Research Discussion Paper – RDP 2019-06 The Effect of Mortgage Debt on Consumer Spending: Evidence from Household-level Data

The household debt-to-income ratio has risen to record levels in Australia in recent years, while household spending has been relatively weak (Figure 1). This follows a period of stagnation in household debt relative to income, which included the global financial crisis (GFC). A similar pattern of high household debt and weak spending has been observed across a range of other countries (Bunn and Rostom 2015; Pistaferri 2016; Lombardi, Mohanty and Shim 2017). This has led to concerns amongst policymakers that elevated levels of household debt are holding back the economic recovery and pose risks to future growth (Hunt 2015; Brazier 2017; Lowe 2017).

Supporting these concerns, international research indicates that expansions in household debt (relative to GDP) can increase the risk of financial crises and subsequently lower household spending (Schularick and Taylor 2012; Jordà, Schularick and Taylor 2013; Mian, Sufi and Verner 2017; Mian and Sufi 2018). This research has typically linked the decline in household spending to the balance sheet adjustments that tend to follow either widespread debt defaults or a tightening in bank lending standards that lowers the ability of households to borrow (e.g. Mian and Sufi 2010, 2018). Either way, the decline in spending is related to disruptions of the financial system and lower credit availability for households.

This makes Australia an interesting case study. Australia has seen a strong increase in household debt and weak spending over recent years despite a persistently stable banking system and reasonable economic growth even during the GFC. This suggests that a high level of household debt may weigh on spending even when the economy is in a more ‘normal’ phase of the business cycle.

So do high levels of household debt cause weaker spending? And does such a relationship exist in both an economic downturn and in more ‘normal’ times? We use a rich source of longitudinal household-level data to test whether higher mortgage debt causes lower household spending (which we refer to as the ‘debt overhang effect’). Our unique data allow us to explore the underlying mechanisms of any debt overhang effect by looking at whether financially constrained households or households with strong precautionary saving motives are particularly sensitive to debt in their spending decisions. Relatedly, we also test for the presence of financing constraints and precautionary saving behaviour by examining whether debt matters for households at all times or only when households experience adverse income or wealth shocks (which we label the ‘debt amplifier effect’).

Identifying the causal effect of mortgage debt on household spending is difficult. First, an increase in spending intentions can lead to higher mortgage debt if households withdraw home equity to support consumption (reverse causality). Second, some unobserved factors, such as an increase in income expectations, may lead to higher debt and spending (omitted variables bias). In both cases, it will be more difficult to identify a negative relationship between mortgage debt and household spending. However, there are other non-causal explanations that may lead to a negative relationship between debt and spending. For instance, higher mortgage debt and lower spending on non-housing goods and services could be due to an unobserved shift in preferences towards owner-occupier housing. Alternatively, weak spending and high levels of debt may reflect the return of spending to its normal level after previously high levels of debt-financed spending (Anderson, Duus and Jensen 2016).

From a policy perspective, it is important to understand and distinguish between these mechanisms. If high levels of debt cause households to reduce their spending, providing debt relief or easing financing constraints through lower interest rates or tax incentives may lift spending. In contrast, if weak household spending is instead due to strong debt-financed spending in the past or a shift in preferences towards owner-occupier housing, such policies may merely postpone a downturn. Moreover, for a policymaker, the distinction between the debt overhang and amplifier effects is important. If debt has no direct effect on spending but affects spending only when there are shocks to income or wealth, the main concern is about the resilience of the economy to such shocks. In contrast, a direct debt overhang effect may explain why household spending in Australia has been relatively weak in recent years despite a strong labour market and rising house prices.

We address the three identification challenges outlined above using longitudinal household-level information from the Household, Income and Labour Dynamics in Australia (HILDA) Survey. This survey is rare by international standards. It not only contains detailed annual information on consumption, income, housing assets and debt of a representative group of households over time, but also provides information on their expectations for future employment and debt, their liquidity holdings and their risk preferences. Moreover, since households are directly asked how much they spend each year, we do not need to impute expenditure like other papers in the literature.

Household-level data is imperative to deal with the challenges posed by externalities that mask any household-level effect of debt when using aggregate or regional-level data.^[1] Moreover, the richness of our data allows us to control for a wide range of observable factors in estimating the effect of mortgage debt on spending. By tracking households over time, we are also able to fully control for unobserved household characteristics that likely do not vary over time but may affect the relationship between debt and spending, such as the household's level of patience. In addition, the longitudinal nature of the data allows us to explore the link between debt and spending during the GFC and in more normal times. This is in contrast to the existing research which often focuses on the linkage between debt and spending during rare episodes such as the GFC.

To further alleviate concerns about endogeneity, we adopt an instrumental variables approach using detailed survey information on each household's home purchase history. This allows us to exploit cross-sectional differences between households in the timing and location of their home purchases. We then use this information as an instrument for the level of outstanding mortgage debt today. In this empirical strategy, households that live in the same area are exposed to identical local demand shocks, but differ in the amount of debt they hold based on when they bought their home.

Based on these identification strategies, we find strong evidence for the debt overhang effect. Estimates from our preferred specification suggest that a 10 per cent increase in debt reduces household expenditure by 0.3 per cent. Notably, we find evidence for this overhang effect when we control for either a household's gross or net housing wealth. The latter implies that households lower their spending even when the gross value of both their debt and assets increases by the same amount (that is, when net wealth remains constant). In other words, a deepening of household balance sheets is associated with less household spending, even if it is not associated with rising net indebtedness. This directly violates conventional consumption theories such as the permanent income hypothesis (PIH), which assumes that the composition of household balance sheets does not affect consumption (Garriga and Hedlund 2017).

We do not find that any specific mechanism, such as financing constraints or precautionary saving, is driving the effect; instead, it appears to be pervasive across all mortgage borrowers, even households that are unlikely to face financing constraints or have strong precautionary saving motives. The effect is also pervasive over time and across regions and persists when allowing highly indebted households to respond more strongly to individual or local unemployment or house price shocks than less indebted households. This suggests that our results are not exclusively driven by financing constraints or precautionary saving motives or reflect the presence of a ‘debt amplifier effect’. However, we do find that households are more sensitive to debt during the GFC and local house price shocks, which could suggest that financing constraints or precautionary saving motives play some role. Furthermore, we rule out some of the non-causal explanations. In particular, the debt overhang effect is evident when controlling for past spending, and also when total spending includes the consumption of (owner-occupier) housing services. This suggests that the debt overhang effect is not driven by spending normalisation or a shift in household preferences towards consuming more housing.

Finally, we use our household-level estimates to consider the potential implications of higher debt levels for aggregate consumption. Simple calculations suggest that the observed increase in aggregate mortgage debt since the GFC weighed on aggregate spending, and this debt overhang effect may explain some of the weakness in aggregate household spending since then. Specifically, we estimate that annual aggregate consumption growth would have been around 0.2 to 0.4 percentage points higher had mortgage debt remained at its 2006 level. However, these estimates abstract from other stimulatory effects of debt. The increase in mortgage debt has likely lifted house prices and by this also supported consumption over this period. Our estimates are thus best interpreted as the loss in consumption had all other trends, such as the growth in house prices, occurred even though debt remained constant. As a result, the net effect of the increase in debt since the mid 2000s is unclear.

The remainder of this paper is organised as follows. In Section 2, we discuss our contribution to the related literature. Section 3 provides an overview of the dataset used in our analysis. Section 4 presents the methodology and results for the debt overhang effect, while Section 5 aims to identify the mechanism behind this effect. Section 6 presents the results for the debt amplifier effect. In Section 7, we assess the potential effect of higher debt levels on aggregate consumption. Section 8 concludes.

The study of the effects of debt has a long history in both public and corporate finance. Since the GFC, there has also been growing interest in studying the effects of household debt. In a standard life-cycle model, households borrow and save to smooth their consumption over time and the outstanding stock of debt has no causal effect on spending decisions. Household spending depends on current net wealth, as well as current and expected future income. While debt is a component of net wealth, a change in debt can only have a wealth effect on spending if that change in debt is unanticipated and exogenous (Paiella and Pistaferri 2017).

But the assumptions of the standard life-cycle model may not hold. Most obviously, households may be restricted in their ability to borrow, have limited liquid wealth or face uncertainty about their lifetime income. The importance of borrowing and liquidity constraints as well as uncertainty for consumption has been brought to the fore during the GFC (Pistaferri 2016). To the extent that debt exacerbates borrowing or liquidity constraints or increases uncertainty about future repayment obligations, the composition of household balance sheets and, in particular, the level of debt can matter for spending. In Appendix A, we set out a simple model to illustrate the channels through which higher debt levels might affect spending, and hence explain the debt overhang mechanism.

Previous empirical cross-country evidence shows that expansions in household debt (relative to GDP) driven by (excessive) credit supply can increase the risk of financial crises and subsequently lead to lower spending and economic growth (Schularick and Taylor 2012; Jordà et al 2013; Mian et al 2017; Mian and Sufi 2018). While this research establishes a link between debt and spending at the aggregate level, research at a more disaggregated level can provide important insights into why debt matters for spending. Exploiting regional variation in the United States, Mian and Sufi (2010) find evidence for a debt overhang effect – the regions that experienced stronger pre-crisis increases in household leverage also experienced stronger post-crisis declines in household spending. Mian, Rao and Sufi (2013), on the other hand, provide evidence for the debt amplifier effect by showing that falling housing prices exacerbated the reduction in spending in regions with relatively high levels of leverage. Household-level data is key to further unpacking these effects and identifying the underlying mechanisms.

Our results complement and extend these findings along several dimensions. First, by exploiting rich longitudinal data at the household level, we are able to better identify the causal effect of mortgage debt than studies that use aggregate or regional data. By aggregating across indebted households, region-level data implicitly put more weight on the richer households that spend more and hold more debt. If the debt overhang is less prevalent amongst rich households, this will attenuate any debt overhang effect – a mechanism that applies at the household rather than at the region level.

Second, by studying the Australian case, we generalise the results to show that debt matters even during periods of financial and economic stability. Almost all of the existing debt overhang research at the household level focuses on the GFC period in countries that experienced strong falls in house prices and large increases in unemployment, including the United States (Dynan 2012), the United Kingdom (Bunn and Rostom 2015), Denmark (Andersen et al 2016), and New Zealand (de Roiste et al 2019).^[2] Evidence for the debt amplifier effect at the household level is also typically limited to periods of financial crises or recessions (Yao et al 2015; Atalay, Whelan and Yates 2017; Garriga and Hedlund 2017; Baker 2018). Cho, Morley and Singh (2019) provide further evidence that amplifier effects matter most during episodes of crises by showing that indebted households in the United States exhibited a considerably greater sensitivity to transitory income shocks during the GFC. While we find a similar pattern of stronger effects of debt during the GFC, we show that our results are not driven by this episode alone and that debt weighs on spending even in more ‘normal’ times.

Third, our rich household-level data allow us to inspect the causal mechanisms that link debt to spending. Only a handful of studies have done this. In contrast to these studies, we find the debt overhang effect to be pervasive across indebted households and find only indirect evidence for borrowing or liquidity constraints (Bunn and Rostom 2015; Baker 2018; de Roiste et al 2019; Cho et al 2019) and precautionary saving motives (Bunn and Rostom 2015; Fagereng and Halvorsen 2016) to be possible drivers. However, we can rule out non-causal explanations proposed by the literature. Specifically, our results contradict the finding of Anderson et al (2016) for Denmark that weak current spending can be explained by high levels of past (debt-financed) spending. Instead, we find that higher debt lowers spending even after controlling for past spending and borrowing. In contrast to other studies, we also assess the extent to which a shift in preferences towards housing services can explain the decline in spending, and find little evidence that it does.

Finally, we emphasise that the debt overhang effect persists irrespective of whether we control for gross or net wealth. The latter implies that households that experience an increase in both the asset and liability sides of their balance sheet reduce their spending. This then suggests that any housing wealth effects on spending are reduced if the increase in housing prices is driven by an increase in the supply of mortgage debt. Using household-level data is crucial to isolate this channel, as the net wealth effect of rising housing prices dominates at higher levels of geographic aggregation. We also extend the findings in Mian and Sufi (2015) by showing that higher levels of mortgage debt can have real effects on the economy even if the credit expansion is just a passive response to higher housing prices.

3.1 The Household, Income and Labour Dynamics in Australia (HILDA) Survey

We use household-level panel data from Wave 17 of the HILDA Survey to pin down the causal relationship between debt and spending (DSS and Melbourne Institute 2018). These data allow us to exploit the heterogeneity in debt holdings and spending patterns across and within households over time. The unique scope and depth of the information in the survey also allows us to examine the underlying mechanisms.

The HILDA Survey is an annual Australian survey that has tracked the same representative group of individuals (roughly 17,000 persons from 9,000 households) since 2001. We use data up to 2017. Through personal interviews and self-completed questionnaires, the survey collects detailed information on household economic behaviour, including their spending, income, debt and assets.^[3] The wide range of information collected as well as the panel nature of this dataset makes it well-suited to answering our research questions.

There are a few features of the dataset that are worth highlighting. First, the survey collects information on both non-durables spending (between 2006 and 2017) and durables spending (between 2006 and 2010). Non-durables spending, as defined in this paper, covers frequently purchased items, such as groceries, fuel and utilities, while durables spending covers infrequently purchased items, such as whitegoods, motor vehicles and computers.^[4] While the time series for durables spending is more limited than that for non-durables, it does cover the GFC period. Our model estimates mainly focus on total spending, though in some cases we provide separate model estimates for durables and non-durables spending.

Second, information on owner-occupier housing debt and assets are available each year, while the other components of the household balance sheet (e.g. financial assets, investor housing debt, and non-housing debt) are only available every four years. The model is estimated on an annual basis to capture relatively high-frequency changes in debt and spending. This means that we limit ourselves to studying the effect of owner-occupier housing debt rather than total household debt.

In Australia, owner-occupier housing debt is by far the largest component of total household debt (at more than 60 per cent of the aggregate), so our results for its effect on spending at the household level should also matter for consumption at the aggregate level.^[5]

Third, we use the level of owner-occupier housing debt as our measure of debt whilst controlling for income and housing wealth, as this is most closely tied with the model outlined in Appendix A and allows us to test whether the depth of household balance sheets (the level of both debt and assets) matters. This is in contrast to most of the existing literature, which uses debt relative to income or assets, or mortgage repayments relative to income. In Appendix E we provide results for these alternative debt indicators.

To limit the influence of large outliers on the results and ensure that households with non-positive debt, assets, housing equity or income are not excluded from the sample by default, we follow Dynan (2012) and apply the inverse hyperbolic sine (IHS) transformation to our main variables of interest. This is likely to be a serious problem in household-level studies that use the natural logarithm transformation, as a reasonable share of households have no debt or assets, or negative incomes and housing equity in some years. Also, some households report no durables spending in a given year. The IHS transformation allows us to keep these observations.

We need to place a couple of restrictions on the sample used in our estimations. Importantly, we use the sample of households that held owner-occupier debt in the previous year.^[6] We drop all household-year observations with zero mortgage debt in the previous year for two reasons: first, we want to abstract from any short-term increase in spending due to taking on new debt (e.g. first home buyers furnishing their new home); and second, we observe that some previously indebted households are likely to have misreported having zero debt given that they return to a similar debt level the following year. We also remove some outlier observations since reporting error is an issue, particularly with the spending measures in the HILDA Survey.^[7] Specifically, we remove household-year observations in the top or bottom 1 per cent of income, house prices, and spending growth from the sample.^[8] Table C1 provides some descriptive statistics for the remaining sample as well as for the non-indebted households excluded from the sample.

The data suggest that there are some differences in spending across households with different levels of debt. Figure 2 shows the median level of durables and non-durables spending between 2006 and 2010 for highly indebted households (i.e. those in the top quartile of the debt distribution) compared with less-indebted and non-indebted households. First, highly indebted households tend to have higher levels of spending than other households since they are also more likely to be asset-rich and in the peak spending years of the life cycle (Carroll and Summers 1991; Deaton 1992; Ellis, Lawson and Roberts-Thomson 2003). Second, during the GFC, which had its peak unemployment effect in Australia in 2009, durables spending fell by more for highly indebted households than for other households. This is consistent with durables spending being more discretionary in nature and more easily postponed than non-durables spending.

3.2 The Australian Mortgage Market and Liquidity-constrained Households

In examining the drivers of any debt overhang effect, we use several measures for liquidity constraints. One such measure proposed in the literature is whether households are ‘hand-to-mouth’, that is if they hold little liquid wealth and consume almost all their current income each period. Kaplan, Violante and Weidner (2014) (hereafter, KVW (2014)) provide a framework for identifying such ‘hand-to-mouth’ households – defined as households whose liquid wealth is less than half their income each pay period.

In the KVW (2014) framework, all housing wealth is illiquid. However, this is not the case in Australia due to several important features of the mortgage market discussed in Appendix B. These features increase the ability of Australian households to prepay their mortgages (at near zero cost) and thereby build up prepayment buffers which are essentially liquid wealth. This implies that prepayment buffers should be taken into account when measuring liquid wealth.

A unique feature of the HILDA Survey is the availability of data on the home purchase history of each home owner. Every four years, households are asked how much their home cost when they originally bought it and how much they borrowed at the time. By combining this information with a standard bank (credit foncier) formula that links loan repayments to the interest rate, loan term and loan amount, we can estimate the scheduled mortgage balance of each household:

where the scheduled mortgage debt balance (D_S) is a function of the stock of debt at origination (D₀), the nominal mortgage interest rate (i), the age of the loan in years (k), and the term of the mortgage in years (T).^[9] The scheduled balance is the total amount that the borrower is contracted to repay at any given time based on this formula. But given the capacity to prepay, most borrowers have an actual mortgage balance that is lower than the scheduled balance. The difference between the actual and scheduled balance is an estimate of each household's prepayment buffer.

Based on this, we estimate that around 14 per cent of Australian households are liquidity constrained (Figure 3). This is slightly lower than the estimate based on the KVW (2014) definition that does not adjust for prepayments. However, the adjustment only affects home owners with mortgage debt. So the differences are much more pronounced when focusing solely on indebted households. On average, around 5 per cent of indebted homeowners are liquidity constrained using the new measure compared with 13 per cent using the KVW measure.

4.1 Identification

As discussed above, there are several challenges associated with identifying the effect of mortgage debt on spending. To see this, consider a simple regression of spending on mortgage debt at the household level:

where the dependent variable (E_h,t) is the level of non-housing spending of household h in year t, and the key variable of interest is the level of owner-occupier mortgage debt (D_h,t).

First, reverse causality is a problem. A household may choose to spend more than it earns, implying more borrowing and hence a higher stock of debt.^[10] To partly mitigate this, we estimate the relationship between households' current spending and the previous year's mortgage debt. It is also worth noting that reverse causality would drive a positive bias in the coefficient estimate, implying that it would be harder to pin down a potential negative debt overhang effect.

Second, omitted variables can influence both spending and mortgage debt. For example, an increase in households' income expectations may lift both their intention to spend and their desire to take on debt. Alternatively, an increase in risk aversion may discourage both spending and borrowing. Again, this is most likely to induce a positive correlation between spending and debt and attenuate any negative debt overhang effect.

Some of the challenges in identifying the causal effect of mortgage debt on spending are highlighted in an ‘event study’ around the time of home purchase. Most home purchases in Australia are financed at least in part through mortgage debt. When a household buys a home for the first time (or when they trade up to a larger/higher-quality home) they typically take on a large amount of debt. And when they buy a new home they also tend to spend more on either furnishing the new home or renovating their existing home for sale.

Given the HILDA Survey is longitudinal in nature we can observe the spending, income and debt of a household both before and after home purchase (Figure 4, left panel). In the year of home purchase, there is a notable jump in spending on durable goods. In the year after home purchase, durable goods spending returns to the pre-purchase level. This pattern of home purchase-related debt accumulation and spending in one year followed by lower spending the next year would lead us to empirically find a negative correlation between current spending and lagged mortgage debt. But the relationship would not be causal – it would be driven by an omitted variable – the decision to buy a new home. This would be similar to the ‘spending normalisation’ hypothesis (Andersen et al 2016). In contrast to the findings of Gross (2017) for the United States, we find little evidence of a fall in non-durable spending around the time of home purchase.

Another notable feature of this event study is the rise in household income in the years leading up to the purchase. We find that this is partly due to households working longer hours, presumably to save for a home deposit. But, in line with Gross (2017), it also seems to reflect a ‘selection effect’; the households that choose to buy a new home are those that received an increase in income (through, say, a promotion or bonus). Either way, this event study highlights the need to control for factors that influence both debt and spending behaviour, such as the age, income, wealth and labour force characteristics of the household.

A similar event study can be undertaken around the year in which households fully pay off their mortgage. Under the PIH, household spending should not respond to anticipated changes in scheduled debt. We would expect spending to remain constant. At odds with this prediction, and suggesting that debt may constrain spending, we find that durables spending increases in the year that households fully repay their mortgage debt and non-durables spending increases in the years following (Figure 4, right panel). The increase in spending is larger than both the observed increase in disposable income and the average mortgage payment prior to paying off the debt, which suggests that the spending response cannot be fully explained by cash flow effects.

4.2 Household Fixed Effects Model

To deal with the issues highlighted above, we exploit the rich longitudinal information available in the HILDA Survey. To test whether debt levels directly influence spending, we first estimate the following regression model, which we refer to as the fixed effects (FE) model:

This model includes the lagged level of owner-occupier mortgage debt (D_{h,t − 1}) as the key variable of interest, household disposable income (Y_h,t) and the lagged reported home value (A_{h,t − 1}). The model also includes a set of control variables (X_h,t), to summarise the other observed determinants of spending, including factors associated with a household's permanent income, such as age, education and labour force status of the household reference person.^[11] The model includes a household fixed effect $\left({\delta }_h\right)$ which captures household characteristics that determine spending but are plausibly invariant over time (e.g. degree of impatience and risk aversion). Estimates are presented with and without the household fixed effect to gauge the importance of these characteristics. Our results are robust to including year fixed effects.

4.3 Instrumental Variables Model

To further alleviate any endogeneity concerns about unobserved time-varying confounding factors (such as changes in income expectations or local labour demand shocks), we also adopt an instrumental variables approach. For this, we exploit the home purchase history of each owner-occupier household in the survey. We use information on the timing of their most recent home purchase relative to other home owners in the same postcode as an instrument for the level of owner-occupier housing debt held by the household. The logic behind this instrument is that households living in the same area are exposed to identical time-varying local demand shocks, but differ in their debt holdings based on when they happened to purchase their home. The instrument should therefore be correlated with outstanding mortgage debt but uncorrelated with differences in spending for borrowers in the same postcode other than their level of debt.^[12]

To take a hypothetical example, suppose there are two households that own identical homes in the same street. The only difference between them is that household A bought before a local housing boom happened while household B bought after. It is plausible that household A borrowed less (in dollars) than household B because housing prices in the area were lower when household A made their purchase decision. The timing of the purchase decision should not affect the spending of household A relative to household B over and above its impact on their respective levels of indebtedness.

Australia experienced a large housing price boom in the early 2000s. The timing of this boom varied by state, generally starting in 2001 in the larger capital cities of Sydney, Melbourne and Brisbane, and in 2002 in other capital cities. We can think of the households that bought just before the boom in housing prices as the ‘lucky’ households while the comparable households that bought just after the boom are ‘unlucky’. To the extent there are other differences between households that bought before and after the boom that affect their consumption behaviour (such as the level of housing wealth), we can only control for observable differences in the model.

To gauge the relevance of the instrument, we compare the average initial debt holdings of households purchasing homes just before and just after the boom (Figure 5). There is a clear jump in average mortgage debt for those households that were ‘unlucky’ to buy just after the housing boom compared to the ‘lucky’ households that bought just before the boom.^[13] Based on the weak identification test in Stock and Yogo (2005), the instrument is found to be significantly correlated with the household's current holdings of mortgage debt (even after the age of loan is taken into account).

This second model is based on a two-stage least squares regression, which we refer to as the instrumental variables (IV) model:

where most of the variables are as denoted before and p denotes the postcode in which household h lives. The main difference is that owner-occupier housing debt is estimated in the first-stage regression using a dummy variable (BOOM_hp) as an instrument. The dummy variable takes the value of one if a household purchased their home after the early 2000s housing price boom in the state of purchase and zero otherwise.^[14] To control for the location choice of the household in their home purchase decision, we drop the household fixed effects and instead include postcode fixed effects $\left({\theta }_p\right)$ in the model. Otherwise the timing of home purchase (the ‘birth cohort’ of the mortgage) will be absorbed by the household fixed effect. Thus, whilst this approach may help to alleviate endogeneity concerns by controlling for unobserved time-varying factors such as local demand shocks, the resulting estimator will include the effects of household debt on spending both within households (over time) and between households (at a point in time).

4.4 Results

4.4.1 Baseline models

Table 1 presents the key results from estimating the OLS, FE and IV models for durables, non-durables and total spending (see Appendix F for full table of results).

We find that higher mortgage debt reduces household spending across all specifications. The overall similarity between our FE estimates and the OLS estimates suggests that unobservable time-invariant variables do not play a major role after controlling for a range of household socio-economic characteristics. The FE model specification suggests that the effect of debt on spending is relatively small with a 10 per cent increase in debt reducing households' non-durables, durables and total spending by 0.1, 1.0 and 0.3 per cent, respectively. The stronger response of durables relative to non-durables spending is consistent with the common finding of larger wealth effects for durables spending.

The IV model estimates provide further evidence that the negative debt overhang effect is not driven by unobservable, time-varying factors such as local demand shocks. The first-stage regression estimates indicate that our instrument is relevant and that households that bought after the boom held mortgage debt levels that were about 45 per cent higher than comparable households living in the same area that bought before the boom. The second-stage IV estimates are more economically significant than those in the FE model and suggest that a 10 per cent increase in debt lowers non-durables, durables, and total spending by 1.5, 8.0 and 2.0 per cent, respectively. While the IV model may help to further remove time-varying confounding factors, it does not account for unobserved household fixed effects. We therefore take the FE model estimates as our benchmark estimates for the remainder of the paper.

In these regressions, we control for household income and the lagged value of gross housing assets. As expected, higher household income and housing prices raise spending. In contrast to the estimated effect on debt, the OLS coefficient estimates on income and home value are considerably larger than the FE estimates, suggesting that unobserved time-invariant factors are significant drivers behind the positive relationship between spending and both income and housing prices. For example, it may be the case that impatient households spend more and buy more expensive homes than patient households, and this partly explains the positive link between spending and home values.

We also estimate all three models using three additional debt measures (the debt-to-income, debt-to-assets and debt service-to-income ratios) that are commonly used in the literature. The results are presented in Appendix E. Using the IV model, we find some evidence that higher debt ratios negatively affect household spending, consistent with a number of papers that have found evidence of a significant negative relationship between these debt ratios and spending. In particular, we find that a 10 percentage point increase in the debt-to-assets ratio or debt-servicing ratio significantly reduces total household spending by 0.1 per cent.

4.4.2 Gross versus net housing wealth

In the previous regressions, we control for gross housing wealth. Another possibility is to control for net housing wealth (housing equity), equal to the difference between the reported value of the home and any outstanding mortgage debt. This allows us to directly test whether the composition of household balance sheets matters for spending. The results can be found in Table 2.

Overall, our estimates of the effect of debt on household spending are broadly unchanged when we control for households' housing equity instead of the value of their home. Importantly, this implies that households lower their spending when the gross value of both their debt and their assets increases. In other words, we find that a deepening of household balance sheets is associated with less household spending, even if it is not associated with rising net indebtedness. This directly violates conventional consumption theories such as the PIH that assume the composition of a household's balance sheet does not affect consumption (Garriga and Hedlund 2017). Our results suggest a small, and occasionally negative, effect of lagged housing equity on spending. When using contemporaneous housing equity, we recover the expected positive effect.

The results from our models suggest that there is a significant negative relationship between the level of household debt and spending. However, this need not be driven by the debt overhang mechanism. In this section, we explore possible underlying mechanisms.

5.1 The Debt Overhang Channels

We first explore two mechanisms that imply that the link between debt and spending is causal. Following the literature and our theoretical model in Appendix A, these mechanisms are borrowing and liquidity constraints (which we summarise as financing constraints) and precautionary saving motives. To test these hypotheses, we estimate whether the negative effect of debt is driven exclusively by households that we identify as likely to be constrained or have strong precautionary saving motives. Accordingly, we add to our models an interaction term between debt and household characteristics that are plausibly associated with financing constraints or precautionary saving (Z_h,t). We focus on the FE model and adjust it to the following form:

5.1.1 Borrowing and liquidity constraints

To test whether borrowing and liquidity constraints drive the negative relationship between debt and spending, we interact indicators of leverage and liquidity with our balance sheet and income measures. We expect borrowing constraints to be binding for households with a high stock of debt relative to assets. For this purpose, we test if households with loan-to-valuation ratios (LVRs) above 80 per cent adjust their spending by more in response to changes in income, wealth or debt, than households with a low LVR.

We proceed similarly when testing for the importance of liquidity constraints. Here, the first indicator is a dummy variable that is equal to one if the household is hand-to-mouth (adjusted for mortgage prepayments) and zero otherwise. The second indicator is a dummy variable that is equal to one if the household reports being behind on their mortgage payments and zero otherwise. A household that is behind on its repayments will have no prepayment buffer with which to offset unexpected income shocks. The third indicator measures the financial stress of each household. This dummy variable is equal to one if the household reports financial problems, such as the inability to quickly raise emergency funds, and zero otherwise.^[15]

If financing constraints drive our results, these interaction terms should be negative. However, we find the debt overhang effect to be pervasive across households and not sensitive to our proxies for borrowing or liquidity constraints (Table 3). All interaction coefficients between debt and our proxies for constraints are statistically insignificant. Furthermore, constrained households also do not appear to be more sensitive to changes in their income or home value. While this appears to suggest that financing constraints are not the main drivers of the negative relationship between debt and spending, it is possible that our proxies do not adequately identify these constraints.

We also estimate these models controlling for household cash flow, measured as income after taxes and mortgage payments rather than after-tax income alone. We find that the estimated effect of debt on spending is smaller, but still statistically significant. This provides further indirect evidence that liquidity (or cash flow) constraints are not the sole explanation for the debt overhang effect. So we next consider the role of precautionary saving motives.

5.1.2 Precautionary saving

Unlike most comparable household surveys, the HILDA Survey asks about households' employment type and their expectations of future employment, which helps to identify uncertainty about future income. In this section, we use a household's self-assessed probability of losing their job, as well as the household's casual employment status, to test whether uncertainty significantly influences the effect of debt on spending.^[16] Having a high perceived probability of losing a job or low job security as a casual worker should capture households facing high income uncertainty, and therefore having a stronger motive for precautionary saving.

We find that households with lower job security appear to be more sensitive to higher debt levels, but the estimated effects are not statistically significant and the overall negative relationship between debt and spending continues to hold for all households (Table 4). As a result, based on these proxies, we find a persistent negative relationship between debt and spending for all households, and little evidence that uncertainty strengthens the negative relationship between debt and spending.

5.2 Non-causal Explanations for a Negative Effect of Debt on Spending

5.2.1 The spending normalisation hypothesis

Andersen et al (2016) suggest that the presence of a negative effect of debt on spending could be due to ‘spending normalisation’ rather than the causal debt overhang mechanism. Households take on debt to finance a large purchase and subsequently reduce their spending back to normal levels (as highlighted by the event study earlier). The macroeconomic policy implications of spending normalisation are quite different to those for debt overhang, so it is important to examine this.

To test the spending normalisation hypothesis, we introduce a lagged dependent variable into the FE model and control for any bias introduced by this lag using the Arellano-Bond estimation procedure (Arellano and Bond 1991). The lagged variable should capture any negative effect due to ‘spending normalisation’ between the previous and current period. We refer to this as the ‘dynamic model’.

In addition, we also follow the approach used in Andersen et al (2016) and separately introduce the lagged growth in household debt into our FE and IV models. The advantage of this approach over the dynamic model is that it maintains a larger sample size, since the dynamic model requires multiple lags to be used as instruments. The lagged growth in debt should capture whether the negative effect of debt is due to previous changes in debt (e.g. spending normalisation) or the level of debt (e.g. debt overhang).

The results from the dynamic model, as well as the models controlling for the growth in debt, suggest that spending normalisation cannot explain the negative relationship between debt and spending. This is consistent with the statistically significant negative effect of lagged debt on non-durables spending in Table 1. Non-durables spending is less likely to be affected by spending normalisation since non-durable items tend to be less lumpy and smaller in size. The link between lagged debt and contemporaneous spending persists even after including lagged spending or changes in debt. Moreover, the estimated coefficient remains significant even when using the Arellano-Bond estimation procedure which reduces the sample size considerably. In contrast to Andersen et al (2016), we find no evidence that lagged spending or lagged changes in debt explain lower spending today.

5.2.2 The housing preferences hypothesis

Next, we consider whether a shift in household preferences towards owner-occupier housing might be driving the correlation between mortgage debt and spending. Note that so far our measure of spending relates specifically to non-housing goods and services. A negative correlation between mortgage debt and non-housing spending could be due to households shifting preferences towards housing and away from other goods and services (with this increased housing consumption at least partly financed through mortgage debt).

To address this, we adjust our measures of household spending and income to include the imputed rent on owner-occupier housing.^[17] If the increase in debt reflects a shift towards housing spending alone, we should see no effect of debt when we include owner-occupier housing spending in total spending.

Table 6 presents the results from the FE and IV models using the adjusted measure of household spending and income. In both models, the effect of mortgage debt on total spending is weaker than previously. However, the negative effect of debt persists, suggesting that the shift in household preferences is not the only driver of the relationship between debt and spending. Our estimates suggest that the preference shift accounts for at most one-third to half of the total negative effect of debt on spending.

Next, we explore whether the level of debt makes households more sensitive to other financial shocks (the debt amplifier effect). The debt overhang and debt amplifier effects are intrinsically linked, since financing constraints and uncertainty are likely to be more significant during financial shocks. While this makes it hard to disentangle the debt overhang and amplifier effects, especially at times of large financial shocks, significant debt amplifier effects may provide indirect evidence for the importance of financing constraints and precautionary saving motives in driving the effect of debt on spending.

We measure a financial shock in four ways. First, the GFC is used to capture an aggregate shock.^[18] Next, we use two measures of a local economic shock: an increase in the regional unemployment rate by more than 1 percentage point in a given year; and a fall in postcode-level housing prices by more than 5 per cent over a year. Finally, we use the household reference person becoming unemployed as a measure of a household-level shock.

If debt makes households more sensitive to shocks, then the spending of highly indebted households should fall by more than the spending of comparable households with lower levels of debt in response to a negative shock. This is precisely what we see in the event of a household unemployment shock (Figure 6).^[19] The indebted household reduces its spending on durable goods by more than the non-indebted in the year of the household unemployment shock and spending remains subdued after the shock.^[20] In contrast, the effects of the GFC and local economic shocks appear to have been broadly similar across indebted and non-indebted households, although indebted households tend to increase their spending after local economic shocks.

While these event studies provide some evidence that debt levels could amplify households' responses to financial shocks, they fail to take into account differences in the characteristics of indebted and non-indebted households, making it unclear whether it is debt per se that is behind the differences in responses. To address this more formally, we add an interaction term between the level of debt and the SHOCK into the FE model. We then test whether the interaction coefficient is significantly different from zero. For example, the FE model is:

Table 7 presents the results for each financial shock. As before, we find a persistent negative and quantitatively similar effect of debt on spending. This suggests that our results for the debt overhang effect are not driven by a higher sensitivity of indebted households to income or wealth shocks alone. We find some evidence for the debt amplifier effect, though it is not present in all specifications.

The debt interaction coefficient is negative for all shocks, but only significantly different from zero for the GFC and the local housing price shocks. These shocks increase the negative effect of debt on spending to around −0.05 or −0.06 per cent. While the event study suggests that households with more debt reduce their spending by more if they become unemployed, the regression estimates do not confirm this. However, even though the interaction effect of debt with household unemployment shocks is negative, it could be imprecisely estimated due to the small sample of survey respondents that have debt and have become unemployed.

The results so far provide some evidence for the debt overhang effect at the household level. A key question for policymakers is whether the results generalise to the aggregate level. Few papers have attempted to infer the aggregate implications from higher debt on spending, with the exception of Bunn and Rostom (2015). In this section, we use our preferred estimates from the FE model to draw some conclusions about the potential aggregate implications. We conduct two thought experiments. In the first experiment we use macro data and ask what aggregate consumption would have looked like had household debt remained at its 2006 level. In the second experiment we use our household-level data and compare actual spending of households in 2010 to their spending had their debt level remained at its 2006 level.

There are several limitations of these approaches and the estimates should be used as a rough guide only. Both experiments identify a partial effect of debt on aggregate spending. In particular, we only capture the (negative) effect on spending of home buyers that take on new mortgage debt but do not capture the (positive) spending response of the seller. But, if one household is buying a house for the first time or trading-up to a larger, more expensive house, often another household is on the other side of the transaction and selling their existing house. This household is liquidating their housing wealth. While the seller may invest in another house again, some share of the liquidated wealth will likely end up in consumption at some point, thereby lifting aggregate spending.^[21] Mortgage-financed housing demand may further stimulate construction and, in consequence, lift aggregate consumption. Additionally, an increase in demand for mortgages should raise mortgage interest rates (all else equal). This could induce saving by other households and increase the returns to lenders. Finally, for both thought experiments we do not allow the sensitivity of spending to debt to vary across households.^[22] Moreover, for the first thought experiment we assume that the aggregate effect of higher debt does not depend on the distribution of that debt, or that distribution, to change over time.

7.1 Using Macro Data

Using macro data allows us to gauge the aggregate effect of debt on consumption beyond the HILDA Survey sample period. Our elasticity estimate of –0.03 from the FE model (Table 1) and the observed average annual growth in real owner-occupier housing debt of around 5 per cent between 2006 and 2017 suggests that real annual consumption growth would have been around 0.15 percentage points higher had debt remained at its 2006 level (all else equal). This suggests that by 2016, consumption would have been around 1.6 per cent higher had there been no increase in debt since 2006 (Figure 7).

It is important to stress that this analysis assumes that all other economic trends that supported consumption over the period would have occurred anyway. For instance, housing prices increased strongly between 2006 and 2016, supporting consumption. But some of the increase in housing prices was likely driven by higher debt. Had debt remained at its 2006 level, house prices would not have increased by the same amount. As a consequence, the actual path of consumption includes the stimulatory effect of higher house prices through higher debt. Our estimates of a 1.6 per cent increase in consumption in the absence of an increase in debt therefore assume that housing prices would have increased for non-debt related reasons. Furthermore, we assume that all households increased their debt holdings uniformly and adjusted their spending in the same way.

7.2 Using Micro Data

To address some of the distributional concerns of differential changes in debt across households, we use the HILDA Survey data to examine how much each household changed spending due to changes in its debt holdings between 2006 and 2010, and then sum these differences across all households. In contrast to the previous section, this allows for heterogeneity in debt growth across households.^[23] For each household, we compare the observed level of spending (E_h,t) with a counterfactual level of spending under the assumption that debt did not change $\left({\widehat{E}}_t^{cf}\right)$ .

The counterfactual level of spending is estimated by taking the difference between the observed spending and the attributed effect of the change in debt of each household and then summing these differences across all households. We use the level change in household debt instead of the percentage change in order to keep households in the sample without any debt in 2006 for which the percentage change would be undefined. Using the level change in debt, however, requires us to transform our elasticity estimates (the percentage spending response to a percentage change in debt) to estimates of the marginal propensity to consume (MPC, the level spending response to a level change in debt).

Using the MPC instead of elasticities is in line with the related literature on the effects of changes in wealth on consumption. We can obtain an aggregate MPC out of mortgage debt in two ways. First, we can scale our estimated elasticity coefficient by the ratio of aggregate total consumption expenditure to lagged mortgage debt. From 2006 to 2010, the ratio of annual consumption to owner-occupier mortgage debt averaged at around 105 per cent, which – multiplied with our coefficient of −0.03 from the FE model in Table 1 – implies an aggregate MPC of −3.2 cents per dollar increase in mortgage debt. Second, we can estimate the MPC directly by re-specifying our regression in level changes rather than using the IHS transformation. Our dependent variable is then the change in household spending, and our three main independent variables are level changes in income, lagged housing debt and lagged home value. The average MPC we obtain using this approach is around – 2.0 cents per dollar (not statistically significant at the 10 per cent level).^[24]

Using our estimates for the MPC, we can then obtain the counterfactual level of aggregate spending as:

The results of the counterfactual exercise are presented in Table 8. Interestingly, the micro data estimates suggest that the effect of debt on aggregate spending is of similar magnitude to the effect using macro data estimates. Using the elasticity-implied MPC out of debt of −3.2 cents per dollar suggests that the changes in mortgage debt between 2006 and 2010 reduced annual aggregate total spending growth over this period by 0.36 percentage points. Using the directly estimated MPC of −2.0 cents per dollar lowers this estimate to 0.24 percentage points lower aggregate total spending growth annually.

Overall, both the macro data and micro data estimates imply that higher aggregate debt levels alone can have an economically significant effect on macroeconomic activity. Again, it is important to note that these estimate do not account for possible stimulatory effects of debt. Furthermore, while this exercise captures differences in changes in debt across households, we do not consider how the MPC varies across households. As discussed in Section 7.1, it is possible that there are meaningful differences in the effect of debt across households and this variation could have important implications for the aggregate effect.

Consistent with international research, we find evidence that high levels of owner-occupier mortgage debt reduce household spending. Higher mortgage debt is associated with less spending even when we control for changes to net housing wealth and cash flow (adjusted for mortgage repayments). This implies that a deepening of both sides of the household balance sheet is associated with weaker spending, and that debt matters for spending over and above its effect on net wealth. In other words, we find that the composition of household balance sheets affects household consumption decisions, which contradicts the predictions of conventional theories, such as the PIH.

Overall, the negative effect of debt on spending is pervasive across households with owner occupier mortgage debt. In contrast to previous literature, we find little evidence for borrowing and liquidity constraints or precautionary saving motives to be key drivers of the negative debt overhang effect. While our results do not directly support the debt overhang mechanism, we rule out some non-causal explanations for the negative effect of debt on spending proposed in the literature.

Specifically, we find that neither spending normalisation nor a shift in household preferences to housing consumption can explain the negative effect of debt on non-housing spending. While we find that household spending is more sensitive to debt during adverse shocks such as the GFC and local housing price shocks, the negative effect of debt also persists in more ‘normal’ times.

Our results also suggest that an increase in aggregate owner-occupier mortgage debt can have important implications for aggregate spending, all else constant, and go at least part of the way to resolving the post-crisis ‘puzzle’ of unusually weak household spending in Australia.

Below we set out a highly stylised two-period partial equilibrium model to illustrate the channels through which higher debt levels might affect spending, and hence explain a debt overhang mechanism. The household lives in an endowment economy, faces no uncertainty and has a lifetime utility function given by:

where the discount factor is $0\le \beta \le 1$. For simplicity, assume that the utility function is given in logarithmic form.

The household has a budget constraint in the first period given by:

where the household consumes (C₁) in the first period out of their current income (Y₁) and new borrowing (L₁) less repayments on debt (D₀) that they inherit from period 0. The household is assumed to repay a fixed fraction $\left(0<\delta <1\right)$ of the initial debt, and this fraction is exogenously given. The household is also assumed to be endowed with an asset from period 0 (A₀), which can serve as collateral in period 1 but which cannot be liquidated until period 2.

The budget constraint in the second period is:

where the real interest rate on debt per period is r. We assume that the stock of debt follows a law of motion:

where the total debt at the end of period 1 is equal to any borrowing during the period plus any outstanding initial debt that the household was born with less the repayment made in the period. The repayment of past debt includes both an interest and principal component. A household that makes a larger principal payment than required (or ‘prepayment’) will have more liquidity at its disposal in the future than a similar household that makes a smaller (or no) prepayment. When δ = 1 this corresponds to the case of one-period debt.

The household's lifetime budget constraint is:

This says that the household can consume out of the present discounted value of lifetime income plus any net wealth that they are born with.

However, when taking on debt, the household potentially faces a borrowing constraint:

where the maximum LVR in any period is given by $0\le {\overline{\phi }}_D\le 1$. This states that the total debt for the household is limited by the value of its gross assets (collateral). Given the law of motion for debt, this implies that the household is potentially constrained in the amount it can borrow in period 1. Moreover, the maximum amount of new borrowing depends on the household's previous debt and any prepayments:

where ${\gamma }_0\le {\overline{\phi }}_D$ is the household's leverage from period 0. Accordingly, if the household was born with the maximum amount of debt $D_0=\overline{D}={\overline{\phi }}_D{A}_0$ , it can at most borrow the amount it repaid, that is $L_1\le \delta {\overline{\phi }}_D{A}_0$ .

The household may also face a liquidity (or cash flow) constraint in period 1:

This condition is essentially a debt servicing requirement imposed by mortgage lenders. It states that the household's repayments in period 2 (when it fully repays the loan) $\left(\left(1-\delta \right)D_1\right)$ must be less than some fraction $\left(0\le \theta \le 1\right)$ of its current disposable income, which is equal to its endowment less the interest payments on the initial debt (Y₁ −rD₀). Using the law of motion for debt, this condition again implies a constraint on new borrowing in the first period:

To see how this works, consider a household that was born with debt and no prepayment $\left(\delta =0\right)$. This household will have more disposable income (or cash flow) in period 1, but will be required to make a larger repayment in period 2. This will mean the constraint is more likely to bind in period 1. Another household that fully prepaid the loan $\left(\delta =1\right)$ will have less cash flow in period 1, but also no repayments in period 2, and hence a greater stock of liquid resources during that final period. For this household, the liquidity constraint will never bind.

The household chooses consumption in each period to maximise its lifetime utility subject to the lifetime budget constraint, the borrowing constraint and the liquidity constraint.

For the unconstrained household, the consumption function is:

And the sensitivity of spending to outstanding debt for the unconstrained household is:

This shows that there is a negative wealth (or income) effect of debt on spending for the unconstrained household.

For a borrowing-constrained household, the consumption function is:

And the sensitivity of spending to outstanding debt for this household is:

When $\beta >0$ and the borrowing constraint binds, any increase in debt reduces consumption by more than the wealth effect. This works through a reduced ability to borrow $\left(1-\delta \right)$ and lower disposable income to meet the required repayments $\left(\delta \right)$.

For a liquidity-constrained household, the consumption function is:

And the sensitivity of spending to outstanding debt for this household is:

This condition holds as long as the real interest rate is positive (r > 0). In effect, there is a negative wealth (or income) effect of debt on spending for the constrained households, but also additional borrowing and liquidity effects. So the effect of debt on spending is larger for constrained households than for unconstrained households. This implies that a debt overhang effect will be stronger for households that face binding financing constraints (Eggertsson and Krugman 2012).

Next, we consider the role of uncertainty in driving the debt overhang effect (King 1994; Albuquerque and Krustev 2015). For simplicity, suppose the only source of uncertainty is about the interest rate to be paid on the debt as the household enters the second period.^[25] And consider a mean-preserving spread of the interest rate:

where the interest rate consists of a constant mean $\left(\overline{r}\right)$ and a stochastic component $\left(\epsilon \right)$. Assume that $E\left(\epsilon \right)=0\text{and}V\left(r\right)=\sigma$. For expositional purposes, also assume that $\beta r=1$. In this case, the household in the first period faces the following decision:

where the first line is just the optimality condition when the constraint does not bind, while the second line comes from the household budget constraint in period 1 when the constraint does bind. This condition can be rewritten in compound form:

Given that we have introduced uncertainty in the model, note that the interest rate enters the equation in expectation. Now suppose that uncertainty about interest rates increases. Very high realisations of interest rates $\left(\epsilon >0\right)$ become more likely, which reduces the household's expected disposable income. As a result this makes the cash flow constraint more likely to bind in the future and reduces the value of $\overline{\overline{L}}=Y_1\left(1+\frac{\theta }{1-\delta }\right)-\frac{\theta E\left(r\right)D_0}{1-\delta }-D_0$. To avoid this, the household reduces consumption in period 1. This precautionary saving effect will be larger for households with more initial debt.

In effect, when there is uncertainty about the ability to meet future mortgage repayments due to an increase in uncertainty about future income or interest rates, the household may choose to consume less today even though they are not currently liquidity constrained, but because they are worried about becoming constrained in the future.

The theoretical model indicates that liquidity constraints (either now or in the future) can drive the relationship between debt and spending. KVW (2014) outline a model to identify liquidity-constrained or ‘hand-to-mouth’ households. In this model, the consumption of some households is very sensitive to shocks to current income despite high levels of wealth. Some households hold wealth in illiquid assets, such as housing, and still act as if they are constrained because their liquid wealth is low. These households may adjust their spending even in response to transitory (and predictable) income or wealth changes. KVW (2014) apply this framework to household-level data for a range of countries, including Australia. La Cava, Hughson and Kaplan (2016) extend KVW's Australian results to study the evolution of hand-to-mouth households over time and examine how these households influence the sensitivity of aggregate household spending to monetary policy shocks.

There are a couple of features of the Australian mortgage market that are unique by international standards and affect the extent to which households are ‘hand-to-mouth’. Most borrowers have offset accounts or redraw facilities linked to their mortgages. These loan features make housing wealth more liquid than otherwise; they reduce the transaction costs involved in prepaying mortgages and make it easier to build buffers that can be used to offset income or wealth shocks.

An offset account is an at-call deposit that is directly linked to the mortgage. Funds deposited into an offset account reduce the borrower's net debt and the interest payable. A redraw facility enables the borrower to withdraw principal payments that they have made ahead of the required schedule. Mortgages with offset accounts currently comprise around 45 per cent of the total value of residential mortgage debt in Australia. Mortgages with redraw facilities make up around 70 per cent of the total number of residential mortgages in Australia (APRA 2019).

These mortgage features provide a tax-effective method of saving. First, the deposit rate that the borrower earns on the balance is equal to the mortgage interest rate. Second, the account generates no tax liability compared to depositing money into a separate savings account where any interest accruing adds to taxable income. This means there can be large tax advantages for borrowers to retain funds in a mortgage offset account (mortgage interest payments are not tax deductible for owner-occupier loans, though they are for investor loans, which are not considered here).

The main differences between offset accounts and redraw facilities are the degree of liquidity and the effect of withdrawals on home equity. In terms of liquidity, the funds sitting in an offset account are at call and easily accessible for withdrawal and for purchasing goods and services. The money in a redraw facility, while accessible, is not generally available for same day, at call withdrawal. There may also be a fee associated with redrawing money from the loan. A withdrawal from an offset account does not affect the principal balance of the loan, whereas a withdrawal from a redraw facility increases the principal and hence reduces the equity in the home.

Another feature of the Australian market is the high share of mortgages that do not face prepayment penalties. Over 80 per cent of home mortgages in Australia are originated at variable (or adjustable) rates and these mortgages can be prepaid without penalty. The combination of no prepayment penalties, tax incentives and highly liquid offset accounts means that Australian mortgage borrowers typically build up liquidity buffers by prepaying their mortgages. These buffers imply that housing is not as illiquid as that implied by the KVW (2014) framework.

Albuquerque B and G Krustev (2015), ‘Debt Overhang and Deleveraging in the US Household Sector: Gauging the Impact on Consumption?’, European Central Bank Working Paper Series No 1843.

Andersen AL, C Duus and TL Jensen (2016), ‘Household Debt and Spending during the Financial Crisis: Evidence from Danish Micro Data’, European Economic Review, 89, pp 96–115.

APRA (Australian Prudential Regulation Authority) (2019), ‘Quarterly Authorised Deposit-taking Property Exposures Statistics: December 2018’, Statistical publication.

Arellano M and S Bond (1991), ‘Some Tests of Specification for Panel Data: Monte Carlo Evidence and an Application to Employment Equations’, The Review of Economic Studies, 58(2), pp 277–297.

Atalay K, S Whelan and J Yates (2017), Housing Prices, Household Debt and Household Consumption, AHURI Final Report No 282, Australian Housing and Urban Research Institute Limited, Melbourne.

Baker SR (2018), ‘Debt and the Response to Household Income Shocks: Validation and Application of Linked Financial Account Data’, Journal of Political Economy, 126(4), pp 1504–1557.

Brazier A (2017), ‘“Debt Strikes Back” or “The Return of the Regulator”?’, Open Lecture at University of Liverpool's Institute for Risk and Uncertainty, Liverpool, 24 July.

Bunn P and M Rostom (2015), ‘Household Debt and Spending in the United Kingdom’, Bank of England Staff Working Paper No 554.

Carroll CD and LH Summers (1991), ‘Consumption Growth Parallels Income Growth: Some New Evidence’, in BD Bernheim and JB Shoven (eds), National Saving and Economic Performance, National Bureau of Economic Research Project Report, The University of Chicago Press, Chicago, pp 305–343.

Cho Y, J Morley and A Singh (2019), ‘Household Balance Sheets and Consumption Responses to Income Shocks’, Unpublished manuscript, 13 May.

Deaton A (1992), Understanding Consumption, Clarendon Lectures in Economics, Oxford University Press, Oxford.

de Roiste M, A Fasianos, R Kirkby an F Yao (2019), ‘Household Leverage and Asymmetric Housing Wealth Effects – Evidence from New Zealand’, Reserve Bank of New Zealand Discussion Paper DP2019/01.

DSS and Melbourne Institute (Department of Social Services and Melbourne Institute of Applied Economic and Social Research) (2018), ‘The Household, Income and Labour Dynamics in Australia (HILDA) Survey, RESTRICTED RELEASE 17 (Waves 1-17)’, ADA Dataverse, V2, accessed 29 November 2018. Available at <doi:10.26193/OFRKRH>.

Dvornak N and M Kohler (2007), ‘Housing Wealth, Stock Market Wealth and Consumption: A Panel Analysis for Australia’, Economic Record, 83(261), pp 117–130.

Dynan K (2012), ‘Is a Household Debt Overhang Holding Back Consumption?’, Brookings Papers on Economic Activity, Spring, pp 299–344.

Eggertsson GB and P Krugman (2012), ‘Debt, Deleveraging, and the Liquidity Trap: A Fisher-Minsky-Koo Approach’, The Quarterly Journal of Economics, 127(3), pp 1469–1513.

Ellis L, J Lawson and L Roberts-Thomson (2003), ‘Housing Leverage in Australia’, RBA Research Discussion Paper No 2003-09.

Fagereng A and E Halvorsen (2016), ‘Debt and Household Consumption Responses’, Norges Bank Staff Memo No 1|2016.

Garriga C and A Hedlund (2017), ‘Mortgage Debt, Consumption, and Illiquid Housing Markets in the Great Recession’, Federal Reserve Bank of St. Louis Working Paper 2017-030A.

Gillitzer C and JC Wang (2015), ‘Housing Wealth Effects: Cross-Sectional Evidence from New Vehicles Registrations’, RBA Research Discussion Paper No 2015-08.

Gross I (2017), ‘Anticipated Changes in Household Debt and Consumption’, Unpublished manuscript, 24 November.

Hunt C (2015), ‘Economic Implications of High and Rising Household Indebtedness’, Reserve Bank of New Zealand Bulletin, 78(1).

Jordà Ò, M Schularick and A Taylor (2013), ‘When Credit Bites Back’, Journal of Money, Credit and Banking, 45(s2), pp 3–28.

Kaplan G, B Moll and GL Violante (2018), ‘Monetary Policy According to HANK’, The American Economic Review, 108(3), pp 697–743.

Kaplan G, GL Violante and J Weidner (2014), ‘The Wealthy Hand-to-Mouth’, Brookings Papers on Economic Activity, Spring, pp 77–138.

King M (1994), ‘Debt Deflation: Theory and Evidence’, European Economic Review, 38(3–4), pp 419–445.

La Cava G, H Hughson and G Kaplan (2016), ‘The Household Cash Flow Channel of Monetary Policy’, RBA Research Discussion Paper No 2016-12.

La Cava G and J Simon (2003), ‘A Tale of Two Surveys: Household Debt and Financial Constraints in Australia’, RBA Research Discussion Paper No 2003-08.

Lombardi M, M Mohanty and I Shim (2017), ‘The Real Effects of Household Debt in the Short and Long Run’, BIS Working Papers No 607.

Lowe P (2017), ‘Household Debt, Housing Prices and Resilience’, Address given to Economic Society of Australia (QLD) Business Lunch, Brisbane, 4 May.

Mian A, K Rao and A Sufi (2013), ‘Household Balance Sheets, Consumption, and the Economic Slump’, The Quarterly Journal of Economics, 128(4), pp 1687–1726.

Mian A and A Sufi (2010), ‘Household Leverage and the Recession of 2007 to 2009’, NBER Working Paper No 15896.

Mian A and A Sufi (2015), House of Debt: How They (and You) Caused the Great Recession, and How We Can Prevent It from Happening Again, Paperback edn, The University of Chicago Press Economics Books, The University of Chicago Press, Chicago.

Mian A and A Sufi (2018), ‘Finance and Business Cycles: The Credit-Driven Household Demand Channel’, Journal of Economic Perspectives, 32(3), pp 31–58.

Mian A, A Sufi and E Verner (2017), ‘Household Debt and Business Cycles Worldwide’, The Quarterly Journal of Economics, 132(4), pp 1755–1817.

Paiella M and L Pistaferri (2017), ‘Decomposing the Wealth Effect on Consumption’, The Review of Economics and Statistics, 99(4), pp 710–721.

Pistaferri L (2016), ‘Why Has Consumption Remained Moderate after the Great Recession?’, Paper presented at the Federal Reserve Bank of Boston's 60th Economic Research Conference ‘The Elusive “Great” Recovery: Causes and Implications for Future Business Cycle Dynamics’, Boston, 14–15 October.

Schularick M and AM Taylor (2012), ‘Credit Booms Gone Bust: Monetary Policy, Leverage Cycles, and Financial Crises, 1870–2008’, The American Economic Review, 102(2), pp 1029–1061.

Schwartz C, T Hampton, C Lewis and D Norman (2006), ‘A Survey of Housing Equity Withdrawal and Injection in Australia’, RBA Research Discussion Paper No 2006-08.

Stock JH and M Yogo (2005), ‘Testing for Weak Instruments in Linear IV Regression’, in DWK Andrews and JH Stock (eds), Identification and Inference for Econometric Models: Essays in Honor of Thomas Rothenberg, Cambridge University Press, Cambridge, pp 80–108.

Wilkins R and C Sun (2010), ‘Assessing the Quality of the Expenditure Data Collected in the Self-Completion Questionnaire’, Melbourne Institute of Applied Economic and Social Research, The University of Melbourne, HILDA Project Discussion Paper Series No 1/10.

Windsor C, J Jääskelä and R Finlay (2013), ‘Home Prices and Household Spending’, RBA Research Discussion Paper No 2013-04.

Yao J, A Fagereng and G Natvik (2015), ‘Housing, Debt, and the Marginal Propensity to Consume’, Job Market Paper, John Hopkins University.

HILDA Disclaimer

This paper uses unit record data from the Household, Income and Labour Dynamics in Australia (HILDA) Survey. The unit record data from the HILDA Survey was obtained from the Australian Data Archive, which is hosted by The Australian National University. The HILDA Survey was initiated and is funded by the Australian Government Department of Social Services (DSS) and is managed by the Melbourne Institute of Applied Economic and Social Research (Melbourne Institute). The findings and views based on the data, however, are those of the authors and should not be attributed to the Australian Government, DSS, the Melbourne Institute, the Australian Data Archive or The Australian National University and none of those entities bear any responsibility for the analysis or interpretation of the unit record data from the HILDA Survey provided by the authors.