The second edition of this acclaimed graduate text provides a unified treatment of two methods used in contemporary econometric research, cross section and data panel methods. By focusing on assumptions that can be given behavioral content, the book maintains an appropriate level of rigor while emphasizing intuitive thinking. The analysis covers both linear and nonlinear models, including models with dynamics and/or individual heterogeneity. In addition to general estimation frameworks (particular methods of moments and maximum likelihood), specific linear and nonlinear methods are covered in detail, including probit and logit models and their multivariate, Tobit models, models for count data, censored and missing data schemes, causal (or treatment) effects, and duration analysis. Econometric Analysis of Cross Section and Panel Data was the first graduate econometrics text to focus on microeconomic data structures, allowing assumptions to be separated into population and sampling assumptions. This second edition has been substantially updated and revised. Improvements include a broader class of models for missing data problems; more detailed treatment of cluster problems, an important topic for empirical researchers; expanded discussion of "generalized instrumental variables" (GIV) estimation; new coverage (based on the author's own recent research) of inverse probability weighting; a more complete framework for estimating treatment effects with panel data, and a firmly established link between econometric approaches to nonlinear panel data and the "generalized estimating equation" literature popular in statistics and other fields. New attention is given to explaining when particular econometric methods can be applied; the goal is not only to tell readers what does work, but why certain "obvious" procedures do not. The numerous included exercises, both theoretical and computer-based, allow the reader to extend methods covered in the text and discover new insights.

/pdf/econometric-analysis-of-cross-section-and-panel-data-3jzr8emmtf.pdf

Econometric Analysis of Cross Section and Panel Data

CULTURE, TRADITION, AND OVERT DISCRIMINATION tend to make restrictive the terms by which women may participate in the labor force. These influences combine to generate an unfavorable occupational distribution of female workers vis-a-vis male workers and to create pay differences between males and females within the same occupation. The result is a chronic earnings gap between male and female full-time, year-round workers. Unfortunately, explanations at this level of generality are mainly descriptive. It is the purpose of this paper to estimate the average extent of discrimination against female workers in the United States and to provide a quantitative assessment of the sources of male-female wage differentials.

Male-Female Wage Differentials in Urban Labor Markets

The equilibrium growth model is modified and used to explain the cyclical variances of a set of economic time series, the covariances between real output and the other series, and the autocovariance of output. The model is fitted to quarterly data for the post-war U.S. economy. Crucial features of the model are the assumption that more than one time period is required for the construction of new productive capital, and the non-time-separable utility function that admits greater intertemporal substitution of leisure. The fit is surprisingly good in light of the model's simplicity and the small number of free parameters. THAT WINE IS NOT MADE in a day has long been recognized by economists (e.g., Bdhm-Bawerk [6]). But, neither are ships nor factories built in a day. A thesis of this essay is that the assumption of multiple-period construction is crucial for explaining aggregate fluctuations. A general equilibrium model is developed and fitted to U.S. quarterly data for the post-war period. The co-movements of the fluctuations for the fitted model are quantitatively consistent with the corresponding co-movements for U.S. data. In addition, the serial correlations of cyclical output for the model match well with those observed. Our approach integrates growth and business cycle theory. Like standard growth theory, a representative infinitely-lived household is assumed. As fluctuations in employment are central to the business cycle, the stand-in consumer values not only consumption but also leisure. One very important modification to the standard growth model is that multiple periods are required to build new capital goods and only finished capital goods are part of the productive capital stock. Each stage of production requires a period and utilizes resources. Halffinished ships and factories are not part of the productive capital stock. Section 2 contains a short critique of the commonly used investment technologies, and presents evidence that single-period production, even with adjustment costs, is inadequate. The preference-technology-information structure of the model is presented in Section 3. A crucial feature of preferences is the non-time-separable utility function that admits greater intertemporal substitution of leisure. The exogenous stochastic components in the model are shocks to technology and imperfect indicators of productivity. The two technology shocks differ in their persistence.

/pdf/time-to-build-and-aggregate-fluctuations-1tevhp3oax.pdf

Time to build and aggregate fluctuations

This paper considers whether it is possible to devise a nonexperimental procedure for evaluating a prototypical job training programme. Using rich nonexperimental data, we examine the performance of a two-stage evaluation methodology that (a) estimates the probability that a person participates in a programme and (b) uses the estimated probability in extensions of the classical method of matching. We decompose the conventional measure of programme evaluation bias into several components and find that bias due to selection on unobservables, commonly called selection bias in econometrics, is empirically less important than other components, although it is still a sizeable fraction of the estimated programme impact. Matching methods applied to comparison groups located in the same labour markets as participants and administered the same questionnaire eliminate much of the bias as conventionally measured, but the remaining bias is a considerable fraction of experimentally-determined programme impact estimates. We test and reject the identifying assumptions that justify the classical method of matching. We present a nonparametric conditional difference-in-differences extension of the method of matching that is consistent with the classical index-sufficient sample selection model and is not rejected by our tests of identifying assumptions. This estimator is effective in eliminating bias, especially when it is due to temporally-invariant omitted variables.

Matching As An Econometric Evaluation Estimator: Evidence from Evaluating a Job Training Programme

The relationship between household wealth and educational enrollment of children can be estimated without expenditure data. A method for doing so - which uses an index based on household asset ownership indicators - is proposed and defended in this paper. In India, children from the wealthiest households are over 30 percentage points more likely to be in school than those from the poorest households, although this gap varies considerably across states. To estimate the relationship between household wealth and the probability that a child (aged 6 to 14) is enrolled in school, Filmer and Pritchett use National Family Health Survey (NFHS) data collected in Indian states in 1992 and 1993. In developing their estimate Filmer and Pritchett had to overcome a methodological difficulty: The NFHS, modeled closely on the Demographic and Health Surveys, measures neither household income nor consumption expenditures. As a proxy for long-run household wealth, they constructed a linear asset index from a set of asset indicators, using principal components analysis to derive the weights. This asset index is robust, produces internally coherent results, and provides a close correspondence with data on state domestic product and on state level poverty rates. They validate the asset index using data on consumption spending and asset ownership from Indonesia, Nepal, and Pakistan. The asset index has reasonable coherence with current consumption expenditures and, more importantly, works as well as - or better than - traditional expenditure-based measures in predicting enrollment status. The authors find that on average a child from a wealthy household (in the top 20 percent on the asset index developed for this analysis) is 31 percent more likely to be enrolled in school than a child from a poor household (in the bottom 40 percent). This paper - a product of Poverty and Human Resources, Development Research Group - is part of a larger effort in the group to inform educational policy. The study was funded by the Bank`s Research Support Budget under the research project Educational Enrollment and Dropout (RPO 682-11).

Estimating Wealth Effects without Expenditure Data or Tears: With an Application to Educational Enrollments in States of India

GOVERNMENTAL post-schooling training programs have become a permanent fixture of the U.S. economy in the last decade. These programs are typically advocated for diverse reasons: (1) to reduce inflation by the provision of more skilled workers to alleviate shortages, (2) to reduce unemployment of certain groups, and (3) to reduce poverty by increasing the skills of certain groups. All of these objectives require that training programs increase the earnings of trainees above what they otherwise would be. For example, alleviating shortages by training more highly skilled workers should increase the earnings of these workers. Likewise, the concern for unemployed workers is derived from a concern for the decreased earnings of these workers; and if trainees subsequently suffer less unemployment, their earnings should be higher. Finally, training programs are intended to reduce poverty by increasing the earnings of low income workers. Evaluating the success of training programs is thus inherently a quantitative assessment of the effect of training on trainee earnings.' It is an important process both because it helps to inform discussions of public policy by shedding light on the past value of these programs as investments and because it can provide a means of testing our ability to augment the human capital of certain workers. Although there have been many studies of the effect of post-school classroom training on earnings it is by now rather widely agreed that very little is reliably known about the actual effects of these programs.2 Three main problems account for this state of affairs: (1) the large sample sizes required to detect relatively small anticipated program effects in a variable with such high variance as earnings, (2) the considerable expense required to keep track of trainees over a long enough period of time to measure the full inter-temporal impact of training, and (3) the extreme difficulty of implementing an adequate experimental design so as to obtain a group against which to reliably compare trainees.3 The purpose of this paper is to report on efforts to cope with this third problem using a data collection system that comes some way towards resolving the first two. The basic idea of this data system is to match the program record on each trainee with the trainee's Social Security earnings history. The Social Security Administration maintains a summary year-by-year earnings history for each Social Security account over the period since 1950 that may be used, under the appropriate confidentiality restrictions, for this purpose.4 In this paper I have concentrated on an analysis of all classroom trainees who started training under the Manpower Development and Training Act (MDTA) in the first 3 months of 1964 so as to ensure their having completed training in that year. In choosing to analyze trainees from so early a cohort something is clearly lost. On the one hand, the nature of the participants in these early years was considerably different than in the later years. In particular, programs geared Received for publication February 9, 1977. Revision accepted for publication August 1, 1977. * Princeton University. This research was supported by ASPER, U.S. Department of Labor, but does not represent an official position of the Department of Labor, its agencies, or staff. I would like to thank Gregory Chow, Ronald Ehrenberg, Roger Gordon, Zvi Griliches, George E. Johnson, Nicholas Kiefer, Richard Quandt, and Sherwin Rosen for helpful comments. I also owe a heavy debt to D. Alton Smith for computational and other assistance. 'See Reid (1976), for example, for a clear analysis of how knowledge of these effects is required in order to establish the impact of government training on the black/white wage differential. 2 Surveys of many of these studies may be found in Stromsdorfer (1972) and O'Neill (1973). 3For further discussion of these points see Ashenfelter (1975). 4The idea for using these data to analyze the effectiveness of government training programs is apparently quite an old one, having been suggested by the National Manpower Advisory Committee (U.S. Department of Labor, 1972) to the Secretary of Labor at its first meeting in a letter dated October 10, 1962, the year of passage of the Manpower Development and Training Act. Actual efforts along these lines were ultimately reported by Borus (1967), Commins (1970), Farber (1970), and Prescott and Cooley (1972).

Estimating the Effect of Training Programs on Earnings.

This paper uses a new survey to contrast the wages of genetically identical twins with different schooling levels. Multiple measurements of schooling levels were also collected to assess the effect of reporting error on the estimated economic returns to schooling. The data indicate that omitted ability variables do not bias the estimated return to schooling upward, but that measurement error does bias it downward. Adjustment for measurement error indicates that an additional year of schooling increases wages by 12-16 percent, a higher estimate of the economic retums to schooling than has been previously found. (JEL J31) This paper uses a new survey of identical twins to study the economic returns to schooling. We estimate the returns to schooling by contrasting the wage rates of identical twins with different schooling levels. Our goal is to ensure that the correlation we observe between schooling and wage rates is not due to a correlation between schooling and a worker's ability or other characteristics. We do this by taking advantage of the fact that monozygotic (from the same egg) twins are genetically identical and have similar family backgrounds. In our survey we also took some unusual steps to measure a worker's schooling level accurately. We obtained independent estimates of each sibling's schooling level by asking the twins to report on both their own and their twin's schooling. These new data provide a simple and powerful method for assessing the role of measurement error in estimates of the economic returns to schooling.

/pdf/estimates-of-the-economic-return-to-schooling-from-a-new-45ufjm0qod.pdf

Estimates of the Economic Return to Schooling from a New Sample of Twins

In this paper we set out some methods that utilize the longitudinal structure of earnings of trainees and a comparison group to estimate the effectiveness of training for the 1976 cohort of CETA trainees. By fitting a components-of-variance model of earnings to the control group, and by posing a simple model of program participation, we are able to predict the entire pre-training and post-training earnings histories of the trainees. The fit of these predictions to the pre-training earnings of the CETA participants provides a test of the model of earnings generation and program participation and a simple check on the corresponding estimate of the effectiveness of training. Two assumptions have a strong influence on the magnitude of the estimated training effects: the timing of the decision to participate in training, and the presence or absence of individual-specific trends in earnings. We find considerable evidence that trainee earnings con- tain permanent, transitory,and trend-like components of selection bias. We are less successful in empirically distinguishing between alternative assumptions on the timing of the participation decision. If earnings in the year prior to training are the appropriate selection criterion, then our estimate of the training effect for adult male CETA partici- pants is about 300 dollars per year. Our estimates for female CETA participants are larger and less sensitive to alternative models of program participation.

Using the Longitudinal Structure of Earnings to Estimate the Effect of Training Programs

This paper uses a new survey to contrast the wages of genetically identical twins with different schooling levels. Multiple measurements of schooling levels were also collected to assess the effect of reporting error on the estimated economic returns to schooling. The data indicate that omitted ability variables do not bias the estimated return to schooling upward but that measurement error does bias it downward. Adjustment for measurement error indicates that an additional year of schooling increases wages by 12 to 16 percent, a higher estimate of the economic returns to schooling than has been previously found. Copyright 1994 by American Economic Association.

Estimates of the Economic Returns to Schooling from a New Sample of Twins

In this paper we set out some methods that utilize the longitudinal structure of earnings of trainees and a comparison group to estimate the effectiveness of training for the 1976 cohort of CETA trainees. By fitting a components-of-variance model of earnings to the control group, and posing a simple model of program participation, we are able to predict the entire earnings histories of the trainees. The fit of these predictions to the pre-training earnings of the CETA participants provides a test of the model of earnings generation and program participation and simple check on the corresponding estimate of the effectiveness of training.Two factors appear to have a critical influence on the size of the estimated training effects: the time of the decision to participate in training and the presence or absence of individual-specific trends in earnings. We find considerable evidence that trainee earnings contain permanent, transitory,and trend-like components of selection bias. We are less successful in distinguishing empirically between alternative assumptions on the timing of the participation decision. If earnings in the year prior to training are the appropriate selection criterion, however, our estimate of the training effect for adult male CETA participants is about 300 dollars per year. Our estimates for female CETA participants are larger, and less sensitive to alternative models of program participation.(This abstract was borrowed from another version of this item.)

/pdf/using-the-longitudinal-structure-of-earnings-to-estimate-the-3m86z0cs3b.pdf

Orley Ashenfelter

Papers

Estimating the Effect of Training Programs on Earnings.

Estimates of the Economic Return to Schooling from a New Sample of Twins

Using the Longitudinal Structure of Earnings to Estimate the Effect of Training Programs

Estimates of the Economic Returns to Schooling from a New Sample of Twins

Using the Longitudinal Structure of Earnings to Estimate the Effect of Training Programs