Stability, Empirical Estimates and Scenario Generation in Stochastic Optimization - Applications in Finance

0485151 - ÚTIA 2018 RIV CZ eng J - Journal Article
Kaňková, Vlasta
Stability, Empirical Estimates and Scenario Generation in Stochastic Optimization - Applications in Finance.
Kybernetika. Roč. 53, č. 6 (2017), s. 1026-1046. ISSN 0023-5954
R&D Projects: GA ČR GA15-10331S
Institutional support: RVO:67985556
Keywords : stochastic programming * stochastic dominance * empirical estimates * financial applications
OECD category: Statistics and probability
Impact factor: 0.632, year: 2017
http://library.utia.cas.cz/separaty/2017/E/kankova-0485151.pdf

Economic and financial processes are mostly simultaneously influuenced by a random factor and a decision parameter. While the random factor can be hardly influenced, the decision parameter can be usually determined by a deterministic optimization problem depending on a corresponding probability measure. However, in applications the „underlying“ probability measure is often a little different, replaced by empirical one determined on the base of data or even (for numerical reason) replaced by simpler (mostly discrete) one. Consequently, real one and approximate one correspond to applications. In the paper we try to investigate their relationship. To this end we employ the results on stability based on the Wasserstein metric and L1 norm, their applications to empirical estimates and scenario generation. Moreover, we apply the achieved new results to simple financial applications. The corresponding model will a problem of stochastic programming.
Permanent Link: http://hdl.handle.net/11104/0280355