One of the oddities of the housing boom was the fact that median prices for new home did not rise as quickly as the median price of existing homes. This may not seem so odd if one is able to explain it by claiming that many new mortgages were going to low income households, so that marginal new homes tended to be at lower price points. But, this explanation doesn't work for me, since I have found that the average incomes of new home owners did not actually decline during the period. My explanation has been that this was a product of location. There are clear migration patterns away from high cost cities. Housing starts in Open Access cities were much higher than housing starts in Closed Access cities, so it doesn't seem crazy to expect that the declining relative price of new homes would correlate with rising building rates in Open Access cities. It seems obvious enough to me that I have been, more or less, stating it as fact. But in this case, the data is not quite as clear as I would have thought.
The first graph compares the US Median home price, according to Zillow, to the median new home price according to the Census Bureau. It is a little difficult to see the relative changes on this graph, so on the next graph, I have reproduced this as a ratio. And, I have graphed it alongside the ratio of permits for new housing units in the Closed Access cities / Open Access cities.
The ratio of permits was much higher in the late 1980s, mostly attributable to California. But, it dropped sharply in the early 1990s, and remained low after that.
In this second graph, we can see that there was an earlier period with relatively low new home prices, in the late 1970s. Home Price/Rent was high then, as it was in the 2000s. And, in the 1970s, high inflation made mortgage payments high, too. Nevertheless, homeownership was rising at that time - again, much like the 2000s. It appears to me that the reasons for the lower new home prices then were somewhat similar to the reasons for it in the 2000s. Households were trying to hedge against rising rents, and since home prices were generally efficient, marginal new homeowners were downsizing in order to accommodate the cash flows of ownership. I think this explains why, counterintuitively, in both periods, even with rising relative home prices and rising ownership, growth of real housing expenditures shifted down (shown in the next graph). Relatively lower value new homes is a sign of the answer to this conundrum.
But, what I thought we would find is that as new home prices declined, relative to existing homes, we would see a parallel decline in permits issued in the Closed Access cities. It seems obvious that this should have happened, simply as a product of population growth. In 1988, the three main Closed Access cities had 254% of the population of the four main Open Access cities. By 2007, that was down to 174%. Just by virtue of being larger, we should have expected the Open Access cities to be issuing more permits over time.
But, instead, the ratio remained pretty level from 1992 to 2006. Now, I think partly what is happening is that the way we normally look at housing starts, in raw units, without adjusting for population, the level of recent periods gets overstated a little bit. I have rendered permits as a proportion of population in these next graphs, which I do in order to compare rates of new building among cities. When we do that, the rate of housing starts (or permits, which follow each other closely) are much more level in the 2000s than they appear to be in unadjusted long term graphs.
Even though the rates of new permit issuance in the Closed Access cities is so low that it is hard to tell, there was an upward trend of permit issuance in those cities during the boom.
I still think location played a large role in the relative decline of new homes, but some of it was playing out within metro areas, with existing core city properties rising sharply in price and some level of building in the suburbs serving as an outlet for households priced out of the core areas.
This trend reversed after the crisis. There has been a small rise in the relative level of new building in the Closed Access cities, mostly because the mortgage crisis has hampered building in the Open Access cities. But, the change isn't large enough to explain the extreme jump in relative new home prices. I don't think location has much to do with the recent trend. It has more to do with the fact that middle income families don't have reliable access to mortgage financing, so new building is skewed to the small portion of households at the top of the income distribution who can qualify for mortgages.
Here is one last graph, which I think helps us to think about the similarities and differences between the 1970s and the 2000s. In the 1970s (and early 1980s), high inflation meant that mortgage payments were very high across the country. This graph shows the median home in Houston, San Francisco, the US, and the median new home, all relative to median incomes in the respective areas. The lower prices of new homes in the 1970s were helping to make mortgages affordable for marginal new households, across the country.
Mortgage rates had declined by the 1990s, and during that period, mortgage affordability for new homes moved higher above the median existing home, as affordability was less of a constraint.
By the 2000s, affordability was tied to location. Homes in Houston were as affordable as ever, but San Francisco had nearly as much of an affordability issue as it had in the early 1980s. Here, we can see the mortgage affordability of new homes moving back down closer to the affordability of existing homes. And, if affordability was largely influenced by location at that time, we would expect this accommodation to happen through location.
Here is a graph of lot and home square footage. In the late 1970s and in the 2000s, real long term interest rates were one cause of rising Price/Rent levels. Since real long term interest rates don't particularly effect the cost of building, we would expect low real long term interest rates to lead to a substitution of more structures and less lots. Since high inflation rates in the 1970s and early 1980s made affordability difficult regardless of location while low real long term interest rates were pushing up prices, we see a reaction to these factors in a very sharp decline in lot sizes, and a small decline in structure sizes. There was a general downsizing and a substitution of structure for lot.
From 2006 to 2009, home size remained level during the deep part of the bust, and now median new home size is rising along with the relative price of new homes because they now tend to be purchased by households with higher incomes.