Cohort Studies

April 14-15, 2017
Dora Costa of the University of California, Los Angeles, Organizer

Kent Thornburg, Oregon Health & Science University

Early Life Origins of Disease

Richard Steckel, Ohio State University and NBER

The Health of Past Populations: Skeletal Evidence

Lewina Lee and Avron Spiro, Boston University

Early Psychosocial Experiences and Trajectories of Cardiometabolic Risk in Later Life: Findings from the VA Normative Aging Study

Evan Roberts, University of Minnesota; Kris Inwood, University of Guelph; and Les Oxley, University of Waikato

Such a Rash Act: Wartime Experiences and Suicides After the Great War

World War I exposed sixty million soldiers worldwide to armed conflict and significant psychological trauma. Despite popular assumptions of significant long-term effects of the conflict on veterans, suicide among returned soldiers has received little attention. To address this lacuna Roberts, Inwood, and Oxley carried out the first retrospective cohort study of suicide risk in World War I veterans, following a cohort of New Zealand soldiers until death. Causes of death were ascertained by doctors and coroners. 32 of 1865 men (1.7%) committed suicide over 83,118 years of follow-up (Crude rate: 39/100,000 person years). Suicides were more likely to occur at early ages, accounting for 8% of deaths before 1939, and 11% of potential years of life lost before age 50. Suicide rates in this cohort were 1.5-2 times higher than among veterans of recent conflicts, and remained high throughout their lifetime, showing the significant long-term cost of the conflict on human health. Explanations of suicide by family and acquaintances revealed three key narrative themes: suicide as a "rash act" that family members did not accept; suicide as the culmination of a life course whose trajectory was altered by war; and finally suicide explained without reference to war at all. The majority of informants in the first 20 years after the war identified wartime experience as a contributing factor to suicide.

Daniel Belsky, Avshalom Caspi, Terrie Moffitt, Jasmin Wertz, Duke University, and Richie Poulton, University of Otago

Do Polygenic Influences on Educational Attainment Predict Crime? Findings from two Birth Cohorts

Daniel Barth, University of Southern California; Nicholas Papageorge, Johns Hopkins University; and Kevin Thom, New York University

Genetic Ability, Wealth and Financial Decision-Making

Recent advances in behavioral genetics have enabled the discovery of genetic scores linked to a variety of economic outcomes, including education. Barth, Papageorge, and Thom build on this progress to demonstrate that the same genetic variants that predict educational attainment independently predict household wealth in the Health and Retirement Study (HRS). This relationship is partly explained by higher earnings, but a substantial portion of this association cannot be explained mechanically by income flows or bequests. This leads the researchers to explore the role of beliefs, financial literacy, and portfolio decisions in explaining this genetic gradient in wealth. They show that individuals with lower genetic scores are more prone to reporting "extreme beliefs" (e.g., reporting that there is a 100% chance of a stock market decline in the near future) and they invest their savings accordingly (e.g., avoiding the stock market). These findings suggest that genetic factors that promote human capital accumulation contribute to wealth disparities not only through education and higher earnings, but also through their impact on the ability to process information and make good financial decisions. The association between genetic ability and wealth is substantially lower among households receiving a defined benefit pension. Policies that transfer greater responsibility to individuals to manage their wealth might therefore exacerbate the consequences of labor market inequality.

Weili Ding, Queen's University, and Steven Lehrer, Queen's University and NBER

Are Genetic Markers of Interest for Economic Research?

The idea that genetic differences may explain a multitude of individual-level behaviors and outcomes as studied by economists, is more than a bit controversial. Since an increasing number of datasets now contain measures of genetic variation, it is reasonable to postulate that incorporating genomic data into economic analyses will become increasingly common. However, there remains much debate among academics as to: First, whether and how ignoring genetic differences in empirical analyses would bias the resulting estimates; Second, since genetic characteristics are largely immutable, what types of policy guidance, if any, the incorporation of these variables into economic analyses may yield. In this paper, Ding and Lehrer revisit these concerns and survey the main avenues by which empirically oriented economic researchers have utilized measures of genetic markers to improve our understanding of economic phenomena. The researchers discuss the strengths, limitations, and potential of existing approaches and conclude by highlighting several prominent directions forward for future research.

Maya Rossin-Slater, University of California at Santa Barbara and NBER, and Miriam Wüst, The Danish National Centre for Social Research

What is the Added Value of Preschool? Long-Term Impacts and Interactions with a Health Intervention (NBER Working Paper No. 22700)

Rossin-Slater and Wüst study the impact of targeted high quality preschool over the life cycle and across generations, and examine its interaction with an infant health intervention. Using administrative data from Denmark together with variation in the timing of program implementation between 1933 and 1960, the researchers find lasting benefits of access to high quality preschool at age 3 on adult educational attainment and survival beyond age 65. Further, the benefits persist to the next generation, who obtain higher education by age 25. However, exposure to a nurse home visiting program in infancy reduces the added value of preschool. The positive effect of preschool is lowered by 88 percent for years of schooling (of the first generation) and by 80 percent for survival past age 65.

Mary McEniry, University of Wisconsin; Carmen Elisa Florez, Universidad del Rosario; Renata Pardo, Health Consultant, Bogotá, Colombia; Rafael Samper-Ternent, University of Texas Medical Branch; and Carlos Cano-Gutierrez, Universidad Javeriana

Examining the Multigenerational Effects of Obesity in a Latin American Middle Income Country: The Case of Colombia

The examination of multigenerational effects on health in Latin American populations from grandparents to subsequent generations is a fruitful area for research which has not yet been fully explored. Within this context, understanding the multigenerational effects on obesity — an important health concern both within the U.S. and internationally — merits further consideration. Using comprehensive data collected on households and individuals throughout Colombia in 2010 (ENDS and ENSIN) McEniry, Florez, Pardo, Samper-Ternent, and Cano-Gutierrez examined the degree to which prevalence of obesity in older generations predicts obesity in subsequent generations. They also examined grandparent and mother height in regards to child stunting. They find strong associations between (1) grandmother-mother and mother-female child in terms of obesity; (2) grandfather and female child obesity; (3) grandparent height and child stunting which is mediated by mother height. While it may be too early to fully observe multigenerational effects of obesity, these results suggest the importance of early life conditions. The researchers' examination of multigenerational effects provides a relevant starting point in a field where there are few studies from the LAC region regarding multigenerational effects and health.

Mayvis Rebeira, University of Toronto

The Effect of Pension Income on Mortality - Evidence from Civil War Confederate Veterans

Differences in pension income legislations in the Southern States post-Civil War offer a unique and natural experiment to study the effect of differences in pension income on mortality rates. In this study, Rebeira examined Confederate pensions in two adjacent states, Texas and Oklahoma, which enacted pension laws for veterans in 1899 and 1915 respectively. Since Confederate pensions represent a significant source of permanent, steady income for the veterans, the study was able to determine the role of newly-endowed wealth on longevity with pension laws providing for the exogenous variation in income. Data was collected through primary means mainly from archival records from multiple sources to create a unique database of births and deaths for both states. The results reveal veterans in Texas gained 1.52 years (or 18.2 months) of additional years of life, as compared to veterans in Oklahoma. In addition, for every $10 increase in pension income, the number of years lived or longevity increased by 1.44% when controlled for year of birth. It increased to 1.94% when controlled for all county-level differences including year of birth. The difference in pension income resulted in a decrease of mortality hazard by 12.9% when controlled for year of birth only and a mortality hazard reduction of 17.5% when controlled for year of birth and all other county-level covariates. The effect of an increase in pension income on longevity is substantial and significant which reveal that circa 1900, income effects were large.

Irma Elo, University of Pennsylvania; Arun Hendi, Duke University; and Pekka Martikainen, University of Helsinki

Birth Cohorts, Synthetic Cohorts, and SES Differentials in Longevity: The Implications of Changing Educational Distributions for Life Expectancy Gradients

One of the primary goals of social demographers is to study educational differentials in life expectancy. Findings from recent research have proposed that shifting education distributions across cohorts are influencing estimates of educational-gradients in mortality. In this study, Hendi, Elo, and Martikainen use high-quality Finnish register data for 1971-2010 to identify an empirical regularity in the relationship between percentile rank in the education distribution and mortality at ages 30 and older. The researchers specify and estimate a two-parameter model that allows them to adjust for changes in the education distribution over time and use the model parameters to generate educational quintile-specific life expectancies for eight five-year periods. These distribution-adjusted absolute education differentials in life expectancy have increased for men, but not for women between 1971 and 2010. Although relative education inequality in mortality has increased for both sexes, this effect is not observed for absolute life expectancy differentials for women, partly due to more rapid declines in overall mortality among women than men. Roughly 39% and 100% of the increase in absolute educational differentials in life expectancy for men and women, respectively, can be attributed to shifting education distributions. Had the relative inequality parameters of mortality differentials governing the Finnish population remained constant at the 1971-75 values, the absolute difference in life expectancy between the top and bottom education quintiles would have narrowed by 1.0 years for men and 0.4 years for women. A comparison of similar estimates for the United States suggests that educational expansion in the U.S. has also influenced education-mortality gradients. Estimates of the growth in the absolute life expectancy gradients are 16% and 73% larger using education categories versus distribution-adjusted education quintiles for non-Hispanic white men and women, respectively.

Adriana Lleras-Muney, University of California at Los Angeles and NBER, and Flavien Moreau, University of California at Los Angeles

A Unified Law of Mortality: Implications for Economic Analysis

How do social and economic conditions affect the evolution of health and mortality rates over the lifetime? To answer this question, Lleras-Muney and Moreau build a simple model that combines an age-dependent process with a random health hazard. A key insight of the model is that population mortality rates place constraints on the evolution of the underlying distribution of (unobserved) health, and can be used to infer how health has evolved over time and across countries. Using cohort life tables provided in the Human Mortality Database, the researchers estimate their model and trace out the evolution of structural parameters since 1850. They use the model to understand how unexpected shocks, like wars, affect the age-profile of health and mortality; and investigate implications for SES gradients and optimal health care expenditures.

Audrey Lai and Andrew Noymer, University of California at Irvine, and Tsuio Tai, National Taipei University

The Geometry of Mortality Change: Convex Hulls for Demographic Analysis

Lai, Noymer, and Tai introduce convex hulls as a data visualization and analytic tool for demography. Convex hulls are widely used in computer science, and have been applied in fields such as ecology, but are heretofore underutilized in population studies. The researchers briefly discuss convex hulls, then show how they may be applied profitably to demography. They do this through three examples, drawn from the relationship between child mortality and adult survivorship (5q0 and 45 p15 in life table notation). The three examples are: (i) using convex hulls for outlier identification; (ii) for studying sex differences in mortality; and (iii) for studying period and cohort differences. They find, respectively, that convex hulls can be useful in robust compilation of demographic databases, and that the gap/lag framework for sex differences or period/cohort differences is more complex when mortality data are arrayed by two components as opposed to a unidimensional measure such as life expectancy. The potential applicability of these methods goes beyond mortality.

Vellore Arthi, University of Essex; Brian Beach, College of William & Mary; and Walker Hanlon, University of California at Los Angeles and NBER

Estimating the Recession-Mortality Relationship When Migration Matters

A large literature following Ruhm (2000) suggests that mortality falls during recessions and rises during booms. This relationship, however, tends to be analyzed within a panel-data framework that implicitly assumes either that local economic shocks do not induce migration, or that insofar as they do, these movements are accurately reflected in intercensal population estimates. In this paper, Arthi, Beach, and Hanlon argue that unobserved migratory responses have the potential to bias results. To study the extent of this bias, the researchers draw on two natural experiments: the recession in cotton textile-producing regions of Britain during the U.S. Civil War and the Appalachian coal boom that followed the OPEC oil embargo in the 1970s. In both settings, they find evidence of a substantial migration response. Furthermore, they show that estimates of the business cycle-mortality relationship obtained using the standard approach are highly sensitive to assumptions about both the accuracy of interpolated population values and the short-run relationship between population and mortality. The researchers also show that control regions may be indirectly affected by migration into or away from treatment regions, leading to unobserved treatment spillovers. Together, these findings suggest that, when left unaddressed, large migratory responses can meaningfully undermine inference. Once the researchers adjust for migration, they find no evidence that the coal boom substantially affected mortality, and they find that mortality increased during the cotton recession.

Gunnar Branden, Uppsala University; Mikael Lindahl, University of Gothenburg; and Bjorn Ockert, Institute for Evaluation of Labour Market and Education Policy

The Importance of Nature-Nurture Interactions in Skill Formation: Evidence from a Large Sample of Swedish Adoptees

In this study Branden, Lindahl, and Ockert estimate the importance of nature-nurture interactions for cognitive and noncognitive ability and educational attainment, using data on adopted children and their adoptive and biological parents complied from Swedish registers. There exist very few studies of the importance of gene-environment interactions for skill formation and evidence from these studies is inconclusive. In this study, the researchers use a large sample of adoptees, born during the 1930s, 1940s, 1950s and 1960s. Preliminary results suggest that interaction effects are nonpositive. This is an important result as it suggests that positive environmental shocks will not exacerbate genetic inequality due to inherited differences among children. There is also some evidence of negative interaction effects, especially for sons, for the early periods, possibly indicating that the role of environmental interventions as an equalizer has decreased during the establishment of the Swedish welfare state.

Valentina Duque, University of Michigan; Maria Rosales Rueda, University of California at Irvine; and Fabio Sanchez, University of Los Andes

Integrating Early-Life Shocks and Human Capital Investments on Children's Education: Evidence from Colombia

This study investigates how early-life conditions interact with subsequent human capital investments to influence future educational outcomes. To provide causal evidence, Duque, Rosales Rueda, and Sanchez exploit two sources of exogenous variation: i) variation in early-life environments resulting from a child's exposure to extreme rainfall and drought shocks in the first years of life as a natural experiment; and ii), variation in subsequent investments resulting from the availability of conditional cash transfers (CCT), which promote investments in children's health and education. Using Colombian administrative data, The researchers combine a natural experiment with a regression discontinuity design using the CCT assignment rule and find that, while CCTs have a larger positive effect on children's educational attainment and achievement relative to the negative effects of the weather shocks, there is little evidence on an interaction effect between CCTs and weather shocks. These findings have important policy implications as they provide evidence of the role of social policies in closing gaps generated by early-life trauma.

Achyuta Adhvaryu, University of Michigan and NBER, and Snaebjorn Gunnsteinsson, University of Maryland

Resilience to Early Life Shocks

Can health investments engender resiliance to early life shocks? Gunnsteinsson and Adhvaryu address this question by leveraging the confluence of two independent sources of variation: exposure to a large-scale natural disaster (a tornado) and randomized access to vitamin A supplementation at birth. Tornado exposure in utero and in infancy increased the frequency of fevers and decreased birth size and physical growth. But infants who received vitamin A supplementation, which boosts immune system functioning, were protected from these effects. Tornado impacts and protective effects were larger for boys. These results provide support for wide-scale supplementation policies in disaster-prone areas of low-income countries.

NBER Videos

National Bureau of Economic Research, 1050 Massachusetts Ave., Cambridge, MA 02138; 617-868-3900; email:

Contact Us