《不同类别的公共投资对所得分配之影响-分量回归模型的应用》由会员分享，可在线阅读，更多相关《不同类别的公共投资对所得分配之影响-分量回归模型的应用（59页珍藏版）》请在金锄头文库上搜索。
1、 國中興大學應用經濟學研究所 碩士學位文 同別的公共投資對所得分配之影響 分迴歸模型的應用 The Impact of Different Category Public Investment on Income DistributionAn Application of Quantile Regression Model 指導教授：建強 Chien-Chiang Lee 研 究 生：永 Yung-Lung Liang 中華民國九十九一月 謝謝 辭辭 時光荏苒，歲月如梭，回首自己前順考上中興大學，完成人生 中未盡的夢想，如今這的研究生涯即將隨著本篇文的完成而劃下完美 的點，心中喜悅之情，溢於言
2、表。 文能順完成，尤其要感謝指導教授建強博士。因為有師適時又 明確的指導，讓學生得以能夠正確找尋出研究方向，師悉心指導僅排解 學生心中疑惑及研究瓶頸，巨細靡遺地逐字逐修改，讓本篇文用字遣詞 得以為完美。此外，由衷感謝口試委員簡美瑟博士與許義忠博士，在口試 過程中對於本研究內容的指正與建議，使整篇文內容臻完善。 的研究生涯，感謝系上所有師們，在學習過程中所給予我的指導 與幫忙，要謝謝同學宏绮時地鼓與協助，進富在電腦軟體及操作應用 的傾囊相授，讓我在文或學習上獲多。 最後要感謝我最摯愛的母親范春美士，感謝您含辛茹苦地扶養我長 大，謝謝您在我人生每一階段的極栽培，提供我接受好的教育，讓我 順地考上公
3、職及研究所，雖然求學過程中難免遇到挫折，但有您的支持與 鼓，讓我能克服萬難並心無旁鶩的專心於研究所課業上，以完成研究所的 學業，在此僅以此文獻給我敬愛的師長與家人，一起分享此份喜悅及榮耀。 永謹誌於 國中興大學應用經濟學系研究所 中華民國 99 1 月 i 摘摘 要要 本文採用 1972 至 2004 共 29 國的公共投資及所得的追蹤資(panel data)，用 Koenker and Bassett (1978)的分迴歸法(Quantile Regression，簡 稱 QR)探討公共總支出、公共教育支出、公共保健支出、公共支出及 公共安全支出與所得條件分配之間的關係。使用分迴歸法可以分
4、析在同 分(quantile)之條件分配下，上述五個公共投資對於所得水準的影響。本文 主要的貢獻在於，以全球的觀點解釋同別的公共投資是否影響所得條件 分配？各國政府應致於哪種別公共投資才可提升所得水準，以低國際 間的均？ 本研究的實證結果顯示：一、於最低分(0.1)及最高分(0.9)，公共總 支出(後 2 期)分別對所得呈現顯著地正向及負向的影響，這表示最低所得 水準國家，公共總支出的貢獻較大，有助於提升所得水準。二、於各分下， 公共教育支出，對所得影響效果並顯著；惟在普通最小平方法下，公共教 育支出(當期)對所得呈顯著正向關係，這代表平均而言，公共教育支出對各 種所得水準國家皆有貢獻，其能有
5、效提升所得水準。三、於中低(0.25)及中 高(0.75)分以上，公共保健支出(當期)對所得皆呈現負向的影響，惟考 後效果之公共保健支出對所得之影響並無顯著性。四、於中低分(0.25)以 下及最高分(0.9)，公共支出(當期)，對所得有顯著的正向影響，惟於 中低分(0.25)以下，公共支出(後 2 期)，對所得則呈現顯著的正向關 係，這表示中低所得水準以下及最高所得水準國家，公共支出的貢獻較 大，可有效提升所得水準。五、於中分(0.5)，公共安全支出(後 2 期)對 所得呈顯著正向影響，由此可，於中所得水準國家，公共安全支出的貢獻 較大，有助提升所得水準。最後，我們考慮解釋變可能具有內生性的問
6、題， 採用階段最小平方法修正變內生性的問題，其估計結果均會改變本 文的結，且支持實證結果的穩健性。 關鍵詞：關鍵詞：分迴歸、公共投資、所得條件分配、追蹤資。 Keywords ： Quantile regression, Public investment, Income conditional distribution , Panel data. JEL 分代號：分代號：C21, H54, O15, C23 ii Abstract In this paper, we used a panel data set of 29 countries in public investment and
7、income from 1972 to 2004, which employed quantile regression approach (abbreviated QR), based on Koenker and Bassett (1978), to explore the relationship of the total public expenditure, public expenditure on education, public health expenditure, public welfare spending, public safety expenditure wit
8、h the income conditional distribution. Quantile regression approach can analyze the effects of the above five categories of public investment on income across quantiles in the conditional distribution. The main contribution of this paper is that a global point of view to explain whether the differen
9、t types of public investment affect the income conditional distribution. In order to reduce international inequality, governments should devote to which categories of public investment that can raise income level. The empirical results of the paper is showed as follows. First, total public expenditu
10、re which has two lags is presented significantly positive and negative effects on income at the lowest quantile (the 0.1 quantile) and the highest quantile (the 0.9 quantile) . This means that total public spending will help raise income level in the minimum income level of countries, which has larg
11、er contribution. Second, the effect of public education spending on income is not significant at the 0.1, 0.25, 0.5, 0.75, and 0.9 quantiles. However, in the ordinary least squares, current public expenditure on education has significant positive relationship with income. This means that, on average
12、, public expenditure on education to the countries possess the contribution of various income levels, which can effectively enhance the income level. Third, current public health spending has negative effect on income at the 0.25 quantile and the above of the 0.75 quantile. However, consider the lag
13、ged effect of public health spending on the income is not significant. Fourth, current expenditure on public welfare has a significant positive impact on income below the 0.25 quantile and the 0.9 quantile. However, the public welfare spending which has two lags is presented significantly positive r
14、elationship with income below the 0.25 quantile. This means that public welfare spending has larger contribution to lower and the iii highest income levels of the countries, which can effectively enhance the income level. Fifth, public safety expenditure which has two lags is presented significantly
15、 positive effect with income at the 0.5 quantile. Thus, public safety expenditure has larger contribution to the middle income level of the countries. Finally, we consider the explanatory variables may have endogeneity problem, using a two-stage least squares method to modify the endogenous variable
16、s in the problem, the estimated results will not change the conclusions of this article, and the support of the robustness of empirical results. Keywords ： Quantile regression, Public investment, Income conditional distribution , Panel data. JEL classification：C21, H54, O15, C23 iv 目目 摘 要摘 要i Abstract.ii 目 目 .iv 圖目圖目
