 Welcome from my side to this panel on shadow banking. My name is Stefan Kern. I'm the Chief Economist of the European Securities and Markets Authority in Paris. And I also have the pleasure and privilege to work closely with the ESRB on shadow banking issues. Together with Richard Portas, we chair the shadow banking expert group of the ESRB. And it's great to see so many people from the NCA's, the National Competent Authorities, the central banks that participate in our work on shadow banking so actively and contribute to that work. And without whom, our analytical knowledge here at least in the financial supervisory system would be inconceivable. This panel on shadow banking is ideally placed between the earlier discussions we had today on the banking situation and also then later on on the policy questions in the non-bank sector. And it is also ideally placed because we just had one of the pioneers of shadow banking thought, Tobias Adrien, with us talking to us. So this is a very timely panel for today's sessions. And let's see how we can discuss these very pertinent questions that we have. Just as a little scene setting on shadow banking in his opening speech yesterday, President Draghi already highlighted the role of shadow banking and the interconnections it creates in the financial system, especially here in the European Union. So in terms of the importance of the topic, there is nothing to add at this stage. And this gives us an opportunity here on the panel to go directly and focus immediately on the risks that we have in the shadow banking system and the implications that this has. Now, 10 years after the onset of the financial crisis, we do look back at substantial progress and advances in terms of both the knowledge that we have about shadow banking, but also about a wide range of policy measures that have been taken to contain the risks in the system. And those measures range from in the EU capital requirements on the banks and financial reporting standards, especially when it comes to shadow banking related activities, for example, special purpose vehicles or securitization as well. We have a full regulatory and supervisory framework now for credit rating agencies. We have a full regulatory framework for hedge funds under the AFMD directive. So hedge funds have been brought into the regulatory perimeter as it was called by the G20 at the time. The same applies for money market funds, a regulation that is currently in the process of implementation. In future, we will have a raft of data coming from securities on securities financing transactions, coming from the new regulation that has been passed in Brussels, and that will bring light to a market that is very much in the shadows still today. Now, with all of these measures, and there are many more around this, the EU has, in fact, spearheaded the regulatory risk management, if you want, of shadow banking. Up to the point that if we read the reports by the FSB, for example, we apparently are at a stage, and I'm quoting the FSB in its latest assessment of shadow banking activities. Shadow banking activities are found to have contributed to the financial crisis that these have declined significantly and are generally no longer considered to post-financial stability risks, and that no other new financial stability risks from shadow banking can currently be identified. So this is great news, of course, for all of us, but it would be a big surprise if that was the end of the story, and there are many topics that we need to discuss. The findings of the FSB are reflected in the data. If you look at the FSB Global Shadow Banking Report, but also at the ESRB annual EU shadow banking monitor, both institutions have achieved a great, to add great precisions to their measurement of shadow banking, focusing on the core activities and the core entities involved in shadow banking, and if we look at the European Union as it is today, core shadow banking, if you want to call it that way, has contracted actually by more than 4% between 2012 and 2015 and another 0.6% last year, while other non-bank financial activities, in fact, grew over that period quite substantively, and it also measures favorably with the international experience where narrow shadow banking as measured by the FSB still continues to grow by around 3% per year, in 2015 at least, even though that too is slightly lower than the general non-bank growth, the so-called OFIs. Now, these are important advances, as I said, but there is broad agreement about additional vulnerabilities that we need to be aware of going forward, and I'm just mentioning a few of them. First of all, liquidity risk and leverage in certain fund vehicles, especially those that are exposed to liquidity and maturity transformation. Another one that many people are working on is interconnectedness, especially between the banking sector and the shadow banking world, but also other forms of interconnectedness. Then there is prosyclicality, liquidity and leverage, risk through derivatives and securities financing transactions, and also through the collateral transformation chain. And then finally, there is a wide range of activities that are still in the shadows that are not covered by transparency or the regulatory requirements, and in addition to that, we have now a whole wave of new innovative and emerging technologies and shadow banking style activities that reflect the evolving, if not mutating, nature of the shadow banking system and the risk of regulatory arbitrage. Now, finally, let me say that despite all the progress that we've made, the analytical work on the shadow banking system is still in its infancy, if not at an embryonic state. The G20 process, FSB, the European Union in Brussels, they have given us new data to collect that we've started to collect over the past 10 years, and it is only now that we are in a situation, although slowly but surely in a situation, that we can actually exploit this data. So this is actually the starting point of shadow banking and analytics, and if we look into the next five to 10 years, I think this will be an important era for analyzing and better understanding the shadow banking system better still than we do today. Now, with all of these questions, it is fantastic to have a highly distinguished panel around me that will discuss the issues that we have on the desk. First of all, I am pleased to welcome Juliana Bergenau. She's an assistant professor of finance at Stanford. Her research focus is on the interplay of the real economy with financial marks and financial institutions, and I understand you actually received an award, the Western Finance Association Award for a paper in this context last year. So welcome, Juliana, for joining us on this panel. Second, we have Stephen Ondner, who's joining us from University of Zurich and the Swiss Finance Institute, where he's a professor. He's a research fellow in financial economics of the Center for Economic Policy Research on top of that. Stephen, thank you for joining us. And last but not least, Stain Classens, their financial stability policy at the Bank for International Settlements, where he represents the base in important senior decision-making groups, including at the FSB, the BCBS and the G20. Within the base, you lead the policy-based analysis of financial sector issues, and you oversee the work of the committee on the global financial system, other committee secretariates. Thank you, Stain, for being with us. It's a great pleasure to have you. And as in previous sessions, we have 15 minutes each for introductory presentations, afterwards 20 minutes for initial discussions here on the panel, and then Q&A's with the audience. With that, Juliana, please. Let me start by thanking Richard for inviting me and having me on this panel. So this session is about identifying and assessing the risk in the shadow banking system. And then first, what you wanna do is to define what shadow banks are. And I'd argue that this definition is not really yet sorted out and that matters for all sorts of things. So what I, in a very broad start, I define shadow banks as those that shadow the asset and liability side, and those can be different institutions that do each of traditional banks. So that would be either lending or some form of deposit creations, for example, in money market mutual funds. So to give you a flavor of why I argue that the definition of shadow banks is not really set, let me give you a sense of some of the causes of shadow banks. So some of the causes for why these functions are performed outside of traditional banks. So it can be, for example, technological change. There's a few things that have happened over the last decades that have caused some of the core function of banking. For example, lending to be performed by non-banks. And the reason is that some of these non-banks have cheaper ways of delivering those same banking things. And one of the things that you might fintech companies that have been alluded to in the earlier sessions today do those kind of perform those kind of functions. Another cause of shadow banking activity can arise through regulation that favor the business opportunity, improve the business opportunity for non-banks. So you can also call them shadow banks that would provoke a substitution away from traditional banks. And then finally, there's that lots of regulators, of course, are worried about, it's like a substitution within traditional balance sheet of balance sheet, within traditional banks of balance sheets. It's with the traditional regulatory arbitrage substitution towards off-balance sheet vehicles. So I'd argue that this is an issue we need in functional definition of banks, as well as shadow banks. And as technology is progressing over time, those definitions will also have to be adopted. And to give you a sense of like how you can come, why is this an issue? And this is the first issue is for measurement. Well, I can take the US financial accounts, the flow of funds and can add up all those institutions, sectors that are listed there that do something similar on their asset side as traditional banks that are not depository institutions. The depository institutions are depicted here as the light blue line. This is from 98 to 2014 in trillion. And you see that as we all know, the shadow banking system has grown tremendously over this time. But at the same time, you can, in the shadow banking system here, shadow assets are the dark blue line or violet line or whatever you wanna call it. And you'd see that you can already pick this line apart. Well, why? Because some of these forms of shadow banking, for example, FinTech companies may not be featured in those because their functions are not really, or their business model is not balance sheet based. So the balance sheet will not be reflected in that blue line. Some of them also is activity by traditional banks. The bank holding company can own security broker dealers, can own finance companies, can own SAPX security issuers and all in an umbrella that are part of this dark line. So what it matters for measurement, it also matters for how to develop models and to thinking about shadow banks. So what are the firm characteristics that we should place into these models? What are the frictions? What are the environment that we should focus on? And of course all these gonna matter for policy results as I will show you in a couple of minutes. And it also matters for not thinking about the opportunities that can arise from shadow banking, from different forms of shadow banking that lie in an efficiency gain that can arise through adoption of new policies, new technologies. So I want to add a few remarks on those models and I think about shadow banks and commercial banks. And as an example, I use a work with mine with Tim Landfork where we basically think about what happens if you impose capital requirement on traditional banks, what happens to the entire economy. And in this model it turns out that the presence of shadow banks is rather benign. And I hopefully come get to the intuition and I'll talk a little bit about fintech. So why we started this model? Well, obviously the financial system is composed of both regulated, commercial banks we call them, and unregulated financial institutions. So all of these institutions and that's how we model them provide some access to some intermediated assets or like terms of long-term credit, for example. And they fund that with some form of liquid money-like liabilities. When I talk about liquidity, I typically think about this as in terms of money-like liabilities of banks and non-banks potentially. So the question that we are asking in this paper is quite frankly, what are the effects of regulating a subset of the economy? So a big, large concern is that also Tobias was today alluding to is like, well, then we're gonna ship off, we're gonna increase activity of the shadow banking system. And it's bad because we don't really know what these guys are doing potentially. He didn't say that, but it is the implicit assumption on that, so, and that's totally fair. And then so the question is, does it increase the overall risk of the financial system and therefore we need basically a quantitative model to assess these trade-offs and that's tricky. So the way how we structure that, we modeled basically commercial banks and shadow banks as two separate entities. So it also retells you that here I had to allude to or I had to take a stand on what type of definition I wanna here make of shadow banks. This is not of balance sheet activity of commercial banks. So these are two different entities. We model them to be tractable fairly similar. They provide liquidity services on their liability side. So let's think about deposits and deposit-like products. That households value with the money and utility function specification. And they both can go bankrupt if they take on too much debt and they can't repay it and they go bankrupt. So that's bad for the economy. There's costly bankruptcies. So the difference between them is that, well, commercial banks, their deposits are perfectly safe to depositors because of deposit insurance. They have to pay like some in deposit insurance fee for that. And then they also face capital requirements on those deposit-making activities. Then shadow banks in contrast do not face such restrictions. So the debt that they issue is risky to households because they may not be recovering all of it. We calibrate the model to the U.S. economy and U.S. financial sector. Most notably, we make an effort to model financial, like shadow banks as fairly fragile because in addition, because there's no deposit insurance, which prevents bank runs, these shadow banks may be exposed to bank runs that then have also other ramification in the model. So the basic trade-off is then in this model, well, if you tighten the capital requirement, for sure you're gonna reduce liquidity provision, which is like one of the key elements in the paper, elements, key functions of banks in this paper. But you do not increase the risk of the financial sector even though shadow bank activity goes up. So this trade-off is basically between a reduction in financial fragility and a reduction in liquidity provision. So, and of course, obviously, we actually tried very hard to think about why because that's not what we thought about what we came when we wrote the model. And then you have to note, I mean, the premise here is that the usefulness of liquidity provision in the economy, that's the key thing that the bankings bank do in a useful way. So shadow banks compete with traditional banks in those functions. These two goods are fair, so the deposit of shadow banks and of commercial banks are fairly substitutable. However, up to a certain degree, so particularly in the crisis, the shadow banks' liabilities lose their money-like functions and we model this actually in a neat, indulgent way. The shadow banks debt pricing is, because debt is here risky, is sensitive to the default probability of shadow banks. So there is, to some extent, these shadow banks internalize the risk that they are taking. We also, in the calibration, we impose some risk, we introduce some random bailout of the shadow banking system such that the shadow banks do not fully internalize the costs that they take on if they use more leverage, but in general they take into account the risk that they are alluding to. So when the capital requirement is increased in this economy, because it increased the cost of capital for commercial banks, commercial banks reduce the balance sheet. That's the first step, what happens here in this economy, and this will basically happen in partial equilibrium if prices stay constant. This increases the scarcity of a good that households care about deposits. And because these two deposit types of shadow banks and non-shadow banks are fairly substitutable, the value of all deposits increases in this economy, also that of the shadow banks. So the bond price increases, therefore, and that induces a reduction in the rates so these banks can, all of them can fund themselves as cheaper rates. So it sounds very unintuitive, but it happens because it's in general equilibrium effect. And of course, on empirical data, this is not what we measure, because empirical models or empirical studies that are well-identified are partial equilibrium effects and the margin. So what we are studying here, and that's why these models can be useful, is a general equilibrium effect. So when there's a reduction in the good that households care about, the liquidity provision, there may be a change in the prices that can actually improve, again, the funding conditions for banks, everything else equal. This gives an incentive for shadow banks to fill the gap, right? They face better funding conditions because they can basically substitute for households what they want to do. They can do this either by levering up, increasing more liabilities, levering up, therefore. This would induce, of course, higher leverage and the higher leverage induces more risk-taking and higher probability of default. And would lead eventually to lower bond prices. Or they can just expand on the scale and keep debt and assets in proportion, so keep leverage constant, so they wouldn't change their risk that they're having, but could still satisfy the liquidity need in the economy. What we find, and that's really fairly robust accounts of different specification, is that they, in our model, as long as households care more about the quantity of liquidity provision compared to the composition, the increase in profitability is large enough so that risk-taking incentives is going to be reduced. And if you think about this, they're typically, so banks gonna choose the second part. So as long, as long as households care about the quantity of liquidity provision, the profit conditions are highly favorable to banks, and you can think about it this way, that typically where does excessive risk-taking incentives arise is typically when banks have low net worth and there's a negative shock. If profit conditions are improved, then risk-taking incentives go down, and that's essentially what's happening in this model, and that's the intuition for why we get there. So in the paper, we have several of these tables that I will not go for in time, but it's basically here showing that if the capital requirements in these other theaters are increased from 10 to 20%, there is an increase in the shadow banking share, but no corresponding increase in the total risk that there is the economy faces, which are here measured as basically in realized risk as the deadweight losses from both sectors, even though the share has gone up from shadow banking activity. So our takeaway is, well, that higher capital that will lead to more shadow banking activity do not necessarily need and have to lead to more financial fragility and response. This has been robust to various specifications, conditional on our setting and our premise, how we modeled and how we defined shadow banks in the first place. We made an effort to model shadow banks as fragile, however, we didn't model many other things as Tobias was alluding to, we base our framework on a rational expectation. There's no extrapolative or any other behavioral assumptions in this framework. We do not model shadow banks as off-balance sheet vehicles for regulatory arbitrage. We do not model strategic interactions and you can make this list of this the very, very long. We also didn't consider this would be also on the downside why we don't find like a negative effect, but we also didn't model any positive effects that can arise through an efficiency gain that kind of rise from shadow banks. And on that I wanna conclude basically that what is often not considered is that financial intermediation is very costly, a site of cost of capital. So today we are listening to some speakers saying, oh well, the cost of capital of banks and all in is like 2% of whatever. But in addition to that, there's also non-financial costs. So it's just like expense, non-interest expenses that can be fairly large. And Philippon and others have measured that to be around 2% of assets. So if you think about the size of the US banking sector of like 15 trillion, just like I mean back on the envelope there's like 2% of that. That's a floor cost per year that's fairly, fairly large. So to this point, so FinTech and other and I mean those similar technologies are also now adopted by the banks themselves. Those give rise, give opportunities to lower these costs that the financial sector actually is using up in order to provide these financial services that are important for the economy. And there's several examples where this is more or less successful and I'm sure that there's a lot of ramifications for risk considerations. However, we have to think about that financial intermediation can also learn or can improve an efficiency and those efficiency gains can be induced from more competition also from the shadow banking system. And so to summarize my points, I'm gonna say that we need a common definition of shadow banks to both measure and identify the risk and understand what are actually the activities they're doing, also properly frame the issues so that we can study policy relevant questions. And also to note that not all forms of shadow banking activities are not necessarily using a wider definition are not all that bad. And with that I would like to conclude and thank you for your attention. Thank you very much, Juliana. Hugely interesting outcomes of this. That give us a lot of food for thought but let me immediately hand it over to Stephen for his presentation. Stephen, thank you. So when Professor Port has invited me a couple of, quite a while ago actually to sit on this panel on shadow banking, I was a bit stressed because I hadn't been working on shadow banking that much at all. But I took it as an admonition of saying you should be working on shadow banking. So what I'm going to present today at the end of my talk is some ongoing work on this account. But before doing that, I want to also talk a little bit about ongoing work on spillos, regulatory arbitrage in general because it's clear that one important aspect of the shadow banking activities may well be this type of regulatory arbitrage. Juliana was already talking about that. And so on that account, I'm going to talk also a little bit on ongoing work on this account talking through a few empirical exercise I've engaged in. So without much ado, it's all about spillovers, not the type of spillos that are being depicted here. That comes later, I assume, I hope. But so I'm going to talk about spillovers across sectors, across countries and then into the shadows. So across sectors and here, to some extent, what I'd like to talk about two minutes is work that actually indeed corroborates what Tobias was alluding to earlier that targeted micro-potential policies may as well lead to spillovers into different areas of the lending by banks. And so here this is the micro-potential policy that targeted, they're going to go to specific agents, specific sectors, and the real estate business, residential mortgage lending is a particular one we're going to focus on. Of course, these targeted micro-potential policies may affect other parts of the economy and may be circumvented by creative agents. And so the piece with Raphael Auer in BIS working paper where we look at indeed the compositional effects of the introduction of the CCYB in Switzerland. And so here there's going to be additional capital charges on residential property mortgages, but not so on commercial mortgages or other business loans. And we're going to look at the Greta register of Switzerland to see indeed to try to identify whether or not there are indeed these effects in the sense that banks start shifting some of their lending towards the commercial sector because the data comes from the Greta register covering all commercial credit. And so allowing us given as micro data we can sort of try to identify within Firm or at least within cluster how indeed banks that are more heavily exposed to these additional charges that come as a consequence of these CCYB are going to start shifting their lending. And so what we find indeed is that those banks first of all, they're going to shift more to commercial lending so they expand their commercial lending. We don't have access to detailed data on the residential property mortgages, but we do observe indeed that they are increasing their commercial lending. They're going to lend a relatively high interest rate but maybe more importantly in this context is also they're going to start lending to small commercial real estate and risky firms. And so to that extent even though it's impossible with the current data to actually investigate this further in terms of maybe relabeling that they may be doing in terms of making some of these residential mortgages partly into commercial ones or whether or not also some of the property developers that seem to be obtaining more loans from these banks start fulfilling an intermediary role at least on the basis of the extent data what seems to be present there is that there is a shift in terms of from residential to commercial also commercial real estate and potentially to the extent that we can adequately measure this more risky lending. So clearly in terms of other types of spillovers of these specific real estate focused charging provisioning in paper on Spain we did show that indeed this increase in provisions to construction real estate firms led to also some volume effect in terms of lending to non-construction real estate related firms. So really a compositional effect and in a paper by Jose Luis and Gabrielle and the quarter and they also show that there's also compositional effect as a consequence of capital new capital requirements. All of this suggesting that there's this type of spillovers going on of course also paper by Ireland all showing that there are these spillovers taking place. Now this is within the bank if you want. Clearly there's also for global banks lots of opportunities for arbitrage regulatory arbitrage and again on this account identification there is more difficult because typically it's hard for the time being to get full-fledged access to microcredit that covers many countries at once which would be truly needed to identify or at least take steps towards identifying spillovers if not regulatory arbitrage where of course the hurdle is even much harder because you need to show sort of one for one whether or not this type of activities take place but at least it seems to suggest that from some of the extent evidence regulations where regulations are laxer banks are going to start lending more and more riskily and so that extent country specific regulatory constraints may also lead to some risk taking. Now on this account may be worthwhile to point out that banks are if you start looking at microdata on one country and observe incoming shocks to which banks in these countries are subject to their very nimble if you want in adjusting their lending. Now let me then sort of conclude by talking a little bit about an ongoing project with Natalia von Westernhagen and Peter Bednarik who's there both sitting here in the audience on the lending to other financial intermediaries. So the idea is there to look at how indeed for a subset of banks that may be specifically hit by the capital charges that are present in the standardized approach how they may because this is very much ongoing work so I have to be very careful how they may engage certain office who are then potentially engage if not the same firm a same set of firms that these banks otherwise would have engaged. The data is here from the German credit register and so the idea would be in the end also to show that the construction that is then being present there one for one at the firm level within firm results in sort of a if you want substitution for the bank lending that is taking place before by the small private bank and then afterwards in the sense as these banks are trying to get away from this binding constraint of these capital charges present in their standardized approach compared to the other banks that potentially are alleviating some of these constraints they are then substituting this type of lending with lending through office. As I said this is very much work in progress so now what is the data that we have currently lined up so we have the Bacchus potential database and then the merge with credit register. Now notice that in this credit register there's also present regulated and not regulated other financial institutions both as borrowers and creditors. Okay so to this extent what we're going to try to investigate here in the process of investigating is that indeed these small banks that are going to be subject to these capital adequacy regulation are going to try to engage certain both regulated and unregulated financial other financial institutions so they are highlighted their financial services institutions but there's actually also five categories within those which are going to be subject to different types of reporting standards then there's the not regulated there also present in the database in terms of their credit activity financial holding companies financial companies and financial asset management companies. Okay now as I already said there's much more to be done so let me just report a few snippets of what we've sort of so far been lining up in terms of data and in terms of if you want to universe statistics so all German banks are going to lend to this set of office I just introduced. Now big banks, landers bank, corporate center banks lend to many office but the small private banks the savings and cooperative banks they're going to lend to very few office actually in many cases only one and we think that is interesting because this potentially could be part of sort of a configuration that is present there. Now over time this exposure has increased dramatically so small private banks sharply increase their exposure to this office after 2008 as you can see there which is now amounting to 30% of their core capital. Okay so now here it's also the case that the lending of this office at least what we are currently having being reported increase quite substantially. Now there is an issue here about reporting so we want to be very careful with this slide having said that the fact that they actually appearing given certain requirements of reporting is interesting that they suddenly are there. So the office are now at least more actively or even to a larger extent potentially with all the caveats involved acting both as lender and borrowers. Now office are distinctively less risky than the real sector firms that are engaged but they are going to finance real sector firms more risky real sector firms than this German bank so this is sort of present in all these small numbers there but especially when the office are going to be financed by German banks themselves. So all in all this is interesting for sure tells us that we need to dig more into this. Anyway is there something worth investigating further potentially so after 2008 there's a very sharp increase in how these private banks engage the singular office. These office are going to lend to risky firms especially when obtaining credit from bank. Again this need to be teased out further empirically but this could be one motive could be to say from regulatory capital and so we plan to investigate what happens after regulatory to this regulatory change try to identify it better also in time how the share of the corporate credit wallet how this substitution took place potentially between this private bank and the if you want the favorite OFI that is being lent to by this bank of course there may be others that are engaging the firm and that makes it even more interesting in terms of empirical exercise and identification. Okay very tentative conclusion so clearly there's a lot of data coming here so there's more need for empirical studies spillover effects using low level data I mean more research is warranted also to keep empirical researchers like myself off the street. Thank you so much Stephen. So we've got from the academic community one impulse which is yes there is an arbitrage problem between banks and non-banks but it might not in all cases actually increase financial stability risks if I'm not mistaken then Stephen proposed that in the especially through the research that you're undertaking at the moment it might actually be that a lot of this arbitrage happens but concentrated in a small number of entities and that may be a very risk prone, very important findings and now over to the dimension and perspective of the international regulatory and analytical body the Bank for International Statements. Stein, thank you. Thank you very much and thanks again Richard because you're also the one that invited me. So this is my perspective not necessarily always as a policymaker because part of it of my involvement with shadow banking predates my involvement as a policymaker so to speak. So I'm gonna give a little bit of a different slant on how to define shadow banking and then I'm gonna talk about what to do about shadow banking possibly in terms of what we have done already to date and talk a little bit about what we maybe can do going forward and end up in terms of monitoring which already as Stefan said is an area where we have to do more. So both Juliana and Stephen have given me some good lead ins as to what it is that we need to think about harder. I being not a theory person, I can only do simple theory so I have this slide as to what the theory of financial intermediation was before we kind of got shadow banking onto the picture and this was my old way of thinking of it. We have savers, we have investors, we have these banks and there's a box around the banks because there's a balance sheet or something tangible in terms of a balance sheet and they intermediate based on soft information that I get in monitoring. There's a huge amount of literature people here have contributed to that in various ways and then we have this more of morphers market based intermediation but it's largely based on hard information verifiable direct financing. So that was my simplistic theory that we could think of textbook kind of way of thinking and then we had this more morphers new animal being introduced shadow banking. The term I'm taking literally a little bit more than maybe other people do. I'm really talking about shadowy form of banking sometimes. Nevertheless, it was based on hard information and the sense that was verifiable meant to be verifiable so the securitization of whatever your CDOs in the US and what have you. The same time was being intermediated because it was not done through the cap of the market. You wasn't like standing up into the market saying, oh, who wants to buy my CDO? There was some involvement of some balance sheet, some intermediary that involved. So this is partly I think what a theory that both Juliana and what Steven on the empirical side is trying to address and we're still not all the way there and that's why I think we have confusion in terms of how to address shadow banking and it's been hard. So we started and this was the FSB giving us a definition. This is the first one credit intermediation involving entities and activities outside the regular banking system. That was a very broad definition. I think it's one of these committees. All of you have participated probably in those. You come up with something that has to work for everybody and you end up with something that nobody understands anymore except for two people in the room. So that was one problem maybe that was underlying that definition. I wasn't there myself. Second, we can go a little bit more. Okay, let's do it functional and Juliana is pushing very hard on that. Let's talk about activity base. What is it exactly that we mean? What kind of a service do we want to describe with shadow banking? And then we can say it's a collection of specific services. That's fine. Tobias has given a speech last week in Helsinki. He's a frequent European visitor and he told us banking and shadow banking and market-based finance is kind of continuous but there are some discreet things that you can maybe draw a line here and there between these various forms. Not to disregard anybody but I think the drawbacks to all of these definitions I think that people probably recognize that we already talked about the first. It's hard to think of all credit that's intermediated outside the regular banking system as being necessarily shadow banking. I mean finance companies, what's wrong with those? I mean the intermediating credit, that's fine. Secondly, there's a lot of intermediation of shadowy activities inside the banking system. So the definition doesn't necessarily work perfectly well. The functional one, theoretically appealing but it's then hard to say shadow banking around the world. Because we then talk about something completely different wealth management products in China are very different than the CDO stuff that we saw in the US as very different than what we see in Indian terms of non-bank financing structures or what have you. So in order to bring it all together then we're gonna have to say everything is eclectic shadow banking. That doesn't work necessarily either. And lastly, the definition in terms of kind of the continuum but then we're back to more or less the same as number two. It can shift over time across countries so not to disregard any of these definitions. I think they clarify but I think maybe I wanna go a little bit more towards the systemic risk part because in the end why we wore it and as Tobias has given us in the speech we should be worried about financial system from a systemic perspective. We can't be regulating each and everything of course. And there I think it's where the shadow banking kind of hits us in the face because there are some forms of shadow banking that can lead us to systemic risk and those are the ones that we wanna focus on. Now how do we then define those that are of systemic concern? The definition I'm giving here and I've been using that for a while and is all financial activities accept traditional banking which require private or public backstop to operate. So something that is similar to banking in the sense that there's credit risk, there's maturity risk, there's liquidity risk probably involved here. However it uses capital markets tools like I said it's in that middle box somewhere. Yet it also is not exactly capital markets because it does require a backstop. As I said in the middle there is somewhere a balance sheet, somewhere something involved that makes sure that the final investor who is of course there's always a demand and supply to shadow banking that the final vest is willing to hold that risk because the tail risk is being covered by some backstop somewhere. And that needs to be there because otherwise you couldn't do your CDOs, you couldn't do your securitization, you couldn't do your well management products, you couldn't do your non-bank financing in India or wherever it is. And a shadow bank has to show that a backstop is there to minimize that risk, to make the investor willing to hold that exposure otherwise he wouldn't get involved into it. Now what's the problem in that sense? Well that shadow banking that looks for that backstop in many different ways. So it's not easy and always practical to identify it and that's why of course why we call it shadow banking otherwise on the balance sheet of the bank we knew it and we can deal with it, we have our regulations, we have our supervision presumably although not always successful we can deal with those kind of risks to some degree. Now the shadow banking is particularly problematic because it has a very low margin for operating. It's not necessarily that it's operating at very high margins so without its own balance sheet without its own profit generation it cannot really provide the backstop so it's not a credible way of doing that. So what it does and it does however at the same time want a large scale of operations to be economically feasible it looks for that backstop somewhere else. It can do that in the private side so back to the banks there's always a capital there that it can look for. Existing institutions can also be insurance companies for that matter. Or it looks for the public backstop that can be explicit or it can be implicit. Many forms of guarantees were used in the bad times of shadow banking but there still are there in many ways when we see this shadow banking forms operating and I gave you some examples as well. But if you think more generally of safe asset provision which is an important component as Juliana mentioned of shadow banking that typically looks for something of a backstop that can be an exemption from the bankruptcy code in the derivatives forms that can be in the form of some stay exemption there. It doesn't have to be explicit it can be implicit forms. What does it mean for policy? Well for policy means that we can zoom in a little bit more concretely I think when shadow banking becomes a systemic issue. So we need this franchise value the shadow banking activity will look for that. That also means it's not easy to regulate because it is a little bit where the market is not able to provide the discipline markets are not good at disciplining tail risk obviously. And that's exactly where the backstop becomes relevant and that the market will not discipline the shadow bank for that because it's tail it's outside of their typical way of thinking. At the same time it is within regulatory reach because after all the backstop is typically either from a regulated institution or from the public sector. So we could do dramatic things like exempt stays or get rid of stays for example if you wanted to reduce certain types of shadow banking activity. It also means that migrations may be less of a concern because you can't get things can move away all the way to hedge funds but then I'm not worried about it and I wouldn't call the shadow bank anymore because there's no systemic risk necessarily because I don't have the backstop there in the first place and hedge funds were not a concern in the last crisis in terms of systemic risk. Policy issues, what has been done the list of first five points is basically FSB agenda for the last few years. I think 0.1 much progress is kind of cutting the ties between the banks and the shadow banks in terms of regulations, step in risk and the like minimizing the direct ties to indirect ties between banks and shadow banking activities. We're not all the way there as the monitor for the EU for example shows lots of interconnection still exist so there are potential concerns here. Money market funds, US of course major reforms maybe not the first best but nevertheless something that is reducing susceptibility to runs, transparency, securitization maybe even more than we needed but in a sense, securitization has been maybe a little bit less than we would like to have been but nevertheless we have done quite a bit. Dampening for security locality, very difficult no clear approaches that I can think of that we have really solved the issue with. Yes, the interact ties we can reduce by making security financing more costly so that sense of reducing proceed locality but the other approaches I think we still have a hard time making operational and then the framework for monitoring I think that's still an ongoing task so I'm gonna give you quickly some data as to where we are so the data is still imperfect I will tell you that but importantly the governance of this sector is by nature still imperfect because we don't have a single entity like a bank supervisor looking at this risk at the overall level the ESAB, et cetera, you've all made great progress but we do have to acknowledge that this is an area where there's multiple regulatory agencies cross-border components to it all of that makes it very hard to think of a single approach to this. The two things on my list here that are unusual I'm not necessarily pushing them but you could think of a demand side approach driving the bad forms of shadow banking out in some sense by providing public safe assets that has been proposed think of T-bills if you have many T-bills you're not necessarily gonna get the securitization whether we want that is a question but it is an approach and I think the last point is a more part and Tobias also alluded to it financial cycle how does it interplay with these risks how do we get these leverage cycles to spill over into the non-bank parts I think we know less about that so I'll finish quickly on the monitoring this is really just giving you a snapshot of what the FSB came out with in terms of monster reports this year how they've been trying to go down from the very big picture of the total and the numbers are out on the web so I'm not gonna give you them one by one but it's going down from a 320 trillion total financial assets down to a shadow banking of about 34 trillion which is obviously still a very large number and then on the right hand side trying to slice that by what they call economic functions by different types of shadow banking so they're making progress in terms of what I would call back to the functional approach the next slide gives it a little bit better as to what the kind of breakdowns they have used these are the five economic functions the first one is the most important is kind of in investment, collective investment vehicles so I think this progress being made on the data and the way they're going about it is really trying, I would say, triangulating entities with activities and then trying to zoom in on risks so the chart on the left gives you back these five economic functions of which the first one was the collective investment vehicles and you see that that has been growing where the other components have been shrinking and that's where we consider them to be the most risk involved so in that sense, in a relative sense the shadow banking has been less risky and then on the right hand side this is a triangulation between the entity and activity approach as to where are the risks, risky components slotted so to speak, is it back in the investment funds or is it in the finance companies or in the insurance companies, broker dealers so I think we're making progress here but I think it's still early days because this does not necessarily give you connectedness so this activity measure is not giving you the links back to the banks or even among all the entities in the system it doesn't give you a sense of what's prosyclical or prosyclicality here it doesn't, of course, allude to what Stefan was saying what's in the shadows it doesn't at all address what's in the shadows because it only covers what's out there and it doesn't cover things like fintech that are only gonna be the unknowns that are coming to the forefront at some point in time but nevertheless it is an attempt and if you try to them see what the regulation has done to reduce the risks the big item here is like I said the increase in collective investment vehicles so the funds so to speak they have increased a lot as in absolute terms but also as a share of total shadow banking so and as you can arguably say money market reforms and other issues have done that then if you go by the other forms the specific issues, regulations that you can attribute, try to attribute the decline in the more risky components but I think it's still early days because it's not necessarily addressing all the risks that we know that exist in shadow banking so to conclude I think we still need to define shadow banking better with a good systemic perspective it's partly about shadow banking all the examples Steven has given us there's a lot of regulatory arbitrage I think that is a component but there's more than that there's also genuine demand for shadow banking we need to consider those systemic risks there are policy issues I gave you five and two additional ones that's probably not a complete set but I think we definitely need better data we need to do that from the bottom up more micro based and then we need to analyze it obviously as to what is causing the growth in shadow banking regardless I think we're always gonna need this somewhat of a backstop as a smell test I think as regulators, as overseeing of financial systems we always have to be willing to step in and say at some point in time we don't like this and we do this for systemic reasons that's tricky because Mark has one certainty but I think there's no way around at some points you have to say that thank you thank you so much Stain for this very systematic overview of the state of the discussion and let me just follow up on your presentation and the systemic risk based look at the shadow banking system where you make the backstops a central element in sort of detecting whether an entity would be falling in that group or no and you're saying that most of these backstops in reality would be implicit step in risks I think is one of these things that would probably count in this regard but the problem with that wouldn't it would be that you would only realize that there is a backstop after the fact, after something has happened so on an ex-anti-basis measuring or trying to identify implicit backstop guarantees would be very difficult to do after all wouldn't it? I think this is always a regulatory game that we play and it's always the regulator being a little bit behind so yes we need them but the regulator needs to take a dynamic perspective as to where the thing the market is going so if you go back to the developer shadow banking in the US particular context with SPVs that were getting liquidity and other support from banks that started out as a very initial credit line provision of liquidity support that we know from banking of over centuries, right? So and then suddenly start to grow into something more that should have raised alarm bells I think ex-anta to say that this is only exposed something that we discovered I think is a little naive so let me leave to that so I think that that is where this smell test is going back to somewhat of the governance issues that I mentioned before we have gonna have this struggle and this was true in the morning as well when we talked about macro potential issues and the like from time to time we have to go into the market and say listen we don't like necessarily what you're doing is because we don't trust that you have covered all the risk on a systemic level and it's not just only entity base, right? Because it's the collective misinterpretation of the market of certain risks. Maybe also bringing that to Juliana and Steven, this definitional question around shadow banking and the perimeter question how relevant is that for your own work and for the analytical work is that a core issue? And I think you made an argument for also the activities based sort of perspective on the issue is that immediate concern to your immediate work or is that something where you would say the most important thing is that we dedicate our attention to individual risks that we see or that we smell in one way or another? I mean personally operationally of course this going to be depending on sort of if you want a prevailing view on where exactly these shadow banks activities are located. I think from a sort of pure empirical point of view one can sort of be agnostic and just see the type of activities that take place and so in that respect, I think it doesn't necessarily preclude further empirical evidence or something so. Yeah I mean if you model things you have to know what you model so you have to make a case for a definition. I think what is key problem though with any of those definitions I mean the question is whether shadow banks I mean ideally we want to have something that is like systemically important and what would that mean in many models is sometimes like risks that are not well understood. And so you have to think about like what are the functions where there are risks that might not end up to be well understood. So that's tricky. Yeah and was that a second question related to the definitional issue? How oftentimes we perceive the definitional issue as one of incoming new risks that we look at and therefore enhancing the perimeter of the definition time and again. So that temptation is there and it's important. Do we also have already an idea and given the advances at the policy level how we deal with taking certain entities or activities out of our definition once we have the feeling and feel comfortable that the risks have been addressed think of money market funds in the European Union now that they are under a full regulatory framework. Does this mean that we if we're rigorous about it take this out of our definition of shadow banking? I think the term shadow banking was always an unfortunate term so there's some elements of it that maybe shouldn't have entered in the first place. But then we should also be willing to take it out. I think yes if we have confidence that these regulations are fully sufficient. At the same time and again it goes back to the point risk will shift so it's an endogenous animal that you're facing right. So to say that risk is always gonna be reduced in one element of the financial sector. I'm low to go for that conclusion. So that definitional issue will be necessarily a dynamic thing that we'll need to catch up with and correct. But it is important you pointed that out in order to focus and help us focus on the risk analytical side and exactly follow up on what we're identifying and maybe that risk side is the most prevalent one also on this panel. Let me go right at it. We've got in the especially amongst the international organizations an activity based sort of approach. Where we look at things like liquidity and leverage and the transformation of credit risks and also the interconnectedness. Are we getting that right? Are these the right risk issues to look at or are we missing out on something at the moment? Technically speaking or also from the perspective of profound analysis, Julianne. I would refer to the empirical crowd here. Yeah, I mean, I guess it's early days. Let me write. I mean, in terms, I guess having micro data on this account will greatly help. Also, I think they're also actually seeing that there is so little. I think, of course, a lot is coming. I know, I guess will help us a great deal in identifying some of the risks that are being taken over time. So I think in that aspect, yeah. If you want a wish list of what will be nice to have that's not going to be easy. So interconnectedness already mentioned, we need much better data there that has been done at the national level to some degree, and then including at the ESOB level, but globally for sure we don't have that data yet. So we need to do more drilling down individual institutions sometimes, but at least to the sectoral level. I think on the cross-border side, we have very limited data yet and we know that's very important in many respects. So we would like to see more there. Inconsistency of data is getting better, but still is not there. I mean, Richard Portas last week pointed out the discrepancy between the micro and the aga data. There's a huge gaps there. If you look at individual data, we can do better flows. The funds data is nice, but that should match up to some degree with the micro data that get commercial sources, right? And that gap is in some circumstances quite large. I think we can do better in terms of thinking in a structured way of how we monitor innovation. So the smell test is, I was using it very loosely, but of course we would like to give the regulatory community a little bit more of a template as to how to go out and monitor this. There are lots of networks of people that talk about how innovation comes about and how we can monitor it, but the more we can give that support and academics can help us in that respect of how we document that. Clearly, market intelligency has become a bigger area of central banks and regulatory agencies, but it's not yet a science and it's still a little bit more of an art. So I think they're going beyond case studies and it would be very, very useful. And then, yeah, drilling down, where is the risk, where is the externality, right? So why are we concerned about this because sometimes systemic risk builds up to some of the mechanisms that were mentioned earlier, but we may be able to document it if we had very granular micro data. So it is indeed the behavior of one individual contingent on the behavior of the other individual through which I see the complementarity that gives me the concern that I should have. That data isn't always there, but nevertheless, sometimes we have it. Security, financing data, more available, trade or process data is more available. It's been collected. It's a huge data exercise and I think we've only started to really get into it to address some of these issues in terms of shadow banking and risk that are out there. I know these have been other ones that are starting a little bit with the academics. Maybe you have to give them a little bit more access. I'm sure, maybe not Juliana, she's more a theory person, but I'm sure Stephen would happily help you out in terms of some other research. We'll probably take data also. So I think one of the key things that we have to think about is like, where does the risk capacity of the residual claimant of whatever entity there is? And if that entity is a person or whatever collection, this residual claimant also aware of the risk and is able to bear it. So I think even leverage may not be a problem if that, like hedge funds are oftentimes very highly leveraged, but if that's then, if there's just a billionaire that is wiped out, it's not an issue potentially. Question if it's that the billionaire then triggers another chain reaction. So that's then the complementarities that can be played. Yeah. Which would speak directly to, again, to that systemic risk sort of identification of shadow banking activity. And I wanna come back to that specific point because it shows the interlinketers through the system. You're saying like, if we're looking at just one asset management entity, maybe a vehicle in it by itself, the question is who does it affect? Who are, which exposure exists in particular if it is leveraged, for example, synthetic or not? The question is how can we best actually monitor these horizontal interconnectedness in the system? We are already struggling, I understand, with identifying individual entities or very limited activities and find out how exactly the risks function and what the channels are, but how do we deal with much more complex layers of interconnectedness? Well, I think, I mean, that's at the moment as a huge problem of data collection and as well as probably processing power, but certain FinTech companies already, I mean, they do this on a very small, a smaller case, like there's this company called Interactive Brokers, they do real-time monitoring of your portfolio and they allow investors to borrow against that and so they allow you to, yeah, basically borrow against margin and they liquidate you immediately once this margin position and the value of your portfolio in real-time measured is below what you've basically had borrowed. So that, I mean, in the long run, I mean, I'm guessing we're far away from that, but it's not implausible to think why banks, for example, or even other shadow banks may not even want to have that something like that themselves, right? So because that can prevent them from then raising equity in times when it's particularly expensive as to be a set today. So they are monitoring your own risk and help by technology can also be done across entities, although there's still like a little idealistic. Stay. Well, I agree completely with you, I mean, I think the big data development that I've seen, I mean, to the extent that Facebook can tell us what we're doing on a daily basis, we should be able to do that and what kind of connections we have around the world. We should be able to do it in financial markets as well. We have a lot of data there. So it's putting the machine to work, so to speak with. That requires, of course, legal and other requirements to be in place. So it's not an easy task, but nevertheless, it's on the technology side, not an undoable task, I would say. But there's another component of it which we didn't talk as much about, of course. To some extent, we're gonna break these horizontal links by having central clearing and CCPs in place. So the goal is to, of course, reduce through that netting and otherwise the overall risk in the system. We're not all the way there yet in terms of both, I think, analytical approaches, how much gains can be expected from such an approach. But also, of course, having it all the way down that there is much clearing that way. So some of it is gonna do through the policy side as well, I think. Stephen. If I may add to that, so clearly micro data is no panacea per se, because obviously one can use it to a better extent to identify, but then, of course, the big challenge is also how to aggregate this back up in a way which is then consistent with the identified elements in the empirical exercise and the thing on that, on that front, clearly some structural modeling in combination with the identification scheme that one has will be needed. And that is, I think, sort of the frontier if you want for empirical work right now. To what extent is the heterogeneity of the landscape of activities and also entities involved in that respect still an obstacle to what we're doing, despite the new technologies that may or may not in future be available to do that. But apparently, you already pointed at that. First of all, the landscape we're looking at is much less homogeneous than the banking system. We know a lot about the banking system. There has been now a tradition of more, several decades, probably more than 80 years of profound knowledge about what banks do and what they don't do. In shallow banking, we're at a starting point. Many of these activities' data don't exist. We're only getting there. But it is part of the analytical challenge that the field and the type of activities are just so different, right? How can we best tackle that from an analytical perspective? Because at the end of the day, the risks are in individual institutions and in the activities between them. What is the best way of getting there? Is there innovation also an answer in looking at new analytical tools or is there a new thinking needed for that? Analytical tools will surely be helpful. New thinking will surely, but I think it's for the moment, at least still the practical agenda is number one. So we need to make progress on issues like LEI, the Legal Entity Identifier. We have something called UTI and UPI. Those still have to be there. They have to be harmonized so that we can know what are the transactions, what are the products, what are the entities and how we can combine it. So there's a, I would say, more of a practical challenge for the moment to overcome. And then the heterogeneity, yes, it's there. At the same time, we probably know that if we have 100 banks around the world that we can observe and monitor well, we do a lot better than, we already do a lot better. So we don't have to go to cover each and everybody around the world. So keep in mind that as well. I'd say, I mean, they interact in certain markets and that's the way of connecting again. I mean, I think that's more a question of legal challenges and of if you're able to monitor them rather than like technical challenges in the long run. And it doesn't matter then if you have more and different institutions once, I mean, the data cranking machine is set up and once the legal challenges are out of the way, which is a big if, if, so. That's reassuring. And maybe the last one before I open it up to the audience. Stan, you mentioned explicitly the heterogeneity also across borders that in the European shadow banking system in as far as it is definable, looks different from that in the United States and those two again look very different from what you see in Asia, in particular in China, for example. What is it that characterizes in particular the European side? Is it to what extent, for example, does the dominance of the banking system that has been subject of many discussions, especially today already, to what extent does the dominance of the banking system condition the way we Europeans look at the shadow banking system? Is it, you know, is it a shadow banking system in its own right or is it in Europe actually more a banking system's shadow, so to say, that we're looking at? Is the European shadow banking system something very unique or very specific in that international comparison? It is shadow banking, so it's always gonna be linked to the banks, at least that's how we coined the term. I think, but in Europe, I think it's even more likely to be linked to the banks because the banks are such a bigger component of the overall financial system. So yes, it is different than in the US where it's more capital markets that were seeking out the banks and here it's coming out of the banks going into the non-banking shadow banking. So that back to the interconnectedness issue that was highlighted in the USRB report, I think that's a very important component for the Europeans to track then, be sure that there's no spilled bags coming into the banking system. It seems to me that's in some sense a doable task where the data exists and approaches exist, so in that sense, it's a little bit easier than maybe some other forms of shadow banking. And that would be corroborated, Julianne and Stephen, by your findings that the channels are mainly through from the banking system into ancillary activities or activities to which they are arbitraried out, right? That is where your elements would chime in, yeah. And if I may, for China, I mean, obviously it's a very different system, but also there, I mean, I worked on China showing that maybe the formal informal sector there actually cooperate in the sense of potentially up to a certain point delivering financing performance for firms. So it's, of course, a much larger portion of the financing that these firms receive than in Europe, so, and it is a very different sector altogether. Which is a good point then to open it up to the audience. Please give me a sign, everybody who's got questions here.