ABSTRACT
In the NIF-10 model 1 of the Australian economy, dwelling commencements react quite slowly to housing loans and general liquidity. The nature of the data is such as to suggest that the estimated lags are unrealistically long, at least for housing loans. This paper examines the issue, paying attention to the relationships between lags and error structure. The explanatory variables are initially confined to the two used in NIF-10. A competing model is then introduced in which the explanatory variables used are the asset price of existing dwellings and the cost of construction of new dwellings.
