1/2/21
A discussion of home price values at a U.S. national level using a couple of valuation measures and a number of relevant new charts.
As discussed in 01/01/21 Post - Home Prices Go Up...And Down, despite some beliefs to the contrary, U.S. real (inflation adjusted) home prices do indeed go up and down as shown in the next chart.
If one views homes as consumable items, there is no good reason to believe that they should appreciate in the long run in real (inflation adjusted) terms. With that view, whenever real home prices fall significantly below their long term average, they could be considered undervalued with an expectation of future real price appreciation. Likewise, whenever real home prices rise significantly above their long term average, they could be considered overvalued with an expectation of future real price depreciation.
The next chart adds average and standard deviation lines to the previous chart. By definition, 68% of data points are within one standard deviation of the average, and 95% of data points are within two standard deviations of the average.
With the view that real home prices should not appreciate or depreciate in the long run, one could consider homes to be over/undervalued when they are more than a standard deviation above/below their average price. Likewise, one could consider homes to be extremely over/undervalued when they are more than two standard deviations above/below their average price.
In the next chart, the real home price index is colored to show when home prices would have been considered over/undervalued by the just discussed valuation measure.
One might wonder if this valuation measure would look any different if it only included more recent history. The next two charts are the same as the previous two, except that they only go back to 1953, the year when monthly data became available, and the index is scaled to be 100 at the last data point. These charts show that only looking at post-WWII data doesn't change this valuation measure's results much.
A good valuation measure would detect major market valleys and peaks as they occur. The valuation measure used in the previous charts did detect a major market valley in the 1920's and 1930s. It also detected a major market peak in 2006. It is also suggesting that home prices are extremely overvalued in 2020 at the time of this writing. However, it failed to detect multiple other major market peaks and valleys.
One could argue on a number of grounds for the invalidity of the thesis that homes are consumable items which don't appreciate in the long run. For instance, the price of the land on which the home sits is included in the home price, and the land is clearly not consumed. Likewise, although there is plenty of land in the U.S. on which to build homes, supply of land in the most desirable locations may be in limited supply relative to the growing population. In addition, a significant number of people buy homes as an investment for the purpose of renting them out. In this case, the amount of rent that can be collected for a particular home can play a significant role in the price an investor is willing to pay for the home.
If one views homes as investment assets with real or potential earnings, then valuation measures similar to those used for other investments might be useful. For stocks, various price-to-earnings (P/E) ratios have been used as a valuation measure (see 12/02/20 Post - U.S. Stock Market Valuation Using P/E). For homes, the real or potential earnings is the rent that is or could be received. If the Case-Shiller Home Price index included a corresponding rent index for the same houses, then calculating the P/E for this index would be simple. Unfortunately, that index does not exist. However, for the purposes of a P/E-like valuation measure, the actual value of the P/E ratio is unimportant. Therefore, it is sufficient to have a national rent index which covers similar homes over the same time window as the price index. Such a rent index does exist as a component of the U.S. Consumer Price Index (CPI). The next chart shows CPI and the housing sub-components of CPI.
Note that the previous chart has a component for Owner's Equivalent Rent. This component tracks the price that home owners could charge to rent out their homes. Unfortunately, the data for this component only goes back to the early 1980s. However, as can be seen from the previous chart, the Owner's Equivalent Rent (OER) component very closely tracks the Shelter component, for which data is available back to the early 1950s.
For this Shelter CPI component to be used as an "earnings" index for a real (inflation adjusted) home price index, it needs to be real (inflation adjusted) as well. This is done in the next chart.
The previous chart shows that Shelter (and OER) have increased in price in real (inflation adjusted) terms - that is, relative to overall CPI. It is this fact that the previous valuation measure did not take into account and a P/E-like valuation measure will take into account.
The next chart shows the real home price index, this real (Shelter based) OER earnings index, and the P/E ratio that is created by dividing the former by the later.
The next chart takes the P/E ratio from the previous chart and adds average and standard deviation lines. By definition, 68% of data points are within one standard deviation of the average, and 95% of data points are within two standard deviations of the average.
With the view that homes are an investment asset, there is no reason to believe that the P/E ratio should trend up or down in the long run. That lack of a long term trend is what the previous chart seems to show. Therefore, one could consider homes to be over/undervalued when they are more than a standard deviation above/below their average price. Likewise, one could consider homes to be extremely over/undervalued when they are more than two standard deviations above/below their average price.
In the next chart, the real home price index is colored to show when home prices would have been considered over/undervalued by the just discussed valuation measure.
As mentioned previously, a good valuation measure would detect major market valleys and peaks as they occur. This P/E-like valuation measure based on the ratio of real home prices to real rent prices does a fairly decent job of detecting some of the largest peaks and valleys. This valuation measure historically would have been a helpful indicator of the valuation of the U.S. national home market. However, because housing markets vary quite a bit throughout the U.S., more geographically localized valuation measures would be helpful.