When is corporate debt a threat to macroeconomic stability?

Since the 2007-08 Global Financial Crisis, policymakers have focused on household debt as the primary risk factor for macroeconomic stability. Should they be similarly worried about corporate debt? Based on data from 115 economies for the period 1940-2014, this article shows that an expansion in corporate debt can be just as significant for predicting crises, and also plays a role in the depth of the post-crisis recession.

Since the Global Financial Crisis of 2007-08, household debt has been viewed as a major contributor to the risk of systemic financial crises. This has prompted policymakers in many countries to focus on tightening borrowing constraints in this sector. Indeed, a large body of evidence shows that household debt played a key role in the boom-bust cycle of the 2000s in the US (Mian and Sufi 2009, 2010, Keys et al. 2010). Internationally, increases in household debt have been found to predict business cycle downturns and financial crises (Buyukkarabacak and Valev 2010, Jordà et al. 2016, Mian et al. 2017) and slow recoveries after such crises (Jordà et al. 2015, Jordà et al. 2022).

Perhaps surprisingly, there is relatively little evidence on whether, from a macroeconomic perspective, we should be similarly worried about corporate debt. When looking at firm-level data, there is considerable evidence that excessive leverage can harm firm investment and employment (Giroud and Mueller 2016, Kalemli-Ozcan et al. 2022). But existing research has shown a relatively limited link between an expansion in corporate debt and future economic activity. For example, Mian et al. (2017), find that growth in corporate debt is not associated with boom-bust cycles and has only weak predictive power for subsequent GDP (gross domestic product) growth. Similarly, Jordà et al. (2022) conclude that corporate credit booms leave a lasting imprint only on the macroeconomy of countries with particularly inefficient corporate bankruptcy regimes.

Data and findings

In recent research (Ivashina et al. 2024), we study the role of firm debt in macroeconomic fluctuations using a large historical cross-country panel dataset on credit markets from the Global Credit Project (Müller and Verner 2023). This dataset is unique in the sense that it has considerably broader coverage than existing datasets and, importantly, allows us to decompose corporate credit by industry. In total, the dataset covers 115 advanced and emerging economies from 1940 to 2014. We also construct a new time series dataset on credit to non-bank financial institutions, which has been missing in previous work. This dataset is unprecedented in its breadth and scope, overlapping with 87 episodes of systemic financial crisis.¹

Using this more granular dataset, we find that corporate debt plays a crucial role in boom-bust cycles because of its link with banking crises. As we show, firm debt accounts for about two-thirds of the aggregate credit growth in the three years preceding banking crises. After a crisis hits, the ensuing credit crunch is entirely concentrated in credit to firms, as is the vast majority of loans that become non-performing. In terms of predictive power, we find that an expansion in firm debt can be just as significant as household debt for predicting crises, and it also predicts the depth of the post-crisis recession. We also show that, while growth in corporate debt is not associated with lower average growth, it is highly predictive of higher crash risk in GDP – much more so than household debt.

The main results of our paper can be summarised in two figures: Figure 1 shows that, although household and firm credit each contribute to credit growth in the years leading up to banking crises, the decline in credit growth following the crisis is entirely concentrated among firms. These results contrast sharply with the patterns for household debt, which continues to grow relative to GDP even after a financial crisis has started.

Figure 1. Share of firms in the credit boom and bust around banking crises

Notes: (i) This figure plots average changes in credit to firms and households (relative to GDP) from five years before to five years after 87 systemic financial crises, where we identify the first year of a crisis based on the chronologies in Baron et al. (2020) and Laeven and Valencia (2020). (ii) By definition, the sum of the sectors is equal to total credit.

Figure 2 shows that defaults in the corporate sector also explain most of the non-performing loans following banking crises. While the 2007-08 crisis in the United States has been extensively studied and is arguably at the heart of much of the narrative about banking crises since, it is far from representative. For the typical crises, we find that three-quarters of the non-performing loans on bank balance sheets are accounted for by corporate – not household debt – and this is true both in advanced and emerging market crises.

Figure 2. Share of firms in non-performing loans associated with financial crises

Notes: (i) This figure plots the share of firm credit in non-performing loans for a sample of 23 systemic financial crises, as defined above. (ii) We plot values for the year where the overall non-performing loan ratio reaches its peak after 10 years following a crisis. (iii) * indicates countries that experienced deep recessions during 2008 following large credit expansions but had no financial crisis, according to Laeven and Valencia (2020).

Mechanisms

To understand the underlying mechanisms linking firm debt and crises, we make use of the sectoral variation in our data. In particular, we exploit the fact that sectors vary in their dependence on different types of collateral. Consistent with the existing literature linking real estate collateral values to firm behaviour (Chaney et al. 2012, Bahaj et al. 2020), we find that lending to firms relying on real estate as collateral, such as those in the construction sector, has a relatively stronger association with crises. A one standard deviation² increase in real estate-backed firm credit relative to GDP over the past three years is associated with a 3.7 percentage point increase in the probability of a financial crisis within the next three years. We find a similar result for credit to the non-bank financial sector, which, as we show, tends to lend to non-financial firms. These findings suggest that corporate debt expansions contain important information about future macroeconomic risks.

The sectoral variation in our data also allows us to examine whether bad credit booms are times when some sectors grow ‘out of whack’. We show that the dispersion of credit across sectors systematically increases during credit expansions, and increases in dispersion are in turn predictive of a higher likelihood of banking crises. Our interpretation is that, even in the case of two credit booms of the same magnitude, disproportionate growth in credit flowing to some industries signals heightened risk-taking in certain sectors, with potential implications for the overall stability of the financial system.

We also examine whether firm debt matters for the recovery from crises using local projections (Jordà 2005). In particular, we study how the path of real GDP per capita depends on the type of credit extended before the downturn. Again, we find that corporate debt is important for macroeconomic dynamics. Corporate debt not only makes financial crises more likely; it also prolongs the recovery from the recessions that usually follow, and this result is particularly strong for corporate debt backed by real estate collateral. The likely channel is that corporate debt booms are particularly predictive of post-crisis increases in non-performing loans, again especially when such booms are fueled by real estate collateral, while we find a more muted role of household debt expansions.

Conclusion and policy implications

Taken together, our findings challenge the narrative that household debt is the primary risk factor for macroeconomic stability. While an expansion in household debt is strongly correlated with lower economic growth in the future, booms in specific types of firm debt have a more pronounced link with the macroeconomic disasters associated with financial crises. As such, our findings have potential implications for the design of macroprudential policy, which so far has focused almost exclusively on banks and households. Our takeaway is that corporate debt should be much more carefully monitored, especially in developing countries where financial crises have often been a major disruption to economic growth.

Notes:

A time series dataset contains the values of a variable over time at consistent time intervals. A panel dataset contains the values of a cross-section of variables over time.
Standard deviation is a measure used to quantify the amount of variation or dispersion of a set of values from the mean (average) value of that set.