| dc.contributor | 國貿系 | |
| dc.creator (作者) | 郭炳伸 | |
| dc.date (日期) | 2017-10 | |
| dc.date.accessioned | 7-Apr-2026 13:28:40 (UTC+8) | - |
| dc.date.available | 7-Apr-2026 13:28:40 (UTC+8) | - |
| dc.date.issued (上傳時間) | 7-Apr-2026 13:28:40 (UTC+8) | - |
| dc.identifier.uri (URI) | https://ah.lib.nccu.edu.tw/item?item_id=181948 | - |
| dc.description.abstract (摘要) | 本研究計畫嘗試以間接推論方法建構在弱工具變數環境下,具有近似不偏性質迴歸係數估計式。工具變數估計式在小樣本常有偏誤,該偏誤在工具變數多或弱時更形惡化。根據前導模擬證據,本計畫建構之估計式不止可以在前述環境下大幅降低偏誤,甚或在迴歸模型無法 「辨識」下亦是如此。相對地,既存的工具變數估計式在前者環境中的表現,正如同最小平方估計式間具有同樣的偏誤。本計畫建構的間接估計式顯然具有非常不一樣的性質。本計畫正是以探討這樣統計性質為主要研究目的。在預期的三年計劃中, 主要工作將包含:(1)發展此一新估計式之漸近理論, 並建構相對應的檢定方法;(2)推導與檢視該估計式的有限樣本性質,尤其在無法辨認的模型下;與(3)將該估計方法與理論推廣至更一般的環境中。 | |
| dc.description.abstract (摘要) | The project proposes an indirect inference approach to yielding approximately unbiased estimator of the regression coefficients in the context of many weak instruments. It is known that the instrumental variables estimation suffers large bias in small samples that can sub-stantially deteriorate with more or/and weaker instruments. Some pioneering simulation reveals that the suggested indirect inference estimator in the project can greatly reduce the bias not only in the situation but also in the case where the regression is unidentified. While the existing estimators are capable of reducing bias in the former context, they remain to exhibit the same bias property as the least squares estimator does in the latter. The pro-posed indirect inference 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 the statistical properties of the proposed estimator. Major investigations of the planned 3-year project should constitute: (1) To develop the asymptotic theory of the indirect inference estimator; (2) To examine the finite-sample prop-erties of the estimator, in particular for models that are not identified; and (3) To extend the results obtained from (1) and (2) to more general contexts. | |
| dc.format.extent | 116 bytes | - |
| dc.format.mimetype | text/html | - |
| dc.relation (關聯) | 科技部, MOST103-2410-H004-019-MY2, 103.08-105.07 | |
| dc.subject (關鍵詞) | 工具變數估計式; 偏誤; 間接推論; 弱工具變數 | |
| dc.subject (關鍵詞) | instrumental variable estimator; bias; indirect inference; weak instrument | |
| dc.title (題名) | 工具變數迴歸之間接推論 | |
| dc.title (題名) | Indirect Inference for Instrumental Variables Regressions | |
| dc.type (資料類型) | report | |
