James A. Ohlson’s research while affiliated with City University of Hong Kong and other places

This article develops an analytically coherent yet parsimonious framework which explains market returns in terms of contemporaneous information. It anchors on the idea that valuation (static perspective) can be connected to the dynamics that explain returns, and vice versa. The framework requires two components. First, an explicit function that maps information to an estimate of value—a valuation heuristic. Second, the framework assumes that the difference between a firm’s actual value and value-per-heuristic follows an autoregressive stochastic process with a contraction parameter and no intercept. The contraction parameter can be estimated efficiently and nonparametrically. This modeling suffices to derive implied returns. Using scaled Earnings Per Share (``EPS'') forecasts as valuation heuristics, we empirically evaluate the framework’s validity and robustness. Its explanatory power compares favorably to that of traditional ordinary least squares (``OLS'') regressions, despite only requiring a single parameter. In a setting with pooled annual data, the implied and realized returns correlations range between 64% and 73%.

This paper examines the current empirical accounting research paradigm. We ask: In general, do the estimated regressions support the promoted narratives? We focus on a regression model’s main variable of interest and consider the extent to which it contributes to the explanation of the dependent variable. We replicate 10 recently published accounting studies, all of which rely on significant t-statistics, per conventional levels, to claim rejection of the null hypothesis. Our examination shows that in eight studies, the incremental explanatory power contributed by the main variable of interest is effectively zero. For the remaining two, the incremental contribution is at best marginal. These findings highlight the apparent overreliance on t-statistics as the primary evaluation metric. A closer examination of the data shows that the t-statistics produced reject the null hypothesis primarily due to a large number of observations (N). Empirical accounting studies often require N > 10,000 to reject the null hypothesis. To avoid the drawback of t-statistics’ connection with N, we consider the implications of using Standardized Regressions (SR). The magnitude of SR coefficients indicates variables’ relevance directly. Empirical analyses establish a strong correlation between a variable’s estimated SR coefficient magnitude and its incremental explanatory power, without reference to N or t-statistics.

Attendees of accounting empirical research seminars all too often come to view the conclusions presented in the papers as non-persuasive. This disappointing situation indicates that researchers employ data analysis methodologies which inherently support conclusions they are looking for. Such issues are rarely discussed because many participants have relied on the same methodologies – thus they have firsthand knowledge about the inherent deficiencies. The mantra becomes: “We are all aware of uncomfortable aspects of the methodologies used in our research, so why dwell on it?” Because these potential questions tend to be outside normal and acceptable bounds, I term them “elephants in the room”. Five such cases are delineated to illustrate incontrovertible problems therein. To sum it up, the elephants highlight that the purported substantive contents of most published papers will be taken with a grain of salt for the foreseeable future.

This paper examines commonly applied methods of data analysis. Predicated on these methods, the main issue pertains to the plausibility of the studies’ end products, that is, their conclusions. I argue that the methods chosen often lead to unwarranted conclusions: the data analyses chosen tend to produce looked-for null rejections even though the null may be much more plausible on prior grounds. Two aspects of data analyses applied cause obvious problems. First, researchers tend to dismiss “preliminary” findings when the findings contradict the expected outcome of the research question (the “screen-picking” issue). Second, researchers rarely acknowledge that small p-values should be expected when the number of observations runs into the tens of thousands (the “large N” issue). This obviously enhances the chance for a null rejection even if the null hypothesis holds for all practical purposes. The discussion elaborates on these two aspects to explain why researchers generally avoid trying to mitigate false positives via supplementary data analyses. In particular, for no apparent good reasons, most research studiously avoids the use of hold-out samples. An additional topic in this paper concerns the dysfunctional consequences of the standard (“A-journal”) publication process, which tends to buttress the use of research methods prone to false or unwarranted null-rejections.

The paper discusses two issues related to ‘growth’. Both lack a proper recognition in the literature. The first concerns growth and accruals. Research practice combines two components of accruals, asset accruals and liability accruals into a ‘net’. This aggregation causes problems: income-increasing asset accruals correlate positively with growth whereas the opposite holds for income-increasing liability accruals and growth. In light of this property, the paper re-configures Jones’s model. The second issue addresses forward P/E-multiple dependence on future growth in expected eps. It is shown that the textbook Gordon-Williams 1/(r-g) approach makes no sense. An alternative modeling views growth as information – not a parameter like Gordon-Williams. Thus, the P/E-multiple is shown to depend on the growth in expected earnings, Y2 vs. Y1. A similar g-parameter dependency is noted in the residual income model and the so-called OJ earnings growth model; both include a 1/(r-g) type of multiplier. This is viewed as a model-deficiency, that is, a ‘mistake’.

Sales and profit margins are two popular earnings components discussed in the media. We study properties of one-year-ahead analyst forecasts of these two components. As sales are in dollar amounts and profit margin is a ratio, we propose robust statistical methods to assess and contrast their forecast properties. We find that four performance properties associated with earnings forecasts—optimism, relative accuracy with respect to benchmark model forecasts, forecast suboptimality, and serial correlation of forecast errors—apply to both sales and profit margins. Sales forecasts, in general, perform better than profit margin forecasts. Further evidence also shows that sales forecasts perform better than profit margin forecasts in terms of how their forecast errors explain earnings forecast errors and how realized surprises affect adjustments of the respective forecasts. We also find that a better information environment, surrogated by size, improves sales forecasts more than profit margin forecasts. All of these findings suggest that forecasting profit margins is inherently more difficult than forecasting sales.

Recent research, due to Patatoukas and Thomas (2011) and Ball et al. (2013), focuses on Basu's (1997) conditional conservatism measure and the existence of a denominator effect—whether the difference between the earnings-return coefficients of bad and good news firms (“the Basu coefficient”) is only due to the beginning-of-year price deflator. We address this issue head-on by applying the Theil-Sen (TS) estimation method, which obtains the same coefficient estimate regardless of the chosen deflator and is robust to outliers. Results show the following: (i) the Basu coefficient remains positive using TS; (ii) the Basu coefficient using TS are similar to those using OLS without scaling but much smaller than what scaled OLS show; (iii) the scaled OLS estimates appear to be influenced by a few outliers; and (iv) OLS estimates are more volatile due to estimation error. In sum, the denominator effect does not overturn Basu's hypothesis but the magnitude and variation of the Basu coefficient is much smaller than what traditional results show.