Association for Public Policy Analysis & Management

This study examines how increased Immigration and Customs Enforcement (ICE) activities impacted marriage probabilities for pregnant Hispanic women in North Carolina. We hypothesize that ICE activities decreased marriage rates among Hispanics along three margins: deportations of Hispanic males; selective migration of Hispanics away from the areas under ICE scrutiny; or strains on relationship quality (either through increased psychological stress or decreased employment and earnings of the father). Our study suggests that ICE enforcement influences marriage, with likely ramifications for the health and well-being of the mothers and children involved.

In 2006, NC became the first state east of Arizona to have a county that implemented a 287(g) program (named for the Section 287(g) of the Immigration and Nationality Act). These programs establish partnerships between ICE and local law enforcement that allows local offices to act as immigration enforcement agents. 287(g) programs were first introduced in Mecklenburg county (home to the NC’s biggest city, Charlotte) and then expanded to other NC counties. Nationwide, the implementation of 287(g) programs has led to an increase in unauthorized immigrants being removed from the United States, but has also led to increases in fear and mistrust among Hispanics.

We focus on the largest counties on the state, and use administrative birth record data on approximately 430,000 live births between 2001 and 2006. The birth record data has information on mother’s country of origin, her Hispanic ethnicity, and education level, which we triangulate to identify Hispanic mothers who are most likely undocumented. The primary dependent variable is the share of mothers who were married at the time of birth in a given month in a given county, contrasting counties in which immigration enforcement was increased with those in which it was not.

We use difference-in-difference and triple difference models comparing marriage likelihood for likely unauthorized Hispanic mothers before and after the 287(g) program implementation with those of mothers from the same demographic group but who reside in non-adopting counties around the same time. We also compare marriage likelihood among other demographic groups (e.g., U.S. born) unlikely to be directly affected by ICE activity. Our method assumes that, in the absence of the 287(g) program, changes in marriage among those most likely affected by increased enforcement in a given county would be equivalent to changes in marriage amongst similar mothers in the rest of the state.

Preliminary results suggest that the implementation of 287(g) programs decreased the share of likely unauthorized Hispanic mothers who were married at the time of the birth. The biggest decrease occurs about six months after a 287(g) program was implemented. Estimates were robust to the inclusion of county and time fixed effects or flexible controls for county-specific seasonality. A similar decrease in the share married was not observed among U.S.-born whites or blacks.