Research Statement Anthony Wray∗ November 7, 2016 Click here for most recent version.

Introduction The frequency of health shocks, the capacity for remediation, and dynamic complementaries are important factors that vary considerably between rich and poor countries today, and potentially determine the strength of the relationship between early life health and adult well-being (Currie and Vogl, 2013). To better understand how different health environments, institutional settings, and levels of economic development influence fetal health and the pathways linking it to adult outcomes, researchers can turn to the health transition that developed countries underwent in the late-nineteenth and early-twentieth centuries (Costa, 2015). Cutler et al. (2006) argue that, first, public health infrastructure and, later, medical innovations made large contributions to the mortality decline during this period. However, it is less clear how these factors affected the health of individuals and whether their influences persisted over the life cycle and between generations. My research interests contribute to literatures in applied microeconomics and economic history that highlight the importance of human capital development for long-run outcomes. The first branch of my research examines the long-term influence of in utero and early childhood health on labor market outcomes. The second component extends the inquiry to consider the consequences of the childhood health environment for inequality and intergenerational transmission. The third segment evaluates the short- and long-term effects of public health policies that aim to improve the quality of the water supply and reduce disparities in access to health care. I address these issues using a combination of modern microeconometric techniques and individual-level micro data from England and the United States that span the late-nineteenth and early-to-mid-twentieth centuries. A significant feature of my research is the use of a wide range of historical sources to construct large-scale data sets for each of my projects. In particular, I draw on transcriptions of hand-collected archival material, complete-count census records, publicly available administrative data, and spatial data. For example, where existing complete-count census records lacked the necessary information, I collected over two million records from the World War I Draft Registration Cards, turning online indexes into machine-readable form. By incorporating elements from existing methods, I have also developed procedures for linking records across censuses and other administrative databases. The process of developing new historical data sets has involved a considerable investment of time and resources in traveling to archives to digitize historical documents, such as individual-level inpatient hospital admission register and monthly reports on water quality in London, and managing the transcription of the records. In the sections below, I discuss a number of completed and ongoing research projects that have used these data sets, in addition to future projects that will make further use of the data. ∗ Postdoctoral

Researcher, Hitotsubashi University; Email: [email protected].

1

Fetal and early life health

A substantial body of evidence has shown that health and investment during the in utero and earlychildhood periods have persistent effects on human capital accumulation and adult socioeconomic outcomes (Almond and Currie, 2011). These effects can work through a number of channels, such as prenatal stress, pollution, disease, and nutritional deficiencies. Biology studies have hypothesized that maternal stress during pregnancy can have harmful consequences for the neurodevelopment of a child (Glover et al., 2010). However, it is not clear whether exposure to prolonged prenatal stress from the mild shock of commonplace events has consequences for the longer-term outcomes of children. In work with Krzysztof Karbownik of Northwestern University (Karbownik and Wray, 2016), we study the long-run effects of exposure to prenatal stress from low intensity (Category 1 and 2) hurricanes in the U.S. South during the late-nineteenth century on human capital development and labor market outcomes. To address this question, I collected data from the individual-level World War I Draft Registration Cards and constructed a data set linking these records to spatial data on hurricane tracks and complete-count census data. Whereas historical census records almost always report only the state of birth and the age in years on the day of census enumeration, we make use of the exact date of birth and place of birth at the city or town level reported in the war records, which are necessary to reduce measurement error and bias in the identification of fetal exposure to the storms. Our quasi-experimental design exploits the random nature of hurricanes. The treatment will operate primarily through prolonged stress induced by uncertainty, loss of normalcy, or temporary displacement from unanticipated and potentially life-threatening natural phenomena, rather than major destruction to property and infrastructure, or long-lasting disruption to economic activity. We use a difference-in-differences identification strategy in which we compare individuals exposed to a hurricane in utero or during infancy to those born before or after a storm, as well as to those born in surrounding locations. The long-run consequences of hurricanes are relevant to policymakers, given that a large fraction of children around the world now face the risk of prenatal exposure to these recurring events. In the U.S. context, 9 to 14 percent of births from 1992 to 2010 occurred in coastal shoreline or watershed counties along the southern Atlantic coast, which is vulnerable to hurricane-force winds. Our micro-based evidence can also inform discussions on the macroeconomic consequences of environmental disasters, such as the negative effects of exposure to tropical cyclones on a country’s long-run growth documented by Hsiang and Jina (2014). In a sample of white males, we estimate that in utero exposure to hurricanes reduces income at ages 42 to 53 by 7.5 percent, while exposure at ages 0 to 6 months is associated with a 5.1 percent decline in income. When we examine hurricane exposure by trimester of pregnancy, we find an effect of first trimester exposure on educational attainment and income, while third trimester exposure affects income and full time employment. This pattern of results is consistent with the fact that the brain develops rapidly during first trimester, while the fetus grows the most during the third trimester. A natural question that arises from the results of our study is whether the long-run consequences of prenatal stress affect individuals over the entire life cycle and contribute to intergenerational transmission. Administrative data currently available to researchers lack the combination of educational attainment, income during adulthood, and precise enough details of birth needed to address these questions in a contemporary developed country setting, but this will soon change. In a follow up study, we will utilize the one-in-six restricted-use sample of the 2010 U.S. Census, which has been linked to unique Personal Identification Keys (PIK) that correspond to Social Security Numbers, and via SSNs, to other administrative data sets (Alexander et al., 2015). This resource provides us with a nationally representative sample that contains 2

appropriate data on educational and labor market outcomes, in addition to the details of birth and death. We can then examine the role of prenatal stress at different points in the life cycle, including effects on longevity, and by incorporating information on parents and children, examine whether the effects persist across generations. The Social Security records can also be linked to Medicare data to study effects on morbidity and health care expenditures. Alcohol consumption during pregnancy is another detrimental shock to fetal health that can affect the labor market and educational outcomes of prenatally exposed children (Nilsson, 2016). Together with David Jacks and Hitoshi Shigeoka of Simon Fraser University, I am studying the long-run effects of in utero exposure to prohibition in the U.S. during the late-nineteenth century. We make use of the data set linking the World War I Draft Registration Cards to the complete-count file of the 1940 U.S. Census, which I have expanded from the nine Southern U.S. states examined in Karbownik and Wray (2016) to cover the entire country. The latter decades of the nineteenth century were characterized by significant county-level variation in prohibition status, which enables us to compare the relative impact of policy changes that ban the consumption of alcohol to those that allow it, in contrast to previous studies that have examined the impact of a single policy change.

2

Childhood health While in utero and early-childhood health have persistent effects on human capital development, the

empirical evidence on the consequences of health in late-childhood and early adolescence is mixed. Historical quasi-experimental studies support the notion that health matters for long-run outcomes (Bleakley, 2007), but recent research utilizing detailed public health insurance data from Canada (Currie et al., 2010) suggests a much weaker and heterogeneous link between health and later-life outcomes. Joint work with Karbownik (Karbownik and Wray, 2015) studies the consequences of childhood illness in late-nineteenth century London, which is an environment with much larger variation in health deficiencies than in modern developed countries. Our primary objective is to identify whether poor childhood health has long-lasting effects on individual occupational success, intergenerational mobility, and marital status. To address these issues, we construct a unique longitudinal data set consisting of inpatient hospital admissions, with detailed information on the cause of admission, linked to complete-count records from the Census of England. We compute the in-hospital mortality rate for each admitted condition, which provides a proxy for the totality of health deficiencies from the in utero period through childhood. When the first children’s hospital was established in London in 1852, the mandate of the charity-run voluntary hospitals was to provide palliative care for the sick working-class poor. While the wealthy had no interest in entering an institution, middle class patients were under-served by hospital care. The growing knowledge of the infection process made it clear that hospitals ought to provide care to all classes of society (Abel-Smith, 1964). By the late-nineteenth century, voluntary hospitals had developed into institutions where doctors treated middle- and upper-class patients, trained medical students, and gained authority as sources of new medical knowledge. Meanwhile, the privileged upper-classes of England continued to rely on general practitioners who operated private clinics and treated patients in the privacy of their homes, even though medical education focused on hospital care and doctors could not make use of new medical technologies such as x-ray machines in the context of home care (Carpenter, 2010). We use a sibling fixed effects identification strategy in which we compare patients hospitalized between the ages of 0 and 12 to their siblings of the same gender, who lived in the same household during childhood

3

and did not appear in the hospital registers. Health deficiencies during childhood decrease the occupational log wage, and increase the probability of downward mobility relative to one’s father, for male patients in comparison to their brothers. Using the occupation of one’s father, we compute an intergenerational occupational elasticity and find that the effect of health deficiencies offsets 30 percent of the elasticity of a son’s occupational log wage with respect to his father’s occupational log wage. We are one of the first papers in the literature using historical data to provide estimates for both males and females. Females have often been ignored in previous research because surname changes at marriage prevent the linkage of census data from childhood to adulthood. We use marital status as a measure of economic opportunity for women, because marriage insured women against poverty by providing access to a share of income from multiple wage earners in a household. We overcome the obstacle of name changes at marriage by linking census records to marriage certificates, and provide estimates for the effects of poor childhood health on the probability of being married as a young adult. We find that female patients were less likely to be married as adults in comparison to their sisters, but there is no evidence of a marriage penalty for males. Newly released de-anonymized complete-count census data for England has made it possible to address a number of additional questions that emerged from our study of childhood hospitalization in Victorian London. Do the siblings of hospital patients have better than average outcomes due to resource allocations that reinforce differences in health status? Did parents identify the “best” child and invest all resources in their “champion”? I have constructed samples that link census records from childhood to occupational outcomes in adulthood and use them to make comparisons between patients and neighboring or singleton children with similar observable characteristics, as well as comparisons between siblings of patients and children in households with no patients. We have also digitized additional hospital records and expanded the sample of both male and female patients, which allow us to consider a wider range of outcomes for women, including the quality of the marital match and the occupational status of the spouse. I am also interested in understanding the determinants of childhood health, and whether it can be influenced by policy changes. In the context of Victorian England, and in many developing country settings, the occupational hazards of child labor can have persistent effects on late-childhood and adolescent health. Although studies have examined the link between child labor laws and the decline in child labor (Moehling, 1999), the literature has remained silent on whether these factors affect childhood health. Work in progress with Prashant Bharadwaj at the University of California, San Diego, and Karbownik, exploits educational reforms that extended compulsory schooling to older ages, and asks whether delayed entry into the labor market during childhood affected the likelihood of hospitalization. We hypothesize that compulsory schooling would have reduced participation in child labor, and we examine the effects on the probability of being admitted to a hospital for an accident or injury, the severity of the admitted conditions, and the length of stay in the hospital, using the inpatient hospital admission records analyzed in Karbownik and Wray (2015). We will be able to quantify the potential tradeoff for long-run outcomes that arises from the choice between the benefits of earlier entrance into the labor market via an apprenticeship, on one hand, and the potential health consequences of starting work during the period when children undergo intense physical development, on the other.

4

3

Public health policies

My research discussed thus far has focused on exploiting exogenous shocks to fetal health and withinhousehold variation in childhood health to study the long-term socioeconomic consequences of poor earlylife health. Another branch of my research examines the effects of public health interventions over which policymakers have direct control. Recent studies have found a positive impact on health and long-run outcomes from public policies that eradicated infectious diseases (Bleakley, 2007), expanded access to medical care (Bailey and Goodman-Bacon, 2015), and increased a family’s economic resources (Hoynes et al., 2016), but many questions remain. In particular we know little about the consequences of inequalities in health access for the racial gap in health and socioeconomic outcomes. An ongoing project with Karbownik studies policies aimed at reducing racial disparities in access to health care in the context of the segregated U.S. South. During the early decades of the twentieth century, the rural black population faced chronic poverty and extreme shortages of physicians and hospital beds. These conditions are similar to, and may shed light on the experiences of the rural poor in developing countries today. More specifically, our project examines the effects of improved access to medical care on infant mortality rates. We construct measures of health care access, such as the number of physicians or hospital beds per capita, and the opening of new hospitals, by county, year, and for some variables, by race as well. Our identification strategy exploits the variation in hospital and physician access across counties at baseline, and the staggered rollout of new hospitals. This project makes three contributions to the literature. First, by obtaining data on the number of hospitals and physicians per capita, we can estimate the first stage relationship between medical care access and early-life health outcomes. Such a relationship has been hypothesized, but not directly estimated, by Chay et al. (2009), who use changes in post-neonatal mortality rates among cohorts born in the 1960s and early 1970s as a proxy for improved hospital access following the integration of hospitals. Second, in contrast to ongoing work using Scandinavian registry data (Hjort et al., 2014; Bhalotra et al., 2015; B¨ utikofer et al., 2015), we can estimate differences in health care access by race and examine whether they led to racial convergence in health and socioeconomic outcomes. Third, we can study the consequences of health care access over the life course, from infant mortality, to socioeconomic status in adulthood and longevity. A related project explores the long-run benefits of access to medical care in the context of England during the late-nineteenth century. In this setting, I focus on the role of physician education and improvements in hospital medical technology in contributing to adult well-being. First, I inquire whether children living in neighborhoods with access to better hospitals and better quality physicians attain higher occupational status as adults in comparison to children living in neighborhoods with lower quality health care or without formal access to care. Second, I ask whether hospital access and physician quality can mediate the negative effects of childhood health shocks, particularly the outbreaks of local epidemics. The analysis sheds light on whether the benefits of childhood access to public health infrastructure and medical care in a pre-modern disease environment extend into adulthood and old age, and potentially across generations. While this project draws on evidence from historical data, its findings are relevant to the developing world, where we observe a rapid expansion in access to medical care today. The provision of safe drinking water is another important means by which public health policies can improve health outcomes of the population, but is one that remains a challenge in the developing world. In 2012, nearly 750 million people lacked access to clean drinking water and over 500,000 diarrhea deaths were attributed to contaminated water. Waterborne diseases such as cholera, dysentery, and typhoid fever were also prevalent in the developed world in the nineteenth century, but the incidence of these diseases declined 5

with the introduction of clean sources of water, as well as the filtration and chlorination of the water supply. Wray (2015) examines the effects of improvements in water quality on typhoid fever morbidity and mortality rates in London during the early decades of the twentieth century. The introduction of chlorination to the public water supply in London in 1916 was arguably unrelated to health concerns, but rather a part of an effort to reduce coal consumption during wartime, as the existing filtration technology for water purification had been dependent on coal to pump water into reservoirs prior to filtration. Technological constraints in the form of storage capacity meant that chlorination could only be implemented in boroughs in northwest London that obtained water from the Thames River. Thus, variation in the sources of water supply and the introduction of chlorination create a setting in which a difference-in-differences identification strategy can be used to examine the effects of improved water quality on population health. I draw on quarterly data to help explain the considerable seasonal variation in water quality and waterborne disease rates, which is completely obscured in previous research that relies on city-by-year panel data (Alsan and Goldin, 2015). I also collect evidence on direct measures of water quality instead of inferring quality improvements from declining typhoid mortality rates or discrete changes in water supply or sanitation infrastructure across cities, as has been the norm in the literature. I find that chlorination accounts for 16 percent of the decline in the typhoid mortality rate during the sample period, with larger effects in the fourth quarter when contaminated river water entered the water supply due to flooding. I also show that higher rates of bacteria present in the water supply were associated with higher rates of typhoid fever mortality and morbidity. In a follow up project, I plan to examine the long-run implications of improved water supply infrastructure on labor market outcomes, mobility and fertility.

4

Summary Much of my early work has involved developing methodologies for linking census records and administra-

tive data to form large historical data sets. I anticipate a positive return on the fixed investment as I make use of the data for multiple projects. Over the past couple of years, an enormous amount of complete-count census records and registration data have become available to researchers. In coming years, the potential for new research using these data will grow considerably as historical and contemporary census records are linked to large administrative data sets. I intend to continue to take advantage of these new opportunities and develop new lines of inquiry to expand my existing research agenda. My research has focused on exploiting exogenous shocks to fetal health and within-household variation in child health, and studying their long-term socioeconomic consequences. Going forward, I plan to provide direct evidence on pathways from fetal programming to adult well-being, and place greater emphasis on the positive impact of public health policies in child development. Part of this focus involves my work in progress on the role of health care access in contributing to racial disparities in health and socioeconomic outcomes. I also plan to take advantage of the longer time span covered by historical data sources to explore determinants of intergenerational transmission and the persistence of inequality. The study of slower-moving processes such as inequality require data with longer time panels, which are typically lacking in modern administrative data. The next stage of my research will address these issues using rigorous empirical methods and newly available data sources.

6

References Abel-Smith, Brian, The Hospitals, 1800-1948: A Study in Social Administration in England and Wales, Cambridge, MA: Harvard University Press, 1964. Alexander, J. Trent, Todd Gardner, Catherine G. Massey, and Amy O’Hara, “Creating a Longitudinal Data Infrastructure at the Census Bureau,” Working Paper, 2015. Almond, Douglas and Janet Currie, “Human Capital Development Before Age Five,” in “Handbook of Labor Economics,” Vol. 4, Elsevier, 2011, pp. 1315–1486. Alsan, Marcella and Claudia Goldin, “Watersheds in Infant Mortality: The Role of Effective Water and Sewerage Infrastructure, 1880 to 1915,” National Bureau of Economic Research Working Paper Series, 2015, No. 21263. Bailey, Martha J. and Andrew Goodman-Bacon, “The War on Poverty’s Experiment in Public Medicine: Community Health Centers and the Mortality of Older Americans,” American Economic Review, 2015, 105 (3), 1067–1104. Bhalotra, Sonia, Martin Karlsson, and Therese Nilsson, “Infant Health and Longevity: Evidence from a Historical Trial in Sweden,” IZA DP No. 8969, 2015. Bleakley, Hoyt, “Disease and Development: Evidence from Hookworm Eradication in the American South,” The Quarterly Journal of Economics, 2007, 122 (1), 73–117. B¨ utikofer, Aline, Katrine Vellesen Løken, and Kjell G. Salvanes, “Long-Term Consequences of Access to Well-Child Visits,” Working Paper, 2015. Carpenter, Mary Wilson, Health, medicine, and society in Victorian England, Santa Barbara, Calif.: Praeger, 2010. Chay, Kenneth Y., Jonathan Guryan, and Bhashkar Mazumder, “Birth Cohort and the BlackWhite Achievement Gap: The Roles of Access and Health Soon After Birth,” National Bureau of Economic Research Working Paper Series, 2009, No. 15078. Costa, Dora L., “Health and the Economy in the United States from 1750 to the Present,” Journal of Economic Literature, 2015, 53 (3), 503–570. Currie, Janet and Tom Vogl, “Early-Life Health and Adult Circumstance in Developing Countries,” Annual Review of Economics, 2013, 5 (1), 1–36. , Mark Stabile, Phongsack Manivong, and Leslie L. Roos, “Child Health and Young Adult Outcomes,” Journal of Human Resources, 2010, 45 (3), 517–548. Cutler, David, Angus Deaton, and Adriana Lleras-Muney, “The Determinants of Mortality,” Journal of Economic Perspectives, 2006, 20 (3), 97–120. Glover, Vivette, T. G. O’Connor, and Kieran O’Donnell, “Prenatal stress and the programming of the HPA axis,” Neuroscience & Biobehavioral Reviews, 2010, 35 (1), 17–22. Hjort, Jonas, Mikkel Solvsten, and Miriam W¨ ust, “Long-run effects of the Danish home visiting program on adult health,” Working Paper, 2014. Hoynes, Hilary, Diane Whitmore Schanzenbach, and Douglas Almond, “Long-Run Impacts of

7

Childhood Access to the Safety Net,” American Economic Review, 2016, 106 (4), 903–34. Hsiang, Solomon M. and Amir S. Jina, “The Causal Effect of Environmental Catastrophe on LongRun Economic Growth: Evidence From 6,700 Cyclones,” National Bureau of Economic Research Working Paper Series, 2014, No. 20352. Karbownik, Krzysztof and Anthony Wray, “Childhood Health and Long-Run Economic Opportunity in Victorian England,” Working Paper, 2015. and

, “Long-Run Consequences of Natural Disasters,” Working Paper, 2016.

Moehling, Carolyn M, “State child labor laws and the decline of child labor,” Explorations in Economic History, 1999, 36 (1), 72–106. Nilsson, J Peter, “Alcohol Availability, Prenatal Conditions, and Long-term Economic Outcomes,” Journal of Political Economy, 2016, forthcoming. Wray, Anthony, “Water Quality, Morbidity, and Mortality in London, 1906-1926,” Working Paper, 2015.

8