"Using a switching regression technique we provide unique evidence on three questions concerning the consumption behaviour of UK households. First, what percentage of households display excess sensitivity to income? Second, what affects the likelihood of being in that group? Third, is there a collateral channel from house prices to consumption? We find 20%-40% of households display excess sensitivity. These households may be liquidity constrained or saving for other precautionary reasons. This is found to be more likely for those without liquid assets, with negative home equity, the young, unmarried, non-white and the degree-educated. According to the 'collateral channel', house prices influence consumption by allowing households that would otherwise be liquidity constrained to borrow on more attractive terms. A key implication of that view is that capital gains on housing should influence the consumption of the liquidity constrained/precautionary saving households, but not other households. We test that implication for the first time and find direct evidence in support."--Bank of England web site.