How home value shocks drive spending
House prices and aggregate spending tend to move in tandem. However, there is disagreement over the causes of this link. One set of studies emphasises the housing wealth effect hypothesis. According to this, an unanticipated change in the home value can affect total lifetime resources, and this can lead homeowners to revise their consumption plans. Campbell and Cocco (2007), Muellbauer (1990), Skinner (1996) find support for this hypothesis. An alternative set of studies emphasises the collateral hypothesis, whereby an increase in home values generates additional collateral that homeowners can borrow against. Agarwal and Qian (2017), Aron and Muellbauer (2013), Browning et al. (2013), Cooper (2013), Disney and Gathergood (2011), and Leth-Petersen (2010) find evidence in support of this hypothesis, and it is also the explanation emphasised by Mian and Sufi (2011) and Mian et al. (2013) in the context of the recent US housing collapse and mortgage crisis.
A new test to discriminate between the wealth effect and the collateral effect hypotheses
Up to now it has been difficult to perform a direct test of the housing wealth effect hypothesis against the collateral effect hypothesis, because data on subjective expectations to home values along with data on mortgage borrowing and spending have been lacking. Identifying unexpected movements in home values is fundamentally a matter of how subjective expectations about home values align with realisations. In a recent study (Andersen and Leth-Petersen 2019), we collect a new individual-level longitudinal dataset for about 5,000 Danes covering the period 2011-2014 with subjective expectations of future home values. Our data documents exactly what homeowners believe about their wealth, and not what the econometrician believes, and they enable us to calculate subjective unanticipated home value changes. We link this information to high quality third-party-reported administrative records with information about deposits, financial assets, bank and credit card debt, and detailed information about mortgages and the timing of refinancing decisions. This unique dataset enables us to directly test the housing wealth hypothesis against the alternative hypothesis that the association between spending and home values is driven by the loosening of borrowing constraints when prices increase.
Spending responses driven by mortgage borrowing
We find that unanticipated home value gains lead to increased mortgage borrowing. The marginal propensity to increase mortgage debt is 3-5% of unanticipated home value gains. Unanticipated home value losses are not associated with any change in mortgage debt. This pattern is documented in Figure 1. Because we have data for practically the entire budget constraint, we can back out the implied spending response, which turns out to exhibit the same pattern and the same magnitudes as for mortgage debt growth. The asymmetric response to positive and negative home value changes is indicative of the fact that the wealth effect is not likely to be the most relevant explanation. When homeowners are entirely unconstrained, spending adjustments should be symmetric for positive and negative price changes. The expected pattern is different when mortgage borrowing constraints are at play. They only bind at the time of loan origination, such that banks do not ask borrowers to pay back the loan more quickly when prices decline. For this reason, when borrowing constraints drive the association between home values and spending, we should expect to see an effect only when prices increase. To further support this conclusion, we consider homeowners with high and low mortgage loan-to-house value (LTV) ratios separately and find that the response is concentrated on homeowners who are close to their borrowing constraint.
Figure 1 Change in mortgage debt against unanticipated home value growth
Notes: The horizontal axis shows annual unanticipated home value growth in the period 2011-2014. The vertical axis shows mortgage debt growth. Mortgage debt growth is derived directly from records reported by mortgage banks. The panel shows a binned scatterplot (red circles) where the bins are defined over equal intervals of the unanticipated home value growth and the size of the circles scaled is by the number of observations in the bins. Regression lines weighted by number of observations are estimated separately for positive and negative home value growth (blue) are overlaid. All variables are normalised on average income during 2008-2010.
Next, we analyse mortgage borrowing activity directly and find that the effect is driven by homeowners who actively take out a new mortgage. This is illustrated in Figure 2, showing the propensity to take out a new mortgage against the magnitude of the unanticipated price change. There is clearly a higher propensity to take out a new mortgage among homeowners faced with unanticipated home value increases than among homeowners faced with home value falls.
Figure 2 Propensity to actively take out new mortgage
Notes: The panel has annual unanticipated home value growth in the period 2011-2014, on the horizontal axis and a dummy variable indicating whether a loan has been refinanced on the vertical axis. The graph shows a binned scatterplot (red circles) where the bins are defined over equal intervals of home value growth and the circles vary in size according to the number of observations in the bins. Regression lines weighted by the number of observations and estimated separately for positive and negative home value growth (blue) are overlaid. The unanticipated home value growth is normalised on average income during 2008-2010.
We further find that holders of fixed-rate mortgages who are faced with an incentive to refinance to lock in a lower market rate take advantage of this opportunity and extract equity at the same time. This further amplifies the effect of home value growth on borrowing and spending. These findings point to the importance of the mortgage market in transforming price increases into spending and suggest that monetary policy can play an important role in transforming housing wealth gains into spending by affecting interest rates on mortgage loans.
Authors’ note: This column builds on the analysis presented in Andersen and Leth-Petersen (2019).
Agarwal, S, and W Qian (2017), “Access to Home Equity and Consumption: Evidence from a Policy Experiment”, Review of Economics and Statistics, 99 (1), 40-52.
Andersen, H Y, and S Leth-Petersen (2019), “Housing Wealth or Collateral: How Spending and Home Equity Extraction Respond to Unanticipated Housing Wealth Gains”, Journal of The European Economic Association, forthcoming. Also available as CEPR Discussion Paper no. 13926.
Aron, J, and J Muellbauer (2013), “Wealth, Credit Conditions, and Consumption: Evidence from South Africa”, Review of Income and Wealth, 59, S161-S196.
Browning, M, M Gørtz, and S Leth-Petersen (2013), “Housing Wealth and Consumption: A Micro Panel Study”, Economic Journal, 123 (563), 401-428.
Campbell, J, and J Cocco (2007), “How Do House Prices Affect Consumption? Evidence from Micro Data”, Journal of Monetary Economics, 54 (3), 591–621.
Disney, R, and J Gathergood (2011), “House Price Growth, Collateral Constraints and the Accumulation of Homeowner Debt in the United States”, B.E. Journal of Macroeconomics, 11 (1), 1-30.
Leth-Petersen, S (2010), “Intertemporal Consumption and Credit Constraints: Does Total Expenditure Respond to An Exogenous Shock to Credit?”, American Economic Review, 100, 1080-1103.
Muellbauer, J, and A Murphy (1990), “Is the UK Balance of Payments Sustainable?”, Economic Policy, 5 (11), 347-396.
Skinner, J (1996), “Is Housing Wealth a Side Show?”, In D Wise (ed.), Advances in the Economics of Aging, 241–68, Chicago, IL: University of Chicago Press.
Mian, A, and A Sufi (2011), “House prices, Home Equity–based Borrowing, and the US Household Leverage Crisis”, American Economic Review, 101 (5), 2132-2156.
Mian, A, K Rao, and A Sufi (2013), “Household Balance Sheets, Consumption, and the Economic Slump”, Quarterly Journal of Economics, 128 (4), 1687-1726.