Basel III allows for the use of statistical models. It is called the internal-ratings-based (IRB) approach and is based on the (Vasicek, 2002) model. It assumes assets returns are standard normally distributed. It suggests incorporating different asset correlation (R) functions to assess credit risk for the loan portfolio, or the risk-weighted assets (RWA). The asset correlation function solely depends on the individual default probability (PD) given certain credit exposure type. At the same time, the IRB approach requires developing PD models to predict the discrete default event occurrence. This means that the IRB approach is based on the Bernoulli trials. We investigate the impact of the asset returns’ correlation for the Bernoulli trials. We show that when Bernoulli trials are considered, the credit risk estimation significantly deviate from the values derived under the normality assumption of asset returns. We investigate the simulated and real-world credit rating agencies’ data to specifically demonstrate the scale of the credit risk underestimation by the IRB approach. Therefore, macroprudential add-ons are of use to offset such IRB limitations.