California Policy Lab

Abstract: As US college costs continue to rise, governments and institutions have quadrupled financial aid. Yet, the administrative process of receiving financial aid remains complex, raising costs for families and deterring students from enrolling. In two large-scale field experiments (N= 265,570), we test the impact of nudging high-school seniors in California to register for State financial aid. We find that simplifying communication and affirming belonging each significantly increase registrations, by 9% and 11% respectively. Yet, these nudges do not impact the final step of the financial aid process -- college enrollment. In contrast, a simplified letter that affirms belonging while also making comparable cost calculations more salient significantly impacts college choice, increasing enrollment in the lowest-cost option by 10.4%. Our findings suggest that different nudges are likely to address different types of administrative burdens, and their combination may be the most effective way to shift educational outcomes.

This work has been supported, in part, by the University of California Multicampus Research Programs and Initiatives grants MRP-19-600774 and M21PR3278.

We use traditional and non-traditional data to measure the collapse and partial recovery of the U.S. labor market from March to early July, contrast this downturn to previous recessions, and provide preliminary evidence on the effects of the policy response. For hourly workers at both small and large businesses, nearly all of the decline in employment occurred between March 14 and 28. It was driven by low-wage services, particularly the retail and leisure and hospitality sectors. A large share of the job losses in small businesses reflected firms that closed entirely, though many subsequently reopened. Firms that were already unhealthy were more likely to close and less likely to reopen, and disadvantaged workers were more likely to be laid off and less likely to return. Most laid off workers expected to be recalled, and this was predictive of rehiring. Shelter-in-place orders drove only a small share of job losses. Last, states that received more small business loans from the Paycheck Protection Program and states with more generous unemployment insurance benefits had milder declines and faster recoveries. We find no evidence that high UI replacement rates drove job losses or slowed rehiring.

This work has been supported, in part, by the University of California Multicampus Research Programs and Initiatives grant MRP-19-600774.

Most labor market policy in the United States is designed for long-term employment relationships. Self-employed workers, including independent contractors and on-demand platform (“gig”) workers, are excluded from labor market protections such as wage and hour laws, occupational safety and health regulations, unemployment insurance, and employer-provided health insurance and retirement programs. They are also poorly covered by our tax collection system, which relies heavily on employer reporting of worker earnings for enforcement. Growth in independent contracting could undermine labor market arrangements, with implications for regulation, tax collection, and worker wellbeing.

This paper uses California tax data to provide an alternative lens on many of the outstanding empirical questions about independent contracting. We use de-identified, individual-level data from California personal income tax returns for tax years 2012 through 2017 to measure the prevalence and nature of self-employment and independent contracting. We estimate that 14.4% of California workers aged 18-64 in tax year 2016 had some independent contracting income. Over half of independent contractors also had traditional jobs generating W-2s, and most of these received the bulk of their earnings from their traditional jobs. Workers with low earnings are significantly more likely to earn independent contracting income and to rely primarily or exclusively on that income. We explore the characteristics of independent contractors and their distribution across family type, geography, and industry.

This work has been supported, in part, by the University of California Multicampus Research Programs and Initiatives grants MRP-19-600774 and M21PR3278.

In this study, we assessed equity in criminal record relief eligibility in California, one of the first states to pass automatic record relief legislation. Our analysis included three components. First, using criminal history data from the California Department of Justice (CA DOJ), we assessed the share of people with criminal records who are eligible for automatic relief under current laws, and how this eligibility varied across racial and ethnic groups. Second, we evaluated two hypothetical reforms in how eligibility is determined that might alter equity across racial and ethnic groups: (a) relief for discretionary cases, and (b) a sunset rule that would automatically grant relief for convictions more than 7 years old. Finally, we estimated how each of these hypothetical reforms would alter population-level disparities in conviction records statewide.

This work has been supported, in part, by the University of California Multicampus Research Programs and Initiatives grants MRP-19-600774 and M21PR3278

As evidenced by both the Pandemic Recession and Great Recession, labor markets can change rapidly. More granular data on the time-path of these changes, and the role played by firm closures, layoffs, hours changes, and worker turnover can help us better understand how the labor market is evolving. On this site, we will post weekly updates of labor market information from Homebase’s timecard data to shed light on the details of a rapid evolving labor market. We aim to measure the short- and medium-term evolution of the size of the small business sector and of the health of employers in this sector, by tracking whether firms in Homebase’s userbase are expanding or contracting the number of hours that they use each week and the rate of turnover among their workers.

This work has been supported, in part, by the University of California Multicampus Research Programs and Initiatives grants MRP-19-600774 and M21PR3278.

In 2014, the California Department of Corrections and Rehabilitation began a demonstration of theEnhanced Drug and Contraband Interdiction Program at 11 prisons in California. Using data provided bythe Department, this study finds that the intensive version of the program yielded a 23% decline in failurerates of random drug tests over the period studied, and a reduction in the number of cellphone violations,but that these same institutions experienced increased levels of drug-related rules violations. Themoderate program had no discernable impact on drug abuse in the prisons in which it was tested.

As described in A Rising Tide, student participation in dual enrollment has been growing steadily over the last four years. Yet, participation varies across racial/ethnic subgroups and special populations of students. This online appendix reports the variation that also exists across California’s 58 counties. For each county, we provide the rate of dual enrollment participation overall in the last four years and depict the differences in participation rates for the four largest racial/ethnic subgroups (Asian, Black, Latinx and White) and for subgroups of socioeconomically disadvantaged students, English learners, students with disabilities, foster youth, and homeless students. In some county-level graphs, student subgroups are omitted due to cell size restrictions for reporting on subgroups with few students.

This work has been supported, in part, by the University of California Multicampus Research Programs and Initiatives grants MRP-19-600774.

Over 150,000 low- and moderate-income California high school graduates each year are eligible for CalGrant entitlement awards, which can cover full tuition and most fees at any of the three public higher education segments in the state, or can make substantial contributions toward tuition at private colleges. Unfortunately, many eligible students don’t take up the awards. Many may not be aware of their eligibility, know how to navigate the system, or feel like these funds are truly meant for them. In 2017-8, the California Policy Lab worked with the California Student Aid Commission to design and test more effective notifications to eligible high school seniors. The redesigned letters were clearer, shorter, and encouraged students to think of themselves as college-bound. The results were promising. Students who received the redesigned letters were much more likely to take the first step toward claiming the award than a randomly selected comparison group. Future analyses will measure impacts on college enrollment, CalGrant payouts, and eventual college completion.

In the admissions cycle that began in November 2016, UC Berkeley carried out the second year of a pilot experiment with letters of recommendation. Unlike other highly selective universities, Berkeley has never previously asked applicants to submit letters from teachers and guidance counselors. This may limit the information available for use in holistic review, and some at Berkeley think that as the university gets more selective it is getting harder to make informed decisions with the evidence available. Others, however, are concerned that students from disadvantaged backgrounds may not have access to adults who can write strong letters, and that the use of letters will further disadvantage these students.

In the pilot experiment, a subset of applicants was invited to submit letters of recommendation if they wished. Any submitted letters were incorporated into the “second read” evaluations of their applications. I evaluate the impact of this on the outcomes of applicants from four groups underrepresented among successful applicants to Berkeley: students from families with low incomes, students whose parents did not attend college, students from low-scoring high schools, and students from underrepresented racial and ethnic groups. I use a variety of methods, including a within-subject design that compares application scores when readers had access to letters with scores from a parallel process that suppressed the letters and a regression discontinuity design that exploits sharp distinctions between otherwise similar students in the selection of students to be invited to submit letters.

In 2014, the California Department of Corrections and Rehabilitation began a demonstration of theEnhanced Drug and Contraband Interdiction Program at 11 prisons in California. Using data provided bythe Department, this study finds that the intensive version of the program yielded a 23% decline in failurerates of random drug tests over the period studied, and a reduction in the number of cellphone violations,but that these same institutions experienced increased levels of drug-related rules violations. Themoderate program had no discernable impact on drug abuse in the prisons in which it was tested.

On any given night, nearly 60,000 people experience homelessness in Los Angeles County, and an estimated 141,000 are homeless in any given year. In response to this growing crisis, voters in Los Angeles County passed Measure H, agreeing to increase their taxes to add an estimated $355 million in homeless services each year. As reported in the 2018–19 Measure H 15-Month Report Card, tens of thousands of people were housed and/or linked to intensive services as a result. Yet, the homeless population continues to grow as inflow outpaces exits to permanent housing. In 2019, despite the fact that thousands of people were served by Measure H services, the homeless population in Los Angeles County (as measured by the Greater Los Angeles Homeless Count) grew by 12%. To help reduce inflows and to reach people before they become homeless, the Board of Supervisors approved Measure H spending plans for Fiscal Years 2017–18 and 2018–19 that included $5.5 million and $17 million, respectively, for prevention strategies. These strategies included short-term financial assistance, case management, and legal services.

This work has been supported, in part, by the University of California Multicampus Research Programs and Initiatives grant MRP-19-600774.

The top one percent of arrestees in San Francisco (“high users”) account for approximately seven percent of all arrests. Property crimes, both felony and misdemeanor, are the most frequent charge in both high user arrests and cases filed by the District Attorney. High users are predominantly male and fall between 30 and 50 years old. African Americans, though 6% of San Francisco’s population, constitute almost 50% of the high user cohort. San Francisco’s high user cohort also faces significant economic insecurity: more than half accessed safety-net benefits from the Human Services Agency during the study period.

California needs a centralized authority for linking the state’s administrative data. Legislators are focusing on new datasets and data systems, which is a step in the right direction. But what the state truly needs is a new office with a clear mandate to link the state’s core data assets, a clear set of tools for doing so, and governance that ensures data are used to inform program improvement. Think of it as the state’s Census Bureau – or “Statistics  California.”

We propose here a roadmap toward that goal: (1) create a new, independent office with the mandate and expertise to link data across siloes, (2) sequence the linkage process by starting with education and expanding outward, and (3) establish streamlined governance that makes data available to improve state policies and programs.

This work has been supported, in part, by the University of California Multicampus Research Programs and Initiatives grant MRP-19-600774.

We review the linking of datasets that contain identifying information (e.g., names, birthdates) but not unique common identifiers for each individual. We discuss strategies for identifying matches in three families: rules-based matching, supervised machine learning, and unsupervised machine learning. These vary in the ways that they combine human knowledge with computing power. We define different measures of accuracy and explore the performance of common algorithms in test data.

Our goal is to de-mystify data linking for non-technical readers. We attempt to explain the criteria that should inform the choice of linking methods, and the decisions that need to be made to implement them.

Additional resources, including code and public data referenced on pp. 26-34 is available at: https://github.com/californiapolicylab/data-linking.

Policymakers rely heavily on the tax system to distribute direct payments to lowincome families. Anti-poverty tax credits such as the Earned Income Tax Credit, the advanced Child Tax Credits, and the federal stimulus payments combined to keep millions of Americans out of poverty during the pandemic. Such credits have strong potential to continue to reduce poverty.

These credits only work, of course, if eligible families receive them. To do so, they must file a tax return. But many low-income families who are at or below the federal poverty level are not legally required to file taxes. Policymakers need a better understanding of how many low-income families don’t file taxes (and therefore miss out on these valuable credits) in order to address this problem.

While state and federal tax agencies know who files taxes, they have very little information on the families who do not file, especially those below the poverty level with little or no earnings. State and local human-service agencies, however, serve many families below the poverty level, placing them in a unique position to assist eligible families to receive these credits.

To help the State of California understand who may be at risk of not receiving anti-poverty tax credits, the California Policy Lab (CPL) facilitated a linkage of two individual-level datasets held by state agencies: one with safety-net enrollment data and one with state tax filing data. CPL served as a trusted third party by implementing a “hashed linkage” — linking data that was de-identified through “hashing” (a one-way encryption process) by each agency. 

By linking this data, we were able to help California measure how many Californians receiving safety-net benefits were at risk of not receiving federal stimulus payments, the state Earned Income Tax Credit, and the advanced Child Tax Credit. We also helped the state learn that the majority of its safety-net beneficiaries were already receiving benefits through tax filing — allowing the state to focus limited resources on those who were not receiving these benefits. This linked data also equipped the California Department of Social Services (CDSS) to conduct targeted outreach to Californians who had not filed state returns (and therefore were missing out on thousands of dollars in credits) in recent years and to direct them towards intensive resources that could help them file a return and claim these credits. This linkage led to millions of dollars in tax credits delivered to Californians who otherwise may not have received them.

The benefits of linking administrative data go beyond the take-up of anti-poverty tax credits. Administrative data can help answer many vexing policy questions faced by policymakers. However, much of the value in administrative data can be obtained only when data can be linked across multiple sources. By linking across systems at the individual level, administrative data, which is often topically narrow, can replicate the cross-domain scope of survey data. 

This toolkit is intended to help staff in state governments outside of California who are interested in using administrative data and linking it across agencies to measure the take-up of safety-net benefits. We are also releasing a technical how-to toolkit for those interested in operationalizing a hashed linkage.

This work has been supported, in part, by the University of California Multicampus Research Programs and Initiatives grants MRP-19-600774 and M21PR3278