>evaluate what up to a 20% increase in housing supply would bring to any one given area and they never say it would fall by anything more than 2%
Hold on. That study is about an increase in the housing stock of a single neighborhood in a city, only for the top decile of expensive neighborhoods, and the resulting effect on rents within that same neighborhood. Its counterfactual is somehow taking the entire private expenditure used to build that housing and putting into public works projects in the bottom 9 deciles of neighborhoods in order to attract demand there instead.