Title: A generalized normal distribution with applications to fit financial and economic data
Authors: Carl Lee - Central Michigan University (United States) [presenting]
Abstract: The distributions of financial and economic data are often highly skewed. During the recent decades, some new methods have been developed for generating highly flexible distributions with four or more parameters. These flexible distributions, although are capable of fitting highly skewed data, they suffer a common weakness of requiring four or more parameters, which have no practical and meaningful interpretations. This article presents a three-parameter generalized normal distribution capable of fitting highly skewed data. One parameter characterizes the location, two parameters together characterize a very wide range of scale, skewness and kurtosis. The distribution is applied to fit the World Trade import and export data, and the time and cost to start a business of 200 countries in the world. Ten years of World Trade import and export data (2008 to 2017) are extracted from WTO (World Trade Organization) database. Ten years of time and cost to start a business (2008-2017) are extracted from the World Bank database. The goodness of fits are presented and the pattern of the distribution shift during the ten years period are investigated.