John Bound

Problems with Instrumental Variables Estimation when the Correlation between the Instruments and the Endogenous Explanatory Variable is Weak

Abstract We draw attention to two problems associated with the use of instrumental variables (IV), the importance of which for empirical work has not been fully appreciated. First, the use of instruments that explain little of the variation in the endogenous explanatory variables can lead to large inconsistencies in the IV estimates even if only a weak relationship exists between the instruments and the error in the structural equation. Second, in finite samples, IV estimates are biased in the same direction as ordinary least squares (OLS) estimates. The magnitude of the bias of IV estimates approaches that of OLS estimates as the R 2 between the instruments and the endogenous explanatory variable approaches 0. To illustrate these problems, we reexamine the results of a recent paper by Angrist and Krueger, who used large samples from the U.S. Census to estimate wage equations in which quarter of birth is used as an instrument for educational attainment. We find evidence that, despite huge sample sizes, th...

Changes in the Demand for Skilled Labor within U. S. Manufacturing: Evidence from the Annual Survey of Manufactures

This paper investigates the shift in demand away from unskilled and toward skilled labor in U. S. manufacturing over the 1980s. Production labor-saving technological change is the chief explanation for this shift. That conclusion is based on three facts: (1) the shift is due mostly to increased use of skilled workers within the 450 industries in U. S. manufacturing rather than to a reallocation of employment between industries, as would be implied by a shift in product demand due to trade or to a defense buildup; (2) trade- and defense-demand are associated with only small employment reallocation effects; (3) increased use of nonproduction workers is strongly correlated with investment in computers and in R&D.

Self-Reported Versus Objective Measures of Health in Retirement Models

Labor supply models are sensitive to the measures of health used. When self-reported measures are used, health seems to play a larger role and economic factors a smaller one than when more objective measures are used. While this may indicate biases inherent in using self-reported measures, there are reasons to be suspicious of more objective measures as well. A statistical model incorporating both self-reported and objective measures of health shows the potential biases involved in using either measure or in using one to instrument the other. The model is initially unidentified, but incorporating outside information on the validity of self-reported measures confirms fears about both the self-reported and objective measures available on such data sets as the Retirement History Survey or the National Longitudinal Survey of Older Men.

The Extent of Measurement Error in Longitudinal Earnings Data: Do Two Wrongs Make a Right?

This article examines the properties and prevalence of measurement error in longitudinal earnings data. The analysis compares matched Current Population Survey data to administrative Social Security payroll tax records. In contrast to typically assumed properties of measurement error, the results indicate that errors are serially correlated over 2 years and negatively correlated with true earnings (i.e., mean reverting). In a cross section, the ratio of the variance of the signal to the total variance is .82 for men and .92 for women. These ratios fall to .65 and .81 when the data are specified in first differences. Longitudinal earnings data may be more reliable than previously believed.

Demand Shifts, Population Adjustments, and Labor Market Outcomes During the 1980s

In this article we explore the effects of labor demand shifts and population adjustments across metropolitan areas on the employment and earnings of various demographic groups during the 1980s. We find that population shifts across areas at least partially offset the effects of these demand shifts, but less‐educated workers showed substantially lower population adjustments in response to these demand shifts. These limited supply responses apparently contributed importantly to relatively greater deterioration of employment and earnings of these groups in declining areas during the 1980s.

Excess Mortality Among Blacks and Whites in the United States

BACKGROUND Although the general relations between race, socioeconomic status, and mortality in the United States are well known, specific patterns of excess mortality are not well understood. METHODS Using standard demographic techniques, we analyzed death certificates and census data and made sex-specific population-level estimates of the 1990 death rates for people 15 to 64 years of age. We studied mortality among blacks in selected areas of New York City, Detroit, Los Angeles, and Alabama (in one area of persistent poverty and one higher-income area each) and among whites in areas of New York City, metropolitan Detroit, Kentucky, and Alabama (one area of poverty and one higher-income area each). Sixteen areas were studied in all. RESULTS When they were compared with the nationwide age-standardized annual death rate for whites, the death rates for both sexes in each of the poverty areas were excessive, especially among blacks (standardized mortality ratios for men and women in Harlem, 4.11 and 3.38; in Watts, 2.92 and 2.60; in central Detroit, 2.79 and 2.58; and in the Black Belt area of Alabama, 1.81 and 1.89). Boys in Harlem who reached the age of 15 had a 37 percent chance of surviving to the age of 65; for girls, the likelihood was 65 percent. Of the higher-income black areas studied, Queens--Bronx had the income level most similar to that of whites and the lowest standardized mortality ratio (men, 1.18; women, 1.08). Of the areas where poor whites were studied, Detroit had the highest standardized mortality ratios (men, 2.01; women, 1.90). On the Lower East Side of Manhattan, in Appalachia, and in Northeast Alabama, the ratios for whites were below the national average for blacks (men, 1.90; women, 1.95). CONCLUSIONS Although differences in mortality rates before the age of 65 between advantaged and disadvantaged groups in the United States are sometimes vast, there are important differences among impoverished communities in patterns of excess mortality.

On the Validity of Using Census Geocode Characteristics to Proxy Individual Socioeconomic Characteristics

Investigators of social differentials in health outcomes commonly augment incomplete micro data by appending socioeconomic characteristics of residential areas (such as median income in a zip code) to proxy for individual characteristics. However, little empirical attention has been paid to how well this aggregate information serves as a proxy for the individual characteristics of interest. We build on recent work addressing the biases inherent in proxies and consider two health-related examples within a statistical framework that illuminate the nature and sources of biases. Data from the Panel Study of Income Dynamics and the National Maternal and Infant Health Survey are linked to census data. We assess the validity of using the aggregate census information as a proxy for individual information when estimating main effects, and when controlling for potential confounding between socioeconomic and sociodemographic factors in measures of general health status and infant mortality. We find a general, but not universal, tendency for aggregate proxies to exaggerate the effects of micro-level variables and to do more poorly than micro-level variables at controlling for confounding. The magnitude and direction of these biases, however, vary across samples. Our statistical framework and empirical findings suggest the difficulties in and limits to interpreting proxies derived from aggregate census data as if they were micro-level variables. The statistical framework we outline for our study of health outcomes should be generally applicable to other situations where researchers have merged aggregate data with micro data samples.

Inequality in life expectancy, functional status, and active life expectancy across selected black and white populations in the United States

We calculated population-level estimates of mortality, functional health, and active life expectancy for black and white adults living in a diverse set of 23 local areas in 1990, and nationwide. At age 16, life expectancy and active life expectancy vary across the local populations by as much as 28 and 25 years respectively. The relationship between population infirmity and longevity also varies. Rural residents outlive urban residents, but their additional years are primarily inactive. Among urban residents, those in more affluent areas outlive those in high-poverty areas. For both whites and blacks, these gains represent increases in active years. For whites alone they also reflect reductions in years spent in poor health.

Trade in university training: Cross-state variation in the production and stock of college-educated labor

The question of this analysis is how the production of college graduates at the state level affects the stock of college-educated workers in the state. The potential mobility of skilled workers implies that the number of college students graduating in an area need not affect the number of college graduates living in the area. However, the production of relatively large numbers of college graduates in a state may lead to increases in the employment of university-trained manpower if industries expand production of goods and services that use college-educated workers intensively. We find at best only a modest link between the production and stock of baccalaureate degree recipients.

Industrial Shifts, Skills Levels, and the Labor Market for White and Black Males

In this paper we estimate the effects of industrial shifts in the 1970s and 1980s on the wages and employment of black and white males. We use micro Census data for 52 MSAs, and estimate effects separately by age and education group. The results show that industrial shifts did reduce demand for blacks and 1essskilled males in 1970s and 1980s. Demand shifts away from manufacturing, in particular, reduced employment and wages for black and white males. While the magnitudes of these effects are fairly small for many groups, they can account for one-third to one-half of the employment decline for less-educated young blacks in the 1970s. These results imply fairly large effects on the earnings of less-skilled males in the 1980s as well.