Taxation, Bank Leverage, and Financial Crises. By Ruud de Mooij, Michael Keen, and Masanori Orihara
IMF Working Paper No. 13/48
Feb 25, 2013
Summary: That most corporate tax systems favor debt over equity finance is now widely recognized as, potentially, amplifying risks to financial stability. This paper makes a first attempt to explore, empirically, the link between this tax bias and the probability of financial crisis. It finds that greater tax bias is associated with significantly higher aggregate bank leverage, and that this in turn is associated with a significantly greater chance of crisis. The implication is that tax bias makes crises much more likely, and, conversely, that the welfare gains from policies to alleviate it can be substantial—far greater than previous studies, which have ignored financial stability considerations, suggest.
The onset of the financial crisis of 2008 quickly prompted many assessments of the role that taxation might have played.1 Their consensus was clear, but vague: tax distortions did not trigger the crisis, but may have increased vulnerability to financial crises. Prominent among the reasons given for this was ‘debt bias’: the tendency toward excess leverage induced, in almost all countries, by the deductibility against corporate taxation of interest payments but not of the return to equity.2 By encouraging firms to finance themselves by debt rather than equity, this might have made them more vulnerable to shocks and so increased both the likelihood and intensity of financial crises. The point applies in principle to all firms, but is a particular concern in relation to financial institutions; and these are the focus here.
This potential link from tax design to financial crises is now widely recognized. But analysis has not progressed beyond metaphor and speculation. Shackelford, Shaviro, and Slemrod (2010, p. 784), for instance, stress “the possibility that the tax biases served…as extra gasoline intensifying the explosion once other causes lit the match”, and the European Commission that “The welfare costs related to debt bias might not be negligible [because] excessive debt levels increase the probability of default” (European Commission, 2011; p. 7), with both the ‘might’ and the ‘not negligible’ leaving much doubt and imprecision. This paper aims to provide a first attempt to establish and quantify an empirical link between the tax incentives that encourage financial institutions (more precisely, banks, the group for which we have data) to finance themselves by debt rather than equity and the likelihood of financial crises erupting; and then to try to quantify the welfare gains that policies to address this bias might consequently yield.
The approach is to combine two elements in a causal chain. The first is that between the statutory corporate tax rate and banks’ leverage. This has received substantial attention in relation to nonfinancial firms,3 but very little in relation to the financial sector. Keen and De Mooij (2011), however, show that for banks too a higher corporate tax rate, amplifying the tax advantage of debt over equity finance, should in principle lead to higher levels of leverage; the presence of capital regulations does not affect the usual tax bias applying, so long as it is privately optimal for banks to hold some buffer over regulatory requirements (as they generally do). Empirically too, Keen and de Mooij (2012) find that, for a large crosscountry panel of banks, tax effects on leverage are significant—and, on average, about aslarge as for nonfinancial institutions. These effects are very much smaller, they also find, for the largest banks, which generally account for the vast bulk of all bank assets. One task in this paper is to explore these findings further, using data now available to extend coverage into the crisis period that began in 2008—enabling a comparison of tax impacts pre- and post-onset—and applying the same estimation strategy to country-level data for the OECD.
Importantly, the finding that tax distortions to leverage are small for the larger banks, which are massively larger than the rest, does not mean that the welfare impact of tax distortions is in aggregate negligible: even small changes in the leverage of very large banks could have a large impact on the likelihood of their distress or failure, and hence on the likelihood of financial crisis.
This is where the second link in the causal chain explored here comes in: that between the aggregate leverage of the financial sector and the probability of financial crisis.4 We estimate such a relationship for OECD countries, applying the estimation strategy of Barrell et al. (2010) and Kato, Kobayashi, and Saita (2010) but, in contrast to these earlier studies, capturing data on the recent financial crisis from Laeven and Valencia (2010). The results suggest sizeable and highly nonlinear effects of aggregate bank leverage on the probability of financial crisis.
Combining the results from these two estimating equations enables simple calculations of the impact of a variety of tax reforms on the likelihood of financial crisis. Linking this, in turn, with estimates of the output loss that is historically associated with such crises gives some rough sense of the potential welfare gains from policies that mitigate debt bias in the financial sector. Putting aside the overarching debate as to the proper roles of taxation and regulation in addressing the potential for excess leverage in the financial sector,5 we consider three tax reforms that would reduce the tax incentive to debt finance: a cut in the corporate tax rate; adoption of an Allowance for Corporate Equity form of corporate tax (which would in principle eliminate debt bias); and a ‘bank levy’ of broadly the kind that a dozen or so countries have introduced since the crisis.6
All this gives a very different perspective on the nature and possible magnitude of the welfare costs associated with debt bias. Previous work, which has not reflected considerations of financial stability, has concluded that these are small: Gordon (2010) estimates the total efficiency loss from debt bias in the U.S. to be less than 1 percent of corporate income tax (CIT) revenue and concludes that: “tax distortions from corporate financial policy are not an important consideration when setting tax policy”; Weichenrieder and Klautke (2008) put the marginal welfare loss from debt bias somewhat higher, but still only at 0.06–0.16 percent of the capital stock. The question here is whether considerations of financial stability imply much higher welfare losses—and the conclusion will be that it seems they do.
The analysis here is in several respects simplistic and limited. In particular, we have not uncovered a direct link between tax incentives favoring debt finance and the probability of financial crisis. But the evidence presented here does suggest the real possibility of such a connection. If debt bias leads to higher aggregate bank leverage than would otherwise be the case—and it seems that it does—and if higher aggregate bank leverage makes financial crisis more likely—and it seems that it does—then debt bias increases the chances of financial crisis. This, in turn, can imply welfare gains from mitigating debt bias far higher than the small amounts found in previous work: noticeably more, in some of the calculations reported here, than 1 percent of GDP. Regulation, of course, has historically had the dominant role in addressing such problems of excess leverage in the financial sector, and the higher and tighter capital requirements of Basel III should to some degree reduce the welfare costs of debt bias. How much comfort is taken from this will depend on one’s evaluation of these reforms. What the evidence assembled here suggests, however, is that the tax incentive encouraging banks to use debt finance is not just an inelegant inconsistency with regulations intended to do the exact opposite, but a potential risk to be recognized, and, as need be, addressed, in the pursuit of financial stability.