B1986
Title: Investigating different parameter estimation techniques for the Lomax distribution
Authors: Thobeka Nombebe - North-West University (South Africa) [presenting]
James Allison - North-West University (South Africa)
Jaco Visagie - North-West University (South Africa)
Leonard Santana - North West University (South Africa)
Abstract: The performance of a variety of estimation techniques for the scale and shape parameter for the Lomax distribution is investigated. These methods include the L-moment estimator, the probability-weighted moments estimator, the maximum likelihood estimator, the maximum likelihood estimator adjusted for bias, the method of moments estimator and three different minimum distance estimators. The comparisons will be done by considering the variance and the bias of these estimators. Based on an extensive Monte Carlo study, we found that the so-called minimum distance estimators are the best performers for small sample sizes. However, for large sample sizes, the maximum likelihood estimators outperform the minimum distance estimators. We conclude with a practical example applied in the context of duration models.
