When is Mortgage Indebtedness a Financial Burden to British Households? A Dynamic Probit Approach

Abstract

Since the mid-1990s the volume of secured lending to households has expanded rapidly, both in absolute terms and in relation to household incomes. This paper examines the determinants of households' ability to service this stock of secured debt. It estimates a random effects probit model for the probability of households having mortgage payment problems. It is found that past experience of payment problems increases the probability that the household has difficulties servicing its secured debt today. At the household level, becoming unemployed, interest income gearing of 20% and above, high loan to value ratios and having a heavy burden of unsecured debt are all associated with a significantly higher probability of mortgage payment problems. Saving regularly and having unsecured debt which is not a problem are both associated with a significantly lower probability of mortgage payment problems. The only non-household-specific variable to have a significant effect is mortgage interest rates - the probability of payment problems increases with the level of mortgage interest rates. An aggregate measure of debt at risk is calculated. This has decreased between 1994 and 2002, as falls in the robability of mortgage payment problems have more than offset increases in the stock of mortgage debt outstanding. It is found that the fall in the probability of mortgage payment problems has been greatest among the most highly indebted households.

Keywords: mortgage debt, dynamic probit

JEL Classification: D14, C23, C35.