‘Stress tests’ for banks as liquidity insurers in a time of COVID
During the last few weeks, the spread of COVID-19 has rattled the global economic prospects and financial markets. Growth and employment forecasts for the global economy have been revised substantially downward to recession levels for the next two quarters; global stock prices have declined, especially for banks and other financial intermediaries; credit spreads have surged, notably for junk-rated paper in developed economies; and central banks have reacted with substantial rate cuts and/or liquidity provision and asset purchase programs.
The root of the present stress is different compared to the primary cause of the 2008-2009 global financial crisis (GFC). Importantly, the present stress did not originate in the banking system, as was the case for the GFC; pre-GFC, the banking system was over-leveraged with poor underwriting decisions in the housing sector, and the household sector was over-leveraged too. The root cause of the current stress is a pandemic. Containing the virus and the drastic steps (from social distancing to isolation) that governments need to undertake have an immediate impact on the real economy through the simultaneous occurrence of both demand and supply shocks, with attendant financial sector spillovers and side-effects such as the oil-price war. In particular, debt repayments will come due as usual, but liquidity appears to be quickly evaporating for both small and large companies as economic activity grounds to a virtual halt.
In spite of their better capitalisation and liquidity positions relative to pre-GFC, the banking sectors in developed economies, notably in the US, have already been under severe pressure as bank stock prices have declined about 40-50%. One important reason behind this could be the role of banks as liquidity insurers for the real economy. While non-bank financial institutions took over a large share of corporate financing over the past decade, particularly of highly leveraged firms, banks remain the main source of liquidity insurance for all firms – investment-grade, non-investment grade and also unrated firms. Indeed, as leveraged loans and bonds financed by the non-bank financial sector come due, firms may draw down on the bank lines of credit, causing credit to re-intermediate (involuntarily) to the banking system.
In other words, if firms experience liquidity problems or are uncertain about future liquidity needs or availability, they might use the insurance arranged with banks and start drawing down their credit lines. Worse, liquidity needs can potentially become highly correlated among firms and they could start running on their banks en masse. Banks could then experience substantial liquidity problems themselves, which could be further elevated if other short-term creditors of banks stop rolling over their funds.1 Moreover, drawdowns require additional bank capital as they manifest as loans on bank balance sheets, constraining the ability to provide further, new loans to the economy, and in some cases, potentially also bringing banks closer to insolvency. Liquidity problems can this way quickly morph into solvency problems.
It thus makes sense as a ‘liquidity stress test’ to quantify the likely insurance function banks may have to provide involuntarily at this stage of the COVID-19 scenario and how large it looks relative to their balance-sheet buffers in terms of capital and liquidity needs. It is also useful to investigate what the past stressed experiences for bank credit-line drawdowns can tell us about the likely outcomes in terms of financial markets and real lending.
We use data for non-financial firms headquartered in the US to illustrate the nature of the present liquidity stress for banks. We obtain data on undrawn credit lines as well as balance sheet information from Capital IQ as of Q4 2019. Overall, our sample comprises 2,425 US non-financial firms with non-missing information on outstanding credit lines. We do not condition on the nationality of banks, i.e., liquidity insurance is provided by both US as well as non-US banks.
Figure 1a shows the industry distribution of our sample firms. Firms are highly concentrated, with about 95% of all firms operating in five industries: manufacturing (54%), services (20%), transportation (9%), mining (7%), and wholesale (4%).
Figure 1a Industry distribution of US non-financial firms with undrawn credit lines
Quantifying the liquidity insurance of US non-financial firms
The total amount of outstanding credit lines at the end of 2019 is $958 billion – in other words, on average, each firm in our sample has outstanding loan commitments of about $397 million. Assuming that firms have not significantly drawn down these commitments in the first month and a half of 2020, this is the amount of liquidity insurance banks had provided to the US non-financial firms when the COVID-19 outbreak started. The outstanding amount of undrawn credit lines represents, on average, 81% of total committed credit lines and is about eight times the amount of bank debt that is already on firms’ balance sheets at the end of 2019.
Figure 1b shows the rating distribution of these US non-financial firms as of Q4 2019 – 70% of firms are unrated, 18% non-investment grade rated, 8% BBB rated, and only 4% are AAA/AA or A-rated. The construction sector has the largest share of credit lines outstanding to non-investment-grade rated firms (43%), followed by the services industry (39%) and then mining (35%).
Only 18% of firms in the manufacturing sector, which have the largest dollar amount of outstanding credit lines, are non-investment grade rated. However, this sector comprises a large number of firms that are BBB rated (i.e. these firms are at the investment-grade boundary). Their access to credit can worsen substantially if they are (close to being) downgraded, which might increase their demand for liquidity.
Figure 1b Rating distribution of US non-financial firms with undrawn credit lines
Table 1 shows the bank exposure to undrawn credit lines by rating class of firms. More than 57% of the commitments are provided to firms either below investment-grade or which are at the investment-grade boundary; 27% only to high quality (AAA, AA or A-rated) firms.
Table 1 Exposure by rating class
Why is this important? Because not all firms are equally likely to use their liquidity insurance. Firms exercise their trade-off between using insurance now versus saving it for later; low-quality firms, however, are in general more likely to draw down their credit lines when credit markets tighten as they are closer to their default thresholds. Another important implication is that if central bank support in the form of asset purchases is primarily geared towards high-rated firms’ corporate bonds (as is usually the case), then it is again the low-rated segment of firms that is likely to rely upon the banking sector by drawing down its credit lines.
Link between the commercial paper market and bank credit lines
The liquidity insurance function of banks is intimately linked to the commercial paper market – a short-term, unsecured credit market for rated non-financial firms – that directly finances a wide range of economic activity. Many firms issue credit lines as a backstop at the same time they issue commercial paper, as credit lines are funds that are available through the banking sector if firms experience problems in refinancing their outstanding commercial paper. This form of credit ‘enhancement’ enables the commercial paper to acquire credit rating that makes it eligible for investment by money-market funds.
Changes in the commercial paper rates can be used to gauge stress in the commercial paper market. For example, starting 16 March 2020, the rate of lower rated A2/P2 commercial paper increased from about 1.5% to 3.5%, and even the rates of highly rated A1/P1 commercial paper increased. To avoid a disruption in economic activity, by close of 17 March 2020 the Federal Reserve had established a Commercial Paper Funding Facility (CPFF) of $10 billion as credit protection against rollover risk.2 This is aimed at allowing the commercial paper to get rolled over, reduce (any) money-market fund redemptions, and in turn, substantially reduce the likelihood that commercial-paper issuers will use their liquidity insurance and drawdown bank credit lines.
At the end of 2019, the US firms in our sample had a combined volume of outstanding commercial paper of about $131 billion. Even if we assume that this is fully backed by credit lines which firms do not intend to use, or are unlikely to use given the Fed backup of the commercial paper market, the remaining exposures to credit lines is nevertheless still quantitatively large at $887 billion as this amount might potentially get drawn down.
Credit markets have also come under stress since the outbreak of the crisis. Figure 2 shows the LSTA Leveraged Loan Index since January 2018 which shows the daily price of the index that contains about $1.5 trillion in outstanding leveraged loans. Loan prices dropped by 9 percentage points (or more than 10%) within a few days.
Saunders et al. (2020) show that changes in the spreads of secondary loans forecast business cycle activity in the US (GDP, employment, investment) because they reflect loan supply conditions in the primary loan market that feed into the wellbeing of the real sector. Loan market spreads contain additional information vis-à-vis other credit spread measures such as bond spreads, as the loan market features a different borrower segment compared to the one that accesses bond markets – in particular, firms that feature higher default risk and whose access to external funding is more limited compared to other firms (for example, unrated firms). In other words, tightened spot credit market conditions might encourage particularly these firms to draw upon their bank credit lines.
Figure 2 US leveraged loan index (LSTA)
How much do firms draw down their credit lines?
Berg et al. (2016) provide some evidence as to which firms utilise their credit lines using US data over the 1986 to 2011 period. Table 2 (taken from this paper) shows the percentage draw-down of firms as a function of their respective stock market performance over the last 12 months. Using average draw-down rates over the long time-series of 25 years allows the authors to use information from two recessions, the 2002 to 2003 period as well as the GFC.
Table 2 Drawdown of credit lines of US non-financial firms
Berg et al. (2017) emphasise the sensitivity of credit line drawdowns of US non-financial firms to changes in the performance of the S&P 500 index. For example, a 25% decline in the S&P 500 index over the previous 12 months increases the usage of credit lines by 5 percentage points; a decline of 40% increases the usage by about 8 percentage points. Given an average usage rate of 24% during their sample period, an 8 percentage point increase in usage rates implies an increase of drawdowns of about one-third.
It is a reasonable ‘stress scenario’ in the present context that firms will experience a stock performance consistent with the lowest quintile of equity returns shown in Table 2 for the two preceding recession periods. We can thus apply these historical draw-down patterns to the current environment as a liquidity stress scenario. Table 3 shows the calculation of the expected drawdowns of US firms, which amounts to about $264 billion based on the historical drawdown rates and present outstanding amounts for undrawn credit lines.
Table 3 Expected drawdown versus total exposure
Using average usage rates at the end of 2008 provides a similar picture (Table 4). Expected usage based on the 2008 draw-down rates is about $244 billion. Note that drawdown rates spiked around the Lehman Brothers default in Q3 2008 and already abated in Q4 2008, which explains the difference with the expected usage rates based on equity performance.
Table 4 Expected drawn-down vs. total exposure – 2008 usage rates
If we even assume a 40% decline in the S&P500 index in a stress scenario, the predicted drawdowns would increase to about $312 billion – a 20-30% increase relative to the baseline calculations in Tables 3-4.
How big is the estimated liquidity stress due to credit line drawdowns for the banking sector?
While we do not know the identity of the banks providing the undrawn loan facilities that we analyse above, banks provide information regarding their aggregate undrawn commitments in their quarterly filing. We collected this information for the 100 largest US banks at the end of 2019 from SNL Financial as well as other balance sheet characteristics.
Assuming that their outstanding commitments are drawn as assumed in Table 3 (using an average draw-down rate of 33.2%), banking sector Tier 1 capital ratio (defined as Tier 1 capital as a percentage of risk-weighted assets) declines from 12.7% to 11.8% on average as a result of loans being added to bank balance sheets. In comparison, banks entered the GFC with substantially lower Tier 1 ratios (lower by more than 20% of the present levels). Given the better capitalisation and liquidity position of US banks and the additional liquidity that can be accessed through the Federal Reserve, a potential solvency problem for US banks due to (even correlated) credit line drawdowns appears to be limited.
If the actual market performance scenario turns out to be more adverse than the scenario assumed in Table 2 for the weakest quintile by stock performance, or worse than the GFC – possibly because market corrections turn out to be far greater – then the resulting strain from credit line drawdowns on US banks would bring them closer to the regulatory capital requirement. In fact, data collected by S&P Global Market Intelligence based on public company filings show that 71 firms have almost fully drawn down their credit lines since 5 March 2020 ($73 billion out of $86 billion in commitments). The drawdowns are concentrated in the most affected industries (hotels, restaurants and retail) and concentrated among few large banks.
Thus, if we assume that firms fully draw down their credit lines, banks’ Tier 1 ratio would decline by 2 percentage points to 10.7%. There is some heterogeneity across banks as to how much their Tier 1 ratio declines, with some banks even falling below the 8% capital ratio. Worse, this would be an extremely adverse liquidity stress scenario that might coincide with other stress factors eroding bank capital such as an increase in default rates on banks’ loan portfolios. Realised volatility in some asset classes has risen to close to 100% and the volatility index VIX (the ‘fear gauge’) has reached levels of above 80 in March 2020, emphasising the high level of uncertainty around the fallout from COVID-19 for financial markets going forward.
In summary, it is a testament to the success of the post-GFC reforms of the financial sector that the present levels of capitalisation in the banking sector appear adequate to deal with liquidity stress tests that feature undrawn credit line withdrawals in line with the past two recessions or the global financial crisis. Regulators should persist with the success of this effort and plan in advance for possibly more severe stress, given the heightened uncertainty at present for the global economy and financial sectors. A minimum immediate regulatory response could be to preserve bank capital by ruling out any capital erosion from dividend payouts or share buybacks. Such capital preservation may even allow for countercyclical relaxation in capital requirements down the line should severe liquidity stress materialise.
Acharya, V and N Mora (2015), “A Crisis of Banks as Liquidity Providers”, Journal of Finance 70(1): 1-44.
Acharya, V and S Steffen (2015), “The Greatest Carry Trade Ever? Understanding Eurozone Bank Risks”, Journal of Financial Economics 115(2): 215-236.
Berg, T, A Saunders and S Steffen (2016), “The Total Cost of Borrowing in the Loan Market – Don’t Ignore the Fees”, Journal of Finance 71(3): 1357-1392.
Berg, T, A Saunders, S Steffen and D Streitz (2017), “Mind the Gap: The Difference between U.S. and European Loan Rates”, Review of Financial Studies 30(3): 948-987.
Ivashina, V and D Scharfstein (2010), “Bank Lending During the Financial Crisis of 2008”, Journal of Financial Economics 97(3): 319–338.
Saunders, A, A Spina, S Steffen and D Streitz (2020), “What’s in the Spread? The Predictive Power of Loan vs. Bond Spreads”, Working Paper.
1 See, for example, Ivashina and Scharfstein (2010) and Acharya and Mora (2013) on liquidity-shortfalls at U.S. banks during 2007-08 GFC, and Acharya and Steffen (2015) in relation to the ‘slow run’ by US money-market funds on European banks in the autumn of 2011.