Calculating Tail Value at Risk Using a EGARCH-Extreme Learning Machine Model And The long-term forecast approach in the insurance industry

One of the most important methods for market risk measurement is Value-at-risk (VaR) that financial institutions such as banks, insurers and investment funds use them extensively. VaR as a risk measure is heavily criticized for not being sub-additive, thus the researchers focuses on the assessment of the Tail value-at-risk (TVaR), and this measure is using on the Basel Committee on Banking and Solvency II of Europe and Swiss Solvency Test (SST). this paper focuses on TVaR to measure the risk of the stock market. Considering that the time horizon of the risks of an insurer unlike banks is annually. thus, to calculate the TVaR, we use of the two methods of the variance-covariance approach with the EGARCH-Extreme learning Machine model to volatility forecasting and use of square-root-of-time rule; and Filtered Historical simulation model. The results of using the daily returns of the Tehran Stock Exchange Index for 1388 to 1396 confirm that the EGARCH-Extreme learning Machine model with use of square-root-of-time rule performs better in TVaR calculation in terms of efficiency and accuracy.