2009-08-012024-05-15https://scholars.lib.ntu.edu.tw/handle/123456789/665701摘要：在本研究當中，我將利用隨機係數的離散選擇模型來分析選擇就讀高中的問題.。產業組織領域已發展出許多應用產品特徵變數的方法來分析產品差異化的產業，我將延伸既有的模型來探討教育部門。本研究的焦點將是在聯考制度之下的台灣高中入學選擇。主要的研究目標是希望可以運用學校層次的資料，來估計出影響個別學生選擇的 因素。 我的研究將會以Berry, Levinsohn 和Pakes 於1995 年發表在Econometrica 的架構來作延伸。在他們所研究的汽車市場，廠商從事價格競爭，因而各家廠商的市場佔有率可用來找出消費者對於各款汽車平均的效用水準。也就是消費者偏好可藉由市場佔有率來identify。相對地，在我所要探討的學校選擇問題，學校的招生名額是由政府官員所決定的，而不是由學生的偏好所決定。針對此特點，我將利用學校在聯考的最低錄取分數來估計出各校帶來的平均效用水準。在其它條件相同下，較高的錄取分數代表較高的效用水準。然而，學校的招生人數，也會影響到錄取分數，因此我將延伸Berry等人的計量方法來解決此問題。 我也會考慮到學生之間的異質性。隨機係數的模型可以讓不同學生對於不同的學校特質給予不同的重要性。此外，通勤的距離或許是選擇學校的重要因素，所以我會在估計中考慮學生居住地點的異質性。 我將要證明聯考的錄取分數可identify 學校的平均效用水準。我也會證明資料當中的變異性可以identify 我希望估計的異質性程度。<br> Abstract: In this project, I will empirically analyze the choice of attending a senior high school by using random-coefficient discrete choice models. Economists in industrial organization have developed econometric methods to study the discrete choice problem in a differentiated-product market by using product-level data. I plan to extend these methods to study the education sector. Specifically, I will focus on the school selection problem in Taiwan, where senior high school admissions are determined by scores in the joint entrance exam. The main objective is to recover the factors which affect a student’s school choice by using school-level data. My estimation approach will be an extension of Berry, Levinsohn, and Pakes (Econometrica, 1995). In their study of automobile choice, firms compete in prices, and market shares can be used to recover the mean utility level for each automobile. Therefore, consumer preferences are identified from market shares. However, in the school choice problem that I am going to study, the number of students admitted to each school is determined by government officials, not by students’preferences. Instead, the minimal required scores in the entrance exam can be used to recover the mean utility level for each school. Other things being equal, a school with a higher mean utility level would have a higher required score. Nonetheless, when two schools with the same mean utility level, the one admitting fewer students would have a higher required score. Consequently, to correctly identify the mean utility level of a school, I need to extend Berry et al.`s empirical framework to account for the number of students admitted. I will also consider the heterogeneity among students. A random-coefficient model allows different students may put different weights on a school characteristic. Moreover, since commuting distance might be an important factor in choosing a senior high school, I will also account for the heterogeneity of students` residences. I will prove that the observed required scores in the entrance exam can identify the mean utility level in my estimation. In addition, I need to show that the variation of coefficients across students can be identify from the data, and the disutility of commuting distances can also be identify from the variation in student population.學校選擇離散選擇隨機係數同儕效果空間競爭聯合招生考試school selectiondiscrete choicerandom coefficientpeer effectspatial competitionjoint entrance exam