Did credit really replace wage growth in the mid-2000s?
The bursting of the U.S. housing bubble in 2006 and 2007 revealed a mountain of debt that households had accumulated in the preceding years. This massive debt overhang obviously had severe consequences for the U.S. economy, which is why economists and policymakers have spent the years since the bust arguing about the best way to deal with the overhang.
But with households starting to increase their leverage once again, perhaps reconsidering the dynamics that led to the housing bubble might be more useful. A rush of new middle and high income mortgage borrowers may have been a significant factor in the run up to the housing crisis of 2006 and 2007, according to a new paper. This new analysis challenges the findings of other recent research into the origins of the crisis that suggest lending to lower-income borrowers who were experiencing no wage gains in this period was the primary cause.
So first, let’s briefly present the currently favored theory. One argument that garners quite a bit of attention is from economist Raghuram Rajan, now the governor of the Reserve Bank of India but formerly of the International Monetary Fund and the University of Chicago. In his book, Fault Lines, Rajan argues that governments loosened credit so that low-and medium-income households could borrow to make up for stagnant wage growth.
Economists Atif Mian of Princeton University and Amir Sufi of the University of Chicago build on Rajan’s research, finding evidence that credit and relative wage growth in zip codes were negatively correlated—meaning that credit grew as relative wages went down—as the housing bubble inflated. But importantly, Mian and Sufi didn’t look at individual lenders, only the activity in the zip code as a whole.
The new National Bureau of Economic Research working paper does look at individual borrowers. The authors, economists Manuel Adelino of Duke University, Felipe Severino of Dartmouth College, and Antoinette Schoar of the Massachusetts Institute of Technology analyze individual mortgage and income data from the Internal Revenue Services to tackle this question. They find similar results to Mian and Sufi on a variety of questions. Their data, for example, also show that credit flowed to areas where housing prices increased the most and that there was a negative correlation between relative income growth and credit growth at the zip code level.
But when Adelino, Severino, and Schoar look at the individual level they find a positive correlation. In other words, mortgage credit was going to individuals who were seeing positive income growth. The authors show that the borrowers who were receiving the credit weren’t those at the bottom of the income ladder, but rather those at the middle and the top. And the credit growth appeared to be in proportion to income growth: The debt-income-ratio didn’t appear to change much during the bubble years.
So were the middle class and the rich were taking out much larger loans than before or more mortgages? According to Adelino, Severino, and Schoar it was both, but more so the latter. They find that the effect on the extensive margin, meaning new borrowers entering the market, was larger than the effect on the intensive margin, or borrowers increasing the size of loans. As the title of their paper says, the composition of home buyers changed during the period.
Adelino, Severino and Schoar’s paper would seem to indicate that what caused the run up in mortgage debt wasn’t due to a change in “lending technology” such as securitization or looser government policies. Rather, the debt was built by the same kind of bubble dynamics that leads to investors betting that an asset will never lose value. Which story is true is still up for debate, but it could just be that this time wasn’t no different after all.