The potency of time series outliers in volatile models: An empirical analysis of fintech, and mineral resources

Understanding the importance of sustainability, the increasing ecological concerns need resource management under the cumulative mineral dependency. In this vein, prices of financial technology and mineral resources influence environmental sustainability; hence, sudden price variations must be evaluated carefully. For this, detecting outliers in financial and economic variables is gaining attention as economies are regionally integrated and influenced by external factors. Such peripheral events highly influence the statistical properties of model parameters and error distributions. In this vein, the current study proposed the method of outlier detection in various time series volatile models by extending the procedure of Frances and Ghijselse (1999) based on the framework of Chen and Liu (1993). The method is based on three phases: First, the outliers are detected, then the outliers are removed and estimated using the linear interpolation method, and third, the data values are forecasted by re-estimating the model with a corrected series. Forecasting accuracy has also been compared with and without outlier series to observe the impact of outlier presence in the series. For empirical analysis, the S&P Kensho Democratized Banking Index (USD), S&P 500 Environmental and Socially Responsible Index, and Gold price per Troy Ounce from July 15, 2013, to July 14, 2023. The outcomes retrieved from the proposed method suggest that the error distribution of volatile models is affected by outliers and manifestly forecasting performance is better with outlier corrected series. The diagnostic of the model is also analyzed. The study presents robust policy implications for outlier detection procedures in volatile models.