RDP 2003-08: A Tale of Two Surveys: Household Debt and Financial Constraints in Australia 2. Previous Research

There has been little previous work linking debt levels to household financial constraints. However, there has been substantial work focusing on the closely related link between consumption behaviour and constraints. This work infers the level of household constraints from macroeconomic consumption behaviour. Other work of a more microeconomic nature looks directly at what factors affect people's access to credit.

Most of the macroeconomic work considers the effect of household constraints in the context of testing the Rational Expectations Permanent Income Hypothesis (REPIH).[1] This theory holds that households choose the path of consumption that maximises their expected lifetime utility. When households are forward-looking in this way, changes in current income should have little effect on their consumption patterns: it is lifetime income that matters. The seminal work in this area was done by Hall (1978). In Hall and in subsequent work, empirical tests have consistently rejected the hypothesis, showing that consumption is ‘excessively sensitive’ to changes in current income.[2] One explanation for this excess sensitivity is the possible existence of liquidity constraints. If households are denied access to credit they will be unable to borrow against future income to optimally smooth their consumption. Instead, these households must resort to consuming solely out of current income.[3] The extent to which aggregate consumption follows aggregate income can then be used to infer the proportion of households who are credit-constrained. The emerging consensus is that about 20 per cent of the population are credit-constrained. However, these estimates do vary significantly, both across countries and across time. This is likely to reflect both structural and cyclical factors. For example, the deregulation of the financial system over the 1980s and 1990s is expected to have made it easier for households to access credit in most industrialised countries. It follows that liquidity constraints are now less likely to affect consumption.

There are a number of Australian studies in this area. Debelle and Preston (1995) suggest that the proportion of liquidity-constrained (current income) consumers has fallen significantly from 40–45 per cent in the 1970s to 20–25 per cent in the 1980s–1990s – as expected given financial deregulation. Blundell-Wignall, Browne and Tarditi (1995) examine similar sub-periods and find a similar decline in the sensitivity of consumption to current income for a large number of countries. However, unlike Debelle and Preston (1995) they do not find support for declining constraints in Australia, a result upheld by de Brouwer (1996).[4]

Household level studies have also tested the REPIH. The main advantage of micro studies is that, in their data sets, they are generally able to directly observe constrained consumers rather than having to infer the presence of liquidity constraints. For example, Jappelli (1990) and Cox and Jappelli (1993) use the Survey of Consumer Finances (SCF) to study the characteristics of liquidity-constrained consumers in the US. Credit constraints can be directly observed in their micro data as the SCF provides information on which consumers had their request for credit rejected by financial institutions. Jappelli (1990) shows that economic characteristics (such as current income, wealth and unemployment) are important determinants of whether a household is credit-constrained. However, Jappelli (1990) also shows that demographic characteristics (such as age, marital status and household type) are highly significant. As such, macro studies that ignore demographic change may not capture changes in the true distribution of liquidity constraints across time.

Cox and Jappelli (1993) estimate the extent to which borrowing constraints reduce the levels of debt held by credit-constrained households. They find that desired debt exhibits a pronounced life-cycle pattern, increasing until the age of the household head reaches the mid-30s, and then declining. Also, the gap between desired and actual debt is highest for younger households, indicating that they would benefit most from the easing of liquidity constraints. The probability of being constrained falls with age and is negatively related to permanent earnings and net worth. Duca and Rosenthal (1993) extend their work by also examining the manner in which lenders vary debt limits across borrowers. They find that debt limits are affected by household income, wealth, credit history and ethnic background.


The literature on the effect of liquidity constraints on consumption is extensive and well summarised by Deaton (1992), Muelbauer (1994), and Attanasio (1998). [1]

Studies that are particularly relevant to the current paper are summarised in Appendix A. [2]

For liquidity constraints to affect consumption behaviour, households must be unable to borrow as much as they want, face an increasing income path and be impatient enough to want to bring resources from the future to the present. [3]

The disparity between the results of Blundell-Wignall et al (1995) and Debelle and Preston (1995) may be due to the latter's sample period extending into a more deregulated financial environment during the mid 1990s, de Brouwer's (1996) results may not be directly comparable as he uses a different range of proxies for liquidity constraints, none of which prove to be significant for Australia. Moreover, he uses annual data where Debelle and Preston (1995) use quarterly data. [4]