By William Song and Theresa Tong
A substantial body of literature on the wage effects of marriage finds that married American men earn anywhere from 10% to 40% higher wages than unmarried men on average, while married American women earn up to 7% less than unmarried women, even after controlling for traits such as background, education, and number of children. Because this literature focuses heavily on men born in a single time period, we study both men and women in two different generational cohorts of Americans (Baby Boomers and Millennials) from the National Longitudinal Surveys of Youth to examine how the wage effects of marriage differ between genders and across time. Using a fixed effects approach, we find that Millennial women—but not Baby Boomer women—experience an increase in wages after marriage, and we replicate the finding from the literature that men experience an increase in wages after marriage as well. However, after controlling for wage trajectory-based selection into marriage by using a modified fixed effects approach that allows wage trajectories to vary by individual, we find that the wage effects of marriage are no longer statistically significant for any group in our data, suggesting that the wage differences between married and unmarried individuals found in previous studies are primarily a result of selection.
Advisors: Professor Marjorie McElroy, Professor Michelle Connolly | JEL Codes: C33; D13; J12; J13; J22; J30
By Grace Peterson
This research determines how automation affects immigrant wages in the US and how closely this impact follows the skills-biased technical change (SBTC) hypothesis. The present study addresses this question using American Community Survey (ACS) data from 2012 to 2016 and a job automation probability index to explain technological change. This research leverages OLS regressions to evaluate real wage drivers, grouping data by year, immigration status, and education level. According to the SBTC hypothesis, high skill immigrant wages should be less negatively affected by technological change than low skill immigrant wages. Univariate analysis suggests that the SBTC hypothesis is even stronger for US = immigrants than native-borns, as high skill immigrants have a lower average probability than low skill immigrants of having their jobs automated, and the difference in effect on high versus low skilled workers is larger for immigrant than native-borns. However, multivariate analysis asserts that technological change affects low skill immigrants’ wages less than high skilled individuals’ wages, which counters the SBTC hypothesis.
Advisors: Professor Grace Kim | JEL Codes: J15, J24, J31, J61, E24
By William J. Battle-McDonald
This paper examines how the quantity and quality of admissions applications to Division 1 colleges and universities were affected by two non-academic factors: (1) performance of a school’s men’s basketball and football teams; and (2) scandals associated with these athletic programs. Admissions data from 2001 – 2017 were compared to team performance during their football and basketball seasons in order to understand how these non-academic factors contribute to an individual’s decisions to apply for admission. A multivariate linear regression model with school and year fixed effects supported the hypothesis that athletic success positively affects the quantity of applications, increasing them by up to 3% in basketball and 11% in football in the following application period. Seasonal football success was also shown to have negative impacts on the distribution of standardized testing scores of future applicant classes, however these scores were shown to increase when a team played their best season in five or more years. Additional analysis of the effects of athletic program scandals reveals a significant negative effect on the number of applications received, although a deep dive into a few of the most prominent scandals suggests that the benefits associated with violating NCAA rules may, under the right circumstances, be well worth the risk.
Advisor: Dr. James Roberts | JEL Codes: I23, J24, L82, L83, Z2
By Hayley Jordan Barton
This research explores the potential mitigating effect of Mexico’s conditional cash transfer program, Oportunidades, on the education and labor impacts of increased homicide rates. Panel data models are combined with a difference-in-differences approach to compare children and young adults who receive cash transfers with those who do not. Results are very sensitive to specification, but Oportunidades participation is shown to be positively associated with educational attainment regardless of homicide increases. Homicides are associated with decreases in likelihood of school enrollment and compulsory education completion; however, they also correspond with increases in educational attainment, with a larger effect for Oportunidades non-recipients.
Advisors: Dr. Charles Becker, and Dr. Michelle Connolly | JEL Codes: C23; D15; I20; I38; J24
By Meredith Parenti
This paper offers an empirical analysis of the effects of division of employment between parent figures on future outcomes of children. Using propensity score matching, this study extends upon previous research to offer an understanding of the joint role played by maternal employment and that of a present or non-present husband. Data from the National Longitudinal Survey of Children and Young Adults and the National Longitudinal Survey of Youth 1979 are used to create and examine a comprehensive set of respondents’ backgrounds and labor outcomes. Relative to children from two parent homes with only working fathers, children from two parent homes with part-time working mothers and non-working fathers appear to have less positive labor outcomes. Conversely, children with full-time working mothers without spouses in the home have more positive labor outcomes. These findings demonstrate the mediation of each parent figure on the role of the other in determining outcomes and suggest maternal work is beneficial to children, or plays no significant role, unless a mother has to support not only her children, but also her husband through her employment.
Advisors: V. Joseph Hotz and Michelle Connolly | JEL Codes: J01, J12, J22
By Nicholas Thomas Gardner
This paper works towards developing the narrative of orphans whose parent or parents died from natural disaster. By taking advantage of the unanticipated nature of death from the 2004
Indonesian tsunami, orphanhood can be treated as much closer to random than similar literature using data centered on HIV/AIDS related deaths. We use a community level fixed effects model to attempt to derive a causal relationship between orphanhood and both education and log wages. Our models suggest that orphaned males aged 14 and older at baseline complete 1-2 fewer years of education than their cohorts. The adverse effects persist in the long-term, as these orphans earn 26% less than non-orphan cohorts.
Advisors: Duncan Thomas and Kent Kimbrough | JEL Codes: I24, I25, I31, J24, J31
By Shafiq Haris, Alexander Prezioso, Michael Temple, Logan Turner, Kevin Zipf, Elizabeth Di Giulio, and Joseph Ueland
This paper analyzes the impact of exogenous shifts in the labor market on the marriage market. The relationship between these two markets is complicated by their reverse causality. That is to say, labor market decisions play into marriage market decisions, and vice versa. In order to mitigate this simultaneous determination, this paper adopts and furthers a methodology utilized by Autor, Dorn and Hansen (2015). Henceforth referred to as ADH, the authors analyze the effects of trade on local labor markets between 1980 and 2007. All 722 commuting zones in the continental United States were evaluated with respect to their level of exposure to increasing competition from Chinese imports, and the share of jobs within the commuting zone considered “routine,” and thus susceptible to computerization and/or mechanization. The authors analyze the impact of these independent variables on labor force participation. This paper takes Autor, et al’s analysis one step further by using the routinization and trade variables as instruments through which we can observe the exogenous impact of the labor market on marital status shares. This paper progresses through two specifications before ultimately utilizing a Two–Stage Least Squares analysis with Autor et al’s instruments to isolate the impact of decadal changes in the labor market on decadal changes in male and female marital status shares. Analysis is performed on different age groups, as both the marriage and labor market are different for people of different ages. The first specification applies Autor, et. al’s right–hand side with marital status shares as dependent variables. The second specification adds labor market ratios, which relate male and female labor market status. The previously mentioned final specification offers easily interpreted results and is the most encompassing model. Overall, we find that the labor market affects the marriage market much like the current literature would suggest. For example, as male employment increases, the share of females never married decreases and the share of females married increases. This relationship is consistent with existing marriage market theory. However, the results suggest that the literature does not hold in the oldest age group in the data, as power dynamics in the marriage market shift. Our methodology and findings are unique, as we explore this field through a new lens. Future research can expand upon this by incorporating a dataset with information regarding cohabitation habits and consistent longitudinal variable measurements for controls.
Advisor: Marjorie McElroy | JEL Codes: J1, J12, J21 | Tagged: Employment, Marriage
By Meghan Mcaneny
In this study, I investigate the relationship between the percentage of women in leadership in a company and its employee benefits. This paper uses data on individual firms’ benefits such as paid parental leave, familial support, and flexibility arrangements. Using OLS, I conclude that benefits that shift familial burdens from women to men, specifically paid paternity leave, result in more women in leadership. This creates an even playing field for women to be promoted as the company environment does not penalize women for using benefits. I also find a negative relationship between reimbursement for fertility procedures and women in leadership.
Advisor: Alison Hagy, Marjorie McElroy | JEL Codes: J22, J32, M51, M52 | Tagged: Employee Benefits, Women in Business Leadership, Work-Life Balance
By Rachael Anderson
Although Turkey ranks among the world’s 20 largest economies, female labor force participation in Turkey is surprisingly low. Relative to other developed countries, however, the proportion of Turkish women in senior management is high. One explanation for these contrasting pictures of Turkey’s female labor force is education. To better understand how women’s education and household characteristics explain variations in Turkey’s female labor market, I use annual Turkish Household Labour Force Survey data from 2004–2012 to estimate five probabilities: the likelihood that a woman (1) participates in the labor force, or is employed in an (2) agricultural, (3) blue collar, (4) lower white collar, or (5) upper white collar job. I find that labor force participation is relatively high among female primary school graduates, who are most likely to work in agricultural and blue collar jobs. Highly educated married women are the most likely group to participate in upper white collar jobs, and families favor sending single daughters over wives to work during periods of reduced household income.
Advisor: Kent Kimbrough, Timur Kuran | JEL Codes: C51, J21, J23 | Tagged: Employment, Labor-force Participation, Occupation Women