The timing of dividend declarations based on a forthcoming change in dividend tax regime

Abstract

Background: The anticipated change in tax regime in South Africa during 2012 provided the opportunity to investigate the role of taxes in affecting corporate payout behaviour.

Aim: The aim of this study was to investigate whether dividend declarations were accelerated or postponed during 2012 for a sample of companies listed in South Africa based on the financial years 2009 to 2015.

Methods: Firstly, a mixed model analysis of variance was employed to investigate the trend in mean days-to-declaration of final and interim dividends and whether the days-to-declaration of dividends in 2012 differed significantly from other years. Secondly, an investigation at individual company level was performed to gain an insight into the timing of declarations before and after 1 April 2012, the non-declaration of dividends during 2012 and special dividends during 2012.

Results: For final dividends findings do not suggest an acceleration or postponement during the 2012 financial years of companies selected. For interim dividends, a significant increase in the days-to-declaration during 2012 was noted (indicative of a postponement during 2012). An investigation at individual company level of interim dividend declarations before and after 1 April 2012 furthermore supports a tax explanation for the postponement noted during 2012. Non-declarations of dividends and special dividends during 2012 were not noted as being utilised for the postponement or acceleration of dividends during 2012. The findings of this study contribute to dividend policy literature by providing empirical evidence that the timing of interim dividend declarations was adjusted in the year of an anticipated tax reform.

Keywords:

• dividends
• dividend days-to-declaration
• dividend declaration
• dividend policy
• dividends tax

Data availability statement

The dataset related to this manuscript is available on request from the corresponding author.

Acknowledgments

Thanks are extended to Prof. Martin Kidd (Centre for Statistical Consultation, Stellenbosch University) for the statistical analyses performed in respect of this study. Thanks also go to Adrian Samuels and Frances Smit for their involvement in the data collection in the initial exploratory research relating to this study.

ORCID

Rudie Nel http://orcid.org/0000-0003-0869-1996