Housing Leverage and Consumption Expenditure - Evidence from New Zealand Microdata

Release date
Karam Shaar; Fang Yao

Understanding household consumption spending is crucial for modelling business cycles and designing macroeconomic policy. This paper investigates how household debt affects the marginal propensity to consume out of housing wealth.

We use microdata from Statistics New Zealand’s "Household Economic Survey" (HES) to investigate how household leverage affects the marginal propensity to consume (MPC) out of housing wealth. HES data provide detailed information on household spending, income and loans. Empirically, estimating the effect of housing wealth changes on household expenditure faces two types of endogeneity issues. First, any evidence of an association between housing wealth variations and consumption changes could be driven by unobservable confounding factors such as future income expectations or household preferences. Second, naive regressions with total household spending can suffer from reversed causality, in which high housing-related spending leads to higher property values. We combine HES data with Real Estate Institute of New Zealand (REINZ) micro house price data to address the endogeneity issues that arise from using household-level cross-sectional data.

In the empirical analysis, we first assess the validity of average local house prices as an instrument for individual house prices. The first stage regression suggests that the instrument can explain up to 22 percent of the variation in individual house prices reported in HES. We then run a benchmark regression of total household expenditure excluding housing-related spending on housing wealth. The IV estimation suggests that using household-level prices leads to downward bias, which is the result of various causes of endogeneity issues discussed above. The average MPC out of a one-dollar increase in exogenous housing wealth is around 2.2 cents. All regressions control for income, household characteristics, and regional and time fixed effects. We also split non-housing expenditure into durables and non-durables. In line with other studies in the literature, we find that durable consumption is more sensitive to changes in housing wealth than non-durables.

We then focus on the role of household leverage in determining the MPC out of housing wealth. In this analysis, we study how leverage measures, such as the loan-to-house-value ratio (LVR) and the DTI, affect the estimated MPC out of housing wealth. Overall, we find that household leverage weakens the MPC associated with housing. To examine the robustness of these findings, we investigate whether household spending responds differently depending on the age and type 5 of home ownership. The findings confirm that the consumption of mortgagors is less sensitive to housing wealth as compared to outright homeowners. The regression with an age-housing wealth interaction also shows that the response of younger households to changes in their housing wealth is weaker than the response of older households, which tend to be less leveraged.