This thesis concerns the highly complex relationship between cost and design decision-making. A theoretical cost function is described by examining the output requirements of the design activity, and the constraints placed on the input to the process by current cost generation procedure. The description takes the form of a set of preferred characteristics or alternatively, criteria on which the potential of a future costing process might best be judged. A particular interpretation of this set of characteristics is described, which produces the tentative specification for a new generation of cost models. An implementation of the specification resulted in the computer-based cost simulation model ACE (Analysis of Construction Economics). This interactive, knowledge-based cost model is used to investigate a series of cost relationships and cost thresholds, both to exemplify its possible applications, and to produce some means of determining the validity of both the general approach, and the particular interpretation which ACE represents.

The building industry is complex, diversified, and labour-intensive. These aspects, together with its inherent instability, are analysed. Improved forecasting methods can assist in economic planning within the industry and formulation of public policy. Economic stabilisation policies can benefit participants in the industry and society at large. In this study leading indicators are developed for the South African building industry to assist in forecasting future demand levels. Use is made of qualitative survey data and quantitative time series. The quarterly qualitative data emanate from the Bureau for Economic Research, University of Stellenbosch. These data are gathered by questionnaire from building contractors and sub-contractors according to the Konjunkturtest developed by the lfo Institute, Munich, Germany. Principal component analyses of the business survey variables reveal that respondents behave purposefully and that these qualitative data are suitable for use as cyclical indicators in a composite index. The monthly quantitative data are compiled by the South African Reserve Bank and the Central Statistical Service, Pretoria, South Africa. The variables used in the construction of the leading indicators are weighted according to the scoring system developed by the National Bureau of Economic Research, United States of America. The six criteria applied in this scoring system are: economic significance of the variables; statistical adequacy; timing at turning points; conformity to historical business cycles; currency; and smoothness. Separate composite leading indices are compiled from 33 qualitative variables and 8 quantitative time series, with the relevant scores as weights. It is found that these indices lead turnjng points of the reference cycle by between three and a half months and ten and a half months. However, the lead times are not consistent. This finding is in accordance with international experience. A combined leading indicator is constructed from these qualitative and quantitative indices (1971 to 1991). It is found that the statistical performance of the final composite leading indicator does not surpass the performance of the individual composite indices. It is suggested that the best forecasting results can be achieved if the qualitative and quantitative leading indices are · used independently, yet in conjunction with other economic indicators and other forecasting models.