Sabine Lautenschläger: A stable financial system - more than the sum of its parts

Keynote speech by Ms Sabine Lautenschläger, Member of the Executive Board of the European Central Bank and Vice-Chair of the Supervisory Board of the European Central Bank, at the Dutch Banking Day 2018, Amsterdam, 15 February 2018.

The views expressed in this speech are those of the speaker and not the view of the BIS.

Logic can be a tricky thing. Apply it in the right way, and you always arrive at a consistent conclusion. But apply it in the wrong way, and it can lead you astray. And that happens all too easily.

There are indeed many wrong ways in which we can apply logic.

One of them is known as the fallacy of composition. It refers to the idea that the whole always equals the sum of its parts. Well, that idea is wrong. As we all know, the whole can be more than the sum of its parts - or less.

Consider this statement: if each bank is safe and sound, the banking system must be safe and sound as well. By now, we have learnt the hard way that this might indeed be a fallacy of composition.

Let me give you just one example. Imagine that a certain asset suddenly becomes more risky. Each bank that holds this asset might react prudently by selling it. However, if many banks react that way, they will drive down the price of the asset. This will amplify the initial shock, might affect other assets, and a full-blown crisis might result. Each bank has behaved prudently, but their collective behaviour has led to a crisis.

The business of banking is ripe with externalities, with potential herding and with contagion. These factors may not be visible when looking at individual banks, but they can threaten the stability of the entire system. This is one of the core insights from the financial crisis.

Countering the cycle and boosting resilience

Indeed, the financial crisis showed us that we had to do more than just look at each part of the system, at each bank. The system as a whole had to be taken into account as well. The microprudential approach must be complemented by a macroprudential approach.

And thus, a new policy area was born out of the crisis. Age-wise, macroprudential policy could be the grandchild of policy areas such as banking supervision and monetary policy. It still has to grow up. Theories need to be elaborated, empirical evidence gathered, and practical experience gained. A long childhood lies ahead.

But at the same time, much has already been achieved. Over the past few years, we have set up a macroprudential framework for Europe. We have defined objectives, developed tools and set out how different policy areas interact. Let us have a closer look at how macroprudential policy works in Europe.

A core concept of macroprudential policy is the financial cycle, that steady succession of ups and downs, of booms and busts. And each time the cycle turns, each time boom goes bust, everyone runs for the exit. Market participants sell the assets concerned, and prices tumble; investors incur losses, as do the banks which financed their investments. This shock might even hit banks which were prudent enough not to participate in the boom. The financial system is so interconnected that even they cannot escape contagion. And as goes the financial system, so goes the real economy. Credit dries up, growth drops off and everyone suffers.

This story is a bit simple, of course. Still, it allows us to define two goals of macroprudential policy:

• First, make the financial system as a whole more resilient.
• Second, dampen the cycle of booms and busts. In very general terms, this is what macroprudential policy aims to do in Europe.

The tools of macroprudential policy reflect these objectives. Some of them focus on increasing the resilience of market participants. Take, for example, the countercyclical buffer. It requires banks to set aside more capital in upturns to make them more resilient. Once the downturn sets in, banks can draw on the buffer, which helps them absorb losses. Furthermore, having to set aside more capital in an upturn makes it more costly for banks to grant loans. This might, as a positive side effect, dampen excessive credit growth and hence the cycle. Likewise, reducing capital requirements in a downturn may help to spur lending.

Other tools focus on the general resilience of the financial system, going beyond the financial cycle. Various capital buffers have been designed for that purpose.

There is, for instance, the systemic risk buffer, which can be applied to all banks or to specific groups of banks. It aims to make banks more resilient to structural risks that affect the entire system.

And then there are buffers for globally systemic banks, the G-SII buffers, and for locally systemic banks, the O-SII buffers. By making systemic banks more resilient, these buffers address potential sources of contagion. At the same time, the buffers put a price on the impact systemic banks would have on the financial system and the economy should they fail. This helps to internalise this impact.

Micro or macro? Separating the tools

Ladies and gentlemen, at the beginning of my speech, I said that macroprudential policy is a relatively new policy area. However, the relevant tools look and feel familiar to banking supervisors. They mostly apply to banks, and often focus on capital. Thus, the distinction between micro and macro is less clear than it should be. This ambiguity might cause conflicts of interest, and this leads us to the question of how to deal with them.

The answer is simple: we have to clearly separate the tools of macro- and microprudential policy. To channel the spirit of the famous Dutch economist Jan Tinbergen: each risk needs its own tool.

Let's start with what is known as Pillar 2 in the context of the Basel framework. Under Pillar 2, supervisors can apply a range of measures to individual banks. The goal is microprudential: address bank-specific risks and bank-specific resilience. Now, under current European rules, Pillar 2 measures could also be used to achieve macroprudential goals.

And this could lead to a conflict of interest. Imagine that there is a downturn in the financial cycle. From the viewpoint of each bank, credit risks increase and microprudential supervisors may want to increase Pillar 2 capital demands. Looking at the same trend, macroprudential supervisors might want to support credit growth and counter the cycle over a longer time horizon and from a systemic point of view. Thus, they may want to decrease Pillar 2 capital demands.

Who should prevail? And how should Pillar 2 be calculated when two different approaches are used by two different stakeholders? To avoid such tricky questions, Pillar 2 should be defined as a micro tool only - the current review of the Capital Requirements Directive offers the chance to do so. This would help to focus on the initial idea of Pillar 2: determining a bank's individual level of capital based on the bank's specific risks.

But looking at the macroprudential framework, is it enough just to define Pillar 2 as a purely microprudential tool, or is the macroprudential toolbox lacking in efficient tools?

And indeed, the toolbox defined by the EU legal framework may be too small to address all types of systemic risk. The available tools address only those who grant loans - the banks. They neglect those who take out loans - households, for instance.

Take, for example, the real estate sector, which often plays a big role in the ups and downs of the financial cycle. When house prices are on the rise, the potential gains from buying a house might be high enough to dwarf any increase in the cost of credit triggered by the countercyclical buffer. Here, borrower-based tools would dampen the boom more effectively. I am thinking, for instance, of loan-to-value ratios, LTVs. In some countries in Europe LTVs and other borrower-based measures are already part of the macroprudential toolbox. Harmonising these tools at European level might be a good next step.

At the same time, we should acknowledge that there is a drawback to tools which only focus on banks. The crisis has shown that the financial sector reaches far beyond banks. And when we only tighten rules for banks, certain activities might be shifted to other, less strictly regulated parts of the financial system, namely the non-bank sector. Consequently, risks build up outside the scope of banking regulation and supervision. Eventually, these risks might spill over into the banking sector. So banks might be protected from the first wave of a crisis, but would be hit by the second wave.

Ignoring this part of the financial system would leave the door wide open to risks and crises. Thus, there is a need for tools that address non-banks and their business.

One of these tools concerns margins and haircuts for securities financing transactions and derivatives. As they stand, these margins and haircuts are pro-cyclical: they go down in a boom, and rise in a bust. Thus, there might be a case for having tools that would make margins and haircuts vary in a more countercyclical manner.

To sum up, we need to clearly separate micro- and macro tools, and we need to expand the macro toolbox.

But there is more: we also need to clearly separate macro tools from each other. This is particularly relevant in respect of various structural buffers: the O-SII, the G-SII and the systemic risk buffers. Each of these buffers is meant to address a specific structural risk. In practice, however, they are often used to achieve the same goals. In this context, the fact that currently all three buffers are treated in combination might play a role: only the highest of them applies.

Given that the buffers for systemically important institutions and the systemic risk buffer are supposed to address different risks, they should rather be set up separately and then added up.

The institutional backdrop to macroprudential policy

Ladies and gentlemen, so far, I have discussed the tools of macroprudential policy. What I have not discussed are those who apply the tools and how they do so. And this brings me to the institutional backdrop to macroprudential policy. As you can imagine, in Europe this is a complex issue.

Macroprudential powers were assigned to the ECB when European banking supervision came into being. This sounds straightforward, but it is not.

First of all, the ECB does not have exclusive powers. It shares them with the national authorities. And this makes sense. After all, there is a national angle to systemic risk: financial cycles and structures vary across countries. So sharing powers makes sense but, at the same time, it makes things a bit trickier.

How does this work? Well, in fact, it is the national authorities that take macroprudential measures. The ECB then has the power to tighten - or "top up" - those measures that are covered by EU law.

This approach seeks to achieve two goals: first, it helps to counter any inaction bias on the part of national authorities. And second, it helps to take into account any cross-border effects.

And it's not just the ECB and the national authorities that are involved. The European Systemic Risk Board, or ESRB for short, and, in certain cases, the European Banking Authority also have a role in assessing the adequacy and implementation of certain macroprudential measures. All in all, a number of different EU institutions may be involved in macroprudential policy, depending on the measure used.

The process of activating macroprudential tools is a long and complex one, and it does not promote the proactive and timely use of those tools. In addition, each tool comes with a different activation mechanism.

So I am convinced that these activation procedures need to be simplified. And the notification and information procedures need to be harmonised and streamlined too. This could be done, for instance, by establishing the ESRB as the central information hub for all notifications regarding macroprudential measures in the EU. This would reduce the overall notification burden on national authorities.

Furthermore, one might think that the authorities were free to choose the tool that they deem best suited to address the risk at hand. Well, not quite.

With regard to certain macroprudential tools, a rigid sequence has to be followed. You can't implement certain tools without first having considered using other tools. More specifically, if you want to apply tools such as macroprudential liquidity requirements or risk weights to target asset bubbles, you first have to consider tools such as Pillar 2 requirements or various capital buffers. This sequence is known as the "pecking order" of tools.

But an effective and efficient macroprudential framework should permit the flexible use of all available tools laid down in EU law to prevent systemic risk. Instruments should be chosen on the basis of their effectiveness, not as a result of mandatory sequencing.

The conclusion therefore is that the pecking order should be removed and the activation mechanisms should be harmonised and streamlined.

Now, as I said before, macroprudential policy is a relatively new field in terms of both theory and practice. So, two things will happen over time: the theory will advance, and we will gain practical experience.

Against this backdrop, the European macroprudential framework should be reviewed on a regular basis. This would help to make it more effective and more efficient.

The current situation

But the framework is just a means to an end. And that end is to ensure a stable financial system. So, where do we stand on this front? Are there any risks to financial stability in the euro area?

We at the ECB currently see four major risks.

The first risk is a sudden reversal of risk premia. Recent events have shown that markets can be subject to corrections even when the macroeconomic environment continues to be supportive. So far, the high volatility in the equity market has not spilled over into other markets, which is certainly good news. That said, some market segments need to be closely monitored.

Residential real estate is one such segment. Prices are broadly in line with the average valuations recorded over recent decades. However, in some large cities, real estate prices have increased more rapidly than household income. In addition, prices for prime commercial properties rose strongly in 2017, and the fundamentals suggest that valuations are a bit stretched here. Euro area corporate bond spreads for some lower-rated issuers also look increasingly tight [compared with fundamentals.

The second risk lies in the low profitability of banks. As only profitable banks can reliably serve the economy, weak banks are indeed a concern.

The third risk is that markets might again doubt the ability of public and private borrowers to service their debt.

And finally, there are liquidity risks in the non-bank sector which might spill over into the entire financial system.

So, there are risks, but, from the ECB's point of view, they are not too pronounced. In addition, the euro area economy is recovering, and this will help to alleviate some financial stability concerns. That said, the situation differs from country to country, of course.

And as this is Dutch Banking Day, let's turn our attention to the Netherlands.

There are two things in this country that the ECB is watching closely. The first is the real estate sector. Since 2013, prices for residential real estate have been rising strongly. This applies mostly to large cities, where we now see tentative signs of overvaluation. Prices increased faster than household income.

But mortgage lending is still growing more slowly than house prices. This seems to be largely due to the fact that many borrowers are repaying existing loans; since 2013, mortgage repayments have totalled around €60 billion.

This has been driven by low interest rates and a gift tax exemption for home purchases. But even when we take this into account, mortgage lending is still growing more slowly than house prices. So other factors are still in play. These include demography and the limited home equity of many households.

But it's not just residential real estate. There is also commercial real estate, and that market is quite large relative to the size of the Dutch economy, mainly because real estate investment funds play a big role in the Netherlands - more so than in other countries.

And prices for prime commercial real estate are rising. On average, they are still below peak levels, but, again, we see tentative signs of overvaluation - particularly in large cities. At the same time, loans have begun to grow faster, and lending standards have started to ease - but they are still tight compared with the pre-crisis period.

Against this backdrop, both residential and commercial real estate markets need to be closely monitored. And the two markets need to be seen in combination. After all, they are closely linked, and risks might spill over from one to the other.

We also keep a close eye on household debt. Over the past few years, households have reduced their debt, but the level is still high. At 106% of GDP, the debt burden of Dutch households is among the highest in the euro area. The same is true of the debt-to-income ratio, which stands at more than 200%. Moreover, as house prices are rising and more loans are being granted, the reduction of debt might come to an end.

Against this backdrop, there might be a need for more active macroprudential policy. And by this I mean a tightening of existing borrower-based measures.

With that in mind, the Dutch Financial Stability Committee and De Nederlandsche Bank have recommended that maximum loan-to-value, or LTV, ratios be further reduced to 90% over the next ten years.

In my view, this is called for. After all, LTV ratios in the Netherlands are higher than in other countries. Combined with the high levels of household debt, this poses a risk to financial stability.

In addition, some have been calling for the tax deductions applicable to interest on mortgage loans to be brought down more quickly. Here too, the Netherlands stands out, with higher levels of tax relief for mortgage loans than in other countries. This kind of tax policy has proven to be pro-cyclical and has made the household sector more vulnerable. It would thus be sensible to speed up the reduction of tax relief.

Conclusion

Ladies and gentlemen,

Macroprudential policy is a relatively new thing. And what is new is almost never perfect; that is also true of macroprudential policy. It still has to mature. We still need to refine the theory and translate it into tools that are fit for purpose, policies that work and an institutional architecture that is both lean and stable. All this requires further work and regular reviews of the status quo.

Let us now return to the fallacy of composition. Each of you is made up of billions of cells. Each of these cells is unconscious. Therefore, you must be unconscious as well. But unless you fell asleep during my speech, you are very much conscious. I admit that this is a bit more philosophical than macroprudential policy, but the basic idea remains the same: the whole is more than the sum of its parts.

Thank you for your attention.