Household Debt and Recovery Periods

In this paper, I seek to investigate the impact that household debt has on how long it takes an economy to recover from a recession by addressing the following questions: How does increasing household debt impact the length of recovery periods in the business cycle? Are worries related to the long-run trends of debt accumulation justified or should observers be more concerned with issues of debt quality? Do different types of debt have different impacts? Using a dataset of thirty-two countries over twenty-two years (1995 to 2017) to estimate multiple empirical models provides support for both a supply-side and demand-side narrative of debt’s impact. Ordinary least squares and two-stage least squares are used to estimate simple linear, multiple linear, and instrumental variable regression models. Support is presented for both the supply- and demand-side views, which have positive and negative relationships with recovery time, respectively. Thus, they are taken to be not mutually exclusive when considering changes in debt around recessions and the impacts of different debt types. Regarding recent debt trends, I present evidence to justify concerns of debt accumulation, given significant increases in household debt. However, excess debt is estimated to have a more negative impact on recovery speeds than that of debt levels, by a factor of five. This suggests that, in order to shorten recovery periods, debt-related policy should primarily focus on risk from excess debt levels. Analysis of different debt types presents a new narrative, which accommodates a small, negative impact on recovery speeds from short-term consumer credit, and a small, positive impact from other short-term loans. In this narrative, the demand-side view of overborrowing consumers and supply-side view of optimal consumption-smoothing consumers take place in traditional lending channels and less tradition lending markets, respectively. This narrative is left for future research, given the limitations of this paper