CMStatistics 2017: Start Registration
View Submission - CMStatistics
B1543
Title: Consistent asymptotically normal estimators for stable distributions Authors:  Annika Krutto - University of Tartu (Estonia) [presenting]
Tonu Kollo - University of Tartu (Estonia)
Abstract: Two methods for estimating parameters in general stable distributions are proposed. One allows point estimators for the scale and shift, another for the four parameters of stable distributions. Methods involve empirical cumulant function. The main advantage of the methods is computational simplicity. Moreover, both provide consistent estimators. Asymptotic normal distributions for the proposed cumulant estimators are specified. The efficiency of the method relies on two arbitrary real points. The points selection problem is treated and suggestions stipulated. Comprehensive simulations are performed. Comparison to other estimation procedures is presented.