Abstract:
In this paper, we develop models for estimating the time varying probability that there will
be a price reversal in the property market. Knowledge of such price reversals may be helpful
in forming property trading strategies, and providing confirming evidence of turning points
in property cycles. Using an index from the UK property market, we obtain the time varying
probabilities by estimating state transition rates. State transition rates are estimated for
cases where the up-state (a run of positive price changes) switches to the down-state (price
falls) and vice versa. We also estimate a model for absolute transitions, where we take no
account of the direction of the state transition. The predictive power of our models is
assessed using data from a holdout period. We find that the absolute transition model
performs worse than separate state transition models for the up-state and the down-state.
For these latter models, the more rapid the decline in the forecast probability of a run
continuing beyond a certain time, the less likely it is that the run will actually continue.
Further, the probability profiles provide a perfect rank ordering of the length of runs in the
holdout period. This suggests that the probabilities could be used to predict whether a
sequence of price rises or falls will be long- or short-lived.
