Economic subjects | Investments, Stock exchange » Hahn-Jacobs - Dividend Change and Longterm Performance of Utilities

Source: http://www.doksinet DIVIDEND CHANGE AND LONGTERM PERFORMANCE OF UTILITIES TeWhan Hahn, Auburn University, Montgomery, Alabama, USA Fred Jacobs, Georgia State University, Atlanta, Georgia, USA ABSTRACT This study investigates dividend change of utility firms and its impact on their long term performance. It is shown that utility firms that decrease dividends improve performance after the dividend change period. This result holds regardless of measures of return performance used: raw returns, market adjusted returns and Fama-French three factor model to measure abnormal performance. However, the future return performance of dividend increasing utility firms is various depending on the measures of return performance. Keywords: Dividend Change, Signaling, Utility Firms 1. INTRODUCTION Utility stocks are often considered high dividend paying stocks. In fact, before the deregulation of utility industry during 1980s, most utility firms’ dividend payments tended to be very stable.

Studies found that dividend change announcements by utility firms attract more stock market response than by non-utility firms since investors’ expectation for dividend stability for utility firms are very high. While announcement effect of dividend change in the stock market has been studied extensively, whether dividend changing utility firms exhibit the future performance implied by the dividend change as signaling theory of dividends conjecture has not been studied in the literature very rigorously. Assuming stock return performance follows the changes in the corporate operating performance in the long term, this study investigates whether dividend changing utility firms show the future performance implied by the dividend change. 2. RELATED LITERATURE Several studies have investigated stock market reaction to dividend change announcements by utility firms. Aharony et al (1988) find that, on average, the stock market responds more positively to dividend increase announcements by

utility firms rather than those by un-regulated firms. Impson (1997) in the same spirit presents the evidence that the stock market responds more negatively to dividend decrease announcements by utility firms rather than those by unregulated firms. The positive correlation between dividend change and stock market response has often been interpreted as dividend announcement signaling future earnings or cash flows of announcing firms (Lintner (1956), Carroll (1995)). This study tests whether future stock return performance of dividend changing utility firms confirms prediction of signaling theory of dividends. 3. SAMPLE AND ESTIMATION PROCEDURES We restricted our sample to utility firms changing dividends by more than 10%. This ensures that the event is economically significant. This screen generated 151 utility firms during time period 1985-2005. We selected publicly traded firms whose stock trade data can be found in CRSP Daily Returns database. The final sample consists of 95 utility

firms increasing dividend payments more than 10%, and 32 utility firms decreasing dividend payments more than 10%. Monthly returns of dividend changing utility firms, market (CRSP value weighted) returns and t-bill 3 month rates were pulled out from CRSP (Center for Research in Security Prices). FamaFrench’s monthly three factor data was pulled out from Ken French’s data library (http://mba.tuckdartmouthedu/pages/faculty/kenfrench/data libraryhtml) Source: http://www.doksinet To measure risk-adjusted return performance reported in tables 2, and 3, Fama-French three factor was used in time series regression. The model used is: r it – r F = α + β 1 (r M - r F ) + β 2 (SMB) + β 3 (HML) + e it r it is the month ‘t’ return of portfolio ‘i.’ r F (r M ) is the T-bill rate (market return) in the month t and time subscript is dropped to avoid the clutter. α captures abnormal returns of portfolios and betas are risk factor betas as in a usual asset pricing regression. e it

is the residual 4. EMPIRICAL RESULTS Table 1 summarizes compound mean raw returns and market adjusted compound mean returns for dividend changing firms, before and after the change. Compound raw returns are calculated as (end of period price minus beginning of the period price) divided by the beginning of the period price. Averaging compound raw returns across sample firms results in compound mean raw returns. Market adjusted compound returns are calculated as each month’s firm return minus market return compounded over the period. Then market adjusted compound mean returns are calculated as average of individual firms’ market adjusted compound returns. TABLE 1. AVERAGE COMPOUND MEAN RAW RETURNS AND MARKET ADJUSTED COMPOUND MEAN RETURNS Compound Mean Raw Returns Before Change (24 mos) (t-value) After Change (48 mos) (t-value) Market Adjusted Compound Mean Returns Div. Decrease Div. Increase Div. Decrease Div. Increase -6.76% 36.63% -42.84% 1.10% 11.071* 58.81% -10.011*

-9.88% 0.334 -18.87% $ -1.621 51.24% 4.613* 10.494* -0.835 -2.938* NOTE: The symbols $,*,, and denote statistical significance at the 10%, 5%, 1% and 0.1% levels, respectively, using a generic one tail test. For dividend decreasing samples, in the 24 month period before the dividend change, a negative compound mean raw return, -6.76%, is observed and this return is statistically different from zero as the significance of t-values indicate. After market return is adjusted, return (market adjusted compound mean return) is more negative with -42.84%, indicative of the difficulties facing the dividend decreasing utility firms. For dividend increasing samples, in the 24 month period before the dividend change, a positive compound mean raw return, 36.63% is observed and this return is statistically different from zero as the significance of t-values indicate. After market return is adjusted, return (market adjusted compound mean return) is merely positive with 1.1% and not

statisctiacally significant In the 48 month period after the dividend change, for both samples, significantly higher compound mean raw returns are observed compared to those before the dividend change. After the market return adjustment, however, return becomes negative for both samples with that of dividend increase sample statistically significant. While results in table 1 is informative, one can argue that these are obtained ignoring risks involved. Thus in tables 2 and 3, Fama-French’s three factor time series regression model was used to measure risk-adjusted abnormal performances of dividend changing utility firms before and after dividend change. In table 2, for dividend decreasing utility firm sample, in the 24 month period before the dividend decrease, a negative average abnoromal return of -1.36% per month is observed, which is Source: http://www.doksinet statistically significant. This means that dividend decreasing utility firms significantly underperform the stock

market avearge performance before the dividend decrease, which makes sense since most likely, difficulty in operations would be reflected before the dividend decrease. In the 48 month period after the dividend decrease, a negative average abnormal return of -0.01% per month is observed but this figure is not statistically significant. TABLE 2. EXCESS RETURN TIME SERIES REGRESSION: DIVIDEND DECREASE SAMPLE R-Square Div. Decrease β2 β3 α β1 Before Change (24 mos) (t-value) After Change (48 mos) -0.0136 -4.16* -0.0001 0.5909 7.73* 0.6448 -0.1938 -1.46$ 0.0317 0.4496 3.12* 0.6532 0.2129 0.2515 (t-value) -0.04 7.54* 0.34 5.28* $ NOTE: The symbols ,*,, and denote statistical significance at the 10%, 5%, 1% and 0.1% levels, respectively. The model used is: r it -r ft = α + β 1 (r M -r F ) + β 2 (SMB) + β 3 (HML) + e it r it is the month ‘t’ return of portfolio ‘i.’ r F (r M ) is the T-bill rate (market return) in the month t and time subscript is dropped to avoid the

clutter. In table 3, for dividend increasing utility firm sample, in the 24 month period before the dividend increase, a positive average abnoromal return of 0.33% per month is observed, which is statistically insignificant. In the 48 month period after the dividend increase, a positive average abnormal returns of 0.12% per month is observed but this figure is not statistically significant TABLE 3. EXCESS RETURN TIME SERIES REGRESSION: DIVIDEND INCREASE SAMPLE R-Square Div. Increase β2 β3 α β1 Before Change (24 mos) (t-value) After Change (48 mos) 0.0033 1.21 0.0012 0.5302 7.55* 0.6405 0.0381 0.32 0.0316 0.4722 3.99* 0.7594 0.2427 0.3613 (t-value) 0.48 11.15* 0.35 6.53* NOTE: The symbols $,*,, and denote statistical significance at the 10%, 5%, 1% and 0.1% levels, respectively. The model used is: r it -r ft = α + β 1 (r M -r F ) + β 2 (SMB) + β 3 (HML) + e it r it is the month ‘t’ return of portfolio ‘i.’ r F (r M ) is the T-bill rate (market return) in the month

t and time subscript is dropped to avoid the clutter. 5. CONCLUSION This study finds that dividend decreasing utility firms experience significant improvements in the long term, regardless of the return performance measures used, while poor performance preceded the dividend decrease. Meanwhile, dividend increasing utility firms exhibit mixed return performance results, dependent on the return performance measures used: When raw return is used, improved performance is detected. When market adjusted return is used, significant drop in performance is detected. However, with risk-adjusted return, there is no performance change Overall, results in this study are in sharp contrast to the prediction of the signaling theory of dividends. REFERENCES: Aharony, J., Falk, H and Swary, I, 1988, Information Content of Dividend Increases: the Case of Regulated Utilities. Journal of Business Finance & Accounting, 15: 401–414 Carroll, T. J,1995, The Information Content of Quarterly Dividend

Changes, Journal of Accounting, Auditing and Finance, 10, 293-319. Source: http://www.doksinet Impson, M.,1997, Market Reaction to Dividend-Decrease Announcements: Public Utilities vs Unregulated Industrial Firms, Journal of Financial Research, 20,407-22. Lintner, J., 1956, Distributions of Incomes of Corporations among Dividends, Retained Earnings, and Taxes, American Economic Review, 46, 97-113. Michael, F., 1996, Dividend Changes, Abnormal Returns, and Intra-Industry Firm Valuations Journal of Financial and Quantitative Analysis, 31, 189-211. AUTHOR PROFILES TeWhan Hahn received his Ph.D in Finance from the University of Alabama in 2000 Currently he is an associate professor of finance at Auburn University - Montgomery. Fred Jacobs earned his Ph.D in Accounting from University of Georgia 1973 Currently he is an Emeritus professor of Accounting at Georgia State University