Filtering as a source of low-income housing in Australia: conceptualisation and testing

Publication Name

AHURI Final Report


Key points Filtering is a market-based process whereby the supply of new, higher quality dwellings, for higher- and middle-income households also leads to additional supply of dwellings for lower income households as higher income households vacate their former homes. Theoretically, dwellings filter over time through successively lower market segments or sub-markets, becoming a supply of ‘naturally occurring affordable housing’. This report tests for filtering dynamics in Melbourne and Sydney, using differently sourced data and different methods of analysis. Theoretically, filtering is traceable in the price dynamics of dwellings, and in the occupancy characteristics of residents in properties over time. Filtering dynamics are tested across the entire housing stock and in the private rental sector specifically. Overall, there is evidence that new dwellings – whether sold or rented – trade at a market premium. This is consistent with filtering dynamics. However, there is little corresponding evidence that these age-related price dynamics are sufficient to generate a supply of affordable housing for low-income households. Under the prevailing housing market and planning conditions, the evidence does not support relying on filtering as a substitute for the non-market provision of affordable housing for low-income households. The results show that the successive transfer of housing to lower income households as properties age (downwards filtering) may be negated or weakened by countervailing neighbourhood socio-economic determinants and locational characteristics (such as proximity to labour markets or urban amenities). While some properties are thus occupied by lower income households over time, the process is conditional on a number of locational and neighbourhood characteristics. For instance, the evidence from Melbourne suggests that many areas have largely retained their relative income status over the past 20 years. Supply is central to addressing the affordability of housing for low-income households. However, a key condition for market-based filtering to deliver affordable low-income housing is that the rate of housing supply exceeds the rate of household formation and new demand. In Melbourne, the price elasticity of supply at the neighbourhood level (Collection District) is almost zero. Additional demand is therefore not met by additional supply, resulting in price appreciation that potentially also generates additional economic rent for older and existing properties in high demand areas. The private rental market analysis focuses on two cohorts of rental properties: one-bedroom units and three-bedroom houses. The evidence shows that the rate of rental depreciation is insufficient to offset the overall increase in rental values, leading to worsening affordability for low-income households. The evidence from Sydney suggests that dwellings are removed from the filtering process before they reach lower income households. While additional research is required to establish and identify more detailed housing market dynamics, the results in this report suggest that filtering may be inhibited and/or interrupted by local housing market contexts such as social interactions, low price elasticity of supply and a lack of substitutability between dwelling typologies and locations. From an affordable housing policy perspective, the results in this report suggests that filtering cannot be considered in isolation from the remaining determinants of occupancy or price change. That is, new supply will not necessarily free up existing housing for lower income households, unless the remaining determinants of demand for housing in specific locations is also addressed; and new supply also results in sustained downward pressure on sub-market specific property prices (that is, the rate of supply exceeds the rate of new household formation and demand; and new housing constitutes effective substitutes for existing housing). Greater use of market signals in planning and regulation can assist in improving the responsiveness of new construction to market signals. However, importantly, supply dynamics are also determined by landowners and property developers that benefit from withholding new supply housing. Public land banking can contribute to overcoming some of the private sector based incentives that inhibit greater supply responsiveness. Key findings Filtering is a market-based process whereby the supply of new dwellings for higher- and middle-income households leads to additional supply of dwellings for lower income households. As higher income households vacate their former homes, lower income households move in. If new dwellings provide a higher level of housing services (quality) and demand for housing services is income elastic, then new dwelling construction may lead to older properties becoming cheaper. As properties become cheaper they, according to the filtering thesis, successively move down through lower quality market segments, ultimately providing affordable housing for lower income households. However, this rests on a set of assumptions, the most important of which are that: The supply of new dwellings exceeds the rate of new household formation (new demand). New dwellings provide effective substitutes for older dwellings. The quality of housing depreciates independently of the characteristics of the residents of the dwellings or neighbourhoods. Empirically, filtering is observable in the price and occupancy characteristics of dwellings. In the housing literature, property age is frequently taken as a proxy for the quality of housing services provided. This yields a number of stylised outcomes indicative of filtering. In this report, we test whether these stylised outcomes are evident in metropolitan Australia by using different datasets and analysis techniques for Melbourne and Sydney. Dwellings age and cyclicality in neighbourhood status Since housing is immobile, when dwellings filter it is actually the characteristics of occupants that change; higher income occupants vacate and lower income occupants move in. Therefore, filtering also leads to change in the relative income status of neighbourhoods. In the Melbourne analysis, relative income is measured as the median income at Collection District (CD) level (2006 boundaries), relative to the median income for the metropolitan area as a whole. Over the past 20 years or so (1996–2016), there has been a degree of stability in relative income levels across metropolitan Melbourne. Twenty years, however, is a relatively short period of time for the housing market. Over a longer period (since the 1970s), the social geography of Melbourne has changed more substantially. Overall, the relationship between change in neighbourhood relative income levels and the age profile of the housing stock at first declines and then rises again. In other words, areas with the newest and oldest housing tend to improve their income level relative to other areas (a u-shaped relationship). This is consistent with filtering. However, when examining the role of dwelling age on the change in relative income at the neighbourhood level, the analysis suggests that the impact of dwelling age is not independent of the socio-economic characteristics of occupants. That is, in neighbourhoods with higher levels of education or owner-occupiers, a concentration of older dwellings (pre-WWII) did not typically result in the neighbourhood filtering down market. A potential explanation for this persistence in the occupancy characteristics of higher (and lower) income areas are endogenous social interactions between residents in these areas. Social interactions refers to the relationships and behaviours between people. These can condition the demand for specific locations when behaviour and interaction give rise to externalities, such as social capital (positive) or costs (negative), disproportionately to the concentration of the specific social characteristic. In this case relative dwelling prices may exhibit persistence also following new dwelling developments. Dwelling age and relative price depreciation In both Melbourne and Sydney, the evidence shows that new properties trade at a market premium. New properties for sale in Melbourne are typically more expensive that older properties. However, the relationship between dwelling age and transaction price is not straightforward. The relationship depends upon, and interacts with, local housing market context, locational and institutional determinants of demand. In Sydney, the results also show that the rental price for new dwellings is higher than comparable older dwellings. This is the case for both three-bedroom houses and one-bedroom flats. However, the relationship between age and rental price differs between these two market segments. Rental prices for three-bedroom houses have a u-shaped relationship with age, whereas rental prices for one-bedroom flats shows a near linear rate of depreciation. However, in neither Melbourne or Sydney does the price premium for new dwellings appear sufficient to generate systematic supply of affordable low-income housing via filtering. In Melbourne, the improvement in relative income status is associated with a higher proportion of both new and old (pre-WWII) housing. However, relative income change is greater in areas with greater concentrations of older housing than new housing.

Open Access Status

This publication may be available as open access



Funding Sponsor

Australian Government



Link to publisher version (DOI)