Shrinkage Methods for Estimating the Shape Parameter of the Generalized Pareto Distribution
Algorithm 1
Algorithm for comparing the performance of the GPD estimators.
Step 1. Generate a sample of size from a GPD with known shape and scale () parameters such that both light and heavy-tailed distributions are realized.
Step 2. For each number of top order statistic or equivalently threshold value compute estimates of where {MLE, LME, PB, IPB, Proposed 1 and Proposed 2}.
Step 3. In order to compute the bias and mean square error for each estimator, repeat Steps and a large number of times, to obtain the estimates, with
