How Household Debt Contributes to Unemployment: Mian and Sufi
The weakness in household balance sheets and the associated pullback in spending are directly responsible for the lion’s share of employment losses in the U.S. economy. This deficiency remains the most significant impediment to a robust recovery.
Our research suggests that 65 percent of the job losses from 2007 to 2009 came from the drop in household spending induced by the collapse in home prices and its effect on a highly levered household sector.
The first observation we made was that there was a large amount of variation across the U.S. in household-debt levels just before the recession began. In areas that experienced strong increases in home values, debt skyrocketed from 2001 to 2007. However, there were many places that avoided the housing boom and experienced no significant house-price appreciation. In these areas, household-debt levels remained steady in the years before the recession began.
By examining the differences in household balance sheets as of 2006, we were able to tease out how the weak ones are affecting the economy. In a study with Kamalesh Rao of MasterCard Advisors, we showed that areas of high debt experienced a severe shock to house prices and spending from 2007 to 2010.
For example, in U.S. counties in the top decile of the household-debt distribution, house prices fell 30 percent from 2006 to 2009 and spending dropped 15 percent from 2007 to 2009. The consumption decline was across the board; even grocery spending was significantly lower in U.S. counties with severe debt problems.
The declines in consumption are far too large to be explained by the drop in house prices alone. It was the combination of collapsing home values and high debt levels that proved disastrous. High-debt areas have been plagued with delinquencies, deleveraging, and the inability to refinance into lower rates -- all characteristics of overleveraged households.
Further, low levels of consumption in high-debt areas continue to be a major drag. For instance, in the second quarter of 2011, auto sales in U.S. counties with the most debt remained a whopping 40 percent below their 2006 levels. By contrast, in areas that had healthy balance sheets before the recession began, the declines in spending were short-lived and a robust recovery is under way.
So how does this dramatic decrease in spending translate into job losses? It’s difficult to answer that question using geographic variation because goods are often produced far from where people buy them.
That means that when Californians sharply reduce spending on automobiles, there will be job losses at the state’s car dealers. But the drop will also cause employment to fall in Michigan, where autos are produced. In other words, one cannot simply examine job losses in California to determine the effect of the state’s weak household balance sheets.
We overcame that problem by using employment data that breaks down jobs by both county and industry. We then classified jobs into those that depend on the local economy -- such as retail employees, waiters and barbers -- and those that cater to the national or even the global economy, such as those involved in producing manufactured goods. (We labeled the local category “non-tradable” industries and the national or global category “tradable.”)
If weak household balance sheets are responsible for a large share of job cuts, we expected losses in non-tradable industries to be much larger in U.S. counties with weak household balance sheets. That is exactly what we found. In counties in the top 10 percent of the 2006 household-debt distribution, employment in non-tradable industries declined 5.1 percent from 2007 to 2009. In the counties in the bottom 10 percent, the losses were only 0.3 percent.
These numbers match the spending figures above. Where weak household balance sheets have led to reduced spending, local- economy jobs disappear.
Our theory predicted that the decline in employment in tradable industries would be uniform across the country. In other words, when Californians reduce their spending on goods produced throughout the country, employment in tradable industries nationwide will decline. This, too, is exactly what we found.
Making the assumption that the effect of household debt on non-tradable employment within the county is similar to the effect on tradable employment on a national scale, we found that 65 percent of the jobs lost during the depths of the recession were directly related to weak household balance sheets and the associated decline in spending.
We would be happy to entertain other theories to account for the economy’s continued weakness. Any hypotheses, however, must also be able to explain the extremely strong relation between household debt, consumption and unemployment at the county level.
(Atif Mian is professor of economics, finance and real estate at the Haas School of Business at the University of California, Berkeley. Amir Sufi is professor of finance at the University of Chicago Booth School of Business and a contributor to Business Class. The opinions expressed are their own.)
To contact the writers of this column: Atif Mian at firstname.lastname@example.org Amir Sufi at email@example.com
To contact the editor responsible for this column: Max Berley at firstname.lastname@example.org