Abstract. This paper reports an investigation of some data and method effects on the predictive accuracy of LIFO/FIFO classification models. The methods compared were probit, ID3, and neural networks. Experiments were conducted to study the effect of data characteristics on classification accuracy and the situations under which a particular method performs better. Hold‐out samples were used to calculate the predictive accuracy. The results indicate that (1) different methods identify different factors that affect the LIFO/FIFO choice and (2) in hold‐out tests, neural network models have the highest average predictive accuracy, whereas ID3 models have the lowest. Neural network models are the best when dominant nominal variables are present; otherwise, probit models are the best. Résumé. Les auteurs rapportent les résultats d'une analyse de l'incidence de certaines données et de certaines méthodes sur le pourcentage de prévisions exactes dérivées des modèles de classification selon l'épuisement à rebours et l'épuisement successif. Ils comparent la méthode probit, la méthode ID3 et la méthode des réseaux neuronaux et procèdent à des expériences destinées à l'étude de l'incidence des caractéristiques de certaines données sur ce pourcentage et des situations dans lesquelles une méthode particulière donne de meilleurs résultats. Les auteurs ont recours, pour la démonstration, à des échantillons à partir desquels est calculé le pourcentage de prévisions exactes. Les résultats révèlent que 1) les facteurs influant sur le choix de l'épuisement à rebours ou de l'épuisement successif diffèrent selon la méthode utilisée et que 2) dans les tests ayant servi à la démonstration, les modèles de réseaux neuronaux présentent le meilleur pourcentage moyen de prévisions exactes, alors que les modèles ID3 ont le plus faible pourcentage de prévisions exactes. Les modèles de réseaux neuronaux donnent les meilleurs résultats lorsqu'il y a des variables nominales dominantes, faute de quoi les modèles probit sont les meilleurs.

At the microeconomic level, this paper revises and broadens the theory of the equilibrium rate of unemployment, the "natural rate" in a monetary model. We begin by recreating Salop's turnover model of the natural rate in its naturally intertemporal version. Useful findings on impact effects and the adjustment process at the individual firm, necessarily excluded by the static version, are shown to derive from the dynamized model. At the macroeconomic level, we then provide a general-equilibrium analysis of some shocks showing how they drive the equilibrium unemployment rate and in varying ways also disturb the real rate of interest.

We develop a private-information model of union contract negotiations in which disputes signal a firm's willingness to pay. Previous models have assumed that all labor disputes take the form of a strike. Yet a prominent feature of U.S. collective bargaining is the holdout: negotiations often continue without a strike after the contract has expired. Production continues with workers paid according to the expired contract. We analyze the union's decision to strike or hold out and highlight its importance to strike activity. Strikes are more likely to occur after a drop in the real wage or a decline in unemployment.

This study examines whether security analysts underreact or overreact to prior earnings information, and whether any such behavior could explain previously documented anomalous stock price movements. We present evidence that analysts' forecasts underreact to recent earnings. This feature of the forecasts is consistent with certain properties of the naive seasonal random walk forecast that Bernard and Thomas (1990) hypothesize underlie the well-known anomalous post-earnings-announcement drift. However, the underreactions in analysts' forecasts are at most only about half as large as necessary to explain the magnitude of the drift. We also document that the “extreme” analysts' forecasts studied by DeBondt and Thaler (1990) cannot be viewed as overreactions to earnings, and are not clearly linked to the stock price overreactions discussed in DeBondt and Thaler (1985, 1987) and Chopra, Lakonishok, and Ritter (Forthcoming). We conclude that security analysts' behavior is at best only a partial explanation for stock price underreaction to earnings, and may be unrelated to stock price overreactions.

ABSTRACT This paper analyzes how the daily opening and closing of financial markets affect trading volume. We model the desire to trade at the beginning and end of the day as a function of overnight return volatility. NYSE data from 1933–88 indicate that closing volume is positively related to expected overnight volatility, while volume at the open is positively related to both expected and unexpected volatility from the previous night. We interpret the symmetric response of trading at the open and the close to expected volatility as being due to investor heterogeneities in the ability to bear risk when the market is closed. This desire of investors to trade prior to market closings indicates a cost of mandating marketwide circuit breakers.

We adapt a contingent claims model of the firm to reflect the incentive effects of the capital structure and thereby to measure the agency costs of debt. An underlying model of the firm and the stochastic features of its product market are analyzed and an optimal operating policy is chosen. We identify the change in operating policy created by leverage and value this change. The model determines the value of the firm and its associated liabilities incorporating the agency consequences of debt.

ABSTRACT Many firms with dividend reinvestment plans also allow their shareholders to voluntarily invest supplemental funds to purchase additional shares. The purchase price for newly‐issued shares often is determined by the average stock price over a prespecified time period preceding the investment date. This gives the firm's shareholders an option to invest in additional shares only when the stock price exceeds the computed average. This paper uses both theoretical and numerical methods to analyze the value of these voluntary purchase options in theory and practice.

The authors develop a two-factor general equilibrium model of the term structure. The factors are the short-term interest rate and the volatility of the short-term interest rate. The authors derive closed-form expressions for discount bonds and study the properties of the term structure implied by the model. The dependence of yields on volatility allows the model to capture many observed properties of the term structure. The authors also derive closed-form expressions for discount bond options. The authors use Hansen's generalized method of moments framework to test the cross-sectional restrictions imposed by the model. The tests support the two-factor model.