Across the globe, economies have been hit hard and fast by COVID-19 (IMF 2021). In the US, unemployment spiked in the first half of 2020 at an unprecedented speed. While such a shock is unlikely to leave banks unaffected, equity buffers have improved significantly since the 2007 financial crisis, and monetary and regulatory policy responses were swift and radical to strengthen the resilience of the financial system (Feyen et al. 2020). In addition, governments stepped in to support the real economy, which indirectly benefitted banks.
What has been the effect of the pandemic on banks’ health and their ability to support the economy with lending? In recent work, we explore if and how US banks’ health has been affected and if there have been changes in lending growth, including in reaction to government support programmes, and in loan conditionality (Beck and Keil 2021).
The US offers a unique laboratory to test the impact of the pandemic and policy reactions. COVID-19 outbreaks were initially concentrated in urban centres on both coasts before the pandemic moved Mid-West and ultimately into the South and Southwest. Similarly, state and county governments across the US have shown quite some variation in lockdown policies. Given the variation in regional exposure of banks, different banks were affected to a different degree by the pandemic as well as at different points in time. This allows us to not only document average trends across banks but also link geographic exposure of banks to pandemic and lockdown measures to banks’ performance.
Granular data on COVID-19 and lockdown policies
We capture exposure to the pandemic by COVID-19 related deaths per 100,000, based on data from the New York Times, except for the five counties that form New York City, which the New York Times sums up into one metropolitan aggregate. For consistency we use CDC data for these counties. To capture lockdown policies, we use the non-pharmaceutical intervention (NPI) index from Olivier Lejeune.1 The NPI index is defined on the state level (there is little to no variation within states), ranging from 0 (no or few containment measures in place) to 6 (harsh lockdown where residents are not allowed to come out of their home) and is computed as the arithmetic average of all days in a quarter.
While there is a high correlation (0.67) between COVID deaths and the NPI index, we find independent effects of pandemic incidence and lockdown policies on unemployment rates across counties and over time.
Measuring banks’ exposure to COVID
We introduce a novel measure to gauge banks’ exposure to COVID-19 and lockdown measures using data on bank branch deposit distributions. Specifically, we use the 2019 bank branch deposit shares in total bank deposits as weights for each county and then – using county-level COVID-19 death rates per capita and a lockdown index – calculate bank-quarter specific measures of COVID-19 and lockdown exposure. We illustrate this idea visually with the examples of Citibank and Zions Bancorp in Q2 2020 in Figure 1. Citi branches (solid red dots) are concentrated in city centres, with a particularly heavy exposure to the New York City metropolitan area – the early epicentre of the US pandemic. Zions (hollow blue circles) is a counter example, operating a relatively dispersed network of locations across the western US with a presence in rural areas and cities less affected by COVID in the first half of 2020. Computed on the bases of new Q2 deaths, this exposure amounts to 67 for Citibank and 13 for Zions.
Figure 1 Examples for differential exposures: Citibank and Zions Bankcorp
Notes: Red dots (blue circles) represent June 2019 Citibank (Zions Bancorp) branches. Citibank (Zions) is an example for a commercial bank with a relatively high (low) geographical exposure to COVID deaths, especially in the first half of 2020. Colouring of contiguous U.S. counties follows a heat map scheme, corresponding to the number of new Q2 2020 COVID-19 related deaths per 100,000 inhabitants. The darker the grey, the higher the death rate.
COVID-19, loan loss provisions and NPLs
Figure 2 shows a general increase in loan loss provisions and NPLs, but more so for banks more exposed to COVID-19. Our first set of regressions results shows that bank exposure to COVID-19 and to NPIs can explain bank variation in loan loss provisions and NPLs over time. The results are not only statistically but also economically significant. In the second quarter of 2020, there was an average 69% increase in loan loss provisions across all banks. The growth rate of loan loss provisions increases by 5 percentage points when bank exposure to COVID deaths doubles, while increasing the NPI index by one notch implies a 20-percentage point increase in the growth rate of loan loss provisions. We have similar, though somewhat weaker findings for NPLs.
Figure 2 Health of banks with differential exposures to COVID
Notes: This figure shows US banks’ median quarterly loan loss provisions on the left and non-performing loans on the right (all indexed to 100 in Q4 2019). The figure differentiates according to banks’ geographical exposure to COVID. The latter is the deposit weighted number of cumulative COVID-19 related deaths / 100,000 inhabitants during the three quarters of 2020. The black (red) line represents the group of banks below (above) the median exposure. The vertical black dashed line indicates the pre- COVID quarter Q4 2019.
COVID-19, lending, and the role of government support
We find a higher increase in total lending growth by banks more exposed to COVID-19 and lockdown policies, but only if we include loans provided under the Paycheck Protection Program , while there is a reduction in lending growth over time when excluding such loans. This contrast is even more striking for lending to small businesses – while small business lending growth increased more for banks more exposed to COVID-19 and lockdown policies when including PPP loans, lending growth was lower for such banks when excluding PPP loans. Figure 3 illustrates this striking difference.
Figure 3 Small business lending volumes by banks with differential geographical exposures to COVID
Notes: This figure shows U.S. banks’ median small business lending volumes (all indexed to 100 in Q4 2019). On the left it includes PPP loans, while it excludes them on the right. Both figures differentiate according to banks’ geographical exposure to COVID. The latter is the deposit weighted number of cumulative COVID-19 related deaths / 100,000 inhabitants during the three quarters of 2020. The black (red) line represents the group of banks below (above) the median exposure. The vertical black dashed line indicates the pre-COVID quarter Q4 2019.
These findings are consistent with an increase in loan demand during the COVID-19 pandemic that outweighed any possibly negative effect of the crisis on loan supply, and are consistent with Acharya and Steffen (2020), Chodorow-Reich et al. (2020) and Li et al. (2020). The substantially larger effects for small business than overall loans (including PPP) are an indication that loan supply to this specific group was supported by policy measures (in line with findings by Chodorow-Reich et al. 2020), while smaller firms also rely more on banks than larger firms that have access to public capital markets.
Comparing changes between the number of small business loans in 2019 and the number of PPP loans in 2020 across banks within counties allows to gauge whether this is driven by supply or demand. Using data for the second quarter of 2020, when most of the PPP loans were given, we find that banks with higher exposure to COVID related deaths or NPIs expanded their lending activity during the pandemic with PPP loans, both along the extensive (lending in counties they have not lent before) and the intensive (higher number of loans in counties where a bank was lending) margins. While it is also consistent with small businesses requesting substitution of regular loans with PPP loans, we do not find a loan demand effect, as PPP lending does not increase (relative to pre-pandemic small business lending) in counties more affected by the pandemic or lockdown policies.
COVID-19 and loan conditionality
Syndicated loan data allow us to track loan conditionality for medium-sized and large firms. Unlike the PPP, there were no direct fiscal policy support measures (though the easing of monetary policies including asset purchases targeted financing conditions for large firms), so that we can more directly observe the impact of pandemic and lockdown measures in this market segment.
Combining our bank-level data on exposure to COVID-19 and lockdown policies with data on syndicated loan contracts, we find a tightening of loan conditionality due to COVID exposures of banks, both in terms of higher interest spreads and shorter loan maturity. Specifically, the interest spread on a new loan increases by 30 basis points as a bank’s exposure to COVID-19 deaths doubles; while there was no bank variation in maturity shortening, loans were, on average, 27 months shorter in the second quarter of 2020. We also find that a doubling in the COVID-19 exposure decreases syndicated bank loan extensions by 15%, while a doubling in the exposure to NPIs reduces average loan volumes by 20%.
Our initial findings suggest that banks are indeed catching Corona, although the effect has not become as obvious from banks’ balance sheets given the easing of regulatory requirements on loan classification and provisioning. We find an increase in lending to support the economy, but driven by government support programmes, as well as a tightening in loan conditionality.
Our findings are consistent with previous research showing an increase in corporate and small business lending during macroeconomic shocks and with work that shows an increase in interest spreads. More generally, our findings are consistent with Gertler and Gilchrist (1993) and Greenwald et al. (2020) of a positive effect of macroeconomic shocks on lending, but also consistent with evidence of an increased risk premium following such a shock (Santos 2010). Compared to previous work, however, we show an important role of bank-level factors in driving the increase in small business loans, especially with government support, rather than a demand-driven take-up of such loans. We also look beyond liquidity and solvency factors in explaining bank reaction to the exposure of banks to pandemic and lockdown.
Acharya, V and S Steffen (2020), “The Risk of Being a Fallen Angel and the Corporate Dash for Cash in the Midst of COVID”, The Review of Corporate Finance Studies.
Beck, T and J Keil (2021), “Are Banks Catching Corona? Effects of COVID on Lending in the US”, CEPR Discussion Paper 15869.
Chodorow-Reich, G, O Darmouni, S Luck, and M Plosser (2020), “Bank liquidity provision across the firm size distribution”, NBER Working Paper 27945.
Feyen, E, T Alonso Gispert, T Kliatskova and D S Mare (2020), “Taking stock of the financial sector policy response to COVID-19 around the world”, VoxEU.org, 17 December.
Gertler, M and S Gilchrist (1993), “The cyclical behavior of short-term business lending: Implications for financial propagation mechanisms”, European Economic Review 37(2): 623–631.
Greenwald, D L, J Krainer, and P Paul (2020), “The credit line channel”, Federal Reserve Bank of San Francisco.
International Monetary Fund (2021), World Economic Outlook, January.
Li, L, P Strahan and S Zhang (2020), “Banks as lenders of first resort: Evidence from the COVID-19 crisis”, The Review of Corporate Finance Studies 9(3): 472–500.
Santos, J A C (2010), “Bank Corporate Loan Pricing Following the Subprime Crisis”, The Review of Financial Studies 24(6): 1916–1943.
2 Under the April 2020 CARES Act’s Paycheck Protection Program loans were made available by the Small Business Administration through banks to small businesses to encourage them to retain or rehire employees that have been furloughed. The loans will be forgiven if certain requirements are met. The PPP lending program was generally targeted at small businesses with at most 500 employees, with an interest rate of one percent and maturities of two years.