Quantile autoregression with censored data

Quantile autoregression (QAR) provides an alternative way to study asymmetric dynamics and local persistence in time series. It is particularly attractive for censored data, where the classical autoregressive models are unidentifiable without further parametric assumptions on the distributions. There have been prominent works by Powell (1986), Portnoy (2003) and Peng and Huang (2008) on estimating the conditional quantile functions with censored data. However, unlike the standard regression models, the autoregressive models should take account of censoring on both response and regressors. In this dissertation, we show that the existing censored quantile regression methods produce empirically consistent estimator on QAR models when using only observed part of regressors. A new algorithm is proposed to improve a censored quantile autoregression (CQAR) estimator by adopting an idea of imputation methods. The algorithm distributes probability mass of each censored point to any sufficiently large value appropriately, and iterates towards self-consistent solutions. Monte Carlo simulations are conducted to examine the empirical consistency of the CQAR estimator. Also, empirical applications of the algorithm to the Samish river water quality study and dry decomposition of NH4 demonstrate the merits of the proposed method.

Name thesis:degree_name

Ph.D.

Level thesis:degree_level

Dissertation

Discipline thesis:degree_discipline

Statistics

Grantor

University of Illinois at Urbana-Champaign

Year dc:date

2014

Author dc:creator

• Choi, Seokwoo

Contributors dc:contributor

• Portnoy, Stephen L.
• Monrad, Ditlev
• Koenker, Roger W.
• Chen, Xiaohui

• Censored time-series
• Autoregression
• Quantile
• Self-consistent
• Kaplan-Meier

Statement dc:rights

• Copyright 2014 Seokwoo Choi

Language dc:language

en

Handle dc:identifier

http://hdl.handle.net/2142/50583

OAI identifier oai:identifier

oai:www.ideals.illinois.edu:2142/50583