RDP 2006-08: A Survey of Housing Equity Withdrawal and Injection in Australia 4. How was Equity Withdrawn and Injected?

Aggregate data on housing equity withdrawal provide little insight into how households withdraw and inject equity, and how widespread such activities are. This section provides results from the survey on these questions.

According to the survey, 42 per cent of households changed their housing equity over 2004; 12 per cent of households made a net withdrawal of equity over 2004, while 30 per cent made a net injection (Table 1). The remaining households neither withdrew nor injected equity, largely because they did not own any property, or owned their property outright.

By number, the bulk of households changing housing equity were non-transactors – 33 per cent of households versus 9 per cent that were property transactors. Around 7¼ per cent of households made a net equity withdrawal by increasing debt on their existing property; for these households, the median increase in debt over the year was $20,000, while the mean was considerably larger. A much larger number of households injected equity into their existing property, with 19 per cent of all households injecting equity through scheduled and additional payments on their housing loans, and a further 6½ per cent injecting equity through renovations. The median value of injections by non-transactors was considerably smaller than the median withdrawal made by non-transactors.

The finding that 9 per cent of households were involved in at least one property transaction in 2004 is broadly consistent with the available housing turnover data. These households were almost equally split between those withdrawing and injecting equity. However, the median change in equity resulting from these transactions was considerably larger than for non-transactors, such that property transactions contributed the bulk of the value of gross injections and withdrawals.

4.1 Withdrawals

Almost three-quarters of the value of all (net) withdrawals by households that were net withdrawers over 2004 were accounted for by those that engaged in property transactions (Table 2). Of the net withdrawals by property transactors, around three-quarters of the value was accounted for by the 2.7 per cent of households that sold more properties than they bought. This large contribution in part reflects the larger median withdrawal by such households – $125,900 versus $33,500 for withdrawals based on other combinations of property transactions. These other property transactions were fewer in number and smaller in value, but nonetheless remained significant as a share of overall withdrawn equity – accounting for almost one-fifth of the total value withdrawn.

Sales of owner-occupied property – which include last-time sales of elderly households' properties – appear to be associated with larger net equity withdrawals than sales of investment property. This is consistent with the finding that for those that sold more properties than they bought, the median LVR of owner-occupied properties sold was slightly lower than it was for investment properties (Table 3); this is not surprising given the tax advantages of interest deductibility for investment properties.^[5] This is despite the fact that the typical investment property had been held for slightly longer than the typical owner-occupied property, allowing more time to accumulate capital gains and pay down debt. Owner-occupied properties also tended to sell for more than investment properties and second homes, consistent with investment property being generally more concentrated in cheaper housing stock such as units (see Appendix D).

Of the households that sold more properties than they bought, 36.6 per cent sold their main residence. Of these households, around 40 per cent moved into rental accommodation; most of the rest moved into a property which they already owned. A small number of these transactions appeared to reflect just one leg of a transaction, with households either moving into a property that had been purchased in 2003 or planning to purchase a property in 2005.

Investors selling more properties than they bought appear to have typically been experienced property investors, with a median holding period of six years for the properties that they sold. Despite the sales, these households finished the year with an average of 2½ properties.

Of the non-transacting households that withdrew equity, by far the most common methods were to refinance an existing loan and increase the outstanding balance or to take out a new loan. Two other common methods were drawing upon previous excess principal payments or drawing on a revolving or home-equity type facility. Around 20 per cent of non-transactor households that withdrew equity also undertook renovations. The methods these renovating households employed to increase their debt were in similar proportions to the overall group, though the median amount these households withdrew was slightly larger at $22,500.

4.2 Injections

In contrast to the results for households withdrawing equity, for households that made a net equity injection over 2004, the value of injections was split fairly equally between non-transactors and transactors. This reflected a large number of non-transacting households making small injections by paying down debt or renovating, balanced by a small number of households making large injections through property transactions (Table 4).

Within the 19 per cent of households that injected equity by reducing debt on their existing property, 9.6 per cent reported that they simply made the regular scheduled repayments, while an additional 6.7 per cent made regular repayments above those required by their lender. A further 2.1 per cent indicated that they made irregular lump-sum repayments. These one-off lump-sum payments tended to be relatively large, so that they accounted for a disproportionately high share of the total equity injected.

Around 6½ per cent of households injected equity over 2004 through renovations, financed, at least partly, from their own savings. In total, this amounted to around 18 per cent of the total amount of equity injected by households that made a net injection over 2004.

Within the 4.6 per cent of households that injected equity and undertook a property transaction, most purchased more properties than they sold, accounting for the bulk of equity injected by property transactors. Over half of the properties purchased by this sub-group were owner-occupied homes (Table 5), with around 40 per cent of these purchased by first-home buyers. The owner-occupier purchases tended to be associated with more expensive properties and lower debt levels compared to those for other properties. These results are consistent with investors' preferences for relatively cheaper property and higher gearing mentioned in Section 4.1. It is worth noting that the LVRs on the purchased properties were significantly higher than the overall LVR of households undertaking these transactions (that is, these households often had other, less indebted, property holdings).^[6]

A comparison of the results regarding the methods of housing equity withdrawal and injection underscores the importance of transactions to overall flows of housing equity withdrawal. In particular, for the sub-groups of property transactors most important for overall housing equity flows, sellers typically withdrew more equity than buyers injected, partly reflecting much higher debt levels among buyers. This is consistent with the influences of life-cycle factors and house price gains discussed in Section 2 and further explored in Section 5. It also follows that shifts in the level of aggregate transaction activity will likely be associated with changes in the value of aggregate housing equity withdrawal, as canvassed in Section 7.

Footnotes

Valuations were provided by the household. However, we believe that our analysis is unlikely to be biased by subjective valuations for the same reasons described in Ellis, Lawson and Roberts-Thomson (2003). In addition, it may be that households' perceptions of their financial position are more relevant to our analysis than is their actual position. [5]

These LVRs are also higher than previous estimates (Coleman et al 2005). [6]