MODELING AND EVALUATION OF RISK MEASURES FOR THE RESIDUALS OF FINANCIAL TIME SERIES WITH UNOBSERVED VALUES USING R

MODELING AND EVALUATION OF RISK MEASURES FOR THE RESIDUALS OF FINANCIAL TIME SERIES WITH UNOBSERVED VALUES USING R

ABSTRACT

In this work ARMA+GARCH model is adopted for the daily stock price of First Bank Nigeria, Plc. The methods of weekly average, regression imputation and repetition were used in computing the unobserved values. An alternative method was also used which involves deleting the days with unobserved values. The method of transformation was determined in each of the series and log transformation was adopted for the four series. In the model selection, the model of the repetition had the minimum AIC. The distribution of the residuals was found to be Frechet case of Generalized Extreme Value Distribution (GEVD). This was approximated to the Generalized Parato Distribution (GPD). The parameters of this distribution were used in computation of risk measures. The computed Value eat Risk (VaR) has a value of N49438.79 and that of the Expected Short Fall (ES) is N49291.24 with position of N1,000,000.