Recently there has been a growing debate as to whether the empirical properties of production response derived from the multi-product framework are consistent with the behavioral assumption in the duality theory of production. This issue could in turn affect the reliable estimates of elasticities which are fundamental requirement to accurate economic forecasting and valid analyses of the impacts of changes in government policies or international events. The crux of the debate is particularly related to whether to estimate cost or profit function and which types of functional form should be used, as well as imposing restrictions on profit and cost functions when those properties are not satisfied by the estimation models. After reviewing the literature, we found that the duality theory may not always hold in empirical work. This depends on many factors such as risk, stochastic error or data quality, and selected functional forms. We also found that the normalized quadratic function has more advantage than other functional forms, although all of flexible functional forms often fail to pass the regularity condition in the duality theory. As such, there is a requirement to impose curvature restriction if normalized quadratic function is used and monotonicity condition needs to be checked if the cross price elasticities are more than unity. We examined the above-mentioned issues by estimating production response for broadacre farms in Western Australia. A normalized quadratic profit function is estimated using the ABARE’s quasi-micro farm level data for the period 1977/78 to 2005/06. The result reflects the imposition of curvature restrictions for a normalized profit function, and estimated elasticities are found to be less elastic in