James A. Ohlson

FINANCIAL RATIOS AND THE PROBABILISTIC PREDICTION OF BANKRUPTCY

This paper presents some empirical results of a study predicting corporate failure as evidenced by the event of bankruptcy. There have been a fair number of previous studies in this field of research; the more notable published contributions are Beaver [1966; 1968a; 1968b], Altman [1968; 1973], Altman and Lorris [1976], Altman and McGough [1974], Altman, Haldeman, and Narayanan [1977], Deakin [1972], Libby [1975], Blum [1974], Edmister [1972], Wilcox [1973], Moyer [1977], and Lev [1971]. Two unpublished papers by White and Turnbull [1975a; 1975b] and a paper by Santomero and Vinso [1977] are of particular interest as they appear to be the first studies which logically and systematically develop probabilistic estimates of failure. The present study is similar to the latter studies, in that the methodology is one of maximum likelihood estimation of the so-called conditional logit model. The data set used in this study is from the seventies (1970-76). I know of only three corporate failure research studies which have examined data from this period. One is a limited study by Altman and McGough [1974] in which only failed firms were drawn from the period 1970-73 and only one type of classification error (misclassification of failed firms) was analyzed. Moyer [1977] considered the period 1965-75, but the sample of bankrupt firms was unusually small (twenty-seven firms). The

Aggregate accounting earnings can explain most of security returns: The case of long return intervals

Abstract The paper analyzes the contemporaneous association between market returns and earnings for long return intervals. The research design exploits two fundamental accounting attributes: (i) earnings aggregate over periods, and (ii) expanding the interval over which earnings are determined, is likely to reduce ‘measurement errors’ in (aggregate) earnings. These concepts lead to the level of (aggregate) earnings as a natural earnings variable for explaining security returns. We hypothesize that the longer the interval over which earnings are aggregated, the higher the cross-sectional correlation between earnings and returns. The empirical findings support this hypothesis.

Volatility increases subsequent to stock splits: An empirical aberration

Abstract This paper analyzes the empirical behavior of stock-return volatilities prior to and subsequent to the ex-dates of stock splits. The evidence demonstrates rather unambiguously that there is, on the average, an approximately 30% ‘arbitrary’ increase in the return standard deviations following the ex-date. The increase holds for both daily and weekly data, and it is not temporary. No explanatory confounding variables, such as institutional frictions affecting price observations, have been identified. We view the findings as being essentially inconsistent with the notion of ‘rational pricing’.

Earnings, Book Values, and Dividends in Equity Valuation: An Empirical Perspective*

This paper revisits Ohlson 1995 to make a number of points not generally appreciated in the literature. First, the residual income valuation (RIV) model does not serve as a crucial centerpiece in the analysis. Instead, RIV plays the role of condensing and streamlining the analysis, but without any effect on the substantive empirical conclusions. Second, the concept of “other information†in the model can be given concrete empirical content if one presumes that next†period expected earnings are observable.

On Transitory Earnings

The paper develops a concept of transitory earnings and contrasts this source of earnings to “core” (or recurring) earnings. It is shown that any two of the following three attributes of transitory earnings imply the third: (i) forecasting irrelevance with respect to next-period aggregate earnings, (ii) value irrelevance, and (iii) unpredictability. The paper makes the case that the current “dirty surplus” items make sense, especially if one expands the valuation perspective to also allow for agency considerations.

Disaggregated Accounting Data as Explanatory Variables for Returns

Modem market-based accounting research rarely refers to the elaborate concepts dictating the preparation of financial reports. The research does not explicate the reasons for the relevance of accounting data in valuation, and it depends only on the broad notion of “accounting data facilitates investors’ assessments of firms’ future cash flows. ” The popular construct of unexpected earnings, which researchers employ to explain returns, by contrast, is irrelevant to the practicing accountant who implements accounting principles. A more traditional way of looking at accounting recognizes the process as one of measurement. That is, the analysis of transactions leads to line items in the financial statements, which in turn aggregate into the “bottom line” numbers: (net) earnings and book value (net worth). These two summary measures achieve preeminent status by serving as primary indicators of a firm’s value. However, the disclosures of the line items clearly suggest that the accountant is aware of the insufficiency of earnings and book values as determinants of values. This paper incorporates a measurement perspective in addition to an information perspective to understand how accounting data relate to security returns.’ Earnings (and book values) derive from line items, and these line items may have differential valuation implications because investors perceive differential measurement errors. Hence, aggregation is not generally satisfied unless the measurement errors in the line items are relatively insignificant. The latter could occur for long reporting periods, as is suggested by the notion that a firm’s lifetime earnings are measured without error. In any event, the language of accounting provides some straightforward hypotheses concerning the realizations of line items and returns. Revenues are ‘‘good,’’ whereas expenses are “bad,” and returns should respond accordingly. More-

Portfolio Selection in a Lognormal Market When the Investor Has a Power Utility Function

Multiasset portfolio selection models stated in terms of the expected utility criterion generally require the evaluation of multiple integrals. This reality has severely hindered attempts towards the development of computation methods to determine optimal portfolio allocations when there are a large number of assets. Aside from special cases, expected utility is not convergent into a simple closed form; the complexity from the point of view of computation is then perhaps most easily appreciated if one realizes that every iteration in a nonlinear program demands the estimation of several integrals (see Ziemba [23] for details). Such calculations are extremely costly when the number of assets is large. It is, consequently, of interest to approximate the expected utility function by a function which is easier to optimize over the set of feasible portfolios.

The Stationary Distribution of Returns and Portfolio Separation in Capital Markets: A Fundamental Contradiction

In a general equilibrium model of risky assets, prices would be determined by the interaction of the supply and demand. Unpredictable events would impact one or both sides of the asset market and thereby influence the return on assets. The probability distribution of returns would thus be endogenous. Since expectations as to subsequent asset returns influence asset demand, the probability distribution of returns is a crucial element in the general equilibrium system; it is the consequence of the demand and supply of assets and, at the same time, a central determinant of expectations and, hence, of the demand for assets. Nevertheless, it is possible to undertake a partial equilibrium analysis in which the behavior of asset demand, conditional upon a postulated probability distribution of returns, is examined. In such a limited context, it is natural to postulate a “convenient†probability distribution of returns, namely, one that facilitates the analysis of demand. A leading assumption has been that returns are serially independent and obey a stationary distribution.

Risk, Return, Security-Valuation and the Stochastic Behavior of Accounting Numbers

There is a considerable body of empirical research in accounting devoted to the analysis of relationships between accounting numbers and security prices. Very roughly, this body of research may be classified into three different categories: (i) share price valuation models and the determination of market equity values; (ii) the measurement of “unexpected earnings†and their contemporaneous association with security returns; (iii) the forecasting of future security returns. The selection and definition of accounting numbers in most of these types of studies have, by and large, been quite heuristic. The accounting variables are usually selected with little consideration given to their empirical time-series behavior; more important appears to be their intrinsic economic connotations. The approaches can thus be thought of as stipulating the existence of “real†economic variables, e.g., real income for a period, and then using numbers of published accounting statements as estimates of the real variables. The errors in estimates of the true variables are then often minimized by the use of aggregation procedures and the diversification effects of such procedures. The postulating of real economic variables has another methodological advantage: it permits the use of comparative statics analysis of corporate behavior and its effect on equity risk and return. For example, Hamada [7], among others, has shown that leverage affects risk in the usually hypothesized manner but this analytical result depends on the assumption that leverage and earnings are real and unambiguous economic variables without “measurement†errors.

On financial disclosure and the behavior of security prices

Abstract The paper addresses the question: what impact does a change of a firms information-disclosure environment have on its stock price behavior? Assuming that disclosure does not have any effects on production financing programs, that the CAPM obtains, and information can be characterized as ‘linear’, a number of results are provided. It is shown that disclosure leads to increased variability in stock price; the result is consistent with several empirical studies [Beaver (1968), May (1971), and Patell and Wolfson (1979)]. Further, the price that would be expected in a richer information environment is simply equal to the price that actually obtains in a less rich environment. Finally, risk and return parameters are essentially independent of changes in the disclosure environment.