The Calculation of Marginal Effective Tax Rates

Transcription

1 The Calculation o Marginal Eective Tax Rates by Kenneth J. McKenzie University o Calgary Mario Mansour Department o Finance Ariane Brûlé Secretariat o the Technical Committee on Business Taxation May 1998 WORKING PAPER Prepared or the Technical Committee on Business Taxation Working papers are circulated to make analytic work prepared or the Technical Committee on Business Taxation available. They have received only limited evaluation; views expressed are those o the authors and do not necessarily relect the views o the Technical Committee, or the Department o Finance.

2

3 The Calculation o Marginal Eective Tax Rates by Kenneth J. McKenzie University o Calgary Mario Mansour Department o Finance Ariane Brûlé Secretariat o the Technical Committee on Business Taxation May 1998 WORKING PAPER Prepared or 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 o Finance Ottawa, Ont. K1A 0G5 Fax: (613) Berg-Dick.Paul@in.gc.ca

4 Kenneth J. McKenzie Department o Economics University o Calgary 2500 University Drive N.W. Calgary, Alberta N6A 3K7 Fax: (403) kjmckenz@acs.ucalgary.ca Mario Mansour Business Income Tax Division Department o Finance 140 O Connor Street Ottawa, Ontatio K1A 0G5 Fax: (613) mansour.mario@in.gc.ca Ariane Brûlé Personal Income Tax Division Department o Finance 140 O Connor Street Ottawa, Ontatio K1A 0G5 Fax: (613) brule.ariane@in.gc.ca For additional copies o this document please contact: Distribution Centre Department o Finance 300 Laurier Avenue West Ottawa K1A 0G5 Telephone: (613) Facsimile: (613) Also available through the Internet at Cette publication est également disponible en rançais.

5 Abstract The objective o this working paper is to describe the methodology used to undertake the marginal eective tax rate (METR) calculations contained in the inal report issued by the Technical Committee on Business Taxation. The data underlying the computations are also described and presented, as are various illustrative calculations that supplement the cases covered by the eective tax rates presented in the Technical Committee s report. The model used to calculate the METRs updates, and expands on, earlier Canadian work. The main extensions to the model are: or METRs on tangible capital, incorporation o ederal and provincial capital taxes, and provincial sales taxes on capital inputs, in addition to ederal and provincial corporate income taxes; METRs on research and development (R&D), labour, and total production cost; METRs or non-tax-paying irms; and inally, METRs on tangible and intangible capital using industry-speciic debt-asset ratios. These additions allow or an analysis o METRs on a broader range o production inputs and accounting or a larger variety o ederal and provincial taxes. This has become more relevant or the study o the impact o taxation, given the increase in the relative importance o taxes that are not sensitive to proits (i.e. payroll and capital taxes). The METRs or the current Canadian corporate tax system vary across irm size, asset type and industries. Generally, small irms ace lower METRs than large irms, regardless o the type o investment they make. Investments in machinery, R&D, and exploration and development (E&D) are subject to lower METRs than investment in structures, land and inventories, and METRs on labour are appreciably lower than METRs on capital (except R&D). Finally, a comparison o METRs across industries shows that, in general, service industries are treated less avourably than manuacturing and resource industries. Also provided, to allow international comparisons, is a calculation o METRs on tangible capital in the G-7 countries and Mexico. Canadian METRs on tangible capital are generally lower than those in Germany, Italy and Japan, but higher than those in the rest o the G-7 countries and Mexico. However, investment in manuacturing equipment is treated more avourably in Canada than in most o the other G-7 countries, including the United States, and investment in R&D is treated more avourably in Canada than in the United States. Finally, a simulation o the Technical Committee s policy package shows that the Committee s recommendations would result in little change in the overall level o METRs, but that the variation in METRs would be reduced substantially, especially those across industries.

6 Table o Contents 1. Introduction Methodology Motivation Methodology Results The Base Case Alternative Scenarios International Comparisons Simulating the Technical Committee s Policy Package Concluding Thoughts...50 Appendix A Structure o the METR Model Appendix B Estimated Eective Payroll Tax Rates or Canada Appendix C Estimated Eective Payroll Tax Rates or the United States Reerences...108

7

8 Acknowledgments We would like to thank Duanjie Chen, who provided invaluable assistance in the early stages o development o the METR model, particularly on the international and non-renewable resource aspects, and Jack Jung, who provided assistance and advice relating to the calculation o the capital stock weights. Special thanks to John Sargent, Jack Mintz and Paul Berg-Dick or valuable eedback throughout the project.

9 The Calculation o Marginal Eective Tax Rates 1 1. Introduction The purpose o this working paper is to describe the methodology used to undertake the marginal eective tax rate (METR) calculations contained in the report issued by the Technical Committee on Business Taxation. The data underlying the computations will also be described and presented, as will various illustrative calculations, extending somewhat the eective tax rates presented in the Technical Committee s main report. As a part o the documents laying the groundwork or the 1987 tax reorm, various calculations o the marginal eective tax rate on capital or dierent types o assets employed in various sectors were presented. 1 A technical paper explaining the methodology used to compute those eective tax rates, as well as an extensive description o the underlying data, was released by the Department o Finance shortly thereater. 2 The current exercise represents a substantial update and revision o the Department s eective tax rate model. The model has been revised along several dimensions. First, various tax changes that have occurred over the past 10 years have been incorporated into the model. This has involved not only updating tax parameters such as statutory corporate tax rates, capital cost allowance (CCA) rates, investment tax credits rates (ITCs), etc., but also incorporating new taxes into the model, such as the large corporations tax (LCT) at the ederal level and various capital taxes at the provincial level. Second, the 1987 tax reorm documents reported METRs on capital or ully tax-paying corporations only; the current version allows or the incorporation o non-tax-paying irms. Third, much o the underlying data have been revised and updated. Economic parameters such as interest and inlation rates have changed, but so too has the structure o the Canadian economy. For example, the service sector has grown, and small (or tax purposes) corporations have increased in prominence. As such, the capital and industry weights used to aggregate the METRs on capital have been substantially updated. Fourth, several methodological innovations have been incorporated into the model. In 1987 the analysis was devoted exclusively to METRs on various types o tangible capital structures, machinery, land and inventories as well as exploration and development (E&D) in the resource sector. The current version also calculates METRs associated with other inputs in the production process, such as research and development (R&D) and labour. The eective tax rates on the dierent inputs used in production are then aggregated together into a METR on production costs, ollowing a methodology developed by McKenzie, Mintz and Schar (1997). As described in more detail in Section 2, the METR on costs measures the contribution o various taxes levied on inputs to the marginal cost o production. As such, it provides a convenient measure o the extent to which the tax system impinges upon the cost o doing business. Finally, the model has been expanded to allow or the calculation o METRs or purposes o international comparisons. Marginal eective tax rates on tangible capital or the manuacturing and services sector in the G-7 countries, as well as Mexico, are computed. For the United States, METRs on the other inputs as well as production costs are calculated or numerous industrial sectors. 1 Canada, Department o Finance (1987). 2 Jung (1989).

10 2 WORKING PAPER Finally, while the METR calculations contained in this document are state o the art, the methodology does have its limitations. Many aspects o the tax system do not lend themselves to the approach, as they cannot be modelled analytically. Moreover, the estimates can be sensitive to the data and assumptions regarding key parameters. While we do perorm some sensitivity analysis, there is clearly scope or more. Thus, one additional purpose o this document is to give the interested technical reader some eel or the limitations o the estimates. The remainder o the document is organized as ollows. The ollowing section provides an overview o the methodology. Section 3 presents and discusses various METR calculations, under dierent assumptions regarding the tax system and the economic environment. Section 4 presents METRs or the G-7 countries plus Mexico. Section 5 presents the METRs or the Technical Committee s policy package. Finally, section 6 concludes. Three data and technical appendixes are also included. Appendix A outlines the data and structure o the METR model. Appendix B provides the data and methodology used to calculate marginal eective payroll tax rates in Canada, while Appendix C does the same thing or the United States. 2. Methodology In this section, an overview o the basic methodology underlying the calculation o METRs is presented. The section begins with a brie discussion o the motivation behind the approach to the calculation o METRs taken in this study. We then move on to the methodological overview. For the most part, the basic concepts underlying the methodology are quite simple and have been documented elsewhere 3 ; the discussion is thereore largely intuitive. 2.1 Motivation While businesses are subject to various types o taxes, analysts have traditionally ocussed on the corporate income tax (CIT). Since the CIT can be thought o as a tax on the return to capital, the tendency to ocus on it is perhaps understandable in light o prevailing concerns regarding the impact o taxes on capital accumulation and investment. 4 However, other business inputs, most notably labour, are subject to various taxes as well, such as ederal payroll taxes to und Employment Insurance (EI) and the Canada Pension Plan and Quebec Pension Plan (CPP/QPP), and provincial payroll taxes to inance health and education. Moreover, while the implementation o the Goods and Services Tax (GST) largely eliminated the taxation o business inputs that occurred under the ederal manuacturing sales tax, provincial retail sales taxes still result in the taxation o some business inputs. 3 See McKenzie, Mintz and Schar (1997) and reerences therein. 4 Bird (1996) discusses various ways o viewing the CIT, and provides some justiication or its existence. See also Mintz (1995). It is also important to note that while the CIT can be thought o as a tax on the return to capital, the ultimate burden, or economic incidence, o the tax need not all on the owners o capital (shareholders). Depending upon the economic environment, and on the technological characteristics o the corporate sector, the burden o the CIT may be borne by (some combination o) the owners o capital, consumers o goods produced by corporations, other actors o production (e.g. labour, land), or even oreign treasuries in the case o non-resident multinational irms subject to the CIT. For a discussion o some o the issues in this regard, see Whalley (1997).

11 The Calculation o Marginal Eective Tax Rates 3 When analysing the incentive eects o taxation, it is important to consider the impact o taxes on marginal, or incremental, economic decisions (i.e. the decision to employ one more unit o capital, hire one more worker, or produce one more unit o output). In this spirit, we thus ocus on the calculation o marginal eective tax rates. The METR is deined as the amount o tax arising rom the decision o a irm to undertake one more unit o an economic activity. In the case o capital, the activity is the employment o an incremental unit o capital, in the case o labour, it is an incremental worker; and in the case o production, it is an incremental unit o output. Because we measure the taxes arising rom an incremental unit o economic activity, a positive METR associated with that activity indicates that it is discouraged by the tax system, a negative METR indicates that the activity is encouraged by the tax system, while a zero METR indicates that the tax system is neutral with respect to the activity. In the case o a neutral tax system, the activity generates no tax revenue at the margin, but inramarginal units o the activity may generate tax revenue. Taxes imposed on capital (like the CIT) can aect both the level and composition o investment in an economy by distorting the return to an incremental unit o capital. The METR on capital is a summary measure o the size o this distortion. There is a growing body o empirical research that suggests that corporate taxes levied on capital can dampen investment, although there is some disagreement over the magnitude o the impact. 5 Moreover, taxes on capital may aect not only the level o investment, but its composition as well: dierences in METRs across types o capital can give rise to inter-asset distortions; dierences in METRs on capital employed in dierent sectors can give rise to inter-industry distortions; and dierences in METRs on capital employed in dierent jurisdictions can give rise to inter-jurisdictional distortions. Thus, taxes can lead to distortions in the allocation o capital over time, across assets, among sectors, and among jurisdictions. The distortions in the allocation o capital caused by the tax system are important or two reasons. First, the distortions can lead to a reduction in the level o goods and services produced in the economy. This gives rise to what are reerred to as the eiciency costs o capital taxation. Second, although both the theoretical and empirical literature is somewhat less developed on this issue, there is some evidence that there exists a positive correlation between investment and economic growth in cross-country comparisons. 6 Thus, taxes on capital may also aect the rate o growth in the economy through their impact on investment. 7 These linkages between taxes on 5 Auerbach and Hassett (1992), Cummins and Hasset (1992, 1994), Cummins, Hasset and Hubbard (1996), and Chirinko, Fazzari and Meyer (1996) ind signiicant CIT eects on investment, with the elasticity o investment with respect to the tax adjusted cost o capital (discussed below) between.25 and 1. McKenzie and Thompson (1997) also ind that dierences in the tax-adjusted cost o capital between Canada and the United States help to explain dierences in investment rates between the two countries in a statistically signiicant way. 6 See Levine and Renelt (1992). 7 There is some question in the literature as to whether or not changes in investment caused by the tax system are permanent or transitory. In either event, a temporary change in growth may still last or a substantial period o time. As discussed above, even i changes in growth are transitory, they still lead to a permanent changes in the level o economic output (GDP). Also, De Long and Summers (1991) argue that it is not so much the level o savings and investment that matters or economic growth, but rather whether that investment is allocated "appropriately." Thus, inter-asset, inter-industry and inter-jurisdictional distortions may also be important or economic growth.

12 4 WORKING PAPER capital and both the level and growth o economic activity provide much o the motivation or the traditional ocus on measuring METRs on capital. Yet taxes on other business inputs, such as payroll taxes levied on labour, can also have important economic eects. The eect o such taxes on wages and employment are theoretically ambiguous, and the empirical evidence is somewhat mixed. 8 Nonetheless, there is some evidence that taxes imposed on labour, such as payroll taxes, dampen employment, at least in the short run. Moreover, some representations o endogenous growth theory, which ocus on the role that technology and human capital play in contributing to economic growth, suggest that taxes on labour can stile growth by lowering the incentives o individuals to invest in human capital. Aside rom the traditional supply and demand responses to taxes levied on actor inputs such as capital and labour, i input METRs dier across actors, irms will, to the extent that it is technologically possible, alter the mix o inputs used in production. Moreover, taxes levied on business inputs, such as labour and capital, can aect the cost o providing the goods and services that help to produce. Policy analysts oten voice concern with the implications o taxation or competitiveness; 9 yet the meaning o the term is not always well deined. A natural way to think about competitiveness is as the cost o doing business. Taking this view, the extent to which the tax system aects a irm s, or an industry s, or a country s cost o doing business, is a key determinant o competitiveness. More precisely, we are interested in measuring the extent to which the business tax regime might impinge upon the marginal cost o producing an incremental unit o output. The METR on costs does just this it measures the extent to which taxes levied on irm inputs, such as capital and labour, contribute to the cost o producing one more unit o output. It thus aggregates the various METRs on irm inputs into a summary measure o the extent to which the business tax regime adds to the cost o doing business. Measuring the extent to which the business tax regime impinges upon marginal costs is a natural way to think about the implications o taxation or competitiveness. For example, in their examination o border-tax adjustments on imported goods, Poterba and Rotemberg (1995) speciically identiy the preservation o the competitive position o domestic producers vis-à-vis oreign producers, which they deine as the relative marginal costs o domestic and oreign producers, as the objective pursued by governments. The marginal cost o production is a key determinant o the level and composition o goods and services produced in an economy. Businesses maximize proits by producing output up to the point that marginal revenues (the revenue accruing to the irm due to the production o one more unit o output) are just equal to marginal costs. An increase in marginal costs will cause production to decrease, a decrease will cause output to expand. To the extent that government policies aect the marginal cost o production, through the tax system or otherwise, they can thereore aect the level o goods and services produced in an economy. O particular concern is 8 See Dahlby (1992), Di Matteo and Shannon (1995), and Wilton and Prescott (1993). 9 See, or example, Porter (1993).

13 The Calculation o Marginal Eective Tax Rates 5 the extent to which the business tax regime may result in an uneven playing ield by generating tax-induced dierences in marginal costs across producers within a market, across producers rom dierent jurisdictions, and across industries. To close this subsection, it is important to reiterate the distinction between the two basic types o METRs that will be employed in the subsequent analysis. The irst is the METR on inputs, which includes tangible capital such as buildings, equipment, land and inventories, intangible capital such as E&D and R&D, and labour. These METRs measure the taxes collected on an incremental unit o the input employed by a irm. The second is the METR on costs, which measures the contribution o the taxes levied on the various inputs to the cost o producing an incremental unit o output. As discussed above, both sets o eective tax rate measures are motivated by dierent, though not necessarily mutually exclusive, considerations: the input METRs provide an indication o the extent to which the tax system impinges upon the incentive to employ these inputs, while the cost METRs, which are a unction o the input METRs, measure the impact o the tax system on competitiveness, or the cost o doing business. 2.2 Methodology With the above motivation in hand, we now proceed by presenting an explanation o the basic idea behind the methodology. We begin with a discussion o the METR on costs, which is grounded in the undamentals o elementary price theory. We then move on to a discussion o the METRs on the individual inputs labour, tangible capital and then intangible capital METR on Costs To begin, presume that both input and output markets are perectly competitive. The assumption o perect competition is or expositional purposes only, as the analysis also applies to non-competitive markets (with minor modiications). Consider the output market or some good or service. Figure 1 illustrates the equilibrium in the absence o any taxes. The equilibrium occurs at the intersection o the market supply and demand curves, denoted S(p;W 0 ) and D(p) respectively, where p is the output price, and W 0 is a vector o input prices, or user costs. The equilibrium price in this case is p 0 and the quantity is q 0. Ignore, or the moment, the rest o the diagram. The approach incorporates the vertical linkages between input and output markets. The output market is connected to the input markets by the act that the aggregate supply curve is the (horizontal) sum o the marginal cost curves or the individual suppliers. The marginal cost o providing an additional unit o output thus relects the user cost o the various inputs, which in turn relects the supply and demand conditions in the input markets. This is emphasized by writing aggregate supply as a unction o the vector o input prices W 0. The connection between the input and output markets provides the key to the measure o the METR on production costs. 10 We do not provide a ormal derivation here, as it has been done elsewhere. In particular, the discussion related to the METR on costs closely ollows McKenzie, Mintz and Schar (1997), while the discussion related to the METR on capital ollows Chen and McKenzie (1997) and McKenzie and Mintz (1992).

14 6 WORKING PAPER The various taxes applied to irm inputs aect the marginal cost o providing the product by changing the user costs. For example, i the tax system causes the user cost o, say, labour to rise, the marginal cost o providing an additional unit o output will rise as well, and the industry supply curve will shit up. Reerring again to Figure 1, the ater-tax aggregate supply curve is designated S(p;W ), where W denotes the vector o gross-o-tax user costs. The ater-tax equilibrium price and quantity are p and q respectively. Figure 1 S ( p;w ) p = MC ( q ;W ) S( p;w 0 ) 0 p MC ( q ;W 0 ) D( p ) 0 q q The gross-o-tax marginal cost o production at the ater-tax equilibrium is MC(q ;W ), which is expressed as a unction o the gross-o-tax user costs, W. Associated with this gross-o-tax 0 marginal cost is a net-o-tax marginal cost, deined as MC( q ;W ). The METR on costs is then the tax rate T, which, i hypothetically applied to production costs directly, would yield the same gross-o-tax marginal cost that results under the existing tax regime. Thus, T solves the equation 0 (1 + T )MC( q ;W ) = MC( q ;W ), which gives: T = MC (q ;W ) MC (q ;W 0 ) - 1 (1) As deined in equation (1), the METR on the cost o production gives the rate o tax on marginal costs implied by the various taxes levied on business inputs. It can be viewed as aggregating these various taxes together into a simple eective excise tax rate. In terms o Figure 1, the tax 0 wedge, MC( q ;W ) MC( q ;W ), is expressed as a percentage o the net-o-tax marginal cost, 0 MC( q ;W ).

15 The Calculation o Marginal Eective Tax Rates 7 An important issue concerns the incidence o the taxes levied on irm inputs, or the extent to which those taxes are relected in higher user costs, and thereore in marginal production costs. This is important because, under some conditions, taxes levied on inputs may not aect user costs at all, and will thereore not eed through to marginal costs. In general, the extent to which taxes will be relected in the user cost o the inputs depends upon the supply and demand conditions in the input markets. Consider the introduction o a tax at the METR o t i on input i, where w i = w i(1 + ti ) is the user cost o input i (the ith element o the vector W ), and w i is the equilibrium supply price. Assuming that the input market is competitive, equilibrium is determined by: Di ( w i (1+ ti )) = S i ( wi ) (2) where D i (.) is the demand unction or input i and S i (.) is the supply unction. The demand unction or input i is expressed with respect to the gross-o-tax price o the input, and the supply unction is expressed with respect to the net-o-tax price. For simplicity, prices o other inputs are suppressed. Dierentiating both sides o (2) with respect to the tax rate t i, and evaluating the derivative at zero, gives, ater some algebraic manipulation: w i t i S ηi = wi S D ηi + ηi (3) D S where ηi is the elasticity o demand or input i and ηi is the elasticity o supply. Equation (2) implicitly determines the equilibrium user cost o input i as a unction o the tax rate t, w w i( ti ). A irst-order Taylor series approximation o this implicit unction yields: i i = Using equations (3) and (4) gives: o w i w i = wi + ti (4) ti 0 w = w i (1 + ti β ), i i where S η i β i = S D ηi + ηi (5) and 0 wi is the ith element in the vector W 0. The parameter 0 < β i < 1 is a tax-shiting actor. When β i = 1 the tax is ully shited orward to the demander o the input, and the user cost changes by the ull amount o the tax, as would be the case when the input supply unction is perectly elastic, or demand is perectly inelastic.

16 8 WORKING PAPER When β i = 0, none o the tax is shited orward to the user, and the user cost is unaected by the tax or subsidy, as would be the case when supply is perectly inelastic, or demand is perectly elastic. In the intermediate case, the user cost increases by some raction o the tax Functional Form or Cost Function To calculate the eective tax rate on marginal cost, as in equation (1) above, the marginal cost unction must be parameterized. In general, marginal costs will depend upon the level o output, productivity parameters, input shares, actor prices, and the degree o substitutability between actors. In this section, we illustrate the methodology by employing the commonly used linearly homogeneous Constant Elasticity o Substitution (CES) production unction, which has the orm: q = H i 0 x, ai = i i a i 1 ρ i 1 (6) where q is output, x i is the quantity o input i employed, H, a i, i and ρ are production parameters, and the elasticity o substitution is σ=ρ/(ρ-1). The gross-o-tax marginal cost unction that arises rom the CES production unction is: 1 MC( q ; W ) = H a i b ρ i ( iw i ) 1 b (7) where b=ρ/(ρ-1). 0 MC( q ;W ) is determined by evaluating (7) at W 0 rather than W. Recalling rom our earlier 0 discussion that w i = wi ( 1 + ti β i ), equations (7) and (1) give an eective tax rate on marginal cost or a CES production unction o: T = 1 b 0 b b ρ a ( w ) i 1 b i i i A (1 + ti β i ) 1, where Ai =, A = b i i ρ 0 b i ai ( wi i ) i (8) and A i is the actor share or input i. It is possible to show rom equation (8) that as the elasticity o substitution increases, the eective tax rate on marginal costs decreases. This is because as the degree o substitutability between inputs rises, the irm is better able to respond to changes in relative actor prices by changing the input mix. As such, a tax-induced increase in the relative price o an input has a lower impact on marginal costs, the higher the elasticity o substitution.

17 The Calculation o Marginal Eective Tax Rates 9 Two commonly used special cases o the CES production unction are the Cobb-Douglas (CD) and the Leontie, or ixed proportions (FP), production unctions. The elasticity o substitution or the CD case is unity, in which case equation (8) reduces (in the limit) to: T = i Ai (1 + ti β i ) 1 (9) For the FP case, the elasticity o scale is zero, and the eective tax rate on marginal costs becomes: T = n A i( 1 + ti β i ) 1 (10) i=1 It is interesting to note that the eective tax rate on marginal costs or the FP case relects a simple arithmetic-weighted average o the user costs o the inputs (with net o tax prices normalized to unity), while the rate or the CD case relects the geometric-weighted average. The arithmetic average or the FP production unction relects the act that under this technology, irms are not able to respond to tax-induced changes in user costs by substituting away rom (relatively) highly taxed actors. Factors are employed in ixed proportions, and the eective excise tax rate on marginal costs is simply the arithmetic-weighted average o the METRs on the inputs. In the case o a CD production unction, with the elasticity o substitution equal to unity, there is some scope or substituting between actors, and the eective tax rate on marginal costs relects the geometric-weighted average o the tax rates on the inputs, which is lower than the arithmetic average. The empirical calculations o the METR on cost reported later in the paper are based on the CD parameterization o the cost unction. Although they are not reported here, calculations based on the FP parameterization yield similar results; the eective tax rates are slightly higher because no input substitution is possible METRs on Inputs In order to calculate the METR on costs, we must determine the METRs on the various inputs (the t i s). Moreover, as discussed in Section 2.1, the marginal eective tax rates on irm inputs are o independent interest. To make the measurement task manageable, a certain amount o aggregation is required. In the empirical analysis which ollows in Section 3, we include our types o tangible capital (structures, machinery, land and inventories), two types o intangible capital (E&D and R&D), and labour Labour To calculate the METR on labour, we incorporate the various payroll taxes or other levies or social security, employment insurance, etc. Since taxes on labour are oten applied at variable rates, depending upon income, and may also vary by individual characteristics, the dierence between average and marginal rates can be quite important. A key consideration is the meaning o the marginal unit o labour. The approach we use here is to presume that the employment o a

18 10 WORKING PAPER marginal unit o labour involves hiring an additional worker with typical characteristics. We use employment data to construct a proile o a hybrid marginal worker in each sector being studied, and then calculate the eective tax rate on labour or this hybrid worker. The data used to do this, and the computations themselves, are discussed in detail in Appendix B. The incidence o payroll taxes levied on labour is an important consideration. As discussed above, this depends upon the demand and supply elasticities or labour. There is not a broad consensus among economists regarding the incidence o payroll taxes. The literature does seem to suggest that in the long run, the supply o labour is relatively inelastic, which suggests that in the long term, labour bears the bulk (though not all) o the burden o payroll taxes. 11 However, labour demand is relatively elastic in the short run, which suggests that in the short term, the burden o payroll taxes may well all on producers, at least to some extent. Moreover, payroll taxes may not be shited at all onto workers at the minimum wage. The approach that we take here is to assume or the base case calculations that the economic incidence o payroll taxes coincides with the statutory incidence. Thus, both employers and employees are assumed to bear the share o payroll taxes that they are legally required to pay. This coincides roughly to an assumption that the tax incidence o payroll taxes is split equally between employers and employees; we also consider the implications o alternative shiting assumptions. Another issue that arises in the case o payroll taxes is the question o the extent to which they should be viewed as taxes at all. In many cases, payroll taxes inance certain programs rom which employers and employees directly beneit. The most obvious example is workers compensation, which can be thought o as insurance or on-the-job accidents paid or workers by their employers. Although in some provinces workers compensation payments are not actuarially air, or the most part they are a reasonably good approximation. Thus, in our base case we do not include workers compensation payments made by employers with other payroll taxes. At the other end o the spectrum, since provincial payroll taxes go into general revenues, we treat them as taxes. CPP/QPP taxes and EI premiums are more problematic. This is particularly true or EI premiums, as there are substantial variations in the net EI contributions (contributions less beneits) arising rom EI across sectors. Sectors that are systematically net beneiciaries o EI (with beneits exceeding contributions) can be thought o as receiving a labour subsidy, as they are able to pay lower wages because o the income support oered to their employees by the EI system. Our approach is to consider various scenarios regarding the eective CPP/QPP and EI rates. In the base case, payroll taxes are expressed on a net-o-beneits basis, as the rates are adjusted to relect their direct beneit component. In the case o EI, the adjustments are based upon the net inlows or outlows o EI contributions and beneits. We also make adjustments to relect the recent changes to the CPP/QPP system, which will increase payroll tax rates over the next ive years in order to achieve a sustainable system by These adjustments are described in detail in Appendix B. 11 See Dahlby (1992).

19 The Calculation o Marginal Eective Tax Rates Tangible Capital Tangible capital inputs such as structures, machinery, land and inventories present other problems. The diiculty here is that these inputs give rise to a low o output over time, which requires that we impute a per-period cost o holding capital, and calculate the METR applied to this imputed cost. As is the case with the other inputs, some aggregation is required. For physical assets, the approach adopted here is similar to that taken in the user cost o capital literature, which examines the impact o corporate income taxes on physical investment. 12 To begin, it is again useul to consider an economy without any taxes at all. Firms invest in projects that generate a rate o return in excess o a hurdle rate required by inancial markets. This hurdle rate relects the real (inlation adjusted), net-o-depreciation rate o return that investors (debt and equity holders) could earn in the next-best alternative investment opportunity with similar characteristics. I we presume that the proportion o investment inanced by debt is β, the expected rate o inlation is π, the nominal interest rate on debt is i, and the nominal required rate o return on equity is ρ, then the real hurdle rate o return is a weighted average o the required rate o return on debt and equity, or R = βi + (1 β ) ρ π. 13 I we assume that investment is continuously divisible, and that the marginal revenue product o capital (the increment to revenue arising rom investing in one more unit o capital) eventually declines as the amount o capital employed increases, value-maximizing irms will invest in capital up to the point where the rate o return on the last unit o capital employed is equal to the real opportunity cost o the unds tied up in that capital, R, plus the loss in the value o the capital due to economic depreciation, δ. I we denote the relative price o a unit o capital with respect to output by q K, the expression q K ( R + δ ) is reerred to as the cost o capital. Denoting the marginal revenue product o capital by MRP K, the equilibrium condition is MRPK = qk ( R + δ ), where capital is accumulated up to the point that the marginal unit breaks even in the sense that it earns just enough to cover the cost o capital. 14 This condition implicitly determines the demand or capital by irms as a unction o the rate o return on capital net-o-depreciation, R. We are now in a position to see how taxes on capital can aect capital accumulation. Investors are concerned with the rate o return on their capital, net o both corporate and personal taxes. Consider irst the imposition o personal taxes on interest income and on the return to equity. Denoting by m the personal tax rate imposed on nominal interest income, and by c the eective tax rate on equity, the weighted average net-o-personal tax real hurdle rate o return becomes R n = βi( 1 m ) + (1 β ) ρ( 1 c ) π. In capital market equilibrium, in the absence o risk and capital market imperections other than taxes, the ater-tax rate o return on equity must equal 12 See King and Fullerton (1984), and Boadway, Bruce and Mintz (1984). 13 In the absence o taxation, or other capital market imperections, and risk, i=ρ. For expositional ease, we abstract rom risk in the discussion. However, the analysis and calculations are consistent with the presence o what is commonly reerred to as income risk. Income risk involves uncertainty regarding uture revenues or operating costs. I we express rates o return net o risk, the METR calculations are identical to those presented here. For a discussion o the implications o other types o risk see McKenzie (1994). 14 We are implicitly treating capital as the numeraire.

20 12 WORKING PAPER the ater-tax rate o return on debt; thus i(1 m ) = ρ (1 c ), which implies that ρ = i( 1 m ) /(1 c ), in which case R n = i( 1 m) π. The imposition o corporate taxes aects investment by altering the cost o capital. There are several ways in which this can occur, depending upon the details o the tax system. Here we present a somewhat simpliied representation, which ignores some o the details o the CIT in Canada. Modiications required to take some o these considerations into account are discussed below. As discussed above, irms will accumulate capital up to the point where the rate o return generated by the last unit o capital is just equal to the cost o capital. To yield R n ater personal taxes, the marginal investment must earn R = βi + (1 β ) ρ π ater corporate taxes (and net o g depreciation). Denote by R K the gross-o-corporate tax, net-o-depreciation rate o return on a g marginal unit o capital required to yield R ater corporate taxes thus R K = MRPK / qk δ. g R K will relect various provisions o the corporate tax system. For example, the deductibility o nominal debt interest expenses or corporate income tax purposes lowers the nominal cost o debt inance to i(1-u), where u is the combined ederal-provincial statutory corporate income tax rate, including the ederal surtax. 15 Nominal interest deductibility thus lowers the cost o capital to the irm by reducing the average opportunity cost o inance to R = β i( 1 u ) + (1 β ) ρ. 16 The cost o capital is also lowered by the reduction in the eective purchase price o capital, due to the presence o tax depreciation allowances and investment tax credits (ITCs). A company that is provided with an ITC at the rate φ and annual depreciation allowances that generate a reduction in taxes o ua in present value terms, A being the present value o the tax depreciation allowances on one dollar o capital, 17 aces an eective purchase price o capital that is lowered by the amount φ+ua per dollar. Taking all o this into account, the ater-tax cost o capital or a irm is qk ( R + δ π )(1 φ ua ). Recognizing that the incremental revenue generated by an additional unit o capital is taxed at the statutory tax rate, leaving MRP K (1 u ) ater tax, the irm's value maximization condition becomes MRP (1 u ) = q ( R + δ π )(1 φ ua ), or MRPK = qk ( R + δ π )(1 φ ua ) /( 1 u ), whereby the marginal revenue product o capital is equal to the gross-o-depreciation, tax-adjusted user cost o capital. Using this equilibrium g condition, and recalling the deinition o R above, the gross-o-corporate-tax, net-o-depreciation rate o return on a marginal unit o capital is: K K R g K = ( R 1 -φ - ua + δ - π ) 1 - u - δ (11) 15 Thus, u = u (1+ s ) + u p where u is the ederal CIT rate, s is the ederal surtax rate and u p is the provincial CIT rate. 16 Note that the required rate o return to equity, ρ, is not deductible or tax purposes. 17 In the absence o various provisions such as the hal-year rule (which is included in subsequent calculations but t t ignored here or simplicity), A is the ininite sum A = (1 φ ) [ α(1 α ) /(1+ R ) ] = (1 φ ) α /( R + α ), where α is the declining balance Capital Cost Allowance (CCA) rate. This relects the reduction in the tax deprecation base by the ITC. t= 0

21 The Calculation o Marginal Eective Tax Rates 13 As indicated above, some aspects o the business tax system in Canada that are relevant to g physical capital are omitted rom the ormulation o R K given in equation (11). Three o these will be discussed here. Provisions that relate speciically to the resource sector are discussed in Subsection below. First, as discussed above, although the GST largely eliminated the taxation o business inputs, including capital, at the ederal level, various provincial retail sales taxes still result in some sales taxes being levied on some types o capital. This is particularly relevant or machinery and equipment; the problem is less acute or structures and land. Second, the ederal government levies a tax on the capital o large corporations with assets in excess o $10 million (the Large Corporations Tax). The ederal surtax is creditable against the LCT. This means that the LCT impacts only on large corporations with an LCT liability in excess o the surtax this is why the LCT is sometimes reerred to as a minimum tax on large corporations. Third, some provinces impose capital taxes on the capital o large corporations as well. Taking these actors into account, the gross rate o return on a marginal unit o capital becomes: R g K = (1+t s )( R 1- φ - ua+( + δ -π ) t +t P(1 - u))/( R 1- u L + α ) - δ (12) The term t s is the eective sales tax rate on machinery and equipment; it serves to increase the cost o a unit o capital by ( 1 + ts ). This term is relevant or machinery and equipment only, as it is presumed that buildings bear no sales taxes. The term t L is the LCT rate, 18 and t P is the provincial capital tax rate. Thus, ( t L + t P ) /( R + α ) is the present value o the ederal and provincial capital taxes that arise due to the purchase o an incremental unit o capital; t P is multiplied by (1-u) to relect the act that provincial capital taxes are deductible or CIT purposes. The expressions to this point have ocussed on investments in machinery and equipment and buildings. The equations may be modiied to develop similar expressions or investments in land and inventories. Land is similar to investments in machinery or buildings, except that we presume that there is no physical depreciation, so δ is set to zero. Moreover, there are no capital cost allowance deductions, no sales taxes and no investment tax credits, thereore A=φ=t s =0, leaving, g 1+( t L +t P(1 - u))/ R RL = ( R - π ) (13) 1- u For investment in inventories, the relevant eature o the tax system is the act that inventory purchases are ully deducted or tax purposes when used, sold or otherwise disposed o. However, the deduction is based on the irst-in, irst-out (FIFO) method o inventory accounting. This means that a deduction or the original cost o the inventory item is made at the time o disposition, and there is no accounting or any increase in that cost due to inlation. As such, 18 It is presumed in equation (12) that the irm pays the LCT [i.e. the surtax is zero]. The METRs calculations relect a weighted average o irms that pay the LCT and those that do not; see Appendix A or more detail.

22 14 WORKING PAPER an inlation tax is imposed on inventory holdings. 19 I we presume that inventories somehow increase the productivity o the irm, then we can use an approach similar to that used or physical capital. The ater-tax marginal revenue product rom investing an incremental unit in inventories is, as above, MRP I ( 1 u ). The opportunity cost o inventories consists o the orgone return on the capital invested in the inventories and the various tax costs. The ater-tax cost o a marginal unit o inventory capital is thus q ( R π + uπ + ( t + t (1 u )). As above, proit maximization requires MRPI (1 u ) = qi ( R π + uπ + ( t L + t P( 1 u )), or, in rate o return orm: I L P R g I = R -π + u π + ( t 1- u L + t p (1 - u)) (14) The Non-Renewable Resource Sector The discussion above ocusses exclusively on tangible capital employed in the non-resource sector. The non-renewable resource sector mining and oil and gas deserves separate treatment, because o its special characteristics and the special eatures o the tax system applied to it. Most important in this regard are the tax treatment o E&D and the incorporation o special resource levies imposed on the resource sector by the provinces. With respect to the latter, the treatment o provincial resource levies, speciically mining taxes imposed upon businesses involved in mining and royalties imposed upon oil and gas companies, is an important consideration. A somewhat contentious issue concerns whether these levies should be treated as a tax or a cost in the calculation o METRs. While rom the perspective o the irms employed in the extraction o non-renewable resources this distinction is largely meaningless at a simple level they are concerned with the cost o employing inputs and producing output, not with how or why a cost arises the matter can be important to policy makers. The approach taken in this study is to treat provincial resource levies as a cost; however, we also present calculations where they are treated as a tax or comparative purposes. Even in the case where resource levies are treated as a cost, mining tax and royalty rates still aect the METRs because o the way that such levies interact with the corporate income tax system. Particularly important in this regard is the act that such levies are not deductible or ederal corporate income tax purposes, rather, a ederal resource allowance is granted instead (more on this below). The Oil and Gas Sector Oil and gas royalty systems are quite complex. For our purposes, it is convenient to simpliy the regime by assuming that a simple average royalty rate applies. In act, royalty rates vary by production, year o discovery, and the price o oil or gas. As indicated above, royalties are not deductible or ederal CIT purposes, but are or provincial purposes. 20 In lieu o the deduction o 19 For more on taxation and the cost o holding inventories, see Boadway, Bruce and Mintz (1982). 20 With over 90% o oil and gas production occurring in Alberta, we model the Alberta system only.

23 The Calculation o Marginal Eective Tax Rates 15 royalties, at the ederal level, a resource allowance equal to 25 percent o revenues net o operating costs and CCA deductions is granted. Notably, the resource allowance base is not reduced by interest expenses or Canadian Exploration Expenses (CEE). The implications o this will be discussed below. For the oil and gas sector, the ater-cit and ater-royalty marginal revenue product or input i is MRPi ( 1 g u (1 + s )(1 σ ) u p (1 g)), where σ is the ederal resource allowance rate and g is the eective royalty rate; this can be written as MRPi ( 1 g)(1 u (1 σ )(1 + s )(1 g) u p ). The ater-tax marginal revenue product relects the act that royalties are deductible or provincial CIT purposes, but not or ederal CIT purposes, where a resource allowance is granted instead. For physical assets in the oil and gas sector (structures and machinery), the ater-tax cost o capital is qk(1+ ts )(R + δ π )(1 φ (u (1 σ ) + up )A+(tL + t p(1 u))/(r + α)), which incorporates the LCT and provincial capital taxes and any sales taxes levied on physical capital, and relects the deductibility o CCA deductions rom the resource allowance base. As above, the optimality condition involves the employment o capital up to the point where the ater-tax marginal revenue product o capital is equal to its ater-tax marginal cost, or MRP K ( 1 g) ( 1 u (1 σ )(1 + s ) /(1 g) u ) = q (1 + t )( R + δ π )(1 φ ( u (1 σ ) + u ) A + p K S ( t L + t p (1 u)) /( R + α)). As beore, this condition can be expressed in rate o return terms by deining the gross-o-tax and royalty rate o return on a marginal unit o capital as g R K = MRPK / qk δ. Expressed in this way, oil and gas royalties are treated as a tax, and we have: p R g K = (1 + t s )( R + δ -π )(1-φ - ( u (1 -σ ) + u (1- g)(1- u (1- σ )(1 + s p ) A + ( t L + t p(1- )/(1- g) - u p ) u))/( R + α )) - δ (15) I royalties are treated as a cost rather than a tax, we deine the gross-o-tax, net-o-royalty rate o g return as R = MRP (1 g ) δ, giving: K K R g K = (1 + t s )( R p + δ -π )(1-φ - ( u (1 -σ ) + u p ) A + ( t L + t (1- u (1- σ )(1 + s )/(1- g) - u p ) (1- u))/(r + α )) - δ (16) Equations (15) and (16) dier rom one another in that the ormer (where royalties are treated as a tax or measurement purposes) multiplies the denominator by (1-g), while the latter (where royalties are treated as a cost) does not. However, in the latter case, royalties still aect the gross-o-tax, net-o-royalty rate o return, so long as the resource allowance rate (σ) does not equal the royalty rate (g). The ormulas or the gross-o-tax marginal rates o return on investments in inventories and land in the oil and gas sector are straightorward to adjust in a similar way, so they are not reported here.