The relationship between the dividend payout ratio and the subsequent earnings growth : a South African study : an analysis into the relationship between the dividend payout ratio and the subsequent earnings growth at the market level

Traditionally, it has been widely accepted that current reinvestments leads to future growth, and hence that there exists a negative relationship between dividend payout ratios and subsequent earnings growth at both the company and market level. Surprisingly, more recent research found a positive relationship between these two variables, rationalised in terms of management signalling positive future prospects through higher dividend payouts. Following the methodology developed by Arnott and Asness (2003), this study conducted an analysis in to which hypothesis is supported at the market level (proxied by the All Share Index) in South Africa, and how these findings compared to international market level findings. Furthermore, the findings were analysed within the context of the political and economic conditions unique to the South African market over the 1960-2014 study period. The results indicated a negative relationship between the payout ratio and subsequent earnings growth from 1960 to 2014. When the time period was subdivided into periods before and after 1994 (a year considered to be a structural break in South African political-economic history), a positive relationship was found from 1960-1994, while a negative relationship was found from 1994-2014. The possible explanations for this contradiction were investigated. For the counterintuitive pre-1994 result, several literature-based tests were conducted in order to eliminate possible explanations other than the link between payout ratio and earnings growth. Firstly, the tests were repeated for three and five year earnings growth periods. With regard to the five year period, although the coefficient for the payout ratio was positive from 1960-2014, the results were statistically insignificant. The subdivided regression periods, 1960-1994 and 1994-2014, also indicated less significant results. For the three year earnings growth period, the regression test generated positive coefficients for the payout ratio variable for all the time periods, with both sub-periods being statistically significant. The difference between the ten and three year subsequent earnings growth findings may have been as a result of South Africa companies making better long-term (ten year) than short-term (three year) investments during that time period. In addition, the possible impact of share repurchases, earnings yields as earnings growth predictors (i.e. the market valuation impact), and the possible role of mean reversion were all either considered, or statistically tested. Although the former two played no role, it was found that mean reversion was a statistically stronger predictor of future earnings growth on the index level than payout ratio. As the above result s differ not only from most (but not all) international studies, and also from the one company-level study previously conducted for South Africa, these differences were further investigated. In terms of the latter, it was found that the most likely reason for the discrepancy was that the previous researcher, who confined her study to JSE-listed industrial stocks, used nominal earnings growth data as opposed to real growth data. It is argued that this is an incorrect approach. Nonetheless, given that means reversion is as good, if not a better, predictor of earnings growth at the market level in South Africa, the traditional hypothesis that a higher payout ratio implies reduced future earnings growth, cannot be ruled out for the JSE