Risk Measures in Optimization Problems via Empirical Estimates
Risk Measures in Optimization Problems via Empirical Estimates
Author(s): Vlasta KaňkováSubject(s): Economy
Published by: Univerzita Karlova v Praze - Institut ekonomických studií
Keywords: Static stochastic optimization problems; linear and nonlinear dependence; risk measures; thin and heavy tails; Wasserstein metric; L1 norm; empirical distribution function
Summary/Abstract: Economic and financial activities are often influenced simultaneously by a decision parameter and a random factor. Since mostly it is necessary to determine the decision parameter without knowledge of the realization of the random element, deterministic optimization problems depending on a probability measure often correspond to such situations. In applications the problem has to be very often solved on the data basis. It means that usually the “underlying” probability measure is replaced by empirical one. Great effort has been made to investigate properties of the corresponding (empirical) estimates; mostly under assumptions of “thin” tails and a linear dependence on the probability measure. The aim of this paper is to focus on the cases when these assumptions are not fulfilled. This happens usually just in economic and financial applications (see, e.g., Mandelbort 2003; Pflug and Römisch 2007; Rachev and Römisch 2002; Shiryaev 1999).
Journal: Czech Economic Review
- Issue Year: 7/2013
- Issue No: 03
- Page Range: 162-177
- Page Count: 16
- Language: English