When shocks collide: AI, war and financial stability – In Conversation with John Fell

war in the Middle East, rising oil and gas prices, upside risk to inflation, and pretty much daily reports or updates on whether peace talks are progressing or not. At a time when markets were already grappling with potentially stretched valuations, uncertainty around artificial intelligence, and growing concerns about private markets, the Euro area is facing another test of its resilience. So, how resilient is it? How resilient is the financial system in particular? What happens if the shock lasts longer than expected? That's what we're looking at in today's episode of Euro Matters, the podcast by the European Central Bank, where we unpack the stories, ideas, and decisions shaping Europe's economy. My name is Paul Gordon and today I'm back in conversation with a regular guest John Fel who is the deputy director general for macro credential policy and financial stability here at the ECB uh who has just published the latest financial stability review and uh good to have you back on the program. >> Thank you. Thanks a lot Paul and great to be back. >> So let's take a look at uh the state of Europe at the moment uh and the state of the world. The last time you joined us, uh, you were saying the Euro financial system was, um, relatively resilient. The risks from tariffs and geopolitical conflicts had been successfully contained as far as we could tell. Developing basis still looking good. >> Good memory. Yeah, that's where we ended last time. Uh, we we said the resilient, the system was resilient. And yeah, I mean, that continues to be >> our core assessment. Um but now we are seeing a situation where this resilience is being tested by not one but two potentially colliding shocks. So we've got the AI disruption shock. It's a kind of a more slowm moving shock and then we have also um an energy shock uh resulting [clears throat] from tensions in the Middle East. Um [snorts] we also have vulnerabilities and I think this is kind of important. you'll see that we structure our FSO around three main themes. And they're usually around the three main vulnerabilities um that we see. And you know, vulnerabilities um can persist, they can self-correct, or they can be unraveled. And they can be unraveled if there are shocks. And we have shocks. >> And we have two, not one. Uh normally, I think when we've done these podcasts in the past, we were talking about one shock, but now >> we've got two and they're also interacting with one another. AI disruption. Just think about economics of AI. It also depends on the cost of energy. Um [clears throat] to kind of put all of this together into a single number that uh readers might be able to you know get a get an overall sense for for where we are. We have developed with the help of AI a sentiment indicator. Um and this is based on how readers might typically interpret the tone of what we've written in in the financial stability review. And I mean, I have to say that this is one of the most negative readings that we've had in in quite some time in in a number of years. It's not quite where it was before the pandemic, but it's it's marketkedly marketkedly higher. Um, I think our overall message this time, if you were to say it in a sentence, uh, the system is holding up, um, but it's now having it's being confronted with having to absorb a combination of shocks. And these are shocks that it has not yet been tested against. So in a way the tectonic plates uh may be moving. Okay. So kind of under assault if you like um from these we've talked about perma shocks in the past but here we are pretty intense sentiment marketkedly negative as you say. Let's take a drill down a little bit. Let's take a look at some of these on financial markets. I mean there's you mentioned the uh AI what some people call a bubble others of course deny it's a bubble but either way valuations are high by historical standards um and we did see some repricing of software related stocks earlier this year how does the current geopolitical shock fit into this financial markets >> yeah so this is um yeah this makes the whole thing complex this time I mean if you take the AI >> first I I mean I think what we would say is that views on AI appear to have flipped. Um so you know we had this concern >> I think we talked before about these you know these diagrams of you know financing relationships between AI firms and you know the interconnectedness and the big question was can this investment ever be paid back? will those firms generate the revenue that would be needed to you know pay the investors back now it's kind of shifted to um is AI already disrupting existing business models and we've seen you know some of the software companies in particular um being affected by this change in view uh you know and I mean if that is true then you could say yeah the likelihood of getting the the money back is higher uh it's disrupting uh those business model software in particular we saw um yeah repricing and substitution on substitution fears um at the beginning of the year um and then you've got kind of offsetting forces you've got kind of you know the short term and the long term of AI the short term is the is the investing and then the the question will AI deliver the productivity that everyone everybody is hoping for and you know lift growth uh long term and I think that's kind of you know the disrupting business models view would kind of support that longerterm optimism view and I think that's why you get this offsetting uh impact from the negativity that comes uh from the war but then on top you've got this energy shock um you know if you just pick an example of you know how how closely AI is related to energy you know we hear a lot about um data centers data centers cannot run without without energy so of course that's going to change economics um of of AI what we what we're trying to look at is you know this collision uh that is taking place is taking place through valuations it's taking place through confidence it's also taking place through uh balance sheets um and so um yeah I mean what we're seeing kind of now is that the equity price decline was was broad-based it hit cyclical sectors um but I think where we are now is that it's a kind of a known unknown own but this interaction element of it is new the AI shock plus energy shock um we don't have experience of it you know so I mean it leaves a cloud of uncertainty um as we as we head into the as we head into the future and of course that has >> had broader impacts in markets >> okay that's uh so uncharted waters uh essentially um another element of this and and this is a particular concern. It's been raised in the past, but I I I believe it's becoming more of a concern is that a lot of the money that's swirling around all of this is in private markets. How how does that affect the situation when you look at financial stability? So look, I mean, one of the questions that we get most often, uh, people look at the size of of of private credit today and they compare it with subprime 20 years ago, >> just before the crisis and ask the question, do we have a subprime 2.0 2.0 >> situation? I mean to be very clear on this, uh, we don't think so. Um, but we do think that there are a number of issues that are that give rise to concern. One of them is data gaps. Um the opacity of the sector because it's private. You don't have that disclosure that you would have if these companies were listed on the stock market say. And uh and then there's also the potential for confidence effect uh spillovers um to you know related markets. So lower tier credit that is that is traded in the markets you could get a read over uh between them. I think you know if I was to pick three points why private credit is not subprime 2.0. Um the first is >> just just to interrupt before you go to these first two points for our younger viewers and our younger listeners. Subprime this was the trigger for the global financial crisis in 2008. So it was it was a real major market meltdown. >> Yeah. I mean it's >> with economic consequence. You're absolutely right. I mean I kind of lost track of the fact that it's like half a half a career ago a normal career ago. um when the you know when people have done the expo expost diagnosis of what went wrong what were the vulnerabilities that caused that subprime thing to explode in the way that it did and what's different now I mean I would point to maybe three things patient capital is the first uh investor lockups um investors are locked in more more or less for the for the maturity of the investment um so that limits kind of the the fire sale dynamics that you would see if you had short-term funding of long-term assets which is which was one of the vulnerabilities that um that was exposed in subprime leverage um it's a lot lower than it was pre GFC lower than instructor credit actually private credit is much less leveraged than normal banking activity you can get as you know as as lowly leveraged as onetoone 50% credit 50% debt >> and then the last is a bit the third factor is a bit linked to the first Uh it's maturity transformation was one of the issues that became prominent um short-term funding long-term assets. You don't have this um in this um in in the in the private credit story. But Paul, you know, I mean, I don't know, you're a communications guy. You you'd know the name of uh William Randolph Hurst. You know, he was he was the one I think who said if it bleeds, it bleeds. And um the headlines you know FT economist all the all the financial press that we we read I mean the headlines focus on the on the business that the US business development companies the BDC's for redemption requests um have outrun [snorts] uh amounts honored we have a we have a chart in the office showing that that you've got requests made for redemptions by investors and how much they got back and there's a lot less coming back than was asked for. Now that means of course that okay investors are not getting all their money back but that also shows the redemption gates are working they're not failing. Um now there's not enough time to go through all of it now in this but we have a special feature this time on dedicated to the private credit issue because we kind of expected that you know readers of would expect >> yeah it's an issue that has been prominent [clears throat] in the press for some time. What our overall conclusion here is I have to be careful with the language what we're saying we don't think it is systemic per se in the euro area uh but we do think that there are situations that could arise unlikely ones but still plausible ones where it could become a source of stress for the financial system but we don't think it's going to affect banks the core of the financial system but more the natural investors in the type of assets that are produced by the industry long-term assets. So that's insurers, pension funds, and so on that could experience that could experience losses. But again, those are financial institutions that don't ordinarily or cannot ordinarily suffer runs. >> Mhm. Okay. So, uh just to sum up in in very simple terms there, uh subprime 1.0, there was unexpected contagion, I guess you would say, from the subprime market through the banks. Do you think that's less likely? um this time. >> Yeah, I mean I think one of the big things I mean we used to talk about too big to fail. There was too interconnected to fail that came out as one of the big lessons out of subprime subprime 1.0. >> There is interconnection here but it's not in the same way that it was in the you know we had all this credit transfer that took place and people were asking where is the where is the credit being transferred to and then we found out that the banks that were trying to transfer credit it ended up back on their balance sheets in different form. So now that risk is being spread more widely uh in the financial system and I think one could one could also say you know the regulatory reform was really kind of aimed at moving these risks off the balance sheets and the business models of banks and in somewhere else non-banks um are doing we don't call this a stress test but it is a test of the resilience in a way of non-bank finance for the first time. >> Okay. All right. So we drill down into that aspect of financial stability and the risks are out there. I want to pull out a little bit again because the other big thing is the war. Um when I say the war, there are two significant wars going on that affect Europe. I'm talking here about the war um against Iran. Um there's a lot of narrative out there in the media saying and analysts are saying this as well. Why are markets so benign? Why the straight of hormuz is closed? Energy prices are soaring. Um, on the day we're speaking now, uh, which is May the 26th, and I say this because things can change every day. We've just seen more attacks by the US on some facilities in Iran. [snorts] Um, would you say the market is underpricing the risk or [laughter] >> everything you're saying is absolutely correct and I think it's been a puzzle to many. Um I think you know Paul that we we rely quite a bit on market intelligence preparing the disability review and one term that I came back from a recent visit to the US with was a term that is normally used in psychology um cognitive dissonance um and it's this kind of the unusual cam against in markets against a background of you know a lot of surprising ing developments, let's say. So, we got stretched valuations, low volatility, and it looks like the market is priced um for a for a short war. What cognitive distance by the way, you know, it's this, you know, how can you reconcile two conflicting views or beliefs at the same time. So, smokers know that smoking kills. So, so why do they smoke? >> Villages um stay at the foot of of active volcanoes. People know the volcano is going to explode someday, but still um you can hold these conflicting uh beliefs. Um and it can be rational uh especially when you've got binary outcomes. So AI optimism and and ceasefire hopes, they're real factors. They're positive and you've got also all the downside points that you mentioned as well. So I think the first point we don't go into the psychology in so much detail in the in the episode but I think what we're what we're saying is that it doesn't necessarily look irrational. Um but high valuations plus comp compressed risk premia plus very low volatility is um a classic um underpricing uh configuration and it does leave markets vulnerable to shocks. Um so they I think we we can be confident on that point. Volatility does look a bit low and I mean and that's everything else given what's happening in the world right now. So these are the a lot of the risks that are out there. >> Um Euro area banks obviously are striving to and we expect them to strive to build a resilience against this but there are still going to be vulnerabilities and part of the job of the SPSR is to point those out. So what have you been looking at? >> So first point um banks are not the weak point today. I mean that I think is something and we said it last time that's where you started. um the resilience of the banks is something that does bring um a pos a big positive element to the overall financial stability uh assessment. But banks are exposed to nearly every channel u that we've discussed um through the energy uh shock uh AI maybe long-term exposure to private credit and so on. Um why why do we say banks are are resilient? I mean just look just look at the numbers. I mean just maybe three numbers. Um or we return on equity is around 10%, it's been there for a while. Um core equity tier one is at a record high. NPLs are 2.2% uh which is like more or less historical lows. And then in addition um in every banking union country now we have releasable capital buffers. We did not have that on the when the pandemic came and and shocked us all on the energy vulnerabilities. I I think and and and exposures to the Middle East more generally. I think direct exposures of banks are are very small here. Um so what we're really saying is that it's the second ch the second round so called what what macroeconomists call the second round channels the macro impacts the energy trade sensitive firms smallmediumsized enterprises that are energy reliant and there's also the interlinkage between the non-bank financial uh institution sector and banks through funding linkages. So banks are net debtors uh to the to the non-bank sector. Um so they borrow short term. Um in a stressful situation, you know, I mean, you could you could write a a narrative for for a stress test around a liquidity squeeze coming from NBFIs that stop that stop funding banks working through overnight deposits, repo market, and so on. That could be uh potential source of stress. But I mean overall I mean we we we think banks are resilient. Um but we need to watch we need to watch these links uh to the vulnerabilities that were we're highlighting in this issue of the FSLR. So correct me if I'm wrong but a sort of a takeaway from this is that the uh the shocks are serious. Uh the risks are uh very um uh weighty. >> [snorts] >> um the resilience is to a large extent there for the financial system but overall are you calling for new policies improved policies does more need to be done here >> I think [clears throat] it's my answer to that is echoing a bit what I what I the advice that we would give to to sovereigns I think it applies equally to macro credential authorities um familiar policy messages but the urgency has changed um so The messages that we have this time for banks are things like maintain releasable capital buffers. We have that uh this is not a time for releasing. Uh keep borrower based measures things like loan to value ratios, debt service to income ratios and so on. keep them in place and also um well we have this initiative for simplifying supervision but supervision we I mean our view is that supervision so fine to to simplify but do it without diluting uh resilience and then for non-banks uh you know it's all about implementing agreements that have already been made on on leverage liquidity reforms build an EU macro credential framework and maybe one of the new ones that we've added this time is close data gaps, especially on private credit. Uh to assess exactly what the implications would be of a full-blown private credit stress. It can't be done without more more complete data. And then I think the last thing and I mean this is becoming more and more important and I'm actually I've been involved with the financial integration report this time as well and it's all about advancing the savings and investments union and financial stability can and a solid market potential framework I think can ensure can make Europe attractive for a successful uh savings and investments union. So, three pillars, Brazilian banks, non-bank uh reform framework and deeper EU capital markets. Okay. All right. It's uh for anyone out there who wants more information on financial stability. All the materials about our so-called FSR that comes out twice a year are on the website of course. So, have a look at those. There's some good stuff including and John's too um modest to mention it but an article in there by John himself on the housing market. Um John thank you very much indeed but we're not done because it's hot tip time. What have you got for us? So you know there isn't a space every time in the efaur to cover you know all topics that everybody is interested in. So um I thought an unrelated topic, one that we are looking at um and we've discussed before is the whole issue of blockchain uh crypto. So I I'd like to make a book recommendation. Uh it's called the blockchain scholars book uh current academic insights condensed for busy practitioners and it's edited by Daniel Lebo and Simon Trimborn. Uh why read it? Uh it's got wide coverage. It covers issues from tokconomics to stable coins to CBDC. Articles are short. They're a few pages each, designed for busy readers. And I think one of the key points is that the articles are written by academics, but they are not written for academics to read. They're they're written for practitioners to read. So, it's informed um with the by the leading academics um in this area. And uh you can read one of the articles in less than half an hour. >> Okay. >> Hard to beat. >> Yeah. Wonderful. Thank you very much, Steve. Great tip. John Fel, deputy director general for macroential financial stability here at the ECV. Which brings us to the end of this episode. You've been listening to Euro Matters with me, Paul Gordon. Of course. If you like what you've heard, subscribe to the podcast. Wherever you're listening, do leave us a review. We're always open to improvements. And in the spirit of Europe, I'd like to end in Irish and say for until next time, thanks for listening.