Knowledge that Transforms

This study establishes and tests, within the framework of a noisy rational expectations equilibrium model, the existence of a formal linear relationship between security prices, the average (consensus) and the dispersion of agents' expectations. Variations in the average and in the dispersion of agents' expectations, measured by the earnings forecasts produced by financial analysts, which are gathered and made available by The Institutional Brokers Estimate System (I/B/E/S) , have respectively a positive and negative effect on security prices. The difficulties raised by this estimation, as well as the institutional dimensions of the financial analysis industry are examined. The main results are the following: (1) the most important changes in consensus (in absolute value) correspond to the most important changes in dispersion in the analysts' forecasts, (2) the changes in the consensus and the dispersion of forecasts are respectively positively and negatively linked to Canadian security returns, but given the delay between the production and the public availability of the forecasts, an important part of the price adjustment occurs before the disclosure of forecast changes, (3) the effect on security returns of variations in the consensus dominates the effect of variations in the forecasts' dispersion. Thus, it seems that the impact of information arrival on security prices does not only depend on the direction and the magnitude of the expectations' average revision, but also depends on the direction and the magnitude of the change in the expectations' dispersion.

The study examines how the risk of exhausting corporate tax liabilities before deducting interest expense affects corporate leverage. It differs from prior studies in three ways: (1) it uses data compiled by the Internal Revenue Service (IRS) from corporate tax returns rather than accounting data; (2) it measures risk of tax exhaustion more accurately; and (3) it adopts a first‐difference time‐series approach, so that firms act as their own control between adjacent years. These methodological innovations reduce biases caused by measurement error and omitted variables that were present in prior research. The results suggest that, all else being equal, high risk of tax exhaustion reduces firms' use of leverage. As well, the study provides the first evidence that personal taxes significantly affect corporate leverage. The effects on leverage decisions of other variables are also tested and the results are consistent with predictions from prior theoretical work.

This study investigates coefficient bias and heteroscedasticity resulting from scale differences in accounting levels‐based research designs analytically and using simulations based on accounting data. Findings indicate that including a scale proxy as an independent variable is more effective than deflation at mitigating coefficient bias, even if the proxy is 95 percent correlated with the true scale factor. In fact, deflation can worsen coefficient bias. Also, deflation often does not noticeably reduce heteroscedasticity and can decrease estimation efficiency. White (1980) standard errors are close to the true ones in regressions using undeflated variables. Replications of specifications in three recent accounting studies confirm the simulation findings. The findings suggest that when scale differences are of concern, accounting researchers should include a scale proxy as an independent variable and report inferences based on White standard errors. Résumé. Les auteurs examinent, tant sur le plan analytique qu'au moyen de simulations basées sur les données comptables, la distorsion des coefficients et l'hétéroscédasticité résultant des différences d'échelle dans les plans de recherche comptable basés sur les niveaux. Leurs constatations révèlent que l'inclusion d'un substitut d'échelle à titre de variable indépendante est plus efficace que la déflation pour atténuer la distorsion relative au coefficient, même si le substitut présente une corrélation de 95 pour cent avec le véritable facteur d'échelle. En fait, la déflation peut accentuer la distorsion relative au coefficient. Aussi, il arrive souvent que la déflation, sans réduire de façon appréciable l'hétéroscédasticité, puisse diminuer l'efficience de l'estimation. Les erreurs‐types de White (1980) se rapprochent des erreurs véritables dans les régressions faisant appel à des variables non déflatées. La répétition des mêmes caractéristiques dans trois études comptables récentes confirme les résultats de la simulation. Les conclusions de l'étude donnent à penser que lorsque les différences d'échelle sont sujet de préoccupation, les chercheurs en comptabilité devraient faire intervenir un substitut d'échelle à titre de variable indépendante et formuler les inférences à partir des erreurs‐types de White.