Knowledge that Transforms

We revisit Reinhardt's (1972) commonly used production function for physician services. This production function, although appropriate in some settings, is not adequate for more detailed studies. The generalized linear production function (Diewart, 1971) is an attractive alternative for this application: it admits zero values for inputs and allows the estimation of complex technical relationships. We find that, from 1965-1988, the technical relationships that describe the production process for physician services are remarkably stable. By empirically estimating the parameters of this more general production function, we provide crucial evidence about the q-complementarity of health worker inputs, the first of its kind in the literature on health labor markets.

This paper analyzes both theoretically and empirically how an absolute grading standard that allows only a small number of distinct grades affects student course performance outcomes. The clearest prediction of the model is that course performance of “grade-motivated” students will tend to be clustered slightly above the boundaries that separate grades, as long as the variance of the random component of performance is not too large. A more tenuous prediction is that the proximity of a grade-motivated student to a grade boundary going into the final exam should influence final exam performance, after controlling for prefinal exam performance. An empirical investigation of the course performance of university students who were enrolled in introductory economics classes that used an absolute grading standard finds evidence in favor of both of these predictions. The results suggest that student effort decisions respond to the incentives created by the grading system.

This paper uses data from the Census Bureau's Longitudinal Research Database to describe the dynamics of geographic concentration in U.S. manufacturing industries. Agglomeration results from a combination of the mean reversion and randomness in the growth of state-industry employment. Although industries' agglomeration levels have declined only slightly over the last quarter century, we find a great deal of movement for many geographically concentrated industries. We decompose aggregate concentration changes into portions attributable to plant births, expansions, contractions, and closures. We find that the location choices of new firms play a deagglomerating role, whereas plant closures have tended to reinforce agglomeration.

This study evaluates the roles of ideology, constituency, and political party for roll-call voting in the U.S. Senate on a broad set of environmental issues. The study estimates a model of political support using voting scores from the League of Conservation Voters (LCV) for the period 1988-1998, including observations on 91 senators for 130 roll-call votes. The study decomposes the scale-adjusted scores into relative weights due to the general electorate, the senator's support constituency, party leadership, and ideology. The main findings are that a senator's ideology is by far the most important consideration for voting profiles on environmental issues, and that party affiliation and regional loyalty explain about 74% of measured ideology. Hence, “green” voting tends to be highly partisan.

In many surveys, multiple observations on the dependent variable are collected from a given respondent. The resulting pooled data set is likely to be censored and to exhibit cross-sectional heterogeneity. We propose a model that addresses both issues by allowing regression coefficients to vary randomly across respondents and by using the Geweke-Hajivassiliou-Keane simulator and Halton sequences to estimate high-order probabilities. We show how this framework can be usefully applied to the estimation of power outage costs to firms using data from a recent survey conducted by a U.S. utility. Our results strongly reject the hypotheses of parameter constancy and cross-sectional homogeneity.

In light of recent, state-level actions banning racial preference in college admissions decisions, we investigate how whites and minorities differ in their college-going behavior. Using data from the National Longitudinal Survey of Youth, we estimate a sequential model of college attendance and graduation decisions that allows correlations among the errors. Our estimates reveal that minorities are more likely than observationally equivalent whites to attend colleges of all quality levels. Being a minority has a positive effect on graduation probabilities, but, overall, minorities are less likely than their white counterparts to complete college because they possess fewer favorable unobserved factors.

Proponents of environmental policies based on liability assert that strict liability imposed on polluters induces firms to handle hazardous wastes properly. We examine unintended pollution releases, relating them to the liability imposed on polluters by state mini-Superfund laws. We find that strict liability reduces the frequency and severity of pollution releases, but that its effects vary with firm size. Smaller firms are associated with more pollution releases in states with strict liability, suggesting that in these states smaller firms may have specialized in riskier production processes. Their number, however, has not necessarily grown in response to the state's liability policy.

I examine the effects of FCC regulation on the innovation and introduction of advanced telecommunications services in the United States. An interim of lighter regulation provides an "experiment" to test the regulatory regime's impact. The econometric model comprises an arrival process (for service innovation) followed by a duration process (for regulatory delay). The number of services the firms created during the interim is 60%-99% higher than the model predicts they would have created if the stricter regulation had still been in place. Overall, firms would have introduced 62% more services to consumers during the study period if the regulation had not been in place. © 2002 President and Fellows of Harvard College and the Massachusetts Institute of Technology.

This paper investigates nonlinear features of FX volatility dynamics using estimates of daily volatility based on the sum of intraday squared returns. Measurement errors associated with using realized volatility to estimate ex post latent volatility imply that standard time series models of the conditional variance become variants of an ARMAX model. We explore nonlinear departures from these linear specifications using a doubly stochastic process under duration-dependent mixing. This process can capture large abrupt changes in the level of volatility, time-varying persistence, and time-varying variance of volatility. The results have implications for forecast precision, hedging, and pricing of derivatives.

Previous literature indicates that stock returns are predictable by several strongly autocorrelated forecasting variables, especially at longer horizons. It is suggested that this finding is spurious and follows from a neglected near unit root problem. Instead of the commonly used t-test, we propose a test that can be considered as a general test of whether the return can be predicted by any highly persistent variable. Using this test, no predictability is found for U.S. stock return data from the period 1928-1996. Simulation experiments show that the standard t-test clearly overrejects whereas our proposed test controls size much better.