California, (2) to provide recent trend data on poverty, inequality, and safety net use in California; and (3) to monitor and evaluate recent developments in poverty and the safety net in California. The mission of the laboratory is to make research on California’s welfare programs accessible to all and thus facilitate an informed discussion of what is working and what needs to be improved.

Colleges and rising income inequality: The ongoing increase in income inequality has led to new worries that (a) children from poor families are finding it difficult to beat out children from well-off families in the competition for admission to desirable colleges, and (b) the colleges that poor children do end up attending may not be delivering much in the way of labor market and poverty-protecting returns. These two worries are being addressed in a project carried out by Raj Chetty, John Friedman, Emmanuel Saez, Nicholas Turner, and Danny Yagan. The analysis, again based on de-identified tax data, will reveal where students from rich and poor families attend college, which colleges project poorer students into the upper part of the income distribution, and what explains the substantial variation in outcomes across similarly-ranked colleges. The researchers will characterize each college’s contribution to upward mobility by constructing a “Mobility Report Card” that conveys the joint distribution of parent and student incomes for every Title IV institution in the United States.

Deep poverty and TANF add-ons: The CPI is actively involved in research examining the effects of various types of TANF supplements. For example, David Grusky, Jonathan Fisher, and Charles Varner are carrying out a project on TANF “add-ons,” where these refer to the additional programs that California nonprofit organizations use to supplement TANF programming. Because of California’s very heterogeneous TANF landscape, the State is implicitly running dozens of natural experiments on TANF add-on policy, yet these experiments are not typically rigorously evaluated. The researchers will conduct quasi-experimental evaluations to evaluate add-on programs and assemble high-quality evidence on how TANF could be reformed to reduce rates of deep poverty.

Differential EITC effects: In collaboration with David Figlio, Krzysiek Karbownik, and David Simon, Hilary Hoynes is examining how the effects of the EITC on a child’s cognitive development may differ by age of exposure. Although in principle the EITC, as a pure earnings subsidy, affords an attractive test of the early-intervention hypothesis, analyses to date haven’t been able to determine when an additional increment of EITC matters the most for children. By matching vital statistics data to administrative data for school children born in Florida between 1992 and 2002, it becomes possible to ascertain birth date, birth place, mother’s demographic characteristics, birth order, test scores (for grades 3 to 8), and behavioral outcomes (suspensions, absence rate) for approximately two million children. The resulting research design will exploit variation in the generosity of the federal EITC to examine the effects of contemporaneous exposure to the EITC, the effects of longer term cumulative exposure, and the differential effects by age of exposure.

Disability and poverty: The federal government provides income protection to individuals with disabilities via the Social Security Disability Insurance program (SSDI), the Supplemental Security Income program (SSI), and the Disability Compensation program (DC) for veterans. In 2014, approximately 17 million disabled adults were enrolled in at least one of these programs, with benefits totaling approximately $250 billion. These disability programs are thus very large and are accordingly a major part of the country’s effort to reduce deep poverty. In evaluating their effectiveness, it is important to determine whether the low employment rates among program recipients reflects an underlying (low) capacity for employment, as opposed to the labor-supply effects of the programs themselves. Although there is much research on this question, it has been difficult to estimate the labor-supply effect of SSDI and SSI because the eligibility rules and benefit formulas are uniform nationally. In an ongoing research program, the leader of CPI’s safety net research group, Mark Duggan, is exploiting a change in medical eligibility criteria for the DC program (i.e., adding diabetes for “in-theater” veterans to a list of covered conditions), a change that allows him and his coauthors to identify the labor supply effects of disability programs (by comparing in-theater and not-in-theater veterans). The initial analyses, carried out by combining administrative data from the U.S. Army, the Department of Veteran Affairs, and the Social Security Administration, indicate that benefits receipt reduced labor force participation by 18 percentage points, even as net-of-transfer income rose. These results are now being supplemented with additional analyses exploring a wider range of outcomes (e.g., self-employment, spousal labor supply) and longer-run employment effects. The results will speak to whether modifications that reduce any undesirable labor-supply effects should be considered.

Distributional National Accounts: In an ambitious infrastructural project, Emmanuel Saez and his team are building a “Distributional National Accounts” based on tax returns, a new data set that will eliminate the current gap between (a) national accounts data based on economic aggregates and (b) inequality analysis that uses micro-level tax data to examine the distribution of income but is not consistent with national aggregates. When this integrated micro-level system is assembled (using micro-level tax data available as early as 1962), it will be possible to understand long-term trends in the relationship between overall growth and the distribution of that growth. The resulting data set will include (a) individual and family pre-tax and post-tax income, (b) wealth estimated using the Saez-Zucman capitalization method, and (c) imputed cash transfers and all in-kind transfers and public goods. It will thus be possible to distribute total National income both on a pre-tax market income basis and on a post-tax, post-government spending basis. These data will allow the Bureau of Economic Analysis (BEA) to include distributional statistics in the National Accounts that are based on high-quality tax data rather than inferior survey data. This will in turn make it possible to evaluate the extent to which economic growth, which has long been represented as a preferred poverty-reduction approach, is indeed delivering on that objective.

The effects of TANF: In California, those in poverty face a very heterogeneous TANF landscape, not just because of county-level differences in programmatic decisions but also because many counties benefit from nonprofit programs that supplement TANF with their own “add-on” programs. As a result, the State is implicitly running dozens of natural experiments on TANF add-on policy, yet these experiments are not typically rigorously evaluated (given the high costs of doing so). For example, some “TANF add-on” programs enable qualified families to buy a car, allowing them to pursue jobs that cannot easily be accessed with public transportation. By garnering the permission of participants, the California Longitudinal Administrative Database (CLAD) can be used to identify a “car effect,” most obviously by comparing car recipients against non-recipients who are still on the (very long) waiting list for a car. This type of quasi-experimental evaluation can likewise be carried out for other add-on programs and thus allow us to assemble high-quality evidence on how TANF could be reformed to reduce rates of poverty.

Evictions and poverty: The leader of CPI’s Housing research group, Matthew Desmond, is carrying out the country’s leading research on evictions and poverty. In an ongoing project, Desmond is assembling public records of all eviction records from every county in the U.S., thus providing the first comprehensive national estimate of eviction frequencies. This analysis of prevalence will be joined to the first causal analysis of the effects of evictions on poverty and labor market outcomes. The key question here: Are evictions an important cause of deep and extreme poverty? In collaboration with Raj Chetty, Desmond is starting a project on the long-term consequences of eviction on poverty by merging public eviction records with tax-return data, with variations in housing laws providing the key instrument. The results will reveal the extent to which deep and extreme poverty can be reduced with a “housing first” policy that ramps up federal housing programs.

Frequent Reporting Project: The CPI is building a suite of “frequent poverty measures” that will make it possible for the country to take current poverty data into account when tailoring its labor market and program policies. As it stands, the Census data used to compute annual poverty rates are released well after the year to which they pertain, thus requiring the U.S. to run its poverty policy largely in the blind. It would be considered deeply problematic, by contrast, to attempt to run the country’s employment policy with labor market data that, instead of being at most two months out of date, were approximately two years old. The Stanford Center has thus developed new frequent poverty measures using the monthly Current Population Survey (CPS). We have carried out nearly three years of research testing various specifications of these frequent measures and examining how they track the traditional poverty measure.

Housing voucher policy: The purpose of this project, which is a collaboration with HUD and local housing authorities (PHAs), is to design housing voucher policies that increase the number of families moving to “high opportunity” neighborhoods in which there are reasonably good chances for upward mobility out of poverty. The U.S. currently spends approximately $20 billion per year on subsidized housing vouchers, but 80 percent of these vouchers are used in moderate- or high-poverty neighborhoods, where opportunities for upward mobility are typically limited. In an experiment undertaken by Raj Chetty, Nathan Hendren, and Lawrence Katz (in collaboration with Peter Bergman, Stefanie DeLuca, Christopher Palmer, and Rebecca Toseland), the goal is to evaluate interventions that entail (a) reducing search frictions by identifying listings in high-opportunity neighborhoods and actively pushing these listings to voucher holders via phone calls and text messages, (b) encouraging landlords to rent units in high-opportunity neighborhoods by providing insurance to protect them from early tenant departures and property damage, (c) providing comprehensive mobility services to tenants (e.g., pre-move counseling, search assistance) that can help them navigate the complexities of moving to and staying in a high-opportunity neighborhood, and (d) offering project-based vouchers in opportunity areas and thereby increasing the supply of Section 8 housing within these areas. By evaluating the effects of these experiments (using data from housing authorities and tax returns), the research team will be able to develop a set of recommendations for reforming the housing voucher program.

Income and the developing brain: The leader of CPI’s life-course research group, Greg Duncan, is planning a new randomized experiment on the cognitive and brain development of low-income children. The prevailing view is of course that poverty is especially likely to shape children’s early development because of the high plasticity and rapid growth of the brain during the first three years of life. There has not, however, been a rigorous study of how income support for families affects the brain function and development of infants and toddlers. In Greg Duncan’s research (in collaboration with Lisa Gennetian, Katherine Magnuson, Kimberly Noble, and Hiro Yoshikawa), approximately one thousand low-income mothers and their newborns from several ethnically and geographically diverse communities will be randomly assigned to either (a) an experimental group that receives $333 in cash payments each month ($4,000 each year) for each of the first 40 months of the children’s lives, or (b) a control group that receives much smaller payments ($20 per month). To understand how poverty-reduction improves brain functioning, the researchers will measure (a) parent stress, family expenditures, time use, parenting practices, and child care arrangements at ages one through three, and (b) children’s cognitive and brain development at age three (with rigorous laboratory measures). The results will provide new evidence on the magnitude and pathways of causal connections between enhanced income and early cognitive and brain development. If the effects are very strong, the case for tax and income-enhancement policies targeted at young children will of course be strengthened.

Income, geography, and life expectancy: The relationship between income and life expectancy can be estimated by combining deidentified tax data and Social Security Administration death records. This project, which was carried out by Raj Chetty and others, revealed that the gap between the richest and poorest 1 percent is as large as 14.6 years. This gap is also increasing: Over the last 13 years, life expectancy remained roughly the same for poor men and women, but increased by 2.3 and 2.9 years respectively for men and women in the top 5% of the income distribution. Although poor people have much shorter lives, the extent of this disadvantage depends dramatically on the place of residence. For example, poor people who reside in affluent areas, like New York, live longer than poor people who reside in less affluent areas, like Las Vegas. This line of research, which is ongoing, provides important baseline evidence on the life-reducing costs of poverty and the opportunities for policy to reduce those costs.

Income supports and deep poverty: The CPI is assisting Y Combinator in examining the effects of unconditional monthly cash transfers on deep poverty and other outcomes. The impetus for this project is the rising concerns, especially prominent in the Silicon Valley, that technology and automation may ultimately lead to substantial further reductions in the prime-age employment rate. The experiment will proceed by drawing an experimental and control group from young adults in deep poverty and regular poverty, delivering unconditional income to the experimental group for five years, and examining cross-group differences in investments in schooling and training, academic achievement, food security, and mental and physical health. These analyses will provide the first U.S. evidence on unconditional income support since the negative income tax experiments of the 1970s.

Infant health and poverty: It is well known that children born into poor neighborhoods are more likely to have low birth weights and other problematic birth outcomes. It is also well known that these disadvantages at the “starting gate” parlay into later developmental disadvantages and increased risks of poverty. We do not know, however, whether the effects of neighborhood conditions on these starting-gate disparities are growing larger. Using the census of U.S. birth records between 1970 and 2014 (with attached geographical information about place of birth at the detailed county level), Florencia Torche will provide the most comprehensive evidence yet on the relationship between local disadvantage and birth outcomes. The results will (a) reveal whether geographical disparities in birth outcomes have changed over time, and (b) identify those neighborhoods that are especially at risk (and hence should be targeted by home visiting and related programs).

Intergenerational elasticities in the U.S.: There remains some debate about the size of intergenerational elasticities (for both earnings and income) in the U.S. In a project led by Pablo Mitnik, David Grusky, Michael Weber, and Victoria Bryant, a 1987 sample of tax filers is exploited to secure new estimates of mobility, with the results underlining the extent to which the family into which one is born affects opportunities to get ahead. 

Long-run effects of SNAP: Because SNAP is very expensive, it is especially important to assess those costs as against its full complement of returns, but we know comparatively little about the long-run effects of early-childhood SNAP exposure on adult economic outcomes. With Martha Bailey, Maya Rossin-Slater, and Reed Walker, Hilary Hoynes is linking adult labor market outcomes to the timing of food stamp introduction in the child’s county of birth. This linked file can then be used to estimate the long-term effects of food stamps on productivity and human capital, economic self-sufficiency, physical ability and health, and neighborhood quality.

Long-run effects of work incentives: It has been difficult to reach any comprehensive conclusions about the payoff to work incentive programs because the effects of such programs are likely to extend to the children of the parents receiving the benefits. The existing assessments of these programs have largely been short term and thus potentially very misleading. In collaboration with Craig Riddell, David Card is undertaking a new long-term assessment of the most famous work incentive program, the Self Sufficiency Project (SSP). This program, administered in Canada in the 1990s, provided generous earnings subsidies to individuals who agreed to forego welfare. In analyses of the SSP to date, the program has been shown to increase participant earnings over the three-year experimental period, but the increases then faded after the subsidy ended. The key question is whether the subsidy had substantial longer-run effects on the children exposed to them. This question can be addressed by linking to individual tax return and unemployment insurance data and then comparing outcomes (e.g., employment, earnings, income, and program participation) across children of the SSP treatment and control group families.

Measuring family complexity in the AOS: Will the American Opportunity Study (AOS) capture the rise of ever more complicated family forms? By linking tax, census, and birth records, the AOS can distinguish biological parenthood (via Social Security filings at the point of birth), social parenthood (via coresidence in census and ACS data), and financial parenthood (via dependent claims in tax records). This in turn makes it possible to analyze both the intragenerational and intergenerational persistence of poverty and program participation.

Minimum wages and poverty: Throughout the west coast, there are a host of minimum wage “experiments” underway, experiments that have the potential to reset the low-wage labor market in quite fundamental ways. With the support of the Laura and John Arnold Foundation and the James T. Irvine Foundation, David Grusky, Bruce Weber, Charles Varner, and Jonathan Fisher are using CLAD (which links census data with California tax returns) to evaluate the effects of these experiments for new labor market entrants. In a related project, Jesse Rothstein aims to go beyond conventional analyses of possible disemployment effects, his key insight being that such effects do not speak decisively to the welfare impact of a minimum wage increase because low-skill workers may well be better off with higher unemployment but better wages (when employed) than with easier access to lower-pay jobs.

The National Poverty Study: The purpose of the National Poverty Study is to develop a protocol for the qualitative measurement of poverty that will allow us to measure trends in the everyday experience of poverty. Although the U.S. is doing an ever better job of counting the poor, it lacks a national infrastructure for examining how the poor live in their natural environment and how they forge their lives in the context of stress, disruption, and deprivation. We have selected 14 sites in the U.S. representing different types of poverty (e.g., deindustrializing poverty, rural poverty, border-community poverty, reservation poverty) and have developed a standardized protocol that allows us to assess cross-site differences and changes in how the poor find jobs, make ends meet, and interact with the safety net. This project is a joint undertaking with the American Institutes for Research (AIR).

A new round of Fragile Families data collection: The Fragile Families Study (FFS), which is designed to enrich our understanding of how experiences in infancy, early childhood, middle childhood, and adolescence are associated with health and socio-economic wellbeing, has long had a special focus on children born to disadvantaged parents (in part because the FFS includes a large sample of poor families). Under the leadership of Sara McLanahan, the coleader of CPI’s Family research group, new FFS data are being collected to better characterize (a) the socioeconomic standing of FFS mothers and children (including obtaining permission to access administrative records), (b) the economic, social, and physical environments of FFS mothers and children, and (c) the health of FFS children (via metabolic and immune markers, genotyping and methylation).

Optimal timing of interventions: It has long been argued that interventions in the earliest years of childhood have larger long-term returns that interventions in later years. The evidence on behalf of this claim is not wholly definitive because it mainly comes from studies based on very small samples (esp. the Perry Preschool Study). Moreover, this research typically compares rates of returns for different interventions at different ages, meaning that one is varying at once both the type of intervention and the age at which it is applied. It would be better to identify the maximal rate of return that one can achieve at each age using the best possible intervention at that age. In a project led by Raj Chetty, Nathan Hendren, and John Friedman, unusually systematic evidence on the optimal timing of interventions is garnered by using administrative data covering the full population from the U.S. The simple insight here is that, if one is willing to assume that parents allocate their income in ways that optimize the outcomes of their children, then the “early-intervention” hypothesis implies that the correlation between parents’ income and children’s earnings should be higher when the children are young. This hypothesis implies, in other words, that children whose parents are relatively rich when they are young but poorer when they are older should fare better than those whose parents became richer when they grew older. In preliminary analyses of the data, the results contradict this prediction: The correlation between children’s outcomes and parent’s income is virtually constant irrespective of the child’s age when the parent’s income is measured. It follows that poverty at younger ages does not appear to be associated with significantly worse outcomes than poverty at older ages. The latter approach is limited, however, because it is correlational and because it relies on the assumption that parents use incremental resources to optimize on behalf of their children. The second approach that will be taken relaxes these assumptions by studying the long-term impacts of a set of interventions at different ages (using earnings data drawn from tax records between 2010-2014). The researchers will consider (a) the effects of moving to a better neighborhood at different ages to estimate rates of return at each age, (b) the effects of having better teachers at different ages, and (c) the long-term impact of the Nurse-Family Partnership (NFP). The latter analysis, which will be carried out in collaboration with David Olds, will be the first long-term assessment of the NFP program. By comparing the effects of the NFP intervention with the effects of the other interventions described above, it becomes possible to assess rates of return for several interventions at different ages, all evaluated in the same era (2010-2014) using the same data (administrative tax records) with minimal attrition and sample selection. These results, when taken together with the correlational evidence from the first line of analysis, will provide high-quality evidence on the best age to intervene to lift children out of poverty. This will of course be a foundational result for future policy decisions.

Poverty among refugees: The U.S. refugee population faces very high rates of poverty, major health problems, and much discrimination. The country’s resettlement program, which is the largest in the world, has faced increasing scrutiny with the global refugee crisis and recent terrorist attacks. We know very little, however, about the best ways to assist with resettlement and to increase self sufficiency. This is partly because, despite the difficulty and high cost of relying on surveys, our data collection system for refugees is still largely survey-based (e.g., the Annual Survey of Refugees). In partnership with the Bureau of Population, Refugees, and Migration (at the State Dept.) and the Office of Refugee Resettlement (at the Dept. of Health and Human Services), David Laitin and Jeremy Weinstein are developing an administrative-data system to monitor the state of the refugee population and to assess the effect of refugee policies and programs.

Poverty and schooling on reservations: The test scores and educational outcomes on Native reservations are, like those in high-poverty black neighborhoods, very low relative to the national average. For the Native population, a special difficulty arises in reconciling traditional and formal education, indeed these are typically viewed as competitive rather than complementary. The purpose of this project, which is led by the noted ethnographer Martin Sánchez-Jankowski, is to investigate how traditional and formal education are viewed and the ways in which they might be better integrated. The sites for the study will be two Native American reservations, San Carlos Apache in Arizona and Rosebud Sioux in South Dakota, both of which are characterized by high rates of dropout, poverty, and suicide.

Poverty and the decline in prison population: There is much evidence that unemployment rates are lowered by incarcerating many of the potentially unemployed. It is likely that other poverty-relevant outcomes, such as homelessness or the use of mental health services, might also be affected by incarceration rates. This possibility takes on immediate importance because of ongoing declines in some prison populations. In October 2011, California passed legislation that led to an 18 percent decline in its prison population, a decline that was partly achieved by discontinuing the practice of returning individuals to custody because of technical parole violations. This decline in the state’s prison population may bring about an increase in homelessness, mental health service use, and other poverty-relevant outcomes. The co-leader of the CPI’s incarceration program, Steven Raphael, is investigating this possibility by seeking permission to link data from California’s Automated Criminal History System (ACHS) with data from the California Department of Healthcare Services. The results from this project will reveal whether ongoing declines in incarceration should be coordinated with increased funding for programs that may substitute for incarceration.

Poverty Relief Project: The Poverty Relief Project, led by Stanford political science professor Karen Jusko, develops a new measure of the effectiveness of cash-based poverty relief programs. The "poverty relief ratio" provides an estimate of the extent of redistribution relative to what would be required to bring all low-income households to a well-defined poverty line. With Kate Weisshaar, Jusko uses the poverty relief ratio to evaluate the effectiveness of anti-poverty programs over time, across states, and across countries. The first set of national, state-level, and cross-national results were released in our 2014, 2015, and 2016 State of the Union reports. This new measure will continue to be applied in further research projects.

Reducing the race gap in test scores: Although the gap between black and white achievement test scores has been narrowing in the last 20 years, it remains very large and is an important source of the persistently high poverty rates for blacks. The key question here: How can the gap be reduced? In a project using the new Stanford Education Data Archive (SEDA), Sean Reardon is examining the substantial variability across school districts in the size of this gap, a variability that opens up the opportunity of understanding the conditions under which the gap can be reduced or even eliminated. The SEDA, which includes harmonized data from 200 million test scores for public school students, allows Reardon and other researchers to examine the effects of district policies, finances, and socioeconomic conditions on the size of the gap. Are there any districts in which the gap has been eliminated? Are gaps typically larger in segregated districts? Are gaps smaller in the well-off schools? The answers to these questions will provide the most systematic evidence to date on the capacity of school-district policies to reduce the racial gap.

Rent and inequality: It is increasingly fashionable to argue that “rent” accounts for much of the takeoff in income inequality. But there haven’t been any efforts to build a comprehensive model of income inequality that accounts for the many forms of rent and assesses their full and complete contribution to the takeoff. In a project led by Kim Weeden, David Grusky, and Beth Red Bird, the Current Population Survey is used to build just such a model covering the 1970s to the present day.

The rise of between-firm inequality: In collaboration with Jae Song, David J. Price, Fatih Guvenen, and Till von Wachter, Nicholas Bloom is examining the role of firms in the increase in U.S. earnings inequality. The key questions: How much of the rise in earnings inequality can be attributed to increasing between-firm dispersion in the average wages they pay? How much is instead due to rising wage dispersion among workers within firms? These questions are addressed by constructing a matched employer-employee data set for the United States using administrative records. The initial results show that, between 1978 to 2012, virtually all of the rise in earnings dispersion between workers is due to increasing dispersion in average wages paid by employers. The amount of inequality within employers has, by contrast, remained virtually unchanged, a finding that is robust across industries, geographical regions, and firm size groups. The wage gap between the most highly paid employees within these firms (CEOs and high level executives) and the average employee has increased only by a small amount.

Small place estimates: The Equal Opportunity Project, led by Raj Chetty, uses tax return data to monitor opportunities for mobility out of poverty. By examining families who move across areas, it is possible to identify the causal effects of neighborhoods on opportunities, a line of analysis that has demonstrated that a child’s chances of success are strongly influenced by where that child grows up. To date, this line of research has relied on relatively large aggregates (i.e., “commuting zones”) to examine the effects of place, aggregates that are arguably too large for the purpose of establishing how place-specific policies are affecting the likelihood of mobility out of poverty. The next step, therefore, is to characterize such “place effects” at a more local level. By carrying out analogous analyses at the census block and tract level, Raj Chetty and Nathan Hendren will be able to evaluate policies implemented at very detailed levels, such as a change in a local school policy, a change in local policing tactics, or a change in local homeless policy. In collaboration with Matthew Desmond, Chetty and Hendren will identify areas that are relatively affordable for low-income families, yet nonetheless have proven to move children out of poverty at very high rates. The researchers will identify 5-10 such “opportunity bargains” and then complete intensive ethnographic research to find out what makes these communities so successful.

Trends in income and wealth inequality: The country’s leading analyses of trends in income and wealth inequality are carried out by Emmanuel Saez and his research team. This line of research, based on annual updates to tax data, is released annually in the CPI’s State of the Union publication and in many other leading outlets.