The Methuselah Effect: The Pernicious Impact of Unreported Deaths on Old-Age Mortality Estimates.

We examine inferences about old-age mortality that arise when researchers use survey data matched to death records. We show that even small rates of failure to match respondents can lead to substantial bias in the measurement of mortality rates at older ages. This type of measurement error is consequential for three strands in the demographic literature: (1) the deceleration in mortality rates at old ages; (2) the black-white mortality crossover; and (3) the relatively low rate of old-age mortality among Hispanics, often called the "Hispanic paradox." Using the National Longitudinal Survey of Older Men matched to death records in both the U.S. Vital Statistics system and the Social Security Death Index, we demonstrate that even small rates of missing mortality matching plausibly lead to an appearance of mortality deceleration when none exists and can generate a spurious black-white mortality crossover. We confirm these findings using data from the National Health Interview Survey matched to the U.S. Vital Statistics system, a data set known as the "gold standard" (Cowper et al. 2002) for estimating age-specific mortality. Moreover, with these data, we show that the Hispanic paradox is also plausibly explained by a similar undercount.

Combining forward and backward mortality estimation.

Demographers often form estimates by combining information from two data sources-a challenging problem when one or both data sources are incomplete. A classic example entails the construction of death probabilities, which requires death counts for the subpopulations under study and corresponding base population estimates. Approaches typically entail 'back projection', as in Wrigley and Schofield's seminal analysis of historical English data, or 'inverse' or 'forward projection' as used by Lee in his important reanalysis of that work, both published in the 1980s. Our paper shows how forward and backward approaches can be optimally combined, using a generalized method of moments (GMM) framework. We apply the method to the estimation of death probabilities for relatively small subpopulations within the United States (men born 1930-39 by state of birth by birth cohort by race), combining data from vital statistics records and census samples.

The Impact of the Great Migration on Mortality of African Americans: Evidence from the Deep South.

The Great Migration-the massive migration of African Americans out of the rural South to largely urban locations in the North, Midwest, and West-was a landmark event in U.S. HISTORY: Our paper shows that this migration increased mortality of African Americans born in the early twentieth century South. This inference comes from an analysis that uses proximity of birthplace to railroad lines as an instrument for migration.

The effect of early-life education on later-life mortality.

Many studies link cross-state variation in compulsory schooling laws to early-life educational attainment, thereby providing a plausible way to investigate the causal impact of education on various lifetime outcomes. We use this strategy to estimate the effect of education on older-age mortality of individuals born in the early twentieth century U.S. Our key innovation is to combine U.S. Census data and the complete Vital Statistics records to form precise mortality estimates by sex, birth cohort, and birth state. In turn we find that virtually all of the variation in these mortality rates is captured by cohort effects and state effects alone, making it impossible to reliably tease out any additional impact due to changing educational attainment induced by state-level changes in compulsory schooling.

Predictive Inference Using Latent Variables with Covariates.

Plausible values (PVs) are a standard multiple imputation tool for analysis of large education survey data, which measures latent proficiency variables. When latent proficiency is the dependent variable, we reconsider the standard institutionally generated PV methodology and find it applies with greater generality than shown previously. When latent proficiency is an independent variable, we show that the standard institutional PV methodology produces biased inference because the institutional conditioning model places restrictions on the form of the secondary analysts' model. We offer an alternative approach that avoids these biases based on the mixed effects structural equations model of Schofield (Modeling measurement error when using cognitive test scores in social science research. Doctoral dissertation. Department of Statistics and Heinz College of Public Policy. Pittsburgh, PA: Carnegie Mellon University, 2008).

Are Children "Normal"?

We examine Becker's (1960) contention that children are "normal." For the cross section of non-Hispanic white married couples in the U.S., we show that when we restrict comparisons to similarly-educated women living in similarly-expensive locations, completed fertility is positively correlated with the husband's income. The empirical evidence is consistent with children being "normal." In an effort to show causal effects, we analyze the localized impact on fertility of the mid-1970s increase in world energy prices - an exogenous shock that substantially increased men's incomes in the Appalachian coal-mining region. Empirical evidence for that population indicates that fertility increases in men's income.

THE ROLE OF LOCATION IN EVALUATING RACIAL WAGE DISPARITY.

A standard object of empirical analysis in labor economics is a modified Mincer wage function in which an individual's log wage is specified to be a function of education, experience, and an indicator variable identifying race. We analyze this approach in a context in which individuals live and work in different locations (and thus face different housing prices and wages). Our model provides a justification for the traditional approach, but with the important caveat that the regression should include location-specific fixed effects. Empirical analyses of men in U.S. labor markets demonstrate that failure to condition on location causes us to (i) overstate the decline in black-white wage disparity over the past 60 years, and (ii) understate racial and ethnic wage gaps that remain after taking into account measured cognitive skill differences that emerge when workers are young.

The use of cognitive ability measures as explanatory variables in regression analysis.

Cognitive ability measures are often taken as explanatory variables in regression analysis, e.g., as a factor affecting a market outcome such as an individual's wage, or a decision such as an individual's education acquisition. Cognitive ability is a latent construct; its true value is unobserved. Nonetheless, researchers often assume that a test score, constructed via standard psychometric practice from individuals' responses to test items, can be safely used in regression analysis. We examine problems that can arise, and suggest that an alternative approach, a "mixed effects structural equations" (MESE) model, may be more appropriate in many circumstances.

Gender Wage Disparities among the Highly Educated.

In the U.S. college-educated women earn approximately 30 percent less than their non-Hispanic white male counterparts. We conduct an empirical examination of this wage disparity for four groups of women-non-Hispanic white, black, Hispanic, and Asian-using the National Survey of College Graduates, a large data set that provides unusually detailed information on higher-level education. Nonparametric matching analysis indicates that among men and women who speak English at home, between 44 and 73 percent of the gender wage gaps are accounted for by such pre-market factors as highest degree and major. When we restrict attention further to women who have "high labor force attachment" (i.e., work experience that is similar to male comparables) we account for 54 to 99 percent of gender wage gaps. Our nonparametric approach differs from familiar regression-based decompositions, so for the sake of comparison we conduct parametric analyses as well. Inferences drawn from these latter decompositions can be quite misleading.

Optimal wages in the market for nurses: an analysis based on Heyes' model.

In a recent paper "The economics of vocation or 'why is a badly paid nurse a good nurse?"' Anthony Heyes posits a novel adverse selection mechanism whereby the average quality of nurses employed is declining in the wage, and evaluates the consequences for an optimizing monopsonistic National Health Service (NHS). I conduct a standard welfare analysis here, based on the proposed model. It turns out that, contrary to concerns raised by Heyes, the NHS will always choose a wage that is lower than the wage chosen by a welfare-maximizing social planner. Nurse are underpaid. I also extend the analysis to the case of a competitive labor market for nurses.