Thursday, July 7, 2016

Housing prices discounted by local inflation

Recently, I looked at the Shiller Real Home Price Index, discounted by general CPI, discounted by Shelter CPI, and in individual cities, discounted by local CPI owner equivalent rent inflation.

Click on the link, and you will see the two graphs there.  On the national level, using shelter inflation instead of general inflation brings down the relative price of homes considerably.  And, during the recent period, where divergence between cities has been strong, we can see that in many parts of the country, home prices have been reasonable, while cities like San Francisco are extreme outliers.

Here I have added the other cities which have both Case-Shiller price indexes and monthly CPI measures of local owner equivalent rent inflation - a total of 12.  All are indexed to 1995, or to the earliest date available.

Generally, these show the same pattern, with various cities falling somewhere between Dallas and San Francisco.  Case-Shiller tends to measure Closed Access type cities, so they are heavily represented here.  Most of the secondary regional centers and rural areas across the country (outside of California and New England) look like Dallas and Atlanta.

The one thing we might note on these two graphs is that the variance between cities is tighter when we use local rent inflation as the discount factor.  The reaction between price and rent is pretty systematic.  I believe that the exaggerated reaction of price comes from a mixture of (1) rent expectations, which act like a growth rate on home valuations and (2) a natural increase in price/rent levels as prices rise, which is probably mostly due to tax benefits.  I haven't posted all of my work on those issues on the blog, but in both cases, the relationships are strong - especially the latter factor.

At the national level, before the 1950s, the home price index had tended to range between 85 and 115, except for during the interwar period, where it was lower.  Using shelter inflation to adjust the index since then, it had remained in that range until the late 1990s, and was, in fact, at the bottom of the range when the housing boom started.

So, we can see that now there is a split.  The open access and Midwestern cities, adjusted for local rent inflation, have continued to move in that range, moving back to the bottom of the range after the bust.  The Closed Access cities tend to be far above that range.  And, the other cities moved well above that range during the boom and are now in the middle of the long term range, along with the national average.

The influence of the Closed Access cities on the national average is evident here by the fact that Seattle follows closely with the national measure in both graphs.  Seattle is sort of at the mid-point between fully closed access and a functional housing market.  It will be interesting to see if they can manage to maintain a healthy tech industry and an open access housing market.  It could go either way.  But, we can see here what a chimera the national measure is.  Seattle's housing market is nothing like the market that most American's experience between the coasts.

Most Americans live in the housing market of the green and grey lines.  The high rents and higher prices of the Closed Access cities cause them to have an outsized effect on a value-weighted average of the country's home values.  Also, the vast majority of new homes were built in places representative of those green and grey lines.


  1. Great post. Constructing a national CPI and a monetary policy in light of property zoning laws....well, very tricky.

  2. As a Seattle resident now trying to buy a house, the magnitude of the issues you've raised in this series are becoming very real. Homes built well out from the core in the 50s (with moderate updating) get 5-10 competing offers (1.6k sq ft on a 10k sq ft lot lists for 450k and goes 25k above ask). The trendy spots where you have to compete with Chinese all-cash buyers... well, yeah, forget those. The city has done decent work on trying to promote high density new construction, but there are only so many places to put it (mountains bound to the East, water to the West, lake in the middle to make the traffic flow nuts with Bellevue added as a satellite downtown core). The value investor in me hates to join what looks like a momentum chase, but failing to cover my household housing short (also known as renting) holds a very real risk of being priced out should 10%+ YoY price increases continue.

    As an aside, I think it's sort of fascinating that Cleveland and Seattle appear to be the only two series in the shelter inflation indices with a significantly high amplitude seasonality. I have no economic prior for this; also, seems more visible in the national value than in other series (Seattle and Cleveland excepted). Thoughts?


    1. Thanks for the input, Andy.

      I haven't looked at the seasonal issues much. Mostly, I think this has to due with the effect large numbers of foreclosure sales had on seasonal patterns, which caused an atypical pattern when foreclosures increased and then made the ex ante seasonal adjustments difficult to execute in the indexes as foreclosures declined and the pattern returned to normal.