John M. Abowd

Privacy: Theory meets Practice on the Map

In this paper, we propose the first formal privacy analysis of a data anonymization process known as the synthetic data generation, a technique becoming popular in the statistics community. The target application for this work is a mapping program that shows the commuting patterns of the population of the United States. The source data for this application were collected by the U.S. Census Bureau, but due to privacy constraints, they cannot be used directly by the mapping program. Instead, we generate synthetic data that statistically mimic the original data while providing privacy guarantees. We use these synthetic data as a surrogate for the original data. We find that while some existing definitions of privacy are inapplicable to our target application, others are too conservative and render the synthetic data useless since they guard against privacy breaches that are very unlikely. Moreover, the data in our target application is sparse, and none of the existing solutions are tailored to anonymize sparse data. In this paper, we propose solutions to address the above issues.

Job queues and the union status of workers

This paper develops a model of the determination of the union status of workers that allows for the possibility of queuing for union jobs. The empirical results derived, using a sample from the University of Michigan Panel Study of Income Dynamics, are supportive of the queuing hypothesis. The no-queue model can be rejected using a likelihood-ratio test. This suggests that a simple probit or logit model for union status is misspecified because it is not based on any consistent behavioral theory. An important implication of the model is that because most new entrants to the labor market prefer union jobs but cannot get them and because accrual of nonunion seniority makes workers progressively less likely to desire union jobs, the union status of workers is largely determined by their success in being selected from the queue early in their working life.

DOES PERFORMANCE-BASED MANAGERIAL COMPENSATION AFFECT CORPORATE PERFORMANCE?

The author, using 1981–86 data on more than 16,000 managers at 250 large corporations, investigates whether the sensitivity of managerial compensation to corporate performance in one year is positively related to corporate performance in the next year. Accounting-based measures of performance yield only weak evidence of such an association, but economic and market measures yield stronger evidence. Payment of an incremental 10% bonus for good economic performance is associated with a 30 to 90 basis point increase in the expected after-tax gross economic return in the following fiscal year; and payment of an incremental raise of 10% following a good stock market performance is associated with a 400 to 1200 basis point increase in expected total shareholder return.

Estimating Gross Labor-Force Flows

We present and apply an adjustment procedure for the Bureau of the Census and Bureau of Labor Statistics gross labor-force flows data that addresses two major defects in the data. First, an adjustment procedure is developed to take account of individuals with missing labor-force classifications who are not missing at random. Second, we provide a procedure for adjustment for individuals with spurious labor-force transitions arising because of classification errors in either the current or the previous Current Population Survey. Our procedures are applied to compute adjusted monthly gross change data for the period January 1977–December 1982. The average adjustment for nonrandom missing classifications ranges from –12% to 15% of the unadjusted gross change data. The average adjustment for spurious labor-force transitions reduces estimated movements by 8%–49%. The classification adjustment also increases estimated consecutive periods of unemployment by 18%. We apply several internal and external consistency ...

The Entry and Exit of Workers and the Growth of Employment: An Analysis of French Establishments

Using data that permit a distinction between flows of workers (directly measured) and job creation and destruction (again, directly measured), we develop employment and job flow statistics for a representative sample of French establishments from 1987 to 1990. Annual job creation can be characterized as hiring three persons and separating two for each job created in a given year. Annual job destruction can be characterized as hiring one person and separating two for each job destroyed in a given year. When an establishment is changing employment, the adjustment is made primarily by reducing entry and not by changing the separation rates. There is considerable simultaneous hiring and separation, even controlling for skill group. Two-thirds of all hiring is on short-term contracts, and more than half of all separations are due to the end of these short-term contracts.

The costs of hiring and separations

In this article, we estimate the costs of hiring, separation, and retirement of employees for a representative sample of French establishments in 1992. The estimates are computed using data from three sources: the Wage Structure Survey (ESS), the Workforce Movement Questionnaire (DMMO), and the Occupational Structure Survey (ESE). We show that the estimated costs are generally asymmetric (hiring is cheaper than terminations), increasing, and concave functions of the number of entries or exits (either retirements or terminations). There is a fixed component to each of these costs that is related to the structure of the firms personnel department. Our estimates imply that firms should not adjust gradually to the desired level of employment.

Wages, Mobility and Firm Performance: Advantages and Insights from Using Matched Worker-Firm Data

To illustrate the wide applicability of longitudinal matched employer-employee data, we study the simultaneous determination of worker mobility and wage rates using an econometric model that allows for both individual and firm-level heterogeneity. The model is estimated using longitudinally linked employer-employee data from France. Structural results for mobility show remarkable heterogeneity with both positive and negative duration dependence present in a significant proportion of firms. The average structural returns to seniority are essentially zero, but this result masks enormous heterogeneity with positive seniority returns found in low starting-wage firms. Copyright 2006 Royal Economic Society.

Econometric Analyses of Linked Employer–Employee Data

There has been an explosion in the use of linked employer-employee data that we have documented in our Handbook of Labor Economics chapter (Abowd and Kramarz, 1999a). New econometric techniques have been developed to address the problems raised by integrating longitudinal employer and employee data via the job relation. We first described these techniques in Abowd and Kramarz (1999b). In this chapter, we present new developments since that article. The key feature of linked longitudinal employer-employee data is that individuals and employing firms are identified and followed over time and the relation between them, called a job relation, is continuously monitored. Statistically there are three populations under simultaneous study. Individuals are sampled from the population of households, workplaces are sampled from the population of businesses, and jobs are sampled from the population of employment histories. Because of the multiple sampling frames involved, it is necessary to be precise about the statistical structure of the variables under study, since they may come from the individual, employer, or job data. Measured characteristics of the individual, employer, and job are collected at multiple points in time, which may or may not be synchronous. To make clear the importance of careful elaboration of the sample structure for the variables under study, we will consider some common problems in the assembly of integrated employer-employee data before we look at specific statistical models. The specific statistical models that we consider are generalizations of the specifications we first used in Abowd, Kramarz and Margolis (1999, AKM hereafter) as well as in more recent research (for instance, Abowd, Kramarz, Lengermann, and Perez-Duarte, 2003). ∗Abowd is Professor of Labor Economics, Cornell University, Distinguished Senior Research Fellow at the US Census Bureau and affiliated with the NBER and CREST. Kramarz is head of the Department of Research, Centre de Recherche en Economie et Statistique (CREST) at INSEE, Professor at Ecole Polytechnique and is a fellow at CEPR and IZA. Woodcock is Assistant Professor at Simon Fraser University. Abowd acknowledges financial support from the NSF (SBER 96-18111). 1See also Lane, Burgess and Theeuwes (1997) for a review of uses of longitudinal linked employer-employee data.

The Sensitivity of Economic Statistics to Coding Errors in Personal Identifiers

In this article we describe the sensitivity of small-cell flow statistics to coding errors in the identity of the underlying entities. Specifically, we present results based on a comparison of the U.S. Census Bureaus Quarterly Workforce Indicators before and after correcting for such errors in Social Security Number-based identifiers in the underlying individual wage records. The correction used involves a novel application of existing statistical matching techniques. It is found that even a very conservative correction procedure has a sizable impact on the statistics. The average bias ranges from .25% up to 15% for flow statistics, and up to 5% for payroll aggregates.

Multiply-Imputing Confidential Characteristics and File Links in Longitudinal Linked Data

This paper describes ongoing research to protect confidentiality in longitudinal linked data through creation of multiply-imputed, partially synthetic data. We present two enhancements to the methods of [2]. The first is designed to preserve marginal distributions in the partially synthetic data. The second is designed to protect confidential links between sampling frames.