Which is perfectly controllable if your model includes delayed feedback. Unfortunately, the dominant macroeconomic models contain neither delayed feedback nor nonlinearity, and are not capable of simulating recessions without resorting to arbitrary "exogenous shocks".

In contrast, Steve Keen is actively developing dynamic nonlinear models of macroeconomics that is actually capable of simulating endogenous recessions.

If you're interested in learning more about Steve Keen's work, I highly recommend watching his presentation to the OECD: https://www.patreon.com/posts/my-speech-to-new-14735058