方法对比

	矩估计分位数回归	Quantile ARDL	Quantile VAR
领域	计量经济学	计量经济学	计量经济学
方法族	Regression model	Regression model	Regression model
起源年份	2004	2006	2006
提出者	Roger Koenker and colleagues	Roger Koenker and Zhijie Xiao	Koenker and Xiao
类型	Distribution regression	Conditional distribution model	Distribution impulse response
开创性文献	Koenker, R. (2004). Quantile regression for longitudinal data. Journal of Multivariate Analysis, 91(1), 74-89.	Koenker, R., & Xiao, Z. (2006). Quantile autoregression. Journal of the American Statistical Association, 101(475), 980-990.	Koenker, R., & Xiao, Z. (2006). Quantile autoregression. Journal of the American Statistical Association, 101(475), 980-990.
别名	GMM quantile regression	Quantile ARDL	Quantile-based impulse response
相关	3	3	3
摘要	Method of Moments Quantile Regression combines moment-based estimation (GMM) with quantile regression to estimate distribution parameters while handling endogeneity, panel structure, and dynamic relationships. Introduced by Koenker (2004) and developed by Machado and Mata (2005), it enables distributional analysis (not just mean regression) in complex settings like dynamic panels and instrumental-variable contexts. This approach is powerful for understanding heterogeneity in treatment effects and policy impacts.	QARDL (Quantile Autoregressive Distributed Lag) combines quantile regression with ARDL modeling to estimate conditional relationships at different points of the distribution, revealing heterogeneous short-run and long-run effects. Introduced by Koenker and Xiao (2006) and refined by Cho et al. (2015), it captures how the effect of explanatory variables on outcomes varies across quantiles, essential for understanding tail behavior and distributional impacts rather than just mean effects.	Quantile VAR estimates impulse responses of multivariate systems conditional on different quantiles of the distribution, revealing how shocks propagate heterogeneously across the conditional distribution. Introduced by Koenker and Xiao (2006) and applied to risk measurement by White et al. (2015), it reveals tail behavior and contagion effects invisible to mean-based VAR analysis. This is essential for risk management and understanding how crises propagate differently than normal times.