在小樣本下，利用極端值理論衡量作業風險值

• Main Author: 游依霖
• Other Authors: 張大成
• Format: Others
• Language: zh-TW
• Published: 2004
• Online Access: http://ndltd.ncl.edu.tw/handle/58360948179484256964

碩士 === 東吳大學 === 經濟學系 === 92 === In the history of banking system, the major cause of serious banking crisis appears to be directly connected with the poor operational standards. It is the operational risk of banks. Banks face risk resulting from operation all the time. Banks should supervisor and manage operational risk well. However, Banks in Taiwan neither understand the definition of operational risk well nor record loss data completely. Because operational risk differs from each bank in structure, culture and business units, external loss data can not be used for all banks. Hence, there are only few data to measure operational risk. The purpose of Value-at-Risk (VaR) is to predict down side risk. Extreme Value Theory (EVT) discusses the tail of the probability distribution, reduces the risk of the wrong model which resulted from misspecification the data and predicts the tail VaR correctly. Normally, EVT would use thousands or millions of data to estimate model, but the operational loss data are not large enough. Compare to large data, this study called the sample as small sample sizes. Hence this study would like to use internal data to measure operational risk and reflecting the scale of operational risk exposure of banks. The result shows that EVT would not reduce the power of prediction because of sample sizes. This study would like to use the internal data reflecting actual operational losses, and control the risk. key words：operational risk、Value-at-Risk、Extreme Value Theory