Key words: Quantile regressions, causal and non-causal model selections, VaR modelling, inference testing on CAViaR models, measure systemic risk, MVMQ CAViaR models

"Essays in quantile regression models and their applications to financial time series"

This research studies financial time series by quantile regressions. Quantile regressions are used to study a particular quantile of a specific interest. This is a question that might be asked when a particular quantile is addressed. When there is an interest in an outcome and it is known that the outcome is an uncertainty, the objective is to find a threshold of this random outcome for securing the position when it happens. It is well known that financial return time series exhibit unconditional and conditional heavy tails (like bubble patterns), volatility clustering and time-varying cross-correlations which are researched in this PhD using quantile regressions respectively. And this PhD research has contributed to causal and non-causal model selection, inference testing on quantile (or value-at-risk) regressions, measuring systemic risk of big financial institutions.

Click here for the full dissertation.