The article doesn't mention it - and I can't find it anywhere - but in any neighborhood what is the highest proportion of units that are short-term rentals. Five percent? Ten percent in the super touristy areas? I have no idea and even before the other flaws in the study I am skeptical than anything short of 20% would make an impact on any major attribute of an area. I see one stat in the paper that "40% of buildings…
With regards to your "alternate theory" here's a relevant line in the paper. > To further test the direction of causality for the results, we use a lag/lead analysis in the spirit of Granger [33, 34]. This method is used when the sample includes multiple years and uses both lead and lagged versions of the treatment variable (τ can be both positive and negative). I don't have enough experience in econometrics/statisti…
For example, that type of analysis might suggest that the direct crime -> airbnb increase link is unlikely. However, it's possible that there's some other process that's more like simultaneous perceived neighborhood quality change -> airbnb increase and neighborhood quality change -> crime increase (basically third variable change).
Reading this paper I was persuaded that tourism per se is maybe not the causal factor (although not convinced because their measures -- reviews -- were a little odd). However, I'm less convinced that airbnb rental per se is the causal factor. Maybe more closely related to whatever the causal factor(s) is(are) but not necessarily it.