RDP 2003-08: A Tale of Two Surveys: Household Debt and Financial Constraints in Australia 4. Features of the Data

Before undertaking any formal modelling, it is useful to look at some of the features of the data to be used in the model. The data presented in this section come from the HES for 1998/99. Firstly, given the importance of the cash flow variable in our analysis, it is of interest to look at its composition. As has already been stated, our proxy for cash constraints is whether a household reports having problems on at least one of seven dimensions of financial stress (e.g., could not pay their bills, had to pawn something, went without meals) – 22 per cent of households report being constrained on at least one of these measures. However, some of these measures were more common than others.

Figure 2 shows the relative contributions of each measure of financial stress. For example, the first column shows that around 72 per cent of cash-constrained households fell into this category because they could not pay their utility bills due to a shortage of money. Most of the other cash flow problems can be explained by households being unable to pay registration or insurance on time (29 per cent) or because they had to ask family or friends for assistance (44 per cent).^[8] One other possible measure of cash flow problems is whether the household could raise $2,000 in a week as emergency money. However, around 51 per cent of households reporting cash flow problems on our preferred measure also reported being unable to raise $2,000 in a week. Furthermore, only 10 per cent of unconstrained households reported being unable to raise the emergency money. So there is significant overlap in these measures. Finally, using the emergency money variable in the model does not appreciably affect the results.

Given our motivating question, it is of interest to see whether household debt and financial stress are related in a simple bi-variate analysis. An important limitation of the survey data in this regard is relevant. Debt for investment purposes is not directly measured by the surveys. The HES excludes debt used to purchase a dwelling that is rented out for more than three months in the previous year. The HILDA Survey asks only about debt secured against the principal place of residence of the household. To the extent that some investment loans may be secured against people's principal place of residence, they would be captured by HILDA.

It should also be noted that, given we have access to unit record file data, we measure debt to income ratios by dividing the outstanding stock of debt for each household by the level of income for each household – effectively weighting all households equally. Aggregate measures of debt to income ratios divide the sum of the total debt stock across all households by the sum of total incomes across all households – effectively giving more weight to higher-income households. We measure the debt-service ratio (mortgage repayments as a share of disposable income) the same way. Differences in the weighting schemes can account for differences in both the level and growth rates of these ratios. For instance, our measure of the level of the debt-service ratio (for households with debt) will tend to be higher than the aggregate measure because higher-income households (with debt) generally have lower relative debt burdens. Also, changes in the distribution of the debt will show up in our measures but are unlikely to affect aggregate measures in the same way.

Figure 3 shows that, across all households, individuals living in lower-income households are more likely to suffer cash flow problems. Around 25–30 per cent of households in the two lowest income quintiles are cash-constrained. This falls to around 10 per cent for households in the highest income quintile. So while cash flow problems are more frequent in lower income groups, they still remain prominent for some households at all income levels. And while low-income households appear more likely to have cash flow problems, they also appear less likely to have debt. As a share of income, household debt stands at around 10–30 per cent in the two lowest quintiles. It peaks at around 60–70 per cent in the third and fourth quintiles. On the surface, this suggests that the households holding debt are less likely to be financially stressed.

However, if we focus on households with debt we can see a possible positive correlation between debt levels and the degree of financial stress. For households with debt, debt to income ratios peak in the lowest income quintiles at around 240 per cent. Debt to income falls to around 130 per cent in the highest income group. For households with debt, financial stress peaks in the second quintile as 42 per cent of households with debt in this quintile report having problems.

A different way to look at these data is across age groups rather than income groups. The life-cycle model of consumption posits that younger households should borrow to consume in advance of future income, repay their debt and save through the middle years, and draw down their savings after retirement. As younger households have had less time to build up assets than older households, they are more likely to report cash flow (and other) problems, as supported by Figure 4.

In keeping with the life-cycle model, the majority of household debt across all households appears to be concentrated in the middle-aged households rather than in young households. So, again, across all households there is only tentative evidence that financial fragility and the incurrence of debt are related. Reported cash flow problems generally fall as the household head gets older. In the case of the aged, the low incidence of cash flow problems may reflect prudent financial management, stable income flows and a capacity to draw upon assets.

But if we again just focus on households with debt, the youngest households appear to have the highest debt to income ratios, peaking at around 220 per cent in the 25–29 age group, reflecting the fact they are more likely to have recently taken out a loan. Younger households are also more likely to report having had cash flow problems with around 24 per cent of households in the 25–29 age group having suffered financial stress in the past year. For households with debt, reported cash flow problems also fall as the household head gets older.

So the overall picture we glean from this is that households with debt are generally less likely to be cash-constrained. However, for those households that do hold mortgage debt, the more debt they hold the more likely they are to be financially constrained. To better understand the relationship between financial constraints and debt, and to control for various other factors, we need to employ more sophisticated econometric techniques and it is this to which we now turn.

Footnote

Some of the other components, such as whether the household pawned or sold something (19 per cent), whether they were unable to heat their home (10 per cent) and whether they went without meals (12 per cent) may be better indicators of financial hardship rather than cash flow problems per se (see Bray (2001)). We experimented with different combinations of the various components in the model as proxies for cash flow problems, without significantly altering the results. [8]