The authors establish that the usefulness to investors of accounting data has declined during the past 20 years. The authors investigate the statistical relationship between earnings and returns for each of the years from 1977 through 1996. Earnings and the one-year change in earnings explain 6–12 percent of the variability of returns in the first 10 years (1977–1986), decreasing to 4–8 percent in the 1987–96 time period. A regression of the annual R²s on a time variable indicates that the decrease is statistically significant. The authors also examine two other accounting variables—cash flows and book values—and find similar results: a declining usefulness of each over the time period.

The authors apply the same tests to a subset of 1,300 companies that were in the sample for the full time period. Relative to the entire sample, earnings information for the subset had greater explanatory power for each of the sample years, indicating that earnings are more informative for companies with a long operating history. Nevertheless, the explanatory power for this subset experienced a statistically significant decrease over the 20-year period.

The authors hypothesize that an increase in the rate of change of business experienced by a company is related to the decline in accounting usefulness. To measure the rate of change, the authors construct 10 equal-sized portfolios based on value of equity. Whenever a company moved from one portfolio to another, a portfolio switch occurred. The authors construct a measure of portfolio switching that reflects the magnitude and frequency of change. For portfolios established on both book value and market value, the switching increased significantly over the 1977–96 time period.

The authors divide the companies into two portfolios based on portfolio switching: “high-change” and “no-change” portfolios. Tests indicate that the greater the portfolio switching, the greater the decrease in information over time. In fact, the no-change portfolio did not experience any decrease in informational usefulness. R&D appears to be a substantial contributor to the rate of change in that high-change companies experienced a much larger change in R&D intensity than did low-change companies.

The authors submit two proposals aimed at enhancing informational usefulness. Currently, expenditures on intangible assets, such as R&D and employee training, are expensed because of the uncertainty regarding the realization of prospective benefits. Because the passage of time reduces the uncertainty about the success of the project, the authors propose that accountants capitalize expenditures on intangible assets, including all project expenditures previously expensed.

The authors also recommend a systematic restatement of financial reports. As the results of expenditures on intangibles materialize, accountants would restate financial reports to reflect the results. For example, once a project passes a feasibility test, the financial reports would be restated by capitalizing previous expenses. The authors expect that some people will argue that the restated information is no longer relevant to decision making, but the authors see the restatements as functioning much like the GDP revisions and believe that the revision process would provide new information to investors.

The authors conclude that the social consequences of the decline of usefulness of financial information are tied to the ability of investors to obtain the information at no added cost. The authors acknowledge that researchers have not fully explored the social cost resulting from inadequate accounting data; however, the information available indicates that the lack of information does impose costs on investors.