by Bill McBride on 6/25/2009 11:00:00 AM
Thursday, June 25, 2009
Earlier this week, Charles W. Calomiris, Stanley D. Longhofer and William Miles wrote in Real Time Economics that the wealth effect from housing on consumption should be small. Atif Mian and Amir Sufi of the University of Chicago Booth School of Business respond that their data indicate the opposite.I commented on the Calomiris et. al. piece here: The Housing Wealth Effect? and I noted that Mian and Sufi disagreed.
Here are excerpts from Atif Mian and Amir Sufi's piece today:
... In the June 22nd entry for Real Time Economics, Calomiris, Longhofer, and Miles argue that ... “the reaction of consumption to housing wealth changes is probably very small.”
Findings in our research suggest the exact opposite: the rise in house prices from 2002 to 2006 was a main driver of economic growth during this time period, and the subsequent collapse of house prices is likely a main contributor to the historic consumption decline over the past year.
We agree with two key points made by Calomiris, Longhofer, and Miles. First, from the perspective of economic theory, it is not obvious that housing wealth should affect consumption. Second, it is difficult to measure the causal effect of housing wealth on consumption because other economic factors confound the relation. ...
These factors highlight the importance of quality data and sound methodology to estimate the effect of house prices on real economic activity. Our study samples 70,000 consumers in 1998 who were already homeowners at the time. We then follow the borrowing decisions of these households for eleven years until the end of 2008. Our data set represents a major advantage over prior studies; it allows us to see exactly how existing homeowners respond to increases in house prices.
In order to isolate the effect of house prices on consumption, we rely on a simple insight: in response to an equivalent increase in local housing demand, house prices will increase more in cities where, due to geography based factors, the cost of building a house is high. For example, consider a homeowner living in San Francisco and a homeowner living in Atlanta as of 1998. From 2002 to 2006, house prices rose sharply in San Francisco where it is difficult to build additional houses because of the limited geography. In contrast, in Atlanta, where home construction is cheaper, house price growth was moderate. In economics jargon, cities where housing supply is relatively “inelastic” will experience larger movement in house prices relative to “elastic” cities (see Figure I).
Our experimental design exploits this insight in order to test how house prices affect borrowing behavior. The “treatment” group consists of homeowners in inelastic housing supply cities (e.g., San Francisco) that experienced a sharp increase and subsequent collapse of house prices. The “control” group consists of homeowners in elastic housing supply cities (e.g., Atlanta) that experienced little change in house prices.
Using this methodology, we find striking results: from 2002 to 2006, homeowners borrowed $0.25 to $0.30 for every $1 increase in their home equity. Our microeconomic estimates suggest a large macroeconomic impact: withdrawals of home equity by households accounted for 2.3% of GDP each year from 2002 to 2006. Figure II illustrates the sharp increase in household leverage for homeowners living in inelastic cities.
A concern with our interpretation is that there are inherently different economic conditions in inelastic versus elastic housing supply cities that may have been responsible for the borrowing patterns we observe. However, several facts suggest that this is not a valid concern. First, inelastic cities do not experience a stronger income growth shock (i.e., a larger shock to their “permanent income”) during the housing boom. Second, the increase in debt among homeowners in high house price growth areas is concentrated in mortgage and home equity related debt.The results of Atif Mian and Amir Sufi fit with what I've observed.
Third, renters in inelastic areas did not experience a larger growth in their total debt. Finally, the effect of house prices on homeowner borrowing is isolated to homeowners with low credit scores and high credit card utilization rates. These “credit-constrained” households respond aggressively to house price growth, whereas the highest credit quality borrowers do not respond at all.
Our results demonstrate that homeowners in high house price areas borrowed heavily against the rise in home equity from 2002 to 2006. We also provide evidence that real outlays were a likely use of borrowed funds. Money withdrawn from home equity was not used to buy new homes, buy investment properties, or invest in financial assets. In fact, homeowners did not even use home equity withdrawals to pay down expensive credit card debt! These facts suggest that consumption and home improvement were the most likely use of borrowed funds, which is consistent with Federal Reserve survey evidence suggesting home equity extraction is used for real outlays.
Our analysis of the microeconomic data has led us to the conclusion that the severity of this economic downturn is rooted in the household leverage crisis, which in turn is closely related to the housing market. If the housing market continues to deteriorate, then further de-leveraging of the household sector will likely keep a lid on any rebound in consumption. In other words, the future of consumption and house prices are closely linked.
Posted by Bill McBride on 6/25/2009 11:00:00 AM