Chartered Financial Analyst (CFA) Practice Exam Level 2

Which statistical feature is typically NOT assumed in VAR analysis?

1. Returns are normally distributed

2. Returns are correlated

3. Returns are serially independent

4. Returns are linearly related

The correct answer is: Returns are linearly related

In Value at Risk (VaR) analysis, the typical assumptions include the distribution of returns, correlation among returns, and the independence of returns over time. While many asset return models assume a normal distribution of returns and acknowledge that returns may be correlated, they often also assume that returns are serially independent. The assumption that returns are linearly related is not a standard characteristic in VaR analyses. This is because VaR focuses primarily on the magnitude of potential losses at a certain confidence level without necessitating a linear relationship among returns. The implication is that risk can be effectively measured without relying on a linear model, allowing for a broader application of the VaR methodology across various asset classes and market conditions. Thus, while returns' distribution, correlation, and independence are significant in the context of risk assessment, the linear relationship is not typically assumed in VaR analysis, making it the identified statistical feature that does not hold as a foundational assumption.