Canonical macroeconomic theories warn that negative macroeconomic shocks can induce a contraction of credit supply, pushing up spreads, tightening credit limits, and ultimately depressing investment and output (Bernanke et al. 1999). Following the outbreak of COVID-19, however, the US experienced a massive rise in bank lending to firms of nearly 25% (Li et al. 2020). This perhaps counterintuitive credit response is far from unique to this episode, having also been observed following the Lehman Brothers’ default in 2008 (Ivashina and Scharfstein 2010), and following contractionary monetary policy shocks (Gertler and Gilchrist 1993).
How does the corporate sector keep credit flowing in bad times? In a recent paper (Greenwald et al. 2020), we trace the source of these responses to a specific debt instrument, namely, bank credit lines. These facilities provide firms with credit at pre-negotiated amounts and prices, allowing firms with undrawn capacity on these lines to sidestep the deterioration in credit conditions predicted by theory. Using a rich data set that contains the near universe of lending facilities for large US banks, we uncover a number of new facts establishing the central role of credit lines in the aggregate and cross-sectional response of bank-firm credit to macroeconomic shocks.
Our empirical findings demonstrate that credit lines are responsible for virtually all of the growth in credit following the shocks listed above, providing a potential lifeline to the corporate sector in times of crisis. However, this flow of credit may not be wholly beneficial. Our estimates show that draws on credit lines by large firms, who hold the overwhelming majority of undrawn credit line capacity, crowd out lending to smaller and more constrained firms. A structural model shows that this shift in the allocation of credit can actually worsen the resulting drop in investment, despite the increase in aggregate bank lending made possible by credit lines — a mechanism we denote the credit line channel of transmission.
Our empirical study centres on the Federal Reserve’s Y-14Q collection of corporate loans (Y14), which is used for bank stress-testing. The collection consists of all commercial loan facilities that are greater than $1 million and held by large US banks participating in the Dodd-Frank stress tests. The great strength of the data set is its unparalleled view into the loan contracts and financial statements for a wide segment of the US firm population. Our sample includes more than 4.5 million loan observations on over 200,000 distinct firms, 98% of which are privately held, and therefore absent from typical public data sets.
Central to our analysis, these data distinguish loan facilities between credit lines and term loans, and committed but undrawn credit. This decomposition of credit balances is displayed in Figure 1 for a pre-COVID sample, showing that credit lines account for around half of all used bank credit in ‘normal times’. But behind these already large used balances lies an enormous pool of unused credit line capacity that is more than 40% larger than all used bank-firm credit combined. Even after adjusting for typical covenants at the firm level, which may prevent firms from fully accessing their undrawn credit, the aggregate volume of effective unused capacity, indicated by the red line in Figure 1, is still of around the same magnitude as total used credit.
Figure 1 Aggregate bank term loans and credit lines
Notes: The figure shows total used term loans and credit lines across all banks in billion US dollars and 2015 consumer prices. Unused credit is given by the difference between committed and used credit. Covenant limits are computed by applying generic covenant rules at the firm level. Sample: 2012:Q3 - 2019:Q4.
Source: Greenwald et al. 2020.
This unused credit is far from evenly distributed across firms in the economy. Figure 2 displays the cumulative shares of various types of credit across the firm size distribution, showing that while the largest 10% of firms hold around 40% of used credit, they hold nearly 75% of unused credit line balances. This result is the composition of three forces: large firms (i) are more likely to hold a credit line facility, (ii) have larger commitments on those facilities, and (iii) utilise much less of the commitments they have. Beyond size, we find that firms’ unused credit line capacity is positively related to profitability, age, being public, and being investment grade, implying that firms with the greatest access to credit lines in bad times are those that are the least financially constrained, and that have the best non-bank alternatives for financing.
Figure 2 Cumulative shares across firm size distribution
Notes: Cumulative shares of used term loans, used credit lines, unused credit, and unused credit adjusted for generic covenant rules ("Cov") across the firm size distribution. Unused credit is given by the difference between all committed and used credit, and the covenant-adjusted version is computed by applying generic covenant rules at the firm level. The firm size distribution is obtained according to firms’ total assets. Sample: 2012:Q3 - 2019:Q4.
Source: Greenwald et al. 2020.
The role of credit lines in transmission
We next show that credit lines play a central role in the transmission of shocks. First, at the firm level, we find that credit lines are the dominant instrument used by firms to smooth out changes to their cash-flows. Second, at the aggregate level, we estimate impulse responses of various credit instruments to a change in monetary policy. Figure 3 displays our estimated response to a 25bp contractionary monetary policy shock. Total bank-firm credit rises (panel a), and this response is completely explained by an increase in balances on credit lines (panel b), while term loans actually fall (panel c), showing that credit lines are central to the transmission of monetary policy through bank lending. In further analysis, we find that the aggregate responses are driven by large firms with plentiful unused capacity.
Figure 3 Aggregate credit responses to a monetary policy surprise
Notes: Impulse responses to a 25-basis point surprise increase in the two-year government bond yield. 95% and 68% confidence bands are shown. Sample: 2012:Q3 to 2019:Q4.
Source: Greenwald et al. 2020.
Third, we investigate credit movements following the COVID-19 outbreak in the US in March 2020, during which banks played a vital role in channelling liquidity to the firm sector. Figure 4 shows changes in end-of-quarter levels of used and committed credit between 2019:Q4 and 2020:Q1, relative to total used credit in 2019:Q4. As illustrated by the left-most blue bar in panel a, total used credit increased by nearly 25% over this period. The remaining blue bars decompose this growth, showing that it is almost completely explained by a higher use of existing credit lines, as opposed to new credit issuances (Acharya and Steffen 2020). Committed credit amounts changed relatively little as displayed in panel b, implying that our results are entirely driven by higher utilisation rates on the same committed balances. Decomposing these changes by firm size, the orange and yellow bars indicate that the overwhelming majority, around 95% of total credit, flowed to the top 10% of the firm size distribution, compared to merely 40% of used bank-firm credit held by these firms in normal times, a result made possible by large firms’ massive credit line draws over this period.
The credit line drawdowns by these large firms may have put pressure on banks’ balance sheets, leading them to contract other types of lending. We test for such a spillover effect while controlling for firm credit demand using firm-fixed effects, following Khwaja and Mian (2008). We find evidence that credit line drawdowns induce a contraction in credit supply, as banks that experienced larger drawdowns restricted their term lending by more, especially smaller and non-syndicated loans. These effects persisted and actually increased in magnitude over the course of the pandemic in 2020.
Figure 4 Changes in used and committed credit, 2019:Q4 to 2020:Q1
Notes: Blue bars show aggregate changes in used and committed credit across all banks between 2019:Q4 and 2020:Q1, relative to total used credit in 2019:Q4. The orange and yellow bars display equivalent changes for the top 10% and the bottom 90% of the firm size distribution, also relative to total used credit in 2019:Q4. Firm size distribution is computed according to firms’ total assets in 2019:Q4.
Source: Greenwald et al. 2020.
In dollar terms, we find that a $1 credit line drawdown resulted in a reduction of term lending of around 10-30 cents at the bank-level — estimates that likely represent a lower bound on crowding out, as other forms of credit excluded from our analysis such as consumer loans may have been impacted as well. We find that these crowding out effects are not alleviated by deposit inflows, indicating that they are not driven directly by liquidity concerns. Instead, they can be explained by regulatory limits, since the risk weights on credit lines increase substantially when they are drawn. Consistent with this explanation, we find that banks with lower pre-crisis capital buffers restricted their term lending by more following the same drawdowns on their credit lines.
Last, we test whether the crowding out effects actually translated into changes in firm behaviour. We find that smaller firms were unable to compensate the lending cuts with alternative sources of financing, resulting in declines in their total debt and investment. In contrast, the largest firms experienced no statistically significant impact, consistent with their differential access to both credit lines and non-bank sources of credit.
A macroeconomic model with credit lines
To study the influence of the credit line channel on output and investment, we turn to a structural model. Inspired by our empirical results, we assume that only larger, less constrained firms have access to credit lines, which offer favourable, fixed spreads in bad times. In contrast, smaller, more constrained firms depend exclusively on new term loans whose spreads worsen in bad times. Following a shock to productivity, designed to mimic the COVID-19 outbreak, large firms draw their credit lines heavily, crowding out credit to small firms. As a result, while credit lines increase aggregate credit growth following the shock, they cause a large reallocation of credit away from small firms with the highest propensity to invest, toward large firms with the lowest, worsening the drop in aggregate investment. We conclude that while credit lines may keep credit flowing to the corporate sector in bad times, the overwhelming flow of that credit to large, unconstrained firms can dampen or even reverse the stimulating effects of credit growth on investment and output.
For policymakers, our findings can be seen as a rationale for various policy interventions during severe crises that aim to promote lending to constrained firms, especially small- and medium-sized enterprises that depend on bank term lending. At the same time, a number of the Federal Reserve’s interventions in capital markets likely encouraged large firms to use alternative funding sources instead of their credit lines, and thereby eased the pressure on banks’ balance sheets. Last, regulatory treatment of undrawn lines should be carefully calibrated to account for the macroeconomic externalities that we uncover, given our empirical results linking crowding out to bank capital positions.
Authors’ note: The views expressed herein are solely those of the authors and do not necessarily reflect the views of the Federal Reserve Bank of San Francisco, the Board of Governors of the Federal Reserve, or the Federal Reserve System.
Acharya, V, and S Steffen (2020), “The risk of being a fallen angel and the corporate dash for cash in the midst of COVID”, COVID Economics 10:44–61.
Bernanke, B, M Gertler, and S Gilchrist (1999), “The Financial Accelerator in a Quantitative Business Cycle Framework.” Chapter 21 in J B Taylor and M Woodford (eds.), Handbook of Macroeconomics, Amsterdam: Elsevier.
Gertler, M, and S Gilchrist (1993), “The Cyclical Behavior of Short-Term Business Lending: Implications for Financial Propagation Mechanisms”, European Economic Review, 37(2-3), 623–631.
Greenwald, D, J Krainer, and P Paul (2020), “The Credit Line Channel”, Federal Reserve Bank of San Francisco Working Paper, 2020-26 .
Ivashina, V, and D Scharfstein (2010), “Bank lending during the financial crisis of 2008”, Journal of Financial Economics 97(3), 319–338.
Khwaja, A I, and A Mian (2008), “Tracing the impact of bank liquidity shocks: Evidence from an emerging market”, American Economic Review 98(4), 1413–42.
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), 472–50.