異常值可以對傳統經典的統計量產生相當大的影響，並導致對變量及變量之間關繫的分析發生偏差，從而得出錯誤的結論。不僅傳統的統計量。如何在異常值存在與否的情況下都能獲得更加穩健的結果，已經吸引了大量研究人員的興趣，並且已經發表了大量有影響力的文獻。多種穩健回歸方法被學者們提出來。分位數回歸，作為LAD從中位數向不同分位數的擴展，首先由Koenker和Bassett（1978）提出。由於它的穩健和有效性，同時允許研究人員不僅在中心而且在因變量的整個條件分布上研究經濟變量之間的關繫等優點，分位數回歸被應用於經濟和金融等許多學術領域。
《利用分位數回歸的統計與經濟分析（英文版）》對於在金融風險存在時穩健統計量計算的投資組合的表現進行了研究，並對分位數回歸的理論和應用進行了研究，基於分位數回歸進一步分析我國省際數據下以及86個非石油國家的經濟增長趨同性，外國直接投資對增長的影響以及金融風險測量，以及風險度量等。
《利用分位數回歸的統計與經濟分析（英文版）》讀者適合為經濟學專業高年級本科生及研究生。

霍麗娟，任職於北京理工大學人文與社會科學學院經濟繫講師。作者2013年初博士畢業於韓國延世大學經濟學繫，主要研究方向為計量經濟學、分位數回歸等。在博士導師Kim Tae-Hwan教授的指引和引導下，博士階段便對穩健統計和穩健回歸的相關研究產生興趣，並在畢業後仍在穩健統計和分位數回歸的理論和影響方面進行進一步研究。2017年作者主持的課題《基於高維VAR分位數回歸的繫統性金融風險測度》獲得教育部人文青年基金支持，2019年作者主持的《馬爾科夫區制轉換分位數回歸模型研究》獲得國家自然科學基金青年項目的支持。

• 
• 
• 
• 

Chapter 1 Introduction
1．1 Overview
1．2 Quantile Regression and Its Applications
References

Chapter 2 Robust Statistics and Robust Regressions
2．1 Introduction to Classical and Robust Approaches to Statistics
2．2 Least Squares Linear Regression
2．3 Robust Regression
2．3．1 Least Absolute Values Regression
2．3．2 M-estimator
2．4 Quantile Regression
2．4．1 Quantile Regression Model
2．4．2 The Finite-sample Distribution of Regression Quantiles
2．4．3 Quantile Regression Asymptotics
2．4．4 Wald Tests
2．4．5 Estimation of Asymptotic Covariance Matrix
2．4．6 Quantile Likelihood Ratio Tests
References

Chapter 3 Robust Estimates of Covariance
3．1 Conventional Measure of Covariance
3．2 Robust Measures of Covariance
3．2．1 Median Absolute Deviation About the Median （MAD）
3．2．2 Gnanadesikan and Ketenring Robust Measures of Covariance
3．2．3 M-estimates
3．2．4 Minimum Volume Ellipsoid Estimate (MVE）
3．2．5 S-estimates
3．2．6 Minimum Covariance Determinant Estimate (MCD）
3．3 An Alternative Robust Measure of Covariance
3．4 Monte Carlo Simulations
3．5 Empirical Application
3．5．1 Empirical Comparison of Robust Estimates
3．5．2 Portfolio Performances of Robust Covariances
3．6 Conclusion
3．7 Appendix： Derivation of Conventional Covariance with Outlier(s）
References

Chapter 4 Quantile Regression Serial Correlation Tests
4．1 Spurious Autocorrelation in Quantile Models
4．1．1 Standard LM Test for Linear Model with AR(p） Errors
4．1．2 Theoretical Explanation to the Occurance of Spurious Autocorrelation
4．2 Correctly-sized Tests
4．2．1 QF test
4．2．2 The QR-LM Test
4．3 Monte-Carlo Simulations
4．4 An Empirical Example
4．5 Conclusion
4．6 Appendix
References

Chapter 5 Growth Empirics Based on IV Panel Quantile Regression
5．1 Economic Growth Convergence
5．2 Quantile Regression for Panel Data Model with Fixed Effects
5．3 Growth Convergence at the Conditional Mean
5．4 Growth Convergence at Different Conditional Quantiles
5．5 Empirical Results from 86 Non-oil Countries
5．5．1 Data and Samples
5．5．2 Empirical Results
5．5．3 Conclusion
5．6 Evidence from China Provincial Panel Data
5．6．1 Literature on China's Regional Economic Development
5．6．2 Model and Data
5．6．3 Empirical Results
5．6．4 Conclusion from China's Empirical Results
References

Chapter 6 The Impact of FDI on Economic Growth： an Empirical Evidence from IV Panel Quantile Regression
6．1 FDI and Economic Growth
6．2 IV Quantile Regression Model for Panel Data with Fixed Effects
6．3 Data and Empirical Results
6．4 Conclusion
6．5 Appendix
References

Chapter 7 Financial Risk Measurement： CoVaR
7．1 Financial Risk Transition Mechanism and Source of Risk in China
7．1．1 The Transmission Mechanism of Financial Risk in China
7．1．2 Sources of Financial Risk in China
7．2 Risk Measurements： VaR， CoVaR， and △CoVaR
7．2．1 Definition of VaR
7．2．2 Calculation of VaR
7．2．3 Definition of CoVaR and △CoVaR
7．2．4 Calculation of CoVaR
7．2．5 CoVaR Model Based on Quantile Regression
7．3 Empirical Study on Systemic Financial Risks in China
7．3．1 Data Selection
7．3．2 Data Processing and Descriptive Statistics
7．3．3 Identification of Systemically Important Financial Institutions
7．4 Static Risk Contribution of Financial Sub-industries on Financial System
7．4．1 Data Selection
7．4．2 Data Processing and Descriptive Statistics
7．5 Risk Spillover Effects Between Financial Sub-sectors
7．5．1 Static Risk Spillover Effects Between Financial Sub-sectors
7．5．2 Dynamic Risk Spillover Effects Between Financial Sub-industries
7．6 Conclusion
References

Chapter 8 Markov Regime Switching in Quantile Autoregression Stock Market Return Model
8．1 Introduction to Markov-switching model
8．2 Markov-switching Quantile Autoregressive Model for Stock Market Returns
8．3 Data Description and Empirical Results
8．4 Conclusion
References

Due to measuring errors or unexpected phenomena， such as
earthquakes，strikes，and financial crises ，etc，outliers occur very frequently in real data and may have considerable influences on the classical estimates like the sample mean，sample variance，and mean-based least squares estimation as well.Robust estimates and regressions， which provide a good fit to data while the data contains outliers，have been developed a lot and applied in many fields.Among robust approaches， quantile regression（QR），as an extension of least absolute deviation (LAD）at the median to various quantiles，was proposed firstly by Koenker and Bassett(1978） and has been applied in many academic fields， such as economics and finance.Quantile regression method has a series of advantages.It can get more robust and efficient estimates，and provide a new perspective of allowing researchers to investigate the relationship between economic variables over the entire conditional distribution.
I started to be interested in this field from my doctoral study.At that time，l did preliminary researches on the robust covariance and robust regression，and realized that the quantile regression theory and its application in the fields of economics and flnance will be important and valuable.And then， further research was conducted on the quantile regression under the support of Humanities and Social Sciences Fund of the Ministry of Education，grant 17YJC790057 and National Natural Science Foundation of China， grant 71803009.This book， contauung some of my research on the robust statistics and quantile regression，first compares several robust covariance measurements and their performances of portfolios.This book also applies the robust regression，quantile regression，to investigate growth convergence in China and 86 non-oil countries， FDI effects on growth，financial risk measurements，and stock market regime-switching issue.
Especially during the financial crisis， the volatility of the economic and financial related data is quite abnormal， which brings challenges to investment decision-making using classical statistics and regression estimates based on the mean.At the same time，in the study of practical problems，researchers and policymakers are usually more concerned with the tail of the
overall distribution，such as factors that affect economic growth in less developed and more developed regions.In order to better capture the tail features and obtain more robust estimation results，quantiles and quantile regression have achieved more and more attention.This book has made some explorations and attempts in the quantile regression theory and its application in the economic and financial fields，and I expect more research results coming out in the future.I am very grateful to Tae-Hwan Kim，my Ph.D.advisor，who led me into Econometric study and encouraged me to research on quantile regression.l thank Yunmi Kim，Dongjin Lee，Jinseo Cho，Weiming Li，and Yuchen Jin for their help and efforts in our collaborative work.This book could not have been written without the support of my family.Untold thanks to Jun Kong.