Impact of Early Conditions in Life : A Joint Study of Marriage and Mortality Sumedha Gupta 1 VU University Amsterdam, Tinbergen Institute Amsterdam Gerard J. van den Berg VU University Amsterdam, IZA, IFAU-Uppsala, Netspar, CEPR, IFS-London October 30, 2008 Abstract It is well-known that early life conditions as well as marital status a/ect health and mortality later in life. In this paper we analyze the interplay between these determinants of high-age mortality. First, we study the impact of economic conditions early in life on the in- dividual rate of getting married. Secondly, we examine the protective e/ect of marriage and, in particular, to what extent this protective e/ect depends on conditions early in life. The results shed light on the use of marriage as a compensatory device in case of adverse early-life conditions. We use business cycle conditions in early years of life as an exogenous indicator of early-life conditions. The endogeneity of marriage calls for a simultaneous analysis that allows for selectivity on unobservables. We use individual data records from Dutch registers of birth, marriage and death certicates, covering an exceptionally long observation window from 1815-2000. These are merged with historical data on macro-economic and health indica- tors. The semi-parametric empirical analysis applies the timing-of-events approach in which bivariate duration models with unobserved heterogeneity and causal e/ects are estimated. It turns out that conditions around birth as well as around the school ages are important for marital status and mortality. The results are strikingly di/erent across gender. Men on average enjoy a protective e/ect of marriage on mortality, and this e/ect increases with age. Women born in economic booms gain from marriage during childbearing ages, but women born in recessions su/er a substantial negative e/ect on life expectancy during these years. Keywords : death, longevity, recession, life expectancy, lifetimes, marriage, conjugal bereave- ment, timing of events, selectivity, health. JEL codes : I12, J14, E32, N33, N13, C41. 1 Corresponding author: Sumedha Gupta (e-mail: gupta@tinbergen.nl). We are grateful to Andrew Chesher, James Banks, Ronald Woltho/, Erik Plug, Maarten Lindeboom, and Manuel Arellano, for valuable comments on an earlier version of this paper. We also thank participants of PAA 2008, ESEM 2007, ESPE 2007, IZA Summer School 2007, 3rd meeting of Microdata RTN at CEMFII, Madrid, seminars at Tinbergen Institute and UCL and workshops of the Max Planck Institute for Demographic Research on Causal infer- ence and endogeneity in duration analysis and Early-life vs later-life e/ects on late-life mortality for their useful input. Thanks to Marta Lpez for help with the preparation of the data. The Historical Sample of the Netherlands Dataset release UZF.02 was kindly provided by the International Institute of Social History (IISG), Amsterdam. The authors acknowledge nancial support from the EU 6th Research Framework and Marie Curie Research Training Actions.