Australia’s economy is giving off mixed signals: Even as GDP growth and income slow, household debt appears to be rising. Understanding what is drivinghousehold behavior has important implications for policy and interest rates.
During Australia’s most recent economic upswing, national income outpaced output, but with larger-than-anticipated commodity price drops over the pastyear, this trend has reversed. This is apparent on a per capita basis, with net national disposable income actually falling, as seen in Figure 1. At thesame time, the main macro variables of unemployment, private final demand, wages, inflation and confidence have all weakened. (For a discussion of how theAustralian terms-of-trade cycle is now weighing on nominal growth and national incomes, see Australia Credit Perspectives, August 2014.)
The Reserve Bank of Australia (RBA) has tried to cushion these macro headwinds by lowering the cash rate to historically low levels but is now faced withthe risk of over-stimulating the housing market. Indeed, common metrics of household leverage and house price valuations remain elevated in Australia,particularly compared with those in other developed markets (see Figure 2).
How household leverage evolves will have implications for policymakers and the potential path of interest rates. A positive response to leverage in theface of weakening fundamentals may increase the economy’s vulnerability to a more severe downturn. This complicates the RBA’s reaction function whensetting policy by introducing financial stability risk. Therefore, it is important to understand what is driving household debt and how it may evolve.
The evolution of Australian household leverage
The run-up in debt at the household level was incurred during a very buoyant period when nominal growth and wages were high, as seen in Figure 3. From2000–2008 debt relative to income accelerated from 80% to 130%, while nominal growth averaged nearly 8% and wages grew at 4% annually.
More recently, the end of the commodity cycle has resulted in a larger-than-expected negative income shock. In fact, over the past three years, nominalgrowth and wages have only averaged gains of around 3% per annum, with the trend continuing to fall. If this downward move persists over the medium term,as expected by the RBA, household expectations could start to deteriorate, leading to a slow but persistent need to decrease leverage. As a result,consumers could increase savings, and the response to low interest rates may be more subdued than history would suggest.
However, it is also possible that lower mortgage rates and asset price appreciation may drive a more irrational response relative to the macrofundamentals, a “herd mentality” that assumes continued capital price appreciation and creates the fear of “missing out.”
The role of debt
Households often decide to incur debt to finance assets that will provide a future expected payoff and maximize their net worth. Therefore, the level ofdebt incurred is typically tied to:
- expected future capital gains in the asset
- expected future income related to the productivity of the asset
- debt-servicing costs
Should expectations rise for future capital price or income gains and/or should debt-servicing costs fall, households may increase their desired levels ofdebt. The problem is that expected future capital price gains may become unrealistic when extrapolated from recent history (“irrational exuberance”),leading to excess borrowing, which, when unwound, can lead to large negative externalities.
Rather than commenting on the appropriateness of any given debt level, we have focused on what is driving the change in debtmetrics. Due to the importance of household debt ratios in consumption, which is ultimately what policymakers are concerned about, we use debt-to-income asour preferred measure of household leverage.
Australian housing in an international context
Given the importance of expected future capital gains in household behavior, it is worthwhile putting Australian house prices in an international context.Based on data from the OECD (2014), two commonly used valuation metrics place Australian house prices at the higher end of international comparisons, asseen in Figure 4. With this starting point, it seems questionable to embed expectations of continued high price growth.
Analyzing household behavior
In a recent speech (April 2015), RBA Governor Glenn Stevens touched on the vulnerabilities created by household leverage: “Household leverage starts from ahigh level… the extent to which further increases in leverage should be encouraged is not easily answered.“ He described dwelling prices as having “alreadyrisen considerably from their previous lows, at a time when income growth has been slowing,” adding that “it is hard to escape the conclusion that Sydneyprices – up by a third since 2012 – look rather exuberant.” He also acknowledged the importance of financial stability issues: “The conduct of monetarypolicy can’t allow these financial considerations to dominate the ‘real economy’ ones completely, nor can it simply ignore them. A balance has to befound.”
Given the increased weight the RBA places on these vulnerabilities when setting monetary policy, analyzing what drives household debt is highly relevant.
To determine which factors drive household leverage, we statistically tested the response (the change) in household leverage against changes in several keyvariables including wages, unemployment, asset prices, confidence and the cost of credit.
We first tested the explanatory variables individually to see whether they are significant, and from that subset, we tested them together to see how theyinteracted and whether some variables dominate others. Importantly, we also tested various lags in these explanatory variables to proxy for expectations.That is to say, individuals’ expectations and responses may change under a persistent set of conditions. So, for example, a quarter of strong wage growthmay not be sufficient to shift expectations, but over several quarters, strong wage growth may result in an increase in leverage as expectations for futureincome are revised higher. (An upcoming paper, “A Model of Australian Household Leverage,” will provide a detailed explanation of our methodology.)
The modelling results reveal that asset prices and mortgage rates largely explain the change in household leverage. These two factors appear to dominateall other variables regardless of how significant they are individually.
From this we infer several probable outcomes in Australia:
- Households’ decision to incur debt is dominated by the cost of debt (the mortgage rate) and recent asset price appreciation, which may not besustainable or linked to the productivity of the asset.
- Of concern, households are exhibiting irrational exuberance because they are placing little weight on broader fundamentals like unemployment that may bemore representative of future incomes or asset price returns, increasing the likelihood of asset price bubbles.
- Australian households appear to respond rather quickly, needing only two quarters of favourable changes in asset prices and mortgage rates to increaseleverage.
- Australian households will react faster and more vigorously to a shock in asset prices or mortgage rates. This could result in a feedback loop wherefalling asset values induce further deleveraging.
Based on our model for household leverage alone, if an exogenous shock sparked a deleveraging cycle in Australia, it would be expected to be quite severegiven the larger co-efficient for asset prices and the quicker household response. Clearly other factors not captured in the model may mitigate such asevere outcome, and no model is perfect, but as a starting point this statistically highlights the potential vulnerability of high and rising householdleverage incurred on the basis of past asset price movements. We believe macro-prudential measures should be strengthened to address financial stabilityrisk and give the RBA maximum flexibility when setting policy for the aggregate economy.
Finally, given the sensitivity of households to mortgage rates, the peak in the cash rate in the RBA’s next hiking cycle is likely to be much lower than inprevious cycles. This is in accordance with PIMCO’s New Neutral view that calls for much lower policy rates for an extended period.