| dc.contributor | 國貿系 | - |
| dc.creator (作者) | 顏佑銘 | - |
| dc.date (日期) | 2021-11 | - |
| dc.date.accessioned | 12-May-2026 15:55:09 (UTC+8) | - |
| dc.date.available | 12-May-2026 15:55:09 (UTC+8) | - |
| dc.date.issued (上傳時間) | 12-May-2026 15:55:09 (UTC+8) | - |
| dc.identifier.uri (URI) | https://ah.lib.nccu.edu.tw/item?item_id=182496 | - |
| dc.description.abstract (摘要) | Fizzler and Ziegel (2016)提出一種新的一致性損失函數（簡稱為FZ損失）。本計畫的目的在研究該一致性損失函數在估計上的相關議題。 FZ損失是一個隨機變數的分位數及條件尾端期望值的一致性損失函數，這意味著實證上，我們可以經由極小化FZ損失來同時估計這兩個統計泛函數。這種估計是一種半參數估計，並不需要對該隨機變數的機率分配進行任何假設。關於分位數，我們已知道可以經由極小化特定的損失函數來估計它。而FZ損失的提出，使得我們也可以用類似的方法來估計條件尾端期望值，這在先前是無法想像的。
本計畫會優先著重於研究如何使用FZ損失來估計風險價值（VaR）和預期損失（ES），這兩者都是FZ損失的可引發統計泛函數。這兩種風險衡量指標廣泛地應用於金融領域，而發展一個可以準確估計它們的方法，將有益於監管機構、學術研究人員和相關業者。
我們預計執行以下任務: (1)在使用FZ損失的情況下，我們將構建一個半參數估計方法來聯合估計VaR和ES; (2)我們會在VaR和ES的結構模型中納入已實現波動率，並對不同方法所產生的樣本內估計和樣本外預測表現進行綜合比較; (3)我們將進行模擬，以解釋為何所提出的方法會有如此之表現; (4)我們將研究估計結果的理論性質，並嘗試建立一個在使用FZ損失來做估計的情況下之一般統計理論; (5)我們將討論如何將FZ損失應用於其他計量經濟學之議題，例如因果關係推論。 | - |
| dc.description.abstract (摘要) | In this project, we aim to investigate issues of estimations with a new class of consistent loss functions proposed by Fizzler and Ziegel (2016) (henceforth the FZ loss). The FZ loss is a consistent loss function for both the quantile and conditional tail expectation of a random variable, which means that the two statistical functionals can be jointly estimated by minimizing an empirical version of the FZ loss. Such an estimation is a semiparametric estimation and does not require any assumption on the distribution of the random variable. The FZ loss makes it possible to estimate the conditional tail expectation in a similar fashion as estimating the quantile based on a specific loss function (say tick loss), which was not available previously.
Our prior interest is on using the FZ loss to estimate value at risk (VaR) and expected shortfall (ES), which are elicitable statistical functionals of the FZ loss. The two risk measures are widely used in finance. Developing a method for accurately estimating them will be beneficial to regulation authorities, academic researchers and industry practitioners.
We plan to conduct the following tasks: (1) with the FZ loss, we will construct a semiparametric estimation method to jointly estimate VaR and ES; (2) we will incorporate realized variance measures into the structural models for VaR and ES, and comprehensively compare their in-sample estimation and out-of-sample forecast performances with other existing methods; (3) we will conduct simulations to justify performances of the proposed method; (4) we will investigate theoretical properties of the estimated results and try to establish a general statistical theories for estimations with the FZ loss; (5) we will discuss possible extensions of using the FZ loss to other econometric issues, such as causal inferences. | - |
| dc.format.extent | 116 bytes | - |
| dc.format.mimetype | text/html | - |
| dc.relation (關聯) | 科技部, MOST108-2410-H004-056-MY2, 108.08-110.07 | - |
| dc.subject (關鍵詞) | 預期損失; 預測; 已實現波動率衡量; 半參數估計; 風險價值 | - |
| dc.subject (關鍵詞) | Expected shortfall; Forecast; Realized variance measure; Semiparametric estimation; Value-at-risk | - |
| dc.title (題名) | FZ損失函數之應用：預測風險衡量指標及其他用途 | - |
| dc.title (題名) | Applications of the Fz Loss: Forecasting Risk Measures and Other Extensions | - |
| dc.type (資料類型) | report | - |
