RDP 2014-09: Predicting Dwelling Prices with Consideration of the Sales Mechanism 4. The Persistence of Shocks

We now consider whether alternative measures of average prices, based on the type of sale, can provide information about the persistence of shocks to dwelling prices. To answer this question, we use two conditions supported in the data: (a) that auction and private-treaty prices are cointegrated and can be represented by a VECM; and (b) that private-treaty prices do not Granger cause auction prices. That is:

Together, these conditions are sufficient for identifying the effects of permanent and transitory shocks to auction and private-treaty prices.^[15] A permanent shock is defined as having an effect on long-run forecasts of auction and private-treaty prices whereas a transitory shock has no such effects.

Table 6 reports forecast error variance decompositions for Sydney and Melbourne of the permanent and transitory shocks assuming that both conditions hold. We only report the decomposition for private-treaty prices, as conditions (a) and (b) together necessarily imply that variation in auction prices can only be attributed to permanent shocks. Relaxing assumption (b) and assuming that only the long-run adjustment parameter in the auction price equation is zero^[16] leads to very similar results.

Table 6 highlights that almost half of the forecast error variation in private-treaty prices one-quarter-ahead is due to transitory shocks. At the two-quarter- and four-quarter-ahead horizons, transitory shocks account for about 20 and 10 per cent of the forecast error variances respectively.

Figure 3 graphs estimates of the permanent and transitory shocks over time. We see very clearly that the estimated permanent shocks are much larger than the estimated transitory shocks. In particular, the period covering the mid 1990s to the 2000s is a period in which permanent shocks were having a noticeable positive effect on dwelling price growth. This is consistent with the typical explanations for changes in dwelling prices during this period, including the effects of financial deregulation and productivity improvements, and a shift to easier access to credit and lower real interest rates.^[17]

In contrast, the transitory shocks to auction and private-treaty prices are smaller in magnitude. The most prominent periods of positive transitory shocks were in the recovery from the early 1990s recession and around 2001 to 2003. Even during the global financial crisis, the estimates suggest that there were no large transitory shocks to dwelling prices. This is interesting given that for other countries the crisis has generally been interpreted as a demand shock, and the conventional wisdom is that demand shocks have only transitory effects on dwelling prices.

Two key properties of the propagation of permanent and transitory shocks are highlighted by the impulse response to a one standard deviation permanent shock, and a one standard deviation transitory shock respectively (Figure 4). The first is that auction prices adjust more quickly than private-treaty prices in response to a permanent price shock. Calculating the fraction of the long-run increase in prices (lim_h→∞E_t (y_t+h) for y_t = a_t, p_t) that has occurred in a given period, we see that around 80 (Sydney) to 60 (Melbourne) per cent of the long-run increase in prices occurs within the first quarter for auction prices, but only around 35 to 25 per cent of the adjustment has occurred for private-treaty prices. After four quarters, roughly 95 per cent of the adjustment to the long-run auction price has occurred for both Sydney and Melbourne. In contrast, when using private-treaty prices approximately 85 to 75 per cent of the adjustment to their long-run price level has been completed after four quarters.

The second property to note is that transitory price shocks have smaller effects on private-treaty prices than do permanent shocks. The restrictions supported by our previous empirical findings – that auction and private-treaty prices can be represented by a VECM and that private-treaty prices do not Granger cause auction prices – imply that transitory shocks have no effect on auction prices.

Footnotes

For further discussion on this point, see Fisher and Huh (2007) and Pagan and Pesaran (2008). [15]

That is, if we only impose the restriction that α_a = 0 rather than for all j = 1,…,J with respect to Equation (4). [16]

See, for example, Ellis (2006) and Yates (2011). [17]