Extension of the Forward Search Estimation Method: Robust Analysis of Energy Markets and Asset Pricing Models

In this thesis we develop a new weighted robust estimation method based on the Forward Search (Atkinson and Riani, 2000). Through extensive simulation experiments we show that the method is almost as efficient as OLS when data is free of outliers and is very robust when data are contaminated. The method is applied to energy market data as well as asset pricing models. Results indicate that data used in both contexts contain outliers and conclusions based on estimated parameters should be interpreted with care, as the most commonly used estimation methods are very sensitive to outliers. Through the application of the weighted forward search estimator more reliable coefficients are obtained which better describe the generating process of the observed data.