Applied micro ... *sigh*
You spend your time pretending to obtain clean, causal identification. We all know this is not reality. And, to be frank, the very best scenario you can achieve is a natural experiment. Yet, these natural experiments are, in many or most cases, no more externally valid than well-designed lab or field experiments. Yet, most of you bash experiments and hold yourself somehow above experimentalists.
The saddest part is that the entire modern applied micro research program implies that over a century of economic theory didn't tell us anything useful to put any kind of structure behind observational data. Instead we just have to wait and cross our fingers until nature hands us a "quasi-experiment" to run ivreg over.