En république tchèque, les politiques de conciliation travail/famille ont été profondément remodelées à l’occasion de la transition systémique vers l’économie de marché : l’objectif de cette thèse est de décrire les récentes évolutions de politique familiale et d’estimer leurs effets sur l’emploi des mères. Malgré l’accession du pays à l’UE en 2004 et une disponibilité croissante de données, la littérature économique sur le régime d’Etat social, les politiques sociales et familiales et leur effet sur le marché du travail reste très rare. Je montre que l’orientation des politiques familiales après 1989 a induit un fort recul de garde publique d’enfants et creusé un très fort écart d’emploi entre les femmes avec et sans enfants en âge préscolaire. J’analyse les effets de deux réformes de congé parental : la réforme de 1955 qui a prolongé le paiement de l’allocation parental à 4 ans par enfant sans prolonger la durée de la protection d’emploi (3 ans), puis la réforme de 2008 qui a au contraire encouragé un retour en emploi plus rapide qu’auparavant. J’utilise l’Enquête Emploi et j’applique la méthode des Différences-de-différences pour estimer l’impact sur l’emploi des mères à court et moyen terme. Enfin, j’examine les déterminants culturels de long terme des préférences des ménages tchèques en termes de conciliation travail/famille, et je mets en évidence une évolution des valeurs de genre vers un modèle conservateur de la division des tâches. Cette évolution, qui court sur les années 2000, s’oppose à la tendance Européenne générale et est susceptible d’influencer l’orientation des politiques familiales ainsi que leurs effets sur les ménages. / Czech work-life conciliation policies and practices have gone through dramatic changes since the 1989 transition from centrally planned to market economy. The objective of this thesis is to describe the recent evolutions of family policies, and to assess their effects on maternal employment. Surprisingly, despite the country’s EU accession in 2004 and an increasing data availability, the economic literature on the Czech welfare state regime, its social and family policy and its effects on labour market outcomes is extremely scarce. I show that post-transitional policies differed from the former interventionist and paternalist orientation, and resulted in a sharp decrease in public childcare supply and the widest parenthood-related employment gaps among OECD countries (41 pp in 2011). I focus on two reforms of the parental leave system: the 1995 Parental Benefit reform which extended the payment of universal parental benefit to 4 years instead of 3 without an equivalent extension of the job protected parental leave; then the 2008 Multi-Speed Parental Benefit reform, which encouraged yet again a faster return to employment. I use the Labour Force Survey and rely on a difference-in-differences strategy to assess the net effect of these reforms on mother’s labour market participation, in both short and medium run. Last but not least, I investigate long-run cultural determinants of the observed work-life conciliation preferences and show that a significant evolution towards conservative gender roles has been taking place in the post-transitional decades. This opposes the general European trend, and is likely to influence family policy orientations as well as the reforms’ outcomes.

The aim of this thesis is to analyse empirically, proposing new methods to tackle disputed questions in the literature of political and labour economics, the Italian institutional setting both in a political competition context and in the occupational structure. The first paper explores the relationship between transfers from central state to political aligned municipalities and the effect of these transfers on local electoral consensus. This study contributes to the empirical literature of the political determinants of spikes in central transfers in pre-electoral periods and of the electoral benefits of pork barrel measures for incumbent politicians. Despite several findings of strong evidence that intergovernmental fiscal transfers rise during election years, in the Italian case researchers investigated little the political incentives that lay behind these increases or the success of these transfers in attracting votes. We focus on the so called swing municipalities, defined as those in which the probability of winning is close to one-half, analysing data of Italian comuni with more than 15 000 inhabitants, in the period 2007-2014. From an empirical perspective, every attempt to estimate the causal impact of political alignment on the amount of federal transfers is clearly complicated by endogeneity issues. Without a credible source of exogenous variation in political alignment, the empirical correlation between alignment and transfers (if any) can be completely driven by socio-economic factors influencing both dimensions. We propose a new model specification to account for the endogeneity issue arising when estimating the causal impact of political alignment on transfers: the unpredicted change in the government occurred in 2011 after the resignation of Silvio Berlusconi and the following appointment of Mario Monti as prime minister. We perform our empirical estimation in two steps: first, we apply the close-race RDD setup (Lee 2008) to assess the impact of political alignment on transfers. Results from the close-race RDD show that aligned municipalities receive more grants, with this effect being stronger before elections. At a second empirical stage, we perform a local linear regression of the re-election probability of the local incumbent on transfers, including the first stage error term to have our coefficient of interest measuring only the effect of politically-driven transfers on electoral outcomes, and we conclude that this probability increases as grants increase. The second paper stems from the observation of the most recent phenomena in the domestic and foreign labour market: technological progress has been associated to a crowding-out of cognitive-skill intensive jobs in favour of jobs requiring soft skills, such as social intelligence, flexibility and creativity. Soft skills can be defined as interpersonal, human, people or behavioural skills necessary for applying technical skills and knowledge in the workplace. The nature of the soft skills make them hardly replaceable by machine work, and Among soft skills, creativity is one of the hardest to define and to codify, therefore, creativity-intensive occupations have been shielded from automation. In our work, we focus on creativity, starting from its definition in order to get significant insights on which occupational profiles in Italy can be considered creative and to explore their dynamics in the labour market. A possible analytical definition of creativity comes from the seminal work of Edward De Bono. According to his pioneering research in the field, lateral thinking is strictly related to creativity and it can be described along four dimensions: 1) fluidity, as the ability of a subject to give the highest possible number of answers to a certain question; 2) flexibility, as the number of categories to which we can bring back these questions; 3) originality: ability of expressing new and innovative ideas; 4) processing: ability of realizing concretely one’s ideas. We apply this definition to a uniquely detailed occupational dataset on tasks, skills, work attitudes, and working conditions regarding all Italian occupations: the Inapp-Istat Survey on Occupations (Indagine Campionaria sulle Professioni, ICP hereafter), an O*NET-type dataset developed by the Italian National Institute for Public Policy Analysis. The Survey on Occupations, in fact, presents a list of skills and competences and workers are asked to identify those they make use of in performing their job. Inside this list, we identify 25 skills associated to creativity and we formulate a Matrix Completion (MC) optimization problem, as discussed theoretically in Mazumder (2010). Matrix Completion is the task of filling in the missing entries of a partially observed matrix, which we generate by obscuring randomly 10%, 25% and 50% of the entries in the columns associated with the creative skills, given a fixed row (occupation). In our analysis, we use a formulation of the problem known as Nuclear Norm Minimization and we solve it with the Soft Impute Algorithm. We conclude our analysis on social skills in our third paper where we analyse the effects of Covid-19 pandemic on soft skills in the context of Italian occupations, operating in about 100 economic sectors. We leverage detailed information from ICP, the Italian O*Net, and we simulate the impact of Covid-19 on those workplace characteristics and working style that were more seriously hit by the lockdown measures and the new sanitary dispositions (physical proximity, face-to-face discussions, working remotely, ecc.). We simulate three possible scenarios based on the intensity of the effects of COVID-19 on some working conditions, such as working from home, keeping physical distance and so on. We then apply matrix completion, a machine learning technique used in recommendation systems, in order to predict the levels of soft skills required for each occupation when working conditions change, as these changes might be persistent in the near future. Professions showing a lower intensity in the use of soft skills, with respect to the predicted one, are exposed to a deficit in their soft-skill endowment, which might ultimately lead to lower productivity or higher unemployment, thus enhancing the negative effects of the pandemic.

<div>State and local adoption/repeal of highly polarized policies causes migration responses both out of and into the affected region. Interpreting the responses as revealed policy pref?erences leads to the conclusion that marijuana legalization and abortion waiting periods had been favored nationally, while gay marriage had been opposed. Policy preferences are geographically heterogeneous, which leads to different responses across counties. From 1992- 2017, these policy changes reduced domestic migration by two percent, which is approxi?mately 20% of the total migration decline. The migration changes, via partisan sorting, accounted for a significant share of the increased political polarization from 2012-2016 in western, urban, and swing counties. <br></div><div><br></div><div>In cases where unmarried parents have joint physical custody of their child(ren), there is a wide range of default relocation restrictions that depend on their state of origin. Using IRS county-to-county migration data, demographic data from the ACS, and state relocation restrictions gathered from divorce law websites, I study the impact of these default reloca?tion restrictions on domestic US migration. Results from both regression discontinuity and selection on observables designs, find about 10% - 30% less migration to counties that are outside the allowed relocation range. This migration friction is shown to strengthen from 1992 - 2012, as both joint physical custody and unmarried parents became more common, thereby contributing to the decline in domestic US migration. <br></div><div><br></div><div>In the United States, between 2004 and 2008, 28 states increased their minimum wage; the national minimum wage was increased in 2007. The average migration response to these increases was a 3% change in migration away from a one dollar increase. These effects are not distributed evenly across the population. People from more impacted demographic groups are more likely to move away from minimum wage increases.</div>

<p>This dissertation explores the issue of under-representation of women in STEM fields in high school and the early years of college. One of the major contributors to the persisting gender earnings gap is male-domination in the STEM workforce. Women are under-represented in STEM occupations since they are less likely than men to take advanced STEM courses in high school and to choose STEM majors in college. While the gender STEM gap does not exist at early ages according to most studies, it has been shown that girls start to lag behind boys in Math tests after middle school.</p> <p><br></p> <p>In Chapter 1, I investigate the STEM gender gap in the context of teacher-student gender matching. Using a fixed-effects regression model, and Chilean administrative education data on SIMCE and PSU exams and college application, I explore whether high school girls perform better in Language and Math when they have female teachers, and whether a female Math teacher impacts girls’ preference towards STEM programs when enrolling in college. I find that female teachers improve girls’ overall performance in high school Math exams for all school types, and college entrance exam Math scores for public school girls. However, they negatively affect girls’ probability of choosing STEM majors when enrolling in college. They negatively affect boys’ high school and college entrance exam Language performance and private school boys’ college entrance exam Math performance, but positively affect boys’ college STEM preference. The presence of female Math teachers in high school has negative effects on both boys’ and girls’ college entrance exam Science scores. There is significant heterogeneity in these effects between public, voucher and private schools. The negative preference effect is significant only for</p> <p>high-performing girls.</p> <p><br></p> <p>Chapter 2 uses restricted NCES data (HSLS:2009 and ELS:2002) and difference-in-difference methodology to explore whether the $4.35 billion federal Race to the Top (RTT) program of 2009 had impacts on overall educational and enrollment outcomes, and gender gaps in these outcomes for high school students in the US. Besides the major objective of making students better prepared for college and future careers, a significant aspect of the RTT program was its emphasis on reducing barriers to women’s entry and success in STEM fields in higher education and the STEM workforce. I find that the program was not successful in fulfilling the major objectives of improving students’ educational outcomes, reducing achievement gaps or improving women’s representation and performance in STEM fields. It prompted students to take fewer and easier courses in high school and increased gender gaps in 12th grade GPA and SAT Math score. While there was a modest reduction in the gender gap in first year college GPA, there were neither any improvements in boys’ or girls’ college STEM credits and grades, nor</p> <p>any reduction in gender gaps in these outcomes.</p> <p><br></p> <p>In Chapter 3, I use the same restricted NCES data as in Chapter 2, data on state policy obtained from Howell and Magazinnik (2017) and difference-in-difference methodology to explore whether states’ adoption of “college and career ready” common K-12 standards affected the overall educational and enrollment outcomes of high school students in the US and gender gaps in these outcomes. I use the 2009 Race to the Top (RTT) program as a source of exogenous variation, since one of the major policies promoted by the program was the adoption of higher K-12 standards across the US. I find that the tougher standards led to students taking relatively more non-STEM oriented, and thus arguably “easier” courses and increased gender gaps in STEM coursetaking.</p> <p>Notably, they drove low performing girls out of college education, which resulted in a more competitive college-going female population. This in turn, led girls to outperform boys once enrolled in college, specially in STEM courses. Thus, common standards-adoption whose goal was to improve college and career readiness failed in this endeavor, but made the pool of college-going women more competitive and inadvertently levelled the playing field</p> <p>for college-bound women.</p>

<p>This dissertation consists of three chapters on economic topics related to crime and early childhood. In the first chapter, I explore the effect of a criminal justice policy on crime. The second chapter examines the relationship between gender nonconformity in childhood and life outcomes. In the third chapter, I study the relationship between access to local mental health care for children and juvenile crime.  </p> <p><br></p> <p>In my first chapter, I study the causal impact of post-release supervision on recidivism and new crime. Prior to 2011, inmates who committed lower-level offenses in North Carolina were not subject to post-release supervision. The North Carolina \textit{Justice Reinvestment Act} changed policy to require nine months of post-release supervision. Leveraging a discrete policy effective date in a regression discontinuity in time model and using administrative data from the North Carolina Department of Public Safety, I explore the effects of this legislative change on criminal outcomes. Evidence indicates that post-release supervision decreases property and violent crimes, but these changes do not persist beyond the supervision period. Results suggest that supervision leads to more individuals returning to prison at a faster rate due to technical, not criminal, violations; however, requiring lower-level offenders to undergo post-release supervision is a cost-effective program.  </p> <p><br></p> <p>The second chapter of this dissertation is coauthored with Torsten Santavirta and Miguel Sarzosa. We study the role of childhood gender conformity in determining gender gaps. We present a conceptual framework that uses gender norms to explain why some women make less profitable choices than comparable men. Using unique longitudinal survey and register data, we show that gender-nonconforming girls have substantially better education and labor market outcomes than gender-conforming girls. In contrast, gender-nonconforming boys perform substantially worse at school, sort into lower-paying occupations, earn less, and have a greater incidence of mental health disorders and substance abuse during adulthood than gender-conforming boys. Our analyses suggest that such divergence develops from an early age. </p> <p><br></p> <p>In my last chapter, I explore the relationship between mental health care for children and juvenile crime. Using data with information on facilities that specifically treat children, I exploit the county-level variation in the number of mental health treatment facilities for minors in a two-way fixed-effects model to explore the relationship between access to mental health care in a given year and juvenile crime the following year. I find that outpatient and inpatient mental health facilities for children have heterogeneous effects on juvenile crime.</p>

This thesis is an investigation into the microeconomic mechanisms that govern some of the occupational choices faced by workers in Sub-Saharan Africa, and into the monetary and non-monetary returns to their decisions. Chapter 1 begins by exploring the decision process that leads workers to allocate themselves to different occupations within the economy. In particular, I investigate the role of risk-aversion in the allocation of workers between formal and informal jobs in Ghana, hence attempting to explain a fundamental dimension of duality through an investigation into workers' preferences. In my model of sectoral allocation risk-averse workers can opt between entering the free-entry informal sector and queuing for formal occupations. Conditional on identifying the riskier option, the model yields testable implications on the relationship between risk-aversion and workers' allocation. My testing strategy proceeds in two steps. First, using the first three waves of the Ghana Household Urban Panel Survey (GHUPS) dataset, I estimate expected income uncertainty and find it considerably higher in the informal sector than in formal employment. Second, using experimental data to elicit risk-attitudes I estimate the effect of risk-aversion on occupational choices and I find that, in line with the first result, more risk-averse workers are more likely to queue for formal jobs and less likely to be in the informal sector. The conclusion of the first chapter is that attitudes to risk should feature more prominently in models of sector allocation and in the design of labour market policies, in particular when those policies aim to impact workers' vulnerability to risk and uncertainty. Chapter 2 focuses on the largest occupational category in the Developing world, self-employed workers with small productive activities, and it tries to estimate the returns to different productive assets, namely physical capital, labour and human capital. These are the workers that form most of the informal sector analysed in chapter 1, which allows me to draw a direct link with the analysis so far. The chapter begins by specifying a model for the income-generating process grounded in the literature on firms' production and hence abridging the gap between the analysis of individual earnings and the study of firms' value added. Identification in the empirics is achieved by means of panel estimators that are suitable to address the endogeneity of input choices, which derives from both time-varying and time-invariant unobservable heterogeneity. The use of these estimators is made feasible by the length of the Ghanaian Household Urban Panel Survey dataset at CSAE. I also explore issues of endogeneity in the selection of different technologies, defined by their relative capital and labour intensity. Finally, I analyse the shape of returns to capital, with the aim to detect potential non-convexities in technology. The results show that capital and work-experience play the strongest role in income-generation, while the shares of value added attributed to labour and to formal schooling are low. Marginal returns to investment are high at low capital levels and they decrease very rapidly, pointing against the existence of non-convexities due to minimum scale requirements, but implying that real income gains resulting form micro-investment are modest. Chapter 3 returns to the issue of earnings uncertainty and risk-aversion explored in Chapter 1, but it now takes the allocation choice as given and explores the direct welfare implications of income uncertainty for worker's well-being. Namely, the chapter explores the relationship between income and welfare, with a particular attention on the link between income vulnerability and happiness. Using unique longitudinal data on life-satisfaction and labour market outcomes, I estimate an individual measure of vulnerability (defined as the probability of falling below a low-income threshold) and investigate its effect on well-being. After controlling for unobservable individual fixed effects, work-satisfaction, relative income and other relevant worker characteristics, I find a sizable impact of vulnerability, over and above the income effect. When I explore the mechanisms behind my results, I find that aspiration adaptation to current income may result in a transitory income effect. Moreover, using my direct measure of attitudes to risk from field-experiments (already used in chapter 1), I can test directly the hypothesis that more risk-averse agents suffer more heavily from a given increase in income vulnerability. Overall, my findings support policy interventions that aim to reduce vulnerability, as I expect such policies to have a 'direct' impact on agents' happiness given the prevailing attitudes to risk and uncertainty in the population. Finally, from the point of view of overall social welfare, my results suggest that non-Rawlsian growth models, whereby 'someone may be left behind', may fail to enhance general welfare, for high enough levels of risk-aversion in the population, if the risk of falling behind is sufficiently widespread.

Varieties of Capitalism theory predicts that the skill specificity of workers determines their demand for social protection. In this thesis, I test this assumption using a measure of occupational mobility between pre- and post-unemployment, which I apply to European workers in different skill groups as defined by Fleckenstein et al., (2011). Using this measure as an indicator of the portability of workers' skills, I then evaluate whether the lower marketability of human capital investments is associated with greater demand for unemployment protection. The findings demonstrate that whilst this relationship is apparent in certain countries, notably Coordinated Market Economies such as Germany, the assumptions do not apply across institutional settings. Consequently, skill specificity cannot explain variation in attitudes towards unemployment protection policies between countries.

Agriculture is a major employer of labour in South Africa with about 8.8% of the total labour force directly involved in agricultural production (StatsSA, 2007a). Farm wages and working conditions in the Albany district were researched in 1957 by Roberts (1958) and 1977 by Antrobus (1984). Research in 2008, involving face-to-face interviews of a sample survey of 40 Albany farmers, was undertaken to update the situation facing farm labourers and allowed for comparisons with the work previously done. Farm workers were governed by common law until 1994 when the government intervened with legislation. The introduction of the Basic Conditions of Employment Act (1997) for farm workers, amended in 2002 to include minimum wage legislation, and the Extension of Security of Tenure Act (ESTA) of 1997 impacted the supply and demand of farm workers. Other impacts have been due to the Albany district experiencing an increase in the establishment of Private Game Reserves and game-tourism with a simultaneous decline in conventional farming. It was concluded from the survey conducted that minimum wage legislation decreased the demand for regular and increased the demand for casual labour, which incur lower costs including transaction costs, than their regular counterparts. The ESTA of 1997 contributed to a decreased number of farm residents, which had spin-off affects on the supply of labour. Farmers experienced a simultaneous price-cost squeeze, which furthermore decreased the demand for labour. Studying the working and living conditions showed that farm workers had limited access to educational and recreational facilities which negatively impacted the supply of labour.

<p>  </p> <p>My dissertation is composed of three independent chapters in the field of labor economics and international trade. </p> <p><br></p> <p>The first chapter studies marriage market signaling and women’s occupation choice. Despite the general closure of gender disparities in the labor market over the past half century, occupational segregation has been stubbornly persistent. I develop a new model that explains these occupational outcomes through marriage market signaling. Vertically differentiated men have preference over women’s unobservable caregiving ability. Heterogenous women choose caregiving occupations to signal their ability to be caregivers. My model generates unique predictions on the influence of marriage market conditions on women’s occupational choices. I find empirical support for these predictions using longitudinal data on marriage rates, policy shocks to divorce laws, and shocks to the marriage market sex ratio driven by waves of immigration. </p> <p><br></p> <p>The second chapter investigates Covid19 and consumer animus towards Chinese products. Covid19 has tremendously affected all areas of our lives and our online shopping behaviors have not been immune. China is the first country to report cases of Covid19 and suffers from rising animus in the U.S. In this paper, we study consumer animus towards Chinese products post Covid19 using Amazon data. We tracked all face masks sold on Amazon between Sep. 2019 to Sep. 2020, and collect product information that is available to a real consumer, including reviews. By analyzing both seller-generated (e.g., product name, description, features) and user-generated (e.g., reviews and customer Q&A) content, we collect information on the country-of-origin as well as consumer animus for the products. Under a fully-dynamic event study design, we find that the average rating drops significantly after a product is identified as made in China for the first time, while no such drop is found for products with other countries-of-origin. This negative impact is U-shaped, which quickly expands in the first five weeks, and then gradually fades out within six months. An informative-animus review affects the average rating of a Chinese product both directly (through its own rating) and indirectly (through other future ratings), with both mechanisms supported in data. We also provide strong evidence that the drop in average rating is driven by consumer animus instead of product quality. </p> <p><br></p> <p>The third chapter explores how cultural transmission through international trade affects gender discrimination. In this paper, I propose that international trade helps alleviate gender discrimination. With imperfect information on workers’ ability, there is statistical discrimination towards female workers. Through international trade, culture transmits asymmetrically between firms located in countries with different gender cultures. This cultural transmission benefits women because it transmits only in one direction from more gender-equal cultures to less gender-equal cultures. I prove this by linking the Customs data to the Industrial Firms data of China in 2004, and find that Chinese firms trading with more gender-equal cultures hire a higher fraction of female workers and enjoy higher profits. Similar patterns are not found in Chinese firms trading with less gender-equal cultures. The impact of cultural transmission goes beyond the firms engaged in international trade to have spillover effects onto purely domestic firms. Comparing across skill groups, cultural transmission benefits high-skill female workers more.</p>

<p>This dissertation is a collection of three papers, each one being a chapter. In the first chapter, I examine how career interruptions related to child-raising duties affect married women’s labor market trajectory, lifetime earnings, and Social Security retirement benefits. To address this question, I develop and estimate a dynamic life-cycle model of female labor supply, savings, and Social Security benefit claiming. I explicitly model the Social Security benefits system and the federal and payroll tax structure in the United States. The framework, thus, allows me to unravel the complex interdependencies between women’s participation decisions, accumulated work experience, lifetime earnings, and public pension benefits. I estimate the model using the Method of Simulated Moments and match data for the cohort born in 1943-1954. I use the model to simulate labor supply behavior of married women in a revenue-neutral counterfactual scenario where I introduce Social Security caregiver credit that covers the lost earnings during the first 5 child-rearing years through changes in retirement benefits. I find that the model predicts that introducing the provision of earning credits for child care in the Social Security system would reduce participation of married women during the child-rearing years, but the contributory nature of the caregiver credits creates incentive to work in the post-childrearing period.</p> <p><br></p> <p>In the second chapter, I evaluate the implication of introducing Social Security caregiver credit by examining whether and to what extent implementing this child-related policy to the existing Social Security system would help reduce the motherhood earnings penalty over the life cycle. I utilize the dynamic lifecycle model developed in Chapter 1 and conduct revenue-neutral counterfactual policy experiments by introducing caregiver credits in in various settings. First, I analyze the effect of the policy at different generosity levels; and second, I implement the policy in the presence as well as in the absence of the marriage-based noncontributory Social Security benefits (spousal and survivors benefits). Third, I conduct an alternative counterfactual policy experiment where mothers could drop five additional work years from the average career earnings (or equivalently Social Security benefits) calculation. The model predicts that lifetime pre-tax labor earnings of married women increase significantly under when the caregiver credits are introduced in a setting where spousal and survivors benefits are eliminated, and there is a sizeable reduction in gender gap in average career earnings at the Social Security Early Retirement Age. The alternative policy of computing average career earnings based on a mother's 30 years of earnings has significantly smaller impact on offsetting motherhood earnings penalty through retirement benefits compared to the provision of Social Security caregiver credits.</p> <p><br></p> <p>In the third chapter, I examine the effect of Supplemental Nutrition Assistance Program (SNAP) work requirement reinstatement on food insecurity outcomes of able-bodied adults without dependents (ABAWDs). The policy restricts SNAP benefits of ABAWDs to 3 months in a 36 month period if they are not working or participating in any work program for at least 20 hours a week. In the aftermath of the 2008 recession, the American Recovery and Reinvestment Act of 2009 waived work requirements nationwide, and many states reimplemented the work rule at different times beginning in 2011. I employ a difference-in-differences approach utilizing this cross-state variation in the reimplementation of the policy. Using rich information on food affordability and food intake behavior from the Food Security Supplement of the Current Population Survey (CPS-FSS), I find that promoting work for food assistance improved the overall food security status of ABAWDs by reducing disruptions in food intake, anxiety over food affordability and dependency on emergency food receipt. Subsample analyses indicate that effects are stronger for never married and less educated ABAWDs.</p>