The note provides a high-level summary of work attempting to predict extreme
movements in house prices.
Extreme house price movements can have significant economic implications.
House prices have an effect on savings rates, resulting in redistribution between
homeowners and renters and large falls may adversely affect the retirement of
elderly people. If these groups have different consumption and investment patterns
a change in the relative price of housing can lead to changes in non-housing
expenditure. Housing is also an important channel to facilitate access to credit
for liquidity constrained households. Many households do not hold significant
non-housing assets that can be used as collateral, so extreme volatility in house
prices may adversely affect certain households access to credit.
There are also increased risks to the financial sector from falling house prices.
Residential mortgage lending is a high proportion of bank assets, and sharp falls
in house prices can result in negative equity loans. High loan to value ratios,
in particular negative equity loans, are positively correlated with the probability
of default, which can put stress on the financial position of the bank, leading to
a tightening in lending standards, caution in the financial markets and possibly
unanticipated adjustments to interest rates, exchanges rates and monetary policy
Indicators of extreme house price movements would be of value to policy makers.
As house prices show mean reversion over the longer run, advance warning
of an extreme appreciation in house prices would allow policy makers more
flexibility in responding. Similarly, indicators of an increased risk of house price
depreciation would allow policy makers more time to respond, possibly preventing
crisis interventions typically seen when nominal house prices decrease.
Extreme movements are, by definition, rare. Data on New Zealand house prices
consists of 40 years of quarterly data, or 160 data points. This necessitates pooling
internationally comparable data, which include countries such as Australia, U.K.,
U.S.A., Canada, France and Germany. Variables such as interest rates, exchange
rates, population, private sector credit and household income are used as possible
indicators. From these data tail dependence, the probability of coinciding extreme
events, is estimated between house prices and each of the indicators.
Extreme increases in house prices are significantly more likely after an extreme
increase in household income or population across a 12 month horizon, however
extreme house prices exhibit strong extremal serial dependence which prevent
causal insights. The results suggest that extreme house price increases are
correlated with increases in the fundamental drivers of house price inflation; as
the demand for housing increases, or as incomes increase, it would be expected
that house prices would similarly increase.
Extreme decreases in...