Knowledge that Transforms

Pseudo maximum likelihood techniques are applied to basic Poisson models and to Poisson models with specification errors. In the latter case it is shown that consistent and asymptotically normal estimators can be obtained without specifying the p.d.f. of the disturbances. These estimators are compared both from the finite sample and the asymptotic point of view. Quasi generalized PML estimators, which asymptotically dominate all PML estimators, are also proposed. Finally, bivariate and panel data Poisson models are discussed. THE ANALYSIS OF ECONOMIC BEHAVIOR often leads to the study of characteristics taking a small number of positive values. The classical linear model is not adapted to explain how such discrete variables depend on other quantitative or qualitative variables. The reasons are similar to those usually given in the case of an endogenous qualitative variable: the shape of the observation set does not correspond to a linear model, the assumption of normality of the disturbances cannot be made, since the endogenous variables take a small number of values with strictly positive probabilities, and the prediction formulae which are deduced from a linear model give impossible values. In the models considered in the literature to describe discrete variables (Cox and Lewis [2], El. Sayyad [3], Frome, Kutner, and Beauchamp [4], Gilbert [5], Hausman, Hall, and Griliches [8]; see also Lancaster [12]) the endogenous variable is assumed to have a Poisson distribution conditional upon the exogenous variables. The parameter of this distribution is a function of the values of the exogenous variables. The choice of such a model is justified if the dependent variable counts the occurrence of a given event during a fixed period and if the usual assumptions of the Poisson process are satisfied. For instance, the model is adapted to describe daily numbers of oil tankers' arrivals in a port, the number of accidents at work by factory, or the number of patents applied for and received by firms (Hausman, Hall, and Griliches [8]).

This paper examines how measurement error biases longitudinal estimates of union effects. It develops numerical examples, statistical models, and econometric estimates which indicate that measurement error is a major problem in longitudinal data sets, so that longitudinal analyses do not provide the research panacea for determining the effects of unionism (or other economic forces) some have suggested. There are three major findings:

Participants in labor market experiments are aware that the experiments run for a limited amount of time. Thus behavior during a temporary experiment will be different than if the experimental change in constraints confronting participants were permanent. In evaluating changes in policy, however, the response to permanent changes is of primary interest. The paper analyzes conditions under which permanent responses can be inferred from data on temporary experiments.

This paper formulates and estimates a model for the determination of employer and union demand for multiemployer (vs. single employer) bargaining units. Utility-maximizing, risk-averse firms and unions are both assumed to weigh the impact of each type of bargaining unit on the expected level and variability of profits and wages, respectively. The model is tested on a 1975 sample of 3,486 individual collective bargaining agreements. Because either party can generally leave a multiemployer unit without the other party's consent, a partially observed bivariate probit model is used to estimate demand for structure. It is found that the forgone profits due to a multiemployer unit (relative to a single-firm unit) lower firm demands for this type of unit, while forgone wages in a multiemployer unit (relative to a single-firm unit) lower union demand for this type of unit.

Affirmative action under Executive Order 11246 ranks among the most controversial of domestic federal policies. This study asks whether affirmative action has been successful in promoting the employment of minorities and females. It compares the change in demographics between 1974 and 1980 at more than sixty-eight thousand establishments, and finds that both minority and female employment have increased faster at establishments subject to affirmative action. Compliance reviews, while not well targeted are also found to have been effective.

H. Gregg Lewis' estimates of the relative wage effect of unionism between 1920 and 1958 are routinely cited though they have rarely been subject to scrutiny. This paper extends Lewis' data to 1980 and, in particular, we construct a series on union membership that links up with the data available in the 1970's from the Current Population Surveys. We proceed to reexamine the effects of trade unions both on relative wages and on relative manhours worked.

This paper extends the theory of self-selection to circumstances in which economic agents have some access to markets. We use the analysis to explain the existence of multidimensional compensation packages in the presence of limited (re)marketability. Employment contracts that include fringe benefits are prominent examples of such multidimensional packages.