The frequent occurrence of financial crises and specifically the 2007-08 subprime global financial crisis brought a series of major macroprudential reforms in the regulation of financial institutions and markets. Among such reforms is to periodically gauge the resilience of institutions through stress testing and to mitigate system-wide risks. Stress testing is a risk management process in which various historical and hypothetical shocks are applied to test the resilience in worst-case scenarios. Over a period, there have been major advances in global financial markets and increased complexity in financial instruments and risk identification. As a result, stress testing methodologies have also improved from basic sensitivity analysis to sophisticated partial equilibrium models. The study employs the Dynamic Stochastic General Equilibrium (DSGE) model calibrated for Pakistan using the quarterly data for 02:2002 - 02: 2018. The DSGE model constructed three shock scenarios of technology, demand preference, and monetary policy and produced stressed forecasts. These forecasts were later used in the VAR model to come up with the future path of credit risk indicator (GNPLR) of the banks. Similarly, the forecasts based on VAR was also computed. It was found that the DSGE based stressed forecast of GNPLR tends to be slightly higher in the initial quarter and as time horizon increases, the deviation between forecasts tends to grow. It is expected that as DSGE models grow in coverage and include financial sector and risk transmission channels, there will be greater acceptability of these sophisticated measures in stress testing the banking sector