The elasticity of substitution for land: The effects of measurement errors

The explanatory variable used in most previous studies of land-nonland substitution, assessed or appraised land value, is subject to considerable measurement error. There is evidence that this biases the substitution elasticity from a true value of unity to a measured value in the 0.5 to 0.8 range. An hedonic method of measuring land value is proposed and applied to the Chicago residential sales data developed by Berry and Bednarz. Substitution elasticities are estimated within a CES framework from the two measures of land value; after correction for a vintage effect, appraised land value indicates that the elasticity is 0.57 whereas the hedonic measure of land values supports a unitary substitution elasticity. Therefore, the errors-in-variables hypothesis is confirmed and the use of a Cobb- Douglas production function is suggested.