Economists have been debating land value taxation ever since the Physiocrats proposed the impot unique at the very birth of their discipline over two centuries ago - and they are still deeply divided over its effects and significance. The Australian states have been implementing a variety of land value taxes for a century, and so furnish some of the best opportunities for resolving the economists' disagreements empirically. However, before examining the evidence on the ground we must first attempt to untangle the obstructing thicket of theory which they have woven.

The concensus amongst economists today is to affirm the first two, deny the third, and remain agnostic about the fourth. But the qualifications (even outright contradictions) hedging about the affirmations are such as to give the impression that it is a fiscal blunderbus best left in the policy cupboard.

The fourth question has been treated by the author in his contributions to the first volume of The Sisyphus Syndrome. The other three are the subject of this chapter.

Frustrated by the intellectual deadlock in the economics profession over these questions, an American industrialist, who had stood for Vice-President of the US in 1924 specifically to lift LVT out of the cupboard, left a bequest which is used to fund a non-profit making educational institute to explore land policy issues - the Lincoln Institute of Land Policy, based in Cambridge, Massachusetts. In September 1991 the Institute organised an International Conference on Property Taxation and its interaction with Land Policy.

Amongst the papers presented were two on "current data on land taxation", one by Wallace Oates and Robert Schwab of the University of Maryland, dealing with Pittsburgh, USA,[1] and one by Kenneth Lusht of the College of Business Administration, Pennsylvania State University, covering Melbourne, Australia.[2] Both papers began by summarising the theoretical state of play on the impact of LVT. The inconsistencies which emerge between the two treatments - and with other economists whom they do not cite - provide an entry into this debate.

A tax in proportion to the annual (rental) value of land, or to its capital (sale) value, has generally been recognised as avoiding distortion of markets, unlike many taxes. Thus, Dick Netzer wrote in 1966 for the Brookings Institute, in The Economics of the Property Tax,

Location rents constitute a surplus, and taxing them will not reduce the supply of sites offered; instead, the site value tax will be entirely neutral with regard to landowners' decisions, since no possible response to the tax can improve the situation, assuming that landowners have been making maximum use of their sites prior to imposition of the tax (quoted in Oates and Schwab, p.614).

It follows that substituting a tax on location rents for a tax which does reduce the use landowners make of their sites removes distortions in the economy. For example, replacing an ad valorem tax on buildings (as contained in most real property taxes), which discourages the supply of buildings, with one purely on land values, constitutes a change that "will result in a higher level of improvements to the land (e.g., a higher capital-land ratio). We will refer to this as the capital-intensity effect," write Oates and Schwab.

Turning to Lusht's paper, there is a reference to an "incentive effect of the use of the site value tax" due to the accompanying removal of a selective tax on a particular product (i.e., buildings): "The reduction in the tax on improvements shifts the supply curve by an amount equal to the tax reduction (Bourassa, 1987) decreasing price and increasing supply" (Lusht, p.519).

The presence of an incentive effect is intuitively appealing. It results from the removal of the economists' "excise effect" which produces a "deadweight loss" to society when a discriminatory tax is applied to a product (see Figure 1 - a). Enthusiasts for a shift of real property taxes to site value taxes stand by the slogan "untaxing buildings promotes building".

But some professors of economics have noted that buildings are a special case of the simple excise tax analysis.The supply of buildings is relatively inelastic - totally inelastic in the short term, by definition. This means that the supply curve in the short term is vertical and cannot shift. The construction industry has no power to raise prices for buildings by taking existing buildings out of use. An annual ad valorem tax imposed upon buildings cannot be shifted to customers immediately. So the prospective owners of buildings cannot offer as much for buildings with the tax as without it. The construction industry would have to cut back production for many years before an overall shortage forced up the market price of buildings significantly. Does this mean that it has to bear the whole of the tax, reduce output, and await the natural growth of demand?

Fortunately for builders there is a factor of production which is so specialised in supplying inputs to builders that it usually has very little exchange value without them. That factor is land. The supply of land is even more inelastic than the supply of buildings - even in the long run its stock cannot be reduced. So in the event of a tax related to building prices, all builders have to do to maintain the profits they require to keep them in business is bid less for building sites. They all face the same proportionate tax on their output; none of the firms is enabled by the tax to reduce its bid by a lesser proportion than its competitors without loosing customers. Landowners cannot take their sites to other markets where comparable prices are paid for uses not mediated by buildings or improvements of one sort or another. So they must accept the level of bids offered. The market prices of building sites fall by the capitalised amounts of the taxes the improvements are expected to incur over their lifetimes.

Professor Raymond Richman of the University of Pittsburgh caused a stir when he elucidated such matters, in a conference on fiscal policy and land values in 1970 sponsored by the Committee on Taxation, Resources and Economic Development (TRED). Noting that capital improvements to land are not totally elastic in supply, he concluded: "If this is true, the bulk of the burden of a tax on improvements, that is, the tax on capital, must be borne by the landowner and is capitalised." [3]

The main cause for concern was his deduction from this fact: capitalisation of the buildings tax into lower land values means that lower interest costs on the purchase price of land counterbalance the effect of the tax on builders' profits, so the rate of development is unaffected. Only in marginal areas where land prices are too low to absorb the full weight of the tax is development reduced.

At TRED's previous annual conference, Professor Mason Gaffney, now at the University of California (Riverside), presented a paper which included a lengthy explanation of why "all property taxes come out of land rent." [4] He went even further and suggested that taxes on buildings reduce annual land rent by more than the annual amount of the tax (p.191), other things being equal. But unlike Richman, Gaffney did not conclude that removal of the buildings tax has little effect on the rate of development; he went on to list at length (pp.192-206) the disincentive effects associated with taxing buildings. He argued that the removal of these mean additional upward pressures on land rents as the buildings taxes are removed.Econometric models [5] and empirical studies (see Lusht's below) have provided support for this view of the effect on land prices.

Gaffney's reasoning on the effects of the removal of buildings tax will be examined below. Meanwhile, Stephen Bourassa, postdoctoral fellow in the Urban Research Unit of the Australian National University, who was cited by Lusht in connection with the simplistic excise tax interpretation of the buildings tax, must be pursued on this point. In a sequel to the article cited, Bourassa provided a review of the theory of the impact of the land tax and mentioned two "incentive effects" resulting from the presumed accompanying buildings tax reduction:

The global effect is a reduction in the real rate of return to capital by the average property tax [footnote: reproducible capital tax, not land tax] rate.... [T]he excise effect, which depends on geographical variations in tax rates, with low tax communities having a lower cost of capital than high tax jurisdictions. Given the assumption of highly mobile capital, it is reasonable to expect that changes in tax rates will result in flows of capital from jurisdictions with high rates to those with low rates.[6]

Both of these incentive effects appear vulnerable to the Richman critique above. Capitalisation of the buildings tax into lower land values should protect real rates of return on reproducible capital. The average property tax rate across all jurisdictions reduces average land values. Local tax differences carve out local undulations in land values. Bourassa nevertheless claimed that his study of the data for Pittsburgh in the years 1978-1984 unearthed significant incentive effects. We shall return to this claim below.

In their "overview" of land taxation theory, Oates and Schwab move on to "a second effect that is less widely appreciated."

In two important papers, Brian Bentick (1979) and David Mills (1981) have shown that in an intertemporal context LVT clearly is non-neutral. While the taxation of land rents retains the property of neutrality [because it lowers the net return at each point in time proportionately], the taxation of land value [i.e., capital value, anticipating future rents] changes the comparative returns of land-use projects with differing time horizons. In particular, LVT favors projects that yield their returns sooner...it encourages the earlier development of unused parcels. We shall call this the timing effect.

LVT...taxes future returns in advance of their receipt. It makes it relatively more expensive to hold land idle in anticipation of future returns. As Mills (1981) shows, a development project, in order to be profitable, must promise a rate of return in excess of the market interest rate - but in the presence of LVT, it must, in order to pass muster, offer a prospective return that exceeds the sum of the rate of interest and the rate of taxation of land values.

It seems that Richman saw through this effect, too. As with capitalisation of the tax on improvements, capitalisation of the tax on land means that the developer pays less land price, hence lower interest charges on loans for land purchase (or foregone interest on own money), and more taxes, in equal amounts.The annual amount of the land value tax simply substitutes for the annual amount of the lower interest costs (see Appendix 1).

Gaffney confirmed this position: "It is widely believed that they [land taxes] speed up ripening [of land for development], but the belief has been wrongly rationalised. It rests mostly on assuming that land taxes are piled on top of interest costs of holding land. But land taxes are capitalised into lower values, and thereby supplant interest costs rather than supplement them." [7]

Land value is the net present value (NPV) of the optimal development project after all future taxes and developer's necessary profit and other costs are taken into account. It is the NPV of the project's expected flow of future revenue surpluses over costs.[8] It is the price that the developer of the optimal project must bid to secure land in a competitive market. The developer's target profit is a cost of production to be paid out at the completion of the project. Tax liabilities while holding land are also a cost. Expected revenue is reduced by anticipation of land value tax liabilities on the part of the developer's customers. Thus, anticipated land value taxes reduce the NPVs of projects and the price that developers are able to bid for land. They do not reduce the anticipated profitability of the projects themselves. In the presence of LVT, "to pass muster" a project does not have to promise a rate of return in excess of the market rate of interest to secure the developer's target profit. Instead, it is enabled to bid less for land by the fact that the NPVs of all competing projects are equally affected.

It makes no difference to the essence of the "timing effect" theory whether the tax reflects currently realisable rents or future rental values. Either, according to Mills' erroneous logic, would raise the required rate of return that the project would need to achieve, though the latter would do so more. Either would make it relatively more expensive to hold land wherever higher development possibilities are anticipated, though the latter would do so more. The continuum of "highest and best use" land rents (the basis of land rent taxation) rises above existing use land rents well in advance of redevelopment for a higher and better use (see Appendix 2).

The "timing effect" argument appears to have two unconnected strands tied together: a pull effect and a push effect.The supposed pull effect, or attraction, of earlier yielding developments depends on the rate of LVT supplementing rather than supplanting the investor's interest rate. The push effect involves a reduction in the capacity of landowners to hold out until investments have borne fruit. Bentick wrote that "a tax on market value causes taxes to be levied ahead in time of the returns on which the tax is based, creating a liquidity problem which cannot be solved by a perfect capital market." [9]

While it is true that taxing capital gains in land as they accrue causes a liquidity problem, it is not true that such a tax is "levied ahead in time of the returns on which such a tax is based." Capital gains form part of the total rate of return (the other part being annual income) as they accrue, not when they are eventually realised on sale (see Appendix 1; also Gaffney's 1970 paper, pp.183-187). Each year the capital gain accrual can be cashed in by sale of the land at market value. But even if that occurs in order to realise cash to pay a land value tax, that does not mean that development is hastened. For whoever owns the land will still find that the annual accrual of the NPV of the optimal development project from one year to the next (plus any existing use rent) still outweighs the annual return on land value that can be gained from the current optimal development; in other words, postponing development yields a higher return than does precipitating it (see Appendix 1). Developing prematurely lowers the market value of the site, given that the market anticipates development at maturity, and thus adds capital devaluation to lower income yield. Any existing land value tax regime cannot alter these relativities.

Given the apparently baseless claims for the "timing effect" of a steady or even increasing tax on land values, it is curious that Oates and Schwab should find the results of their preliminary empirical tests of the theory "encouraging: new building activity clearly picked up following the striking rise in land taxation in the city of Pittsburgh - and this increase in new building permits is not to be found in other cities in the region.... The results are in a sense, too good: it is difficult to believe that city property tax reform should by itself produce such dramatic results" (p.619).

Adding to that difficulty is the fact that their theory is not that the tax on land per se may have contributed to the increased activity, but that only that part of the tax on land values which distinguishes it from a tax on land rents contributed. They even stress that whatever incentive effects may flow from the removal of buildings taxes are ruled out in the case of Pittsburgh, for the situation in the period studied (1960-1989) was one divided by a sudden doubling around 1980 of the overall nominal rate of tax on land values (from about 1% to about 2%) accompanied by a slight increase in the rate of tax on structures (from about .07% to about .09%). "The tax rate on structures was not reduced - so what we are examining is a case of a dramatic increase in the rate of taxation of land in the city" (p.617).

These facts make even more curious Bourassa's claim, mentioned in the previous section, that his empirical "results for Pittsburgh indicated a significant incentive effect, but no liquidity effect" (p.107). In other words, he found that the non-existent reduction of the tax on structures is influential, while Oates and Schwab found encouraging preliminary support for the timing effect - a liquidity effect - of the increase in the tax on land values - and each denied the conclusion of the other.[10]

We have not finished with the "liquidity effect". In Lusht's "summary of theory" (and Bourassa "review of economic theory") the "liquidity effect" encompasses the "timing effect", but extends beyond it:

Assuming the supply of land to be inelastic, the supply cannot be changed in response to increased changes in the tax burden, and the tax cannot be shifted. Faced with an increased cost of holding developed sites, owners are encouraged to develop (p.519).

This is the timing effect, and therefore ignores tax capitalisation. If interest foregone on developed sites falls by the same amount as the tax increase, there is no increased holding cost. The statement also has a second aspect: it refers to a transition phase to higher LVT rates rather than a steady state. But again there is no reason why, in a perfect market with economically rational behaviour, the transition period should spur site development any more than other periods, despite the capital losses to landowners. It would not be to land developers' advantage to buy land in a falling market, for there would be no related increase in the profitability of building to override the losses in capital value.

Ironically, the extension beyond the "timing effect" depends upon tax capitalisation. In Lusht's words (p.519):

As the tax rate increases, the value of land decreases. In turn, lower land values encourage liquidity and development (Becker, 1969).

Bourassa's statement of the same point indeed furnishes a quote from Arthur P.Becker:

The benefit would be the equivalent of an automatic perpetual loan to the developer for purposes of land acquisition in the amount of the capitalized value of the land tax.

But to the extent that increased holding costs are illusory, so must decreased acquisition costs be illusory. If capital markets are perfect it makes no difference whether acquisition costs are borne in terms of annual land value tax bills, annual interest charges on loans for land purchase, or interest foregone on own money used for land purchase. Each is an opportunity cost of land ownership.

Only the title given to this group of effects - "liquidity" - provides a cluue as to why a tax on land values might be expected to stimulate development. There is an important difference between the three types of carrying cost of land which none of the authorities so far cited mention, other than Gaffney. Although the three costs are "economically equivalent", only two involve actual cash flow, and only one involves taxes. Gaffney has observed:

A common misuse of theory is the notion that people react to opportunity cost [alternatives foregone] as alacritously as to cash costs, because that is what "rational" people "should" do. This belief is a severe case of doctrine overriding observation.... A cash drain is what attracts the attention of any seller and moves him. [11]

A cash drain to the taxman is something to be avoided most of all! Gaffney observed that the motivated seller of surplus land is someone "subject to debt and/or property taxes." In his 1973 paper, he elaborated:

If money talks, the tax dollar outtalks the interest dollar, at least the dollar of foregone interest on equity, which speaks in a whisper.... According to the portion of tax theory that looks at marginal incentives and ignores the wealth and liquidity effects of taxes, land taxes are simply neutral, and in an important sense that is true.... In practice they accelerate renewal because they drain cash from holdouts waiting for high bids from builders.... The effect of a cash drain on a holdout far outweighs the effect of foregone interest on equity because the cash drain lowers his wealth and liquidity. The cash drain of land taxes also conveys information to many owners who are only vaguely aware that they are holding a resource of high salvage value to society. Land taxes build a fire under sleeping owners (p.133, italics added).

Equity ownership of surplus land is like ownership of Premium Bonds in the UK. They may not bring in any income, not even to stave off the ravishes of inflation - but there is always the chance of hitting the jackpot. The success of lotteries shows that it is natural to forego income regularly for just that chance. But there must be a limit to the cash haemorrhage - especially to the taxman - that owners of land surplus to requirements will put up with.

More generally, equity ownership is a licence to behave as an economically irrational person: to hold on to vacant land well beyond its ripe-for-development date, for example. Gambling spirit, inertia, ignorance and incompetence mean that much land is well past its optimal redevelopment date. Society is already suffering loss, and there are customers queuing up to put those sites to use. It only requires the introduction, or an increase, of taxation on land values tax to flush them onto the market. After the first flush, the flow should steady - but remain higher than before.

Land value or rent taxation of all landowners transfers the financial equity in land to the public sector. This is the "weath effect" that Gaffney referred to. So long as the stewardship of the financial equity transferred to the public sector does not produce decisions as economically irrational, from both private and social viewpoints, as those produced by erstwhile private, and to a large degree land-sated, stewards, then the economy will function more efficiently. Economically destructive forms of taxation will be reduced, and sub-optimal land uses will be replaced.

The cash drain effect of LVT is purely to stimulate development which would otherwise be impeded or delayed by economically irrational decisions or imperfect markets. This does not mean that LVT is "distortionary" and only has the secondary virtue of distorting in the opposite direction to other distortions, tending to iron them out, as Oates and Schwab suggest (pp.620-621). Where it is efficient to reap cash rents - that is to develop - the cash drain will make the reaping necessary to conserve the landowner's other assets. Where it is not efficient to reap cash rents, the cash drain will not encourage developers to buy land for development. Undeveloped land will remain in the hands of owners who have sufficient liquidity to take advantage of the competitive investment opportunities, partly secured by land value tax payments, offered by market value gains as sites ripen for development.[12]

In the absence of LVT those investment opportunities would be the same, only the source of funds would be different. Rents foregone on alternative uses of the sites would be the effective source, rather than cash from sources of income unrelated to sites.[13] For the reasons stated above, the cash investments would have a higher chance of being economically rational. They would also be a component of a more equitable society, which in turn would be a more economically rational society.

One aspect of LVT not mentioned in the theoretical overviews is its progressive nature. It is important that land taxes reduce land prices, because capital markets are not perfect and credit is rationed to those with greater credit-worthiness - the rich. So the rich have an unfair advantage in purchasing the fixed quantity of land. LVT reduces this advantage.

We have noted that in the absence of LVT rents foregone on alternative uses of ripening sites are effectively reinvested in additional capital gains from those sites. The wealth of those investors feeds itself. However, LVT requires that capital gains from vacant sites be "purchased" with cash, either from current income from labour or capital, or by liquidation of other assets.The sites themselves cease to provide the income invested.

The progressive effect of LVT is therefore compounded. Not only do the rich have less preferential access to land, but the land itself produces less private income for those who own it. A positive feedback circuit which polarises society in terms of income and wealth is weakened.

The mechanics that initiate the vicious circle were termed by Mason Gaffney "differential capitalisation":

Interest rates vary among people. They are regressive - the poor pay more. Land taxes, assuming true assessment, are not regressive. Substituting taxes for interest therefore undoes the effect of regressive interest rates. It hits the rich owner harder than the poor... increases the bidding power of the poor for land, causing them to encroach on lands held by the rich (1973, p.131).

As with the cash-drain effect, this has the effect of removing distortions in the market which delay development and cause sub-optimal intensity of land use. The effect is not uniform, but subtly dependent upon place. Enthusiasts for LVT claim that it "forces land into use" and thus physically intensifies land use (apart from where, as a consequence, it relieves demand pressure towards the external margins of land uses). But Gaffney pointed out that the effect

is not a simple plus or minus. The effect is equalizing as among classes. Land taxes let the poor, who live crowded on poor land, live less crowded and move to better land. They lower density for the poor by raising it for the rich, who own most of the land (1973, p.132).

As a rule, both the physical and the economic intensity of land uses are higher in poorer than in richer areas. But the overall effect of equalization alone is unpredictable. If the increase in LVT revenue is at the expense of revenue from taxes on buildings, however, the overall effect should be higher physical intensity, due to removal of delays in redeveloping (see Figure 6).

Despite capitalisation of taxes on building values into lower land values, there is still theoretical room for the operation of disincentive effects, hence incentive effects from switching to direct taxation of land values.

As with the effects of LVT, the effects of taxes on construction are more subtle than appear at first sight. They are dependent on time and place. Appendix 2 presents the necessary background for understanding them. It shows that the primary effect of buildings taxes is to delay the replacement of buildings, both as they wear out physically and as they become obsolete economically. The latter point is specified in the appendix, but the former also deserves elaboration.

A tax on building value delays replacement of an existing improvement in the same land use category as before. With the market value and optimal-use value of the site both rising at the same rate, as the economy grows, challenger and defender buildings will be battling it out in terms of the annual land rent they can provide. A fully depreciated existing building will cease to earn income sufficient to cover even the rental value of the land, and gradually the existing use land rent will decline. The optimal building for the site, meanwhile, will continue to offer a rising land rent to the site owner. A tax that bears proportionately on land rents does not change the competitive situation (unlike in the special case discussed in footnote 12). However, a tax on structures bears down on land rent maximally with the optimal building and not at all with the valueless building, with the result depicted in Figure 7. Renewal of the structure on the land is delayed for many years, production equal to the value of the squandered land rents is aborted, and revenue that the government might have raised by taxing land rents instead of buildings is lost.

Builders may not be concerned about the buildings tax if it lowers what they must bid for building sites - but landowners are. They are aware that de-intensification of land use targets, both physically and economically, may raise bids simply by reducing the taxes on construction values more than the project surpluses which enable the bids. Builders will therefore have to adjust the physical capital content of their projects downwards to pass muster. Referring to figure 6 in Appendix 2, it is apparent that the potential for reduction of taxes on buildings relative to revenues is greatest at the external margins of land use bands, where construction costs are highest in proportion to property values. The overall effect will therefore be to lessen the slope of land values, and reduce the delays of redevelopment dates. In aggregate, the physical intensity of land use will be reduced by a mixture of two alternative means - larger gaps in the land use pattern or lower intensity of use between the gaps.[14]

Thus the response to a tax on capital is to substitute land for capital, and knock out marginal uses.

Ironically, an implication of attempts to brand LVT as encouraging quick pay-off developments is the need to retain building taxes instead. Apart from delaying development, these encourage quick pay-off choices where development does occur. Acting like higher mortgage interest rates, less durable buildings are built because future returns do not enable present high interest rates to be paid.

Considerations of equity are intimately bound up with considerations of efficiency, so we have already explored the main equity issues as we have attempted to unravel the efficiency implications of land value taxation.

It has become apparent that part of the effect of LVT is to prise land from the hands of the relatively wealthy. We have called this the "redistribution effect". Another effect of LVT is to shift sites into the hands of those with sufficient liquidity to handle with the "cash drain effect". The latter influence appears at first sight to be at odds with the former, and accounts for the irony that LVT is commonly thought to be regressive. However, the fact that many landowners hold most of their wealth in the form of illiquid land does not make them any less wealthy; only less able to find the cash to pay taxes if they choose to consume the rental income from their wealth in the form of owner-occupied living space, or to invest it in the form of vacant sites.

Both difficulties for landowners are soluble. In the first case, the owner may move to a lower value rented accommodation and let his or her own property. The land value tax would then be paid out of cash income, as would be the accommodation rent. The individual would be no less wealthy;[15] he or she would simply be paying the same taxes as others who own equally valuable land. In the second case, footnote 12 suggests that an interim land use should usually solve the cash flow problem. If that is impossible, perhaps due to planning law, then premature development is the individual's short run solution. As this harms both society and the individual owner in the long run, agreement over interim solutions is probable.

These special cases should not be allowed to distract from the larger picture, which is that the proportion of land value in real property assets tends to be proportional to the wealth of the owner, and the proportion of real property assets per se in wealth tends to be proportional to the wealth of the owner. The relatively poor live in - may even own - houses the value of which is mainly in the buildings; the relatively rich occupy mainly land value. A tax on buildings is therefore regressive, and a tax on land value progressive - at its introduction and after.

This wider picture is confirmed by the evidence of choices that have actually been made wherever communities have been given the chance to choose directly between property taxes based on land values and those based on building values. Australia and New Zealand both inherited the British rating system which taxed the imputed rental value of occupied properties, and hence the existing use value of buildings and land. However, many parts of those countries enacted legislation from the 1880s onwards to allow ratepayers to change the basis of rating on a majority vote. In New Zealand 80% of local authorities had changed over to rating of land values by 1985; in Australia 65% had done so by 1976. In Melbourne, Australia, half of the 56 local government areas were rating site values by 1989, 70 years after the option had been introduced. Approximately 25 attempted changes have been defeated by petition and popular vote, but most of these have been attempts to change back .[16]

Robert Hargreaves of Massey University, New Zealand, noted in a paper for the 1991 Lincoln Institute conference that "the popularity of the land value system in New Zealand can be attributed to the fact that it tends to favour residential ratepayers...[hence] the majority of taxpayers." In the residential sector the ratio of land value to improvements value is lower than the average for all sectors, so the sector as a whole benefits from a revenue neutral shift to land value rating.[17] Moreover, most homeowners within the sector probably benefit more than the average homeowner.

Kenneth Lusht noted somewhat disparagingly about property tax changes in Melbourne that "the pre-vote debates have been highly emotional, with emphasis on the relative 'fairness' of the taxing systems rather than their developmental impacts" (p.527) Such an attitude is understandable given the purpose of his study, and his frustration that because existing study of land use patterns in Melbourne had been conducted mainly by land tax enthusiasts, it "consistently (and predictably) supports the notion that site value tax stimulates development, the tone tends to be exhortatorial and the quality of the methodology at best uneven" (p.521). However, it overlooks the possibility that "fairness" is itself a factor affecting development patterns, and suggests a lack of awareness of the larger issues that were responsible for the introduction of LVT in the first place - in Melbourne, in Australia as a whole, and indeed in many places around the world around the turn of the twentieth century.

The original settlement of Australia was mainly a mercantilist project; the establishment by Britain of self-supporting naval bases to safeguard its commercial interests in the Pacific. The dumping of convicts was an ancilliary activity. The interior of the continent was of no relevance to this enterprise, which displayed, in the words of Dr J.F.N. Murray, an Australian valuer of global reputation, "an almost complete disregard of the value and potentialities of land, and of the fundamental principles applicable to its allocation and use." [18] Theories of uniform price, and grants of free land, led to sporadic settlement, with settlers fanning out to select the areas most suited to immediate use as opposed to long term development.

This was the veritable apogee of the "timing effect": areas would be selected "for their immediate response to a minimum application of labour and capital." [19] And it was the result of a diametrically opposite policy to allocating land on the basis of cash payments related to its long term value. The only notional qualification for the receipt of land was the possession of sufficient capital to use it. All grants in the early nineteenth century carried a quit rent determined by area, not by quality. Later, land came to be sold at a uniform price of one pound per acre, and then with uniform minimum prices at auction."[T]he uniform price led those with speculative tendencies, or who possessed special information, to pit their knowledge or estimates of the future trends of the market prices for land, against those bidders who were concerned only with the probable future incomes to be derived from farming...[aided] by the exceptional facilities for obtaining credit which were available." [20]

Edward Gibbon Wakefield's infamous land settlement scheme of the second quarter of the nineteenth century completely overlooked the possibility that the land of Australia might not be of uniform character. He relied on setting a price per acre for land release - uniform within each colony - which was sufficient to prevent labourers from obtaining land too soon, and thus reducing the supply of labour at low wage rates, but not too high to prevent them from purchasing within a reasonable time, thus filling the public coffers, funding further immigration, and providing a demand for labour.

The combination of earlier land grant and later land famine policies produced the "squatter" problem of the second half of the nineteenth century. The original "squatters" overran Crown lands outside the areas granted or offered for sale, were then converted into nominal fee-paying licencees or tenants by a pragmatic government, and finally offered their holdings at one pound an acre - the outer areas of which they did not buy but became the de facto owners of anyway.[21] Along with purchasers and grantees they were occupying for pastoral purposes huge areas of land which included the "eyes" of the country (its waterholes) and much of the most fertile land. Latecomers to the colonies were forced to accept an urbanised future, "shut out of their rightful patrimony" (Peter Burroughs [22]).

This was the setting for the birth of land value taxation in Australia: a bitter realization of injustice on the part of the urban masses. As early as 1854 the Melbourne Argus began a campaign demanding that the government "unlock the land" and impose a land tax. Public policy was also set on fostering an agricultural industry, which required closer settlement.

After 1860 the governments of the colonies were faced with another problem that made land taxation look tempting: "revenue from the sale and rental of land had been diminishing as the areas of unalienated useful land decreased.... In 1877 the Victorian Parliament enacted a provision for the imposition of a tax on pastoral land to vary with the capacity of the land to carry sheep."[23] This was the forerunner of the State land taxes. By 1915 every State in Australia was levying an annual tax on the "unimproved value" of land.

LVT was thus originally introduced throughout Australia because it was popularly thought to be effective in promoting more efficient and equitable land use, and raising revenue.

In 1910 the Land Tax Assessment Act became law, introducing the first direct tax to be levied by the federal government. "The objects of the tax were stated to be the breaking up of big estates and provision of funds for defence purposes."[24] The former objective was challenged unsuccessfully in the High Court in 1911. The Federal land tax was abolished in 1952 as the result of a move to secure sole authority to tax incomes to the Federal government. In exchange the States claimed most of the revenues from unimproved values that had been going to the Commonwealth through their own land taxes. Both were levied at steeply progressive rates, and only on properties above a high minimum unimproved value (5,000 pounds in the case of the Federal tax, 1910-1952). That was deemed a necessary part of the attack on big estates and the promotion of smallholdings and home ownership. In reality, it undermined the aims of the taxes.

K.C.Taeuber, Commissioner of Land Tax and Chief Valuer, South Australia, told the Lincoln Institute's International Seminar of 1966 that closer settlement of primary production land in Australia has been only slightly influenced by the incidence of the land taxes. He gave four reasons:

2. Avoidance of the steeply progressive rates by subdivision of nominal ownership within families, or between corporations owned by the original landowners.

4. Insignificance of land tax rates beside 5-10% per annum untaxed rates of capital gain in land values.[25]

Steeply progessive tax rates mean that rates are very low at the bottom end. Add to this a substantial minimum taxable value allowance and numerous other exemptions and it is clear that the price of land cannot be reduced for the majority of sites. The Collins Commission's Review of the State Tax System (1988) estimated the revenue worth of New South Wales' exemptions from land tax (12 large classes, including the owner's principal home and the family farm) at A$2bn per annum - equivalent to about half of the State's revenue, and almost six times the land tax collected.

The State land taxes are concentrated in their effect on the areas of highest land value. Taeuber went on to note that "there appears to be a ground for qualifying the statement that property taxation is insignificant relative to income taxation [and income tax deductibility of capital expenditures] in its application to central city properties."

An empirical study [26] by R.W.Archer for the Washington-based Urban Land Institute confirms this conjecture. Archer studied the effect of site value taxation (SVR) on redevelopment in the central business district of Sydney in the 1960s. He chose the Sydney CBD "because it is one of the few areas in Australia where site value taxes are high enough to be an influential proportion of property values, where the site value tax system as a whole is administered efficiently and effectively, and where there has also been a large amount of redevelopment activity.... [T]he municipal rate and the state land tax amount to about five per cent per annum of site valuations [which] are full market valuations at the time of valuation and have been made at six yearly intervals or less" (p.38).

In the redevelopment situation the site value tax system acts to increase the supply of sites for redevelopment... to accelerate the status of marginal properties to the status of economic redevelopment sites.... It has the greatest impact on those properties with the greatest redevelopment potential.... [and] reinforces concentration of redevelopment by stimulating the greater availability of sites in this area (p.38).

He emphasised that "Site value taxation can only encourage the redevelopment of a site when its redevelopment is economically feasible; that is when the site value and improved value of a property are similar" (p.36). Does that still leave open the possibility of development before the economically optimal moment? Not according to his study: "Developers would seek to obtain the maximum income from their redevelopment project as a safeguard against the expected long-term rise in site values, site valuations and tax charges" (p.41).

In the Sydney CBD it would seem that there was no room for the existence of the vacant sites that are a prerequisite of premature development. Properties became ripe for development while the buildings were still valuable. Indeed, Archer diagnosed an "equity problem" arising for owners of properties with a long transition period to the status of redevelopment sites. "Because the property's current use value is still greater than its redevelopment potential-use value, it is not possible for the property owner to redevelop or to sell the property for redevelopment. This dilemma can cause financial hardship.... Thus, the ratio of site to improved value for one store rose from 52% percent in 1956 to 70 percent in 1962 and 86 percent in 1968, when it was paying 38 percent of its [gross property] income in site value taxes" (p.36, 41).

If LVT's "timing effect" exists it would trigger redevelopment in such a situation. But it cannot.

Probably the most important effect of a site value tax system is the pressure on owners to sell their property for redevelopment if they cannot or will not redevelop it themselves. This is accomplished by the cash outflow required by the site value system. This mechanism encourages the availability and use of sites ripe for redevelopment and encourages early use of the most suitable sites since they attract the largest tax charges. This facet of the system also encourages the sale of properties for redeveloment and thereby:

1. promotes a more active and informed market in redevelopment properties [which discourages speculation],

3. allows developers to expand their market role in the redevelopment building market at the expense of single project redevelopment by property owners (p.38).

This would appear to dispose of the claim that LVT leads to premature and piecemeal development.

As for its credentials as a weapon against land speculation: The study attempted to find "person-firms buying redevelopment sites to hold out of the market for later resale at higher prices. This speculation would act to accelerate site value increases by reducing the effective supply of sites. The author found no evidence of such speculation activity although there were persons and firms operating as dealers in redevelopment sites. These dealers were anxious to resell as quickly as possible" (p.40).

The centre of Sydney is the showpiece of LVT at work stimulating efficient land use in Australia. It is given little chance to work elsewhere due to the slightness and inconsistency of its application. There are, indeed, many measures specifically to narcotize the stimulating effect of LVT. In Queensland, for example, where all municipal rates are on unimproved value, existing use value rather than highest and best use value is the valuation standard. The Committee of Inquiry which considered the rating system in Brisbane in 1989 noted that, apart from making "a gift of... a substantial capital gain when they [owners] subsequently sold their properties for market prices which reflected the value of their more intensive use...[, t]he concession moreover negated the incentive to put land to its highest and best use, whereas this was an intended effect of rating on unimproved values" [27] They therefore made several recommendations to rectify this situation.

Such concessional valuation practice is widespread in Australia. It is regarded as more reasonable to protect people from being compelled to move than to have a discipline which fosters efficient and equitable land use. Since the 1920s, revenue has been almost the sole purpose of land taxation.[28]

Archer considered a possible solution to the "equity problem" which he identified in the Sydney CBD. The office redevelopment potential of large department stores caused the land taxes to drive an increasing wedge into their existing use property incomes before it was economic for the owners to actually redevelop. The indignant companies complained of the "intolerable and unjust burden of land tax on business premises", presumably without mentioning the capital gains in site value which were more than offsetting the tax inroads into current property - not retailing - incomes.

Archer toyed with the idea of recommending current use value assessments until properties become ripe for development, at which point assessors would switch to potential use valuations. But he decided that in practice that would be complicated and confusing, and concluded: "There would seem to be no way of curing this inequity without undermining the usefulness of site value taxation in redevelopment" (p.36). Yet that is the route that has often been chosen - without the reversion to potential use valuation before sites are redeveloped.

It is hardly surprising that the efficiency benefits of LVT are not obvious in Australia. The Australian dream is ownership of a detached home on a quarter acre block. But over 60% of Australians live in five coastal cities which are experiencing a population explosion. 80% of houses are single detached homes designed for families, but only 30% of households are families. The federal housing ministry has become concerned at the prospect of another 700,000 quarter acre blocks being added to the fringes of cities over the next decade. It calculates that providing public infrastructure for those low density settlements would cost about A$6bn.Journey to work times would become even longer; fuel consumption, already at least double that of the average European city, even greater.

So the federal government launched a Building Better Cities programme in August 1991, proposing to invest A$0.8bn over 5 years in medium density housing (6 or more per acre) and conversion of redundant industrial or institutional land to housing.[29] Meanwhile, for the sake of "equity" (financial?) for landowners, potential for subdivision of dwelling sites continues to be explicitly ignored in many property tax valuations, and rates of property tax everywhere remain immobile upwards.

Taeuber stated that the aggregate of property taxes raised by State and local governments and ad hoc authorities for water and sewage works was "insignificant in comparison to the other economic forces motivating land development in Australia.... In the metropolitan areas of the cities of Melbourne and Adelaide, the boundaries of local authorities levying taxes on each base [unimproved or improved values] adjoin. It is impossible to distinguish in any way, either on, or within, the boundaries any difference in the standard or the degree of development between the areas." [30]

Advocates of LVT claimed to distiguish differently, and it was to test this dispute rigorously that Kenneth Lusht conducted his statistical enquiry. Melbourne provided the ideal laboratory. In the 1980s 27 local government areas (LGAs) rated entirely on site value (SV, defined as the market value of unimproved land, but including the value of invisible improvements, such as drainage, deemed to have merged with the land), 28 rated entirely on net annual value (NAV, defined as 5% of the market value of residential properties and "one year's rent" for other types of properties), and one rated on both bases, with a higher rate on SV than on NAV. Lusht presented a straight comparison of data on these LGAs for 1986, excluding the LGA with the combined rate and the NAV LGA which included the central business district.

Lusht also analysed data on the stocks of residential improvements and of all improvements in 53 of the LGAs for the year 1984. 29 LGAs were counted as SV areas, including the combined rate LGA, which was SV between 1922 and 1982, and another that was SV from 1946 to 1983. Three NAV LGAs were omitted - the one with the CBD and two that had switched more than once.

Finally, Lusht compared the flows of new housing, new industrial development and alterations and additions to residential properties over five-year periods in the 1980s in the 28 LGAs (15 SV, 13 NAV) that had significant areas of vacant and developable land.

1. Overall, the SV areas appeared to have less capital value (improvement value) per acre, that is they had a lower physical intensity of land use. Analysis of the data suggested that there was a 90% probability that SV rating in an LGA reduced the total value of improvements per acre by between 1% and 9% (p.551).

2. Straight comparison of the average numbers of industrial and retail establishments and occupied homes showed SV areas to have 16%, 15% and 12% less units per urban acre, respectively (p.534) - another indication of lower physical intensity of land use.

3. Straight comparison of the average ratio of assessed total property value (improved value) to assessed site value showed SV areas to have a lower economic intensity of land use. The ratios were 2.30 (SV) and 3.24 (NAV), implying economic intensities of 56.5% and 69%. Thus 43.5% of real property values were site values in the SV areas, 31% in NAV areas (p.534).

4. Statistical analysis suggested that there was no significant overall relationship (that is, not explainable by chance alone) between SV areas and the value of residential improvements per acre (pp.545-546). However, there was a 95% probability that the LGAs that adopted SV rating earliest (in the 1920s) had a lower value of residential improvements per acre as a consequence. Limiting the acreage in the analysis to residentially zoned land supported the pattern of an inverse relationship between development and length of SV rating experience - the latest adopters (in the 1960s) even having higher residential improvements as a consequence - though the zoned land exercise had low explanatory power.

5. Statistical analysis of 28 LGAs with unconstrained land supplies between 1983 and 1987 indicated that there was an 86% probability that higher numbers of permits were issued for single family detached houses in SV areas. The association of SV areas with higher values of permits was weaker (p.559).

6. Analysis of the variation in residential lot prices between the 28 suburban LGAs indicated that a combination of seven variables, including two tax variables, accounted for about half the variation. There was a 94% probability that SV rating gave a residential land price premium of between 5% and 21% over equivalent land in NAV areas; and a 90% probability that between 4% and 10% off the property tax rate (a reduction, for example, from 8% to 7.5%) gave a 1% land price premium (p.563-564).

7. Statistical analysis of the number of permits issued for new industrial facilities and extensions in the same 28 LGAs between 1981 and 1985 produced the strongest results of the study. Seven variables were found which "explained" 83% of the variation in the number of permits between LGAs, with each "significant at the 1% level" (i.e., having only a 1% chance of randomly falling into the same pattern).

One of these was SV rating. "Thus, the [average] number of new firms and expansions of existing firms in site taxing communities was about double the [average] number (24.15 versus 12.55) in capital value taxing communities" (p.580). Another was the effective property tax rate of either system: "For example, a community taxing at the mean rate of .008 of value would attract, on average, about 18 fewer firms and additions to existing firms than would a community taxing at .006 of value" (p.580). Lusht commented that "The findings with respect to the site value tax are...the first empirical confirmation of extant theory" (p.584).

8. Analysis as in 7. above, but for the value of permits issued between 1983 and 1987, again produced "satisfactory" statistical results. The tax variables had the same associations as in 7. above "though at slightly lower significance levels". One combination of variables explaining 80% of the variation in the value of permits gave SV areas an 85% chance of attracting between A$5m and A$19m more industrial development than NAV areas, the average development attracted to all areas being A$21m. And it gave a fall in the effective tax rate from .008 to .007 a 90% chance of raising the average value of permits per LGA by between A$2.3m and A$5.9m (pp.582-584).

9. Analysis of the variation in average industrial land prices between the 28 LGAs indicated no significant association with either SV rating or property tax rates (p.587).

10. No significant relationship was found between expenditures for residential alterations and additions in 1986 and the effective property tax rate on improvements (p.597, 561).

According to Lusht these results were in line with the "new view" based on LVT's hypothesised "timing effect". He interpreted them thus: the SV rating areas exhibited faster development after they switched away from NAV rating, but lower physical intensities of land use overall, suggesting that they had developed, or were developing, too quickly to the prejudice of long run intensity. "In Melbourne, there is no evidence that two to five decades of site value taxation has had a positive effect on long-run development intensity" (p.553).

1. One of Lusht's four "indications" of the new view was that "descriptive statistics show that communities which tax site value only have on average a larger number of retail and industrial establishments, and a larger number of residential units" whilst having less capital value per unit of land, the implication being that the average building is less valuable in SV areas (p.603). Closer attention to the table of descriptive statistics shows that if the "existing acres in agriculture" are stripped out, the SV areas in fact have a lower number of units in each of the three urban land uses per urban acre than the NAV areas. The facts are as in 2. above. The capital value per unit of urban land was not given.

Lusht mentioned that the 27 NAV LGAs in the comparison included four that switched in the 1980s from SV. He supposed that, as they had more than average industrial and retail units, they smoothed the true differences (p.535). But that deduction cannot be made. The four LGAs were all relatively urban "cities" as opposed to relatively rural "shires"; they took with them higher than average proportions of urban acres as well as urban units. Smoothing of urban units per urban acre does not follow.

It would, indeed, be astonishing if the SV areas, with 69% more urban area on average, did not have more buildings. To compare such a descriptive fact with the synthetic "finding of less capital value per unit of land" is meaningless. The latter "fact" was the resultant of attempts to strip out the five or six most likely sources of variation other than SV rating. Distance from the centre of Melbourne was one, location south and east of the centre another.

2. Another of Lusht's "indications" was "the consistently stronger association" between SV areas and the number rather than the value of permits issued for new development (see 5, 7 and 8 above). For this to be an "indication" it must be assumed that lower income households or smaller businesses are associated with less intensive development. That is not the case.

Two explanations were advanced for the relative (not absolute) lack of stimulation by SV rating of more capital intensive projects: the burden of property taxes may be less important for higher income households (p.558); taxes may affect firms' location decisions more than their decisions on size of investment once located (p.583). Neither of these implied that more capital intensive projects are pre-empted by smaller projects encouraged by the tax on land - indeed the data was specifically garnered from LGAs which had plentiful supplies of vacant and developable land. Rather they were attempts to explain an apparent shortcoming in the "incentive effect" of removal of taxes on buildings.

Somewhat inconsistently, the "incentive effect" was called upon to explain another phenomenon - why SV rating stimulates industrial building most, residential building least and residential alterations not at all: "the most capital intensive uses tend to avoid taxes on capital" (p.604). That particular advantage of LVT over taxes on buildings appears to be indisputable. The point is: it is illogical to find evidence for the "timing effect" of taxes on land in phenomena which are attributed to dilution of the effect of the removal of taxes on buildings.

3. Lusht's third "indication" was the finding (4. above) that the longest established SV areas had the least intensive residential development. These LGAs, however, have other characteristics which would be expected to hold down physical and economic land use intensities. They tend to be the most prestigious residential areas - to the east and south of the centre, either not too distant or lining the east coast of Port Phillip Bay (Map 1, p.529). The SV LGAs as a whole have 11% higher median household incomes (p.534), and these would be the richest of them.Old, rich residential neighbourhoods tend to resist subdivision. This fact is overridden in the statistical analysis by the fact that areas east and south of the centre also attract more additions to the residential area due to their relative desirability and superior transport infrastructure (p.543).

We have argued that the "cash drain effect" of switching to LVT on sleeping owners would produce a temporary flush of development, and this is compatible with Lusht's conclusion of "faster but not necessarily more intensive development per unit of land."Theoretically, the effect on the "economic intensity" of land use due to switching from SV to NAV categories is not clear-cut, but physical intensity should be increased (away from the urban fringes) as illustrated in Figure 6. Social factors would seem to have dominated the pattern of intensity in Melbourne, however.

The "redistribution effect" may have reinforced the social factors. We have argued that it lowers the physical intensity of development in relatively poor areas in the long run. Relatively poor areas are more likely to vote for a switch to SV rating because they have relatively high building value to land value ratios. Therefore, the areas which switch to SV rating see initial flushes of redevelopment as the poor are enabled to acquire more land, but redevelopment in the direction of lower physical intensity - detached bungalows instead of apartment blocks.

But the main factor determining the decision to switch from NAV to SV rating must be the presence of sufficient non-residential property in an LGA to shift a significant proportion of the tax burden off the predominant voter - the residential taxpayer. "The historic `in-to-out' movement of the tax is consistent with this notion" (p.605). "In" in this statement does not refer to the centre of Melbourne, however. The inner ring of 10 LGAs around the central LGA has remained almost exclusively NAV. This fact must tend to bias descriptive statistics on intensity of urban land use in favour of NAV areas.

Furthermore, inner areas were developed before town planning constraints, the first of which were uniform building regulations, post-WWI. More controls followed in the 1940s and 1960s. There is less site coverage permitted in the younger, outer municipalities - which tend to be the site value areas. Differential incidence of planning regulations may, indeed, be the crux of the matter.

Unusually low ratios of improvements to land value in older, more prestige and hence more land-intensive areas may help explain why the residential sectors in four of the most central SV LGAs failed to resist switches to NAV rating, or dual rating, in the 1980s.

4. SV rating has demonstrably had a significant intensifying effect in the industrial sector, and this provides Lusht with his main "indication" of faster development. But the problem with this indication is that it does not touch the essence of the "new view". It provides no evidence that the areas currently developing fastest will prove to be the least intensively developed half a century hence. It only provides evidence that firms prefer SV areas, or that they find land easiest to acquire in those areas.

The latter conclusion has one proviso. Lusht noted that firms are actively discriminated against in NAV areas. Their tax base is "one year's rent" of industrial property, which was equivalent to 7-10% of market value in the 1980s (p.526), whereas homeowners' tax base is 5% of market value, which was probably less than one year's rent in the 1980s. In SV areas all sectors have the same tax base. So firms would be expected to prefer SV LGAs on this ground alone. However, the discrimination is less than the bald figures suggest. 5% of market value may not have been much less than "one year's rent" in the residential sector. Certainly, houses would have had considerably lower rent yields than factories, their imputed rents not being subject to income tax or profits tax, amongst other factors.

It cannot be argued that firms are attracted to SV areas simply because more firms are already there. They had 16% less industrial establishments in 1986. The switch to SV rating appears to take place in relatively residential LGAs, with industrial catch-up a consequence. But the catch-up process must be concentrated in limited areas. One of the unambiguous findings of the study was that there was "a strong negative association between the number of manufacturing facilities and the residential stock [p.556]... residential and industrial development are clearly separated" (p.579. Also pp.557, 583).

We turn now to the other broad finding of the study, regarding land prices (6. and 9. above). Lusht's explanation of higher residential land prices in SV LGAs depends upon each being a submarket in "a `balkanised' urban area, parts of which tax site value and parts of which tax capital value" (p.565), so that "the incentive effect of reducing the tax on improvements will tend to dominate the liquidity effect of increasing the tax on land, resulting in greater demand for land in that submarket and higher, not lower land values" (p.519, emphasis added). Choice between proximate LGAs makes the supply of capital particularly elastic in individual LGAs. Reduce the tax on capital in some and capital flows into them at the expense of others; the rearranged pattern of derived demand for land produces "a quilt of land values" (p.604).

The evidence (5. above) does suggest a higher supply of residential capital in the relevant SV areas during the study period, hence reason to suppose that the derived demand for land was significantly higher. But it points even more (7. and 8. above) to a substantially higher flow of industrial capital into the same SV sample during the same period. Yet there was no significant difference in industrial site prices compared with NAV areas.

To account for this differential Lusht supposed that "prices are more endogenous for industrial than for residential land" (p.587). By that he meant that the industrial land market in the 15 SV LGAs was of sufficient relative size and coherence to shield it from external influence and allow the negative impact of the tax on land to fully offset the positive impact of untaxing buildings. By contrast, "the price of housing is exogenous" (p.560), that is, the attraction of tax-free buildings is so concentrated in a limited area that demand for housing is stimulated in that area despite the higher tax on land. This distinction is not very convincing, depending as it does on SV LGAs being a relatively larger proportion of the industrial land market than of the housing land market, which does not appear to be the case.

There is a way in which removal of the tax on improvements may dominate the increase in the tax on land and raise land prices. The tax rate on improvements may fall by more than the tax rate on land rises, so that the overall property tax rate becomes less than it was before. In fact, that is what is likely to happen. Mason Gaffney argued the case in his 1970 paper, and predicted that shifts from NAV to SV rating would cause land prices to rise in the SV areas for that reason (p.191).

All site values are currently depressed by taxes on the full improvements that redevelopment of each site would attract. But each property is only currently paying improvements taxes on the depreciated value of the existing building. Many old buildings are paying no improvements tax at all. Abolition of the tax on buildings would allow all site values to rebound by the full amount by which they had been depressed. To collect the same revenue as the buildings tax had collected, therefore, the site value tax rate - which would fall on all site values - would not have to rise by the same number of points as the improvements tax rate had fallen. The lower tax rate would be spread more equally over a greater number of properties.

In the Melbourne study, lower effective property tax rates were treated as an entirely separate factor from the choice of tax base. It may be that some of the higher land price attributed to lower tax rates should be seen as flowing from the choice of tax base.

But why should residential land exhibit higher prices in SV areas, but not industrial land? The price effect depends on the age structure and rate of obsolesence of the buildings in each LGA. Residential and industrial land "are clearly separated", which must mean (as the data are LGA aggregates) that some LGAs are distinctly residential and others distinctly industrial. One would suppose that industrial buildings are on average owned by more profit-orientated owners than residential buildings, so they tend to be renewed more promptly as economics dictates. These assumptions lead to the conclusion that the effective property tax rate advantage of SV areas should be less in more industrial LGAs than more residential LGAs.

Before leaving this review of empirical evidence, we may return to the confusing picture presented above of the outcome of higher LVT in another continent - in Pittsburgh, USA. Pittsburgh moved rapidly to a higher overall rate of property taxation by raising the rate on land value alone. Relatively cheap property and faster development appears to have been the outcome. The former may be explained by the higher overall rate of taxation. Both elements of the graded tax - on structures and on land values - bear down on land prices. Increase either, and land values are depressed. But one bears down more than the other - the tax on construction, for the reason mentioned above. Hence a shift from one to the other can raise land prices. In Pittsburgh, however, there was no shift, just an increase in the tax on land value. This may be predicted to trigger the "cash drain effect", and the results are consistent with that. The "redistribution effect" would have unpredictable results depending on the social composition of the city. The increased tax revenue may also have funded public spending which promoted development.

With property tax rates in Australia generally around the 1-2.5% level in metropolitan areas, innumerable exemptions, and deductibility of those taxes against income tax until the 1970s (when they ceased to be deductible for non-income producing properties), it is not surprising that they have not been regarded as important revenue-raisers, except at the municipal level.

In the late 1980s, State land taxes and municipal rates were raising about 4.5% of total tax revenue, just less than in the late 1940s - when the Federal land tax was also being levied - and considerably less than in the early 1970s (see Figure 1: "site taxes/tax"). Adding in the rates levied by separate sewage and water authorities raises the revenue from land values by about 40%. The States were relying for about 5% of their tax revenue from land taxes; the municipalities over 90%.

The total tax base for LVT in Australia, in annual income terms, may be approximated by discounting the aggregate land value of the country by a representative interest rate for real property to find the rent remaining in private hands. Bryan Kavanagh, a Melbourne valuer, has carried on the earlier work by Allan Hutchinson of updating estimates of the aggregate land value of the country. Official assessment information is provided by the Commonwealth Grants Commission, adjusted according to its known shortcomings.

FIGURE 1: Effective tax rates [To be printed out, unless Microsoft Works spreadsheet can be used[MAO2]]

Relationships between land rents, taxes (and "captures", which include utility rates), and national income over time, are shown in Figure 1, which is derived from Table 1, Appendix 2 of the author's UK/Australia chapter in Volume 1 of The Sisyphus Syndrome. The private annual rent of land in 1985/86 amounted to almost six times the amount of property tax revenues and water/sewage rates each year ("uncaptured rent/NY" divided by "captured rent/NY" in the figure). Public receipts (captures) from site rents thus tapped 17% of the annual tax base ("% rent captured" in the figure) whereas national taxation tapped about 38% of national income. Together the private rents and the publicly received rents (apart from utility rates - not part of taxation) were equivalent to about 37% of total tax revenue. Including utility rates they were 14.5% of national income ("all rent/NY" in the figure). Mineral rents are not part of this calculation, though they are part of the national income. Account is taken of public receipts of mineral rents at the end of the second appendix of the author's chapter on Australia and the UK in volume 1 of The Sisyphus Syndrome.

As property taxes are (or were) set-off against taxable incomes, they reduce the income tax paid by owners by their marginal income tax rates. Purchasers are therefore enabled to pay more for land. However, income tax rates may be higher to offset the loss of revenue. A small subsidy from non-owners to property owners, especially with higher incomes, results, inflating land prices somewhat. Property market yields, however, fall by the same token, so the rents that are calculated from the land prices are not exaggerated.

On the other hand, the 25% or so of municipal rates revenue that derives from buildings understates the increase in revenue that would result from switching to site taxation. The land valuation is also understated by some use of existing use values, and generous apportionment of the value to the improvements component of the property.

In a land of over 70% owner-occupancy there is very little income taxation of land rents. For a source of public revenue that promotes efficiency and equity like no other, the tax potential of sites in Australia is grossly undertapped.

Simple tax capitalisation (assuming no income growth) is described by the following equation:

This equation describes the net present value of any perpetual fixed annual income subject to a special ad valorem tax. Like the stream of income from fixed interest bonds, the stream of annual income provided by a site is converted into the current market value (net present value) which provides a rate of return on that value competitive with other interest rates after paying the special tax on land.

However, the simple tax capitalisation equation does not allow for changing land rents over time, or changing tax rates. The expectation of changing land rents adds a speculative element to the market price of land. Thus, the equation becomes:

This signifies that the annual cost of land ownership (market value multiplied by the sum of the interest rate and the land value tax rate) equals the annual benefit (annual rent plus the annual accrual in market value). A land value tax does not add to the overall annual cost of landownership, nor does it subtract from the annual benefit. This equation assumes, however, that future changes are correctly anticipated. This is most likely to be the case in an economy where inflation is under control and the tax regime is stable, or its direction of change is agreed well in advance.Unanticipated changes are a problem for all asset owners and tax payers; they are not a special problem for landowners and land value tax payers.

Differential tax capitalisation: Given that the rich are able to borrow at lower interest rates than the poor, and that their relative security also allows them to take a longer term view of their investments and thus invest at lower interest rates, it follows that they are able to pay more for the same net income yielding sites than the poor. i in the equations above is smaller for them, so V is larger.

It also follows that any LVT rate, t, lowers the bids (V) of the rich more than the poor. In the first equation above, with t = 0, and a = 100, i = 0.04 supports V = 2,500, but i = 0.05 supports V = 2,000. With t = 0.01, the Vs become 2,000 and 1,667, respectively. The first is cut by 20%, the second by 16.7%. Bids are cut in half when t = i, and that occurs for the rich at a lower tax rate than it does for the poor.

Appreciating land strengthens the effect of differential capitalisation. In the second equation above, a given g has a greater effect on V the lower is i, and a smaller effect the greater is t. In our examples, with no LVT, g = 0.01 raises V by 33% from 2,500 to 3,334 when i = 0.04, and by 25% from 2,000 to 2,500 when i = 0.05. With t = 0.01, g = 0.01 raises the Vs by 25% from 2,000 to 2,500, and by 20% from 1,667 to 2,000, respectively.

Thus appreciating land gravitates into the hands of those with the wealth to be able to borrow at low interest rates - or not borrow at all (forego interest on equity at low rates).

In order to predict the effect of different taxes on landowners' decisions to initiate developments it is necessary to understand the factors which influence the development decision.

Imagine a point on the Earth's surface, A, in the year 1800, say. It is part of a wild, unsettled region. As time passes, farmers begin to settle around it. A farmhouse is built. The area has become developed for agricultural use. But no settler is willing to pay more for the property enclosed by the first settler than the construction cost of the farmhouse and other fixed improvements to the land. The land is right at the margin of production and yields no more income to the farmer than is necessary to compensate for the labour and fixed and circulating capital expended. There is no surplus, or rent, which may be capitalised into site value on sale of the property.

As time passes, the demand for farm products increases. Public works, such as roads, are extended in the direction of point A, making it more accessible to markets, further increasing demand. The costs of transport fall and enable inputs to be acquired more cheaply. Thus as the farm increases its output (and takes advantage of economies of scale), revenue grows faster than costs. More expensive structures need to be erected, but the resulting revenues yield surpluses over costs. The farm enterprise enjoys growing annual surpluses over all costs of production (including the farmer's entrepreneurial and labour inputs) which are made possible by the location of the land, and which may be sold as the site value of the farm.

In Figure 2, the property will now be located between 1 and 2, if we imagine that discrete land use bands are mobile, encircle the economic centres of regions, and expand outwards from the centre over time. Between 1800 and 1900, say, point A shifts from location 1 to 2 as land uses migrate out from the centre. As it does so the net revenue attributable to the property alone grows faster than the capital cost of structures, and the sale value of the site apart from the structures grows accordingly, being the difference between the two. As the site value grows, so the security of the property improves and the credit rating of the owner improves. The owner is able to borrow at lower interest rates, further lowering costs of production and raising site rent. Less risk also means that purchasers of the land are prepared to capitalise the rents at lower interest rates, thus levering site value up further.

By 1900 the land is "ripening" for redevelopment at the urban fringe. This means that the landowner is presented with the possibility of enjoying a higher site value by changing over to a "higher" use of the land than would be possible by continuing to intensify the site's agricultural use. Note that as this change-over approaches - indeed throughout the whole period of agricultural use - the "economic" as opposed to the "physical" intensity of land use has been falling. Economic intensity compares property values with construction values; physical intensity compares construction values (or simply units) with area units. Thus, a property value of $1 million and construction values of $750,000 means that the economic intensity of land use is 75%, and site value is 25% of the use value of the property. The higher is the proportion of site value in the value of the property, the lower is the economic intensity of land use. Falling economic intensity usually accompanies rising physical intensity (more building units or values per unit of land) as land ripens towards its date for change-over to a higher use (see Gaffney's 1973 paper, pp.148-149).

The change-over to a higher use involves a sudden reduction of economic intensity and a sudden increase in physical intensity. Meanwhile, the market has been anticipating the higher use, and the higher land rent and site value that accompanies it. So the market value of the farm's land has been bid up in line with the discounted value of the increased site value at the date of development. This means that market value of the land has begun to exceed the value of the land in its present use, or use value.

This divergence began as soon as the shape of future developments was discerned by the market, and strengthened as the approach of the redevelopment date diminished the dilution effect of the discounting process on future values. The further in the future is a value, the smaller is its discounted present value, or capitalised value, at any given interest rate. For example, if the ruling interest rate is 5%, $1m in 20 years time is worth 1m/(1 + 0.05)²⁰ today, which works out at $359,000. But $1m ten years away is worth $585,000 today. In other words, $585,000 put in the bank today at 5% compound interest would be worth $1m after ten years. So the present value of $1m receivable ten years hence, with a 5% discount rate, is $585,000. That is what the market would add to the current price of a site if it expected a lump sum capital gain with those particulars. Each year approaching that sudden use value increase, the market value of the site would increase by 5%, by the very nature of the discounting process that set the original present value of that increase when it was first anticipated.

The particulars are grossly simplified in this illustration. The use-value of a site increases gradually as the optimal redevelopment date approaches. Each succeeding year's hypothetical optimal redevelopment yields a larger future stream of site rents (net revenues minus construction costs), which capitalises into a higher use value. The market value of the site began to grow faster than the use vale when the market perceived a use value peak on the temporal horizon, so the use value must at some stage begin to climb faster than the market value if it is to reach that horizon and merge again with the market value. While use value is growing faster than market value, it is economically irrational to develop the site. The landowner would only be able to sell the site at the use value supported by the premature construction upon it, and that is less than its market value - what it would fetch without development. The next year's building would confer a higher use value on the site and the growth in use value would be at a percentage rate greater than the interest that could be gained by selling at market value and putting the money in the bank. The profit-maximising landowner would wait until the growth rate of use value had fallen back to growth rate of market value, in other words the two values had merged and they were equal to the interest rate on money in the bank. That is the optimal date for redevelopment of the site. That use is now feasible which maximises the stream of rents over time. Future developments after that date cannot provide rent increases sufficient to cover the rents foregone in the meantime.

Figure 3 presents the situation described graphically, using illustrative figures. The line " V accrual" represents the annual accrual to the market value of a site which is growing at a ruling interest rate of 5%. The line "S/mkt i (chall)" represents the use rent of the site in its optimum use for each year (which is capitalised into use value, S, at the market rate of interest). The line "S accrual (def)" represents the annual accrual to the use value of the site. It is assumed for simplicity of presentation that the site is vacant, the tag rent from its earlier use having expired (an assumption also made in the previous exposition). Until 1985 the site earns more for its owner in its vacant state than either selling the site at market value and banking the proceeds, or initiating optimal present development, would. The amount of the accrual to the use value of the site as each year brings greater potentialities may be regarded as the defending use's (vacancy's) income. The rent achievable each year by the optimum development of the site is the challenging use's income. In 1985 the challenger catches up with the defender and is poised to overtake. That is D-date, the optimal year for redevelopment. That is the year that has set the market value of the site in previous years. Delaying for a further year will squander potential income.

Returning to Figure 2, a series of D-dates is indicated representing transitions every generation to higher discrete uses as the rapidly growing settlement expands past point A throughout the twentieth century, until it is abutting on the commercial centre. The central point of a settlement clearly cannot migrate out from the centre, so there is a limit to the analogy that can be drawn between the passing of time and the migration of land uses across space in the diagram. But it serves the purpose of setting the development decision in its widest perspective. Figure 3 focuses on a particular development decision, and demonstrates that though the lines are drawn straight in Figure 2, they are in reality curved. The use value line, S, traces a wave pattern, with the crest of each wave, or swell, coinciding with the transitions between each land use. The revenue and cost curves are by implication curves as well. The market value of land, if drawn, would look more like the use value line as drawn.

The crucial fact is that the market value and use value curves are identical at the optimal development dates. This is the criterion that determines the development decision (as stated by Gaffney in his 1973 paper, p.138). It is the date at which the three curves intersect in Figure 3. Using this criterion, it is possible to judge the effects of various taxes on the date of the development decision. Figures 3, 4 and 5 compare the changes to the non-tax situation affected by 1% annual taxes on the market value of land (which includes speculative value), the use value of land (the land's current optimal rent capitalised) and the value of the optimal new building in each year, respectively.

Neither of the taxes on land alter the optimal development date. The tax on improvements value, however, sets back the date by 5 years. This differential occurs despite the capitalisation of all three taxes into lower land values. The explanation lies in the crucial observation above that land generally ripens into a lower "economic intensity" of use. As the market and use values of land rise, the proportion of construction value in the value of a property with an optimal new building (never mind depreciated buildings) falls. This means that equal rates of taxes on each of these values impose different tax liabilities, to be subtracted from land values, at different stages of ripening. A buildings tax falls relatively heavily on unripe sites - just after they have been redeveloped - and least of all on ripe sites - not at all in the case of vacant sites. It therefore steepens the slope of land values between redevelopment dates. Direct taxes on land values, however, lower land values proportionately, so do not increase their slopes. Alteration of the growth rate of the use value of sites shifts the development date, as illustrated in Figure 6.

[1] "Urban Land Taxation for the Economic Rejuvenation of Center Cities: The Pittsburgh Experience (Progress Report, June 1991)."

[2] "The Site Value Tax and Land Development Patterns: Evidence from Melbourne, Australia."

[3] "Capitalisation of Taxes and Subsidies," in C.Lowell Harris (ed.), Government Spending and Land Values, Madison: University of Wisconsin Press, 1971, p.21.

[4] "Adequacy of Land as a Tax Base," in D.M.Holland (ed.), The Assesssment of Land Value, Madison: University of Wisconsin Press, 1970, p.187.

[5] Shawna P.Grosskopf and Marvin B.Johnson, "Land Value Tax Revenue Potentials: Methodology and Measurement," in Richard W. Lindholm and Arthur D. Lynn, Jr. (eds.), Land Value Taxation, Madison: University of Wisconsin Press, 1982, p.65.

[6] "Land Value Taxation and Housing Development," in American Journal of Economics and Sociology, Vol.49, No.1, 1990, January, p.102.

[7] "Tax Reform to Release Land," in Marion Clawson (ed.), Modernizing Urban Land Policy, Baltimore: John Hopkins Press, 1973, p.140.

[8] W.D.Fraser, Principles of Property Investment and Pricing, London: Macmillan Education Ltd, 1984, p.248.

[9] Brian L.Bentick, "The Impact of Taxation and Valuation Practices on the Timing and Efficiency of Land Use," in Journal of Political Economy 87, 1979, p.860.

[10] Bourassa has also recently refuted Bentick's analysis, which he argues "is incorrect": "Economic Effects of Taxes on Land: a review," in American Journal of Economics and Sociology, Vol.51, No.1, 1992, January, p.110.

[11] "Whose Water? Ours", paper presented at "Whose Water?" conference, sponsored by Polly Dyer, Institute For Environmental Studies, University of Washington, September 29-30, 1989.

[12] This statement must be qualified.In Figure 3 in Appendix 2, the site owner pays a tax of $7.84 per annum on land value in 1990 and receives a capital gain of $31.34 in that year. The investor's interest rate of 5% means that the site owner can receive $31.34 in interest by selling the site and banking the proceeds. But site owners who have to pay 25% or more income tax on interest find that vacant land is a competitive investment - and this fact holds true for every other year in Figure 3. The capital gain minus the land value tax equals interest in the bank minus 25% income tax. Even if capital gains tax is expected on sale of the site the percentages do not change, for capital gains tax, like land value tax, lowers all land values by the same proportion.

Thus, in the absence of any other taxation LVT might encourage premature development of vacant or near vacant land. In Figure 3, if there were no income tax the landowner would wish the accrual of use value, S, to cover the cash payment of tax as well as the interest foregone on tax. The landowner would, as it were, wish the site to pay for that portion of the equity which is rented from the state, as well provide competitive returns. The effect would be to precipitate development in 1993, by subtracting the tax from "S accrual (def)" a second time. However, any existing-use land rent from the tag end of a previous, or an interim use, would offset the need for optimal-use value accrual to cover the tax. It would provide cash without affecting tax liability. Indeed it seems likely that an interim use simply to pay the tax would in most cases be the preferred option to developing prematurely and locking into an underimprovement while the market value of the site keeps on rising, and with it LVT liability.

[13] A point that has not yet been made is that "rents foregone" are not essentially a "cost" at all - that they serve the allocating function of "opportunity cost". They are only a cost in the sense that savings. or consumption foregone are a cost. Rents foregone are income which is not consumed but invested in the right to receive future land rents and capital gains. Rents foregone are income provided by the land and reinvested, or added to the wealth of the landowner. Interest paid on loans to buy land, or land value taxes paid, are income really foregone. They have to be paid out of a landowner's other assets or incomes; they clearly do not add to the wealth of the landowner. That is why, as Gaffney pointed out, they have cash and liquidity effects which mere opportunity cost does not.

[14] Gaffney, 1973, op. cit., pp.149-151)

[15] We are not talking about the introduction, or raising of the rate, of LVT here, only a steady rate of LVT.

[16] Lusht, 1991, op.cit., p.527.

[17] "Is Site Value Still an Appropriate Basis for Taxation?" pp.94-96.

[18] J.F.N.Murray, Assessment Practices and the National Economy: An historical study in the United Kingdom and Australia, Hartford, Con.: John C.Lincoln Institute, 1967, p.46.

[19] Ibid., p.78.

[20] Ibid., p.81.

[21] Frank Brennan, Canberra in Crisis, Canberra: Dalton Publishing Company, 1971, p.6.

[22] Ibid., quoted, p.6.

[23] Kenneth Taeuber, "A Century of Experience with Land Value Taxation," in A. Woodruff, et al., (eds.) International Seminar on Land Taxation, Land Tenure and Land Reform in Developing Countries, Phoenix: John C. Lincoln Foundation, 1967, pp. 137-138.

[24] Murray, op.cit., p.86.

[25] Taeuber, op.cit., pp.150-151.

[26] "Site Value Taxation in Central Business District Redevelopment (Sydney, Australia)," ULI Research Report 19, 1972.

[27] Brisbane City, Committee of Inquiry into Valuation and Rating, 1989: Vol. 1, pp.11-12.

[28] Taeuber, op.cit., p.148, 155.

[29] Financial Times, "Survey: Australia," November 7, 1991, p.5.

[30] Op.cit., p.157.

Return to Contents