QUANTILE REGRESSION: CONCEPT AND RISK MEASURE PREDICTION

• Main Author: Pusparani Wibisono, Aqilah
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/47708
• Institution: Institut Teknologi Bandung

Time series often exhibit distinct changes in data pattern. Data with changed pattern can be modeled with markov switching regression. Markov switching model is dynamic regression whose switching between regimes follows the markov process. The expansion of markov switching model is markov switching quantile regression model which is a dynamic quantile model. A special form of markov switching quantile regression is quantile regression. Quantile regression can provide a model for each quantile, thereby the picture of relationship between random variables studied is more complete. Quantile regression has properties that are an advantage of quantile regression namely robustness and the absence of distribution assumptions. Quantile regression is the right choice to predict quantile risk measures, particularly in Covid-19 risk prediction. The most commonly used risk measure is Value-at-Risk (VaR). VaR prediction can be done with a quantile regression models including QAR,QARCH, dan CAViaR. VaR prediction with quantile regression model was applied to nancial and health data. In the insurance sector, quantile regression can be used to determine premiums and reserves.