I have been playing around with Zillow data on a set of 151 metropolitan areas (MSAs), and I decided to double check myself on the extent to which Americans did not experience a pricing bubble in the 2000s. I have put together histograms of Price/Income, Mortgage Affordability, and Rent Affordability, by city. For each measure, I have one graph with the count of cities and one graph with the total populations. For the total populations, I am associating the entire population of each city to its median value. In reality, each MSA will have a range of values, but accounting for that would require a lot more effort. This helps us get the basic idea, I think.
These cities represent about 70% of the US population. I don't currently have direct data on the other 30%, but since it is largely rural, I expect it to have low valuations. So, including it would mostly add to the lower part of the distributions of these measures.
First, here is Median Price/Income, by city. Blue bars are 1995, Green are 2005, and Red are 2014. By city count, we can see that in 1995, the distribution was relatively tight. Price/Income was under 4x in almost every city, and was under 6x in all cities. There was a slight shift higher in 2005, with the mode moving up from the 2.7x bin to the 3.0x bin, and a very long right tail developed. By 2015, the bulk of the distribution had moved most of the way back to 1995 ranges, except for some of the extreme cities at the upper end.
This is the sort of distribution I am thinking about when I say that most Americans did not experience the price boom. But, I may have been overstating the case, slightly, because the outlier cities tend to be the larger cities, so that if we graph it based on populations, the tail of outliers claims a lot larger part of the country. There are definitely two stories here - Americans in the hump and Americans in the tail.
Since low real long term interest rates increase the cash value of homes, some of the shift to higher home values should be attributed to interest rates, which have fallen since 1995.
Here is a measure of Mortgage Affordability, which measures the portion of the median family's income needed to make the payments on a conventional 30 year mortgage. Since the low long term interest rates we have experienced would have a similar effect on the value of a home and the affordability of a mortgage, changes in Mortgage Affordability can give us some insight into the relative causes of home price changes.
The basic distribution of mortgage expenses in 2005 is very similar to 1995 for cities in the hump. The hump is slightly smaller, because some portion of the cities have moved to the tail. But, the cities that didn't move to the tail have a similar distribution to cities in 1995. This suggests to me that the relatively small changes in Price/Income in the hump cities were largely related to falling interest rates. This is why debt service ratios were relatively stable until very late in the boom. This measure also has a much larger tail when rendered by population. So, again, there were many more cities that didn't experience the price boom, but they tended to be smaller cities. About 1/3 of US population lived in cities where Mortgage Affordability moved outside the 1995 distribution. The other 2/3 of the country lived in the hump (assuming that the remaining rural areas tended to have hump characteristics).
In 2015, Mortgage Affordability has moved lower than it was in 1995, but with the remnants of the outlier cities remaining. This is a picture of the basic error I believe we made. We have moved the hump to the left when our problem was the tail. We have solved a problem we didn't have while the problem we did have remains.
Next is Rent Affordability. (A caveat. I think the Zillow data for this measure includes some linear extrapolation for the earlier time periods, but the estimated levels should be relatively accurate.) What we see here is a slight march to higher rents in each period, especially from 2005 to 2015, across the distribution.
There isn't so much of a hump & tail shape here, but we can see affordability in some of the larger cities moving away from the norm in 2005 and especially by 2015. The skew isn't as high here as with the range of home prices. That is because for the outlier cities, I think the price reflects both the current high rent level and the expectation of persistently high rent inflation. Note that while the skew has declined in the price and mortgage measures, the outlier rent cities have moved further from the norm in 2015.
This is why I insist that homes now are significantly underpriced. The Price/Income hump is basically where the hump was in 1995. But the Rent/Income hump has moved to a much higher level in 2015. This also reflects our misdiagnosis. We have hobbled the credit markets that fuel housing supply, so home prices cannot be bid up to their intrinsic values. The continuing lack of supply has pushed rents relentlessly higher. I suspect that home prices will need to rise substantially before they induce reasonable levels of housing starts - which at this point needs to be more than 1.5 million units per year for some time to bring these rents down.
With regard to the mismatch of rents and prices, I am referring to the hump - the majority of the country. The tail cities are expensive because of localized supply constraints, so their correct prices are necessarily more speculative. And, to the extent that aggregate national numbers are skewed higher by those outlier cities, it distracts us from the clear policy needs of the majority of the country that is capable of building homes, if we can revive our mortgage market.