Knowledge that Transforms

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at

In the context of the classical linear model, the problem of comparing two arbitrary hypotheses on the regression coefficients is considered. Problems involving nonlinear hypotheses, inequality restrictions, or non-nested hypotheses are included as special cases. Exact bounds on the null distribution of likelihood ratio statistics are derived. In an important special case, a bounds test similar to the Durbin-Watson test is proposed. Multiple testing problems are also studied

Larry G. Epstein, Stanley E. Zin, Substitution, Risk Aversion, and the Temporal Behavior of Consumption and Asset Returns: A Theoretical Framework, Econometrica, Vol. 57, No. 4 (Jul., 1989), pp. 937-969

This paper presents an empirical analysis of changes in individual earnings and hours ov&r time. Using longitudinal data from three panel surveys, we catalogue the main features of the covariance structure of changes in earnings and hours. We then present an interpretation of these features in terms of both a life-cycle labor supply model and a fixed-wage labor contract nDdel. Our major findings are: (1) there is a remarkable similarity in the covariance structure of earnings and hours changes across the three surveys; and (2) apart from simple measurement error, the major component of variance in earnings and hours affects earnings and hours equi-proportionately.

Methods for the systematic application of Monte Carlo integration with importance sampling to Bayesian inference are developed. Conditions under which the numerical approximation converges almost surely to the true value with the number of Monte Carlo replications, and its numerical accuracy may be assessed reliably, are given. Importance sampling densities are derived from multivariate normal or student approximations to the posterior density. These densities are modified by automatic rescaling along each axis. The concept of relative numerical efficiency is introduced to evaluate the adequacy of a chosen importance sampling density. Applications in two illustrative models are presented. Copyright 1989 by The Econometric Society.

This paper models occasional, discrete shifts in the growth rate of a nonstationary series. Algorithms for inferring these unobserved shifts are presented, a byproduct of which permits estimation of parameters by maximum likelihood. An empirical application of this technique suggests that the periodic shift from a positive growth rate to a negative growth rate is a recurrent feature of the U.S. business cycle, and indeed could be used as an objective criterion for defining and measuring economic recessions. The estimated parameter values suggest that a typical economic recession is associated with a 3 percent permanent drop in the level of GNP. Copyright 1989 by The Econometric Society.

IN MANY AREAS OF SCIENCE, when drawing conclusions concerning a set of phenomena, individual or linear combinations of parameters are assumed to be nonnegative, nonpositive, or to lie between upper and lower bounds. Likewise, many functions are assumed to be monotonic, convex, or quasiconvex. Consequently, questions naturally arise as to the correctness of the assumptions and, within the context of statistical inference, how to evaluate the evidence against these assumptions and to use the data at hand to seek the appropriate statistical model and the corresponding method of estimation. Within the context of the general linear statistical model, multivariate analogues of one-sided (inequality hypotheses) tests have been explored by Bartholomew (1959), Kudo (1963), Osterhoff (1969), Barlow, et al. (1972), Yancey, Judge, and Bock (1981), Yancey, Bohrer, and Judge (1982), Gourieroux, Holly, and Montfort (GHM) (1982), Judge and Yancey (1986), Wolak (1987) and Shapiro (1988). In general, the likelihood ratio framework has been used in developing a test statistic and determining acceptance and rejection regions. In order to facilitate the interpretation of these test results in econometric practice, in this paper we evaluate, under a squared error loss (SEL) measure, the risk characteristics of the preliminary test (PT) estimators that evolve when, based on the data at hand, an estimation decision is taken as a result of the outcomes of two one-sided multivariate hypothesis tests.

The error Components regression model is now widely applied in econometics and statistics. Given the potentially high costs of incorrectly excluding the component from the model, an error components test should have high power. In addition, if a test is to gain acceptance from practitioners, the test should be computed easily. To obtain improved critical-value approximations, we introduce a standardized Lagrange multiplier (SLM) test statistic, which is centered and scaled to have a zero mean and unit variance under the null hypothesis. We also examine the F test, which is easily computed and has a well-known exact distribution under the null hypothesis if the regression errors are normally distributed