This forthcoming AER paper uses housing price shocks to study the effects on corporate investment. One reason that it is particularly good is that it uses datasets everybody else uses. I find it very valuable if a paper can use “old” data to say something new. What’s a bit surprising is that they employed a wrong 2SLS: they forgot to include the controls in the second stage in the first stage regressions. Don’t know why no referee pointed this out.
As mentioned in this paper, the first papers using this idea are done by Jie Gan, now working at HKUST.