Conceptual Challenges to Taxing Imputed Housing Consumption through Pre-Collection

Transcription

1 Conceptual Challenges to Taxing Imputed Housing Consumption through Pre-Collection Wei Cui China University of Political Science and Law (Preliminary and incomplete draft, June 15, 2011) i Abstract The pre-collection of tax on consumption benefits generated by durable assets such as housing plays a crucial role both in consumption tax theory and in real-world consumption tax regimes. It is also regarded as a potential cure for the non-taxation of imputed rent under many real-world income taxes. However, even under current VAT systems with the widest tax bases, the taxation of imputed housing consumption through pre-collection is incomplete. In response, some have recommended the taxation of resales of residential property in order to capture previously untaxed consumption value. Such proposals have been critically examined in terms of their consistency with traditional VAT doctrines, but their relationship with the basic aims of the consumption tax has not been carefully analyzed. This paper highlights several conceptual issues that proposals for taxing the resale of residential property must confront in order to be consistent with consumption tax theory. Most importantly, they must overcome the presumption that the incidence of any tax on resale will fall on the seller and therefore fail to achieve efficiency gains in either consumer or production choice. Indeed, taxing resale might distort consumption choice by locking in owners in unanticipated rises in consumption. A number of adjustments also must be made to ensure that owner-occupiers of housing are not taxed on the riskless rate of return, risk premium, or inflation-induced increase in value. All such adjustments make clear that taxing resale is indeed unlike other VAT mechanisms. Given the apparent difficulty of normal consumption tax mechanisms in reaching certain significant forms of imputed consumption, the paper finally considers the ability of a property tax to complement the consumption tax by addressing such shortcomings. i I am grateful for the support of a grant from the Peking University-Lincoln Institute Center for Urban Development and Land Policy for the research that produced this paper. 1

2 I. Introduction The taxation of imputed income from the ownership of assets has long been viewed as an important topic in income tax design. If the benefits derived from owning an asset is not included in the owner s income, but if costs (such as interest expenses) incurred in the ownership of the asset are deductible, 1 then there could be significant under-measurements of the asset-owner s income. This could lead to both over-investment in the classes of assets that generate under-taxed net income (due to the non-taxation of imputed income) and to distortions in the choice between ownership and rental. For this reason, many countries have adopted at some point the practice of taxing the imputed rent from ownership of real property, and some countries continue to do so. 2 By contrast, the decision not to include imputed rent from owner-occupied housing in the income tax base in countries like the U.S. and Canada has long been criticized from a policy perspective. 3 The taxation of imputed consumption under a consumption tax raises related but distinct problems. On the one hand, because a consumption tax is intended not to tax the normal return to savings, 4 it generally has no need for depreciation and interest expense deductions. Under the typical consumption-type value added tax (VAT), for example, interest income and expenses received and incurred are ignored in the computation of tax liabilities of businesses, and the cost of any purchase of fixed assets by business taxpayers is immediately deducted instead of depreciated. This means that there are fewer offsetting factors that would mitigate the failure to tax imputed consumption. On the other hand, it has been thought that the consumption value derived from the ownership of a consumer durable asset may be adequately taxed under the pre-collection method. 5 That is, a tax can be imposed at the desired rate on the purchase of the durable asset, even though the purchase itself merely converts one type of asset, cash, into another, and therefore is not consumption. If the purchase price reflects the present discounted value (PDV) of the stream of consumption benefits that the asset is expected to produce, then the tax on the purchase price should equal the PDV of the tax that the government can expect to collect on such consumption benefits as they transpire. Pre-collection, therefore, could substitute for current taxation of imputed consumption. However, although it is straightforward to implement the pre-collection method to tax the imputed consumption generated by a wide range of durable goods e.g. house appliances, automobiles, etc its application to owner-occupied housing, the most significant form of imputed consumption, 6 has been incomplete even in countries with the widest consumption tax bases. This 1 It appears rare to allow depreciation deductions for assets that generate imputed income. This partially compensates for the non-taxation of imputed income. 2 The United Kingdom, for example, adopted the taxation of imputed income arising from the occupation of property as far back as in 1803, and kept the system in place until Other countries known for taxing imputed rent at some point include Germany, Australia, and Sweden. France and the Netherlands still have regimes that tax imputed rent on some types of property. Ault and Arnold (2009), pp. 3 See e.g. U.S. Department of Treasury (1977), pp More precisely, the consumption tax is not intended to tax the time value return to savings. Whether it should tax risk premia is less clear if one thinks that the normative justification for taxing risk prema is unclear. See Zelenak (2006). In any case, it is perfectly coherent for the consumption tax to tax any supra-normal (or infra-marginal ) return to savings. See, e.g. Bankman and Griffith (1992) and Bankman and Weisbach (2006). 5 See, e.g. U.S. Department of Treasury (1977), p 123; Bradford (1988). 6 According to the Bureau of Economic Analysis, U.S. Department of Commerce, for example, expenditures on housing services constitute approximately 15% of total personal consumption expenditures in the United States in 2

3 is mainly due to two factors. First, the relatively young nature of the VAT (the most common form assumed by broad-based consumption taxes) for most countries means that there is a significant portion of housing stock that has not been subject to tax on their initial sales. Second, almost no country subject the resale of residential property to the VAT. This not only means that previously untaxed residences continue to be untaxed, but also that the increased consumption value of housing that is attributable to unexpected appreciation fall outside the reach of the pre-collection method. 7 Consumption tax theorists who have identified this issue seem to agree that it is a flaw in existing tax systems. (Ebrill et al 2001, pp 98-99; Cnossen 1995). Some continue to advocate for improved design in the context of either the reform of existing VATs in some countries or the first-time introduction of VATs in others (Schenk and Oldman 2007, pp ), while others believe that political resistance would render more fulsome consumption taxation of housing unlikely in most countries (Poddar 2009). Overall, though, the issue has not received close theoretical examination: it has the awkward status of an institutional design topic still waiting for the right type of institutional setting to become salient. Once one takes a global perspective, however, the topic will inevitably assume practical urgency on some occasions. For example, taxing the resale of residential properties was reportedly seriously discussed when Canada enacted its Goods and Services Tax (GST) legislation (Poddar 2009, p 459). Such taxation is also embodied in the current Chinese consumption tax treatment of housing, and the question of whether it should be continued during further VAT reform is being considered (Cui 2011). This paper argues that the main challenges faced by proposals to extend the pre-collection method for taxing imputed consumption to resales of residential properties may not be political or administrative. Instead, they may be more fundamentally conceptual. The paper highlights three conceptual challenges. The first arises from reflections on the economic incidence of a tax on resale: there are strong reasons to believe that such incidence is likely to fall fully on the seller of property, whereas the actual consumers of housing, per consumers, do not bear the tax. This is both because the existing housing stock is by definition in-elastically supplied, and because any additional taxable value found in previously taxed property represents infra-marginal rent. This means that a key objective of consumption tax policy, namely ensuring that different consumption goods are taxed at similar rates and that consumption choices are not distorted by taxation, is not furthered by taxing the resale of residential property. Similarly, since exiting housing by definition has already been produced, taxing it cannot enhance production efficiency. Moreover, because the burden of the consumption tax on resale is likely to fall on the seller, its economic effect will be similar to a capital gains tax. If the latter is already imposed on the resale of residences, the imposition of an additional tax may aggravate the lock-in effect on investment decisions. A second challenge to the idea of applying pre-collection to resales of residential properties is that it is at best an approximation of taxing imputed consumption. Such taxation may be avoided by deferring resale, even when such deferral otherwise implies lower utility. The distortion is distinct from the lock-in effect associated with realization-based income taxation in that it leads 2008, whereas other household related consumption (e.g. utilities and appliances) represented an additional 10%. See Poddar (2009) at These two factors are discussed in Poddar (2009). 3

4 to tax-induced over-consumption by the owner of housing. From this perspective, taxation of resale is inferior to taxing housing consumption on a current basis, e.g. through a property tax on housing based on market value. The third challenge to proposals of taxing residential resales in order to implement a consumption tax is that one needs to avoid overtaxing the return housing may generate as an investment asset. At the minimum, this requires one to exempt, from the VAT imposed on resale, the amount of gain that reflects a risk-free rate of return, as well as any nominal appreciation due to inflation. Somewhat less obviously, a risk premium should also be excluded, because the government does not undertake any risk in pre-collecting tax on the purchase of a risky asset (leaving the risk entirely to the taxpayer), and the consumption tax generally does not tax risk-taking. What these challenges show is not only that there is much more work to do for consumption tax advocates in devising real world rules (even abstracting from the politics of particular countries), but also that it may be important to step back and re-consider the relative advantages of different taxes. Specifically, it may be that any transaction-based tax cannot play the role of taxing imputed consumption as well as a wealth tax could, and the property tax should be more heavily relied on for taxing imputed housing consumption. That is, one may ultimately have to look to a broader set of tax policy instruments rather than devising cleverer mechanisms within the VAT. This paper is organized as follows. Section II reviews the pre-collection method for taxing imputed housing consumption, its flaws if only applied to first sales, and existing proposals for taxing residential resales to remedy such flaws. Section III discusses the first and second challenges to the idea of taxing residential resale, i.e. why taxing resale may fail to affect consumption choices regarding housing, and why it may be inferior to a property tax in reaching imputed consumption. Section IV considers how, even putting the first two objections aside, the taxation of residential resale must be designed in order to avoid undue taxation of the returns to investment. Section V distinguishes the foregoing challenges from other criticisms levied at the proposal of taxing residential resale, and also examines whether there are other policy justifications for taxing residential resale that are not subject to the foregoing objections. Finally, Section VI discusses why consumption tax theorists may have to look to other taxes (such as the property tax) to achieve some of the inherent goals of the consumption tax. II. The Pre-Collection Method of Taxing Imputed Housing Consumption Buying a house or an apartment is, in some fundamental ways, just like buying durable consumer goods like refrigerators, cars, and furniture. The act of the purchase is not itself, strictly speaking, an act of consumption. Instead, consumption happens when the durable good is used. A uniform tax on consumption should tax the value of such imputed consumption just as it taxes other forms of consumption of goods and services. However, for durable goods, the way this is usually done is not to collect the tax when the imputed consumption happens but when the durable good is purchased. The purchase price of a consumer durable generally reflects its consumption value during its useful life. Assuming that the purchase price equals the sum of the PDV of the value of the use of the good during each future period, a tax imposed at rate T on the purchase 4

5 price equals the sum of the PDV of tax payments that could be collected at the same rate, T, from the imputed consumption of the good during future periods. Algebraically, (1) V*T= t [T*c t /(1+r) t ] where t = time period; c t = the value of the consumption use of the good in period t; r= rate of discount; T= tax rate applicable to all periods; and V= purchase price of the durable good. Where this equivalence holds, taxing purchases is equivalent to pre-collecting the tax on the imputed consumption of the purchased goods. This is how countries with VATs generally tax the consumption of durable goods. 8 Extending the pre-collection method to imputed housing consumption implies taxing at least initial purchases of housing, which is what Canada, Australia, New Zealand and a number of other countries currently do under the modern VAT. 9 In the context of traditional discussions of VAT design, this method is considered superior to subjecting only the material and labor inputs to housing construction to the VAT but exempting the sales of residential properties themselves. 10 Our focus, however, is not on the comparison with traditional VAT regimes but on how successfully the pre-collection method achieves the ideal of taxing housing consumption. 11 The holding of equivalence (1) above depends on various assumptions, 12 and for housing, the most important that may be violated is that the purchase price fully reflects future consumption value. In certain historical periods and in certain locations, housing, like the real estate sector in general, may witness unexpected appreciation increases in value that have not been fully anticipated nor capitalized into purchase prices. Although such changes may occur for all types of reasons and there may be unexpected devaluation as well an important reason is enhanced locational premium. Urbanization, the building of new transportation pathways and amenities, unexpected rises in income in the local population, etc. may all enhance the value of real property in ways that could not easily be predicted. 13 Such unexpected appreciation normally does not happen to other durable consumer goods, although it may happen to artworks, collectibles, etc. Where it does happen, the pre-collection method under-taxes imputed consumption. In addition, even where equivalence (1) does hold, taxing imputed housing consumption only through pre-collection may result in under-taxation if a residence was bought before the VAT is introduced. While this transitional issue is arguably trivial for consumer durables that have limited useful lives and tend not to be resold for substantial values, the fact that residential properties have long useful lives and their resale represents a substantial part of the housing market makes such properties different. If residential properties are subject the VAT only on their first sales, then the 8 The superiority of this method from an administrative perspective is easily appreciated, when compared to the alternative of having all consumers pay tax for the consumption of the goods they own on a current basis. 9 By contrast, under the traditional European VAT, sales of residential property used to be subject only to turnover taxes that are not part of a comprehensive consumption tax. 10 The superiority lies in the fact that while the tax on such inputs is to some extent borne by the final consumers of housing, it causes distortions in productions decisions (for example by encouraging house builders to self-supply in order to avoid VAT in the production process.) 11 Existing discussions of VAT design also dwell on how equal treatment between rental and owner-occupied residences can be attained. See Poddar 2009, pp (comparing three options in addition to the approach of the modern VAT ). This paper focuses on the problems of taxing imputed housing consumption, i.e. owner-occupied housing. 12 For one, the tax rate obviously needs to remain constant. The discussion in this paper will focus only on those assumptions that are especially likely to be problematic for real estate. 13 Taking such enhancements into account in purchase price would be essentially an act of speculation. 5

6 housing stock existing at the time of the introduction of the VAT will never bear VAT. This will deliver existing housing that is resold a price advantage relative to newly constructed housing, and therefore windfall gains to the owners of existing housing. In this regard, owner-occupied housing is different from other existing capital stock in an economy in terms of the impact of the introduction of a VAT. The consequences for existing capital stock of the introduction of a consumption tax like the VAT, particularly to replace an income tax, have been carefully analyzed 14 and can be thought of in terms of two subcategories of effects: carryover effects and price effects. 15 The former refer to the possibility of assets purchased with income that had already been subject to the income tax being taxed again under the consumption tax. This would result in a substantial windfall loss to the owner of such assets. 16 Existing unsold housing inventory and housing that would be rented on the rental market would be subject to this fate if no special transition rules are designed. However, as has been observed, existing owner-occupied housing would not suffer in this regard insofar as imputed rent is (aside from the pre-collection method) not taxed (Bradford 1996, p 140). Price effects, on the other hand, refer to the impact on the relative prices of assets due to new tax treatments and changes in interest rates. As an example of the former, if owner-occupied housing is favorably treated under the income tax but not under the consumption tax, a transition from the income to the consumption tax would lower the relative price of housing. However, in the event that a portion of existing housing stock, i.e. those that are already owner-occupied, is not subject to such uniform consumption tax treatment, such assets again suffer no detriment from the transition. 17 A natural response to problems of the failure to tax consumption corresponding to (i) unexpected appreciation, and (ii) existing housing, is to implement the pre-collection method with respect to secondary sales as well. For properties that were not subject to the VAT on their initial sales, 18 imposing the VAT on the values of their secondary sales obviously can play the role of pre-collecting tax on their inherent consumption value for the rest of their useful lives. What is perhaps less intuitive is how imposing the VAT on the resales of residences that have already been subject to the VAT on previous sales can be justified. Currently the most explicit proposal, advanced by Poddar (2009), is that while the standard VAT rate is applied to the value of the resale, the seller can take a credit for any VAT paid on his earlier purchase of the residence. 19 If one assumes the VAT rate has remained the same between the earlier purchase and the resale, this is equivalent to imposing the VAT rate on any difference between, on one hand, the selling price and, on the other, the sum of the original purchase price and any cost of improvement already subject 14 See, e.g. U.S. Department of Treasury (1977), Chapter 6; Sarkar and Zodrow (1993); and Bradford (1996). 15 U.S. Department of Treasury (1977), pp See also Kaplow (1995). 17 In terms of the transition effect through interest rates, if eliminating the income tax increases the after-tax return to saving, it diminishes the valuation of all existing assets (which had been priced in light of a lower after-tax return). But to the extent that the before-tax return to saving is lowered as a result of the greater supply of capital, existing capital may reap a windfall gain during the period of adjustment when new capital stock is constructed. Where one is considering only the introduction of a VAT or an increase in VAT rates (and not a replacement of the existing income tax), this type of effect presumably will be less pronounced. 18 In this paper, initial sales may be understood to include the first-time putting into use of self-constructed residences. Mature VAT systems tend to have self-supply rules that ensure self-constructed property is treated in the same way as property constructed and sold to third-parties. 19 Credit can also be taken for any VAT paid when making improvements to the property. The total credit cannot exceed the amount of VAT chargeable on the resale price. 6

7 to the VAT. The idea is that this difference approximates the unanticipated appreciation in consumption value of the residence, neglecting which would result in under-taxation. To the Author s knowledge, this proposal for taxing residential resale has so far been discussed only in the context of VAT design. As reviewed in Section V below, it has been criticized (e.g. in Millar 2011) for stretching the conception of VAT too far. However, the more fundamental question is likely not whether the proposal can cohere in legal and other terms with normal VAT apparatus, but whether it is justifiable in terms of consumption tax theory. It is this latter question that motivates the inquiry in the next two sections. III. Creating instead of Removing Distortions in Consumption Choices Consider the economic incidence of a pre-collected consumption tax on housing. Because the tax is a part of a general consumption tax that applies to all or most goods and services, the analysis of the incidence of the tax on the capital and labor that go into the production of housing may seem not to promise any distinctive results. However, it is notable that the tax on the portion of the property s purchase price that is attributable to land (and implicitly the location) would be borne by the owner of the land. It is only the remainder of the tax that is (potentially) borne by capital, labor and consumers. In other words, a portion of the tax on the consumer good a residence is entirely shifted back to the seller (insofar as he is the landowner). This suggests that the consumer of housing, in her capacity as such a consumer (i.e. disregarding the fact that she might also be a landowner) is bearing less of the burden of the consumption tax than on many other consumption goods and services. This itself is arguably unremarkable in the context of newly constructed housing, since the factors that go into the production of consumer goods and services may all have different price elasticities: there may be other factors that are relatively inelastic in supply, and land may be just an extreme example. 20 However, these reflections take on an unusual significance in the context of a VAT imposed on housing resale. First, consider an instance of resale where the property sold had already been previously subject to the VAT (i.e. where the consumption tax had already been pre-collected on the then-expected consumption value of the property). The point of subjecting the property to a VAT again, recall, is mainly to capture any unexpected appreciation in the (expected) consumption value of the property. 21 However, such unexpected appreciation generally cannot be explained by the properties viewed as physical structures: the value of that should generally depreciate. Any actual improvements made to the physical structures should also have been subject to the VAT and therefore would not create a net tax liability on resale. Instead, the most important reason for such rise in housing value is locational premium. But such premium, besides being tied to land, is by definition a form of infra-marginal return. A tax on it is likely to be borne by the claimant to such return the seller of used residential property and unlikely to be shifted 20 A VAT applied at a uniform rate to all consumption does not attempt to be an optimal tax by taking elasticities into account. Instead it aims at achieving production efficiency i.e. causing minimal distortions in the choice of factor inputs. Ebrill et al 2011, pp. 21 Taxing resale may also have the effect of not distorting choices of inputs (between those subject to the VAT and those, such as self-supplied services, that are not) for any renovation of the property. Some countries (such as Australia) that do not generally tax the resale of residential property have rules that treat sales after substantial renovations as first sales, thus dealing with the most important range of cases where such distortions need to be avoided. 7

8 forward onto the buyer. A similar conclusion holds if one considers taxing the resale of owner-occupied housing that had not previously been subject to the VAT, i.e. the stock of owner-occupied housing that existed at the time of the introduction of the VAT. Existing housing stock is just like land and is inelastic in supply (except, of course, to the extent that such stock depreciates). Accordingly, the burden of the VAT on the sale of this type of property should also not shift forward to the purchaser. 22 However, if the burden of any net tax liability generated by the tax on the resale of owner-occupied housing is not borne by the buyer, then the pre-collection method must be viewed differently. Even if the taxation of resale is designed not to burden the normal return to investment (we will discuss in the next section what is required to achieve this end), the burden of the tax would not fall on the actual consumer. It would thus not be able to correct the over-consumption of housing due to under-taxation. 23 Besides not being able to help impose the same tax burden on all forms of consumption, pre-collection, insofar as it is applied to consumption value attributable to locational premium and past production decisions, also has no effect in enhancing production efficiency in an economy. At best, it may capture some of the windfall gains that accrued to owner-occupiers of housing constructed and sold before the VAT s introduction. Extending the pre-collection method to resale thus seems only tenuously connected with the basic aims of consumption taxation. Indeed, in substance as well as appearance, it has more affinity to a capital gains tax (or, if it is designed to exempt the normal return to investment, to a pure profit tax). As such, it may be imposed at very different rates from the regular consumption tax. This view of the nature of a consumption tax on resale of owner-occupied housing is the first, and perhaps most fundamental challenge, for proponents of such a tax. It makes some of the flaws of the resale method less tolerable. We now turn to these flaws, relating to mis-measurement and a lock-in effect for consumers. Because the expected and actual consumption values of an owner-occupied residence may differ, the pre-collection method could easily under-tax or over-tax the imputed consumption. For example, a mere rise in the consumed value of the property (e.g. because of better transportation and other amenities nearby) will itself not create an additional consumption tax liability, as long as there is no sale of the property. When the property is sold, it is only the rise in its future expected consumption value that would give rise to any tax liability, and not the increased consumption value that has already been enjoyed. Moreover, the difference between the resale value and the original purchase price may under-measure the increase in value in future imputed consumption. 24 Similar remarks can be made concerning when the actual consumption value of a property turns 22 One may renovate existing property and the degree of renovation is elastic. However, such renovation is not likely to generate substantial net VAT liability if resale is taxed, and as already observed, many VAT systems already have rules for treating substantial renovation as though it is first-sale. 23 In fact, as discussed below, it may have the effect of forcing such over-consumption by the current owner-occupiers of housing. 24 That is, if the value of a property at its initial sale is V 0 = t=(0,b) [c t /(1+r) t ] (where b denotes the last period of the property s useful life), and if its sales price at time a is V a = t=(a, b) [c t /(1+r) t ], then V= V a - V 0 = t=(a, b) [ c t /(1+r) t ] - t=(0, a) [c t /(1+r) t ]. The second term in the right side of the last equation shows the amount of under-valuation. 8

9 out to the lower than expected (e.g. because an expected amenity was not built). 25 More serious than the issue of mis-measurements is behavioral distortions. If the owner-occupier does not sell a property that has experienced unexpected appreciation, he could enjoy the increased consumption value without taxation under the pre-collection approach. This results in a consumption tax subsidy that the owner would lose by selling, and a corresponding lock-in effect can be expected. This effect is distinct from the lock-in effect associated with the realization principle under the income tax in several ways. First, instead of creating distortions in investment choices, it maintains a price distortion between the consumption value of a house and other consumption alternatives. Second, none of the recent theoretical proposals for undoing the lock-in effect without abandoning the realization principle under the income tax seems to work for locked-in consumption. Traditionally, concerns with the distortions (as well as the unfair advantages of deferral) caused by the realization principle have led to proposals to its abandonment and the adoption of mark-to-market or other similar methods of income taxation. More recently, however, it has been suggested that taxation only upon realization can be retained but its problems eliminated through the adoption of retrospective taxation (Auerbach 1991, Cunningham 2000). Under retrospective taxation, investments are taxed on their assumed gains upon liquidation and interest is charged on any deferred tax payments. Proposals of retrospective taxation, however, are largely framed in terms of the need of the income tax only to reach the risk-less rate of return on investment assets. 26 They do not deal with the need to tax consumption, and therefore not with the difference between taxing consumption currently or on a pre-collection basis. It thus appears that only a tax that accurately measures and taxes imputed consumption on a current basis (corresponding to the mark-to-market method of income taxation for investment assets) may avoid the lock-in effect created by the pre-collection method of taxing imputed consumption. This, of course, amounts to abandoning the pre-collection method. To summarize this section: taxing housing resale appears to be an inadequate solution to the under-taxation of housing consumption because (i) there is continued under-taxation prior to resale, (ii) the tax on resale could not be expected to shift forward to purchasers, and therefore fails to enhance efficiency in consumer choice (or in production choice), and (iii) indeed the tax on resale may distort consumer choice through its lock-in effect. Such mismatches with the basic goals of the consumption tax imply that the pre-collection method suffers from graver problems than its crudeness in measurement. IV. Indexing for the Risk-Free Return, Risk Premium, and Inflation in Taxing Resale Despite the fundamental misgivings just suggested regarding whether taxing residential resale constitutes even the right direction in which to develop comprehensive consumption taxation, this section explores certain refinements of the proposal for taxing resale, with two 25 By contrast, if there was a property tax that accurately measured and taxed imputed consumption on a current basis, no increase in the value of imputed consumption would escape taxation. 26 There is no need to deal with risk premiums because, at least in the idealized settings dealt with in these proposals, taxpayers can adjust their portfolios to neutralize the effect on any taxation on risk-taking. And economic profit is generally not considered. 9

10 different motivations. The first is to formulate such proposal in a way that at least addresses some less detrimental objections to it. The second is to highlight an aspect of the pre-collection method that is somewhat insufficiently discussed, namely how it avoids subjecting risk-taking to tax. The need to refine the proposal for taxing resale arises first as follows. Because of the enduring nature of residential properties and their (perceived) propensity for appreciation, many purchasers of such properties treat them as both consumption and investment assets. That is, when one purchases a house or apartment, one may consider both its rental value and its resale value. While a consumption tax should subject the consumption value of the asset in the form of either actual or imputed rent to tax, any pure investment return generated by the asset should, in theory, be left out of the consumption tax base. The question is how to do this when the pre-collection method is extended to resale. Consider an example. Suppose that an apartment is on sale, and for each of the next 5 years, it is expected to generate zero net income flow the rental value it generates each year is offset by expenses in the same amount. 27 But a potential buyer, A, believes that when the asset may be sold at the end of year 5 for $100x. Suppose A s discount rate is 5% (we will come back to how to interpret this discount rate). Then A should be willing to pay up to $78.35x for the apartment. Suppose that is what A does, and that, as she correctly anticipates, the apartment sells for $100x at the end of year 5. If a VAT is imposed not only on the initial purchase price of $78.35 but also to on resale, how much of the appreciation in the value of the asset should be subject to VAT? If the aim of the consumption tax is not to tax the return to investment, the intuitive answer is, None, as the appreciation merely reflects the accrual of investment returns. If resale were to be subject to tax, such appreciation should be excluded from the tax base. One way of doing this, for the above example, is to adjust the initial purchase price of the apartment upward by the market rate of return. 28 The purchase price so adjusted would be $100x, and the imposition of the VAT on the resale for $100x would generate zero tax liability. There would be an amount subject to VAT on resale only if the resale price exceeded $100x, the adjusted purchase price. Such an adjustment of the initial purchase price would of course not be straightforward if, as is more commonly the case, the expected net consumption value of the apartment between the initial sale and the resale is not zero. If the PDV of the first 5 years' net consumption benefits generated by the apartment contributes a positive amount to the initial purchase price, it would not make sense to adjust that portion of the purchase price by a rate of return when computing tax liability upon resale. In fact, if one could know what that portion was, one would want to subtract it from the initial taxable purchase price of the asset when taxing resale in year 5, because only then could one measure the appreciation in the apartment s expected consumption value after year 5. That is, one would have to adjust the initial purchase price both (i) downwards for actual 27 To make this example more specific, we may suppose that such expenses are incurred for taxable services such as repair and maintenance. Thus if the property were rented out at market value, and the landlord subject to the VAT for rental receipts, he would not subject to any net VAT liability the input tax credit for repair and maintenance services entirely offsets the output tax. If the property is owner-occupied, the landlord effectively would have borne the VAT on his housing consumption for the current period by paying tax on the input service (and not being able to take a credit for such tax). 28 This is the amount that was subject to VAT on initial sale, and treated as an input credit (in Poddar s proposal) or, perhaps less metaphorically, simply as a deduction, when VAT is imposed on resale. 10

11 consumption transpired and (ii) upwards in order to exclude investment returns from the consumption tax base. This complication need not be regarded as detrimental for the resale method. Apparently, under the income tax (not the consumption-type VAT) in Japan, imputed housing consumption is subject to tax to some extent due to the following rule: In general, a deduction for depreciation is not allowed with respect to personal assets. Nonetheless, the taxpayer must adjust the basis of the personal asset as if depreciation had been taken. If the personal asset is later sold at a gain, the gain must be calculated from the adjusted basis. (Ault and Arnold 2009, p. ). The resale method of implementing a consumption tax may adopt similar mechanisms. In any case, the example above shows that some form of indexing may be necessary to distinguish between anticipated and unanticipated appreciation in an asset: it is only the latter form of appreciation that justifies taxing the resale of previously taxed residential properties. One interesting and important question that can be pressed here is whether the rate of return by which the original purchase price is adjusted should be the risk-free rate of return or some risk-adjusted rate. Adjustment by the risk-free rate seems uncontroversial, but it may not be sufficient. One infrequently noted feature of the pre-collection method for implementing the consumption tax is its treatment of risk-taking. Recent scholarship has tended to agree that the consumption tax, just like an ideal income tax, generally does not impose a burden on risk taking. For instance, under a cash flow consumption tax, any amount that is not currently consumed but put into investment would be excluded from the consumption tax base; the amount excluded and the yield it generates would be taxed only when withdrawn from investment for consumption. In effect, then, the public becomes a partner in the taxpayer s investment activities. Accordingly, the taxpayer may make appropriate portfolio adjustments to achieve a risk exposure and expected returns similar to what he could obtain had there been no tax (just as he may do so in response to an ideal income tax).(weisbach 2004) Similarly, under a VAT, at least for business taxpayers, any investment outlays are immediately deducted in the computation of VAT liability. As a result, the general public shares in the investment and payoffs in proportion to the tax rate[; in] making the investment, the taxable firm considers its share. (Bradford 196, p 132) The pre-collection method, however, works differently. The investment is fully taxed upfront and there is generally no subsequent taxation. What this implies in terms of the tax burden on risk taking is less often discussed. 29 But it is not difficult to see how the pre-collection method also avoids taxing risky returns. If two assets, one safe and one risky, sell for the same price, the risky asset should generate higher expected returns. Conversely, if two assets, one safe and one risky, generate the same expected returns, the risky asset should be cheaper, since the greater risk must be compensated by greater yield. However, when a VAT is pre-collected on the purchase of a consumer durable or, more generally, when the consumption tax is pre-paid on an amount that is used for investment and 29 David Bradford has pointed out that, for imputed consumption taking place in the tax-exempt household, the general public does not share in the investment or the return. The investment decision is based on the full cost and the full return. (Bradford 196, p 132) The next paragraph in the text can be seen as a way of further fleshing out this idea. 11

12 deferred consumption no distinction is made regarding whether the investment is made in a safe or risky asset. The government collects the same amount of tax on the purchase of a safe asset as it would on the purchase of a risky asset sold at the same price. This means that the government is not really pre-collecting the tax that would be due on the periodic returns generated by the risky asset. Those returns, being risky, would have generated on average a higher amount of total tax payments, but the government undertakes no such risk in making a pre-collection: instead, it collects a sum certain. To put it differently, if the government were to fully tax the expected returns of the risky asset, it would also have to discount such returns by a higher yield to compensate for the risk. Converting such tax payments into a pre-collected amount changes not only timing but the nature of the risk, and means a lower amount of prepaid tax (just as the purchase price would be lower than a safe asset with the same expected return). By collecting a sum certain, the government cedes the risk in the asset entirely to the taxpayer. Thus the pre-collection method avoids burdening risk taking, by requiring the taxpayer to remit only the certainty equivalent amount of the tax that could be expected to be due on the consumer durable/investment asset. 30 This requires no adjustment of investment portfolios, and is thus quite different from the rest of the consumption tax apparatus (i.e. in a cash flow consumption tax or under the typical VAT treatment of business taxpayers), which adopts the post-paid method. 31 Once we recognize this, it follows that if the pre-collection method were to be extended and imposed on the resale of housing, the original purchase price (possibly depreciated to reflect net positive consumption that has already occurred through the use of the asset) must be adjusted upwards by a risk-adjusted rate of return, i.e. by the riskless rate of return plus a market risk premium. This is because the taxpayer has taken the full risk associated with the asset during the period between its initial purchase and its resale, whereas the government has taken no risk. If the original purchase price were not increased by the risk-adjusted return, the government would certainly be taxing risk-taking, resulting in an inconsistency with the rest of the consumption tax regime. What about the fact that the government only taxes any appreciation on resale but does not compensate for any depreciation? This is the approach of existing proposals for taxing residential resale: if the resale price is lower than the original purchase price (as adjusted), there is no refund of any consumption tax previously paid. 32 In essence, under existing proposals, the government would be taxing any actual return above the expected return of an asset (i.e. when the taxpayer is lucky), but would not be compensating for any actual loss if the expected return is not achieved (i.e. when the taxpayer is unlucky). This changes the payoffs of taxpayers bets, and seems to affect choices involving risk in a way that cannot be neutralized through portfolio adjustments. 30 The conversion to the certainty equivalent amount is at the taxpayer s own rate for discounting risk. 31 I have not seen a full exposition of this feature of the pre-collection method, despite the wide agreement on the importance of the method for real world consumption tax implementation. 32 One might be tempted to rationalize such asymmetrical treatment by claiming that, for legislative and administrative simplicity, the case of devaluation can be ignored, since it is unusual for housing to lose value, but the persuasiveness of such claim has been somewhat diminished in many housing markets since the global economic downturn in

13 If the government were to recognize losses realized on resale, it would then be taking on some of the risk associated with the asset, which it formerly refrained from taking on when collecting only a sum certain on the initial sale. And if the government does assume risk in this way, taxpayers may respond with portfolio adjustments under the normal consumption (and income) tax mechanisms. The tax burden on risk-taking would again be neutralized. But if the proposal to tax resale does not incorporate the loss recognition feature, it can only be justified by the premise that substantial losses are much more infrequent than substantial increases in housing value. Even if one accepts this premise, the adjustment of the original purchase price when computing the tax liability on resale by the riskless rate of return and a market risk premium is necessary. This naturally leads to the idea of indexing for inflation as well. Generally, the consumption tax does not raise the difficulties of avoiding taxing the nominal increase in asset value resulting from inflation that confront the income tax. (Shakow and Shuldiner 2000) Both the taxation of consumption in the current period and the pre-collection of the tax on future consumption do not result in taxing the inflationary increase in value. If the resale of residential property is subject to tax, the problem of such inflationary increase comes back, but it can be easily dealt with by inflation-indexing the original purchase price. In summary, the taxation of the resale of residential property, if it were to be consistent with the rest of the consumption tax in achieving the non-taxation of the time value return of investment, risk taking, and inflationary appreciation in asset, must be designed with a number of mechanisms normally not required for any input credit scheme under the VAT. And even after such refinements, the taxation of resale, by the arguments in the last section, still fails to advance two positive objectives of the comprehensive consumption tax: ensuring as far as possible that all consumption is subject to taxation at the same rate (so as not to distort consumption choices), and that taxes do not distort producers choices of input. In short, the obstacles facing the proposal to tax resale are not primarily political or administrative; they are conceptual. [V. Other Arguments for and against Taxing Resale ] [Section to be completed] VI. Resorting to Other Tax Instruments We saw in Section IV that it is fairly difficult for consumption tax mechanisms to reach certain imputed housing consumption value. The pre-collection mechanism is itself already distinguishable from the rest of the consumption tax apparatus (under the VAT, the cash flow consumption tax, or a number of other variants) in terms both of timing and the treatment of risk-taking. If extended to resale, it would, for some at least, stretch the VAT beyond recognition. But even so, it may not impose a tax burden on the portion of imputed housing consumption attributable to unexpected appreciation and previously untaxed housing stock due to reasons of economic incidence. This means that, as far as owner-occupied housing is concerned, pre-collection is a very imperfect substitute for taxing imputed consumption on a current basis. The taxation of imputed consumption as it transpires has long been regarded as 13

14 administratively difficult, although, as noted in the Introduction, a number of countries have practiced taxation of imputed income from real property ownership and some continue to do so. More common than the taxation of imputed income, however, is the property tax widely imposed on residences (as well as non-business, non-residential property and business property). In fact, some have regarded the property tax as partially remedying the non-taxation of imputed income under the income tax. (Zodrow 2008) Property tax valuation techniques have been improving for decades and continue to improve, which seems to enhance its eligibility as a complement to certain parts of the income tax. Is it possible to conceive the property tax on residential property as a similar complement to a consumption tax in respect of owner-occupied housing? Alarge number of issues need to be considered in answering this question, and the following attempts to sort through some of them. First, a property tax on residences, as a particular instance of a wealth tax, imposes a tax burden on the time value return of investment, and therefore could not be regarded as a part of any consumption tax. 33 Nonetheless, in many countries, a consumption tax and an income tax are imposed simultaneously, and so it would not be terrible if the property tax, at the same time as subjecting imputed housing consumption to tax, also burdens some other forms of income. Any income tax burden in addition to what burden is borne by consumption can, if undesired, be neutralized (for example by allowing a credit in computing property owner s income tax liability). 34 Second, the property tax is generally imposed on all real property, whether business or personal, and in the latter case, whether owner-occupied or rented. Insofar as business or rented residential properties are treated differently under a consumption tax from owner-occupied housing, 35 and insofar as the problems of under-taxation of imputed consumption do not arise for such other types of properties, the property taxation of such other properties of course cannot be regarded as playing the role of complementing the consumption (or income) tax. Perhaps requiring further consideration is the fact that if different uses of real property are subject to different tax burdens (under one tax or the combination of several taxes), the deployment of resources including land among these different uses may be influenced by tax considerations. For example, land may not be in-elastically supplied relative to a particular use of land. Third, the consumption tax, especially the VAT, is often conceived of as a national tax. Even though sub-national jurisdictions may also impose broad-based consumption taxes, the challenges for consumption taxation raised by imputed housing consumption are probably regarded as most important for a national tax. On the other hand, in most countries, the property tax is designated as 33 However, it may be possible to design the property tax in such a way (e.g. excluding from taxable value each year an amount equal to the time value return on the property) as to remove the burden on such return on investment. 34 In terms of its treatment of other components of investment return, Shakow and Shuldiner (2000) argue that a net wealth tax does not burden risk taking and fails to reach infra-marginal rent, although the latter point is not clear. The property tax is imposed typically without taking into account debt encumbering the property, and so may differ from a net wealth tax in that it may subject infra-marginal return to tax. 35 Property held in business use should not attract any consumption tax burden. For rented residences, if they are subject to the consumption tax (e.g. VAT) on current rental value and if the VAT on the purchase of the property is immediately deductible-refundable, then they are essentially treated as business property. On the other hand, if they are not subject to the VAT on rent and if VAT paid on acquisition is not deductible/refundable, they are treated similarly with owner/occupied housing. 14