Resumen de Three essays in empirical public economics
In this dissertation, I study the role of policies and institutions to foster social inclusion. In particular, in two of my projects, I use the Survey of Health, Ageing, and Retirement in Europe (SHARE) dataset to analyze the impact of institutions and policies on individuals' decisions. In the first chapter, I exploit the historical context and study how the Soviet regime changed women's choices within the Soviet sphere regarding educational attainment, labor participation, marriage and fertility. In the second chapter, I consider the role of employment status and the probability to provide informal care to elders in Europe. Finally, in the third part of my doctoral dissertation, I study the role of self-regulation to mitigate the ethnic discrimination on the largest hospitality platform, Airbnb.
The first chapter, ``USSR, Education, Work History, Fertility Choices, and Later-Life Outcomes'' (with Telmo Pérez-Izquierdo), investigates the difference in the impact of the Soviet regime on life decisions within the Soviet sphere. In the 20th century, communist institutions appeared in several forms in Eastern Europe. According to its philosophy, women have the same equal rights as men in economic, political, and family life. In contrast, the male bread-winner family structure was prevalent in capitalist societies. In recent years, economists evaluated the impact of communist regimes on individuals’ decisions, focusing on a communist versus non-communist comparison. However, communist countries were not homogeneous and experienced different forms of the Soviet regime.
In this paper, we are interested in the impact of the different forms of communism within the Soviet sphere. Indeed, communist countries were not homogeneous. Communist institutions greatly differ from the Union of Soviet Socialist Republics (USSR) to the Eastern Bloc. Specifically, we study the effects of different forms of communism on women's schooling decisions, labor participation, and fertility choices during the regime to abstract from the transition period.
Both individual decisions and political regimes are endogenous to country-specific factors, making it challenging to quantify the impact of the different shades of the Soviet regime from other effects. To overcome this problem and get a causal estimate, we restrict our analysis to the former territories of the Russian Empire in Lithuania and Poland. These two countries show a similar history and patterns before the start of the Second World War. Yet, since the Molotov-Ribbentrop Pact (1939), Lithuania became one of the republics of the USSR and got exposed to the same central government as the rest of the USSR. On the other hand, after World War II (WWII), the Polish National Government continued to exist and formed the Polish People's Republic (1947 - 1989). We exploit this divergence as a natural experiment to shed light on the impact of different forms of communist regimes.
We use the retrospective SHARELIFE data to analyze the educational, labor, marriage, and fertility decisions of East Europeans from 1950 to 1990. Our identification strategy is a natural experiment. We assume that Lithuania became part of the USSR and not of the Eastern Bloc due to exogenous factors unrelated to outcome variables relevant to this study. For 40 years, Lithuania was a part of the USSR, whereas Poland was a part of the Eastern Bloc.
We find that during communism, Lithuanian women worked two years more by age 50 relative to Polish women. This effect is half of the one found for the East-West Germany comparison. We refer to working status only before the regime's fall in 1990, not including the transition period. The impact of the USSR gets larger when we consider low-educated women. Accordingly, we account for heterogeneity when estimating the impact on experience. Regarding the intensive margin of the USSR's impact on labor participation, we find that women from early birth cohorts (1935 - 1940 and 1941 - 1946) work more. This is in line with the idea that the treatment intensity is the largest for early-born individuals.
Moreover, we observe that being part of the USSR increased the educational attainment for all residents. Yet, the total impact on the education of women was three times larger than on men. We propose a potential mechanism behind this fact: an indirect channel of improved work opportunities on female education.
Our paper relates to several strands of literature. First of all, we contribute to the literature about the direct impact of communism on residents. We add to this literature by looking at a new environment and a different regime in Europe. It can be of particular interest in light of the recent paper by Becker, Mergele, and Woessmann (2020) who stress the potential bias in East and West Germany comparison due to pre-separation differences along the newly assigned country borders after World War II.
Furthermore, apart from bringing to attention a new setting, our estimates are based on capturing the difference within communism (e.g., due to private ownership or religion in Poland during the communist period). Consequently, results are not driven by the divergence between egalitarian and traditional regimes. Our findings are not affected by how much a male-bread winner family structure discouraged women from working. This is not the case when we consider East-West Germany, and two very distinct leading regimes: communism (East Germany) versus capitalism (West Germany). In addition to our main study, we repeat our specification in the East and West Germany context, using the SHARELIFE data. In this case, the coefficient of communism on women's experience gets even larger, because the control group is different: Poland before, West Germany now. In particular, East German women worked 5 years more by age 50 than West Germans, which is twice larger than in Lithuania-Poland comparison. It confirms the divergence of the regimes during the separation. The third contribution to this literature is an analysis of individual choices under the USSR regime. Instead of focusing on the persistence of the impact of the regime, we study the factors that shaped individual choices across birth cohorts from 1935 to 1950, isolating the period of the transition from planned to market economy.
The second group of articles we contribute to is the impact of gender-egalitarian policies on educational and work choices. Given the potential complementarity among institutions, the only way to estimate policies’ synergy is to look at the historical context. The case of Eastern Europe, that experienced communism, gives us the laboratory to address the research question.
Accordingly, this paper's important implication is the critical distinction between the Soviet exposure between Lithuania and Poland for more than 40 years. Up to our knowledge, this notable difference in the political-economic regime is little pronounced in the literature, and often, researchers pool all Eastern European countries together for what regards the Soviet inheritance. Within post-Soviet states, Baltic countries were part of the Soviet Union, along with other countries that formed the Eastern Bloc. We argue that contemporaneous policies should be tailored to the historical context and not ignore communism's difference in these countries.
The second chapter, ``Impact of Employment on Informal Caregiving to the Elderly Mothers in Europe'', studies the trade-off faced by adult individuals in Europe between participating in the labor market and providing informal care to their elderly mothers. In 2019, almost a quarter of European residents was 65 years old or more, and soon the proportion of elderly is expected to increase even further. Long term care (LTC) expenditures do not necessarily cover all care loads, and elders often need to rely on informal caregivers, mainly their children. In many cases, potential caregivers work or are looking for a job. Thus, for many adults in Europe, the decisions to provide informal care and participate in the labor market influence each other: those with elder parents may have to quit their jobs to provide care, and those with less prosperous work opportunities may be more likely to become caregivers. Understanding how these two decisions are interconnected is critical to guide policy recommendations regarding formal care expenditures and more favorable labor conditions for potential care providers.
Until recently, scholars have mainly studied the impact of informal care provision on individuals' labor participation, whereas the opposite channel - how the work status affects the care choice - has been mostly neglected. Accordingly, this paper aims to close this gap in the literature and identify the causal impact of the working choice on care. Specifically, to solve the simultaneity problem between these two decisions and isolate the endogeneity in providing care, I exploit the shift in a macro-level variable (unemployment rate) due to the Great Recession that affects the individuals' propensity to care through their participation in the labor market. As a result, after controlling for observables, I find that being employed significantly reduces the probability of providing care to an elder.
Using the SHARE, retrospective SHARELIFE data, and the Job Episode Panel, I model both endogenous processes jointly to capture the simultaneity of binary care and work decisions. Moreover, to identify the causal impact of work on care, I consider the following exclusion restriction for the work choice: the country's exposure to the Great Recession measured through changes in the unemployment rate. To quantify the impact of the crisis on labor participation, I consider a change in the unemployment rate in a country. Further, I allow for a heterogeneous effect of the crisis across men and women in the same country. The identification strategy allows for the correlation of unobservables in work and care decisions. The analysis focuses on care provided to mothers as they are more likely to receive care from working-age individuals. Moreover, this paper studies both extensive and intensive margins; the later is measured through care provision frequency.
The main identification strategy assumes that the Great Recession does not directly affect the care choice after controlling for an extensive set of observed variables, such as work experience, the health of caregiver and care receiver, residential proximity, and proxies for prices and availability of formal care. To construct work experience, I exploit the SHARELIFE data and the Job Episode Panel that were part of the SHARE survey in wave 3 (2009) and wave 7 (2017). This retrospective information allows for tracking life decisions from birth to the moment of the interview. Cumulative work experience is one of the key predictors of employment among elders, so I restrict the sample to individuals who participated in the SHARELIFE survey.
This study documents an increase by nine p.p. in the probability of providing informal care to a mother when an adult child does not participate in the labor market. Female caregivers mainly drive this result.
This paper contributes to the literature about informal care provision. Up to my knowledge, this paper is the first one to study the causal impact of labor participation on care provision in Europe. The majority of papers investigates the opposite channel - the impact of care on the employment status. Literature in this field often uses a two-stage least squares strategy and proposes an instrumental variable to correct for potential endogeneity in the decision to give informal care.
Accordingly, this study adds to the literature by considering the impact of participation on care choice. Moreover, I directly exploit both dependent variables' binary nature and apply a bivariate model to compute the discrete change in work over care variable. In this setting, the average partial effect more likely leads to an appropriate approximation comparing with the local linear approximation. Finally, I contribute to the list of articles based on the SHARE data by documenting the sample selection problem in conducting a panel data analysis. Specifically, in this case, the target sample should include respondents who are present at least twice in the survey: 1) being above 50, 2) being below statutory retirement age, and 3) having a mother alive in that periods. These restrictions lead to a positive selection of individuals regarding observable characteristics, such as age, health, and education. Even though controlling for permanent unobserved heterogeneity can be important in this context, any study related to informal care provision to the elderly should not neglect the sample selection problem, in particular, using the SHARE data.
The result of this analysis - a significant increase in care responsibilities in the absence of employment opportunities in Europe - has potential policy implications directly related to the trade-off between work and care to elders. In countries with a growing percent of elders, policymakers may be tempted to keep more individuals into the labor market to sustain a larger share of non-working elder population. However, if potential caregivers stay longer in the market, according to the results of this analysis, they will provide less care to elders. Keeping more individuals in the labor market may be beneficial for a variety of reasons, but policymakers should be conscious that it can reinforce the problem of unmet needs in elders' care when formal care provision is scarce.
The third chapter, ``Online Discrimination and (Self) Regulation: Evaluating the Airbnb's Nondiscrimination Policy'' (with Michelangelo Rossi), is motivated by the following fact digital platforms have changed the ways of doing business in many markets. During the last two decades, the volume of online transactions has enormously risen across different sectors and industries. Platforms such as Airbnb and Uber changed the traditional ways of doing business in the hospitality and transportation markets, respectively. Thanks to these websites, a great number of non-professional sellers entered those markets. However, some characteristics of the transactions that now occur on these digital marketplaces are unaltered relative to traditional off-line settings; unfortunately, one of them regards the discrimination of minorities by sellers and buyers.
In the most recent years, evidence shows the difficulties by non-white users to fully integrate into digital platforms. Regarding Airbnb, discrimination is present on both sides of the market (guests and hosts): booking requests sent by users with African American sounding names are less likely to be accepted relative to users with white-sounding names. On the other side, Asian and Hispanic Airbnb's hosts charge lower prices with respect to white hosts with similar properties.
To reduce discrimination and partially remedy the absence of clear laws, platforms often implement self-regulation to establish users' conduct rules. Among others, Airbnb launched a Nondiscrimination Policy at the end of 2016, prohibiting any discriminatory behavior by hosts and setting up several objectives to be reached in the next years in terms of inclusiveness.
This paper studies this policy in four US cities: New Orleans, New York, Portland, and San Francisco. In the two years after the policy, we document an increase in the share of hosts activating a setting called ``instant book'' that automatically accepts any guests' requests.
With a difference-in-differences and an event study approach, we estimate variations in the number of non-white guests before and after the policy comparing hosts who can choose to reject a guest request (and potentially discriminate) with hosts who already activated the ``instant book'' setting before the policy.
We find that the number of hosts who decide to activate the instant booking option and accept all booking requests gradually increased after the policy. Further, results show hosts who can choose to reject a guest request do not rent more to non-white guests after the policy. Accordingly, non-instant bookable hosts (the majority on the platform) did not adjust their behavior.
We find that, after the policy, the share of non-white guests slightly increases. Moreover, the promotion of the instant booking option leads to a higher share of hosts who accept all booking requests and cannot select guests. Still, the policy was not able to improve the proportion of non-white guests accepted by those hosts who can discriminate among guests (instant booking off). This may signal the reluctance of certain hosts to accept non-white guests in the absence of a proper nudge by the platform.
We conclude that Airbnb's Nondiscrimination Policy may be considered as the first step toward inclusiveness. This self-regulation increases renting opportunities for non-white guests thanks to expanding the ``instant book'' setting among Airbnb's hosts. Still, further measures have to be taken to encourage hosts who do not want or cannot activate this setting.
