Calculations of the Value at Risk for a Diversified Portfolio

This report presents the work produced during a five-and-a-half-months curricular internship at the insurance company Fidelidade, in the Asset and Liability Management team, which is part of the Risk Management Department. The main goal of the internship was to develop an R model that would allow the Value-at-Risk (VaR) of Fidelidade’s portfolio to be calculated efficiently, given the complexity of the calculation required for a diversified portfolio. In order to improve the company’s risk management methodologies, the process of developing the R model began with research into the main methodologies for calculating the VaR. Given the characteristics of Fidelidade’s portfolio, the Monte Carlo simulation method was chosen. Since fixed income instruments have specific characteristics, we used a methodology different from the “VaR mapping” approaches (also analysed), called the “New approach”, so that the VaR calculation could be similar to that applied to other asset classes. Due to time constraints, asset classes such as derivatives, participations and strategic investments were not included in the portfolio analysis. We developed the model in RStudio, using one year of historical data with a reference date of 31st of December 2021. The model was first tested by asset class and then applied to the entire portfolio. To validate the results, we used the “PORT” tool on the Bloomberg platform, which produced results close to our own, thus validating the performance of the model created to a certain extent.