| dc.contributor | 國貿系 | |
| dc.creator (作者) | 郭炳伸 | |
| dc.date (日期) | 2019-10 | |
| dc.date.accessioned | 7-Apr-2026 13:28:41 (UTC+8) | - |
| dc.date.available | 7-Apr-2026 13:28:41 (UTC+8) | - |
| dc.date.issued (上傳時間) | 7-Apr-2026 13:28:41 (UTC+8) | - |
| dc.identifier.uri (URI) | https://ah.lib.nccu.edu.tw/item?item_id=181949 | - |
| dc.description.abstract (摘要) | 實證上已有很多證據發現, 一般追蹤資料不具「合併性」。但奇怪的是, 應用上研究者仍然不顧前述的證據,照樣視追蹤資料可合併,假裝所有橫斷面參數皆相同下進行估計。主要原因在於,在這樣假設下的估計式具有簡約容易特性,因而可以在預測比較上形成較小的均方差。本計畫企圖建構新的平均估計式,用以解決這種實務上的矛盾。本估計的想法源於追蹤資料內的某些組群具有相同的對應參數。這代表還有存在一些參數限制關係可以加以應用,除了所有參數皆相同此一限制之外。本平均估計式的創新之處在將所有這些群組限制估計一起平均。顯然地,新的群組平均估計式將會有截然不同的統計性質。本計畫希望以2年期完成探討下列議題: (1) 決定每一群組估計值的最適權重, 以及推導更重要的漸近性質; (2) 發展本平均估計式的漸近理論, 以及推導其漸近風險; 以及(3) 研究本平均估計式在不同型態追蹤資料下的小樣本性質。 | |
| dc.description.abstract (摘要) | While ample evidence has been established against the poolability null for empirical panel data, researchers have to content themselves with the use of pooled estimates in applications. This is simply because the pooled estimators yield lower mean-squared errors (MSE) in forecasting due to their parsimony and simplicity. This project attempts to propose a new averaging estimator to resolve this conflicting practice. The idea of the estimator rests on the existence of some clusters in panels within which the parameters may be the same. It suggests many more restrictions worth being considered than that all slopes equal across individuals and time. The novelty of the estimator thus lies in that the averaging takes into account all possible clustered estimates in computations, in sharp contrast to the existing estimators just averaging over the pooled and individual estimates. Thus, the proposed clustering averaging estimator appears to carry a very distinct character. This motivates and opens up a handful of research questions to be addressed in the project regarding its statis-tical properties. Major investigations of the planned 2-year project should constitute: (1) To determine the optimal weight assigned to each considered clustered estimate under the MSE framework, and importantly to develop asymptotic theory of the optimal weights; (2) To develop the asymptotic theory of the proposed averaging estimator, and to derive its asymptotic risks; and (3) To examine the finite-sample properties of the estimator for different types of panels. | |
| dc.format.extent | 116 bytes | - |
| dc.format.mimetype | text/html | - |
| dc.relation (關聯) | 科技部, MOST105-2410-H004-009-MY2, 105.08-107.07 | |
| dc.subject (關鍵詞) | 均方差; 平均估計式; 合併估計式; 群組 | |
| dc.subject (關鍵詞) | mean-squared error; averaging estimator; pooled estimator;cluster | |
| dc.title (題名) | 群組平均--合併追蹤資料的方法 | |
| dc.title (題名) | Clustering Averaging as an Alternative to Pooling Panels | |
| dc.type (資料類型) | report | |
