Tax Reforms, Debt Shifting and Corporate Tax Revenues: Multinational Corporations in Canada

Transcription

1 Tax Reforms, Debt Shifting and Corporate Tax Revenues: Multinational Corporations in Canada Vijay Jog Professor of Finance School of Business, Carleton University Jianmin Tang conomist Micro-conomic Policy Analysis, Industry Canada February 998 WORKING PAPR 97-4 Prepared for the Technical Committee on Business Taxation Working papers are circulated to make analytic work prepared for the Technical Committee on Business Taxation available. They have received only limited evaluation; views expressed are those of the authors and do not necessarily reflect the views of the Technical Committee, or the Department of Finance.

2

3 Tax Reforms, Debt Shifting and Corporate Tax Revenues: Multinational Corporations in Canada Vijay Jog Professor of Finance School of Business, Carleton University Jianmin Tang conomist Micro-conomic Policy Analysis, Industry Canada February 998 WORKING PAPR 97-4 Prepared for the Technical Committee on Business Taxation Comments on the working papers are invited and may be sent to: Paul Berg-Dick, Director Business Income Tax Division Department of Finance Ottawa, Ont. KA 0G5 Fax: (63)

4 Vijay Jog School of Business Carleton University 25 Colonel By Drive Ottawa, Ontario KS 5B6 Fax: (63) Jianmin Tang Micro-conomic Policy Analysis Industry Canada 235 Queen Street Ottawa, Ontario KA 0H5 Fax: (63) For additional copies of this document please contact: Distribution Centre Department of Finance 300 Laurier Avenue West Ottawa KA 0G5 Telephone: (63) Facsimile: (63) Also available through the Internet at Cette publication est également disponible en français.

5 Abstract An analysis of Canadian corporate income tax revenues during the period shows a relative shifting of tax revenue shares between Canadian and foreign-controlled corporations, and a substantial change in the debt levels of foreign-controlled corporations, as well as Canadian-based multinationals. In this paper, we explore the hypothesis that these changes in debt levels may have been associated with the tax reforms undertaken by the United States and Canada in the mid-980s. Our empirical analysis places special emphasis on the differences between Canadian-controlled corporations and foreign-controlled corporations. We further separate these two groups into those with foreign affiliates and those without. We present evidence based on the universe of Canadian corporations, as well as that from a longitudinal data set created especially for this paper. We develop a theoretical two-country integrated model to show the directional impact of the tax reforms on the capital structure of corporations and the corresponding impact on Canadian tax revenues. Our results are consistent with the predictions of our theoretical model. Our data reveal that a significant shifting of debt by foreign-controlled firms in Canada, as well as Canadian firms with foreign affiliates has occurred during the period. Also evident is a contemporaneous decline in Canadian corporate income tax revenues from multinationals as revealed by taxes paid relative to operating income before interest and taxes. Analysis of our longitudinal sample and econometric analysis supports these conclusions. xcept for Canadian-controlled corporations with no foreign affiliates, we observe a substantive increase in the reliance on debt suggesting a corresponding decline in Canadian tax revenues. Since a large proportion of the foreign controlled corporations is U.S.-controlled, tax reforms undertaken by the United States in the 980s could be a potential causal factor for this debt shifting. Our results imply that the relative change in the tax rates between Canada and the United States has had a negative impact on the Canadian corporate income tax base. Noting the fact that the debt levels of the Canadian subsidiaries of multinationals are, on average, still much lower than their Canadian counterparts, it is possible that the Canadian corporate income tax base may continue to remain under pressure due to further adjustments to the debt levels.

6

7 Table of Contents. Introduction The U.S. and Canadian Tax Reforms Theoretical Considerations mpirical Analysis Data Source Aggregate vidence conometric Analysis Conclusions... 7 Appendix: One-parent, One-foreign-subsidiary, Two-period Model... 2 References... 36

8

9 Tax Reforms, Debt Shifting and Corporate Tax Revenues. Introduction The 980s saw significant tax reforms undertaken by Canada and the United States as well as other Organization for conomic Co-operation and Development (OCD) countries. These tax reforms have resulted in Canada becoming a relatively high-tax jurisdiction as compared with its trading partners, particularly the United States. Although both Canada and the United States saw an overall reduction in corporate tax rates, the degree of reduction reversed Canada s position from a relatively low-tax to a relatively high-tax jurisdiction as illustrated by Figure. For example, the U.S. tax reform reduced the combined federal-state statutory U.S. corporate tax rate from 50 to 39 percent. In contrast, Canadian tax reform reduced the comparable Canadian rate from approximately 5 to 42 percent for non-manufacturing and 44 to 35 percent for manufacturing., 2 Furthermore, the tax reform in the United States has resulted in changes not only to the statutory corporate tax rates but also to the tax rules for U.S. multinational corporations (MNCs) with respect to the tax treatment associated with the income of their foreign subsidiaries. These changes have been of particular importance to Canada, which has a significant number of U.S.-controlled corporations (USCCs). For example, as of 994, foreign-controlled corporations (FCCs) represented 22 percent of total taxable income in Canada, 23 percent of total corporate These statutory tax rates take into account the federal, as well as state and provincial tax rates. These rates are not fully representative of the situations faced by individual firms, as these would vary from state (province) to state (province) and, in certain instances, industrial sector (e.g. due to industry-specific tax provisions). The numbers shown in Figure for Canada are based on statutory corporate tax rates faced by a sample of 457 large corporations in Canada (the sample will be described in detail later on). These corporations represent both the manufacturing and the non-manufacturing sectors but exclude the financial industry; therefore, the average rate represents a blend of manufacturing and non-manufacturing rates. The statutory corporate tax rate of each corporation is calculated as the sum of federal and provincial tax rates. The provincial tax rate is calculated as the average of the relevant provincial statutory corporate tax rates, weighted by its taxable income in each province. Because of a lack of data, calculations for the U.S. tax rates are simply based on the formula: tax rate = federal rate average state rate x (-federal rate). We assume a reduction from 46 percent pre-reform to a 34 percent post-reform federal rate and a 7-percent average state tax. 2 It may be noted that, in addition to lowering statutory tax rates, tax reforms in both countries introduced several base-broadening measures. For example, the U.S. tax reform eliminated investment tax credits and imposed restrictions on capital cost allowances through modifications to the accelerated cost recovery system (ACRS). In Canada, investment tax credits were effectively eliminated (except for R&D), and capital cost allowances were reduced and streamlined.

10 2 WORKING PAPR 97-4 tax revenues and 4 percent of total corporate assets. 3 Although corresponding figures for USCCs are not available, approximately 90 percent of foreign-controlled corporations are U.S.-controlled. The changes in corporate tax revenues attributed to FCCs along with their share of taxable income and assets are shown in Table and Figure 2. In Table, we list total taxable income, total corporate income tax and total assets of all Canadian corporations, by control for 984 through 994. In 984, FCCs represented 3 percent of total corporate assets in Canada rising marginally to 4 percent in 994. The corresponding figures for taxable income were 28 percent in 984 and 2 percent in 994; whereas, for taxes paid, these figures were 33 percent in 984 and 23 percent in 994. Although the asset share of FCCs increased marginally, the taxable income share of FCCs declined by 7 percentage points, and the tax share declined by 0 percentage points. Thus, over the sample period, FCC tax payments have been dramatically reduced relative to those of their Canadian-controlled counterparts. Figure 2 shows the corporate tax-to-assets ratio for the two groups over this -year period. For Canadian-controlled corporations (CCCs) the value of the ratio fell from 0.46 percent to 0.39 percent, only a 5-percent reduction. For FCCs, this ratio fell dramatically from.55 percent to 0.7 percent, a 54-percent reduction. 4 If FCCs had maintained in 994 the same corporate tax-to-asset ratio as in 984, Canadian tax revenues from FCCs would have been $6.4 billion instead of the actual amount of $2.9 billion, a difference of $3.5 billion. The main thrust of this paper is to investigate whether the changes in the relative tax rates between the United States and Canada had an impact on debt levels of Canadian subsidiaries of foreign multinationals as well as Canadian multinationals. If this is found to be the case, then this debt shifting can also provide an explanation of significant changes in Canadian corporate 3 In the past, these numbers were available in the Statistics Canada publication #6-20. The publication was changed significantly in the early 990s, and no longer publishes this type of information for years after Figure 2 shows that foreign-controlled corporations always have a higher tax-to-assets ratio than Canadian-controlled corporations. This occurs because that the Canadian-controlled group contains many Canadian-controlled private corporations (CCPCs), which enjoy a lower tax rate due to lower statutory tax rates and additional tax deductions (e.g. R&D credits).

11 Tax Reforms, Debt Shifting and Corporate Tax Revenues 3 income tax revenues. The investigation is conducted by comparing the capital structure of CCCs without foreign affiliates with that of all other corporations that is CCCs with foreign affiliates and FCCs/USCCs. Our hypothesis is that the increase in debt by USCCs (and Canadian multinationals) occurred as a direct result of changes in relative tax rates. This debt shifting resulted in higher deductions of interest, thereby resulting in lower tax payments in Canada, everything else being equal. 5 Note that such debt-shifting may affect both the amount of tax revenues and their volatility. This paper examines the debt-shifting behaviour of FCCs and CCCs by using a longitudinal data set that was made available to us by Finance Canada. (The data set will be described in detail later on). Our results suggest that the changes in tax systems resulting from U.S. tax reform indeed influenced the financial policy of multinationals in Canada. Canadian corporate tax rate also played a significant role in determining a corporation s debt-to-asset ratio; the higher the corporate tax rate faced by these corporations, the higher the debt-to-asset ratio. We show that, from both theoretical and empirical perspectives, the tax difference between Canada and the United States does have a positive impact on a corporation s debt-to-asset ratio, indicating a substantial debt-shifting response and a corresponding impact on Canadian corporate tax revenues. This paper is organized as follows. In the next section, we provide a brief overview of the tax reforms in Canada and the United States. In Section 3, we conduct a brief review of the existing literature and provide a theoretical analysis of changes to financial behaviour of a multinational enterprise that is facing changes in tax regimes. In Section 4, we describe the data set used, our econometric analysis and our results. Our conclusions are summarized in Section 5. 5 Three well-known instruments can be used by multinationals to minimize their tax payments. First, multinationals can shift profits from high-tax countries to low-tax countries by manipulating transfer prices on cross-border interplant transactions. For instance, charging a below-cost price on products and services provided by a plant in a high-tax jurisdiction to a plant in a low-tax jurisdiction reduces the profit of the former and increases the profit of the latter (Grubert, 997). Second, multinationals can defer their home-country tax liabilities on repatriated foreign dividend income by reinvesting the profits that would have been used to pay such dividends, in active assets, passive assets or both in host countries (Weichenrieder, 996). Finally, multinationals can reduce their overall taxes by reallocating their debt from a low-tax jurisdiction to a high-tax jurisdiction to take advantage of the high interest deduction in the high-tax jurisdiction. This third possibility is the subject of this paper.

12 4 WORKING PAPR The U.S. and Canadian Tax Reforms 6 The U.S. tax reform of 986 changed not only the statutory corporate tax rates in the United States but also the tax treatment of U.S.-controlled foreign subsidiaries. The Canadian tax reform in 987 made changes to the statutory corporate income tax rates while broadening the base subject to tax, but did not change the Canadian tax treatment of repatriated income from a USCC to its parent. The results of these tax reforms resulted in three major changes affecting the tax planning of a USCC from the viewpoint of its parent, and consequently Canadian corporate tax revenues. First, the post-reform period resulted in the reduction of the average combined federal-state U.S. tax rate from 5 percent to 39 percent, a reduction of 2 percentage points. The Canadian tax reforms resulted in a reduction of 7 percentage points, on average, from an overall pre-reform rate of 5 percent to a post-reform rate of 44 percent for the non-manufacturing sector and 44 to 35 percent for the manufacturing sector. Although the exact rates applicable to individual firms may vary, the direction of the change was clear. In the pre-reform period, Canadian tax rates were lower than the U.S. tax rates whereas the post-reform period changed it in favour of the U.S. Second, the U.S. tax reform changed the U.S. foreign tax credit (FTC) system. As these changes have direct implications for the analysis in this paper, these deserve further elaboration. Prior to the U.S. reform, the U.S. parent could average foreign-taxed income across low- and high- (relatively) tax countries in an overall FTC basket calculation. 7 This averaging implied that excess foreign tax credits could be used to offset U.S. tax liability on income received from a low-tax country. The U.S. tax reform made fundamental changes to this procedure. While continuing to allow pooling of income across countries, it introduced new separate baskets for determining the total FTC limitations. These separate baskets introduced additional restrictions on the ability of the U.S. parent to minimize its U.S. tax liability on a worldwide basis. 6 As noted in the Introduction, this section provides only a brief overview of these two tax reforms. Detailed descriptions of these reforms are available in many papers, including Ault and Bradford (990), Bruce (989), Hogg and Mintz (99), and Jog and Mintz (989). 7 The foreign tax credit includes the amount by which the creditable foreign income tax and withholding tax exceeded the U.S. tax on that income.

13 Tax Reforms, Debt Shifting and Corporate Tax Revenues 5 Moreover, Canada moved from a low-tax country to a high-tax country. This change reduced the incentive of the foreign parent not to repatriate income from its Canadian subsidiary. 8 Thus, if Canada was considered to be a foreign deficit tax credit country prior to the tax reform, it did not remain so after the tax reform. 9 Third, there were changes to the rules governing allocation of interest between U.S. and foreign-source income for the purpose of calculating the FTC limitation. This limit is calculated as notional U.S. tax payable on net foreign-source income. 0 Prior to the tax reform, a taxpayer could apportion interest either on the basis of the value of assets generating U.S. and foreign-source income or on the basis of the amount of gross income generated. Corporate groups filing consolidated returns could allocate interest on a separate company basis. This provided scope to increase the amount of foreign tax credits available for the consolidated group. Post-reform, the allocation of interest was made on the basis of the value of assets (for tax purposes) and on a consolidated basis. Therefore, interest expense incurred by any member of a corporate group proportionally reduced the U.S. and foreign-source income for all members of the group on the basis of the value of assets generating income. Interest allocated against foreign-source income was therefore apportioned to each of the FTC baskets. This approach, which is based on the premise that money is fungible, limited the extent to which corporate groups could engage in tax planning to maximize the amount of available foreign tax credits. From the viewpoint of U.S. taxation, this rule was neutral with respect to the financing decisions of the corporate group. This implied that, given the absence of any tax in the source country, U.S. subsidiaries should be expected to have a capital structure that was more consistent with the underlying business risk rather than a tax-planning perspective. These changes implied that the Canadian subsidiaries were now likely to be capitalized at a higher debt-to-equity ratio and there 8 There are also non-tax reasons why dividend payments can increase. See Hines and Hubbard (990). 9 The pre-tax reform case for not repatriating dividends due to the benefit of the deferral prior to the tax reform is well described in Hines and Hubbard (990), and Leechor and Mintz (993). 0 See Altshuler and Mintz (995), Arnold et al (996) and dgar (987).

14 6 WORKING PAPR 97-4 would be an increase in dividend repatriation. The former would lead to a reduction in Canadian tax revenues, since the interest expenses of the subsidiary would increase as well. Tables 2 and 3 show an illustrative example of the impact of these changes on a USCC, assuming some basic tax parameters, as viewed from its U.S. parent. Table 2 shows the values of the base parameters used. Table 3 shows a simplified income statement for three cases: case represents pre-reform base case. Case 2 represents a post-reform situation with no changes in capital structure. Case 3 represents a post-reform situation where the debt-to-equity ratio is changed so as to revert back to the same FTC position as the pre-reform period. The example assumes that the subsidiary has $3,000 in assets financed by 50-percent equity (parent contributions and cumulative retained earnings) and 50-percent debt from Canadian sources (i.e. non-parent debt) Case. In all cases, all earnings are assumed to be paid out as dividends; this assumption has no impact on the basic conclusions. In Table 3, the first column (Case ) in the simplified income statement provides the base case (pre-reform) situation. In this case, the U.S. parent was, in effect, indifferent between the choice of the capital structure for its Canadian subsidiary. One could even argue that, due to the small positive difference between the U.S. and Canadian tax rates, the U.S. parent may have preferred to: a) raise debt at the parent level; b) finance its Canadian subsidiary through equity investment; and c) not withdraw dividends so as not to have an excess FTC position. Column 2 (case 2) shows the simplified statement after the tax reform assuming no changes in the USCC capitalization. As can be seen, the U.S. parent receives higher income (due to reductions in tax rates in both countries), but increases its FTC from $6 to $7. However, it is possible for the U.S. parent to further decrease the subsidiary's total tax payment by increasing the debt assumed by the Canadian subsidiary. For example, if the U.S. parent was satisfied with the previous FTC position, it can get to that position by increasing the subsidiary's debt to its U.S. parent. The third column indicates that this would increase its interest deductions in Canada thereby further reducing its Canadian taxes. Column 4 shows the net impact of this change in the capital structure. Thus, if the U.S. parent chooses to maintain its pre-reform net tax position, this The actual decision of the U.S.-based multinational would also depend on whether it faces an excess credit or an excess limitation situation. For further details, see Hartman (985) and Slemrod (995).

15 Tax Reforms, Debt Shifting and Corporate Tax Revenues 7 would have an adverse impact on Canadian corporate tax revenues. For example, according to the case in Table 3, the Canadian government would see a reduction of its total tax revenues (as a result of debt shifting alone) to $67 from $82 a reduction of approximately 8 percent. Correspondingly, the U.S. parent s after-tax income increases from $02 to $3, an increase entirely at the expense of the Canadian government as a result of debt shifting. Clearly, the negative impact described in this hypothetical example can be considered speculative at this stage, although there is an incentive to move in the direction represented by the shift from Case 2 to Case 3. It should also be noted here that the above discussion focusses mainly on the capital-structure decisions of USCCs that may be affected by the tax reforms in both countries. There are, of course, other important impacts of the tax reforms, including their impact on real (production and investment) decisions by the U.S. parent in Canada, as well as their impact on transfer pricing policies between the U.S. parent and its Canadian subsidiary. 2 These changes in relative tax rates also impact capital-structure decisions of Canadian multinationals with foreign affiliates. In Section 4, we concentrate on the empirical evidence that allows investigation of the magnitude of these impacts. 3. Theoretical Considerations The impact of changes to corporate taxation in home and host country on the behaviour of MNCs has been a subject of great interest, especially in light of increasing globalization. The literature has centred almost exclusively on the effect of a home-country tax system on a foreign subsidiary s investment decisions. 3 In one of the earliest papers, Horst (977) examined the impact of changing U.S. tax policy (repealing deferral, increasing R&D charges to foreign 2 For the discussion of the impact on investment decisions, please see Hartman (985); for transfer pricing issues, please see Grubert et al. (99). Also see footnote 6. 3 There is a vast amount of literature on the impact of taxation on MNC decisions, including those related to investments, transfer pricing, tax competition, capital mobility, cost of capital and capital sourcing. There is also a large body of literature on optimal capital structure from a purely domestic perspective. A complete review of these strands of literature is clearly beyond the scope of this paper. Some relevant papers are included in the list of references at the end of the paper.

16 8 WORKING PAPR 97-4 subsidiaries and repealing the FTC) on U.S. MNCs. Sinn (984) and Hartman (985) studied the effect of a home country tax system including deferral and an FTC on the mature subsidiary s investment decisions, and showed that the home-country tax system is irrelevant, once a subsidiary has become mature. This area of study has since been extended in several directions. Sinn (993) has explored the investment decisions of an immature subsidiary, and concludes that the overall taxes on cross-border profit repatriations tend to reduce the birth weight of a subsidiary and increase the phase of its growth to maturity. Hines (994) incorporates debt in financing foreign subsidiaries and finds that the availability of debt finance makes it unlikely that a home-country tax system reduces investments by subsidiaries, even in the early stages of investments. 4 With the exception of Horst (977), there have been only a few studies that take into account the interaction between a parent and its subsidiary. Unlike other studies that single out a subsidiary, Horst considers both the parent and the subsidiary simultaneously and assumes that a multinational enterprise strives to maximize its consolidated after-tax income. In his words, the consolidated after-tax income is the parent s after-tax income (which includes dividend income from its foreign subsidiary) plus the subsidiary s retained earnings. With the consolidated after-tax income as the objective, his model captures some interaction between the parent and its foreign subsidiary. The extent of its ability to do so, however, is limited by the static nature of his model. In this paper (and unlike Horst), we offer a two-period model of a multinational enterprise with a parent and a foreign subsidiary, so that the dynamic relationship between the parent and its subsidiary is captured. We also assume that the multinational enterprise strives to maximize its consolidated after-tax income. This feature is important because intertemporal manipulation (deferral, for example) is one of the main instruments used by the multinational to avoid tax payments. Our simple model (see Appendix) allows us to evaluate the impact of tax-regime changes in the home and host countries on the financing of multinational firms. The main assumptions and conclusions from our theoretical model are as follows: 4 Note that most of these models assume that the repatriation tax on dividends is exogenous and constant over time.

17 Tax Reforms, Debt Shifting and Corporate Tax Revenues 9 (i) (ii) (iii) (iv) The parent is financed in the first period as much as possible by the existing stock of equity (or as little as possible by debt). The parent is assumed to be mature, and it is indifferent as to whether debt or retained earnings finance its investment in the second period. The foreign subsidiary is financed in the first period by as little equity or as much debt as possible. When the tax is relatively high in the home country, the foreign subsidiary in the second period is financed as much as possible by retained earnings. When the tax is relatively low in the home country, it is financed in the second period as much as possible by debt. All after-foreign-tax earnings are repatriated immediately as dividends. Incentives exist for the multinational to finance the parent by borrowing through the subsidiary, but not the opposite. Withholding taxes are neutral and have no effect on the multinational s investment decisions. Notwithstanding the fact that our simple model does not account for complex interactions among the excess tax credit positions, a U.S. parent having subsidiaries in other countries in addition to Canada, and the possibility of maximizing after-tax income on a global basis, a basic conclusion seems inescapable. Given the general direction of the U.S. and Canadian tax reforms, there is a greater incentive for a USCC to engage in debt shifting between its Canadian subsidiary and the parent corporation. 4. mpirical Analysis In this section, we describe the results of our analysis of a longitudinal sample of Canadian-based firms, which includes FCCs and CCCs, some with foreign affiliates and others without. The main purpose of the analysis is to determine whether this sample confirms the theoretical conjectures and the conclusions from the aggregate data.

18 0 WORKING PAPR Data Source To investigate debt shifting at the individual corporation level, a longitudinal data set was constructed by Revenue Canada at the request of Finance Canada for the purpose of this study. 5 It is composed of 525 Canadian-based large corporations that survived from 984 to 994. These 525 corporations are sampled from a universe that represents about 750,000 corporations. 6 The data set consists of both CCPCs and non-ccpcs. The latter are restricted to those with annual sales greater than $20 million. Because the financial industry is predominantly Canadian-owned and financial corporations financial behavior is more difficult to predict, we further exclude financial corporations from the sample, which results in a sample of 457 Canadian non-financial large corporations, of which 50, on average, are FCCs. For our econometric analysis, we also exclude non-u.s. foreign corporations from the sample and retain those corporations that survived from 986 to 994. Thus, the final sample for econometric analysis is composed of 388 Canadian non-financial large companies, of which 20, on average, are USCCs. Since USCCs account for a large portion of FCCs, most of the conclusions regarding FCCs in the introductory section reflect changes in the USCCs. 4.2 Aggregate vidence As indicated above, multinationals can reduce their tax payments by shifting debt from a low-tax jurisdiction to a high-tax jurisdiction to take advantage of the high-interest deduction in the high-tax jurisdiction. Tables 4a and 4b shows the values of main variables for the two samples: CCCs and FCCs. The variables considered are total assets (net of accounts payable), fixed assets, equity, total debt (long-term and short-term), revenues, operating earnings before interest and taxes, taxable income as reported and taxes paid. 5 The identity of individual corporations was suppressed in the micro-level (company-level), longitudinal data set constructed for this study. 6 We exclude Crown corporations, banks and trust companies, insurance corporations, co-operative corporations, credit unions, investment corporations, mutual-fund corporations, mortgage-investment corporations, exempt corporations and non-resident corporations carrying on entertainment or travel business in Canada.

19 Tax Reforms, Debt Shifting and Corporate Tax Revenues Some interesting observations can be made from these tables. In 984, taxes as a percentage of revenues for CCCs were 2.7 percent, and this percentage was little changed by the end of the 0 years; the corresponding value for 994 is 2.4 percent. During the 0-year period under consideration, the assets, equity and debt levels of CCCs grew by about 75 percent, revenues by 50 percent and, operating earnings before interest and taxes by 79 percent, whereas taxable income rose by 25 percent and total taxes paid by 33 percent. Thus, even for CCCs, taxable income and taxes did not keep pace with the growth in revenues and operating earnings before interest and taxes. The picture for FCCs is, however, more dramatic. In 984, the tax-to-revenue ratio for FCCs was 5.3 percent almost double that of the CCCs. The primary reasons for this difference were their higher operating margins defined as profit before interest and taxes divided by revenues (9 percent versus 5 percent), and a much lower reliance on debt (a 7-percent debt-to-total asset ratio versus 34 percent for CCCs) thereby leading to a lower interest burden. However, the 994 numbers tell a much different story. In 994, the tax-to-revenue ratio is.7 percent for FCCs, even lower than the corresponding number for CCCs. This decline is not because FCCs had lower operating margins, but because they resorted to higher levels of debt financing, with a corresponding increase in their interest burden. The operating margin in 994 is 8.5 percent, almost identical to that of the 984 level. The drop in this tax-to-revenue ratio can be accounted for, in major part, by the much lower taxable income-to-revenue ratio. In turn, the decline in this ratio is at least partially a result of a much higher reliance on debt by FCCs. During the ten years under consideration, the assets of FCCs rose by 72 percent, equity by only 29 percent, debt by 54 percent, operating margins rose by 4 percent, whereas taxable income declined by 36 percent and taxes declined by 52 percent. Figures 3 and 4 show these dramatic changes in the tax-to-asset ratio and in the operating earnings before interest and taxes-to-asset ratio for the CCC and FCC subsamples. In terms of the potential impact on tax revenues, data in Table 4b can be used to make some crude estimates. For example, if the debt-to-asset ratio of the sample FCCs in 994 had remained at the 984 level, their debt would have been higher by $4.5 billion. At a 0-percent interest rate and a 40-percent tax rate, the tax revenues from the sample corporations would have been higher by $80 million. Since the assets represented by sample corporations are approximately

20 2 WORKING PAPR percent of total assets (see Table ) of all FCCs, this would have meant $.3 billion in higher taxes. If, on the other hand, we use the debt-to-fixed asset ratio as the variable for this analysis, the corresponding numbers for the sample would be $320 million (i.e. potentially higher tax revenues) for the sample corporations and $2.4 billion as an estimate for the universe (using the numbers in Table ). Although not shown here, if one considers results for the Canadian-controlled corporation without foreign affiliates (CC-NFA), Canadian-controlled corporation with foreign affiliates (CC-FA), foreign-controlled corporation with foreign affiliates (FC-FA) and FC-NFA sub-samples, except for CC-NFA, all groups show an increase in the debt-to-asset ratios and a corresponding decrease in tax-to-revenue ratios compared to the operating margins. For example, for the 26 or so foreign-controlled corporations with no foreign affiliates other than the parent (FC-NFA), the operating margins increased by 50 percent whereas tax-to-revenue dropped by 40 percent. The unambiguous conclusion is that there has been a consistent increase in debt-to-asset ratio in all categories of corporations except for those controlled by Canadian parents with no foreign affiliates. Correspondingly, the taxes paid by the foreign-controlled corporations, as well as those Canadian-controlled corporations with foreign affiliates have declined significantly relative to their underlying operating earnings. Figures 5 and 6 provide the subsample values of debt-to-asset ratios, which are consistent with the aggregate evidence in Table. 7 Figure 5 shows the debt-to-asset ratio for all CCCs and FCCs, whereas Figure 6 shows the values for CC-FA, CC-NFA, FC-FA and FC-NFA. A subgroup s debt-to-asset ratio is defined as the subgroup aggregate debt to the subgroup aggregate assets. 8 This breakdown of CCCs and FCCs into those with and without foreign affiliates indicates that, except for Canadian-controlled corporations without foreign affiliates, corporations in the three other subgroups increased their debt-to-asset ratios, meaning an increasing reliance on debt. 7 Classification of corporations by whether or not they have foreign affiliates did not start until 986, so Figure 6 starts from Debt equals the sum of both long-term and short-term debt and assets equal total assets minus accounts payable.

21 Tax Reforms, Debt Shifting and Corporate Tax Revenues 3 This evidence is striking, since it suggests that the changes in the relative tax rates resulting from the tax reforms may have had a strong impact on the financial structures of MNCs. Although there was a reduction in the Canadian statutory corporate tax rate as a result of the Canadian tax reform, by itself, this did not seem to have had a significant impact on the debt-to-asset ratio of Canadian-controlled corporations with no foreign affiliates. 9 This suggests that, even though there was a decrease in Canadian tax rates, the decrease was overshadowed by the reversal in relative tax positions between the United States and Canada. Since these results are based on the aggregated data which may mask potential variation across corporations, an econometric analysis, based on micro data, is in order. 4.3 conometric Analysis To estimate the effect of taxes on debt financing by comparing the financial policy of Canadian-controlled corporations without foreign affiliates with the financial policy of all other corporations which consist of Canadian-controlled corporations with foreign affiliates and U.S.-controlled corporations, we specify the following regression models: 20 Debt Asset i t i t = α ( α α α α ) τ D D D 0 CCFA CCFA FCNFA FCNFA FCFA FCFA i t () (2) Debt Asset i t i t β τ τ β P SD γ ε t p i t i t i t = α ( α α D α D α D ) τ 0 CCFA CCFA FCNFA FCNFA FCFA FCFA i t US β US τ β P SD γ ε τ t p i t i t i t 9 A standard extension of capital-structure theory would lead us to expect that a change in the Canadian tax rates would reduce the incentive of CCCs to finance their operations with debt. However, we see no such reduction. 20 Although not reported here, we also attempted to model the impact of the tax loss carry-forward on the debt-to-asset ratio. Theoretically, the availability of tax loss carry-forward will reduce a corporation s effective tax rate. We tried different ways such as adding an extra explanatory variable, D LCF times the tax rate where D LCF is an annual loss carried-forward dummy with a value of one, where tax loss is carried-forward, and zero otherwise, or is a sample-period dummy with a value of one, where there is a tax loss carry-forward in any year of the sample period, and zero otherwise. Neither of these approaches resulted in the corresponding coefficient being significant, and, therefore, the results are not included in the table.

22 4 WORKING PAPR 97-4 where Debt i t = long-term and short-term debt of corporation i in year t; Asset i t = total assets minus accounts payable of corporation i in year t; D j, j = CC-FA, FC-NFA and FC-FA, is a control dummy; τ i t = Canadian corporate tax rate (which combines provincial and federal rates) of corporation i in year t; τ t = the difference of the average industry corporate tax rates between Canada and the United States for groups CC-FA, FC-NFA, and FC-FA. For CC-NFA, τ t is zero; τ t US = the average tax rate in the United States, in each year, t; P i t = risk - adjusted operating performance of corporation i in year t; SD i t = a vector of size dummies; ε i t denotes the error term, assuming heteroskedasticity across corporations; α, β, and γ are the parameters to be estimated. Note that γ is a vector. The control dummies divide corporations into four groups according to ownership (Canadian-controlled or U.S.-controlled) and the presence or absence of foreign affiliates. Unlike Canadian-controlled corporations without foreign affiliates whose debt-to-asset ratios are affected only by Canadian taxes, other corporations have incentives to change their debt-to-asset ratios whenever there is a change in Canadian taxes or in relative tax rates between Canada and the United States. The risk-adjusted operating performance variable P it is defined as a corporation s operating earnings (before interest and tax), scaled by the corporation s assets in year t and divided by the standard deviation of the corporation s earnings across sample years. The standard deviation is used to measure the risk of operating income of a corporation, and is assumed to be constant for each year t for each corporation. There is a general consensus that the ability (cost) to borrow is negatively related to the riskiness of the corporation s operating performance, since more profitable and less risky corporations can sustain higher debt levels. The size dummies are introduced to capture the size effect on the debt-to-asset ratio, which, to some extent, reflects a corporation s vintage. We divide corporations based on assets into

23 Tax Reforms, Debt Shifting and Corporate Tax Revenues 5 four groups. The first group consists of corporations with assets not greater than $00 million; the second group with assets larger than $00 million but not greater than $500 million; the third group with assets larger than $500 million but not greater than $ billion; the last group with assets larger than $ billion. The first group will be used as the base for regression. Some comments on the variable τ in equation () and τ US t t in equation (2) are in order. Ideally, we would have liked to model this variable at the individual firm level and for each year by calculating (or estimating) the actual rates faced by the parent and its subsidiary. In addition, we would have liked to model the excess foreign tax positions of the parent. Unfortunately, we do not have access to the tax data of the U.S. parent or the foreign affiliate(s) of the Canadian parent, and there is a multicollinearity problem (between τ US t and τ ). Therefore, in t equation (), we have used the Canadian industry average corporate tax rates (based on our sample) minus the U.S. national average corporate tax rates to derive tax differentials that depend on the type of the firm CC-FA, CC-NFA, FC-NFA and FC-FA. Thus, the tax differential estimate is the same for each firm in a group in a given year. The tax differential for group CC-NFA is zero, since its financial policy should not be influenced by U.S. tax rates. To test for the robustness and the impact of the tax-rate changes, we also use in equation (2) just the U.S. tax rate instead of the τ to see whether the debt-to-asset ratio is sensitive to the U.S. tax t rate. Note that we cannot use both the variables in the same regressions due to the multicollinearity problem. Table 5 reports the results of estimating equations () and (2) using the longitudinal data set consisting of 388 large Canadian-based non-financial corporations. Column (i) shows the results for equation () but without the size dummies, column (ii) corresponds to the full specification, column (iii) corresponds to equation specification 2. The regressions reported in column (i) through (iii) suggest the following: a) Canadian corporate tax rates have a significant impact on debt-to-asset ratios of Canadian-based corporations. Canadian-controlled corporations without foreign affiliates tend to be more sensitive to Canadian taxes than Canadian-controlled corporations with foreign affiliates and U.S.-controlled corporations. This is because the financial policy of

24 6 WORKING PAPR 97-4 corporations with foreign affiliates or corporations controlled by U.S. parents is also influenced by foreign tax systems. b) The other key result is that the difference between Canadian and U.S. taxes correlates with an increase in the debt-to-asset ratios of corporations other than those that are Canadian-controlled without foreign affiliates. Under the column (ii) specification, the coefficient for the differential tax rate is statistically significant and positive. That is, on average, the debt-to-asset ratios of firms in groups other than CC-NFA increase as the Canadian tax rates relative to U.S. tax-rates increase. Similarly, the results in column (iii) indicate that even the simple U.S. tax-rate variable has the right sign and is also statistically significant. What is remarkable about these statistically significant results is that neither of these two specifications is exact. Ideally and as mentioned earlier, one would like to have firm-level data for the U.S. parent and then use the matched differential tax rate in the regression specification. Unfortunately, such data are unavailable. c) Corporations with good operating performance have significantly higher debt-to-asset ratios. This simply confirms the link between business risk and financial risk. d) The size of corporations, which is measured by assets, also significantly influences their debt-to-asset ratio. Large-size corporations have higher debt-to-asset ratios. Overall, these results indicate that there has been a change in the debt-to-asset ratios of the multinationals operating in Canada along with the relative shifts in the statutory tax rates between the two countries. Although the Canadian tax reform reduced the statutory tax rates in Canada, it was overshadowed by the declines in the U.S. tax rates. Note that these results do not necessarily indicate that, by lowering its tax rate so as to (potentially) lower the debt-to-asset ratio, Canada would achieve a net gain in its tax revenues. This can be explored with illustrative calculations based on column (iii) of Table 5. Based on the coefficients of FC-FAs and FC-NFAs, a one percentage point reduction in the Canadian corporate tax rate would, on the margin, reduce the debt-to-asset ratio by percentage points on average (weighted by the assets of FC-FAs and FC-NFAs in Table ) for foreign-controlled corporations. Given the 994 data from Table for the total assets of these

25 Tax Reforms, Debt Shifting and Corporate Tax Revenues 7 corporations, this implies a reduction in debt financing of $.05 billion. Assuming an interest rate of 0 percent for simplicity, this reduction in debt would imply a reduction of $0.05 billion in interest payments and a corresponding increase in total taxable income. Given that the taxable income of these corporations was $2.355 billion in 994 (Table ), the total change in tax revenue would be: R = ( ) (τ 0.0) τ = 0.05 τ < 0 Thus, although the interest deductions are potentially reduced, the reduction is not sufficient to fully offset the overall one percentage point decline in the tax rate. In other words, no matter how much the tax rate declines, the overall tax revenues will decline, although debt levels may increase. In a similar vein, one may argue that an increase of one percent in the tax rate may increase the overall tax revenues, all else being held constant. However, relatively higher tax rates may encourage corporations to engage in other instruments (as in footnote 5) to minimize their tax payments. Moreover, relatively high tax rates may induce Canadian multinationals to invest abroad and may deter other FCCs from investing in Canada. 5. Conclusions In this paper, we concentrate on analysing debt-shifting behaviour of Canadian corporations with a special emphasis on Canadian subsidiaries of foreign corporations and Canadian-controlled corporations with foreign affiliates. We hypothesize that shifts in the relative tax rate faced by corporations in the United States and Canada following the tax reforms undertaken by these two countries have led these categories of corporations to increase their reliance on debt in Canada. This debt shifting may be one explanation for the reduced Canadian corporate tax in the early 990s. Our two-country integrated theoretical model also shows that, in a world where host-country tax rates substantially exceed home-country tax rates, the home-country multinational will choose to finance its subsidiary with a much higher portion of debt. This would allow the multinational to maximize its overall after-tax income.

26 8 WORKING PAPR 97-4 Although many potential avenues for tax shifting across national borders are open to a typical multinational, we hypothesize that the multinational will prefer debt shifting as a mechanism, as it is potentially the least costly mechanism. More specifically, debt shifting is free from: a) any audit consequences resulting from aggressive transfer pricing; and b) potential difficulty in evaluating trade-offs between tax differentials and resource rents, or high cost due to the irreversibility of investments. Our empirical and econometric results confirm the significant changes that have occurred in debt levels of multinationals in Canada. Our aggregate results clearly show a reduction in taxes from FCCs in Canada as a proportion of total corporate revenues. Our sample of 457 large corporations indicates that this reduction is consistent with the increase in debt-to-asset ratios of FCCs in Canada following the U.S. and Canadian tax reforms, which significantly changed the relative tax differential and made Canada a country with relatively high tax rates. Similar increases in debt-to-asset ratios are also evident for Canadian-controlled multinationals. Our econometric analysis also confirms these results, even though we do not have a direct measure of firm-specific tax rates for the U.S. parent. In all cases, the debt-to-asset ratios at the firm level are statistically significantly related to the Canadian tax rates as well as the tax differential. Since USCCs control a significant fraction of Canadian corporate assets and operating income, the impact on Canadian corporate tax revenues has been significant. In essence, the evident debt-shifting policies adopted by USCCs have benefited the United States at the expense of Canadian tax revenues. In aggregate, if FCCs had maintained, in 994, the same corporate tax-to-asset ratio as in 984, Canadian tax revenues from FCCs would have been $6.4 billion instead of the actual amount of $2.9 billion, a difference of $3.5 billion. Using the crude estimates based on the sample corporations, this difference could be in the range of $.3 to $2.4 billion still a significant number. These estimates exclude the potential loss of tax revenues due to an increase in debt by Canadian-controlled corporations who now have an incentive to finance their foreign (U.S.) affiliates by raising debt in Canada. Clearly, the impact on Canadian tax revenues of debt shifting by both the FCCs and the CCCs with foreign affiliates is significant. However, it should also be noted that our regression results suggest that tax revenue losses as a result of debt shifting in response to changes in relative tax rates are much lower than the impacts of changes in total corporate tax-to-asset ratios noted above.