Hey Ken, thanks for the comment. The following link shows the National Home Price Index published by S&P/Case-Shiller and estimated rent of primary residence over time (https://fred.stlouisfed.org/graph/?g=jdQD generated by FRED). Price returns have tended to be much more volatile than rents generated. Some of this disparity can be associated with lower overall resolution of estimating rents (rent data is harder to generate so variance in rents may be missed). However, we don’t believe this tells the whole story.

Campbell and Shiller published an important paper studying the properties of price-to-dividend ratios in equity markets (http://pages.stern.nyu.edu/~dbackus/GE_asset_pricing/CampbellShiller%20…) . They were investigating whether the source of variance in price-to-dividend ratios was due to (1) variance in the expected dividends or (2) variance in the discount rate used to turn those dividends into a present value (or price). Their conclusion was that dividends (and their growth estimates) contribute much less noise than changes in the discount rate.

Why is this relevant? We can model home prices as the present value of discounted rents. Rents have historically been very stable over time (and grow at a very predictable rate). However, home discount rates (mortgage rates can be used as a rough proxy here) are much more likely to bounce around unexpectedly. The literature and the net present value model suggest that shocks to these discount rates amplify the variance of home prices in excess of the (relatively small) variance in projected future rent.

To conclude, our claim is that by including rents and not home prices, the CPI misses a large source of variance and increases in the real liabilities of Americans (since a large percentage of Americans are homeowners). By biasing the CPI downwards, Americans whose incomes depend on CPI to properly track their future costs will be forced to cut back on consumption. This violates the intent of CPI.