Some Results on the Variance of Welfare Estimates from Recreation Demand Models

This paper examines three points concerning the variances of consumer welfare measures: (1) the size of variance estimates depend on the assumed source of error, (2) the precision of welfare estimates (as measured by the coefficient of variation) generally falls with an increase in the size of the welfare change, and (3) the difference between Marshallian and Hicksian welfare measures also depends on the assumed source of error and may be small relative to the standard error. These results are demonstrated analytically using the linear or semilog demand function. A simple simulation experiment illustrates the possible magnitude of these points.