The efficacy of Australia’s full dividend imputation system was questioned by the 2009 Henry Review and the Tax Discussion Paper released in March 2015, which started the process for developing the White Paper on the ‘Reform of Australia’s Tax System’. More recently, heated debate has emerged regarding the sustainability of full franking credit refundability to government revenue. The Labor Party announced its intent to deny cash refunds for excess franking credits from 1 July 2019, if elected to power in the 2019 federal election. In response, the report of the Standing Committee on Economics revealed it had formed the view that Labor’s policy was ‘inequitable and deeply flawed’.
Labor’s defeat at the 2019 election may have postponed further debate about franking credits in the short term. However, the debate will inevitably re-emerge in a post COVID-19 environment where all sources of tax revenue will invite detailed scrutiny, and ultimately the question of the sustainability of the current arrangements is an empirical one.
Our study
Our paper contributes to the ongoing debate by empirically examining the benefits of full franking credit refundability in terms of its impact on corporate tax avoidance. We exploit the plausibly exogenous shock that occurred in late 1999 whereby legislation (section 67-25(1) of the Income Tax Assessment Act 1997) was passed allowing Australian resident shareholders to claim all franking credits attached to dividends received. That is, contrary to the prior rules which only allowed credits to be utilised up to the point where they offset a shareholder’s tax liability, the new rules allow taxpayers to claim all franking credits, even if it puts them into a tax refund position.
Effective 1 July 2000, this new rule increased the value and demand for franking credits by certain shareholders (for example, resident taxpayers whose marginal tax rate is less than the statutory company tax rate of 30 per cent). Arguably, this enhancement to shareholder’s after-tax positions provides stronger incentives for some firms to pay corporate taxes (minimise tax avoidance) to generate additional valuable franking credits for distribution to shareholders.
Hypotheses
We argue that the full dividend imputation system provides a strong countervailing influence on the incentives managers of certain firms may have to engage in tax avoidance. This system eliminates the double taxation on profits distributed by companies to shareholders by allowing the tax paid by the company to be credited, or imputed, to the shareholders by way of a franking credit attached to dividends to reduce the final income tax payable by the shareholder.
Consequently, the imputation system likely leads to heterogeneity in tax avoidance incentives across publicly listed firms due to differences in dividend pay-out policy. Australian firms that pay franked dividends are more likely to have stronger incentives to pay corporate tax relative to foreign firms publicly listed in Australia, Australian firms that do not pay dividends, or Australian firms that pay unfranked dividends. Importantly, this heterogeneity facilitates comparisons between these different groups. Prior Australian studies suggest that dividend imputation is associated with lower tax avoidance. However, tax avoidance and dividend policy decisions are jointly determined in equilibrium. Therefore, to mitigate concerns regarding endogeneity, we extend these findings by exploiting the quasi-experimental setting offered by the change in Australia’s full dividend imputation system to provide cleaner identification of the relation between dividend imputation and tax avoidance.
We contend that post-July 2000, shareholder-friendly managers of Australian firms have enhanced incentives to pay company tax (reduce tax avoidance) to generate valuable franking credits to satisfy shareholders’ increased demand for fully franked dividends. However, only Australian resident shareholders can utilise franking credits. Consequently, low marginal rate resident shareholders (for example, superannuation funds) will prefer franked dividends whilst foreign resident shareholders will prefer capital gains or unfranked dividends. Based on the above intuition, the following hypotheses are tested:
Corporate tax avoidance decreased after 1 July 2000 for:
Hypothesis 1: domestic firms relative to foreign firms.
Hypothesis 2: dividend-paying domestic firms relative to foreign firms.
Hypothesis 3: dividend-paying domestic firms relative to non-dividend-paying domestic firms.
Hypothesis 4: domestic firms paying fully franked dividends relative to dividend-paying domestic firms that do not pay fully franked dividends.
Research design and methodology
The final sample consists of 482 companies publicly listed on the Australian Securities Exchange (ASX) during 1997-2004 (2,237 firm-year observations). To test our hypotheses, a difference-in-differences regression specification is employed to analyse the effect of the change in the dividend imputation system on tax avoidance (measured using cash effective tax rates) for several treatment and control group combinations. This research design allows us to control for time invariant differences between treatment and control groups in addition to general economic trends common to both groups. Different variations of the following specification are estimated:
CASH EFFECTIVE TAX RATEi,t =
α0 + α1TREATEDi,t +α2TREATED*POSTi,t + α3SIZEi,t + α4RETURN ON ASSETSi,t +α5LEVERAGEi,t +α6INTANGIBLE ASSETSi,t + α7CAPITAL INTENSITYi,t+ α8MARKET TO BOOK RATIOi,t + α9RESEARCH AND DEVELOPMENT INTENSITYi,t + εi,t
TREATED is a dummy variable equal to 1 for each treatment group (domestic firms, domestic dividend-paying firms, domestic firms where all dividends paid are fully franked, and domestic firms where at least one dividend paid is fully franked), and 0 otherwise. POST is a dummy variable which takes the value of 1 for financial years commencing on 1 July 2000 or thereafter, and 0 otherwise, in other words the first year of the post period will be the 2001 income year (1 July 2000 to 30 June 2001). The coefficient of interest is α2, the coefficient on TREATED*POST, which is the difference-in-differences estimate of the effect of the change in the dividend imputation system on tax avoidance in the post-refund period.
Empirical findings
Consistent with expectations, we find evidence of a statistically significant increase in cash effective tax rates (decrease in tax avoidance) for both domestic firms and domestic dividend-paying firms relative to foreign firms. This finding is even more pronounced when comparing (i) dividend paying domestic firms to non-dividend-paying domestic firms and (ii) fully franked dividend-paying domestic firms to non-fully franked dividend-paying domestic firms. Estimates suggest the decreases in tax avoidance are economically significant with additional cash taxes paid ranging from approximately A$8.6 million to A$16.6 million.
Conclusions and policy implications
Corporate tax avoidance is a major concern for governments, exacerbated in recent years by the global financial crisis and its aftermath. The results of our study are consistent with the notion that firms undertake less tax avoidance after 1 July 2000, and this is attributed to the presence of stronger incentives for them to pay corporate tax following the advent of full franking credit refundability. Accordingly, we suggest that an unintended consequence of a removal of full refundability of franking credits may be an increase in corporate tax avoidance. The evidence provided in our study highlights the benefits of maintaining full franking credit refundability and the potential of the current dividend imputation system to help mitigate corporate tax avoidance and protect corporate tax system integrity.
Overall, our findings advance the understanding of the incentives for corporate tax avoidance in Australia and may assist in designing future policy. Australia’s budget is more dependent on corporate tax than all other OECD countries except Norway and after personal income tax, corporate tax is the second largest source of federal government revenue. Hence, understanding the underlying motives for tax avoidance can help legislators design a more efficient and equitable corporate tax system, particularly in a post-COVID-19 era where governments may need to explore all potential sources of additional government revenue. This study has particular relevance to the discourse surrounding the efficacy of Australia’s full dividend imputation system.
This is an interesting question and study, so thanks for the work. However, it would have been of greater assistance if the empirical results could have been more clearly stated. Is this the range of amounts per firm studied, per variation in study parameters chosen, or the aggregate total for all firms studied (in which latter case it seems minor) and how can that translate to the aggregate economy level. How does it compare with the fiscal cost of imputation refunds (in other words, how policy significant is the finding)?
Greg, thanks for your comments and questions. Our economic magnitude calculations are more detailed in the paper itself (see footnotes 28 and 29). The calculations are based on amounts per firm studied. We then provide a range of estimates based upon the empirical model used (study parameters chosen) i.e., we use different combinations of fixed effects -industry/time and firm/time and we use different control groups. As for how the estimates translate to the aggregate economy level, this is more difficult as our sample is very specific and the results may not be generalisable to the population (Panel A of Table 1 outlines the sample selection process). Essentially, we only analyse publicly listed companies, not private companies which constitute a significant proportion of the economy. We also impose some restrictive sample selection criteria (e.g., we do not include companies that make a loss) which reduces our sample from 1,839 firms initially to 482 firms. Hope this helps.