Working Papers

The Icing on the Cake: The Effects of Monetary Incentives on Income Data Quality in the SIPP (with Jonathan S. Eggleston)

Accurate measurement of key income variables plays a crucial role in economic research and policy decision-making. However, the presence of item nonresponse and measurement error in survey data can cause biased estimates. These biases can subsequently lead to sub-optimal policy decisions and inefficient allocation of resources. While there have been various studies documenting item nonresponse and measurement error in economic data, there have not been many studies investigating interventions that could reduce item nonresponse and measurement error. In our research, we investigate the impact of monetary incentives on reducing item nonresponse and measurement error for labor and investment income in the Survey of Income and Program Participation (SIPP). Our study utilizes a randomized incentive experiment in Waves 1 and 2 of the 2014 SIPP, which allows us to assess the effectiveness of incentives in reducing item nonresponse and measurement error. We find that households receiving incentives had item nonresponse rates that are 1.3 percentage points lower for earnings and 1.5 percentage points lower for Social Security income. Measurement error was 6.31 percentage points lower at the intensive margin for interest income, and 16.48 percentage points lower for dividend income compared to non-incentive recipient households. These findings provide valuable insights for data producers and users and highlight the importance of implementing strategies to improve data quality in economic research.

Money Talks: The Effects of Monetary Incentives on Earnings Non-Response in SIPP (with Kelly R. Wilkin)

The Survey of Income and Program Participation (SIPP) has a history of using conditional and discretionary monetary incentives to induce survey responses. While incentives have been effective in increasing unit response, little is known about their effect on item response. This paper exploits a multi-wave random monetary incentive experiment for the SIPP 2014 panel to examine the effect of incentives on earnings non-response. We show that individuals in incentive-receiving households have a 1.3-percentage-point lower earnings non-response rate than those in non-incentive households. This effect is robust to controls for observed and unobserved individual heterogeneity and non-random panel attrition in a correlated random effects specification. Further, we find the effect is driven by a $40 incentive assignment and not the $20 incentive. Consistent with theories linking unit and item non-response, we find that contemporaneous earnings non-response is associated with a higher probability of attrition in the following wave, but the $40 incentive mitigates this relationship.

Frailty, Thy Name Is Still Woman? The Impact of Local Labor Demand Shocks On The Prevalence of Traditional Attitudes (Job Market Paper)

This paper examines the effects of U.S. state-level labor demand changes on the prevalence of traditional attitudes toward women working outside of the home, and a woman’s emotional suitability for politics. Traditional gender-role attitudes, where women are viewed as homemakers and men as breadwinners, have declined substantially over time. Although, many contributing factors have been previously studied, I examine the contribution of labor demand shifts to these attitude changes. I document that positive labor demand shocks, measured as Bartik shocks, lower the prevalence of traditional attitudes toward women working but find no statistically significant effect on traditional attitudes toward women's emotional suitability for politics. Also, despite finding no evidence of heterogeneous effects of Bartik shocks, I find suggestive evidence that own-group Bartik shocks, defined along gender and education dimensions, are possibly more relevant measures than the overall labor demand shocks, especially among men with less than a high school diploma and men with at least a baccalaureate.

States Taking the Reins? Employment Verification Requirements And Local Labor Market Outcomes (with Benjamin Feigenberg and Darren Lubotsky)

We estimate the impact of state-level “E-Verify” legislation that mandates employment eligibility verification for private-sector workers. We document declines in formal sector employment and employment turnover after mandate passage, with effects concentrated among those likeliest to be work-ineligible. Using newly available data, we show that larger firms are far more likely to comply with mandates. Heterogeneity in adherence leads to substantial within-state employment spillovers from larger to smaller firms, as well as a reduction in the number of large firms. We find no evidence that work-ineligible populations relocate or that native-born workers’ labor market outcomes improve in response to mandates.

Selected Works in Progress

Hot Decks and Cold Values: A Solution To The Missing Data Problem? (with Kelly R. Wilkin)

Hot-deck imputations are commonly used for replacing missing data at U.S. Census Bureau. However, their performance is inherently limited by a trade-off: the higher the number of selected predictive features, the better matched are donors to recipients but the lower is the probability of finding a donor. We propose a strategy to minimize the said trade-off. We use recursive feature elimination to select the most predictive features for employment status by looping over multiple estimation models such as random forests, logistic regressions, and Bernoulli naïve classifiers, and using a grid search over different number of features to select. We choose a model and number of features that produce the highest precision for setting up our hot-deck matrix. If a cell has no donors, we populate cells with “mode of modes” values from sequentially dropping a covariate until a mode is obtained, taking the mode of the different combination of the remaining covariate values, and taking the mode of those modes. We show how to automate this process using metadata, Python, and SAS. This systematic approach ensures higher data quality by removing ad hoc human selection of cold-deck values from the data allocation process.

Checked Out:  Unbanked Households’ Engagement with the Tax System and the Social Safety Net (with Mark A. Klee)

The quickest way to have received the Economic Impact Payments during the Coronavirus pandemic has been through direct deposit payments. Unbanked households, because of their disengagement from the banking system, experience longer waits. The problem is twofold: unbanked households are hard to reach and hard to identify. These difficulties are not specific to the pandemic crisis and have meant that the unbanked households may miss out on receiving government benefits. In fact, the US Treasury and several well-known transfer programs including Social Security and the Supplemental Nutrition Assistance Program, formerly known as food stamps, offer benefits through prepaid debit cards or electronic benefits transfer cards to include unbanked recipients. However, if unbanked households do not earn enough income to meet the tax return filing requirement or if they are more likely to work in the informal sector, it is hard for authorities to identify these households or verify their eligibility. This study sheds light on how unbanked households’ lack of engagement extends to the tax system and the social safety net by linking data from the Survey of Income and Program Participation to administrative tax and program data.

Are Tariffs Biased? The Effects of the 2018 U.S. Tariffs on the Gender Wage Gap (with Neil Bennett)

In 2018, the United States engaged in protectionist policies by imposing tariffs on all imported steel and aluminum. Although the US has a long history of using unilateral tariffs, the breadth and the justification for these tariffs renewed global attention to the impacts of trade policies. Given that these rising tariffs were also a precursor to a global pandemic that disproportionally affected the labor force participation of women, it is becoming increasingly important to understand how women were faring in the labor market prior to the COVID-19 pandemic. Therefore, we examine how the 2018 U.S. tariffs on steel and aluminum changed the gender wage gap in the years leading up to the pandemic. Specifically, pooling cross-sectional waves 1 of the 2018 and 2019 panels of the Survey of Income and Program Participation (SIPP) and using a fixed effects model, we explore the effects of the 2018 U.S. tariffs on steel and aluminum on industry-specific gender wage gaps. Ex ante, the effects of tariffs on the gender wage gap is ambiguous. Tariffs on steel and aluminum as final goods may lower competition by increasing the prices of foreign goods relative to domestic goods, increase demand for domestic products, and therefore increase employment. Given the higher concentration of male to female workers at steel and aluminum industries, the increase in employment is likely concentrated among men, leading to an increase in the male to female wage ratio. Alternatively, tariffs on steel and aluminum as intermediate goods may increase the cost of production at downstream domestic industries, and lead to a price hike. This price hike would lower demand and increase employment layoffs in industries producing final goods. Depending on the composition of these downstream industries, more men may be laid off, relative to women, which would lower the gender wage gap.

Immigration Attitudes and Labor Market Conditions in the United States (with Oleg Firsin)

We examine immigration attitudes in the United States expressed in tweets, newspaper articles, google trends, and surveys of political opinions (e.g., American National Election Studies and National Annenberg Election Survey). After applying machine learning techniques to classify millions of tweets and parse layouts of newspaper articles, we map implicit attitudes to sentiments on a negative-to-positive scale, and compare results across all the alternative measures used. We explore geographical and temporal variation in the imputed sentiments and test the sensitivity of our findings to methodological choices. The results indicate a generally positive association between the measures of immigration sentiment expressed in different sources. At the same time, each measure adds a different dimension to the expression of attitudes. In particular, tweets are often anonymous, numerous and free-form, whereas surveys have more uniform and well-defined questions, newspaper articles are long-form, often cover local events, and go back in time more than other measures, and google trends are high-frequency data that reveal people’s attitudes through anonymous search queries. Hence, a more complete understanding of immigration attitudes emerges through a comprehensive analysis of multiple measures.

Notes on the Evolution of Gender-Specific Language and Job Opportunities for Women

This ongoing research expands my job market paper by documenting changes in evolution of gender-specific language and job opportunities for women by applying computational tools of natural language processing to unstructured newspaper data.