RDP 2006-12: Housing and Housing Finance: The View from Australia and Beyond 3. National Features

3.1 Tax System

Housing is both an asset and a source of housing services (imputed income), and it can be owned either by the occupier or a landlord. Therefore the tax system can affect behaviour in the housing market in multiple ways and at multiple points in the life-cycle of ownership. Some relevant features are summarised in Table 3. One set of taxes can affect a household's decision of whether and when to transact in the housing market; that is, how often they buy or sell property. For example, a transaction tax (stamp duty) affects turnover directly by fixing a wedge between prices paid by buyers and returns received by sellers. This limits buyers' capacity to pay by increasing the ‘deposit gap’ between feasible borrowings and the total cost of the dwelling (RBA 2003b). It also limits the incentive to turn properties over frequently, reducing the extent to which an upswing in housing prices can attract speculative demand aimed at short-term capital gains.[6]

Table 3: Features of Taxation Systems Relevant to Housing Markets
Country Mortgage deductibility   CGT   Land/property tax   Negative gearing Depreciation
Owner Investor Owner Investor Owner Investor Investor Investor
Australia No Yes   No ½ rate(c)   Limited Yes   Yes Yes(d)
Canada No Yes   No ½ rate(c)   Yes Yes   Yes(e) Yes
France No Yes   No No(f)   Limited Limited   Limited(g) Yes
Germany No No   No(f) No(f)   Limited Limited   Yes Yes
Netherlands(a) Yes na   na na   Yes Yes   na No
NZ No Yes   No No   Limited Limited   Yes Yes
Sweden Yes Yes   Limited Limited   Yes Yes   Yes No
Switzerland(b) Yes Yes   Yes Yes   Yes Yes   No Outlays
US Yes Yes   Limited Yes   Yes Yes   Limited(h) Yes
UK No Yes   No Yes   Limited Yes   No No
Notes: Under CGT, ‘limited’ means homeowners may defer payment provided the proceeds of sale are reinvested in housing. Under land/property tax, ‘limited’ refers to property owner charges along the lines of council rates, which are linked to local services and need not move proportionately with property values.
(a) The Netherlands levies a tax on net wealth using an assumed rate of return, so negative gearing is not possible.
(b) Swiss homeowners pay tax on imputed rental income, net of interest and renovation costs.
(c) CGT is levied in Australia and Canada at half the taxpayer's marginal rate if the holding period exceeds one year, but in Canada gains resulting from changes in the cost base due to depreciation are levied at the full rate.
(d) For buildings constructed after 1985.
(e) Only cash expenses, not depreciation, can be negatively geared in Canada.
(f) Provided property owned for at least 15 years (France) or 10 years (Germany).
(g) Negative gearing allowed up to a set limit and interest costs may not exceed gross rent.
(h) Rental property expenses cannot be deducted against unrelated labour income in the US, which effectively limits negative gearing to professional investors and developers.

Sources: adapted from BIS (2006); Ellis and Andrews (2001); RBA (2003b); Scanlon and Whitehead (2004) with some updating from national sources.

Capital gains taxes (CGT), with exemptions or concessions for assets held for longer holding periods, may also influence speculative demand. For example, the half marginal rate paid on capital gains in Australia refers to assets held for at least a year; assets held for less than this period attract CGT at the full marginal rate. In France and Germany the required holding periods to obtain concessional taxation of capital gains are considerably longer, having been extended from two years to ten years in Germany in 1998 (Scanlon and Whitehead 2004).

More generally, capital gains taxes influence the incentive to invest in residential property and other gains-producing assets such as equities, relative to assets that provide an income flow alone. This is particularly pertinent given that purchase of real estate is often highly geared, partly because interest payments for mortgages on rental properties can be written off as an expense against tax in most jurisdictions. The tax regime in Australia is usually considered to be among the most generous towards individual landlords, offering concessional taxation of capital gains relative to income flows, and the ability to negatively gear expenses against other income, including non-cash depreciation expenses. This has been previously identified as a factor encouraging small-investor participation in the housing market in Australia, particularly in an environment of rising prices (RBA 2003b; BIS 2006). On the other hand, some studies have argued that certain segments of Australian landlords – mainly those renting to low-income households – face higher effective tax rates than landlords in the UK, even though the UK tax system does not permit negative gearing (Wood and Kemp 2003).

A second set of taxation arrangements can influence the funding of home purchases. Mortgage interest deductibility affects the capacity to service debt and the incentives to repay principal. This in turn affects incentives to take mortgages with fixed versus variable interest rates. When interest payments are not deductible, mortgage borrowers are effectively paying their mortgage out of post-tax income. This implies that the post-tax return to paying down the mortgage will generally exceed the post-tax return on investing in financial assets, providing an incentive to pay down the mortgage rapidly if possible. Such an incentive encourages the use of variable-rate mortgages, which are less likely to involve prepayment penalties.

Specific tax concessions can influence the structure of ownership of the dwelling stock. For example, real estate investment trusts (REITs) in the US qualify for tax-free status provided they distribute most of their earnings to shareholders and fulfil certain other conditions. This increases the incentive for at least some of the private rental stock to be owned and managed by institutions rather than individual landlords. In countries such as France and Germany, there are concessions designed to encourage the construction of rental housing, particularly in the market segment serving low-income households (see Scanlon and Whitehead 2004 for more details).

Despite the clear incentives for certain patterns of funding and financing embedded in tax systems, it is difficult to show a simple mapping between features of taxation systems and macro outcomes such as debt-income and housing price-income ratios. This is because the tax regime interacts with the other aspects of the housing–finance system in sometimes complex ways. Some of these other features are discussed in the following subsections.

3.2 Structure of the Financial System

Households in different countries access mortgage finance at widely varying terms. As summarised in Table 4, loan terms can vary from 15 to 45 years. The maximum allowable loan-to-valuation ratio (LTV) also differs substantially, as does the typical LTV for new loans.[7] Variable-rate loans predominate in some jurisdictions, while in others fixed-rate loans are more important. The term of the fixed rate need not be the same as the term over which the mortgage loan is amortised and this difference can vary across countries.

Table 4: Contract and Funding Features in Selected Mortgage Systems
Typical loan term (years)
New loans estimated average LTV ratio (per cent) Variable-rate loans (per cent of total)
Prepayment penalties
Australia 25 60–70 ∼85 For fixed rate Extensive
Canada 25 75–95 29 Some fixed rate Extensive
France 15–20 78 20 Limited by statute Limited
Germany 20–30 80–100 30 Only fixed rate (by law) Some
Netherlands 30 87 (max 125) 26 Some fixed rate Extensive
NZ 25–30 80–85 16 For fixed rate Very limited
Sweden 30–45 80–95 98 For fixed rate Limited
Switzerland 15–20 80 33 Some fixed rate Limited
UK 25 70 95 For fixed rate Some
US 30 ∼85 25 None Extensive

Note: ‘Variable-rate’ includes loans fixed for up to two years for most countries shown except NZ, for which only fully floating-rate mortgages are included.

Sources: NZ – RBNZ estimates; other countries – BIS (2006); Green and Wachter (2005); national sources; see notes to Table 3 in BIS (2006) for more detail

A few common cross-country trends emerge nonetheless. Countries where mortgage interest payments are not tax-deductible for owner-occupiers tend to have systems where the predominant mortgage type is either a variable-rate loan or loans with interest rates that are fixed for a relatively short period compared with the contract term. This is partly an endogenous response to the tax incentive described above, which creates an option value on the ability to make prepayments if possible.[8] There is also some tendency for the average LTV to be lower in these countries than in those where deductibility is possible, as can be seen from a comparison of Australia and the UK with the Netherlands and the US. This also seems to be the conclusion drawn from the experience of the UK, where the deductibility of owner-occupied mortgage interest was gradually removed over the 1990s, and the prevalence of high LTV loans fell accordingly (Hendershott, Pryce and White 2002).

Mortgage deductibility is also implicated in the tendency of borrowers in some countries to adopt products like endowment mortgages; these are an interest-only mortgage attached to an account that works like a managed fund. The idea is that the returns on the accumulated assets are more than sufficient to repay the loan principal at the end of the loan term. This can only work if the post-tax return exceeds the interest rate paid on the mortgage debt that would otherwise have been paid down, which is more likely if mortgage interest is deductible. It is therefore no surprise that this type of instrument has lost popularity in the UK, become more prevalent in the Netherlands, and has essentially never been adopted in Australia.

In addition to tax systems, differences in other government regulations and interventions have influenced mortgage markets, particularly the development of funding markets. For example, in many jurisdictions, legislative and regulatory support was needed before securitisation of mortgage loans could occur. In some countries, such as the UK and some in continental Europe, mortgage-backed securities (MBS) or mortgage bonds (which have similar properties to MBS) have required enabling legislation setting out the rights and responsibilities of issuers and bond holders. In others, such as Canada and the US, government support for securitisation markets has been crucial to their development. The Canadian Mortgage and Housing Corporation is a government-owned agency responsible for insuring mortgages. In the US, the so-called government-sponsored enterprises (GSEs) Fannie Mae and Freddie Mac were founded to support mortgage markets and the expansion of homeownership. In both countries, these publicly supported agencies have been instrumental in supporting the MBS market by setting standards for underwriting and, in the case of the US GSEs, holding significant MBS on their balance sheets.[9]

Many other national differences in outcomes have simply reflected the endogenous developments of products and conventions in the light of historical practice and competition between lenders. The past decade or so has seen substantial innovation in the range of mortgage products available, and in the ways that these products are funded, as was already described in Section 2.1. These changes have tended to result in more households having access to more finance than had been the case previously.

In particular, there has been a tendency towards allowing higher LTV ratios. That is, households do not need to make as large a down payment as in the past. This has been an important development, since as disinflation and deregulation allowed households to service larger debts, the down payment required to do so would also have increased substantially relative to income. If this down payment constraint had not eased at the same time as the repayment constraint was eased, the effect on housing prices and household sector indebtedness would probably have been smaller (Ellis 2005).

The institutional framework in the financial sector has influenced the extent of mortgage product innovation in individual countries. For example, the role of technological innovation in driving product innovation has been most apparent in the United States, where the dominant presence of the GSEs and the widespread practice of mortgage originators securitising their loan books have encouraged development of data-driven credit scoring and automated underwriting practices (BIS 2006). These innovations have occurred to a much lesser extent in countries where lenders tend to keep loans on their own balance sheets and therefore face weaker incentives to package loans into standardised types with consistent degrees of credit risk.

Changes in the structure of the financial system have been important drivers of the evolution of the mortgage markets of particular countries. For example, as mentioned in Section 2, the entry of a new class of lender in Australia in the 1990s resulted in lower interest rate margins, new products such as low-documentation loans, and innovations in funding such as the wider use of MBS. Similarly, the entry of banks into the UK mortgage market in the 1980s and 1990s increased competition in a market that had previously been dominated by building societies. The demutualisation of several major building societies may also have contributed to this increased degree of competition. By contrast, one reason why there has been less product innovation in some of the major European mortgage markets could be that there has as yet been little cross-border competition or examples of lenders breaking into new markets in neighbouring countries (BIS 2006).

3.3 Legal System and Housing Policy

Government policy affects housing markets beyond the measures specifically relating to the financial system. Government interventions especially influence structural features of the housing market, including the owner-occupation rate, who owns rental properties, and the types of housing in the rental segment. This can affect housing prices and household balance sheets if different types of owners have different motivations and borrowing capacity. Table 5 summarises some of the relevant features for a selection of countries.

Table 5: Legal and Institutional Features of Housing–Finance Systems
Country Owner-occupation rate (per cent)   Social housing (per cent) Rent control Institutional landlords (private or social)
1980 2002–2004 Latest
Australia 71 72   5 No Almost absent
Canada 62 66   6 Some provinces Some REITs
France 47 55   19 Cost-based 1.2% non-social
Germany 41 42   6 Continuing tenants ∼15% total stock
Netherlands 42 54   35 Yes Social housing
NZ 73 68   6 No Almost absent
Sweden 58 61   21 Yes Mainly municipal
Switzerland 33 35   2 Cost-based ∼⅓; rental stock
UK 65 68   20 Social only Social housing
US 58 69   3 Some REITs

Note: Social housing shares are for 1997 (Sweden; municipal housing), 1998 (Netherlands), 1999 (Australia), 2000 (Switzerland), 2001 (Canada, Germany), 2002 (France, UK, US), 2004 (NZ).

Sources: adapted from BIS (2006); European Parliament (1996); Scanlon and Whitehead (2004) and national sources

One of the most important differences across countries is the importance of the social housing sector, including public housing owned directly by government and housing owned and managed by non-profit organisations, charities and enterprises associated with municipalities. Social housing removes an entire segment of the dwelling stock from ownership by the household sector, with obvious implications for the size and composition of balance sheets and the sector's sensitivity to changes in housing prices. In countries with large social housing sectors such as Sweden and the Netherlands, many of its tenants include middle-income households that might have been owner-occupiers in other jurisdictions. By contrast, in countries with low shares of social housing such as Australia and the US, this housing type tends to be highly targeted to low-income and disadvantaged households, who have lower propensities to own their own homes.

Government regulation of the landlord–tenant relationship, including imposition of rent controls and regulating the terms under which tenants may be evicted, clearly influences the supply of private-sector rental housing. Rent controls reduce their responsiveness to current market conditions and generally reduce rental returns. For example, deregulation of rents in the UK and the introduction of Assured Shorthold Tenancy encouraged the expansion of the rental market in that country. Figures from 1998 showed that net rental returns on properties that were still covered by the old arrangements were less than two-thirds of the returns available on properties under assured shorthold tenancies (Crook and Kemp 2002). Similarly, restrictions on landlords' ability to evict bad tenants or sell the property when they want will increase the riskiness of investing in rental property. This raises returns required to encourage additional stock to be made available to the rental market, thereby constraining the supply of rental properties.

Housing policy can also affect incentives for different parts of the private sector to own the rental properties, and the type of housing that they own. Specific incentives or requirements to invest in rental housing for low-income households have been important throughout Europe and North America (Scanlon and Whitehead 2004); supply of some amount of low-cost housing is sometimes a condition of planning permission for residential developments. On the other side of the market, encouragement of homeownership has been an explicit policy of successive US governments for many decades. This was the motivation for the creation of the GSEs, as well as for many smaller-scale efforts such as the subsidisation of construction of properties for the owner-occupied market; there is evidence that the latter have boosted both ownership rates and housing values in neighbouring areas (Ellen et al 2001).

Another feature of the legal system that can influence the pattern of housing ownership is the way that title to property is allocated and divided. In particular, small-scale ownership of rental property by individual landlords is more likely where individual apartments in a block can have different owners; for example, what is known as condominium structure in most jurisdictions, or strata title in Australia.[10] Even where ownership of condominiums is possible, other aspects of housing policy and the legal system can discourage small-scale landlords. In Canada, the authorities' housing policies are directed towards ensuring sufficient supply of so-called ‘conforming’ rental stock – that is, dedicated blocks of apartments that are all rented out. Other types of rental stock such as individual condominiums and detached houses are termed ‘non-conforming’ and are discouraged, partly because of concerns that these dwellings might be withdrawn from the rental stock at a later date (Clayton Research 1998; Crook 1998).

These policies can have unintended consequences, both in the housing market directly and in terms of macroeconomic outcomes. Rent controls and other measures designed to support tenants can sometimes work to their disadvantage, as they raise required rental returns to investors and restrict supply. For example, the focus on ‘conforming’ rental property in Canada seems to have resulted in relatively high costs of renting compared with owning and very low rental vacancy rates (Traclet 2005), but very little new supply of apartments (Crook 1998).

3.4 Geographical Features

The physical characteristics of a country can affect the outcome of an increase in housing demand brought about by greater availability of finance. Because most people in industrialised countries live in urban areas, housing outcomes can be affected by the characteristics of the cities themselves, including their location, geographical spread and density, and perhaps also by the structure of the relationships between cities.

Traditional urban economics has antecedents in the von Thünen model of rural land rent and crop distribution. This literature assumes cities have a single employment centre to which residents commute each day from their homes further out (Mills 1967; Muth 1969). In this model, living at the fringe of the city is generally less desirable than in the centre because commuting times are greater, so housing prices are lower at the fringe. This pattern is observable in the data for most countries;[11] Table 6 shows the ratio of inner-ring to outer-ring prices for the major cities in Australia. Although there was some tendency for inner-suburban property to become relatively more expensive during the turn-of-the-century price upswing, overall these ratios seem fairly stable within cities. There is a slight tendency for the difference to be greater in the larger cities, although Melbourne's price differential would seem out of line with that relationship.[12]

Table 6: Australian Metropolitan Median House Prices
Ratio of inner-suburban to outer-suburban median prices; December quarter data
City 1998 2002 2005
Sydney 2.19 2.16 2.01
Melbourne 1.72 1.82 1.79
Brisbane 1.65 1.68 1.89
Adelaide 1.52 1.99 1.85
Perth 1.70 1.97 1.95

Sources: Real Estate Institute of Australia; author's calculations

When cities are more spread out, it is harder to provide efficient transportation over the whole city, so the costs of traffic congestion become relatively greater at the fringe than if the city had been more compact. This would suggest that the price gradient from inner areas to outer areas should be steeper for more sprawled cities with poor transport options.[13] The effect of the steeper price gradient on average city-wide prices is ambiguous: it could imply either cheaper housing at the fringe relative to the fringes of other cities, or more expensive inner-city housing. Most evidence from the US suggests that average housing prices are lower in more sprawled cities; as a simple illustration of the point, Figure 5 in the next section shows that prices are quite low in the canonical urban sprawl of Atlanta, compared with other large cities. However, sprawled cities generally differ from denser cities in other ways as well. The denser cities are usually older, larger and more likely to be constrained from expansion by coasts and other natural barriers (Fulton et al 2001). As such, at least some of the higher density is probably an endogenous response to the costs of commuting and congestion in large cities, while some of it can be accounted for by exogenous geographic factors.

On the other hand, sprawl usually occurs when land availability is greater, so plot and dwelling sizes can be larger at the fringe than would be possible in more compact cities. Realised prices of outer-suburban housing might therefore be quite high in sprawled cities, but a regression controlling for housing quality would still identify a steep price gradient. Table 7 shows that countries where overall population densities are lower and cities are more sprawled, such as Australia and the US, generally have larger dwellings with more floor space than countries with higher population densities and more compact cities.

Table 7: Selected Demographic and Urban Stock Characteristics
  Total population
density (2001)(a)
Average size of
existing dwellings
Houses Detached houses
Persons/km2 m2 Per cent of housing stock
Australia 2.5 131.8(c) 85.6 76.5
Canada 3.3 114.0 66.4 55.9
France 107.1 88.0 56.2 na
Germany(b) 230.5 86.7 45.6 31.0
NZ 14.3 132.0(d) 83.0 73.0(d)
UK 243.8 84.0 80.7(e) 25.6
US 30.8 156.5 66.7 60.6

Notes: (a) For whole country
(b) German housing stock data refer to West Germany only
(c) Excludes public housing
(d) Detached house and floor space data refer to Auckland only
(e) House data refer to England only

Sources: reproduced from Berger-Thomson and Ellis (2004); Ellis and Andrews (2001); RBA (2003b)

Another aspect of the role of urban density and the type of dwellings in the housing stock for national outcomes is that certain dwelling types are more conducive to institutional versus individual ownership of the private rental stock. Institutions are more likely to seek to reap the economies of scale in property management by owning whole apartment blocks or housing estates. In countries such as Australia and NZ, where the dwelling stock is disproportionately comprised of detached houses, these scale economies are not available, and individual households are more likely to be landlords than is the case in some other countries.

Whatever the effect of city density on housing prices in the long run, more sprawled cities probably show less tendency towards short-run price surges in response to increased demand for dwelling numbers, for example when population growth increases. This is because the more sprawled a city is, the greater is the proportion of detached single-family homes in the housing stock, as opposed to apartments. The logistics of detached-house construction are much simpler than for a multi-storey apartment block, which requires deeper foundations, lifts and other more complex engineering elements. So although detached houses consume more land than the same number of dwellings in the form of apartments, they can usually be built more quickly. Supply will therefore be more responsive to demand for extra dwellings. This may explain why Berger-Thomson and Ellis (2004) found that the estimated supply curve for the number of dwellings in the UK is steeper than in Australia, the US or Canada, where population density is lower and detached housing is a larger share of the housing stock.

This greater responsiveness to an increase in demand for extra dwellings will make little difference in the face of a surge in demand for the entire stock of housing as described in Section 2. The construction of additional properties at the fringe of a city does not do much to supply in response to an increase in the demand for average housing services per property, that is, average housing quality. To the extent that the newly built properties are of the currently desired quality but existing properties are not, some households might be induced to sell their existing home and move to a newly built one. Vacancy rates in the established areas would rise as a result, with the older, lower-quality homes either renovated or demolished and replaced over time. It seems that this would result in a slower process of adjustment than simply renovating the existing dwelling stock. On the other hand, it is quicker to demolish and replace detached houses than whole apartment blocks, so the renovation process might also be quicker in more sprawled cities.

The implications for prices of a demand surge might be different if the country has only a few big cities, rather than a network of many smaller cities. In general, the larger is a city's population, the more expensive is its housing. This is a well-known result from urban economics (Gabaix 1999). Large cities offer advantages in terms of the range of jobs and products available because the size of the market is larger (Fujita, Krugman and Venables 2001; Fujita and Thisse 2002). There may also be productivity spillovers from living in large or dense population centres where there are more people and firms to learn from – so-called Jacobian externalities (Jacobs 1970); Kohler and Smith (2005) present evidence that wages and housing wealth are both higher in more densely populated areas.[14] In equilibrium, these advantages must be balanced by the disadvantages of living in a big city. Otherwise, the population of one city would all move to one that is more attractive in net terms. Congestion costs such as traffic jams and crime are one type of disadvantage that supports the equilibrium population distribution, but housing costs are a particularly powerful disincentive against cities becoming too large.

This implies that the distribution of population between cities has a compositional effect on aggregate household balance sheets at the very least; a population concentrated in a few cities will result in higher national average housing prices and assets (Ellis and Andrews 2001). Moreover, if there is only one large city, the positive externalities of urban living cannot be provided by alternative locations. An upswing in housing demand in that city cannot easily be siphoned off by households moving to other cities where housing is cheaper.


Transaction taxes or stamp duties apply to property sales in most of the countries listed in Table 3, but the rates usually do not have simple relationships to sale price. Real estate agents' fees and other administrative costs can reinforce the effects of transaction taxes, together adding as much as 10–15 per cent to the purchase price in some European jurisdictions (BIS 2006). [6]

The typical characteristics of loans offered in other markets vary even more widely than for the peer group shown here. Loan terms are generally shorter in emerging markets (for example, 10–15 years in Mexico and 3 years in Korea) and have more stringent down payment requirements, but these features are generally converging towards those seen in the major economies. A similar pattern of historical development was evident in mortgage markets in industrialised countries (Green and Wachter 2005). [7]

This is not a universal trend, however; in some countries not shown in Table 4, such as Spain, prepayment penalties also apply to variable-rate loans (BIS 2006). [8]

Green and Wachter (2005) provide a comprehensive overview of the historical development of the US mortgage market and its peculiar features; see Courchane and Giles (2002) for a comparison of the historical development of the US and Canadian mortgage markets and associated government interventions. [9]

Even if some sort of condominium structure does not exist, as was the case in the UK until 2004, it can still be possible to have separate ownership of apartments, but there must be a residual owner of all the land under the apartment block. [10]

But not all countries: in the US, for example, suburban locations have often been preferred because they allowed higher-income households to avoid city taxes and inner-city schools, while retaining access to the same metropolitan amenities and job market. When US cities have reverted to the more usual situation of inner-city properties being more expensive, it has sometimes been cause for comment (McMillen 2002). There are also some non-US cases of inversion of the land-price gradient, that is, where outer areas are more expensive than inner areas, such as Haifa in Israel (Plaut and Plaut 2003). Some recent research has focused on the development of town centres at the fringe of existing cities (for example, Garreau 1992; Glaeser and Kahn 2001; Lucas and Rossi-Hansberg 2002). [11]

This might be an artefact of the Real Estate Institute of Australia's definitions of inner and outer suburbs. [12]

Consistent with this, improved transport to the outlying areas was one of the suggested causes of the land-price gradient inversion in Haifa cited by Plaut and Plaut (2003). [13]

See also Ciccone and Hall (1996) and Glaeser and Maré (2001). [14]