This research project aimed at evaluating the effects of ante- and neonatal care programmes on children’s and mothers’ health and socioeconomic outcomes later in life. In particular, we exploited two different historical interventions which were carried out in Sweden in the 1930s and 40s. The first intervention was designed as a field trial and took place between 1931 and 1933. It offered free ante- and neonatal care to mothers and infants of 7 chosen health districts. Services included guidance and examinations at the surgeries, home visits and information campaigns. The second intervention involves the subsequent roll-out of these services on a national level from 1938 onward.

Execution of the project

These two natural experiments allow us to draw clear conclusions regarding the effectiveness of ante- and neonatal care programmes since we can estimate their causal effects. Exploiting regional and time variation in the implementation of the programmes, we employed difference-in-differences estimators to study the effects on mortality, health, educational attainment and socio-economic outcomes later in life. Our research regarding the effects of the field trial have been finished and a first paper is published; a second paper has been submitted for publication. A third research paper regarding the national roll-out is in progress.   

To conduct the analyses, several purpose-built datasets were created and linked with standard administrative datasets. The main effort has been put into the digitisation of individual-level information on children born between 1930 and 1934 in treated and matched control regions. Therefore a unique dataset of more than 25,000 individuals containing background information (e.g. age, gender or socio-economic status), test scores and sickness absence in the first and fourth school year (currently for a subsample of about 60% of the individuals) and individual utilisation data has been created. It is complemented by regional aggregates on public health and demography and official medical, population and death statistics, which were needed to identify suitable control regions and to conduct the analyses. These purpose-built datasets are ready for use and will be accessible on a data server provided by Statistics Sweden to Lund University.

The main aim of the intervention was to improve health outcomes of expectant mothers and their children. We therefore explored the effects of the field trial on mortality in a first step. Results regarding the prenatal care component of the intervention were ambiguous so that we decided to focus on the effects of the infant intervention only. Our findings indicate that the average duration of programme exposure in infancy led to significant reductions in mortality for exposed children.

The second part of the project investigates the effects of the field trial on academic performance and later life socio-economic outcomes. We conducted digitization of school records and record linkage at a very large scale (20,000 individuals) and the collection of school data was completed within the duration of the project (we are currently tracking down individuals who moved between birth and school years, which turned out to be very time consuming). Our results in this part indicate that the infant intervention led to better academic performance (especially in fourth grade) and decreased sickness absence in first grade for females. Long-term outcomes show an increase in the propensity to work fulltime and in the likelihood of having secondary education for women, with large associated gains in earnings.

The empirical analysis on the national roll-out, is based on data on the numbers of health care facilities implemented as well as shares of expectant mothers and infants treated. These data have been purposively digitized for the project. We estimate the effects of the intervention on aggregate measures of fertility, length of pregnancy, stillbirths as well as on maternal and infant mortality to find out whether improvements in these rates have been caused by the intervention. For this purpose, we use aggregate data from official medical, population and death statistics. Additionally, we are interested in the long-term effects of the reform. Specifically, we measure reform effects on mortality using data from the Swedish Death Index. Preliminary results from the aggregated analysis provide nearly no evidence on intervention effects on various short-run outcomes. In contrast, the estimates for individual mortality suggest significant negative effects that are sizeable and increasing in age.

Key findings

Several spin-off projects are still in early phases, but the original project has delivered three distinct findings that represent significant contributions to the literature:

1. The neonatal care intervention caused a significant reduction in mortality during two phases in the life cycle: early childhood and from age 50 onward. An important aspect of the first finding is that the effects were persistent: treated children who survived thanks to the intervention did not exhibit higher mortality at higher ages. Moreover, we see this effect reinforced at older ages; for males as well as for females. Using data on death causes, we are able to attribute this reduction in mortality to specific causes. It turns out that males enjoy reduced mortality due to cardiovascular disease – which is a common finding in the literature on the effects of early life health shocks. For females, it turns out that the reduction in adult mortality, which is roughly equal in size, is driven by a reduction in cancer mortality, in particular cancers in the lower gastrointestinal tract. This is arguably the most important finding regarding mortality, since we are able to establish a plausible mechanism running from early-life nutrition via reduced risk of gastrointestinal infection and inflammation, which in turn are important triggers of cancer.

2. We also find that the neonatal care intervention improved various measures of human capital throughout the life course. In particular treated women experienced a large and persistent increase in earnings, which is partly explained by a larger propensity to work full-time. Our central estimate is that earnings increased by 7 per cent, and that this is completely driven by females who increased their earnings by 19 per cent.

3. The increase in earnings reported above – 19 per cent for females – appeared too large to be plausible, and for this reason, we tried very carefully to assess whether it can be explained in the light of human capital investments occurring earlier in the life course. This analysis, which links effects at different stages of the life cycle, represents a third important contribution of our project. Despite similar improvements in primary school performance between boys and girls, only females enjoyed long-term benefits. We were able to show that the effects on female school performance were concentrated in parts of the distribution leading to secondary schooling, whereas male gains were dispersed over the entire distribution. Using a technique developed within the project, we were able to show that this increase in secondary schooling enrolment can explain at least half of the gain in female earnings.

Dissemination

Research output has been presented at numerous leading national and international conferences in the field of economics. These include presentations at the RES in Manchester (2014 and 2015), the ESPE conference in Izmir (two presentations in 2015), the SOLE/EALE conference in Montreal (2015), the Annual Conference of the International Association for Applied Econometrics (2015 in Thessaloniki), the EEA conference (Mannheim, 2015), the Nationell Konferens i Nationalekonomi in Umeå (2014), the Swedish Health Economic Association annual conference in 2014, a workshop in Augsburg (2014) and several seminars at economics departments (e.g. Linneaus University and the Universities of Navarra, Hamburg, Stockholm, Stirling, Prag, Örebro and Lund), workshops (e.g. SDU Workshop on Applied Microeconomics, Odense 2015) and conferences. Another conference presentation at the Annual Congress of the EEA in Mannheim is forthcoming. Furthermore, the project team devoted its 2014-17 editions of the Essen Health Conference to topics which have been covered by this project. Therefore research on the ante- and neonatal care programmes has also been presented to a broad specialised audience at these tailor-made conferences. Concerning dissemination to practitioners, there have been various activities in several countries. Most importantly, the SNS published a policy paper and organised a one-day workshop based on this project in April 2018.

Future work and spin-off effects

The project and its unique and purpose built datasets have generated several spin-off effects and further research ideas.

One project deals with the intergenerational transmission of health. The existence of an intergenerational persistence of health and human capital is well-documented (see e.g. Royer, 2009 and Mazumder, 2005), however the underlying transmission mechanisms are not completely understood. We therefore empirically investigate the intergenerational transmission of health across two Swedish generations using a rich individual-level historical dataset that was tailor-made for the purpose of this study. Our database covers two generations with a total time span of about 130 years (the first generation was born between 1880 and 1918 and the second generation was born between 1930 and 1934). We study whether fetal programming in the first generation transmits to the second generation and we further explore to what extent health insults suffered by the first generation contribute to the second generation’s health and human capital formation, as a source of persisting social inequality.

Another project analyses the short- and long-term effects of sickness absence in school on education, income and health. While most research investigates the effects of additional education and often finds small or no consequences, absence might be a more promising case since it affects students individually (Goodman, 2014, and Aucejo and Romano, 2014). Combining the generated schooling data with tax and health registers, we estimate the effects of sickness absence for a representative sample of Swedish students in the 1930s. By considering income and health outcomes, this is the first study that investigates long-term effects.



Website

https://www.goek.wiwi.uni-due.de/forschung/forschungsprojekte/