On the issue of plan shape complexity: plan shape indices revisited

Plan shape indices are based on the geometry of the plan shape rather than on empirical data, and are problematic as they embrace implicit assumptions that are not tested empirically nor justified theoretically. Empirical research using data in Hong Kong confirms that these plan shape indices are much poorer predictors of unit construction costs than the variables used to construct the indices. This study also develops a method for constructing a plan shape index that is free from the unjustified assumptions implicit in existing plan shape indices; it involves an empirical cost model. Different functional specifications of the cost models are tested. The result rejects the linear model and other special case models such as semi-log and log-linear models. However, the reciprocal model is not rejected. This result suggests that a linear plan shape index can predict the amount of floor area that can be constructed with a fixed sum of money better than the construction cost per floor area.