The financial sector is typically simplistically represented in DSGE (distributed stochastic general equilibrium) macroeconomic models, sometimes as simple as plug variables that assumes perfect competition and deny the possibility of a financially induced recession. Macroeconomics as a discipline has a decidedly poor forecasting track record, Noah Smith, who among other things writes for Bloomberg, has a good deck on the weak state of macroeconomics and forecasting. Yet the regulators and central banks, who themselves are miles away from usefully connecting the real and financial economies, are imposing this on the stress testing regime. Just have a look at how low the R-squareds are in the CCAR world, a solid econometrician would laugh at it. Good luck.