This paper presents a finite-horizon search model that is econometrically implemented using all of the restrictions impli ed by the theory. Following a sample of male high school graduates fr om the youth cohort of the National Longitudinal Surveys from graduat ion to employment, search parameters are estimated. Reservation wages and offer probabilities are estimated to be quite low. Simulations a re performed of the impact of changing the parameters on the expected duration of unemployment. Copyright 1987 by The Econometric Society.

T HE purpose of this paper is to ascertain the extent to which deterrence, environment, and culture can be considered responsible for the observed variance in the propensity for criminal behavior across countries. It is based on time-series observations on robberies from 1955 to 1971 for three countries: England, Japan and the United States as represented by California.1 The differential impacts on the per-capita robbery rate of the certainty and severity of punishment and of economic and demographic characteristics are estimated. Culture is defined to be the set of unmeasured crime determinants that permanently differ across countries and its effect is captured by country-specific dummy variables. The character of this unobservable is inferred from the interrelationship of country-specific effects on crime to those on punishments. The choice of the three countries is based upon the availability of fairly comparable data. We examine robbery because it is defined similarly across countries and because it combines features of both property crimes and personal crimes of violence.2 Indeed, it seems that much of the variation in crime between the United States and England is due to differences in the level of crimes with personal confrontation.3 In discussing the results, I interpret the findings as a consequence of individual maximizing behavior. There has, however, been much debate over the deterrence interpretation, a debate which cannot be resolved by any single empirical effort.4 Given the extent of the controversy, a short digression on the nature of the debate should be valuable at least in placing the present study in the appropriate context. A single equation approach, as pursued in this paper, can provide unambiguous evidence on the deterrence issue only if the level of deterrence is unaffected by variations in the level of crime. Two arguments for this feedback from crime to enforcement have been suggested in the literature, one technical and the other behavioral. The former, the congestion argument, is due to the likelihood that increased crime with fixed law enforcement resources leads to declining rates of capture, conviction, and punishment. This phenomenon is mitigated by the second mechanism which is based upon the societal response to increased crime of expanding law enforcement activity. The extent to which the technical feedback is relevant depends upon the degree to which society anticipates variations in crime and the rapidity with which adjustments in law enforcement inputs can optimally be made, i.e., on costs of adjustment. A priori, it is impossible to tell which one of these will dominate in any set of observations and, thus, impossible to ascertain the direction of the bias in deterrence estimates. Indeed, each of the several law enforcement stages (arrest, conviction, sentencing) may, through these two avenues, be differentially responsive to the crime level. Two strategies can be followed given this problem, each of which may potentially contribute to our ultimate understanding. Which one is chosen depends essentially on the availability of data. One method involves estimating the supply of crime equation within the setting of a multiple equation framework using a suitable estimation technique. Much of the debate over deterrence has been concerned, therefore, with the identification issue, and it has been argued that results which employ simultaneous equation estimaReceived for publication August 15, 1978. Revision accepted for publication November 26, 1979. * Yale University. This paper was prepared under Grant Number 75N1-99-027 from the National Institute of Law Enforcement and Criminal Justice, Law Enforcement Assistance Administration, U.S. Department of Justice. Points of view or opinions stated in this document are my own and do not necessarily represent the official position or policies of the U.S. Department of Justice. I am indebted to John Treat for performing the translations of the Japanese criminal statistics. I The data for the United States as a whole are not as complete as for California alone. 2 Robbery is basically defined as a theft with violence or the threat of violence. 3 See Wolpin (1978) for a comparison of U.S. and English property crime rates. 4 See the collection of papers in Blumstein, Cohen, and Nagin (1978) for a critical view of the deterrence framework and a lengthy discussion of alternative interpretations. [ 417 1

A number of major social policy interventions have been introduced recently in the United States. The new Temporary Assistance for Needy Families (TANF) program, introduced in 1996, was advertised changing the welfare system as we know it. The new Medicare prescription drug benefit, introduced in 2006, was the largest expansion of Medicare in its history. Developing countries are also fertile ground for innovative new policies. Mexico introduced a program in 1997 (Progresa) that provided subsidies to poor rural households contingent upon the school attendance of their children. The distinction between ex post and ex ante policy evaluation is important. Ex post policy evaluation occurs upon or after the policy has been implemented. It is ubiquitous in the social sciences. Such studies make use of existing policy variation. Examples include the study of minimum wage effects on labor market outcomes, the study of the impact of welfare benefits on labor market and demographic outcomes, and the study of how divorce laws affect marital stability. The development of methodological approaches to ex post program evaluation using nonexperimental methods is an active area of research (Petra Todd 2006). There is little methodological or applied research explicitly concerned with ex ante policy evaluation using nonexperimental methods, which is perhaps surprising given its potential value. Interventions that require ex ante evaluation are those that are outside the historical experience. These include a large change in the parameters of existing programs such doubling the (real) minimum wage, adding new features to an existing program such the Medicare drug benefit program, or introducing a completely new program such Progresa. The nonexperimental approach to ex ante policy evaluation must be an extrapolation from existing policy or policy-relevant variation.' Because of that, and unlike ex post evaluation, ex ante valuation must rely on parametric and/or behavioral assumptions (theory).

Section I introduces the material. In section II a model is developed which explores the impact of input-quality uncertainty on factor demand from which is derived a rationale for the use of devices which segment the population into classes differing in their "skill" distribution parameters. The model, however, ignores the motivation of individuals to acquire the characteristics upon which firms screen, in particular, the greater incentive for the more productive to purchase the screen. This aspect has been explored by Spence (1973) and Stiglitz (1973) and will not be explicitly considered here. In section III the social value of schooling's informational context is derived within the preceeding framework. Section IV describes some empirical attempts to isolate the productivity and identification effects. The last section summarizes the paper.

This paper studies the transition from school to full-time employment and subsequent labor mobility during the first five postschooling years for several recent cohorts of black and white male "terminal" high school graduates using unique data from the 1979 youth cohort of the National Longitudinal Surveys of Labor Market Experience. A constrained optimization model of labor force dynamics is implemented empirically integrating features of models previously described in the literature. The estimates of the model provide quantitative evidence on underlying structural differences in labor market constraints faced by blacks and whites. For example, while blacks have overall a substantially smaller wage return to work experience and face a less disperse wage offer distribution, blacks face a higher probability of receiving job offers.