This dissertation is two essays that examine the impact of two distinct structural changes in hospitals. The first essay examines whether legislated changes in the minimum postpartum length of stay improve health outcomes for newborns. The second essay examines the consequences of nurse unions in hospitals.

Much of the previous research about the relationship between postpartum length of stay and patient outcomes are potentially subject to an omitted variable bias because sicker newborns usually stay longer in the hospital.  We overcome this problem by using passages of early discharge laws as a quasi experiment that generated exogenous increase in length of postpartum stays.  The California Newborn's and Mother's Health Act of 1997 effective on August 26, 1997, mandated that private insurance carriers provide coverage for at least 48-hour hospital stays for normal deliveries and 96-hour hospital stays for cesarean deliveries.  A similar federal law went into effect on January1, 1998.  Using an interrupted time series design, we demonstrate that early discharge laws reduced considerably the fraction of newborns and mothers who were discharged early.  In two-stage least square models using the state and Federal law as instruments for the length of hospital stay, we find that an additional day in the hospital reduces the probability of readmission by about one percentage point for complicated vaginal deliveries and c-sections of all types.  For uncomplicated vaginal deliveries, we find there was no statistically significant change in 28-day newborn readmission rates.  

The second essay examines the impact of nurse unionization in hospitals on wages, hours, staffing ratio of nursing personnel.  Using hospital-level panel data merged with data on union elections from the National Labor Relations Board, we compare the outcomes of nursing personnel in hospitals that became unionized during the sample period with hospitals that did not change union status.  The results indicate that unions have a small negative impact on nurse wage rates and they encourage hospitals to use a larger fraction of contract employees.  These difference-in-difference estimates also indicate that cross-sectional regressions tend to overstate the wage gains of union because unions are more likely to appear in higher-wage hospitals.