A unique dataset collected from the personnel records of a large company is used to study the relationship between on-the-job training and worker productivity. The analysis shows how information contained in a company database is useful for eliminating heterogeneity bias in the estimation of training's impact on wages and job performance. Even when selection bias in assignment to training programs is eliminated, training is found to have a positive and significant effect on both wage growth and the change in job performance scores, thereby confirming the robustness of the relationship between training and productivity.

An earlier version of this paper was presented at the Conference on Franchise Contracting, Organization, and Regulation, Michigan Business School. I am indebted to the conference participants, and particularly to G. Frank Mathewson, for helpful suggestions. I also received perceptive comments from the conference organizers, Francine Lafontaine and Scott Masten. I have benefited from discussions of the subject matter of this paper with Tasneem Chipty, James Delaney, Robert Fenili, Tom Hogarty, Robert Lande, Jim Peck, and Anthony Robinson. None of the above should be implicated in the conclusions put forth below.

J. S. Duesenberry's (1949) ratcheting consumption demand is derived as a feature of the optimal dynamic consumption and investment policy given extreme habit formation that prevents consumption from falling over time. Preferences are in effect non-time-separable, extended-real-valued von Neumann-Morgenstern preferences. Consumption increases each time wealth reaches a new maximum. Risky investment is proportional to the excess of wealth over the perpetuity value of current consumption. Extensions constrain the net rate of decrease in consumption with a constant other than zero, add more consumption goods, and constrain on the maximal holding of the risky asset as a proportion of wealth. Copyright 1995 by The Review of Economic Studies Limited.

This paper analyzes a newly assembled data set consisting of subjective indices of corruption, the amount of red tape, the efficiency of the judicial system, and various categories of political stability for a cross section of countries. Corruption is found to lower investment, thereby lowering economic growth. The results are robust to controlling for endogeneity by using an index of ethnolinguistic fractionalization as an instrument.

This article applies economic analysis to rural households in poor countries to see what one may mean by a "population problem." It is argued that there is a serious population problem in certain regions of the world, and that it is in varying degrees linked to poverty, to gender inequalities in the exercise of power, to communal sharing of child-rearing, and to an erosion of the local environmental-resource base. It is argued that some of the links may, to an extent, be synergistic. One manifestation of the problem is that very high fertility rates are experienced by women bearing risks of death that should now be unacceptable. An argument is sketched to show how the cycle of poverty, low birth-weight and stature, and high fertility rates can perpetuate within a dynasty. The one general policy conclusion that emerges is that a population policy in these parts should not only contain such measures as family-planning programs, improved female education, and employment opportunities, but also those measures that are directed at the alleviation of poverty, such as improved credit, insurance, and savings opportunities, and a ready availability of basic household needs, such as potable water and fuel. It is argued that these latter measures lower the net private benefits of procreation.

Externalities between buyers are shown to induce delays in negotiations between a seller and several buyers. Delays arise in a perfect and complete information setting with random matching even when there is no deadline. While with a deadline we identify delays both for positive and negative externalities, without a deadline we find that (1) when externalities are positive, there exists no SPNE in pure strategies with bounded recall that exhibits delay; (2) when externalities are negative, it may happen that all SPNE with bounded recall have the property that long periods of waiting alternate with short periods of activity: This is the cyclical delay phenomenon.

Return models (returns regressed on scaled earnings variables) are commonly preferred to price models (stock price regressed on earnings per share). We provide a framework for choosing between these models. An economically intuitive rationale suggests that price models are better specified in that the estimated slope coefficients from price models, but not return models, are unbiased. Our empirical results confirm that price models' earnings response coefficients are less biased. However, return models have less serious econometric problems than price models. In some research contexts the combined use of both price and return models may be useful.