How far will US home prices fall?

William C. Wheaton, 18 July 2009

a

A

In a New York Times column last month, Robert Shiller painted a scenario in which home prices might fall for years (and even decades) to come. However, this view ignores some critical differences between real estate and other more purely “financial” assets. This oversight often leads to an exclusive focus on demand fundamentals, expectations, and other factors that both create asset price bubbles and their collapse. What is missing from this analysis is the supply side of the market. Real estate is a physical asset and not just a claim on a hypothetical income flow.

As housing is a physical asset, its price must eventually equal or exceed the full cost building or rebuilding it – that is, as long as the market requires the construction of additional housing. So the real questions in the current crisis are a) when and how much future housing will the US need to construct, and b) are prices today so much higher than the cost of construction that they could still fall significantly and have development remain economically viable?

During the last decade, net new household formation averaged approximately 1.4 million per year. Last year, the Census reported that the US added only 544,000 new households – during severe contractions the young stay at home, singles “double up”, and household formation (normally) slows. Even with declining demographics, however, most analysts foresee new household growth resuming to a level of at least 1 million by 2010 and beyond. If we conservatively add 200,000 demolitions per year, the US economy will “need” at least 1.25 million new units yearly in the near future. With today’s currently depressed construction, this generates a yearly deficit of 750,000 units. At that rate, the current excess inventory of units for sale or rent will be back below normal by 2011. Prices historically have a strong relationship with sales “duration” – the ratio of inventory-to-sales. Hence under reasonable conditions, in two years we will have to increase construction considerably and prices will have to justify the cost of that construction.

Even without physical growth, income growth can also help provide a floor on housing prices. With rising prosperity, most nations demand “bigger and better” housing units – even if not more of them. This generates significant residential re-development as well as “alterations and additions”. These investments in turn require incremental increases in value to justify the expenditure. Once again, replacement cost can set a floor on prices.

But could it be that US prices inflated so far above “replacement costs” that in fact decades of decline could occur before prices reach those building costs? Perhaps in Japan, where the value of land at one point represented 80% or more of real estate value, but not in the US. Other economists (e.g. Glaeser and Gyourko, 2005) have amply demonstrated that outside of a handful of US markets, most US homes carry values that are at or below their replacement cost – even in recent years! With a further fall in prices, there simply will not be of sufficient value to justify new construction, which then will lead to a severe shortage of units and, of course, higher prices.

The importance of these “supply side” considerations is born out quite nicely in the experience of Japan, the one country where house prices have declined continuously since 1990. During this period, Japan’s household growth swung from barely positive to distinctly negative. Every year, houses in rural areas and smaller Japanese cities are abandoned and boarded up as the population slowly shrinks. Meanwhile, in Tokyo and other major cities there continues to be extensive reinvesting and rebuilding. This occurs despite the price decline of the last two decades, because in those cities the value of Japanese real estate is still composed of (now only) 40% land. Until that land value is nearly extinguished, Japanese prices could still fall further. The two countries are simply not comparable. Japan has not yet reached the “natural price floor” for housing and has a shrinking population. The US, on the other hand, is growing and has already reached that floor. US housing prices will stabilise and then resume growing again soon.

References

Glaeser, Edward L. & Joseph Gyourko, (2005). "Urban Decline and Durable Housing," Journal of Political Economy, 113(2): 345-375, April.

Topics: Global crisis
Tags: Japanese real estate market, US housing prices

Comments

What about the large number of completely vacant homes?

A very well though out hypothesis, but no mention is made of the large number of completely vacant homes that are not for sale or rent.

Or alternatively...

Professor Wheaton,

I respectfully disagree based on two quick items:

1. Household formation is trending at 772k, nowhere close to 1.4m. (http://www.washingtonpost.com/wp-dyn/content/article/2009/07/09/AR200907...)

2. Existing inventory of unsold new homes are off the chart (http://www.nytimes.com/2009/06/27/business/economy/27charts.html?sq=rece...)

So given the overhang from housing starts that averaged 1.8m units in the bubble period of 2001-2007 AND given the fact that homebuilders would need to exercise discipline in home starts to near post-WWII lows AND given the fact that record low affordability and government incentives have done little to spark demand (that $8k rebate ends in Dec btw), home prices seem headed lower over the next several years.

I agree the Japanese example you cite is prescient... in illustrating the fact that in the long run governments find it difficult to make water flow uphill.

Counting more carefully could give a different result

The premise is that given more or less historic housing demand levels (an awkward premise in a long recession with high unemployment), the lack of new housing will lead to a constrained supply and so higher prices and everyone will be happy. Lolipops (suckers) for everyone.
But it's worth counting more carefully. The housing bubble is a very regional thing. Will all that overbuilding in Florida get sopped up by ordinary demand in short order? I know where I live, Portland, OR, where the bubble was only moderate, there are many many thousands of unoccupied condos and developments. Some converted to rentals. Most are just empty, waiting for the developers' financing to be called in. Then they will be sold and dumped back on the market CHEAP.
CHEAP is bad for housing recoveries.
So that's one "shadow" market.

Then there are all the houses owned by people near default. Those people will become long term renters and those houses will add to the supply.

It seems like it would be worth doing a local analysis of one median bubble impacted area instead of totaling up national averages.
For one thing, a better count could be made of what the real housing supply is or would become.
For another, real estate simply is a local market. Analyzing it in aggregate will lead to interesting (if wrong) conclusions.

Professor of Economics, MIT