An application of General Maximum Entropy to Utility

Methodologies related to information theory have been increasingly used in studies in economics and management. In this paper, we use generalised maximum entropy as an alternative to ordinary least squares in the estimation of utility functions. Generalised maximum entropy has some advantages: it does not need such restrictive assumptions and could be used with both well and ill-posed problems, for example, when we have small samples, which is the case when estimating utility functions. Using linear, logarithmic and power utility functions, we estimate those functions and confidence intervals and perform hypothesis tests. Results point to the greater accuracy of generalised maximum entropy, showing its efficiency in estimation.