Does including more structural shocks help mitigate model misspecification problems over periods like great financial crisis and COVID19?

[jpfeifer]

I don’t know a definitive answer, but let me share my thoughts.

Your question is already full of inherent tensions. If there has not been a structural break, then the shocks you are considering to add were present all time. Given that you did not include them before, your model was misspecified all the time. For example, you could argue that there has indeed not been a structural break, rather there was always disaster risk, with a low probability of large events and that these events simply did not materialize before. Now for the first time we have such an event. But that implies that a disaster risk model would have been appropriate the whole time.

Now the question is how to explain the Corona crisis. You could go the route of shocks simply being bigger as previously discussed in

Including observable data during COVID19 period into DSGE model

For the GFC, there is for example the Stock/Watson paper (https://www.brookings.edu/bpea-articles/disentangling-the-channels-of-the-2007-2009-recession/) arguing that there was no structural break, just a bigger shocks. It is hard to argue the same for the COVID19-period. Also, the current GDP drop is bigger and faster than previous ones. With normally distributed shocks, this should pretty much never happen. Thus, these outliers will massively affect estimation.

But then the question is whether these are simply realizations from the same distribution of whether the distribution has changed to make big shocks more likely. In the latter case, there is a structural break.