Is stock price a good measure for assessing value-relevance of earnings?

Is stock price a good measure for assessing value-relevance of earnings?
An empirical test

Alex Dontoh
New York University
Stern School of Business
40 W. 4th St., Room 300
New York, NY 10012
E-Mail: adontoh@stern.nyu.edu

Suresh Radhakrishnan
University of Texas at Dallas
School of Management
2601 N. Floyd Road
Richardson, Texas 75083-0688

E-Mail: sradhakr@utdallas.edu
Joshua Ronen
New York University
Stern School of Business
40 W. 4th St., Room 300
New York, NY 10012
E-Mail: jronen@stern.nyu.edu
October 1999

Introduction
Recently, a growing body of literature has created a widespread impression that financial
statements have lost their value-relevance because of shift from traditional capital-intensive economy into a high-tech, service-oriented economy. In particular, the claim is that financial statements are less relevant in assessing the fundamental value of high-tech, service-oriented firms/activities, which are by nature knowledge-intensive (see “Jenkins Committee” report of the AICPA Special Committee On Financial Reporting; Elliott and Jacobsen, 1991, Jenkins, 1994, Reimerman, 1990, and Sever and Boisclair, 1990).
In the academic literature, a few studies have documented evidence on the steady decline in the value-relevance of earnings over time (see e.g., Ramesh and Thiagarajan, 1995). Noting that in recent years firms have become increasingly likely to report negative earnings and nonrecurring items, Collins, Maydew, and Weiss (1997) report that when book values are added as independent variables, along with earnings, the value-relevance of earnings and book value
combined do not decline over time. Rather, they find that the incremental value-relevance of earnings (book value) declines (increases) in the frequency of non-recurring items and of negative earnings. These findings prompt the authors to suggest that claims that the
conventional historical cost accounting model has lost its value relevance are premature.
Brown, et. al., (1998), however, argue that a scale factor common to price per share, EPS,and
book value per share induces a spurious increase in R2 over time. After controlling for the scale, they find that incremental R2s of both earnings and book value have, in fact, declined over time. Chang (1998) uses an alternative (to R2) measure of value-relevance -- the natural log of the ratio of predicted value based on earnings and book value to price. He also finds the value-relevance of accounting information (as summarized in earnings and book value) to October 7, 1999 decline over time. This alternative measure may in fact, be controlling for the same scale effects isolated by Brown et al. (1998.)
Liu & Thomas (1998) find that when earnings are introduced along with relevant
information about future earnings as independent variables, the explanatory power of variations in unexpected stock returns is significantly improved. This finding casts doubt on the purported decline of the value relevance of earnings. However, the authors themselves assert that although both current earnings and future earnings forecasts reflect accounting information, the finding does not reflect the extent to which the analysts’ forecast revisions are derived from information in current accounting reports. Moreover, the authors point out that it is not easy to infer valuerelevance from simple valuation regressions if other accounting information that is valuerelevant
is omitted.
In this paper we address the information content of earnings, and not their value
relevance as has been defined in these empirical studies. As noted above, value-relevance is defined in terms of the association of earnings with market values, returns, or unexpected returns. We define the information content of accounting earnings, intuitively, as the degree to which earnings change the market's expectation of the fundamental value of the firm. We derive empirical predictions obtained from examining a noisy rational expectations model that features three rounds of trading and two dates on which a public announcement (accounting earnings) is issued. The analysis shows how the value-relevance benchmark, i.e., the stock price, returns or unexpected returns, is influenced by other forces in the economy (other than those impinging directly on the fundamental value of the firm). Specifically, we show that the information content of stock prices is smaller than the information content of the accounting earnings, with the gap increasing in the level of non-information based (NIB) trading. NIB trading may be induced by October 7, 1999

many factors such as global wealth transfers, activities of speculative day traders, etc. Intuitively, this finding implies that while the information content of earnings may have declined, the information content of stock prices, which is used as the benchmark for value-relevance could have declined even further.
To assess the information content of earnings and stock prices, we regress current
period’s earnings and stock prices on the discounted future period’s earnings and a terminal value. Thus, the empirical specification captures the information content as operationalized in the theoretical model. Specifically, the information content of earnings (stock prices) is the predictive ability of current period’s earnings (stock prices). Intuitively, this empirical specification is attractive also because it allows us to examine whether stock prices contain information about the future in addition to information that is contained in earnings.
We find that the predictive content (R2) of the earnings regression is higher than the
predictive content (R2) of the stock price regression. That is, prices are not a good benchmark for assessing the information content of earnings. And, while the information content of earnings declined over time, the information content of prices declined even more, thus making them (the prices) even poorer benchmarks over time. We purge the effects of the information contained in current period’s earnings from the information contained in current period’s stock prices, so that we can assess whether the information content of stock prices from sources other than earnings has increased over time. We find that the R2 of this regression also declines over time. This suggests the reason for the declining information content over time is not due to the decline in information content of earnings.
More interestingly, we find that the decline in the information content of stock prices
over time is more pronounced for small-sized firms than for large-sized firms. Specifically, the October 7, 1999
ratio of the earnings regression R2 to the stock price regression R2 is almost flat for the large size firms, while for the small-sized firms the ratio has arisen considerably. Similarly, the ratio of the earnings regression R2 to the stock price regression R2 is almost flat for the low book-to-market ratio (low-intangible-intensity), while for the high book-to-market ratio (high-intangibleintensity)
the ratio has arisen. Other sensitivity tests also suggest that the reason for the declining
information content of stock prices over time is not due to the decline in the information content of earnings.
Some economy-wide factors possibly affecting changes in the value-relevance of
earnings were examined in Collins et. al. (1997). We investigate in this study , as well, these possible causes for the decline of the information content of earnings and stock prices by examining the association of the declining R2 with the annual cross-sectional mean loss, mean one-time items and mean intangible intensity of the economy. Similar to Collins et. al. (1997), we find that these factors are associated with the decline in the information content of prices and earnings. Specifically, the mean loss and the mean one time items are negatively associated with the R2 while the mean intangible intensity is not associated with the R2.
The empirical findings and the theoretical model suggest that an increase in NIB trading
over time could degrade the information content of stock prices. That is, NIB traders and other speculators who trade without considering information regarding the fundamental value, create such noise in prices as would attenuate their (the prices') ability to incorporate the content imbedded in earnings and in other sources of information. This results in a loss in the predictive ability of prices. This suggests that the relative predictive ability of earnings (relative to the predictive ability of prices) increases over time. To investigate this, we measure the annual crosssectional mean trading volume and the standard deviation of trading volume as measures of NIB
October 7, 1999
trading. We find that the increase in the mean trading volume over time is associated with the decline in the information content of stock prices. Furthermore, the variables found to be significant in earlier studies, such as the mean one time items and mean loss no longer explain the decline in information content once the mean trading volume is included in the association tests. This suggests that the increase in trading volume, which is consistent with increased NIB trading may have contributed to the decline in the predictive abiliTy (information content) of
stock prices.
An important implication of our finding is that information content cannot be inferred
from association between market value/returns and earnings, since the information content (predictive ability) of market values or returns is inferior to that of earnings and is in fact declining over time for a variety of reasons.