Will housing save the US economy?

By Amir Sufi     

From: Magazine

A growing consensus pins the severe US recession and slow recovery on the tremendous negative wealth shock to households, which resulted from the combination of very high leverage and a sharp decline in house prices. We are now witnessing a recovery in house prices and residential investment, with house prices in January 2013 up 10% year over year according to CoreLogic and permits for new residential construction up 17% in March 2013 compared to the previous year. A heightened focus on housing is warranted because it tends to be a great leading indicator of where the economy is going. 

Many have pointed to the recovery in housing as a major positive for US economic activity going forward. For example, Bill Dudley, the president of the Federal Reserve Bank of New York, said in March that he sees “the recovery in home prices as particularly important” in part because “houses are a significant component of household wealth.” Jan Hatzius, the chief economist at Goldman Sachs, said in January that “the fundamentals for housing activity point to further large gains in the next couple of years.” I bring up these two because I have great respect for both—they have been consistently right about many things over the past few years.

So will housing save the US economy? I won’t leave you in suspense: we need to temper our optimism on what a housing recovery can do. I agree that house prices will continue to rise and new residential construction will steadily increase from its current very low level. This is good news. But we will not be returning to the boom years that preceded the Great Recession. The days when housing was the predominant force driving economic activity are gone, and I view that as a good thing.who withdrew home equity during 2002-2006 boom chart

disappearance of cash-out refinances chart

retail spending on homes versus other goods chart

investor purchases and house price growth chart

us retail sales 1992-2010 chart

household spending across states chart

Housing and the broader economy

An increase in house prices drives economic activity in two ways. First, it induces investment in new residential construction. Second, it leads some households to spend, either for home improvement or consumption. The latter effect has generally been called a “housing wealth effect,” but in my view that’s the wrong way of thinking about it. Instead, the positive effect of house prices on household spending relies crucially on the degree to which a given household is constrained from spending as much as it would like in the short run, either because of borrowing constraints or behavioral biases.

During the housing boom that preceded the recession, both effects were big drivers of economic activity. Single-family housing starts hit 2 million in 2004 and 2005, the highest amount in more than 30 years. The home equity withdrawal effects were likely even larger. We don’t have precise measures of how much spending was fueled by home equity–based borrowing, but research I’ve done with Atif Mian of Princeton suggests that homeowners borrowed $1.25 trillion out of their homes from 2002 to 2006. 

Spending as a response to an increase in house prices was not uniform, which is a critical point often neglected in the discussion of housing wealth effects. In our study of the housing boom, we found enormous differences in the propensity of homeowners to extract equity from their home based on credit scores (see figure 1). Homeowners with the lowest credit scores were very aggressive, borrowing 40¢ against every dollar of increased home equity. Homeowners with the highest credit scores were almost completely passive, pulling almost no equity out of their homes when house prices increased. I refer to the low credit score borrowers as “constrained.” Changes in wealth lead to big changes in spending, which suggests these households are constrained from spending as much as they’d like in the absence of housing collateral.

In research with Kamalesh Rao of MasterCard Advisors, Mian and I also found the exact same relationship during the housing bust. For a given dollar decline in house prices, constrained borrowers cut back on spending much more dramatically than unconstrained households. The marginal propensity to consume out of housing wealth was three-to-four times larger for constrained versus unconstrained households.

The fact that unconstrained homeowners do not consume more out of home equity should not be surprising. Standard economic theory casts suspicion on housing wealth effects, because an increase in house prices is also an increase in implicit rental payments. For an unconstrained household with good access to credit markets, an increase in the value of a home is unlikely to significantly alter spending behavior. But for a constrained homeowner who desperately wants to spend more, a change in home value affects spending. This is exactly what we find in the data for both the housing boom and housing bust. We should not think of the housing wealth effect as a constant parameter in a macroeconomic model. It varies substantially based on which types of homeowners have access to credit.

Mitigated housing wealth effect

Only constrained borrowers consume aggressively out of home equity, but today these constrained borrowers have been shut out of housing and mortgage markets. The only households that can buy a home or borrow against one are precisely the unconstrained households that are least likely to spend out of an increase in housing wealth. Therefore few homeowners are aggressively borrowing against their homes, precisely because they have high credit scores. If we take the results from our previous research, the housing wealth effect for these households may be close to zero, which would substantially dampen the effect of house prices on spending.

The latest version of Lender Processing Services’ Mortgage Monitor illustrates this point. The average credit score for borrowers successfully getting a mortgage (both refinance and purchase) has increased by 40 points from 2006 to 2013 for mortgages backed by Fannie Mae, Freddie Mac, and the Federal Housing Administration, and a whopping 80 points for
private mortgages. 

Figure 2 shows evidence of “cash-out” refinancing, or mortgage refinancing where the homeowner takes additional equity out of the home. The fraction of all refinancing volume where borrowers took cash out was at 2.8% in 2012, the lowest it has ever been in the recorded data. Compare this to 2006, when the fraction had reached almost 30%.

The same pattern can be seen in refinancing volumes. Interest rates on mortgages are the lowest they have ever been, but nobody is taking money out of their homes. This partly reflects tighter credit conditions. But it also reflects the fact that there has been a major compositional shift in the homeowners who are able to refinance their mortgages. Individuals with high credit scores are much less likely to use their home equity to spend.

Another way to see the mitigated housing wealth effect is to look at expenditures on home improvement, appliances, and furniture, which survey evidence from the Federal Reserve suggests are the most likely purchases from home equity withdrawal. Figure 3 (next page) compares retail spending on these goods versus all other goods, and we see a dramatic difference. Spending on home-related purchases in 2012 was still 10% below its 2006 level. This is not just a durable goods issue. Spending on autos has been much stronger than spending on home improvement, appliances, and furniture. Evidence from Neal Soss and Henry Mo of Credit Suisse also shows a mitigated housing wealth effect. They find that the effect of house prices on household spending has been much smaller
since 2008.

Year-over-year spending on home improvement, appliances, and furniture was up 2.4% in January through March of 2013, while other retail spending was up 3.5%. Spending on home-related purchases remained weak even as house prices climbed. In contrast, during the 2002–06 boom, year-over-year spending on home improvement, appliances, and furniture outpaced other retail spending every single year.

One counterargument is that constrained buyers will eventually be able to reenter housing and mortgage markets if house prices continue to rise. This is probably true at some point, but I am doubtful we are close. The homeownership rate in the United States plummeted from 69% to 65% from 2006 to 2013. Most of this decline was driven by low credit quality individuals defaulting and leaving their homes, and I do not see a strong rebound in homeownership in the cards. The homeownership rate has continued to decline even as house prices have started to recover. Further, research suggests that only 10% of borrowers that have a serious default on a mortgage regain access to mortgage markets within 10 years. A very large fraction of the population will be forced to sit on the sidelines even as the housing market gains steam.

The rise of the investor

Which brings me to my second main point: the nature of the housing recovery is quite different than what we’ve seen in the past. Up to this point, it appears to be driven in large part by investors and cash-buyers. Data tracking the fraction of homes purchased by investors are not perfect, but they tell a pretty consistent story. Ilyce Glink, the personal finance writer, used DataQuick information on “absentee” purchases, which are transactions where the property tax bill for the purchaser is sent to a different address than the purchased home. Across the country, she found that absentee purchases have skyrocketed. For example, absentee purchases in Los Angeles have risen to 25% compared to only 12% in 2006. In Cincinnati, absentee purchases accounted for 35% of purchases, compared to the previous ten-year average of 22%.

Another measure is all-cash purchases, or purchases made without any mortgage, which have also dramatically increased over the past two years. Housing economist Tom Lawler posted data on Bill McBride’s calculated risk blog showing the all-cash share of purchases rose from 9% in 2006 to 46% in 2012 in Phoenix. For Orlando, the fraction has risen from 7% to 53% over the same time period. In every market for which he has data, we see the same pattern. 

The data suggest that strong growth in house prices is related to investor purchases such as these. In figure 4, I plot the relation between house price growth from January 2012 to January 2013 against the fraction of houses purchased by investors in 2012. The latter data come from CoreLogic’s March edition of “The MarketPulse.” As figure 4 shows, there is a pretty strong positive correlation for these 16 markets. The direction of causality is difficult to discern: investors may be responding to house price growth as much as driving it. But the recent growth should be understood in the context of the boom in investor activity.

What does this mean for the effect of housing on economic activity? If the trend continues, the most direct effect would be a permanent return to homeownership rates in the United States of 65% or perhaps even lower. I would argue that the marginal 5% of Americans who switched in and out of homeownership over the past decade have high marginal propensities to consume out of wealth and income. During the boom, these households consumed aggressively out of home equity. Now that they are again renters, an increase in house prices represents a loss in wealth, given that they must pay higher rents going forward. This will mitigate any positive effects of house prices on household spending.

Further, investors renting out apartments and single-family homes are likely to invest less in the homes than homeowners would. We still need good theory and data to back up this argument, but it seems to be accepted wisdom among professionals working in housing and durable goods markets. It does make intuitive sense. Landlords tolerate more depreciated washing machines and kitchen appliances, and more transient renters are less willing to pay the landlord for better equipment. The evidence in figure 3 on the weakness in spending on furniture, appliances, and home improvement is consistent with this argument.

Household spending: Permanent damage?

While estimates vary, most agree that the housing boom during the early part of the 2000s was a significant driver of household spending. I have argued that we are unlikely to see housing have such a large effect on economic activity going forward. Evidence suggests we are unlikely to catch up to the household spending rate we saw before the Great Recession.

Figure 5 shows total retail sales for the US economy from 1991 through 2012. The graph is done in logarithm scale with 1991 subtracted. I also plot an estimated linear trend using the 1991–2006 data. The gap between the trend line and actual retail sales from 2007 to 2009 shows the severity of the Great Recession. Even more interesting is the lack of any evidence that the gap is closing through 2012. While spending is increasing, it does not appear that we will catch up to trend the spending growth we saw from 1991 to 2006. 

State-level data help us understand why. In figure 6, I use state level sales tax receipts as a measure of household spending. It is not perfect, given that some states periodically change sales tax rates, but it is the best measure of disaggregated household spending data we can get quickly. Figure 6 splits states based on how large a negative housing-net-worth shock households received from 2006 to 2009. Small housing-net-worth-shock states are areas that escaped the housing crash relatively unscathed, such as Kansas, Kentucky, and Pennsylvania. Large housing-net-worth-shock states, including California, Florida, and Rhode Island, experienced a much more severe decline in household spending during the recession, something I have documented in previous research. 

What is novel here is to look at the recovery. States with the largest housing net worth shock from 2006 to 2009 remain the weakest even through 2012. There is no evidence of catching up. The housing crash appears to have inflicted permanent damage to household spending in these areas.

I believe we have learned a painful lesson from the 2002–12 experience. We cannot rely on debt-fueled collateral-based consumption for economic growth. Easy credit-fueled house prices and consumption during the boom, and the collapse in spending, were exacerbated by excessive debt burdens and a failure to provide any relief to underwater homeowners. Fueling consumption through easy household credit may help in the short-run, but it inevitably has long-run painful consequences. 

Works cited

Glenn Canner, Karen Dynan, and Wayne Passmore, “Mortgage Refinancing in 2001 and Early 2002,” Federal Reserve Bulletin, December 2002.

Robert E. Hall, "The Long Slump," American Economic Review, April 2011.

William Hedberg and John Krainer, "Credit access following a mortgage default," FRBSF Economic Letter, October 2012.

Atif Mian and Amir Sufi, " Household Leverage and the Recession of 2007 to 2009," IMF Economic Review, 2010.

----------, "House Prices, Home Equity Based Borrowing, and the US Household Leverage Crisis," American Economic Review, 2011.

----------, "What explains high unemployment? The aggregate demand channel," Chicago Booth working paper, 2012. 

Atif Mian, Kamalesh Rao, and Amir Sufi, "Household balance sheets, consumption, and the economic slump," Chicago Booth working paper, 2013.