AN ANALYSIS OF THE PERMANENT PRICE CHANGE ASSOCIATED WITH COMMON STOCK SPLITS*

Abstract

THIS STUDY is concerned with an analysis of stock price changes over a period of thirteen months surrounding common stock splits. The purpose of the analysis was to determine whether there is a significant price change associated with stock splits, while controlling for changes in dividends, initial dividend level, change in earnings, initial earnings level, and stock market trend. The effect of other concomitant factors was assumed to average out over the 148 stocks in the study. Some published research (for example, Barker) has concluded that the observed price increase accompanying a stock split is a function of an increase in cash dividends rather than the split itself. The results of this study tend to disagree with Barker's conclusion. Analysis of covariance is the technique employed to test the hypothesis that there is no significant difference between the mean price change of the group of split stocks and the group of nonsplit stocks, after controlling for the effect of the above five concomitant variables. The analysis includes all true stock splits of 2: 1 or greater that occurred during the calendar year 1959 among stocks listed on the New York Stock Exchange. For the purposes of this study, a true stock split is defined as a split in which no part of the retained earnings was capitalized. Seventy-four stocks listed on the New York Stock Exchange that did not split their shares in 1959 were selected at random and were then randomly paired with the split stocks. The price change was determined by expressing the average price of the stock in the fifth month after the split date as a percentage of the average price of the stock in the eighth month prior to the split date. These prices and the control variables were adjusted for any change in the number of shares outstanding. The price change of the nonsplit stock that was randomly matched with each split stock was determined for the same time period. The market trend was included as a control variable by matching each stock with an appropriate Standard and Poor's subgroup stock price index. The percentage change in each index was computed over the same time period and in the same manner as the price change. Unlike the change in stock price and index, the change in earnings and dividends could not be expressed in percentage terms because of the presence of deficits and zero values. The change in dividends was incorporated by the use of two control variables. The dividends per share in the quarter preceding the period under study, the initial dividend, provided the first of these control variables. Any change in this dividend rate prior to the end of the thirteen-month period provided the data for the other dividend control variable. The change in earnings was included in the same manner. The quarterly earnings for the last fiscal quarter prior to the period under study provided the initial earnings variable. The actual difference between these earnings and the earnings in the quarter ending one year later provided the data for change in earnings. Tests of the statistical validity of the data for the two-way analysis of covariance design were carried out and satisfied. The F-test of the difference between the mean