Recent papers by Lamoureux and Poon (1987) and Brennan and Copeland (1988) document a significant permanent increase in average beta subsequent to stock split ex-dates. This paper demonstrates that the shift in estimated beta following ex-dates decays as the measurement interval is lengthened. There is no statistically significant difference between pre- and post-split betas using the Scholes-Williams (1977) estimator and weekly return data, or using monthly returns. We conclude that Lamoureux and Poon's and Brennan and Copeland's results can be attributed to a bias created by using too short a return measurement interval to estimate beta.

[A testable single-beta model of asset prices is presented. If state variables have a long-run stationary joint density function, then the rate return on a very long-term default-free discount bond will be perfectly correlated with the representative investor's marginal utility of consumption. Thus, the covariance of an asset's return with the return on such a bond will be an appropriate measure of the asset's riskiness. The model can be, therefore, applied or tested even though the market portfolio or aggregate consumption may not be observable. It also is shown that the expected rate of return on a very long-term bond is equal to its variance. This proposition can be tested to determine whether state variables follow stationary processes.]

We consider the question, “Under what conditions is the extremum of a random function over a random set itself a random object?” The answer is relevant to problems in both game theory and econometrics, as we illustrate with examples. Our purpose here is to bring the powerful tools of the theory of analytic sets as developed by Dellacherie and Meyer (1978) to the wider attention of the economics profession and to distill Dellacherie and Meyer's work in such a way as to provide some readily accessible theoretical results that will permit relatively easy treatment of economically or econometrically relevant applications.

This article uses aggregate birth year/calendar year level data derived from the Current Population Survey (CPS) to estimate the effect of Social Security wealth on the labor supply of older men in the 1970s and 1980s. The analysis focuses on measuring the impact of the 1977 amendments to the Social Security Act, which created a substantial, unanticipated reduction in Social Security wealth for individuals born after 1916. This differential in benefits has become known as the benefit notch. Results indicate that labor supply continued to decline for the "notch babies" who received lower Social Security benefits than earlier cohorts.

Many accounting research problems involve classification of observations into discrete categories. A number of statistical techniques are used in accounting research involving discrete categories. This study examines the performance of seven techniques that can be used when there are more than two discrete categories. The techniques are linear discriminant analysis, quadratic discriminant analysis, McKelvey and Zavoina n‐chotomous probit, Walker and Duncan ordinal logit, Nerlove and Press polytomous logit, ordered classification trees, and unordered classification trees. Technique performance is measured using classification accuracy. The study finds that the Walker and Duncan ordinal logit, Nerlove and Press polytomous logit, and linear discriminant analysis techniques have the highest performance. The study also finds that the theoretical assignment rule for McKelvey and Zavoina n‐chotomous probit produces lower classification accuracy than an assignment rule based on maximum probability. Résumé. De nombreux problèmes de recherche comptable exigent la classification des observations en catégories discrètes. Maintes techniques statistiques qui font intervenir des catégories discrètes sont utilisées en recherche comptable. L'auteur examine le rendement de sept techniques pouvant être utilisées en présence de plus de deux catégories discrètes. Ces techniques sont l'analyse discriminante linéaire, l'analyse discriminante quadratique, le probit n‐chotomique de McKelvey et Zavoina, le logit ordinal de Walker et Duncan, le logit polytomique de Nerlove et Press, les arbres de classification ordonnée et les arbres de classification non ordonnée. Le rendement de ces techniques est évalué en fonction de la précision de la classification. L'auteur conclut que le logit ordinal de Walker et Duncan, le logit polytomique de Nerlove et Press et l'analyse discriminante linéaire sont les techniques qui ont le rendement le plus élevé. Il en vient également à la conclusion que la règle d'affectation théorique pour le probit n‐chotomique de McKelvey et Zavoina offre une précision moindre sur le plan de la classification qu'une règle d'affectation basée sur la probabilité maximum.

This paper examines problems in the use of divergence of analyst opinion as a proxy for estimation risk in empirical studies of security returns and asset pricing models. We demonstrate that diversity of opinion can increase even though the amount of private information increases, and we show that diversity of opinion may overstate estimation risk if the capital market aggregates the information held by investors. We produce empirical results consistent with our conclusions. Specifically, we find that divergence of opinion can produce measures of estimation risk that are inconsistent with a received proxy for estimation risk and with observed common stock returns.

This study considers the welfare effects of line‐of‐business (LOB) reporting by firms operating in imperfectly competitive markets. The analysis is based on a two‐period model in which a multisegment firm, labeled the incumbent, operates as a monopolist in two markets in the first period and then competes against an entrant in each of the two markets in the second period. The incumbent obtains private information in the first period. LOB reporting fully reveals that information to the entrants; aggregate reporting only partially reveals the incumbent's information. The second‐period consequences on firms and consumers parallel results previously available from the information‐sharing literature. These results serve to explain why some firms opposed the imposition of LOB reporting by accounting rule‐making bodies. The incumbent may distort first‐period production in an attempt to influence entrant beliefs. However, such distortions lower the incumbent's expected first‐period profit and result in no gains in the second period. Nonetheless, in equilibrium, these distortions cannot be avoided unless the entrants observe the incumbent's first‐period production or prices, suggesting a role for nonfinancial disclosures in accounting reports. Résumé. Les auteurs se penchent sur l'effet «protecteur» de la communication d'informations relatives au secteur d'activité par les entreprises qui exercent leurs activités sur des marchés imparfaitement concurrentiels. L'analyse est fondée sur un modèle comportant deux périodes, dans lequel une entreprise oeuvrant dans plusieurs secteurs d'activité‐désignée sous l'appellation d'entreprise installée—exerce un monopole dans deux marchés au cours de la première période et doit ensuite livrer concurrence à un nouveau venu dans chacun des deux marchés au cours de la seconde période. L'entreprise installée obtient de l'information privilégiée au cours de la première période. L'information sectorielle livre intégralement cette information aux nouveaux venus, mais l'information consolidée ne révèle qu'une partie de l'information dont dispose l'entreprise installée. Les résultats de la seconde période pour les entreprises et les consommateurs s'apparentent à ceux qui ont été exposés jusqu'à maintenant dans les écrits relatifs au partage d'informations. Ces résultats viennent expliquer pourquoi certaines entreprises se sont opposées à ce que les organismes de réglementation comptable fassent de la présentation de l'information sectorielle une obligation. L'entreprise installée peut «biaiser» l'information relative à la première période dans l'intention d'influer sur les convictions des nouveaux venus. Ces distorsions diminueront cependant le profit espéré par l'entreprise installée au cours de la première période et se traduiront par des gains nuls au cours de la seconde période. Néanmoins, en situation d'équilibre, ces distorsions ne peuvent être évitées, à moins que les nouveaux venus n'observent la production ou les prix de l'entreprise installée au cours de la première période, ce qui permet de croire que la présentation d'informations non financières dans les rapports comptables a un rôle à jouer.