After the Global Crisis of 2008, the main central banks adopted non-standard monetary measures to provide economic and financial stimulus to their economies. Daetz et al. (2018), De Marco (2019), and Andrade et al. (2016) analyse the effects for the ECB, while Krishnamurthy and Vissing-Jorgensen (2011), Chakraborty et al. (forthcoming), Maggio et al. (2016), among others, do the same for the US.
The assumption is that changes in credit are driven by supply, implying that liquidity and a capital injection in the banking sector can counter a credit shortage. But the interaction with low demand may reduce and even distort the effect of any injection of liquidity. Therefore, it is important to understand the transmission channels of these measures.
Separating supply channels from demand channels requires individual information at the bank and the firm level (Mian 2012). The Banque de France recently opened its credit registry containing every corporate loan provided by every bank in France. Using this, we were able to use a change in collateral requirements in February 2012 as a natural experiment to examine causal effects (Delatte et al. 2019). We find that credit increased after the liquidity injection and this was exclusively driven by supply. We also found strategic risk-taking by a group of banks, which was an unintentional implication of the policy.
A natural experiment
As part of a broader set of unconventional monetary policy measures, the ECB announced on 8 December 2011 that it would allow national central banks to accept additional credit claims (ACC) as collateral from borrowing banks, starting in February 2012. Loans to firms with credit rating of 4 would be eligible as collateral, whereas previously eligibility cut off at 4+. A rating of 4 on the Banque de France’s scale is approximately equivalent to a Fitch rating of BB-. The firm’s debt would be considered speculative, with a one-year default probability of between 0.4% and 1%.
This overnight re-classification of loans on bank balance sheets by the ECB created an ideal setting to discover whether there was a causal effect. We examine the effects of the ACC on the bank-firm level credit volume through two channels:
- A supply channel: The ACC policy implied that loans to 4-rated firms, previously sitting idle on the asset side of the banks’ balance sheets, were suddenly more valuable due to their ability to act as collateral. This created a positive externality for all firms – provided that a bank chooses to use the new collateral to access liquidity from the ECB.
- A demand channel: After the ACC policy, a subset of firms – the 4-rated firms – have a marginally lower external finance premium, because bank loans to the assets of these firms can be used for refinancing purposes.
In the French credit registry, administered by the Banque de France, we can also observe the amount of 4-rated loans on banks’ lending portfolios before the ACC policy implementation, and so identify banks that benefited from a sudden and exogenous increase in their capacity to borrow money from the central bank. These are our treated banks. We compare them to control banks – i.e. the banks that were less active in issuing the targeted loans before the policy.1
The company database compiled by Banque de France also captures firm balance-sheet information and credit-rating information. So, we are able to identify firms with a rating equal to 4 and better (i.e. the firms that benefited from a lower borrowing costs after the relaxing of collateral requirement). We study these effects over a three-year period between January 2011 and December 2013.
Bank and firm response
We find the expected positive effect of the ACC policy on lending volume by the banks that hold larger-than-average amounts of newly eligible collateral in their portfolio. The ACC policy intervention worked as a capital injection for these banks and stimulated their credit supply.
Surprisingly, this effect does not work for all banks holding a large amount of 4-rated loans. On the contrary, banks with the highest concentration of newly eligible collateral contracted credit. These are small, risk-averse banks whose priority after the recession was to strengthen their balance sheets. Given additional liquidity in the form of newly eligible collateral, they would hoard the additional credit if they prioritised liquidity, but that does not explain a fall in credit.
Zooming in, we find that they contracted lending to firms rated 4 and worse (i.e. the riskiest firms). Instead of expanding their balance sheet, they used the policy as a positive income effect to reduce the level of risk of their portfolio; they could maintain its value and reduce the amount of higher risk loans.
This had not been observed before. The supply channel did not work homogeneously – banks reacted differently, depending on their concentration of newly eligible collateral and their need for liquidity.
But, this might be the case if demand for credit was low. To be sure, we zoom in the effect of the ACC policy on credit to the firms eligible for collateral (firms with a credit rating of 4 or better). These firms benefited from a lower risk premium after the ACC. If they had demanded more credit after the reduction in their borrowing cost, then we would observe that they borrow more from all banks. However, we find that in response to the ACC policy, treated banks increase credit supply to them but that same group of firms gets less credit from the control banks. This suggests that the observed credit increase is driven by supply-side factors only.
Figure 1 plots the credit volume (in log) along the different bank and firm dimensions. It indicates shifts in the credit market after the ACC: control banks increased their lending to high-risk firms, while treated banks reduced their exposure to them.
One possible explanation is a voluntary risk-shifting strategy by control banks in search for higher interest rate returns via risk premium. Such behaviour could be motivated by increasing profit-shares to compete with banks that benefited from the ACC policy.
Figure 1 Evolution of credit across treated and control banks, low- and high-risk firms
Source: Delatte et al. (2019).
Last, we examine the effect of ACC on small and medium enterprises, as expanding credit supply to this category of firms was one of the core motivations to accept lower-rated securities as collateral. But we find that treated banks reduced credit supply to small firms and expanded it to large firms.
None of these results overturns the conclusion that overall credit expanded as a result of loose unconventional monetary policy. But there is a fundamental heterogeneity in response across banks – not all banks increased credit, and not all risky firms had improved access to credit.
There might even exist a distribution of bank portfolios in which the policy could lower aggregate credit supply. In this case, the heterogeneity would have first-order importance.
Andrade, P, J H Breckenfelder, F De Fiore, P Karadi, P and O Tristani (2016), “The ECB’s asset purchase programme: an early assessment”, European Central Bank working paper 1956.
Chakraborty, I, I Goldstein, and A MacKinlay (forthcoming), “Monetary stimulus and bank lending”, Journal of Financial Economics.
Daetz, S L, M G Subrahmanyam, D Y Tang, and S Q Wang (2016), “Did ECB Liquidity Injections Help the Real Economy?”, New York University.
Delatte, A L, J Imbs, and P Garg (2019), “The transmission channels of unconventional monetary policy: Evidence from a change in collateral requirements in France“, CEPR Discussion Paper 13693.
De Marco, F (2019), “Bank lending and the European sovereign debt crisis”, Journal of Financial and Quantitative Analysis 54(1): 155–182.
Krishnamurthy, A and A Vissing-Jorgensen (2011), “The effects of quantitative easing on long-term interest rate”, Brookings Papers on Economic Activity 2011, number 2.
Di Maggio, M, A Kermani, and C Palmer (2016), “How quantitative easing works: Evidence on the refinancing channel”, NBER Working Paper 22638.
Mian, A (2012), “The case for a credit registry”, In Markus Brunnermeier and Arvind Krishnamurthy (eds.), Risk Topography: Systemic Risk and Macro Modeling, University of Chicago Press: 163-172.
 One pitfall of identification would be that banks might respond to the ACC policy by altering their lending portfolio. So we used the composition before the ACC policy, because the policy was not anticipated by banks. We did not observe a persistent trend in extending ACC targeted loans before the implementation, though the picture changes after it is implemented.