DISCRIMINATING BETWEEN COMPETING HYPOTHESES
THE COMPETING HYPOTHESES The No-Effect Hypothesis Accounting change has no implication for stock prices Unless it had implication for taxes Accounting be unexpected by the market
THE COMPETING HYPOTHESES The Mechanistic hypothesis Changes in accounting procedures affect stock prices even if those changes have no effect on the firm’s cash flow Assumption: accounting reports are the sole information on the firm
DISCRIMINATING BETWEEN HYPOTHESIS 3 SET OF ACCOUNTING CHANGES: All accounting changes whether they affect taxes or not Accounting changes that do not affect taxes Accounting changes that affect taxes
Kaplan & Roll (1972) Investigate accounting changes that do not affect taxes Investment tax credit: Abnormal rates of return >0 for change sample and = 0 for control sample Support the mechanistic hypothesis rather than no-effect hypothesis
Kaplan & Roll (1972) Depreciation switchback: Abnormal rates of return = 0 Support no-effect hypothesis Contradict with result for investment tax credit
Methodological Issues in Kaplan & Roll (1972) Study Specification of the hypothesis tested The concentration is on the competing hypothesis – the hypothesis that the market is misled Specifying the mechanistic hypothesis Kaplan & Roll (1972) fail to make powerful test when they use two tail test on the abnormal returns Should be one tail: H0 = e0 ≤ 0, Ha: e0 > 0
Methodological Issues in Kaplan & Roll (1972) Study Specification of the hypotheses tested Specifying the mechanistic hypothesis Another way to increase the power of test is to calculate the earnings effect of the accounting change and then investigate the abnormal returns associated with the largest effects Kaplan & Roll use the 1st annual earnings number, which is not necessarily the first earnings disclosed using the change accounting method
Methodological Issues in Kaplan & Roll (1972) Study Specification of the hypotheses tested Testing the no-effect hypothesis The acceptance of EMH led early accounting researchers to accept readily no-effect hypothesis when it was not rejected There are infinite number of alternative hypotheses to the null hypothesis
Methodological Issues in Kaplan & Roll (1972) Study Specification of the hypotheses tested Event study methodology Clustering: investment tax credit changes occur in 1 year and depreciation switchback occur predominantly in 3 year and over represented in several industries Selection bias
Ball (1972) Investigated all types of accounting changes
Sunder (1973, 1975) Test the mechanistic hypothesis Result: Risk increases contemporaneously with a switch to LIFO Observe abnormal price increases associated with switches to LIFO
Ricks (1982) Consistent with mechanistic hypothesis: the market appear to be misled by the drop in earnings caused by the LIFO change Methodological problem: Clustering of changes in 1974 Self selection bias
Biddle & Lindahl (1982) Results: Methodological problem: clustering The larger the tax saving, the larger the change in value of the firm The larger the unexpected earnings, the higher the value of the firm Consistent with EMH Methodological problem: clustering
ECONOMIC CONCEQUENCES
OVERVIEW Economic consequences A concept that asserts that, despite the implications of efficient securities market theory, accounting policy choice can effect firm value
THE RISE OF ECONOMIC CONSEQUENCES Zeff (1978): Economic consequences: the impact of accounting reports on decision-making behavior of business, government, and creditors