Discussion of Analysis of the Usefulness of Accounting Data for the Portfolio Decision: A Decision-Theory Approach

Abstract

Let me start out by saying that Ohlson's research is addressed to a problem which I consider to be of major importance to accounting and to accountants: the question of whether accounting information is useful to investors. In view of my personal and professional prejudice, I am also pleased by the conclusion reached; namely, that it is indeed useful (and this in contrast to some other empirical findings, e.g. Ball and Brown1). I am depressed by the extreme pessimism expressed by my colleague, George Benston who suggests2 that financial statements may only serve to confirm information already available to investors from other sources. It is also clear that a lot of work has gone into the paper-both in terms of thought given to the theoretical framework, and the effort that has gone into carrying out the computations for the empirical tests. For these reasons, it is only with deep regret that I find myself unable to accept the results of Ohlson's research as justifying the conclusions he wants to draw from it. I shall discuss only a number of issues which lead me to this rejection of the conclusion. What was Ohlson's experiment? Put into the most simplistic terms, an individual risk-averse investor selecting an optimal portfolio wants to form expectations about stocks (and portfolios). He has the choice between: (a) a naive model, using no information (Model 3); (b) a formula based on market information only (Model 2); or (c) a formula incor-