So that still assumes that saving is risk-free (maybe) and the discount rate people apply to consumption is 0, right? The latter just doesn't match reality.

But it would definitely be a step up from where we are now, in theory. In practice, measuring inflation is an interesting exercise; there are multiple measures that differ quite a bit once your timescale is "decades" as it can easily be for savings and housing. And yrs, I know I said "the rate of inflation" -- I forget which measure those studies used. For housing specifically, one could use whichever inflation measure best matched housing prices in the past, I guess...