In the previous post of the housing series, I discussed the ironic source of residential fixed investment during the housing boom. Because our major metropolitan areas have limited the expansion of housing in their cities, households have had to substitute suburban homes in less valuable locations for homes in high value, high density cities. The lack of housing stock in the cities that causes this substitution causes real housing expenditures to decrease while at the same time it causes residential fixed investment to rise, because those households tend to build more valuable houses on less valuable lots. They are substituting building for location. In other words, the high level of residential fixed investment was not a result of overconsumption of housing. It was, ironically, a result of falling consumption due to constricted supply.
This graph might give some more clues to this issue. As a starting point, the dark red line is the year-over-year (YOY) change in market values of real estate owned by households. To arrive at the orange line, I have subtracted the YOY change in the S&P/Case Shiller National Home Price Index. This gives us a rough estimate of how much of the rise in the value of residential real estate was the product of real expansion and how much was price appreciation. The real increase in household real estate was normal or slightly low during the boom.
The dark blue line is residential fixed investment minus estimated depreciation, as a percentage of real estate market values. This should give us the same basic result that the price corrected growth of real estate did, and in fact it does give us the same basic behavior, but with less noise. By these measures, new investment in housing has been in decline for decades, including during the so-called housing "bubble".
Now, take a look at the light blue area. This is residential fixed investment (RFI) as a proportion of gross rent, scaled to reflect depreciation. It also tends to follow the same trends as the other investment measures. But, since it has rent as the denominator instead of market values, the difference between this measure and the others is mathematically simply a reflection of changing Price/Rent ratios. So, the three periods where RFI/Rent rises above RFI/Price are the periods were Price/Rent has been high.
A high Price/Rent ratio is caused by low long term real interest rates and high rent inflation. Relatively stable construction costs mean that the higher price accrues to the lot. This graph shows the replacement cost of houses as a proportion of total market value, and this declined as Price/Rent rose and as rent inflation has accumulated.
During these periods, households substitute building value for lot value. And, if we look back at the first graph, during expansions, the price adjusted growth rate in real estate values has tended to fall below the growth rate estimated by RFI when Price/Rent (and RFI/rent) has been high. The RFI measure is a measure of investment in buildings. The price adjusted YOY change in real estate values is a measure of all real estate - land and buildings. During these periods, investment in buildings rose but investment in land lagged.
After the cyclical spike in 2001, growth in real housing was low. We were building a lot of houses where location values were low. So, homes in the problem cities were climbing in value because of supply constraints and rent inflation - no real new housing value was created there. And, households priced out of those markets were building "McMansions" in the hinterlands for a fraction of the price on lots with much lower values.
Economic Rents and Real Value
There is a subtle issue to think about here regarding economic rents on land and real economic production. If we released the constraints on urban building and made it feasible for urban real estate owners to expand housing to its reasonable potential where its value is high, they would earn a large windfall as economic rents on those assets. But, those rents wouldn't so much reflect a transfer of income as they would reflect the creation of value. The added value provided by the higher capacity of the urban land holdings would increase real household consumption and real incomes of urban households (through falling shelter rent). The orange line in our graph above would be higher.
Here's a funding idea for city governments in the worst cities (NYC, LA, San Diego, San Francisco, and Washington DC), if they just can't stand to see their land owners reaping a windfall. Just go around buying up underdeveloped plots in voluntary transactions, then streamline the process for developing them and sell them to developers without restrictions on rent. They could buy properties for a few million dollars, then allow them to have a high-rise condo building, and sell them to developers for hundreds of millions of dollars. They could probably fund the city budget that way. And, eventually rents will fall enough that the new condo buildings will be built for middle class households, and we won't have to keep building "McMansions" in the desert.
The only parties that will be damaged by that program will be the real estate owners who don't sell their properties to the city, and see their property values fall when rents begin to fall. But, maybe they won't be so upset if the city has eliminated property taxes because of all the new revenue. It's not that hard to come up with a win-win scenario when your current policy is to needlessly destroy value.
My Rent Inflation Assumption
I have been using core non-shelter inflation as the baseline to estimate the rent inflation that is caused by supply constraints. I realize this is a little sloppy. Even lacking regulatory constraints, no two items have the same price behavior over time.
As a general point of view, the cost of replacement graph above confirms the idea that there has been significant inflation in residential land values, above the cost of building. Here is a graph comparing shelter prices over time to household furnishings and operations and to lumber. In both of those cases, prices have been falling relative to core CPI, while shelter prices (mostly consisting of rent) have been increasing. I suspect that, if anything, the use of core minus shelter CPI as a benchmark understates the level of supply-based rent inflation.
Over time, there could be some non-inflationary lot appreciation. For instance, in fast growing cities like Phoenix, Houston, or Las Vegas, homes which were purchased and then subsequently had the city grow around them, probably experienced real value gains due to the value of living near newly developed amenities. I don't know if the BLS accounts for that sort of hedonic adjustment or not.
On the other hand, while building in high-rise areas of core cities would be expensive even without regulatory barriers, it would generally be reflected in higher price levels, not ongoing inflation. Skyscrapers may be somewhat more expensive than McMansions, per square foot of living area, but they aren't progressively getting more expensive over time. So, natural price levels of housing in the cities might tend to be high, but there is not a natural reason for rent inflation to be high there.