DETERMINATION OF DEPENDENCY PARAMETER IN JOINT DISTRIBUTION OF DEPENDENT RISKS BY FUZZY APPROACH Fatih TANK KIRIKKALE UNIVERSITY FACULTY OF SCIENCE AND LITERATURE, DEPARTMENT OF STATISTICS, 71100 YAHSIHAN-KIRIKKALE / TURKEY Omer L. GEBIZLIOGLU ANKARA UNIVERSITY FACULTY OF SCIENCE, DEPARTMENT OF STATISTICS, 06100 TANDOGAN-ANKARA / TURKEY Aysen APAYDIN ANKARA UNIVERSITY FACULTY OF SCIENCE, DEPARTMENT OF STATISTICS, 06100 TANDOGAN-ANKARA / TURKEY Abstract Frechet bounds for distribution of sum of dependent risks and Farlie-Gumbel- Morgenstein (FGM) distribution family are suggested for the analysis of dependent risks. Behavior of the coe¢cient of dependency in FGM distribution is investigated by fuzzy approach. Deliberations are made for the dependency and uncertainty context. Key words: FGM distribution family, fuzzy mathematical programming, dependent risks, coe¢cient of dependency, uncertainty, imprecision. 1 INTRODUCTION Generally risks are assumed to be independent in a portfolio. But in many situations this assumption does not t the reality. For instance, in a company, workers group life or health insurance can be a/ected by a signicant acci- dent, re, earthquake or a disease so that risks are not independent among Preprint submitted to Elsevier Science 30 August 2005 Manuscript