INTRODUCTION
"An Empirical Evaluation of Accounting Income Numbers" written by Ball, R. and Brown, P. was published in the Journal of Accounting Research in 1968, which then received the American Accounting Association's inaugural award for seminal contributions in the accounting literature.
The above mentioned article was the first paper undertaking to study the usability of the accounting income numbers. It was carried out to retort a general belief that income numbers cannot be defined substantively. They lack of meaning and are therefore a doubtful utility, which is a traditional view suggested by the accounting theorists following the reason that normative analytic models generally do not take into account their ability in predicting or explaining observed behaviour. As a result, this is an article about an empirical test to claim that an inadequacy of substantive meaning does not normally mean a lack of utility.
OBJECTIVES
To study on the usefulness of the accounting income numbers.
THEORETICAL FRAMEWORK
It is proposed that information is useful in creating capital asset prices under the assumption that capital markets are efficient and fair. Consequently, asset prices would be adjusted promptly by the market to the new information and are therefore able to avoid further abnormal gain. In other words, the flow of information to the market will always be reflected by changes in security prices if the proposition is true. Thus, a conclusion of income numbers is useful can be made if the observed review of stock prices is connected with the release of income report.
The relationship between accounting income numbers and stock prices is built on this theory. An important concept here is that the difference between the actual change in income and the expected change in income will determine the amount of new information which has impacts on stock prices. Such a difference between both figures is called forecast error, which is presumed to be the new information conveyed by the present income number.
The researchers further stated that stock prices and rates of return from holding stocks tend to move together. This could be explained through an idea that describes the release of information which concerns all firms throughout the industry will trigger market-wide variations in stock returns. Equally important, another notion called economy-wide effects also arises besides market-wide effects. This is a situation which says that part of the above movement or change could be estimated if one has the knowledge of the current economy status as well as the financial position for the past few years of the firm invested by him. Accordingly, the contents and timing of income reports are considerable in the assessment relative to changes in the rates of return on a firm's stocks. In short, results should be the net of market-wide effects and economy-wide effects as individual firm is considered.
As mentioned above, the income forecast error is a reflection of the information flow. When the actual change in income is less than the expected change, it is a negative forecast error. This negative value is defined as bad news which predicts if there is a relationship between accounting income numbers and stock prices, the return on the firm's securities would be less than the expected after the release of income number. This result is proven by a negative behaviour in the stock return residuals around the annual report announcement date. For a positive forecast error, it would be the other way round.
DATA
This research is of interest to three types of data, namely, the contents of income reports, the announcement date for annual reports, and the movements of the security prices around the announcement dates.
The record of income numbers for 1946 until 1966 is collected from Standard and Poor's Compustat tapes. The data is used by the researchers to analyze the reaction of stock market to the release of income numbers as mentioned earlier. Again, the observation concentrates only on the existence of the unpredictable income forecast errors at a minimum of 12 months prior to the announcement dates. That is, market-wide effects as well as economy-wide effects are ignored. However, such an assumption is not applied when the errors are auto-correlated.
In addition, the information in regards to the annual report announcements is gathered from three kinds of sources in the Wall Street Journal. They are the forecast of the year's income, the preliminary reports and the completed annual report. In the opinion of the researchers, forecasts are often inexact, while the preliminary report would usually have the same numbers for Earnings Per Share (EPS) and net income as in the final annual report. Therefore, the presumption made is that the annual income numbers will be available to the public as soon as the preliminary report appeared in the Wall Street Journal.
Last but not least, the data of stock prices are gathered from the monthly closing prices on the New York Stock Exchange starting from the period of January, 1946 until June, 1966. These tapes are constructed by Center for Research in Security Prices (CRSP) at the University of Chicago.
However, there are some issues that arise from the data due to the sample selection criteria which may reduce the generality of the results. Basically, the criteria for data inclusion is limited to only nine fiscal years from 1957 until 1965 with the firms' financial year end on December 31. The failed young firms, those which do not report their financial statements on December 31 and those who are not presented in CRSP tapes as well as Wall Street Journal are not included in the data.
FINDINGS
As pointed out, the purpose of this paper is to identify the impact of the accounting income numbers, and this can be achieved by applying the Abnormal Performance Index (API). The statistics derived from the API has been illustrated in the following figure for better understanding.
Figure: Abnormal Performance Indexes for Various Portfolios
Refer to the figure above. The graph plotted API, that is, the value of an investment of one dollar in all securities against a timeframe of twelve months. Month 0 is defined as the month of the annual report announcement. As a result, month -2 is logically referring to the second month prior to the month of the releasing of annual report, and similarly, month -4 is referring to the fourth month prior to the month of the releasing of annual report, and so on.
Based on the figure, the good news portfolios rise before the announcement made at month 0, while the bad news portfolios are also having the same pattern which it falls before the announcement made at month 0. Above all, there are post announcement drifts in both forms, that is, the API continues to rise or fall after month 0. The latter situation demonstrates that the event of releasing the accounting income numbers at month 0 does give some impacts as there are changes reflected in security prices. Furthermore, such a relationship between these two variables is also verified by the Chi-square test which the values of probability for most of the months up to that of the annual report announcement are more than 0.05.
However, the annual income report hardly regards as a timely means, since the movements in the capital market rise or fall soon after month -12, that is, one year before, instead of a large rise or a large fall around month 0. It is believed that the reason contributes to such circumstances is the accessible of public to more prompt media, such as interim reports, other than the final report.
COMPARISON
Our reading is extended to a research paper called "A Reformulation of the API Approach to Evaluating Accounting Income Numbers" presented by Winsen, J. K. in 1977. According to him, he does not agree with the technique of API metric which was initially formulated by Ball, R. and Brown, P. in examining the relationship between the accounting income numbers and stock prices due to three major weaknesses.
First and foremost, the metric does not utilize all of the information potentially available from accounting numbers. Next, the meaning of the metric has been questioned. Finally, the statistical significance of the API metric and of differences between API's has not always been satisfactorily considered. Winsen then tries to resolve those factors by reformulating the API approach in his research paper.
CONCLUSION
About one-half or more of the information regarding an individual firm which is available during a year is captured in that year's income numbers. Therefore, its content is considerable and this study has met its objective to prove the usefulness of the accounting income numbers.
