The Limits of the Banking Channel

Flows of new bank loans to the corporate sector reached a record high in March in the Euro area, and the ECB’s Bank Lending Survey suggests that credit institutions are planning to ease their credit standards to meet additional demand in the coming months. This is the opposite of what was happening during the Great Recession of 2008-2009 and again during the sovereign crisis of 2011 and 2012. The transmission of the central bank’s decisive stimulus is swift.
Still, the ECB chose last week to add another layer of support through the banking channel by lowering further the interest rate on its targeted long term refinancing operations. This will probably help, but we continue to think that the focus should be on supporting the sovereigns better. Indeed, the governments’ pledge of massively guaranteeing corporate loans plays in our view a key role in the banks’ willingness to lend. This is adding heavy contingent liabilities on the governments at a time when their deficits are rising spectacularly. 
The ECB needs to have “the conversation” on how far it can go with its new pledge to be flexible with its limits on public debt holding as well as on diverging from apportioning outright purchases of govies according to the share of each Member State in its capital. This is key to responding to the disparity in the economic reaction to the pandemic across member states which is already reflected in the first estimates for Q1 GDP.
That banks could provide the necessary support to the sovereigns by recycling the proceeds of the TLTRO has become popular since last week. The loop would be closed: governments would backstop banks, and banks would fund the governments. We explore this in some detail. “Circuits” of that nature were frequent in the late 1940s/1950s, with some short-term success but at large long-term cost. We don’t think that the current monetary set-up is actually consistent with such “loop” anyway. Judging by their communication last week both C. Lagarde and P. Lane are fully aware they will have to do more to help government directly. The German Supreme Court’s decision on quantitative easing on 5 May might tell them what their room for manoeuvre is.

Asymmetry confirmed

We expressed in Macrocast last week our concern that the pandemic is no longer a symmetric shock affecting all economies in broadly the same way but is in fact adding to the dispersion in growth trajectories. Unfortunately, the early estimates for Q1 GDP published last week for the US and the Euro area strengthened our view. True, GDP is contracting everywhere, but the decline in the US (-1.2% quarter-on-quarter) was half of what was seen at the worst of the Great Recession of 2008-2009, whereas in large swathes of the Euro area the drop was much more spectacular.

Last week we presented a “mechanical trajectory” in which activity 30% below normal for 2 weeks in March would result in a 5% qoq contraction in GDP in Q1. France, Spain and Italy came out in this ball-park. We don’t have yet the figure for Germany (it will be out on 15 May), but we can infer from the Euro area estimate that GDP there fell by about 2%. This would be consistent with the difference in the severity of the lockdown between “Latin European” countries and Germany revealed by real-time indicators. The same holds for the US, but in addition there the lockdown came later than in most of Europe (on average in the second half of March only roughly 40% of American States were in lockdown). We update in Exhibit 1 a graph originally produced by Oxford Economics, linking the degree of severity of the lockdown and Q1 GDP “de-growth”. The correlation is impressive.

Everywhere, Q2 is likely to be significantly worse than Q1 with more weeks affected by the lockdown and only gradual normalization after that. In our “illustrative trajectory” last week we had a 15% qoq contraction. But divergence is likely to be even more obvious. According to the latest “community report” by Google, movements close to workplaces were only 18% below trend for the week of April 26 in Germany, against 29% two weeks before. There was also some improvement in France but the gap to trend was still at -43%. Germany was having a “mild lockdown” and is also starting its relaxation relatively early. For the US, since the lockdown started later than in Europe, a longer one in Q2 was a plausible assumption. However, a significant number of States are already relaxing the containment measures, for instance Texas, which accounts for 10% of the US GDP. 

Beyond the depth of the recession in 1H 2020, what is crucial is avoiding a relapse in the second half of the year. Last week we were concerned by developments in Singapore, but there authorities have just stated they’ve managed to contain the new clusters (dormitories for immigrant workers). In Denmark so far where the lockdown started being rolled back on 15 April already, the infection rate continues to be in check. Consequently, we maintain our baseline of a decent rebound from Q3 onward, but for this to hold, we need to be able to count on the continuation of a very accommodative policy mix and its swift transmission to the private sector. The behavior of the banking industry in Europe has been reassuring so far, even if we continue to think that ultimately the ECB will have to do more to support the sovereigns, which in our view will remain key to safeguarding the private sector’s solvency. 

Record bank lending in the Euro area

The European Central Bank continues to fine-tune its emergency response in real time, and the number of occasions Christine Lagarde used the word “flexibility” in her Q&A last Thursday was remarkable. Still, the announcements by the ECB were concrete for banks – another cut on the interest rate of the Targeted Longer-term Refinancing Operations (TLTRO) and the creation of a new lending facility at negative interest rate, the Pandemic Emergency Long Term Repurchase Operation (PELTRO) – while we had only hints, albeit strong ones, at the possibility to increase in size and duration the Pandemic Emergency Purchase Programme (PEPP). We were not expecting anything on the latter (the decision by S&P to maintain Italy’s investment grade status gave them some space), but we were a bit surprised by their renewed focus on the banking channel. In the order of priority, we think that clarifying the quantum and duration of support for sovereigns would trump further improving liquidity conditions for banks.

The bank lending survey released last week and conducted between 19 March and 3 April, when almost all of the Euro area was in lockdown, unsurprisingly reflected a very significant rise in loan demand by businesses, reaching its highest level since 2006, and the highest since the survey was created in 2003 for short-term loans, as companies are trying to plug the (hopefully) temporary hole in the cash flows. Banks are expecting a continuation of this trend, as the “expected credit demand” component has reached a record level. Perhaps more surprisingly, the “credit standards” component did not reflect any significant tightening on the supply-side and banks even expect a net loosening in coming months (see Exhibit 12. Less than 10% of banks reported a net increase in their rate of rejection of loan applications from businesses.

At first glance, this would suggest that there is no threat to credit origination, at least for now. The monthly flow of loans to the private sector in March has been by far the highest on record (see Exhibit 3), while there remains a lot of room in the cumulated amount of debt the various Euro area governments announced they were ready to guarantee. We estimate that business cash flows fell by about EUR120-150bn during the 2 weeks of lockdown of March. The flow of new loans to businesses (EUR 115bn) was almost large enough to cover this shortfall – while we know that at the same time governments were providing significant support (e.g. through in-work unemployment benefits). Actually, while businesses were leveraging themselves, they were also hoarding cash, since their bank deposits rose by a whopping EUR 100bn in March. It is likely that firms drew on credit lines early to cover them into Q2 and possibly beyond.

Earlier ECB action – and specifically the first increase in the generosity on the TLTRO announced on March 12 – clearly played a significant role in this. The Bank Lending Survey suggests growing appreciation by banks of the support granted by the TLTROs. 48% of respondents now expect the facility to contribute to their origination volume of loans to corporations, from 2% in January.

This would suggest there was no pressing need to make the TLTRO even more generous last week. While only 32% of the banks in the BLS sample participated to the March operation, 51% stated their intention to take part in the next one before they knew of the ECB’s latest concessions.

It seems the ECB is choosing a “better safe than sorry” approach and of course, the additional drop in the cost of the TLTRO will help further, as banks would either be incentivised to lend more (thanks to a rise in their margins) or to reduce the interest rate on their loans thus supporting businesses.

Still, ECB support through ultra-cheap medium-term liquidity provision is a key contributor to sustained supply of credit, but it is not a sufficient condition. Indeed, cheap liquidity – against which collateral needs to be pledged – does not insure against credit risk.

We could see two non-mutually exclusive explanations behind the restraint on credit standards. First, that banks genuinely believe that we are facing a temporary shock with few medium-term consequences for firms’ credit worthiness. Second, that the guarantees offered by the governments are providing enough reassurance to the banking sector. We put more faith in the latter than in the former.

A key benefit of a state guarantee on a loan is not just that it provides effective support if, and when, the borrower defaults, but that it also immediately protects the banks’ capital ratio. Indeed, banks can substitute to the risk-weight applied to the borrower that of the guarantor (for the share of the loan that is guaranteed). Fortunately, the European regulator has so far resisted calls for introducing non-zero risk weight on sovereigns.

Creating a “backstopping loop”?

But of course, pledging massive guarantees on corporate debt has triggered a huge transfer of risk from the banks to the sovereigns. The Italian government alone has pledged guarantees of EUR400bn via different channels, nearly a quarter of its GDP. The rise in their contingent liabilities would normally result in higher funding costs, especially since guaranteeing bank loans is only one aspect of their effort to ensure the survival of the business sector and fiscal deficits will grow massively. So, in the end we need a succession of interconnected backstops: the banks backstop the corporate sector with emergency loans, the state backstops the banks with guarantees and the ECB backstops the governments with quantitative easing.

However, we argued last week that the ECB’s backstop is not infinite. True, the central bank is certainly open to changing the size or the duration of PEPP, as well as tweaking the eligibility rule to make it explicit that “fallen angels” could still be bought by the central bank were they to lose their investment grade status. Judging on both Christine Lagarde’s Q&A and Philip Lane’s blog the following day, their finger is on the buzzer for this. Still, some thorny issues cannot be eluded forever. How hard is the 33% issuer limit? If the capital key can be suspended for the flows but not on the stocks – implying a re-convergence at some point - how can the most fragile states be protected for long, unless the ECB intends to continue PEPP well beyond December 2020 and re-invest it over a long period of time?

One temptation would be to close the loop and help the banks support the sovereigns, in a nutshell “backstopping their own backstop”. What the ECB could not do directly beyond a certain threshold, it could nudge banks to provide. This has been a popular explanation since Thursday, but we think the conditions for this to work are not met. Our preferred interpretation of the choice to focus on the banks first and the sovereigns later is very simply that this is what is the least problematic politically and legally.

When your humble servant was sitting the Banque de France exams – quite a few geological layers ago – the key to success was mastering a thick book by Jean-Pierre Patat (the Banque’s then head of monetary statistics) titled “Money, Financial Systems and Monetary Policy”. His approach was often historical. In a nutshell his narrative was how France and most other European economies gradually extricated themselves from heavily regulated financial management skewed towards funding the government as a legacy of WW2, to a decentralised, market-based approach.  The similarities between some of the most baroque aspects of the 1950s and the current policy discussions are uncanny, but ultimately misleading.

In 1944, Banque de France was granted the possibility to refinance some categories of loans with a maturity of up to 5 years, usually underwritten by government entities as they were contributing to the country’s reconstruction. That sounds a lot like refinancing state-guaranteed “pandemic emergency” loans through a TLTRO. Nothing new under the sun. But crucially French banks were not free to invest their excess liquidity as they saw fit. They were in fact forced to fund the government at an administratively-set interest rate which was often negative in real term (initially 20% of the growth in their deposit base had to invested in government debt). This was the embodiment of the “financial repression” which we alluded to a few weeks ago in Macrocast.

Let’s fast-forward to 2020. Could the liquidity created by the TLTROs find its way to government debt? Some of it already is, undoubtedly. Indeed, banks are increasing their holdings of govies spontaneously. In Exhibit 2 we plotted in the dotted line the flows of purchases of government bonds by banks together with the flows of loans to the corporate sector. In March, they also reached their highest level ever since the creation of the series in 2003. Interestingly, they amounted to EUR 127bn, a bit more than the net purchases conducted by the ECB through the Public Sector Purchase Programme and the PEPP over the last 4 weeks.

Still, two hurdles would impair the creation of a solid “banks/sovereign loop” funded by central bank liquidity.

First, banks have access to a risk-free substitute to government bonds: holding excess reserves at the central bank. By pushing the funding rate of banks below the deposit rate through the TLTROs, the ECB is supporting banks’ profitability and hence possibly their willingness to lend, but this also means that credit institutions will earn a carry trade of 50 basis points by parking the proceeds of the TLTROs on their account at the central bank. Of course, they would earn an even larger margin by investing their cash into government bonds, at least in the periphery, but they would then take a capital risk which cannot be ignored given the recent market gyrations.

Second, producing a significant relief for the governments’ aggregate funding cost would entail nudging banks towards the long end of the curve (the average maturity of Italy’s public debt is 7.3 years for instance). Despite its extension by one year, the TLTRO’s term is three years. Banks would take a maturity mismatch risk by “transforming” TLTRO money into longer term bonds. The market reaction after the ECB announcements last week was quite telling. While the Italian 3-year yield fell by c.10 basis points, the 10-year yield rose.

Establishing a “proper” “bank/sovereign” loop would entail at least massively extending the term of the TLTRO, but also probably some quantum of coercion which is for now not on the radar in Europe, a complete turning back from the market-based monetary policy which has been the norm there since the 1980s. Trying to replicate with price signals only the characteristics of old administrative management systems based on financial repression is vain in our view. After six weeks of lockdown, you may excuse your humble servant’s final bout of nostalgia when quoting some famous words from the man who was Prime Minister in France at the time when Patat’s book was on his reading list: “one does not play games with the market. Its logic is global”. France dealt with the legacy debt of WW2 and funded two colonial wars with its coercive monetary policy. Still, ultimately the price to pay in terms of stubborn inflationary pressure, exchange rate volatility and under-development of its financial system was high.

We may not come back to the Fifties. We don’t think the ECB is pursuing a “bank/sovereign loop” strategy. The decisions last week were in the continuation of its action in March to deal with liquidity issues. A sign of this is the creation of the PELTRO, which is a way to make sure banks which have exhausted their TLTRO allowance or whose business model is skewed towards lending to non-corporates can access cheap liquidity (only loans to corporations are considered when calculating the TLTRO allowance). The extra nudge to lending to businesses triggered by the further drop in the TLTRO rate is not essential in our view. Governments are currently the guarantor of the solvency of the whole productive system. Ultimately the ECB will have to have “the conversation” on how flexible the limits to QE actually are. The decision by the ECB to postpone decisions on these matters may owe a lot to the decision of the German Supreme Court on the legality of QE, on May 5th. The Court cannot impose anything on the ECB itself, but it could instruct the Bundesbank to stop participating to some of its activities. It probably makes sense not to take risks by “pushing the envelope” on PEPP too far while we wait.