Testing the sensitivity of CGE results: A Monte Carlo Filtering approach to an application to rural development policies in Aberdeenshire

Parameter uncertainty has fuelled criticisms on the robustness of CGE results and has led to the development of alternative approaches to sensitivity analyses. Researchers have used Monte Carlo (MC) for systematic sensitivity analysis (SSA) because of its flexibility. However, MC may provide biased simulation results. Gaussian Quadratures (GQ) have then been developed, but they are much more difficult to apply in practical modelling and may not always be desirable. This report applies an alternative approach to SSA, Monte Carlo Filtering, and examines how its results compare to MC and GQ approaches, in an application to rural development policies in Aberdeenshire.