The economic system is a non-linear dynamical system that is always evolving. What exists today could look dramatically different in a matter of years as banking evolves, the economy expands, innovation changes, etc. An economic model might apply in certain environments and be totally irrelevant in a different environment. This is one aspect of economics that makes it all so difficult to model present and future conditions.
I bring this up on the back of a discussion that was started by Noah Smith in which he took some of the “crisis predictors” to task. I think Noah makes some excellent points. But I also think it’s important to understand and accept the simple reality described above. No one’s model is perfect or always able to explain the world we have or will have because that world is evolving much faster than anyone’s model can conform to it.
The good news, however, is that we can and should quickly dismiss the models that have proven themselves largely erroneous. For instance, the models that predicted the “USA will default” or that “hyperinflation is coming” were proven thoroughly wrong over the last 5 years. If you understood that banks don’t lend reserves then you understood why QE wasn’t going to cause hyperinflation. If you understood how Greece’s monetary system was different than the USA’s then you knew that the USA was unlikely to default. Some models predicted these things using a sound understanding of the system. Others did not and they have been largely discredited as a result.
All of this has been a good thing because it means we can learn from past mistakes. And as the system grows and evolves our models should evolve, grow and improve from these mistakes. It’s unlikely that anyone’s model will always keep up with such a fast evolving system, but it’s becoming clear that some models have served us better than others. And so the very worst thing we can do is allow ourselves not to learn from the past and improve the models we work with so that we can better understand the monetary system and more closely model what might occur in the future.