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

Most papers that employ Differences-in-Differences estimation (DD) use many years of data and focus on serially correlated outcomes but ignore that the resulting standard errors are inconsistent. To illustrate the severity of this issue, we randomly generate placebo laws in state-level data on female wages from the Current Population Survey. For each law, we use OLS to compute the DD estimate of its “effect” as well as the standard error of this estimate. These conventional DD standard errors severely understate the standard deviation of the estimators: we find an “effect” significant at the 5 percent level for up to 45 percent of the placebo interventions. We use Monte Carlo simulations to investigate how well existing methods help solve this problem. Econometric corrections that place a specific parametric form on the time-series process do not perform well. Bootstrap (taking into account the autocorrelation of the data) works well when the number of states is large enough. Two corrections based on asymptotic approximation of the variance-covariance matrix work well for moderate numbers of states and one correction that collapses the time series information into a “pre”- and “post”-period and explicitly takes into account the effective sample size works well even for small numbers of states.

/pdf/how-much-should-we-trust-differences-in-differences-407p70a9b4.pdf

How Much Should We Trust Differences-In-Differences Estimates?

https://www.oxfordmartin.ox.ac.uk/downloads/academic/The_Future_of_Employment.pdf

The future of employment: How susceptible are jobs to computerisation?

This paper examines the reflection problem that arises when a researcher observing the distribution of behaviour in a population tries to infer whether the average behaviour in some group influences the behaviour of the individuals that comprise the group. It is found that inference is not possible unless the researcher has prior information specifying the compisition of reference groups. If this information is available, the prospects for inference depend critically on the population relationship between the variables defining reference groups and those directly affecting outcomes. Inference is difficult to implossible if these variables are functionally dependent or are statistically independent. The prospects are better if the variables defining reference groups and those directly affecting outcomes are moderately related in the population.

/pdf/identification-of-endogenous-social-effects-the-reflection-rurdh6yhqn.pdf

Identification of Endogenous Social Effects: The Reflection Problem

▪ Abstract This paper assesses and synthesizes the cumulative results of a new “neighborhood-effects” literature that examines social processes related to problem behaviors and health-related outcomes. Our review identified over 40 relevant studies published in peer-reviewed journals from the mid-1990s to 2001, the take-off point for an increasing level of interest in neighborhood effects. Moving beyond traditional characteristics such as concentrated poverty, we evaluate the salience of social-interactional and institutional mechanisms hypothesized to account for neighborhood-level variations in a variety of phenomena (e.g., delinquency, violence, depression, high-risk behavior), especially among adolescents. We highlight neighborhood ties, social control, mutual trust, institutional resources, disorder, and routine activity patterns. We also discuss a set of thorny methodological problems that plague the study of neighborhood effects, with special attention to selection bias. We conclude with promising ...

ASSESSING "NEIGHBORHOOD EFFECTS": Social Processes and New Directions in Research

This paper shows the importance of explicitly accounting for the possibility of recalls when analyzing the determinants of unemployment spell durations and the effects of unemployment insurance (UI) on unemployment outcomes in the United States. These issues are examined using a unique sample of UI recipients from Missouri and Pennsylvania covering unemployment spells in the 1979- 1981 period. We find that those expecting recall who are not recalled tend to have quite long unemployment spells. Furthermore, ex-ante temporary layoff spells (the spells of individuals' who initially expect to be recalled) may account for over 60 percent of the unemployment of UI recipients and appear to account for much more unemployment than ex-post temporary layoff spells (spells actually ending in recall). We estimate a competing risks model in which the finding of a new job and recall are treated as alternate routes of leaving unemployment. Our results using this approach show that the recall and new job exit probabilities have quite different time patterns and are often affected in opposite ways by explanatory variables. We also find that the probability of leaving unemployment (both through recalls and new job finding) increases greatly around the time that UI benefits lapse.

https://papers.ssrn.com/sol3/Delivery.cfm?abstractid=428343

Unemployment Insurance, Recall Expectations, and Unemployment Outcomes

The U.S. wage structure evolved across the last century: narrowing from 1910 to 1950, fairly stable in the 1950s and 1960s, widening rapidly during the 1980s, and "polarizing" since the late 1980s. We document the spectacular rise of U.S. wage inequality after 1980 and place recent changes into a century-long historical perspective to understand the sources of change. The majority of the increase in wage inequality since 1980 can be accounted for by rising educational wage differentials, just as a substantial part of the decrease in wage inequality in the earlier era can be accounted for by decreasing educational wage differentials. Although skill-biased technological change has generated rapid growth in the relative demand for more-educated workers for at least the past century, increases in the supply of skills, from rising educational attainment of the U.S. work force, more than kept pace for most of the twentieth century. Since 1980, however, a sharp decline in skill supply growth driven by a slowdown in the rise of educational attainment of successive U.S. born cohorts has been a major factor in the surge in educational wage differentials. Polarization set in during the late 1980s with employment shifts into high- and low-wage jobs at the expense of the middle leading to rapidly rising upper tail wage inequality but modestly falling lower tail wage inequality.

/pdf/long-run-changes-in-the-u-s-wage-structure-narrowing-2vqj8848m2.pdf

Long-Run Changes in the U.S. Wage Structure: Narrowing, Widening, Polarizing

We present the first estimates of the returns to years of schooling before 1940 using a large sample individuals (from the 1915 Iowa State Census). The returns to a year of high school or college were substantial in 1915â€”about 11 percent for all males and in excess of 12 percent for young males. Education enabled individuals to enter lucrative white-collar jobs, but sizable educational wage differentials also existed within occupational groups. Returns were substantial even for those in farming. We find, using U.S. census data, that returns to education decreased between 1915 and 1940 and again during the 1940s.

Education and Income in the Early 20th Century: Evidence from the Prairies

Numerous studies have shown large differences in wages for apparently similar workers across industries. These findings pose a challenge to standard model s of labor market behavior. A problem with past studies of industry wage differences is that they have failed to distinguish between union and nonunion workers. Many economists may expect union workers wages to be set in a noncompetitive fashion but would be surprised if nonunion wages were. We examine the differences in wages across industries for both union and nonunion workers. We find that even after controlling for a wide range of personal characteristics and geographic location large wage differences persist for both union and nonunion workers. Furthermore the premiums of union and nonunion workers are highly correlated. We review past studies which demonstrate that industry wage premiums are also highly correlated across countries and have been very similar over many decades. We present new evidence that the wages of different occupations are highly correlated across industries -- that is if any occupation in an industry is highly paid all occupations are. We also review the evidence which suggests that people who move from low to high paying industries receive a large fraction of the industry wage premium and that those who move from high to low paying industries lose the premium. Finally, we review the evidence on the correlates of industry wage differences. Quit rates, human capital variables, capital labor ratios and market power measures are all positively correlated with industry wage differences individually though the data are not adequate to determine their independent contributions in multiple regression. On the basis of all the evidence we conclude that standard labor market clearing models can not easily explain all the facts. Several alternative models are discussed including efficiency wage and collective action threat mode1 s. These are found to be more consistent with the facts though some troubling problems remain.

Inter-Industry Wage Differences and Theories of Wage Determination

By the mid-nineteenth century school enrollment rates in the United States exceeded those of any other nation in the world and by the early twentieth century the United States had accomplished mass education at all levels. No country was able to close the gap until the last quarter of the twentieth century. For much of its history U.S. education was spurred by a set of 'virtues,' the most important of which were public provision by small fiscally independent districts, public funding, secular control, gender neutrality, open access, a forgiving system, and an academic curriculum. The outcomes of the virtues were an enormous diffusion of educational institutions and the early spread of mass education. America borrowed its educational institutions from Europe but added to them in ways that served to enhance competition and openness. The virtues of long ago need not be the virtues of today, and they also need not have been virtuous in all places and at all times in the past. In this essay we explore the historical origins of these virtues and find that almost all were in place in the period before the American Civil War.

/pdf/the-virtues-of-the-past-education-in-the-first-hundred-years-3uqqzbp8e3.pdf

Lawrence F. Katz

Papers

Unemployment Insurance, Recall Expectations, and Unemployment Outcomes

Long-Run Changes in the U.S. Wage Structure: Narrowing, Widening, Polarizing

Education and Income in the Early 20th Century: Evidence from the Prairies

Inter-Industry Wage Differences and Theories of Wage Determination

The "virtues" of the past: education in the first hundred years of the new republic