Specification Bias in Housing Production Functions

To produce housing, a wide variety of inputs are necessary: land, structural capital, different forms of energy, water and sewer services, management, and other factors. Yet for reasons of analytical convenience and/or lack of data, researchers investigating the production function for housing services have rarely taken a comprehensive view of the production function. This study examines the effects of some of the more common forms of simplification assumed in analyses of the production function for housing. A translog production function was fit to data drawn from the Housing Assistance Supply Experiment, taking land, capital, energy, and other current inputs as its factors. The results indicate that measurement of the first order effects in the production function is relatively insensitive to the way inputs are aggregated or to the use of restrictive functional forms. The second order effects in the production function appear to be less stable.