Dividends and Taxes: The Moderating Role of Agency Conflicts

Janis Berzins
2017 Social Science Research Network  
A B S T R A C T We find that potential conflicts between majority and minority shareholders strongly influence how dividends respond to taxes. When the controlling shareholder has a smaller stake, the incentives to extract private benefits are stronger -a shareholder conflict that can be mitigated by dividend payout. We study a large and clean regulatory shock in Norway that increases the dividend tax rate for all individuals from 0% to 28%. We find that dividends drop less the higher the
more » ... he higher the potential shareholder conflict, suggesting that dividend policy trades off tax and agency considerations. The average payout ratio falls by 30 percentage points when the conflict potential is low, but by only 18 points when it is high. These lower dividends cannot be explained by higher salaries to shareholders or diverse liquidity needs. We also observe a strong increase in indirect ownership of high-conflict firms through tax-exempt holding companies and suggest policy implications for intercorporate dividend taxation. T controlling stake gives majority shareholders sufficient power both to single-handedly make the dividend decision and to extract private benefits at the expense of minority shareholders. 1 The controlling stake also gives the owner strong incentives to monitor management and mitigate the vertical agency conflict. The controlling shareholder captures the entirety of any private benefits, but suffers the cost only in proportion to her stake. Hence, the smaller her stake (i.e., the closer to 50% rather than 100%), the greater her incentives to extract private benefits -and thus the greater the importance of using dividends to mitigate the agency conflict (Gomes, 2000). We therefore use the size of the controlling stake as an inverse measure of the potential shareholder conflict. 2 The tax shock we observe increases the cost of paying dividends and should therefore cause all firms to pay less. However, the controlling shareholder must trade off the positive effect of reduced taxes against the negative effect of higher shareholder conflicts, which is larger the smaller the controlling stake. Thus, because dividends are used to address agency costs, we hypothesize that firms reduce dividends less after the tax shock the smaller the controlling shareholder's stake. We find that the tax shock has a large effect on dividends, reducing the average payout ratio (dividends to earnings) from 43% to 18%. Consistent with our hypothesis, the drop is smaller the higher the potential shareholder conflict. For instance, the average payout ratio falls by 30 percentage points when the majority stake is high (90-99%, low conflict), but falls by only 18 percentage points when the stake is low (50-60%, high conflict). Similarly, multiple-owner firms, which trade off both tax and agency effects, cut dividends less than do single-owner firms, which have no shareholder conflicts and therefore face only tax effects. Moreover, dividends and the largest equity stake are unrelated in firms without a controlling shareholder, where nobody can single-handedly set the dividends. Taken together, these results suggest that, because controlling shareholders trade off the effect of dividends on taxes against the effect of dividends on shareholder conflicts, the relationship between dividends and taxes depends on the severity of agency costs. We consider and reject four alternative explanations. First, we avoid several tax-related complications because, unusually, the tax rate is flat and identical for dividends and capital gains. This fact rules out problems that often plague dividend studies -tax-based dividend clienteles (Elton and Gruber, 1970; Desai and Jin, 2011) , tax arbitrage between dividends and capital gains around tax shocks (Sørensen, 2005) , and firms' using tax-disadvantaged dividends to signal intrinsic value (Bernheim, 1991; Bernheim and Wantz, 1995) . Most tax reforms examined in the literature change the relative taxation of dividends and capital gains, such as the 1986 and 2003 reforms in the United States (Hubbard and Michaely, 1997; Chetty and Saez, 2005) . We study a tax reform designed to affect dividends and capital gains equally regardless of payout type. Thus, the dividend response cannot be due to tax-induced shifts between dividends and repurchases. 3 Second, controlling shareholders might pay themselves larger salaries to offset the lower dividends after the tax decrease. If this were the case, what looks like a tradeoff between tax effects and agency effects of dividends is just a tax-driven switch between two payout forms (Jacob and Michaely, 2017). We find no such evidence. Third, although unequal liquidity preferences among shareholders seem to matter for the tax sensitivity of dividends as in Jacob and Michaely (2017), we find that such coordination concerns do not replace agency concerns. Finally, we find that neither conflicts between shareholders and managers nor shareholder wealth influences the observed relationship between dividends, taxes, and agency costs. Our hypothesis implies that firms with severe shareholder conflicts, which are reluctant to reduce dividends despite the tax shock, will look for ways to mitigate the increased tax burden. While the tax reform raised the tax on dividends paid to individuals from 0% to 28%, dividends paid to firms remained tax-free. This difference creates incentives to own shares indirectly through holding companies rather than directly. Indirect ownership ensures that free cash flow can be taken away from the majority shareholder's control without triggering immediate tax payments. 4 We hypothesize that higher dividend taxation for individuals increases the use of indirect ownership, particularly in firms where potential shareholder conflicts are high. 1 Johnson et al. (2000) and Bertrand et al. (2002) provide evidence that controlling shareholders extract private benefits. Common mechanisms are tunneling, nepotism, and social visibility. 2 This measure also reflects a common agency measure used in the literature, which is the wedge between voting rights and cash flow rights (Faccio et al., 2001) . In particular, our measure reflects the ratio between control rights, which are constant across the sample, and cash flow rights, which vary. However, our measure is not driven by dual-class shares, which are rare in Norway. For instance, Che and Langli (2015) find that only 3.8% of the firm-year observations involve firms with dual-class shares in Norwegian private firms from 2001 to 2011. Up until 1994, foreigners as a group could not own more than one third of a firm's voting shares. The firms adapted to this regime by widespread use of non-voting shares targeted to foreigners. When EU regulation outlawed the discrimination of foreign investors in 1995, however, the use of dual-class shares dropped very strongly and remained low. There are no legal restrictions or corporate governance codes on the use of dual-class shares in private firms in our sample period. 3 We find repurchase activity in only 1.4% of the firm years, varying between 0.9% and 2.0% over the years. Excluding these cases has no effect on any result. Repurchases might be unusually low not just because of tax neutrality, but also because shareholders who sell might lose control. Moreover, sellers must negotiate with the firm at every repurchase because there is no liquid market and no obvious market price for the private firm's shares. 4 Norwegian holding companies have no special tax status. Just as for any corporate owner, dividend income is tax-free. However, a holding company cannot permanently shield its personal owners from taxes on cash needed for consumption, because the holding company must pay this cash to the person as taxable dividends. Nevertheless, the holding company can be used to temporarily store the cash paid from the operating company at zero tax costs. Because the average holding company has only 1.2 operating companies (see Table 3 ), most holding companies cannot be used to reallocate capital across operating companies. Moreover, because holding companies have no operating activity and very few owners (2.5 on average after the tax reform), agency problems in the holding company are negligible. Finally, establishing a holding company is not costless. There are registration costs, reporting costs, and equity requirements. hence lower dividend capacity. We prefer this forward-looking measure to measures using past growth, which might overlap with Free cash flow. Unreported results show, however, that no significant relationship changes if we replace the forward-looking by the backward-looking growth measure.
doi:10.2139/ssrn.2973551 fatcat:ir73whzv5bdjldobg4ayoqnxky