Long-term debt propagation and real reversals

Summary

Focus

Economic propagation mechanisms that capture how disturbances systematically feed through the economy over time are central to macroeconomic models. Such mechanisms allow us to understand the behaviour of key macroeconomic variables and help us make more reliable forecasts. Unfortunately, many macro models lack strong propagation based on understandable economic behaviour and instead rely on mechanisms for which there is no economic rationale.

Contribution

We describe a natural propagation mechanism through which new borrowing can systematically affect future output and lead to reversals in activity. The starting point is simple: the majority of debt contracts are long-term and imply regular future debt service payments (consisting of interest and amortisations). These payments pile up during a credit boom and, as time progresses, eventually outweigh the flow of borrowing. When this happens, the positive output effect from the credit boom reverses and output falls. We confirm this pattern using data from many countries over the last four decades.

Findings

Using a novel multi-country data set of debt flows, we find that the prevalence of long-term debt leads to predictable patterns in the data. In the short term, an increase in new household borrowing is associated with higher output growth. Over time, as the stock of debt increases, debt service payments place an increasing drag on output. Eventually the negative debt service effect outweighs the positive effect from borrowing, leading to a real reversal. We find that this mechanism largely accounts for the well documented fact that growth tends to systematically slow for several years after a credit boom.

Abstract

We examine a propagation mechanism that arises from households' long-term borrowing and show empirically that it has sizable real effects. The mechanism recognises that when there is long-term debt, an impulse to new borrowing generates a predictable hump-shaped path of future debt service. We confirm this pattern using a novel multi-country dataset of debt flows. Whereas new borrowing boosts output contemporaneously, debt service depresses output. Credit booms thus lead to predictable reversals in real economic activity several years later. This long-term debt propagation channel is the main reason for why indicators of credit cycles have predictive power for future economic activity.

JEL classification: E17, E44, G01, D14

Keywords: new borrowing, debt service, financial cycle, financial flows and real effects