Deleveraging may be caused by a declining willingness by households to borrow instead of a tightening of borrowing constraints.
U.S. households decreased their credit card debt substantially between 2007 and 2013. Although this “deleveraging” may be a consequence of credit demand (households want to borrow less) or supply (banks provide less credit), it is usually attributed to the latter. For example, Guerrieri and Lorenzoni (2011) and Eggertsson and Krugman (2012) modeled this deleveraging as a tightening of household debt limits and proposed such forced deleveraging as a possible cause of the Great Recession.
In this essay, we take a step back and reexamine whether this deleveraging of U.S. households was caused by a contraction in credit demand or supply. In particular, we connect this deleveraging process with the evolution of household financial distress. If the deleveraging were generated by tightened debt limits, one should expect a sudden rise in household financial distress because households would have difficulty rolling over their current obligations. We find that, in contrast to that prediction, the data show household financial distress decreased during this period. This evidence suggests the deleveraging may have been caused by the declining willingness of households to borrow (a credit demand story) instead of a tightening of borrowing constraints.
We use data from the Survey of Consumer Finances (SCF) to illustrate the deleveraging process. Figure 1 shows two series. The solid line (the right y-axis) demonstrates the extensive margin: the share of heads of households carrying a positive balance after the most recent credit card payment. This margin decreases substantially—by 10 percentage points—from 44 percent in 2007 to 34 percent in 2013. The dashed line (the left y-axis) illustrates the evolution of the intensive margin: the average debt per head of household for households with a positive balance after the most recent credit card payment. This margin decreased substantially—by 23 percent—from $7,328 in 2007 to $5,589 in 2013.
We use quarterly data from the Federal Reserve Bank of New York Consumer Credit Panel/Equifax to study household financial distress. We use two definitions of household financial distress as follows:
(i) “High credit”: the fraction of households using all credit available from their credit cards1
(ii) “Delinquent”: the fraction of households with credit card payments 90 days or more late2
Figure 2 shows that both the delinquency measure (the dotted line) and the high credit measure (the solid line) imply similar shares of the population are in financial distress. More importantly, both indicators follow very similar patterns over time: Financial distress remained quite constant in the 2004-07 period and then declined steadily over the next six years. In the pre-crisis period, the average distressed household share using the high credit measure was 7.4 percent and 8.3 percent using the delinquency measure. In the 2008-15 period, the distressed household share using the delinquency measure declined by 41 percent—from 8.3 percent to 4.9 percent—while the share using the high credit measure declined even more—53 percent—from 7.6 percent to 3.6 percent.
The evidence suggests that post-crisis deleveraging did not drive more households to financial distress. One possible reason is that households were not forced to reduce borrowing after the crisis but instead decided to do so in light of crisis-related developments. For example, after the Great Recession, households may have lowered their expectations of future income growth and chose to borrow less.
Overall, the analysis suggests that a contraction in credit demand may have played a role in the deleveraging. Credit supply stories could also have played a role. However, the tightening must be such that it does not imply an immediate increase in household financial distress, which would be counterfactual to the data presented previously. For example, one could imagine a tightening of credit that would affect predominantly young individuals without existing debt. Of course, more research is needed to determine what caused the post-recession deleveraging. Combining this type of data with structural models may be useful to evaluate alternative hypotheses.