1 Introduction
In the last years, the Organisation for Economic Co-operation and Development (OECD) and G20 have started an unprecedented initiative against Base Erosion and Profit Shifting (BEPS). This initiative builds on a large body of empirical research that has identified the main channels of BEPS (Dharmapala
2014). One of these channels is the strategic use of debt, exploiting the tax-deductibility of interest payments. Most of this empirical literature, however, excludes the financial sector. This omission is all the more surprising as, in many countries, the financial sector contributes around a quarter of corporate tax revenues.
1 The OECD has recognized this gap and published a discussion draft entitled “Approaches to Address BEPS Involving Interest in the Banking and Insurance Sectors” as part of Action 4 on the limitation of interest deductions in 2016.
2
The tax deductibility of interest payments enables both base erosion and profit shifting: When a bank borrows from a third party, the resulting interest payments lower its taxable profit. Higher risks for shareholders and regulatory requirements limit the extent to which banks can erode their tax base in this way. When a bank affiliate borrows internally, i.e. from a different subsidiary of the same bank group, profit is shifted to that subsidiary. The interest payment is tax deductible in the borrowing affiliate, while the interest income is taxed in the lending affiliate—usually located in a strategically chosen low-tax country. As internal debt does not increase the overall leverage of the bank group, it does not entail risks for the ultimate shareholders. Previous literature (de Mooij and Keen
2016; Heckemeyer and de Mooij
2017) has studied the tax responsiveness of external debt, i.e. base erosion. However, to our knowledge, no previous paper studies how banks can use (internal) interest payments to shift profits to low-tax jurisdictions.
Our paper aims to fill this gap. Using administrative data provided by the German central bank, we show that banks indeed use internal debt to shift profits to low-tax countries, and that they do so more aggressively than non-financial firms. Our dataset, the External Positions of Banks database, has information on bilateral financial flows between the foreign subsidiaries and branches of all German multinational banks. Using panel regressions with and without affiliate fixed effects, we find that a ten percentage points higher corporate tax rate increases the internal debt-to-assets ratio by about 5.1 percentage points. This absolute response is much stronger than the effect that Fuest et al. (
2011) and Buettner et al. (
2012) find for non-banks in a comparable setting.
3 Thus, the banking sector uses internal debt shifting more aggressively than other sectors of the economy.
A second contribution of this paper is methodological. It arises from the importance of conduit entities in internal debt financing. Such conduit affiliates pass through internal loans, without shifting any profits away from themselves. There are three potential reasons why multinationals might use conduit entities: First, the conduit entities might simply serve as financial hubs, coordinating internal financing. Second, passing loans through an additional affiliate also impedes the uncovering of the tax avoidance scheme by the tax authorities or the media. Third, the pass-through loans offer an additional profit shifting possibility by mispricing the related interest rates. Despite the existence of these pass-through loans, the literature on internal debt shifting uses internal gross liabilities as proxy for the volume of internal debt shifting. If the location of the conduit entities correlates with tax rates, this proxy systematically underestimates the true amount of debt shifting.
We show that conduit entities are indeed systematically located in low-tax countries. To account for the arising bias, we propose a new dependent variable, the internal net-debt-to-assets ratio. Briefly put, it captures internal liabilities net of internal claims relative to total assets. Using this variable, we estimate a slightly higher tax rate coefficient, even though the sample mean of the internal net-debt-to-assets ratio is substantially lower. We find that a ten percentage points higher corporate tax rate increases the internal net-debt-to-assets ratio by 5.7 percentage points, which corresponds to an increase of 20% at the mean.
Our results also add to the current policy debate on multinational taxation. A main policy-relevant finding is that profit shifting in the financial sector is substantial: The size of the financial sector in the economy, combined with the high tax elasticities found in this paper, implies a substantial potential revenue loss. At the same time, the main current anti-tax avoidance rules, controlled-foreign-corporation (CFC) and thin capitalization rules, either do not apply to or are not effective in the financial sector.
CFC rules stipulate that passive income from a low-tax subsidiary is added to the profit of the parent company and taxed there. Usually, interest payments on internal debt are passive income, but many countries have exemptions for banks, as interest income is their main business.
4 When interest income is deemed active income, CFC rules are toothless for banks. Moreover, CFC rules are not compatible with EU law and are currently not applied within the EU.
Thin capitalization rules usually limit the deductibility of net interest expenses to a certain fraction (often 30%) of an adjusted earnings measure. Several countries exempt banks from these rules (e.g. Belgium, France, Greece, Italy, Spain) or have more lax rules for banks (e.g. China, Japan, United Kingdom). In other countries (e.g. Germany, Portugal, USA), thin capitalization rules apply to banks; however, as these rules consider net interest (i.e. interest expense minus interest income), they are usually not relevant for banks due to the high levels of interest income. Summing up, the existing anti-avoidance rules tackling debt shifting are not effective for banks even though they also use this tax avoidance channel. Therefore, there is a need for specific anti-tax avoidance rules in the financial sector. Such rules could, for example, limit the deductibility of internal interest payments.
This paper contributes to three different strands of literature. First, it adds to a large literature on profit shifting (recently surveyed by Beer et al.
2020). Within this literature, several papers focus on the use of internal debt, e.g. Fuest et al. (
2011), Buettner et al. (
2012), Buettner and Wamser (
2013), Blouin et al. (
2014), Egger et al. (
2014) and Overesch and Wamser (
2014). Similarly to our paper, Overesch and Wamser (
2014) also use bilateral internal debt data; they also find significantly positive effects of the bilateral tax rate differential, the most precise measure for debt shifting incentives. We contribute to this literature in two ways: As a first contribution, we point out the importance of modelling conduit entities, showing that the dependent variable should be net (and not gross) internal debt. As a second contribution, we focus on banks, an industry where debt financing plays a particularly large role.
Second, we contribute to the emerging literature focusing on profit shifting in the banking sector. Merz and Overesch (
2016) show in a worldwide sample of bank affiliates that corporate tax rates are negatively associated with reported pre-tax profits, indicating that banks indeed engage in profit shifting. While they cannot identify precise profit shifting channels in their data, they find some suggestive evidence that internal debt shifting might play a role. Langenmayr and Reiter (
2017) identify another potential channel by showing that banks shift profits through the relocation of proprietary trading assets to lower-taxed affiliates.
Lastly, we contribute to the literature studying how taxes affect capital structure choices in the financial sector. While there is plenty of evidence on how taxes affect the capital structures of non-financial firms (see, for example, the meta-analysis of Feld et al.
2013), only a few papers look at the financial sector, despite the importance of bank leverage as a risk factor for financial crises. de Mooij and Keen (
2016) first show analytically that higher tax rates should increase both conventional debt as well as hybrid financing in the banking sector, a result reflecting the tax deductibility of interest payments. They confirm this relationship empirically for conventional debt, but not for hybrid financial instruments. Heckemeyer and de Mooij (
2017) compare the responsiveness of debt finance to taxation for banks and non-banks. They find that while small banks are more responsive to taxation than non-banks of similar size, large banks are less responsive compared to both smaller banks and similarly sized non-banks. Gu et al. (
2015) regress overall debt-to-assets ratios on the difference between the tax rate an affiliate faces and the group’s average tax rate. As they cannot distinguish between internal and external debt they also cannot separate this effect into internal debt shifting and the classical debt financing incentive. In contrast, our focus is on the tax responsiveness of
internal debt, which we observe directly and at a bilateral level in the Bundesbank data. The presumed underlying motive is to shift profits from a high-tax to a low-tax location without increasing the indebtedness of the bank group as whole. We will discuss below how this decision interacts with the conventional tax shield motive to take on additional (external) debt in response to higher tax rates.
The next section discusses relevant institutional issues and the role of conduit entities. Section
3 presents the empirical specification that we use for identification. Section
4 describes the data and provides descriptive evidence, and Sect.
5 presents regression results. Section
6 concludes.
5 Results
Table
4 shows the baseline estimation results for the determinants of the internal debt variables in affiliates and headquarters of German multinational banks. For comparability with previous studies, the dependent variable in column (1) is the ratio of internal liabilities to total (external) assets. We find a significant positive coefficient of 0.508 for the corporate tax rate, indicating that a 10 percentage points higher corporate tax rate implies an increase in the internal liabilities to total assets ratio by about five percentage points. At the mean (42.4%) this corresponds to an increase by 12%. This effect of corporate tax rates on internal liabilities in the banking sector is quantitatively larger than previous studies estimated for other sectors relative to the sample mean: Fuest et al. (
2011) and Buettner et al. (
2012) use a similar setting with data on German multinationals and find coefficients for the corporate tax rate of only 0.177 and 0.214, respectively. A 10 percentage points tax rate increase in these studies implies at the sample means (23% and 28%) an increase in the internal debt ratio by around 7% to 8%. One caveat in the comparison of our results to previous studies is that we have to approximate total assets by total external assets. As we compare changes at the mean, this approximation is innocuous if the ratio of total external assets to total assets remains constant over time.
We can also compare our results to the literature on the effect of taxes on banks’ capital structure. This literature finds marginal effects of the tax rate on the total debt-to-assets ratio between 0.25 and 0.3 (Gu et al.
2015; de Mooij and Keen
2016). In further analyses, de Mooij and Keen (
2016) show that the tax effect is twice as large after 2007; Heckemeyer and de Mooij (
2017) find that small banks react much more strongly than larger banks.
20 Whereas estimates on the total debt ratio and on internal debt are not directly comparable, a comparison with these results also suggests that banks’ internal debt reacts particularly strongly to taxation.
21
The greater impact of tax rates on internal debt in the financial sector becomes even clearer if we use internal net debt as the dependent variable in column (2). This variable reflects the effective amount of debt that shifts profits out of an affiliate. As shown in the previous section, as conduit entities in internal debt financing are mainly located in low-tax countries, ignoring conduit debt results in a downward biased estimate of the tax coefficient when using internal gross liabilities as proxy for debt shifting. The tax coefficient in column (2) is 0.565, which is about 11% larger than the estimate in column (1). A Wald test shows that the coefficient estimate in column (2) is significantly larger than the coefficient estimate in column (1) at a significance level of 10%. While the relatively low significance level implies that one should interpret this result cautiously, it does nevertheless suggest that using internal net debt as the dependent variable tends to provide stronger results than using the traditional internal liabilities variable. At the sample mean (28.0%), a 10 percentage points higher corporate tax rate implies an increase in the internal-net-debt-to-total-assets ratio by 20%. Previous literature has not analysed the tax response of internal net debt, and therefore comparability to non-financial sectors is limited in column (2). However, as also non-banks use conduit entities (e.g. internal financing hubs), accounting for conduit debt is an interesting extension for future research on debt shifting in non-financial sectors.
Table 4
Baseline intragroup debt regressions
CTR | 0.508* | 0.565** | 0.363** | | 0.468* | 0.511** | 0.344** | |
(0.270) | (0.225) | (0.146) | | (0.265) | (0.214) | (0.151) | |
GroupCTR | | | | − 0.540** | | | | − 1.036*** |
| | | (0.219) | | | | (0.252) |
Ln(TA) | − 0.024*** | − 0.015*** | 0.009** | 0.006* | − 0.004 | 0.006 | 0.016*** | 0.013*** |
(0.007) | (0.006) | (0.003) | (0.003) | (0.007) | (0.006) | (0.004) | (0.004) |
Ln(GDP) | 0.019 | − 0.000 | 0.010 | − 0.006 | 0.022 | 0.000 | 0.020 | 0.007 |
(0.015) | (0.012) | (0.023) | (0.023) | (0.014) | (0.011) | (0.024) | (0.024) |
Inflation | − 0.007* | − 0.005** | − 0.000 | − 0.000 | − 0.013*** | − 0.008*** | 0.000 | − 0.000 |
(0.004) | (0.003) | (0.001) | (0.001) | (0.004) | (0.003) | (0.000) | (0.001) |
GDP growth | − 0.010** | − 0.008** | − 0.001 | − 0.001* | − 0.011*** | − 0.009*** | − 0.001 | − 0.001 |
(0.004) | (0.004) | (0.001) | (0.001) | (0.004) | (0.003) | (0.001) | (0.001) |
Fin. sector share | 0.879*** | 0.102 | 1.409*** | 1.552*** | 0.495 | − 0.285 | 1.543*** | 1.744*** |
(0.318) | (0.230) | (0.225) | (0.239) | (0.310) | (0.219) | (0.227) | (0.242) |
Reg. index | − 0.018** | − 0.005 | | | 0.003 | 0.018*** | | |
(0.009) | (0.008) | | | (0.008) | (0.007) | | |
Cap. req. | 0.221 | 1.256 | | | − 0.544 | − 0.544 | | |
(2.735) | (1.975) | | | (2.451) | (1.775) | | |
Time FE | \(\checkmark \) | \(\checkmark \) | \(\checkmark \) | \(\checkmark \) | \(\checkmark \) | \(\checkmark \) | \(\checkmark \) | \(\checkmark \) |
Bank group FE | \(\checkmark \) | \(\checkmark \) | \(\checkmark \) | \(\checkmark \) | \(\checkmark \) | \(\checkmark \) | \(\checkmark \) | \(\checkmark \) |
Bank FE | | | \(\checkmark \) | \(\checkmark \) | | | \(\checkmark \) | \(\checkmark \) |
\(R^2\) | 0.359 | 0.332 | 0.799 | 0.795 | 0.403 | 0.484 | 0.781 | 0.777 |
Observations | 21,896 | 21,896 | 21,893 | 21,884 | 16,187 | 16,187 | 16,184 | 16,181 |
In column (3) we additionally include bank affiliate fixed effects. Qualitatively we can confirm that multinational banks shift profits through the use of internal debt; however the estimated coefficient is smaller. As fixed effects models assume that the full effect of a tax rate change on internal debt takes place within the same period, the smaller coefficient may indicate that banks respond more slowly.
22 In addition, the fixed effects specification uses only tax rate changes for identification, and the bulk of these changes took place in high-tax countries. As tax havens (the potential destinations for profits) still offer a much lower tax rate (but did not change their tax rate), the tax incentive for internal debt structures often changes little if only countries with medium or high tax rates make changes. This results in relatively low adjustments to tax rate changes (see also Davies et al.
2018). Still, the estimated coefficient indicates a strong response of internal net debt to corporate tax rates: A ten percentage points rise in the tax rate implies an increase in the internal net debt ratio by 3.63 percentage points, corresponding to an increase by about 13% at the sample mean.
Instead of the tax rate affiliate i is facing, in column (4) we use the average tax rate in the bank group as our main explanatory variable. Relying on variation in foreign taxation allows us to more carefully disentangle profit shifting from tax debt shielding. The group tax coefficient in column (4) is negative and statistically significant, indicating that an increase in the average group tax rate lowers internal debt. If the average group tax rate increases, shifted profits are on average taxed at a higher rate, reducing the incentive to increase internal lending for profit shifting.
Columns (5) to (8) show the results of re-estimating the four specifications with exclusion of German headquarters. This accounts for the idiosyncrasy of the External Positions of Banks database that all headquarters reside in Germany, and Dischinger et al. (
2014) show that multinationals might be reluctant to shift profits away from headquarters. However, we find smaller tax coefficients (0.468, 0.511 and 0.344) when excluding headquarters from our sample of German multinational banks. There are several potential explanations for this finding: First, banks might use debt shifting to substantially shift profits out of their German headquarters. This would be in line with Tørsløv et al. (
2020) who show that the share of corporate tax revenues lost due to profit shifting in Germany is the highest among all European countries. Second, headquarters partially finance their foreign affiliates with internal debt, leading to a ‘base’ stock of internal debt in these affiliates that does not respond to tax rates and leads to the smaller estimated responses in regressions (5) to (7).
Results on control variables furthermore show a small negative effect of an affiliate’s size (measured in total external assets) on the use of internal debt in the full sample without affiliate fixed effects, but estimates in the subsample of foreign affiliates are insignificant. When including affiliate dummies, the effect gets slightly positive. Inflation rates in the host country have a significantly negative impact on both the internal-gross-liabilities-to-total-assets ratio and the internal-net-debt ratio, perhaps reflecting higher risks. When including affiliate fixed effects, this effect vanishes. A negative effect also arises from GDP growth, possibly because banks do not shift funds away from affiliates in fast growing countries. As expected, the share of the financial sector in a country’s gross value added has a significantly positive effect on the internal-gross-liabilities-to-total-assets ratio in regression (1). However, on the internal-net-debt ratio we can only find a positive effect when including bank affiliate fixed effects.
Table 5
Bilateral regression results
\(CTR_{it} - CTR_{jt}\) | 0.033*** | 0.042*** | 0.059*** | 0.094*** |
(0.007) | (0.008) | (0.008) | (0.010) |
Ln(Total assets) | − 0.009*** | − 0.008*** | − 0.005*** | − 0.008*** |
(0.000) | (0.002) | (0.001) | (0.003) |
Ln(GDP) |
Host country i | 0.002*** | 0.005 | 0.001 | 0.005 |
(0.001) | (0.005) | (0.001) | (0.007) |
Counterpart j | 0.008*** | 0.006*** | 0.012*** | 0.013*** |
(0.001) | (0.001) | (0.001) | (0.001) |
Inflation rate |
Host country i | − 0.002*** | − 0.001** | − 0.002*** | − 0.000 |
(0.000) | (0.000) | (0.000) | (0.000) |
Counterpart j | − 0.001*** | − 0.001*** | − 0.001*** | − 0.001*** |
(0.000) | (0.000) | (0.000) | (0.000) |
GDP growth |
Host country i | − 0.001*** | − 0.000** | − 0.001*** | − 0.000 |
(0.000) | (0.000) | (0.000) | (0.000) |
Counterpart j | − 0.001*** | − 0.001*** | − 0.002*** | − 0.002*** |
(0.000) | (0.000) | (0.000) | (0.000) |
Regulatory index |
Host country i | − 0.000 | | 0.001*** | |
(0.000) | | (0.000) | |
Counterpart j | 0.004*** | 0.003*** | 0.007*** | 0.006*** |
(0.001) | (0.001) | (0.000) | (0.000) |
Capital requirement |
Host country i | 0.453*** | | 0.397*** | |
(0.079) | | (0.072) | |
Counterpart j | − 0.344*** | − 0.281*** | − 0.375*** | − 0.365*** |
(0.026) | (0.023) | (0.036) | (0.035) |
Financial sector share |
Host country i | 0.077*** | − 0.143*** | 0.067*** | − 0.161** |
(0.012) | (0.053) | (0.012) | (0.063) |
Counterpart j | 0.031*** | 0.026*** | 0.045*** | 0.038*** |
(0.004) | (0.003) | (0.008) | (0.006) |
Monthly time FE | \(\checkmark \) | \(\checkmark \) | \(\checkmark \) | \(\checkmark \) |
Bank group FE | \(\checkmark \) | \(\checkmark \) | \(\checkmark \) | \(\checkmark \) |
Bank FE | | \(\checkmark \) | | \(\checkmark \) |
\(R^2\) | 0.078 | 0.231 | 0.146 | 0.241 |
Observations | 107,361 | 107,361 | 57,628 | 57,628 |
Table
5 shows results of the bilateral debt shifting regressions that allow use of the precise corporate tax rate differential as a measure for the shifting incentive. For this tax rate differential a significantly positive effect on bilateral-internal-net-debt-to-total-assets of 0.033 arises in the baseline regression, and of 0.042 when including affiliate fixed effects. In the subsample of foreign affiliates these effects are even larger: In regression (3) a coefficient of 0.059 arises, meaning that a 10 percentage points higher corporate tax rate differential leads to an increase in the bilateral-internal-net-debt ratio by 0.59 percentage points. Compared to the sample mean (3.2% in foreign affiliates) this corresponds to an increase of 18%. This result is in line with banks shifting profits through internal debt from higher-taxed to lower-taxed affiliates. Moreover, this implied semi-elasticity also quantitatively confirms the 20% increase in internal net debt ratio (in response to a 10 percentage points corporate tax rate increase) which we find in Table
4. When controlling for affiliate fixed effects in column (4), the estimated coefficient increases even further to 0.094.
Note that in the bilateral debt regressions in Table
5 the estimated tax effect increases when including bank affiliate dummies, whereas with aggregate internal net debt as the dependent variable in Table
4 the tax effect is smaller with bank affiliate dummies. This implies that internal net debt is highly responsive to changes in the internal counterpart’s tax rate (the potential destination for profits), whereas banks do not respond equally strongly to changes in the host country’s tax rate.
Results on host country control variables of affiliate
i are qualitatively similar to the estimates for aggregate debt data in Table
4. In bilateral regressions we also include macroeconomic control variables for the country from which the internal net debt is taken. For the internal counterpart’s country we find a positive effect of the GDP that probably comes from the fact that German banks partially finance a stronger engagement in large countries through internal debt. Interestingly, the capital requirement in the internal counterpart’s country has a significantly negative effect on bilateral internal net debt: Additional claims have to be backed by additional equity to fulfil capital requirements; hence a higher capital requirement can discourage internal lending.
To summarize, both aggregate and bilateral internal debt regressions on German multinational banks indicate that banks engage in debt shifting. Moreover, the estimated effect in the banking sector is larger than previous studies estimated for non-financial firms, both absolutely and relative to the sample average of internal debt ratios. This becomes even clearer when we correct for conduit entities: Since conduit affiliates are taxed lower than the sample average, using the internal-net-debt ratio as the dependent variable leads to even larger estimated tax responses. Accounting for conduit debt is also a more general methodological issue that can be addressed by future empirical internal debt shifting studies on non-banks.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.