The Rent Assistance Programme: A Synthesis and Analysis of Stakeholder Views 

• 4.1 Difficulties in securing affordable and appropriate housing
• 4.2 Government housing programmes
• 4.3 Targeting of RA

This section presents discussion of some of the key issues raised in the discussions. It does not try to reach conclusions about what the policy answers should be. It analyses some of the complex issues that were raised, drawing on factual evidence from previous research studies and considerations of how policies and programmes may interact.

4.1 Difficulties in securing affordable and appropriate housing

Supply shortages
Trends in the supply of low cost rental housing are explored in Yates, Wulff and Reynolds (2004). The authors note that between 1986 and 1996 there was diminishing supply of rental property at the lower cost end of the market (at constant prices) accompanied by a large increase in the number of low income households in the private rental market. They find that between 1996 and 2001 the private rental stock grew, but all of that growth was in properties in the top quintile of the rent distribution. Over the same period rising incomes reduced the number of households who "need" this low rent stock. Consequently, on one measure, the shortage of low rent dwellings declined from 150,000 in 1996 to 134,000 in 2001. But there was an increase in the proportion of the low rent stock occupied by middle and high income households. The authors conclude that the shortfalls were most serious in Sydney and Melbourne.

Non-shelter outcomes
Bridge et al (2003) conducted a very extensive review of the literature relating to housing assistance and non-shelter outcomes. Their findings are not specific to RA. They report that:

• Housing assistance can foster better educational outcomes by freeing up household resources to invest in education and training and by providing security of tenure and a stable learning environment. It can also, by improving the physical, social and economic foundations of neighbourhoods, positively influence education outcomes - but it can too have adverse effects, for instance if it produces spatial concentrations of poor households.

• The impact of housing assistance on labour supply depends on the design of the assistance programmes. (Empirical effects of affordability support are considered in more detail in projects described below.) Tenure is also a relevant factor, and it may be that homeownership and public housing occupancy reduce a person's willingness to relocate in search of employment.

• Housing circumstances are correlated with curative health outcomes for some groups (notably older Australians and people with a mental illness). The connections between housing, ability to access support services, and health outcomes are cloudy. Certainly correlations exist between secure and adequate housing and some morbidities. Housing renewal can reduce morbidity and improve wellbeing and life quality (but is not guaranteed to).

• Overcrowding has been identified as a causal factor in increased infection rates and respiratory disease. It has ambiguous consequences for wellbeing and mental health, and perhaps an important question is how much capacity householders have to stop overcrowding when it does become problematic.

• Concentrations of public housing have been associated with high crime rates. There is a range of possible reasons for this, including poor design, a tendency for public housing allocation mechanisms and/or self-selection to direct individuals with a predisposition to crime into particular estates.

• There are many studies suggesting that homeowners are more socially connected with their communities than renters. Homeownership has been said to have a range of positive non-shelter impacts but the isolation of explicit causal linkages is difficult. Homeownership is clearly associated with wealth accumulation.

• There are many studies that conclude that neighbourhood is an important influence on important outcomes for children and adults. However, there is much uncertainty as to just what features of a neighbourhood play a causal role.

Phibbs (2005) report the results of a study of non-shelter outcomes for people moving into public housing. On average the respondents reported improvements in health, feelings of safety and security, and school outcomes. There were mixed effects on labour market participation.

Bradbury and Chalmers (2003) investigate connections between changes in location, employment and housing affordability. They found that, compared to other income support recipients, unemployed people were more likely to move, and likely to move further. Younger and single people were more likely to move as were those with some non-benefit income or renting privately. People in high housing cost regions were also more likely to move as were those living in low unemployment regions, but often to other low unemployment localities.

Whelan (2004) uses the HILDA dataset to analyse the impact of housing assistance on labour market activity. He finds that "there is some evidence that receipt of RA reduces the likelihood that an individual is observed to be engaging in paid employment. Conversely the analysis of hours worked suggests that receipt of [housing assistance] measures, especially RA, does not impact on the choice of hours worked".

With studies of this type there is uncertainty about the extent to which RA (or indeed any other form of assistance) causes worse labour market outcomes. In the absence of any government interventions at all, one would expect to see varying outcomes in the labour market. But people who would, in the absence of government housing programmes, get poor labour market outcomes, are more likely to be people that actually do take up housing programmes. Researchers in the studies cited attempt to control this. However, it is notoriously difficult to filter out these complications completely, and there is still some uncertainty as to the exact consequences of RA for labour market participation.

Wood et al (2005) also consider the impacts of housing assistance on incentives for labour market participation. In their study incentive mechanisms are explicitly modelled. EMTRs are used to capture incentive effects on hours worked and "Replacement Ratios" (RRs) are used to capture incentive effects on having a job or not.⁷ Inter alia the study shows that:

• Although recipients of government pensions and benefits have EMTRs which are substantially higher than those of non-recipients, RA itself generally does not boost effective marginal tax rates and therefore has relatively limited incentive effects on hours worked. This is in sharp contrast to public housing which does boost EMTRs. RA does not have much impact on EMTRs because it is not income-contingent; public housing rents are.
• Disincentives to have a job are on average smaller for RA recipients who are not on family payments. This conclusion rests on the fact that income support payments tend to be relatively smaller for this group whereas there assumed earning potential is not at so sharp a discount to family payment recipients.⁸
• The impact of RA entitlement itself modestly increases disincentives to have a job, particularly for recipients of non-family payments (e.g. using SIHC data the average exit RR is 63 per cent whereas it would be 59 per cent without RA). However, the disincentive effect is more pronounced for public housing tenants (e.g. using the SIHC data the average RR is 77 per cent compared with 68 per cent without RA).

Hulse and Randolph (2004) report the results of interviews of public and private renters in Sydney and Melbourne. A significant minority of public renters saw increased rents as a major cost of entering paid employment. In contrast, there was no indication that any loss of RA featured in the job decisions of private renters.

Discrimination
Participants said that discrimination is prevalent in the rental market. It is useful to distinguish between two types of discrimination: discrimination on "past history" and "profiling" (known as "statistical discrimination" in the economics literature). An example of the former would be discrimination against a person who has in the past defaulted on tenancy obligations. An example of the latter would be discrimination against a person simply because they are Indigenous. The distinction is relevant because discrimination of past history is in some sense a product of an individual's own actions. Discrimination in the form of profiling definitely is not.

It is not surprising that landlords make use of easily observed characteristics to make risk assessments. The phenomenon exists because it is costly, and sometimes even impossible, to make detailed assessments of tenant risk at the individual level. In lieu of detailed assessments landlords often fall back on "rules of thumb".

For instance, a landlord's past experience may lead her to think that an employed professional is less likely than an unemployed person to default on tenancy obligations. When faced with a choice between a professional and an unemployed person as a tenant, the landlord could conduct a detailed investigation of the unemployed person's reliability as a tenant. But it is costly to do that, and the landlord probably would not. The more likely outcome is that the landlord accepts the employed person. Discrimination of this type, commonly referred to as "profiling" is not a product of malice, rather it is a hard-nosed commercial decision.

While profiling may be rational and without malice, it is of course still potentially tremendously unfair. What it means is that people are judged not on their own past behaviour, but on the past behaviour of other people who a landlord thinks they are "like". In many areas of life profiling is explicitly prohibited. For instances, insurance companies are precluded from charging different premiums to people because of gender, age, race, etc. There are even prohibitions against differentiating premiums in respect of some characteristics that could be regarded as elective - for instance private health insurers cannot charge smokers more. Insurance companies tend to offer policies with standard wording and prices which are open to all comers which makes it easy to monitor and limit the extent of profiling. But rental housing is not offered in this way. Indeed many landlords own just one or a few properties, so transactions are one-off and it is hard to clearly identify any discriminatory treatment. This makes it very difficult to stop profiling in the rental market.

One traditional solution to discrimination has been the public housing system. Homeownership is also a potential solution, albeit one that is not available to people who are unable to afford entry to homeownership.

4.2 Government housing programmes

Lack of an overarching framework
The absence of an overarching, national housing policy framework was often commented on. The development of such a framework would be a useful step. At the very least it would serve as a reference point when considering policy options.

Ideally an overarching framework would identify the horizontal equity consequences of existing government policies. There will always be some debate about what parameters to admit in an analysis of horizontal equity, but a useful starting point would consider the differential levels of support across different tenure types.

Support comes in a variety of forms: concessions on income and capital gains tax, government provision at sub-market rates (with implicit subsidies entailed) and cash transfers.

The following factors would need to be considered:

• The income tax treatment of a homeowner differs from that of a landlord in that: the value of the implicit rental stream is tax-free; capital gains are exempt from gains tax; mortgage interest deductions are not available. The value of these exemptions will generally be greater for individuals on higher incomes because the concessions are received at higher marginal tax rates. They will also be greater the higher the value of the property occupied. Wang, Wilson and Yates (2004) show that the extent of concessions is greatly influenced by mortgage status (based on 1999 figures outright owner households had an average $3,200 concession compared with negative $300 per year for mortgager households).

• Homeowners also receive other tax concessions not generally available to landlords such as land tax exemptions. There may also be concessions embedded in the states' stamp duties applying to housing transactions.

• As far as GST is concerned the implicit rent of owner occupiers is notionally "input taxed" and this is also true for most private rentals, including caravan parks.

• Upon entering homeownership for the first time there is a widely available "First Home Owner Grant" which could be conceived of as capitalised affordability support.

• There is an implicit subsidy to public housing tenants. In contrast to the tax concessions available to homeowners, implicit subsidies for public housing tenants are negatively related to income.

• Some private renters are eligible for RA (Wang, Wilson and Yates report average assistance of $2,470 per assisted household in 2001-02) while some get none. There will be a negative correlation between RA and income. However, the "fit" will be highly imprecise. People with like household and dependent circumstances, and with equal incomes and rents, will get quite different RA support.

• Some specific programmes also provide a significant measure of support to some relatively small populations - viz support to homeless people via SAAP and support to people in aged care residences. While these programmes may be small in the scheme of things, they cannot be treated as peripheral in a holistic framework as they are of great importance to the affected customers who may not access other support programmes.

Significant parts of the data that would support such a framework are collected under the CSHA and analysed by the Australian Institute of Health and Welfare and its associated researchers. Their work paves the way to study how the tenure-dependent character of housing supports leads to breaches of the horizontal equity principle. It would then be interesting to ask why cross-tenure variations exist. Can they be rationalised on incentive and efficiency grounds? Or are they simply a result of administrative and political expediency? The answers to these questions could be extremely useful to make the different arms of housing policy work together better, including in targeting resources to areas of highest priority.

It is interesting, for example, to consider the tax concessions afforded to owner-occupiers. Table 4.1 shows the annual tax concession available to the owner-occupier of a $337,000 dwelling.⁹ The value of the concessions can be determined by calculating the income tax that the owner would be liable for if owner-occupied housing were subject to taxation of yields and capital gains with deductibility of mortgage interest rates. The value of the concessions depends on both the equity in the home and the owner-occupier's marginal tax rate. It can be seen that the concession is larger (a) the greater the degree of equity in the home and (b) the higher is the individual's taxable income. ¹⁰ For a homeowner with an income of $50,000 and a 50 per cent equity stake, the concession is worth about $2,000 per annum. A homeowner with an income exceeding $95,000 and 100 per cent equity receives a concession of about $8,700 for the same house.¹¹

* Calculations assume property value of $337,000, 4 per cent rental yield, 3 per cent average annual capital gain taxed on an accrual basis at half the marginal rate, and mortgage interest rate of 7 per cent.

While an externality argument for different treatment of different tenure types might be put forward, as in Gans and King (2003) in respect of homeownership, it is hard to see why that concession would vary according to income levels and equity levels. It is not clear how the vastly different rates of concession seen in Table 4.1 cannot be justified on welfare grounds.

Role of public housing
On several occasions participants suggested that there should be an expansion of the public housing stock. But this view is not universally shared - for instance Gans and King (2003) are critical of public housing as an assistance measure on the grounds that it denies recipients of housing assistance a degree of choice about what housing to access. In principle a government agency could make the optimal housing choice for an individual but in reality that is unlikely to happen because it cannot have perfect knowledge of an individuals' priorities and preferences. Devolving the decision to individual households may significantly improve outcomes. For instance, public housing tenants who need to move may find that their options are confined to the relatively small pool of vacant housing stock in the public system. The benefits of a wider range of choices could be quite significant (they might, for instance, mean substantially reduced commute times to work or getting a child into a preferred school).

This raises the important question of what it is that separates public housing from private rental housing.

One important difference is that the implicit subsidies to public housing tenants are in some instances more generous than the subsidies available to tenants in private rentals. This in itself would not seem to constitute a very strong ground for public housing in preference to private housing - there could be some harmonisation of the affordability assistance offered in public and private housing. The actual level of the assistance could then be addressed on its merits and applied synchronously to the two tenure types.

But the arguments for public housing go beyond this. Two important additional features of public housing are that (a) it provides greater security of tenure to tenants and (b) it can be operated in a non-discriminatory fashion to assist those who are likely to be precluded in mainstream housing markets.

It is possible that private markets could provide more security of tenure, for instance by offering longer term contracts. There might be more inclination to do this if there was a greater presence of portfolio investors in the rental market. For small investors it is possible, but by no means certain, that long tenure arrangements are seen as adversely affecting the marketability of a property. It is probably also relevant to take into account the ways in which long term leases bind. Public housing tenancies are beneficial to tenants in that they give the tenant security without requiring the tenant to commit to a long term lease. In private rental markets, on the other hand, long term leases would typically involve two-way commitments, and would typically involve higher rentals where the tenant reserves the right to terminate on short notice. In this sense public housing offers a concession to the tenant beyond what is revealed in a straight comparison of rents.

It is possible as well, in principle, that governments could address discrimination in the private rental market. One approach would be to subsidise the rent payments of groups subject to discrimination, effectively paying a premium for certain kinds of tenants. But this would be problematic. Choosing the right premium would be difficult and there would also be potential political difficulties in setting different subsidy rates according to factors such as race. A much more practical approach would be for government to act as guarantor to private rental tenants. A uniformly available guarantee would provide more (implicit) assistance to those who need it more, without government needing to make assessments or manage differentiated rates of assistance.

Guarantees are probably important to an increasing part of the community. According to Powall and Withers (2004), Australians' employment arrangements have become less secure. For instance, between 1993 and 2003 the proportion of Australian employees who were in casual employment arrangements increased from 22.7 per cent to 27.6 per cent (Parliamentary Library 2004). While this sort of labour market flexibility may have benefits to some employers and workers, it is unlikely to be viewed positively by landlords.

The possibility should also be admitted that governments may wish to override the choices of individuals on externality grounds. For instance, in South Australia the public housing authority for many years had a policy of providing public housing in desirable neighbourhoods to avoid concentrations of disadvantaged tenants in areas where they are effectively disconnected from other parts of the community. The operation of individual choice in private markets cannot be expected to produce this result.

Finally, it should be remembered that in some areas there simply is no deep private rental market, and in those cases the private rental market would not appear to offer any advantage in terms of diversity of choice. It is possible that private development could occur if a more supportive institutional environment emerged but government would need to take a much more interventionist role in the private rental market than it currently does - for instance offering guarantees. Public housing would seem still to have an important default role until, if ever, a satisfactorily functioning market solution can be found.

Policies to promote supply
Participants frequently mentioned "supply-side" and "demand-side" interventions. The distinction is set out in Gans and King (2003) in the following terms:

Supply-side policies aim to directly influence the supply of low income housing while demand-side policies aim to influence this supply indirectly, by increasing the ability of low income households to afford housing. Thus demand-side policies aim to improve low income access to housing in the short run and increase the supply of low income housing as the market reacts to the changed demand in the long term.

As Gans and King note, supply- and demand-side interventions can have the same effects. Yet in our discussions there was a common presumption that supply- and demand-side policies would have different effects. Very commonly this was due to differences in the targeting of the policies. For instance, a subsidy to providers of low cost housing should have similar effects to a support payment to occupiers of low cost housing. But it will not have the same incidence as a support payment confined to low-income occupiers of low cost housing. The difference really arises from the targeting regime more than the side of the market on which the intervention occurs. When comparing proposals for intervention in the housing market it is useful to keep the matter of targeting to the fore: Who is being assisted?

A range of measures were proposed to increase the supply of low cost rental housing such as:

• tax concessions to landlords;
• measures to increase flexibility in the allocation of the existing housing stock;
• development quotas;
• a more generous RA scheme; and
• a more certain RA environment.

Wood, Watson and Flatau (2005) simulate the impact of a tax credit for investors in low cost rental housing. What they model is similar in substance to the Low Income Housing Tax Credit (LIHTC) scheme that operates in the US, but with the important qualification that they do not incorporate an allocation mechanism to direct low cost housing to low income families (whereas the US scheme does incorporate this). The simulations find that among the poorest 20 per cent of Australian households, the scheme reduces the incidence of housing stress from 77 per cent to 72 per cent. The authors note that better results could be had if there was also some targeting of the low cost stock to low income households. Their analysis also highlights interactions between supply- and demand-side policies: by causing a moderate decrease in rents, the LIHTC actually reduces RA payments, and this is a budgetary offset.

The work of Wood and his co authors shows just how important targeting mechanisms are in understanding the impact and cost of interventions in rental markets (the point is also made in Yates 2002). It seems likely that many of the differences between demand-side and supply-side policies actually arise because of differences in the targeting regime bundled with each. It is important to know this because it may mean that policy comparisons of affordability measures are to some extent comparisons of different targeting regimes. Of course affordability is not the only relevant dimension of the supply- or demand-side choice - tenure types, discrimination, provision of suitable stock are also important relevant factors.

One of the arguments put forward by participants was that there would be advantages in promoting "portfolio" investment. While this argument may have some substance it needs to be carefully considered. Firstly, the idea was sometimes bundled with community housing, i.e. an NGO running a portfolio of properties. Where that is the intended model, the supply benefits that arise may be more to do with the concessional nature of the cooperative's rental policies than with a portfolio model per se. Secondly, if one considers the case of a portfolio operating to a purely commercial model, the question arises whether it would behave differently from a small investor. While it is true that a portfolio can diversify risks, and therefore reduce exposure to large adverse events, diversification cannot totally remove bad risks.¹² A portfolio may be more willing to "take a chance" but will still be motivated to avoid bad risks and will expect to be compensated when it takes them on.. And while it is possible that portfolio investors could find new ways to deal with bad risks, their options are limited. For instance, requiring larger bonds is often prevented under legislation. Thirdly, incentives to portfolio investment would seem more likely to have an effect if accompanied with incentives to rent to people in housing need. But in that case it might be just as effective to provide incentives to rent to the needy but remain neutral on whether investment is carried out by individuals or channelled through portfolios.

It was suggested in the consultations that "empty nesters", whose children have left home and who therefore find themselves in houses larger than they need, should be subsidised to move. Any consideration of this issue should also consider the extent of any impediments from taxes such as stamp duties on property transfers. This issue is very problematic because these stamp duties are one of the few ways that exist to extract tax contributions from owner-occupier property owners in line with services provided to them. However, they are much inferior to consistently-applied taxes on property, such as land taxes with minimal exemptions, mainly because they tax transactions in property rather than the property itself. They are a tax on moving, rather than a tax on property.

Extension of RA
One of the difficulties with any widespread extension of RA is the budgetary costs. It is possible that with a more clearly enunciated housing framework it would be possible to reallocate resources from other forms of housing assistance.

A second concern that may arise is that enhanced RA would flow through into higher rents. To believe that an increase in RA would be dissipated in higher rents and higher prices one would have to believe that:

1. the enhancement was large in the scheme of things, and
2. the supply of housing is inelastic.
   

Supply elasticity is likely to vary for different sorts of housing. Supply will be inelastic for old, run down stock which is amenable to renovation. Its supply is, definitionally, inelastic - it is not possible to make more old houses. So if there is a surge of interest in occupying residences of this type, then rises in prices can be expected. But the supply of new dwellings is likely to be more elastic. This then leads to the question whether there are, more generally, certain geographic areas in which the supply of housing is inherently inelastic. The answer to this question must be that, in general, there are not physical constraints which make it overwhelmingly difficult to increase the housing stock. By and large it is possible to expand the housing stock within geographic areas by introducing higher-density housing. However, if high density housing is more costly to build it may be accompanied by a permanent ratcheting-up of housing prices. Forces like this may be at work over the medium term for those areas of major cities which have experienced urban consolidation (e.g., Sydney and Melbourne).¹³

Planning policies have a potential impact. To the extent that local planning policies impede increases in urban density, supply will be less elastic and a greater flow through to rents and prices could be expected. In some rapid-growth areas there are planning policies to impede the growth of the housing stock. These policies might be implemented on grounds of infrastructure burdens or in terms of the urban amenity of the extant community.

Planning policies may also have an impact via selective development policies. If councils are not enthusiastic to have low cost housing, and use development controls to discourage it, then the consequences are potentially significant. If low cost housing is associated with households with social difficulties, there is a potential for "not in my back yard" factors to influence development policies.

In our discussions little was said about the impacts - positive or negative - of local government.

Last resort assistance
There is a complex range of government programmes operating in the housing market. Those policies, and the environments in which they operate, will change over time, sometimes with consequences that were not predicted. It is important to pay careful attention to the caseloads of "last resort" programmes, such as the SAAP, and for government to be ready to respond rapidly to changing demands on such programmes. In addition, careful attention should be given to "unmet demands" at a point in time. This would involve analysis of the characteristics of demands that were not met, the reasons that they were not met, and consideration of whether they should be met.

4.3 Targeting of RA

National Shelter and ACOSS (2003) summarise RA's performance in delivering affordability in the private rental market. They report that over 35 per cent of RA recipients spend more than 30 per cent of their incomes on rents, that the RA programme does not extend eligibility to many low income workers and students and thus does nothing to deliver affordability to these groups, and that the numbers affected are large". Australian Bureau of Statistics (2005) reports that among the whole population of private renter households in Australia in 2002-03, 32 per cent spent (541,000) spent more than 30 per cent of their income on rent. 11 per cent (188,000) spent more than 50 per cent of their income on rent. In discussions it was pointed out that RA is also available to many who are not in affordability stress.

It should of course be kept in mind that there is no indisputable standard for calculating housing stress. In fact there are degrees of housing stress. In discussions there seemed to be very great weight put on a 25 per cent benchmark, possibly because this is the standard ratio used in public and community housing.

Wood, Forbes and Gibb (2005) examine the incidence of housing stress in 1996-97, comparing public and private rental tenants using a range of stress measures. They note that public housing rental formulas around Australia typically set rents no higher than 25 per cent of household income. That being so, one might expect to find no public housing tenants in stress using a 25 per cent benchmark. However, there is some variance between states and territories in the income definitions used and there are some unusual consequences arising in multiple income unit households. Moreover, Wood et al produce an "equivalent income" measure which is better suited for comparison of different household types. As a consequence they do report some housing stress in the public housing population.

Table 4.2 summarises some of their results. Using a 25 per cent stress benchmark, and after subsidies, 31 per cent of private tenants are in stress compared with 37 per cent of public tenants.¹⁴ But looking at the bottom of the income distribution for the two renter groups, a much higher proportion of private tenants are in stress than public renters (67 per cent versus 33 per cent). This result is consistent with the targeting arrangements that are in force. Public housing subsidies, other things equal, are larger for low income households. In private rentals, on the other hand, subsidies are not tied closely to income. At higher-level stress benchmarks a similar theme emerges - on average private tenants are no more stressed than public tenants but private tenants on low incomes are much more stressed.

* Calculations for equivalent income are reported.

Wood and his co-authors also consider the "targeting error rate" for RA. The logic of the measure is that if RA were perfectly targeted then every household with rent exceeding 25 per cent of income (or some other such benchmark) would get RA and no house with rent less than 25 per cent of income would get RA. Using the 25 per cent benchmark, the target error rate is 24 per cent (17 per cent of households do not get it, but should, whereas 6 per cent do get it but should not). The error rate is a bit lower using a 30 per cent benchmark. It is higher using a 40 per cent benchmark, particularly in terms of households in stress but not being assisted.

It was often said that RA is not well targeted to households in housing stress. Figures like those above certainly bear out that contention. There are ways in which RA targeting could be improved:

• An income element could be introduced to the entitlement calculation, as happens with public housing assistance for instance. But this has the drawback that it introduces potential work disincentives. There is a difficult decision here as to what is the right balance between ameliorating housing stress and introducing potential work disincentives.
• Another possibility would be to extend RA entitlement more broadly. As it is, "stressed" households that are not eligible for income support cannot receive RA.
• The maximum rate of RA could be regionally differentiated. It seems likely that a reallocation of RA to direct more into high cost markets would reduce the extent of housing stress.

As noted by Hulse (2002) Australian income security policy has tended to emphasise equal treatment across the country. Such frameworks may allow some regional variations in relevant outcomes. To illustrate, there is a rent range over which RA varies. If "fair market rents" in different regions fell within this range, then the RA scheme might be said to allow for regional variations in costs. People in cheap regions would have rents at the low end of the range and smaller RA payments while people in expensive regions would have higher rents and higher RA. But in fact many people are on the ceiling rate of RA. The capacity of the current structure to produce RA outcomes sensitive to regional cost differences would seem to be quite limited.

This in turn raises the question why the 75c in the dollar range exists at all. It would of course be possible to have a flat rate of RA (differentiated by household type) for each housing market. Renters could then go into the market knowing that they would keep any rent savings that they managed to make and would incur the full costs of any excess rent that they accept. As it stands, there is not much incentive to try to save rent over the 75c in the dollar range (the tenant keeps just 25c in the dollar of any saving). This may contribute to some clustering at the rent ceiling.

There are two possible rationales for the 75c in the dollar element. Firstly, if one thought that rent variations within regions were largely out of control of tenants, then one might choose to compensate for them. Secondly, one might think that household is some sort of "merit good" deserving of subsidy. After all, owner occupier exemptions are correlated with the value of the housing too.¹⁵ Just how strongly these reasons support the 75c in the dollar subsidy is an unresolved empirical question.

The first of these two reasons may have a bearing on the issue of regional variations in rent. If one starts from the assumption that peoples' locations are a given, then it seems clear that there are inter-region variations in rent that are unavoidable. The question then is to what extent RA addresses those variations and whether it should. An argument can be mounted that it is horizontally inequitable not to: as it is, for renters in like degrees of housing stress, RA provides a degree of relief that varies significantly across regions. (See Johnston 2002 for further discussion.)

If one does not take location decisions as a given, then a case against regionally based rates of RA could be made along the lines that it is not desirable to promote relocation into what are already high cost markets. This rationalisation sits uneasily alongside policies surrounding unemployment assistance which penalise unemployed people who move to areas with poor employment prospects. However, policy could be constructed on the basis that (a) RA strives to be neutral over location choices and therefore not to exacerbate pressures in the housing market, and therefore does not vary across markets, and (b) unemployment assistance is targeted according to job prospects. In this view each policy is allocated to a single objective. Unemployment assistance may lead to different effects on different housing markets, just as industry location choices, technological changes and lifestyle choices do. An implication of this view is that horizontal inequities (which would appear to be quite large) are simply allowed to persist. In this context it is interesting to note recent comments by the Governor of the Reserve Bank of Australia, Mr Ian McFarlane (2005):

In dollar terms Sydney is still way more expensive than anywhere else in Australia. I think it is so expensive that it is in the interests of people, particularly a lot of young people, to go elsewhere to where their lifestyle is more affordable.

It was argued that in high cost locations it is difficult to attract key workers such as teachers, nurses, police, etc. and that housing policy should seek to address this (the "key worker argument"). It needs to be recognised that the adoption of such an objective is a significant qualitative departure from the objectives that are usually cited in the context of housing assistance. It would mean that, instead of targeting affordability and access across the board, housing policy would be assigned the role of assisting employers to save them the cost of offering higher wages or housing assistance to attract key workers. This potentially means transferring resources from people with housing stress to the employers of key workers. Of course if housing policy maintained an emphasis on providing assistance in correlation with housing stress, this would not preclude key workers having access to affordability supports on the same terms as "non-key" workers, or indeed non-workers.

One issue that was frequently raised was the failure of Indigenous Australians to access RA. This may be because they have a substantial presence in community housing, where rents are highly concessional, and indeed to such an extent that tenants do not qualify for RA or qualify for only small amounts of RA. With this in mind, it is arguable that affordability assistance is delivered, but simply delivered via different channels (i.e. non-Indigenous people are more likely to get assistance via RA and Indigenous people are more likely to get assistance via community housing). This is not to deny that some special response to the plight of Indigenous people is required - in our view it is the deepest failing in the current policy mix - but rather that something more than conventional affordability support is needed.

⁷ The replacement ratio is the ratio of the income support received without employment to the income received in employment. The higher is the ratio, the less is the gain from taking a job, and hence the less the incentive to take a job.

⁸ There are some sizeable differences in the RR results according to whether they are calculated using HILDA or the ABS Survey of Income and Housing Costs.

⁹ This was the median first home price in the June quarter 2005 (HIA-Commonwealth Bank 2005).

¹⁰ The comparisons here are over the same house - i.e. the median first home. In reality high income earners occupy more valuable homes than low income earners and the scale of differences in tax concessions will be much greater than what is shown here. However, confining attention to the median first home owner sets aside the complicating issue of whether externalities are larger when more valuable homes are held by owner-occupiers.

¹¹ The comparisons here are at a point in time. More robust comparisons could be had by using a lifetime analysis, as is done for indirect tax exemptions in Cameron and Creedy (1995), but the data compilation is then considerably more complex. One could predict that considerable differences would still exist between people who rent for their whole life and people who buy homes (especially those who pay them off early in their lives). However, the contrast between recent home purchasers with low equity and long-time home owners with high equity would be less relevant in the lifecycle framework.

¹² The point can be illustrated with an example. Assume there are two types of tenants, "good" tenants and "bad" tenants, and that the landlord expects to receive a rent of $4,000 over a coming six-month period. It is known that with a good tenant the landlord will definitely get his $4,000. But with a bad tenant there is only a 50 per cent chance that he gets the $4,000. For a landlord with just one tenant the "expected" return (in a statistical sense) is $2,000 but there is a high degree of uncertainty about what will be received - it may be zero or it may be $4,000. Thus the bad tenant has two undesirable features to the small landlord. The first is that the expected return to the landlord is just $2,000, compared with $4,0000 with a good tenant. The second is that the return is uncertain. Now with a portfolio approach the uncertainty can be diversified away, but the low expected return cannot. With a large portfolio, say a portfolio with 100 "bad" tenants, there is a high chance that about 50 will default and 50 will not. Thus the return has a reasonably degree of certainty - on average about $50 per tenant. But this is still much inferior to the $100 received from a "good" tenant. The portfolio approach diversifies the uncertainty, but it does not correct for the lower expected return from a "bad" tenant.

¹³ Associated with it there has no doubt been a surge in the value of underlying land parcels (which are in fixed supply).

¹⁴ The figures for public housing are surprising, as public housing authorities usually set rents with something like a "25 per cent of income" rule, in which case there would be no cases of stress using the 25 per cent benchmark. The authors discuss the anomaly and conclude that it relates to differences between the income measures used by state housing authorities and the income measures in the authors' dataset. One important factor is that the figures cited here relate to an "equivalent income" concept that adjusts for household size and composition. If one uses more straightforward income measures the degree of stress among public renters is much smaller: 20 per cent of public tenants spend more than 25 per cent of their “gross income” and 22 per cent spend more than 25 per cent of their "disposable income".

¹⁵ The valuation of tax exemptions for owner occupiers is not entirely straightforward. Account needs to be taken of the facts that: the implicit rental income stream is not taxed at the owner-occupiers marginal tax rate; mortgage interest payments are not deductible; and capital gains are not taxed at the normal rate (which is half of the marginal rate). Moreover, the marginal rate depends on the circumstances of the owner. The value of exemptions is likely to correlate more closely with housing equity than with housing value, but there is still likely to be a correlation with housing value.