Honey, I Shrunk Accounting – The End of Accounting (Wiley, 6/27) Excerpted with permission of the publisher, Wiley, from “The End of Accounting and the Path Forward for Investors and Managers” by Baruch Lev and Feng Gu. Copyright (c) 2016 by Baruch Lev and Feng Gu. All rights reserved. This book is available at all bookstores and online booksellers.
Honey, I Shrunk Accounting
To avoid undue suspense, we will tell you our findings up front, made clear by Figure 3.1. The steady decline of the graph, from over 90 percent in the 1950s (see horizontal scale, spanning the past 60 years) to around 50 percent currently-a fall of almost half-tells vividly the story of the decline in the relevance of corporate financial information to investors. That’s quite a fall from grace of the ubiquitous investors’ information source-accounting and financial reporting.5 You surely wonder: How did we achieve such an accurate measurement of decline in information usefulness? How can information relevance be measured so precisely? Please keep reading. It will be worth your while.
Some Useful Details
First, how do we capture the information content of corporate financial reports extending over 150 to 200 pages? We have to be highly selective, of course. Much of the “information” in these reports, like the pictures of the ever-smiling executives, employees, and customers, or the long discourses on the company’s products and its many do-goodings, is either outright useless for investment decisions or readily available on the Web. Ditto the extended discussion of risk factors and historical stock price data. But then there are scores of seemingly relevant financial items, like revenues, accounts receivable, cost of sales, and earnings. Obviously, not all of them can be captured in a statistical study. And they don’t have to be, if one wisely chooses a few summary measures, reflecting the essence of the financial report.
[drizzle]We chose for our initial study the two most widely used indicators of a company’s operations and economic condition: earnings (net income) and book value, or equity (balance sheet assets minus liabilities). The former reflects the enterprise’s performance during the period-revenues minus all expenses-whereas the latter (net assets) captures its economic position, or net worth at the end of the period: reflecting the company’s underlying external sources of funds (borrowed money and other indebtedness) vis-a-vis the uses of fund (the various assets employed by the enterprise).6 In earnings and book value, we thus have a parsimonious yet representative set of key financial information items, which uniformly ranks at the top of investors’ decision determinants.
Turning to investors’ use of earnings and book values, we focus on companies’ stock prices three months after fiscal-year-end (during which time the annual reports have to be publicly released) to assure that these stock prices impound the most recent information contained in the two financial items we examine. We then statistically relate for each year, over the past 60 years, the market values (the product of stock price and the number of shares outstanding) of all US public companies with the required data to their recent respective earnings and book value (see the Appendix for a more formal discussion of this analysis). Market values (capitalization) of companies reflect, of course, multiple sources of information, such as interest rates, industry conditions (e.g., depressed real estate in the financial crisis), and monetary policy (the Fed’s “quantitative easing”), in addition to companies’ earnings and book values. Accordingly, our statistical methodology (a regression analysis) enables us to answer the following question: Of all the information items reflected in companies’ market values (stock prices), how much is attributed to corporate earnings and book values? This is the message of Figure 3.1: roughly 80 to 90 percent in the 1950s and 1960s versus 50 percent today.
And Now For Some Intuition
To fully appreciate the meaning of the drop to 50 percent in Figure 3.1, one has to intuitively grasp the derivation of these numbers. We mentioned that we applied a regression analysis. But, what is this creature? A regression is a statistical technique, akin to a correlation, relating one variable, or indicator (dependent variable), to a set of other variables (explanatory variables), intended to answer the following question: How much of the variation in the former (different market values of companies in our case) is explained by, or can be attributed to, the set of explanatory variables? In our case, to what extent do companies’ earnings and book values explain their different market values? This question is answered by the regression’s adjusted coefficient of variation, or R2 (henceforth R2), which is depicted on the vertical axis of Figure 3.1. If differences among companies’ market values are mainly attributable to their performance (earnings) and financial situation (book value), then the R2 will be high (close to 100 percent), whereas if other factors are dominant in setting stock prices, the R2 will be low. What Figure 3.1 tells us is that in the 1950s, 1960s, and even 1970s, the key financial report variables, earnings and book values, were critical to investors’ valuation of companies, whereas the usefulness or relevance of these two variables to investors has diminished considerably since then.7
A brief example of a regression analysis in a widely familiar context will solidify your understanding: Suppose medical researchers wish to ascertain the main causes of the different cholesterol levels of people (to understand why yours is so high). They suspect that age, weight, and education level (affecting health awareness) affect cholesterol levels. To determine the impact of these effects quantitatively, the researchers first measure the cholesterol level of members of a sample of, say, 500 persons, and then run a regression of the 500 measured cholesterol levels (akin to our companies’ market values) on the 500 triple measures of each person’s age, weight, and school years (like our earnings and book values). Suppose the measured R2 of this regression is 35 percent. This means that the combined effect of a person’s age, weight, and education level accounts for (or explains) about a third of the (squared) differences in people’s cholesterol levels, implying that almost two-thirds of the determinants of cholesterol level are still unknown to the researchers. The search for additional cholesterol determinants, like food intake or parents’ cholesterol level, should go on. Now you are a statistical maven and appreciate our empirical finding: The role that earnings and book values-the key financial indicators-play in securities valuation dropped by almost 50 percent during the past half century.
Who’s The Culprit – Earnings Or Book Values?
Figure 3.1 reflects the joint relevance-loss of earnings and book values. Since accounting standard setters sometimes change their emphasis from the balance sheet to the income statement (focusing on income measurement as the primary objective of accounting) and vice versa (emphasizing the valuation of assets and liabilities over earnings), it’s instructive to examine separately the change over time in the relevance of earnings, the all-important “bottom line” of the income statement, and that of the book value, reflecting the balance sheet information concerning the company’s assets and liabilities.
Using the same research methodology previously described, Figures 3.2