Research

Does economic depression stimulate unionization? I investigate the first major wave of union formation in U.S. history following the Panic of 1873 and the ensuing Long Depression (1873–1879). Using bank loan data and newly digitized wage data, I construct two new measures of local depression severity and find that unionization was more prevalent in hard-hit labor markets. Through the lens of a model of workers’ unionization decisions, this result reflects the interplay of two opposing forces. A downturn can encourage unionization by reducing the value of incumbent non-union employment relative to workers’ outside options, leaving a failed attempt to unionize and potential job loss less costly. Conversely, deteriorating reemployment prospects can discourage unionization by increasing the cost of a potential job loss. The empirical evidence is consistent with the depression strengthening workers’ incentives to unionize. Heterogeneity analysis suggests the relationship between unionization and depression is even stronger in concentrated labor markets.

We study the impact of Intimate Partner Violence (IPV) on various aspects of the victims’ lives throughout the course of violence, including their labor market performance, marital decisions, fertility decisions, and health outcomes. Our data consists of the universe of administratively reported IPV cases primarily from hospitals, police, and helplines in Taiwan from 2012 to 2018. We distinguish the violence effect and the report effect. We use later-treated victims as control groups under an event study framework. Among all victims who eventually report IPV to the officials, we find (i) women’s employment rate decreases after the onset of violence, but an incremental positive effect follows after reporting, particularly among young and educated women; (ii) the probability of divorce increases for both men and women after the report but not at the violence onset; (iii) reduction in additional childbirth after reporting regardless of the victim’s gender; (iv) rising depression outpatient-visits after violence and reporting among women.

We use population-wide administrative health records from Taiwan to estimate intergenerational persistence in health, providing the first estimates for a middle-income country. We measure latent health by applying principal components analysis to a set of indicators for 13 broad ICD categories and quintiles of visits to a general practitioner. We find that the rank–rank slope in health between adult children and their parents is 0.22, which is broadly in line with results from other countries. Maternal transmission is stronger than paternal transmission, and sons have higher persistence than daughters. Persistence is also higher at the upper tail of the parent health distribution. Persistence is lower when complete data on outpatient care is unavailable. Health transmission is almost entirely unrelated to household income levels in Taiwan. We also find that there are small geographic differences in absolute health mobility across townships and that these are modestly correlated with area-level income and doctor availability.

Taking advantage of merged administrative data, we estimate the intended and unintended effects of two public policies aimed at combating driving under influence (DUI) of alcohol in Taiwan: the three-month-long campaign of random sobriety checkpoints in mid-2012, and the legal reform carried out in mid-2013 leading to much harsher punishments for DUI offenders. Our study gives rise to several important findings: (i) The checkpoint campaign substantially reduced DUI deaths and injuries, and the effect persisted after the campaign ended; (ii) the campaign also had a strong spillover effect of reducing non-DUI deaths and injuries, and this unintended effect was four times the size of the intended effect on DUI cases; (iii) these effects were not driven by social pressure or self-awareness arising from the intensive media coverage taking place at the same time; and (iv) there was little response to either policy among repeat DUI offenders and those who caused serious injuries.

We investigate the causal impact of radio exposure in early life on individuals' subsequent socioeconomic outcomes in the U.S., focusing on education, family formation, labor market outcomes, and migration. Using linked census records, we instrument county-level radio exposure in 1930 with ground conductivity, exploiting plausibly exogenous variation in AM radio reception. Preliminary estimates suggest that greater radio exposure increased educational attainment, wages, and occupational income scores, while reducing marriage rates and the number of own children living in the household by 1940. We also find some evidence of higher labor force participation and increased migration to urban areas.

I examine the short- and long-term consequences of child labor laws in the U.S., focusing on state laws that raised the minimum working age in manufacturing from twelve to fourteen. Combining full-count census data with linked individual records, I use a difference-in-differences design that compares boys below and above the new age threshold before and after reform in northeastern and midwestern states. The employment effects differ across reform waves: laws enacted between 1880 and 1900 had no statistically detectable effect on manufacturing employment by 1900, whereas reforms adopted between 1900 and 1910 substantially reduced employment among affected boys. Following individuals into adulthood, I find that exposure to these laws increased the likelihood of marriage and parenthood by their early twenties, with no effects on educational attainment or occupational scores. These findings suggest that child labor restrictions can shape subsequent family formation without necessarily generating improvements in schooling or later occupational outcomes.

This project examines how institutional environments shape whether and when victims report intimate partner violence and how these decisions affect underreporting. Because administrative records capture only reported cases, they may obscure violence prevalence and introduce selection bias. Using Taiwanese administrative data, I plan to combine evidence from institutional reforms with a dynamic model of reporting behavior. The empirical analysis will assess how access to protective services affects reporting rates, delays, and case severity. The model will characterize how victims weigh protection benefits against retaliation risks, economic uncertainty, and reporting costs. Reporting patterns and survey evidence on prevalence will inform estimates of underreporting and counterfactual evaluations of policies that lower reporting barriers. This research aims to improve the interpretation of administrative records and inform policies that facilitate timely access to protection.

We study the impact of child abuse on children's later life outcomes. Using linked administrative data from Taiwan, we employ an event-study design to compare children whose first abuse reports occur at different times. Our preliminary findings indicate adverse health outcomes and elevated healthcare utilization across multiple categories of medical services for several years after reporting. These patterns suggest sustained health needs beyond the initial reporting period. By documenting the duration and breadth of these changes, the study highlights the importance of continued healthcare support for children who have experienced abuse.

We study the short- and long-run consequences of sexual violence for victims. Using linked administrative data from Taiwan, we employ a staggered difference-in-differences design that exploits variation in the timing of first reports. Among adolescent girls, preliminary estimates indicate a persistent increase in healthcare utilization following reporting. We also find evidence of earlier entry into employment, potentially reflecting efforts to achieve greater financial independence. These findings suggest that adolescent girls who experience sexual violence may require additional economic support as they face pressure to become financially independent.