Four essays on the economics of maternity care for health policy: evidence from Peru

Women's health outcomes and welfare in childbirth are intrinsically related to women's choice of provider and providers' staffing and decision making. But these factors have rarely been assessed empirically for a single population. This study is an attempt to do so in a systematic way. I use a unique and rich patient-level dataset that spans 2011 to 2015 from Lima, Peru, and use a variety of methods to address four research questions: a) conditional logit models to assess whether women trade-off continuity of care in midwife-led maternity units against structural quality in obstetrician-led maternity units; b) synthetic difference in difference methods to evaluate the policy of opening new midwife-led maternity units on use of health care and conditional logit for evaluating its effects on women's welfare; c) linear probability models with provider and obstetrician fixed effects to investigate whether the ratio of obstetricians/child deliveries on Sundays (when there are non-scheduled C-sections) afects the C-section rate upcoding by obstetricians and women's health outcomes, and d) instrumental variables methods to investigate the efect of C-sections on maternal health outcomes. I found that women that receive their antenatal care locally are prepared to travel further afield when it comes to giving birth, suggesting that they are prepared to accept less care continuity in midwife-led maternity units and incur higher travel costs to go to maternity units with better facilities. The introduction of new midwife-led maternity units increases women's welfare, but this efect is small compared with the investment required to set up the new midwife-led maternity units. New midwife-led maternity units have the potential to improve women's health outcomes only if they are close to an obstetrician provider and there are no midwife-led maternity units nearby. More obstetricians on duty on Sunday leads them to perform more C-sections and they justify this behaviour by up-coding comorbidities. The variation in the number of obstetricians has mixed effects on women's health outcomes improving some and worsening others. Finally, I show that C-sections are bad for women's health, especially causing haemorrhage in low C-section-risk women. (Haemorrhage is one of the main causes of maternal death in Latin American countries).