The value of equity investments depends to some extent on the tax consequences for investors. When groups of investors have different tax preferences, this can lead to conflicting pressures on firms to either retain earnings or pay dividends. The findings of this study will be of interest to researchers of taxation and corporate governance alike, as they highlight the role that corporate shareholders play in the decisions of the firm. Investors and regulators will also be interested in the findings as they reveal more about the interaction between shareholders with conflicting interests. Lastly, changes in behaviour as a result of changes in tax legislation are of interest to those with fiscal responsibility.
A 2012 dividend tax change in South Africa, which simultaneously altered the tax preferences of individual and corporate investors, provides a unique opportunity to investigate firms’ reaction to their investors’ tax preferences.
This article seeks to determine whether firms respond to changes in their investors’ tax preferences in their decisions to either retain earnings or pay dividends.
The article investigates the responses of firms to the 2012 dividend tax change using multivariate regressions.
Findings show that firms consider changes in the tax preferences of their investors in setting dividend policies. In addition, it appears that corporates have greater success in lobbying for beneficial dividend changes than individuals.
Changes in investors’ tax preferences impact on firms’ dividend policy decisions. These decisions ultimately affect the value of the firm to its investors.
A general finance principle holds that the value of an investment is the present value of its future post-tax cash flows (Ilmanen
The setting for this article is a recent change in South African dividend tax on 1 April 2012, which replaced a 10% tax on firms with a 15% tax on shareholders. Importantly, this tax change altered the tax preferences of two groups of investors, namely, individuals and corporates, simultaneously. Toerien and Marcus (
This is important, as prior research provides limited insights when tax preferences of shareholders alter in a conflicting manner. Bond, Devereux and Klemm (
Using a multivariate regression approach, this study considers the impact of corporate shareholding on growth in dividends prior and subsequent to the 2012 tax change for a sample of South African firms. This approach effectively tests the significance of changes in dividend growth before and after the tax change (related to a difference in difference approach). In addition, findings are cross-checked against an earlier tax change during 2007 (the control sample). Although this tax change aligned the tax preferences of individual and corporate investors, the South African government simultaneously announced its future intentions in respect of dividend tax changes.
Findings show a significantly positive (negative) association between corporate shareholding and growth in dividends after (before) the 2012 tax change. In other words, after (before) this tax change, firms with higher corporate shareholding grew their dividends significantly faster (slower) than other firms. These findings imply that corporate shareholders had sufficient influence to align dividend policies of their investees with their tax preferences. This could be because of the size of their holdings or merely reflect that corporate investors tend to be better at organised lobbying than individuals. By contrast, findings for the control sample show that corporate shareholding had a limited impact on growth in dividends as a result of the 2007 tax change. In summary, this study, therefore, concludes that tax preferences of investors affect dividends. In addition, when tax preferences conflict, corporates appear to have greater influence over their investees than individuals.
This study contributes to the existing literature by revealing that investors place pressure on firms to minimise their own tax burden and that firms adjust their dividend policies in response. In addition, it shows that, in the face of conflicting tax preferences, corporates appear to have greater influence than individuals.
The findings of this study will be of interest to researchers of taxation and corporate governance alike, as they highlight the role that corporate shareholders play in the decisions of the firm. Investors and regulators will also be interested in the findings, as they reveal more about the interaction between shareholders with conflicting interests. Lastly, changes in behaviour as a result of changes in tax legislation are of interest to those with fiscal responsibility.
The remainder of this article is organised as follows: the next section reviews the prior literature, followed by a discussion of dividend tax in South Africa. Then, the research methodology is set out, followed by the sample selection, descriptive statistics and univariate investigations. The detailed multivariate regression results are discussed in a separate section, followed by the results of additional analyses and robustness tests. The final section summarises and concludes the article.
Investors are concerned with maximising the value of each of their investments, which is affected by the tax consequences associated with it. As a result, investors have an incentive to pressure firms to alter their decisions to improve tax outcomes. Firms have an incentive to comply, as greater investor demand increases their market value, which in turn reduces the risk of takeovers and increases executive compensation.
Some research findings support such a conclusion. Ayers, Cloyd and Robinson (
Research on tax changes in Norway and Finland finds that firms in these countries increased dividends prior to dividend tax increases on individuals (Alstadsæter & Fjærli
In summary, prior researchers find that tax preferences of investors affect their own investment decisions. However, the impact on the dividend decisions of firms is less clear. This problem is exacerbated by the fact that most of the tax changes investigated by prior research affected only individuals. Although prior research shows that corporate shareholders have a significant impact on dividend policy (Bond et al.
In most countries, dividends are taxed in the hands of the recipient. However, from 1993 to 2012, firms in South Africa were taxed on dividends they declared under a system known as ‘Secondary Tax on Companies’ (STC). During this time, dividends were collected as tax free income by recipients. In the main, all dividend declarations were subjected to STC, regardless of the identity or nature of the recipient, although subsidiaries had a choice to exempt dividends paid to their parents. However, if the recipient was also subject to STC, it reduced its own STC liability based on dividends received (if the dividends received had been subject to STC).
The rate of STC initially fluctuated but remained stable at 12.5% from 14 March 1996 to 30 September 2007. During 2007, the South African government announced its intention of aligning South African dividend tax practices with international norms. This entailed that STC (a tax on the dividend payer) would be replaced with dividend tax (a tax on the dividend recipient). As an initial step in this direction, the STC rate was decreased to 10.0% for dividends declared on or after 1 October 2007.
An important aspect of dividend tax is its interaction with capital gains tax. Simply put, the South African capital gains tax is levied on the difference between the proceeds received upon the sale of an investment and the price initially paid for it (base cost). As the tax regimes for dividends and capital gains differ, this may affect investors’ tax preferences for receiving dividends or retaining income in the firm. As Toerien and Marcus (
However, on 1 April 2012, STC was finally replaced with dividend tax in respect of dividends declared on or after that date. This change altered the tax preferences of investors yet again. Dividend tax is levied at a rate of 15.0% and is a tax on the dividend recipient. Importantly, some recipients are exempt from the tax, notably South African firms (i.e. corporate investors) who receive the income tax free (they withhold dividend tax from their own dividend payments). Toerien and Marcus (
From the above discussion, it is clear that the South African situation allows for unique empirical investigations. Prior research on dividend tax changes has been criticised by some, as these tax changes only affected individuals who make up a minority of shareholders (Amromin et al.
This article uses a multivariate regression approach to investigate the impact of changing tax preferences on dividends. To reduce the effect of contemporaneous changes, investigations focus on the impact of corporate shareholding on growth in dividends prior and subsequent to the tax change in 2012. This approach effectively tests the significance of changes in dividend growth (related to a difference in difference approach). It firstly eliminates omitted correlated variables that remain constant over time and secondly removes omitted correlated variables that change by a constant factor over time. In addition, findings are cross-checked against an earlier tax change during 2007 (the control sample).
The following model is regressed for firm
Cash on hand is an important limiting factor for dividend payments. Following prior research (Al-Ajmi & Hussain
Firms with greater growth opportunities tend to reinvest earnings, rather than distribute it. GTH represents the change in the 3-year annual compound growth rate in sales between period
The variable of interest is CORP, which is defined as the percentage of common shares held by corporate investors (i.e. the shares held by one company in another). If corporate shareholders have greater (less) influence than individual investors on firm’s dividend policies, CORP will be significant and positive (negative) after the 2012 tax change. By contrast, insignificance for CORP would indicate that firms ignore the tax preferences of their investors in determining dividend policies or that they treat all their investors alike.
To compensate for the impact of outliers, all variables, other than indicator variables, are winsorised at the 1% and 99% levels.
The initial sample to assess the impact of the 2012 tax change consists of firms with a primary listing on the Johannesburg Stock Exchange (JSE) in South Africa on 31 March 2012 (the day before the tax change). This sample is reduced to include only those firms that were still listed on 31 March 2013 and for which all data items are available. The control sample is selected in a similar manner, using the firms with a primary listing on the JSE on 30 September 2007 (the day before the tax change). Once again the sample is reduced to include only those firms that were still listed on 30 September 2008 and for which all data items are available.
In both instances, only firms that paid a dividend in the year before the tax change are included in the final sample. Real estate investment trusts are eliminated from the sample, as these firms are not subject to dividend tax. The final sample numbers and industry composition are detailed in
Sample composition.
Industry | 2007–2008 sample |
2012–2013 sample |
||
---|---|---|---|---|
Number | % | Number | % | |
Auto parts | 1 | 1 | 1 | 1 |
Business services | 2 | 2 | 3 | 2 |
Cement | 1 | 1 | 2 | 1 |
Chemical and allied products | 9 | 7 | 8 | 6 |
Clothing manufacturers | 1 | 1 | - | - |
Computer equipment | 1 | 1 | 2 | 1 |
Computer related services | 8 | 7 | 8 | 6 |
Construction | 3 | 2 | 5 | 3 |
Electronic and electrical equipment | 5 | 4 | 6 | 4 |
Financial services | 17 | 14 | 21 | 14 |
Food producers | 3 | 2 | 4 | 3 |
Food products | 8 | 7 | 10 | 7 |
Health services | 2 | 2 | 3 | 2 |
Hotels and gambling | 3 | 2 | 6 | 4 |
Industrial machinery and equipment | 2 | 2 | 2 | 1 |
Iron and steel products | 2 | 2 | 2 | 1 |
Media | 1 | 1 | 1 | 1 |
Mining | 10 | 8 | 12 | 8 |
Other services | 4 | 3 | 5 | 3 |
Paper, packaging and publishers | 5 | 4 | 3 | 2 |
Restaurants and pubs | 2 | 2 | 2 | 1 |
Retailers | 16 | 13 | 18 | 12 |
Telecommunications | 2 | 2 | 3 | 2 |
Transport | 7 | 6 | 8 | 6 |
Wholesalers | 5 | 4 | 10 | 7 |
Wood products |
1 |
1 |
- |
- |
The descriptive statistics contained in
Descriptive statistics.
Panel | Variable | 2007 |
2008 |
2012 |
2013 |
---|---|---|---|---|---|
A: Mean | DPS | 0.182 | 0.040 | 0.118 | 0.168 |
ROA | −0.002 | −0.017 | −0.007 | −0.023 | |
LOSS | 0.017 | 0.050 | 0.014 | 0.124 | |
RE | 0.023 | 0.202 | 0.019 | 0.048 | |
CF | 0.099 | 0.050 | 0.097 | 0.103 | |
CASH | 0.124 | 0.340 | 0.249 | 0.889 | |
DE | −0.008 | 0.413 | 0.286 | −0.099 | |
GTH | 0.059 | −0.069 | 0.372 | −0.005 | |
PD | 0.505 | 0.182 | 0.350 | 0.118 | |
RESTR | 0.099 | 0.107 | 0.124 | 0.170 | |
SIZE | 0.312 | −0.265 | 0.174 | 0.170 | |
EMP | 0.034 | 0.033 | 0.085 | 0.081 | |
CORP | 0.177 | 0.182 | 0.162 | 0.163 | |
B: Median | DPS | 0.120 | 0.052 | 0.080 | 0.050 |
ROA | 0.001 | −0.007 | −0.001 | −0.011 | |
LOSS | 0.000 | 0.000 | 0.000 | 0.000 | |
RE | 0.016 | 0.018 | 0.006 | 0.001 | |
CF | 0.000 | 0.000 | 0.000 | 0.000 | |
CASH | 0.077 | 0.093 | 0.022 | 0.046 | |
DE | −0.008 | 0.059 | −0.004 | 0.037 | |
GTH | 0.014 | 0.005 | 0.033 | 0.003 | |
PD | 0.090 | 0.120 | 0.065 | 0.080 | |
RESTR | 0.000 | 0.000 | 0.000 | 0.000 | |
SIZE | 0.296 | −0.267 | 0.181 | 0.165 | |
EMP | 0.000 | 0.000 | 0.000 | 0.000 | |
CORP | 0.100 | 0.100 | 0.070 | 0.070 | |
C: Standard deviation | DPS | 0.985 | 1.801 | 1.326 | 0.966 |
ROA | 0.157 | 0.121 | 0.063 | 0.069 | |
LOSS | 0.128 | 0.218 | 0.117 | 0.331 | |
RE | 0.227 | 2.270 | 0.112 | 0.245 | |
CF | 0.300 | 0.218 | 0.296 | 0.306 | |
CASH | 6.718 | 13.179 | 3.251 | 8.012 | |
DE | 1.524 | 6.681 | 2.911 | 3.365 | |
GTH | 0.302 | 0.659 | 0.227 | 0.298 | |
PD | 3.709 | 0.985 | 1.063 | 1.326 | |
RESTR | 0.300 | 0.311 | 0.331 | 0.373 | |
SIZE | 0.324 | 0.315 | 0.243 | 0.334 | |
EMP | 0.099 | 0.096 | 0.164 | 0.158 | |
CORP | 0.214 | 0.221 | 0.210 | 0.220 |
DPS, change in total dividend per share declared, excluding special dividends; ROA, change in return on total assets for the period, calculated as the change in earnings before interest and tax divided by average total assets in the current and prior period; LOSS, indicator variable set to one if a firm reports a basic loss per share and zero otherwise; RE, change in retained earnings as a percentage of common equity; CF, indicator variable set to one if a firm reports negative cash generated by operations and zero otherwise; CASH, change in cash and cash equivalents per share; DE, change in leverage, calculated as the book value of total assets to the book value of common equity in each period; GTH, change in 3-year annual compound growth rate in sales between the current and prior periods; PD, change in prior year dividend per share; RESTR, indicator variable set to one if the firm has discontinued operations in the current year and zero otherwise; SIZE, change in size, calculated as the natural log of the market value of the firm in each period; EMP, the percentage of outstanding common shares held by employees; CORP, the percentage of outstanding common shares held by corporate investors.
The descriptive statistics in
To investigate the impact of corporate shareholding on dividend per share, the sample is stratified into firms with corporate shareholding above and below the median in each of the sample years. Panel A of
Results of univariate investigations.
Panel | Variable | Subsample | 2007 |
2008 |
2012 |
2013 |
||||
---|---|---|---|---|---|---|---|---|---|---|
Test statistic | Test statistic | Test statistic | Test statistic | |||||||
A: Comparison of dividends and growth (changes) therein | Mean dividend per share | CORP above the median | 2.140 | - | 2.118 | - | 1.921 | - | 2.210 | - |
CORP below the median | 1.676 | - | 1.734 | - | 2.086 | - | 2.084 | - | ||
Difference in means | 0.464 | - | 0.384 | - | 0.165 | - | 0.126 | - | ||
Satterthwaite test statistic | 0.51 | 0.614 | 0.55 | 0.583 | 0.28 | 0.781 | 0.18 | 0.859 | ||
Mean growth (change) in dividend per share | CORP above the median | 0.114 | - | −0.200 | - | −0.069 | - | 0.309 | - | |
CORP below the median | 0.247 | - | 0.272 | - | 0.299 | - | 0.028 | - | ||
Difference in means | 0.133 | - | 0.472 | - | 0.368 | - | 0.281 | - | ||
Satterthwaite test statistic | 0.70 | 0.487 | 1.41 | 0.163 | 1.63 | 0.107 | 0.088 | |||
B: Pearson correlations with DPS | ROA | - | 0.142 | 0.120 | 0.054 | < 0.001 | 0.001 | |||
LOSS | - | 0.003 | 0.970 | −0.065 | 0.477 | −0.025 | 0.762 | 0.007 | ||
RE | - | −0.053 | 0.561 | 0.043 | 0.640 | −0.002 | 0.981 | 0.124 | 0.138 | |
CF | - | −0.010 | 0.916 | 0.005 | 0.953 | 0.014 | 0.864 | −0.053 | 0.523 | |
CASH | - | −0.108 | 0.237 | 0.033 | < 0.001 | 0.120 | 0.152 | |||
DE | - | −0.142 | 0.121 | 0.023 | 0.803 | −0.008 | 0.920 | −0.034 | 0.681 | |
GTH | - | 0.094 | 0.306 | 0.046 | 0.618 | 0.118 | 0.157 | 0.031 | 0.709 | |
PD | - | −0.103 | 0.260 | 0.012 | < 0.001 | < 0.001 | ||||
RESTR | - | 0.083 | 0.047 | 0.607 | −0.049 | 0.560 | −0.090 | 0.282 | ||
SIZE | - | 0.107 | 0.242 | 0.134 | 0.142 | 0.112 | 0.180 | 0.133 | 0.110 | |
EMP | - | −0.025 | 0.784 | 0.014 | 0.875 | 0.001 | 0.995 | −0.078 | 0.351 | |
CORP | - | −0.083 | 0.368 | 0.007 | 0.014 | 0.013 |
DPS, Change in total dividend per share declared, excluding special dividends; ROA, Change in return on total assets for the period, calculated as the change in earnings before interest and tax divided by average total assets in the current and prior period; LOSS, Indicator variable set to one if a firm reports a basic loss per share and zero otherwise; RE, Change in retained earnings as a percentage of common equity; CF, Indicator variable set to one if a firm reports negative cash generated by operations and zero otherwise; CASH, Change in cash and cash equivalents per share; DE, Change in leverage, calculated as the book value of total assets to the book value of common equity in each period; GTH, Change in 3 year annual compound growth rate in sales between the current and prior period; PD, Change in prior year dividend per share; RESTR, Indicator variable set to one if the firm has discontinued operations in the current year and zero otherwise; SIZE, Change in size, calculated as the natural log of the market value of the firm in each period; EMP, The percentage of outstanding common shares held by employees; CORP, The percentage of outstanding common shares held by corporate investors.
Significant at the 10% level;
significant at the 5% level;
significant at the 1% level.
The Pearson correlations in Panel B of
By contrast, prior to the announcement of impending changes in tax policy during 2007, the degree of corporate shareholding had an insignificant impact on growth in dividends (
The above findings, therefore, offer some preliminary evidence that changes in the tax preferences of corporate investors because of the 2012 tax change affected growth in dividends. However, this study relies on the results of multivariate investigations for its conclusions. These results are detailed in the sections that follow.
Main regression results.
Variable | 2007 |
2008 |
2012 |
2013 |
||||
---|---|---|---|---|---|---|---|---|
Coefficient | Coefficient | Coefficient | Coefficient | |||||
ROA | 0.051 | 0.038 | 0.003 | 0.018 | ||||
LOSS | 0.348 | 0.580 | −0.329 | 0.451 | 0.048 | 0.942 | −0.122 | 0.461 |
RE | −0.190 | 0.675 | −0.016 | 0.979 | −0.448 | 0.543 | 0.350 | 0.128 |
CF | −0.066 | 0.800 | 0.384 | 0.375 | 0.121 | 0.564 | 0.083 | 0.604 |
CASH | −0.013 | 0.314 | 0.022 | 0.205 | 0.018 | 0.014 | ||
DE | −0.083 | 0.198 | −0.042 | 0.677 | 0.049 | 0.490 | −0.053 | 0.341 |
GTH | 0.016 | 0.090 | 0.682 | 0.858 | 0.163 | −0.229 | 0.525 | |
PD | −0.038 | 0.710 | 0.098 | 0.370 | −0.100 | 0.276 | 0.001 | |
RESTR | 0.020 | 0.182 | 0.518 | −0.326 | 0.104 | −0.085 | 0.488 | |
SIZE | 0.204 | 0.451 | 0.390 | 0.210 | 0.416 | 0.138 | 0.222 | 0.178 |
EMP | −0.831 | 0.355 | −0.402 | 0.700 | −0.304 | 0.450 | −0.166 | 0.592 |
CORP | −0.262 | 0.476 | 0.099 | 0.040 | 0.040 | |||
N | 121 | - | 121 | - | 145 | - | 145 | - |
14.3% | - | 20.1% | - | 25.6% | - | 32.8% | - | |
Chow-test | - | - | < 0.001 | - | - | 0.004 |
DPS, change in total dividend per share declared, excluding special dividends; ROA, change in return on total assets for the period, calculated as the change in earnings before interest and tax divided by average total assets in the current and prior period; LOSS, indicator variable set to one if a firm reports a basic loss per share and zero otherwise; RE, change in retained earnings as a percentage of common equity; CF, indicator variable set to one if a firm reports negative cash generated by operations and zero otherwise; CASH, change in cash and cash equivalents per share; DE, change in leverage, calculated as the book value of total assets to the book value of common equity in each period; GTH, change in 3-year annual compound growth rate in sales between the current and prior periods; PD, change in prior year dividend per share; RESTR, indicator variable set to one if the firm has discontinued operations in the current year and zero otherwise; SIZE, change in size, calculated as the natural log of the market value of the firm in each period; EMP, the percentage of outstanding common shares held by employees; CORP, the percentage of outstanding common shares held by corporate investors.
Significant at the 10% level;
significant at the 5% level;
significant at the 1% level.
Note: The Chow-test (Chow
More importantly, the results in
In contrast, findings from the control sample show that corporate shareholding had an insignificant impact on dividend growth prior to the announcement of impending tax changes during 2007 (
Overall, the main regression results, therefore, imply that firms consider the tax preferences of their investors in determining dividend policies. In addition, corporate investors appear to have greater influence over their investees than individual investors when tax preferences conflict. This could be because of the size of their holdings or merely reflect that corporate investors tend to be better at organised lobbying than individuals.
The main regression results reveal that growth in dividends is affected by the degree of corporate shareholding. However, it is possible that corporate shareholding affects the dividend itself (i.e. the level thereof) independently of changes in tax preferences. For this reason, the regression is also run using a levels specification. The results from this model specification are presented in
Regression results determined independently per year.
Variable | 2007 |
2008 |
2012 |
2013 |
||||
---|---|---|---|---|---|---|---|---|
Coefficient | Coefficient | Coefficient | Coefficient | |||||
ROA | 1.366 | 0.113 | 0.618 | 0.500 | 0.002 | 0.001 | ||
LOSS | 0.548 | 0.477 | −0.325 | 0.489 | −0.366 | 0.554 | 0.766 | 0.189 |
RE | 0.127 | 0.691 | 0.188 | 0.485 | 0.032 | 0.869 | −0.526 | 0.260 |
CF | 0.148 | 0.643 | 0.110 | 0.807 | 0.215 | 0.337 | 0.318 | 0.588 |
CASH | 0.004 | 0.655 | 0.009 | 0.242 | 0.002 | 0.716 | −0.020 | 0.170 |
DE | −0.266 | 0.285 | −0.008 | 0.745 | 0.049 | 0.093 | ||
GTH | 0.145 | 0.565 | 0.046 | 0.380 | 0.620 | 0.105 | 0.944 | |
PD | < 0.001 | < 0.001 | < 0.001 | < 0.001 | ||||
RESTR | 0.067 | 0.106 | 0.732 | −0.208 | 0.320 | 0.113 | 0.800 | |
SIZE | 0.024 | 0.060 | 0.256 | 0.059 | 0.182 | −0.108 | 0.332 | |
EMP | 0.215 | 0.850 | 0.316 | 0.778 | 0.050 | 0.911 | 0.493 | 0.690 |
CORP | −0.038 | 0.931 | −0.073 | 0.858 | 0.028 | 0.929 | 0.009 | |
81.6% | - | 87.6% | - | 91.0% | - | 79.2% | - | |
Chow-test | - | - | 0.758 | 0.704 | - | - | < 0.001 |
DPS, change in total dividend per share declared, excluding special dividends; ROA, change in return on total assets for the period, calculated as the change in earnings before interest and tax divided by average total assets in the current and prior period; LOSS, indicator variable set to one if a firm reports a basic loss per share and zero otherwise; RE, change in retained earnings as a percentage of common equity; CF, indicator variable set to one if a firm reports negative cash generated by operations and zero otherwise; CASH, change in cash and cash equivalents per share; DE, change in leverage, calculated as the book value of total assets to the book value of common equity in each period; GTH, change in 3-year annual compound growth rate in sales between the current and prior periods; PD, change in prior year dividend per share; RESTR, indicator variable set to one if the firm has discontinued operations in the current year and zero otherwise; SIZE, change in size, calculated as the natural log of the market value of the firm in each period; EMP, the percentage of outstanding common shares held by employees; CORP, the percentage of outstanding common shares held by corporate investors.
Significant at the 10% level;
significant at the 5% level;
significant at the 1% level.
Note: The Chow-test (Chow
In addition, results from a Chow-test (Chow
Results from this robustness test, therefore, imply that corporate shareholding does not affect dividends independently of changes in tax preferences. In addition, the findings also confirm that corporates successfully lobbied for higher dividends subsequent to the 2012 tax change, despite their new tax preferences conflicting with those of individuals.
The main regression results do not control for cross-sectional correlation between firms, mainly because the tax change is itself severely cross-sectionally correlated in nature. In other words, decisions based on the tax change cannot be independent between firms. However, to investigate the impact of cross-sectional correlation, the regression is also run with robust standard errors clustered by firm. These results are presented in
Controlling for cross-sectional correlation.
Variable | 2007 |
2008 |
2012 |
2013 |
||||
---|---|---|---|---|---|---|---|---|
Coefficient | Coefficient | Coefficient | Coefficient | |||||
ROA | 1.574 | 0.150 | 0.084 | 0.023 | 0.046 | |||
LOSS | 0.348 | 0.862 | −0.329 | 0.327 | 0.048 | 0.931 | −0.122 | 0.367 |
RE | −0.190 | 0.682 | −0.016 | 0.972 | −0.448 | 0.476 | 0.050 | |
CF | −0.066 | 0.690 | 0.384 | 0.112 | 0.121 | 0.446 | 0.083 | 0.538 |
CASH | −0.013 | 0.458 | 0.022 | 0.285 | 0.052 | 0.136 | 0.040 | |
DE | 0.083 | 0.327 | −0.042 | 0.523 | 0.049 | 0.253 | −0.054 | 0.316 |
GTH | 0.069 | 0.090 | 0.498 | 0.089 | −0.229 | 0.396 | ||
PD | 0.038 | 0.841 | 0.098 | 0.658 | −0.100 | 0.564 | 0.203 | 0.102 |
RESTR | 0.064 | 0.182 | 0.316 | 0.069 | −0.085 | 0.417 | ||
SIZE | 0.204 | 0.399 | 0.390 | 0.328 | 0.093 | 0.222 | 0.159 | |
EMP | 0.072 | −0.402 | 0.555 | −0.305 | 0.252 | −0.166 | 0.365 | |
CORP | −0.262 | 0.478 | −0.668 | 0.244 | 0.033 | 0.066 | ||
14.3% | - | 20.1% | - | 25.6% | - | 32.8% | - | |
Chow-test | - | - | < 0.001 | - | - | 0.004 |
DPS, change in total dividend per share declared, excluding special dividends; ROA, change in return on total assets for the period, calculated as the change in earnings before interest and tax divided by average total assets in the current and prior period; LOSS, indicator variable set to one if a firm reports a basic loss per share and zero otherwise; RE, change in retained earnings as a percentage of common equity; CF, indicator variable set to one if a firm reports negative cash generated by operations and zero otherwise; CASH, change in cash and cash equivalents per share; DE, change in leverage, calculated as the book value of total assets to the book value of common equity in each period; GTH, change in 3-year annual compound growth rate in sales between the current and prior period; PD, change in prior year dividend per share; RESTR, indicator variable set to one if the firm has discontinued operations in the current year and zero otherwise; SIZE, change in size, calculated as the natural log of the market value of the firm in each period; EMP, the percentage of outstanding common shares held by employees; CORP, the percentage of outstanding common shares held by corporate investors.
Significant at the 10% level;
significant at the 5% level;
significant at the 1% level.
Note: The Chow-test (Chow
Findings show that higher corporate shareholding continues to be associated with higher dividend growth subsequent to the 2012 tax change, albeit at the 10% level (
In summary, controlling for cross-sectional correlation does not alter the main conclusions of this study. Results continue to imply that firms consider the tax preferences of their investors in determining dividend policies and that corporate investors have greater influence over their investees than individual investors when tax preferences conflict. However, it appears that lobbying in anticipation of future tax changes could be weaker than what the main results suggest.
This study was approved by the Research Ethics Committee of the University of Pretoria. It uses only secondary data that are publicly available.
When investors with differing tax preferences invest in the equity of the same firm, this can lead to conflicting pressures on management to either retain earnings or distribute profits to investors. This article investigates the response of firms when their investors have conflicting tax preferences using a change in South African dividend tax on 01 April 2012 as its setting. Importantly, this tax change allows for unique investigations as it altered the tax preferences of two groups of investors, namely, individuals and corporates, simultaneously. Using a multivariate regression approach, this study considers the impact of corporate shareholding on growth in dividends prior and subsequent to the tax change during 2012 for a sample of South African firms. This approach effectively tests the significance of changes in dividend growth (related to a difference in difference approach). In addition, the findings are cross-checked against an earlier tax change during 2007 (the control sample).
Findings show a significantly positive (negative) association between corporate shareholding and growth in dividends after (before) the 2012 tax change. In other words, after (before) this tax change, firms with higher corporate shareholding grew their dividends significantly faster (slower) than other firms. These findings imply that corporate shareholders had sufficient influence to align dividend policies of their investees with their tax preferences. This could be because of the size of their holdings or merely reflect that corporate investors tend to be better at organised lobbying than individuals. By contrast, findings from the control sample show that corporate shareholding had a limited impact on growth in dividends as a result of the 2007 tax change. In summary, this study, therefore, concludes that tax preferences of investors affect dividends. In addition, when tax preferences conflict, corporates appear to have greater influence over their investees than individuals.
This study contributes to the existing literature by revealing that investors place pressure on firms to minimise their own tax burden and that firms respond to this pressure through their dividend policies. In addition, it shows that, in the face of conflicting tax preferences, corporates appear to have greater influence than individuals.
The findings of this study will be of interest to researchers of taxation and corporate governance alike, as they highlight the role that corporate shareholders play in the decisions of the firm. Investors and regulators will also be interested in the findings, as they reveal more about the interaction between shareholders with conflicting interests. Lastly, changes in behaviour as a result of changes in tax legislation are of interest to those with fiscal responsibility.
However, the findings of this study are limited to tax changes that altered the tax preferences of corporate and individual investors simultaneously. In addition, the findings cannot be extrapolated to conflicting tax preferences between other groups of investors. Such questions are left to future research.
The author would like to thank Marna de Klerk, Petri Ferreira and Lizette Kotze for their helpful comments and suggestions.
The author declares that he has no financial or personal relationships that may have inappropriately influenced him in writing this article.
Note that the Worldscope data item on Datastream is adjusted for events such as stock splits and share consolidations. This ensures the comparability of the data across sample years.
Ownership structures tend to be relatively stable over shorter periods of time, which the descriptive statistics of this paper confirm. For this reason, the ownership percentages in this paper are not specified as changes therein.