Abstract:
This study estimates the patterns of dwellings losses from New Zealand housing stock between 1860-1979 by developing mortality models which address the limitations in the available field survey data. As the mortality of New Zealand housing stock is unknown, three hypotheses of mortality are formulated. It is assumed that New Zealand dwelling cohorts have been exposed to regimes of mortality which either remain the same, change, or vary simultaneously over time. Each deterministic mortality model is driven by a predetermined series of net gains variables obtained from census statistics and by a test schedule of probability of loss variables derived from possible functions of mortality. The probability of loss variables in the dynamic mortality model are amplified by a function of predetermined expansion rates of housing stock. The mortality models are tested against a profile of housing stock by age distribution and a series of total dwelling losses which is estimated as a residual of new dwelling permits less net gains from census statistics. Significant levels of uncertainty in the field survey data are handled by estimating upper and lower limiting boundaries to the uncertain data. These boundaries are used to formulate criteria and descriptive tests for a realistic and useful mortality model. The main findings of this study are that New Zealand housing stock has been exposed to a regime of mortality which is a function of age and the expansion rate of housing stock. As a result of fluctuations in the expansion rate, each dwelling cohort has been exposed to different regimes of mortality. About 50% of dwellings have been lost from each cohort by the age of 90 years and the distribution of losses from each cohort follows that of a bell shape skewed to the left. By having been exposed to dynamic mortality conditions, the future mortality of New Zealand housing stock cannot be assumed to be the same as past mortality. Assumptions need to be made as to the future expansion rates of the housing stock.