Boundaries of Correlation Adjustment with Applications to Financial Risk Management

In recent years, stress testing has become a regulatory requirement for risk assessment in financial institutions. To perform stress testing in multi-asset case, adjusting the correlation matrix to an extreme level is an important process. With a larger matrix, it is more difficult to choose the right correlation coefficients such that the newly adjusted correlation matrix is still valid. In this paper, we present a systematic way to obtain the boundaries of a correlation matrix for both single stress and multiple stress cases, which can help determine how much the correlation should be adjusted in the first place