 I'm delighted to introduce to you our first keynote speaker, Wenxin Du. And Wenxin is an example of what we really cherish at this conference. Her work is not only academically outstanding, but highly relevant for central bankers and market protection years. So we are very happy to have you. Wenxin worked, or after earning her PhD at Harvard University, Wenxin worked as economist for the US Fed Board before becoming a professor of finance at Chicago Booth School of Business. And most recently, professor of finance at Columbia Business School. Wenxin, the floor is yours. Great. OK, great. Thank you so much for the invitation. It's a true honor to be here. Today, as you'll see, the title of the talk is Repo and FX Swap, A Tale of Two Markets. In the first three talks, we've already mentioned a lot about the repo market. But hopefully, by the end of the talk, you'll appreciate this other market a lot more. So in this talk, the central theme is that repo and FX swaps are the two most important short term funding markets for the US dollar liquidity. As I will show, the two markets share very many similarities, but also are distinct along several important dimensions. A very important commonality between the two markets is that the dealers and banks play a central role in both markets. And in particular, the time-varying intermediary constraint in part of the post-crisis regulation is the key driver of the pricing in both markets. A big part of this talk is also to use this opportunity to demystify the FX swap markets. And we're going to, for the second time today, make use of the MMSR, the Money Markets Statistical Reporting Transaction Data collected here at ACB. And a central message is that despite the breakdown of a fundamental no-arbitrage condition in FX swap called cover interest parity, so if we think cover interest parity doesn't work, the market isn't functional, we want to really highlight the contrary. The FX swap market is overall a very well-functioning market, and it serves a very important function of fulfilled significant dollar demand during stress days, particularly reaching out to market participants who do not have ready access to the cash market funding alternative. So the structure of the talk has three parts. The first part, I'll give you an overview of the structure of the two markets. Talk about the cash flow market structure and the balance sheet implications. The second part is I'll discuss more on the pricing side in terms of linkages between the two markets. I'll comment on the funding cost comparison on average and also in the cross-section, and also specifically focus on the quarter end dynamics where lots of interesting happening in money markets. And finally, I'll use a few minutes to discuss implications for financial stability policies and also perhaps open that up for Q&A. OK, so first let's start with the overview of the two markets. This is the single most basic cash flow diagram for short-term funding markets. To give example of JPMorgan as the dollar lender in the US dollar repo market, and we have some USD borrower, doesn't have to be a US person, can be somewhere outside the US. The transaction flow is at time t. JPMorgan is going to give dollar cash to the USD borrower in terms of a repo loan. So it's a reverse repo on the bulk of JPMorgan. And the USD borrower pledges the US treasury as collateral. At t plus 1, all the cash flow reverses. We have the USD borrower returns the dollar cash, and JPMorgan returns the collateral. So how does the FX swap market actually compare to the repo transaction? In fact, very similarly, all you need to do is replace the collateral instead of using the US treasury bond but use a foreign currency instead. So here, since we're at the ECB, I'm using this example of JPMorgan lending dollars in the euro dollar swap markets. And the USD lender is going to essentially exchange the dollar borrowing for euro at the inception of the trade. So euro can be viewed as the collateral for the dollar lending. And in this case, the spot exchange rate is used at time t plus t to exchange between dollar and euro. So it's happening at the market rate. And at t plus 1, the USD borrower is going to give back US dollar, and the JPMorgan is going to give back the euro. It is very important to note that the FX swap does not involve any FX risk because the exchange rate at t plus 1 is preset at time t. So the exchange rate between dollar and euro is going to happen at an outright forward rate that is predetermined and pre-agreed upon between the two counter parties. So in short, as you can see, FX swap is a very powerful funding instruments in addition to currency hedging instruments because in this case, the USD borrower achieves the same goal of getting US dollar. And the banks also fulfill the mission of lending the dollar securely. In this case, not collateral by US Treasury, but collateralized by a foreign currency. And with initial and variation margins, it's just as secure as a typical repo transaction. So in terms of the overall market structure, both markets share a lot of commonality. In particular, the global dollar funding market is highly segmented. In a sense, we have a lot of cash-rich dollar lenders who do not have direct access to people who want to borrow dollars. So in this chart, I'm just giving you some example. On the left, we have some cash-rich lenders in the USD markets, such as the money market funds, the prime funds, or the government funds. And towards the end of this chart in the orange box, we have the ultimate dollar borrowers, such as banks, hedge funds, institution investors, and the corporates. It is important to note, because of the strict mandates of US money market funds, they are not allowed to essentially lend to anyone. They can only lend to the US governments, or they can lend to banks with very high ratings, subject to very strict diversification limits. So not everybody in the orange box is going to be able to borrow from the cash-rich money market funds. So this motivates a significant need to have a large amount of intermediation, which are these entities sitting in the green box. So these are the large global banking organizations. They don't have to be US banks. In fact, non-US banks play just as an important role in intimidating dollar funding in the global markets, in addition to the banks. International banks, obviously the affiliated repo and FX swap dealers, are also facilitating the plumbing of the global dollar system. So in terms of repo and FX swaps, we think of the repo markets typically tend to be a bit shorter dated. The overnight is due the most common tenor. And in terms of the size, we're talking about $3 trillion in the US private repo market. In recent years, because of the large expansion and take-up in Fed overnight reverse repo facility, there is essentially additional $102 trillion between the money market funds and the Federal Reserve. That is not part of the $3 trillion. So as most people in this room thought, repo is perhaps the largest short-term funding market. In fact, there is a larger one. And that is the FX swap. So the tenor also tend to be a little short-term, except maybe not as short-term. I mean, you can have overnight and spot next and tomorrow next FX swaps. But a large part of the liquidity is also centered around a few term dates, like a week, a month, for instance. And the size in terms of the USD-linked contract is currently outstanding at close to $100 trillion. So significantly higher in terms of the total size when compared to the repo market. So now let's talk about the balance sheet implications between these two type of activities. On this slide, the title of the figure is Total Activities. So it's not, I mean, as I shortly explained, it's not everything is on balance sheet. When it comes to the repo, you think about the bank's lending repo in the jargon. That's reverse repo position from a bank perspective. Think about this as the JP Morgan's balance sheets. And on a liability side, JP Morgan may borrow in order to lend, also in the repo market. And there could be some additional source of unbalanced sheet funding that JP Morgan is going to be able to source to make up the difference between the repo lending and the repo borrowing, and therefore end up with a net repo lending position, OK? Very similar for FX swaps. Again, JP Morgan could lend dollars in the FX swap markets. And meanwhile, it can borrow dollars in the FX swap markets. And if there's a deficit between the overall gross lending and gross borrowing, they need to source some additional cash market funding in order to support that, OK? So the mechanics is very similar. Large dealers have large gross positions on borrowing and lending sites in both markets. However, there is a key difference. The key difference is, at least in the US dollar markets, the gross repo borrowing and lending positions cannot be netted out, whereas the gross borrowing and lending position for FX swap, because there are derivatives, largely off balance sheet, except for the margins, can be netted. So if we change the title of the figure to unbalanced sheet activities, you see a significant difference here. You see in the most majority of the case where the repo transactions cannot be netted out, the repo is a lot more balance sheet intensive when it comes to just think about the total size of the balance sheet usage. However, if you're doing match book FX swaps, the gross borrowing and lending FX swaps can be netted. However, if there's a then a deficit, if you want to be a net lender or a net borrower in the FX swap markets, and that is supported by additional unbalanced sheet lending or borrowing, and that net lending part cannot be netted out, and therefore stays unbalanced sheet, OK? So that's a very important distinction. It's very useful to keep this in mind when we talk about the quarter end dynamics later. OK, so in addition to the usual balance sheet implications, speaking of the bossel-free and non-risk-weighted leverage ratio, basically it's the total size of balance sheet that matters the most. We just discussed the difference. When it comes to the other type of important regulations and regulatory metrics, I'll just highlight very briefly, like both types of activities are virtually pretty much risk-free. So they involve very, very little risk-based capital charge. So in addition to the implication for the leverage ratio, depending on whether you're doing growth versus net, which market you are operating on, there is also additional consideration, and that is the GSIB, which is Central Global Systematic Important Banks, capital surcharge that has different dimensions. Banks have to score on different dimensions, and in particular, the FX swap activities tick more box for GSIB surcharge. And in managers' decisions, the GSIB surcharge is assessed based on the year and snapshot of bank balance sheets, and therefore, on year ends, the FX swap markets, generally speaking, are more on the distress because the activities tick more GSIB surcharge boxes. All right, so that was the background to set up the stage for us to understand the mechanics of these two markets. Now it comes to the meat of the talk and talk about the pricing of the two markets and the connections between the two markets. So let's start with the repo first. So we know in the US, we have completed the transition to the new benchmark rate, and that is so far, which Central Secured Overnight Financing Rate. And to many of us, it's just like a one rate, but in practice, there are many repo rates. I mean, I guess the first talk talked about the euro area collateral scarcity, but even in the absence of specialness of different QSIBs in the repo markets, even in terms of general collateral borrowing, repo rate can be different across different market participant types. So this first slide is strictly taken from a summer statistics from the Federal Reserve Banks of New York's SOFR publication, where we have reported the average SOFR rate between 2016 to the present at around 177 basis points. And it gives you the distribution of all the individual rates that go into the SOFR, which is a volume-weighted medium rate of all the individual repo transactions. So it is important to note that all these individual transactions could include some special repo. So usually, I wouldn't take the first percentile to indicate the GC repo rate, but once you go to the 25th percentile, it is still generally reflecting the demand for cash as opposed to specific demand for collateral. So what you see here is that the SOFR rate at 177. And what is this rate? This is the tri-party repo rate. This is the specific segments of the repo market that goes into the SOFR that reflects specifically the borrowing from money market funds from the large bank dealers perspective. So just thinking about our diagram, this is the top leg of the chart where money market funds lend to the top tier global banks and top tier global banks enjoy the privilege of borrowing from money market funds at a very low rate. So a tri-party rate at 175 basis points, but SOFR generally lines up decently with the tri-party rate. So you think about this as the large banks funding rate in the repo market. They're doing the business. It's very balance sheet intensive. They have to turn around that dollar at a higher rate. How much can they earn? So if you go to the 75th percentile, they earn on average not tensome amount, like three basis points. But if you go to the 99th percentile, they can earn up to 15 basis points on average doing a matchbook repo in the mediation, borrowing from money market funds, and then turn around lend to someone like a small bank dealers or hedge funds at 190 basis points, capturing the 15 basis point spreads. So that's the mechanics of the repo market when it comes to the price dispersion. How about the FX swap markets? Generally speaking, it is more expensive to borrow in the FX swap market for US dollar funding because we have a deviation from cover interest parity condition. What CIP is is that post crisis, the implied dollar funding rate from the FX swap, the first term on this equation highlighted in blue, is going to be generally greater than the cash market dollar funding cost measured in repo or OIS or LIBOR. And for the euro dollar swap market, at least. But how exactly? How much is this exactly? Like we know in the repo markets, going from the lowest rate to the highest rate is about 15 basis points. But when it comes to the FX swap, so in the ongoing work, I was a drug stressor here at the ECB and Adrian Verdahan at MIT. We make use of the euro dollar FX swap transaction from the MMSR data, where for each transaction, we observe the actual price. So this is not Bloomberg quoted price. This is actual price. So over the three year sample between 2018 to 2021, we get around 30 basis points. Doesn't matter whether the MMSR banks are borrowing or lending on average, doesn't matter if you're borrowing or lending on average, the funding rates implied from these actual transactions are about 30 basis points higher than the US dollar repo rate. So this means the FX swap markets always operate at a higher rate compared to even the most, I guess the highest segments of the repo rates. So there is a higher key going on. So despite the dispersion, but the segments overall is more expensive, okay? And this already rinse about as something's more expensive, sounds like it's not operating as efficiently. And without doing this analysis, when I first go to Bloomberg, I began on to calculate some FX implied funding rates here in the plotted in the form of CIP deviations for the tomorrow next tenor. I see a lot of big, big spikes. And so I started with the Bloomberg quote, which is in blue. I was like, okay, there's so many spikes. It looks like this market again has some issues. How come you have so much volatility? But what's amazing is that when you compare to the actual transaction, which is plotted in red, you can barely see the difference. So the quoted big spikes from Bloomberg's are actually coinciding very closely with the actual transaction price. So price are high for a good reason. There's actually transaction taken place at this high price, okay? And another way to look at whether the FX swap market is functioning or not, is to look at within a certain trading day, looking at price dispersion across different market participants. So we discussed in the repo market that dispersions around 15 basis points. And for the FX swap markets, the nice thing about the MMSR data, it also gives you all the granularity of the counterparty types. So this shows when the MMSR banks are borrowing dollars from the tomorrow next segments of the market against from all these different market participants. As you can see, when the MMSR banks, these large, your area dealers are borrowed from their customers, obviously, they get a better price, but not that much better, okay? So the spreads are like ranges between one, two, three basis points in terms of the markup, if you like to use that word. And in terms of lending, they also get a little better price. And if you go to the three month tenor, which is the typical tenor we look at, or the one month tenor, thinking about institution investors hedging their dollar exposure, overall the price dispersion in a cross-section remain fairly content between one, two, three basis points on average if you do like a volume weighted calculations. So despite these large, large spikes that you saw in Boomburg screen, this market is functioning. And in terms of price dispersion across participant types, it doesn't really strike to us as something highly unusual. But then you might say, wait a minute, there are certain days that are very, very strange about these markets and those are the quarter end days. For most people in this audience, we know quarter ends are very special because in many non-U.S. jurisdictions, the Bossel 3 leverage ratio is assessed based on the quarter end balance sheets and therefore banks have strong incentives to manage their quarter end balance sheet exposure and to cut down balance sheet intensive activities such as the match book repo and let that to a big dollar funding shortage, okay? And these shortages are showing up significantly in terms of the level of these spreads. So what we show here is if you take the GCF repo which is more like the 95th percentile of our software distribution or 99th percentile of our software distribution minus the tri-party repo, that match book repo spreads spikes by about 30 basis points. And if you look at GCF alone relative to interest on reserve paid by the Fed which doesn't spike on quarter end, GCF spikes about 40 basis points. But there are large, large spikes if you calculate the overnight FX swapping platforming rates. If you do it from the euro dollar markets as shown here, on average they spike about 150 basis points on the overnight basis. And if you do the dollar YAM markets they on average spike about 400 basis points. So despite the fact that for the longest time post-GFC both the BOJ and the ECB here pay a negative interest on excess reserves. Once you swap those negative euro rates and negative YAM rates back into U.S. dollar terms, on quarter end there are 150 basis and 400 basis points higher than the interest rate paid by the Fed which never went negative. So it's quite remarkable. So this price, if you just look at the surface it seems like makers may not be functioning well. Liquidity might be highly dried up. Like no one is transacting at these crazy prices but is that really the case? So first as confirmed by the strong intuition the reason that we have these big spikes in funding rates on quarter end is that non-U.S. banks in particular contract their matchbook repo intermediation. As been evident from this charts where we're collecting the data not all for all non-U.S. banks but for the selected non-U.S. banks that file daily reporting to the Fed under the liquidity coverage ratio disclosure. What we have is that these non-U.S. banks contract their repo lending by about $60 billion and in the meanwhile they contract their repo borrowing by $80 billion. So on net it looks as if they are on net lending more in repo but there is a massive contraction in the gross book of repo borrowing and lending. And this additional net lending is financed of your draining reserves on quarter end days. Okay, so there is enough evidence on the repo side we know the cost and the reason for why we have liquidity crunch but the mystery part is what happens to the FX swaps. We know the prices are way higher than the repo spreads but is there any transaction? And the short answer is yes and there's even more transaction on quarter end dates. So this is again demystifying the markets using the MMSR data and what you can see here is this is the event study style diagram looking at the gross borrowing. So this is the MMSR banks gross borrowing of dollar in the repo markets, sorry in the FX swap markets. Instead of going down it actually went up by about 60 billion if you look at the short term tenor. And the overall lending very different from the repo transaction, the overall lending stayed pretty steady throughout the quarter ends. On net it looks like everybody on average is on net lending less dollars in the FX swap but that's not because of a contraction lending it's because increase in borrowing. Okay, so this is a very interesting and to our surprise results without I guess the no hypothesis might be the other way around but after looking at the results and just to do a quick recap, we know that the quarter end balance sheet constraint contracts, I mean quarter end balance sheet capacity contracts significantly because of these regular reporting reasons and instead of having a very obsolete market in fact FX swap markets serves a very important function because euro area banks in particular these money market, these MMSR statistical reporting banks significantly increase their dollar borrowing in the FX swap markets and in short this market acts as important residual markets for the MMSR banks to borrow dollar funding on this very stressful days and overall there is actually higher FX swap volume that actually incurs the significant quarter term premium. Okay, so all right, so we're doing good on time I will use the rest of the time before the Q&A to talk about implications for financial stability. So first let's comment a little bit on the central bank liquidity facilities. So when we look at these large large spikes on quarter end here it's just a recast of the same phenomena as shown in the green which calculates the implied dollar funding rates here I'm switching to the dollar yen markets to make the point even stronger because dollar yen tend to have a higher cost when it comes to the dollar borrowing compared to the euro dollar market. So what you see here is this is the private implied dollar funding rates which spikes every quarter end this is using the one week contract so it spikes the last week within the quarter and stays high and only normalized after the quarter end so it's not a one day phenomenon it's a five business day phenomenon, okay? And what is the red line? The red line is this dollar funding rate that banks can actually borrow from the central bank swap line at the BOJ. So what have we learned by looking at this chart in conjunction to these blue areas which is the actual takeout from the central bank swap lines summing up all these different central banks take up. So what we have learned that it is pretty common especially on these quarter end days for the private money market rates at least in the FX swap markets to exceed the official rates from the central bank liquidity but there isn't always a take up except when we are in a crisis situation. So in other words, to manage these quarter end spikes we shouldn't really rely on the central bank swap line banks are not going to them I mean there is a tiny bit of take up like maybe a cup of billion dollars but it's quite negligible especially compared to the crisis take up. So central banks liquidity swaps are not there to manage the quarter term premium but instead if we have a major market dysfunction in these short term funding markets during the peak of GFC during peak of the European debt crisis and during the peak of March 2020 COVID pandemic the swap line was actively drawn. It was one of the largest liquidity facilities on the Fed balance sheet and the swap line was quite effective in bringing down the dollar funding costs from the FX swap markets. Okay, since the title of the talk is repo and FX swap a tail of two markets so far I've only talked about the FX swap you might wondering how about the repo market didn't we also have a facility for the repo and the answer is yes. So here I'm still plotting the swap line take up in blue and I am plotting the repo facility take up again facility, liquidity facility offered by the Federal Reserve in orange. What you see is that very interestingly both for the GFC period and also for the March 2020 period the demand from liquidity facility in the repo happened a lot earlier but once we recalibrated the FX swap facility parameters that boosted the massive increase in the take up in the FX swap there remain to be very little interest in drawing additional liquidity from the repo market. So even in terms of drawing central bank liquidity in stretch time there seems to be important substitution going on and there seems to be at least with two data points suggesting after massive take up in the FX swap market the unbalanced sheet cash market repo and the un-shown unsecured markets became normalized a lot more quickly. Okay. And this is just to zoom in the March 2020 episode where you see there was initially increasing demand from repo facility but after we changed the auction frequency for FX swaps from weekly to daily so every day the market participants can dollars in the FX swap markets there that boosted the demand for the FX swap significantly and that also coincides with the dwindling demand from the repo facility. Okay. And the next question that is on everybody's mind is how about normal time? How about now? And okay. So maybe before going there let's do a recap of the crisis response first. So my personal view in terms of repo and FX swaps and which markets should we intervene which market is more critical. Obviously both markets are very important but it's very, very important to remember as we mentioned in the first slide the global dollar funding markets are highly segmented and the nice thing about the FX swap is it represents the most challenging segments of this very segmented dollar funding markets. A lot of market participants getting funding in the FX swaps do not have ready access to the repo to the dollar unsecured market. And therefore there is a hierarchy of money market rates where we have the FX swap being the highest rate when it comes to dollar funding followed by the unsecured term funding rates like CDCP used to be live or not died and OIS and then repo and Fed swans are typically the lowest rates. So the natural policy implication when it comes to crisis response is if somehow we can help bring down this highest rate down to normal then all the other market follows through. If you fix the most hardest, the most challenging segments of the funding problem then you don't have to worry as much for the other segments. And therefore in my opinion, the central bank swap line is one of the most effective tools when it comes to crisis response. Okay, now that was it for the crisis response. Let's spend the last few minutes before the Q and A to talk about the normal times. Okay, so normal times thinking about now where in the second round of QT for the US and the first round didn't end up with a lot of I guess comforting news when it comes to the functioning of the repo and FX swap markets, it ended with a ban. And that was September 16, 2019 that many people in this room distinctly remember. The green line, sorry, the blue line made into the headline of all the major newspapers. And that was the mail down in the repo market. So the green line, the blue line plus the GC repo rates in the morning of September 17th, it reached to near 10% and the Fed immediately launched the repo facility and calm the market afterwards. What perhaps didn't make into FT and Wall Street Journal are the red and the green lines. These are implied dollar funding costs from the overnight Euro dollar FX swap. Again, very much consistent with the overall theme of the talk. It's a market that's highly synergistic with repo market. Even though we think for this particular episode, the pressure might be originating from the repo market, but the FX swap market pretty much on an intro day basis had a very similar pressure. And it also rises in the afternoon of the 17th and peaked in the morning of the 17th. So as many people in this room now agree, what was the problem back in September 2019 is that after two years of Fed balancing normalization or the previous round of QT, there was not enough reserve in the system. Even though the number was still pretty impressive, 1.4 trillion, but relative to the demand of the banking sector for, for instance, their intro day liquidity needs, 1.4 trillion is not enough. So what brings the natural question today, what is the right amount of level, right? If you look at the charts, this was the blip that we were talking about. The reserve is showing red, right? I mean, in terms of the trend, it was like definitely shrinking and then we hit the air pocket for both repo and FX swap markets. And this time around, the Fed has normalized this balance sheet more in terms of decline in security holdings. The reserve has staying overall relatively steady because we have this extra stuff that we didn't have as much. And that's the overnight reverse repo facility in terms of extra cash power at the Fed, which was useful in terms of the first area to dissipate if the Fed continues to shrink its balance sheet. But the natural question is how far can we go? And another important factor to come in mind is not only there's a shrinking amount of cash balances held in the commercial banks, there is often time at increasing demand for repo during QT at exactly the same time. Why is that? Because that helps, it helps to understand the mechanics of the rebalancing in the cash US treasury market. So the Fed is not holding the treasury and precisely during a rate hiking cycle, the yield curve of the treasury market is highly inverted and negative. The real money investors are not very interesting loading up longer data duration at this time than who has to pick up the gap between the supply and Fed now rolling over and the real money not willing to buy up. It has to be the intermediaries. Intermediary takes two forms. The primary dealers themselves in the treasury markets and levered investors the hedge funds that rely on primary dealer balance sheets for repo funding in order to finance their treasury purchases. What we have seen is that during the September 2017-20, sorry, during the 2017-2019 QT cycle, the primary dealers increased their treasury holdings by around 150 billion dollars. And the levered investors as proxied by their negative position in the treasury future market increased their position by a slightly less. But this time so far, the primary dealers increased their positions maybe a little less than 100 billion dollars, but there has been a massive take up in the implied levered investors position. This is what is the treasury cash future basis trade that made into the FD Wall Street Journal headline in recent weeks. So even though these are hedge fund positions we think they still rely on dealer balance sheets because dealers are providing the repo financing for them. They're not taking the outright risk as in this case when the rate volatility is high, dealers are more timid taking on their own position, but they're still giving large amount of balance sheet capacity to hedge funds for them to finance their treasury purchases. So the natural challenge is as we have an increasing demand for repo, if you want to keep doing a matchbook intermediation from a dealer perspective, you're going to run into your balance sheet limit pretty quickly. So the workaround is instead of increase the size of balance sheet doing more effects, lending in effect swaps when repo markets, dealers have to drain reserves to meet the demands from hedge funds to finance the basis trades. And this is where the reserve, the optimal reserve level becomes even more important just to in addition to accommodate this additional levered investor repo finance treasury demand. And finally, I will just conclude and dollar funding market is highly global because of the global reserve currency status of the dollar and repo and FX swap markets are intimately connected. Hopefully after this 40 minutes or so, we should reach the understanding that both markets are crucial to global financial stability. And one important finding from this ongoing work with George and Adrian using the MMSR data is that despite the seemingly dysfunctional, no arbitrage violation, despite this large and persistent CIP deviation, the FX swap market is functional and serves the core function of providing dollar funding to the segments of the market which lacks access of direct funding. Just at the higher cost and in stressful days, this is a very, very important the residual market to backstop dollar liquidity. And finally, the resilience of both markets highly depends on the intermediary balance sheet capacity, ample amount of reserves and central bank liquidity backstabs. And I will just conclude here and happy to take questions. Perfect. Many thanks, thanks. If you want, you can also... Okay. I'll come back. Have a drink while we collect questions. Are there any questions here from the audience? Oh, sorry. Yes, please. And you'll get a microphone. Will Diamond from University of Pennsylvania. So very interesting evidence. You've shown connecting US repo markets here and CIP deviations. I'm curious if you think there's a much stronger connection there between say, frictions in a European or another country's repo market and what's going on internationally? Or do you think if you saw stress in Europe, UK, other developed countries that this would also be reflected in currency swap markets having frictions at the same time? I see Christian has another question. Maybe we collect two or three and then... Sure. Christian Kubitze from the ECB. Thanks a lot for the great keynote, Wenschen. I was very surprised to see that euro area banks, US dollar borrowing sharply increases at quarter ends. We would have expected probably the reverse from the previous papers and then we know that CIP deviation also increases. So if the cost of dollar borrowing increases, why would euro area banks also increase their borrowing at quarter ends? Do you have a story in mind that explains this? Why euro area banks need to increase their dollar funding using the most expensive source for that instead of increasing their repo funding or even the euro funding? We're very curious to hear your thoughts on that. Thanks. Yes, and one more question in the middle, please. Hi, Wenschen. I'm Shreya for those people who don't know me. My question is related to sponsored lending because I think you had a bullet on that on your first slide. So I think a lot of the deviations we see, like the spikes in prices, is due to the fact that these rich cash dollar lenders can't lend to the intermediaries that want the dollar funding, right? So with sponsored lending, that's supposed to mitigate that friction and then dealers are also able to net the repo on their balance sheet, not have those balance sheet frictions you illustrated. So I was wondering if you guys had thought on this seemingly new paper about if sponsored lending is going to reduce some of these price spikes we see between these two markets. Okay, would you like to answer now? Yeah, great. Not in any particular order. Thanks so much for the great questions. Going to the question first, why do euro area banks increase their dollar borrowing in the FX swap market? I should have spent a little more time and the reason is they cut back their cash market lending as a cash market borrowing in the repo market because repo is unbalanced sheet except the caveat that Shria was mentioned. So if you think about most of the gross borrowing and lending of repo is unbalanced sheet on quarter end, you want to cut back your balance sheet footprint so you immediately cut back your repo borrowing by a large amount. So in order to fulfill part of that deficit, you go to this other market, which is the FX swap and if you're able as a large dealer to do gross borrowing and lending in FX swap at the same time, you net out your balance sheet exposure. So on quarter end in particular, the FX swap markets from European banks perspective, if you already, especially if they already have euro on their balance sheets, it's costless to engineer that into US dollar and so it's a more efficient way to borrow dollars. An exception is on year end, so we didn't specifically mention the difference in the talk but what you see in the data is on year end, the pattern may not hold because as we briefly discussed, the G-sub surcharge actually has more to do with the FX swap so this relative balance sheet cost consideration doesn't necessarily apply to year end but for most of the quarter ends, it is a very efficient market in terms of balance sheets for European banks to fulfill part of the deficits due to the contraction in the repo market borrowing. Okay, so on Shria's question, yeah, thanks for bringing up. So it is true that most of the gross borrowing and lending in repo markets stays on balance sheet. The exception in the US is if the repo is happening in the sponsor platforms, so these are called sponsor repo and you can think of that as being also centrally cleared and these are not going to show up on balance sheets because the dealers are able to net out the borrowing and lending. This has important implications because a lot of these plumbing issues because we have a bottleneck in the intermediate balance sheet constraint but if we can net out these very safe market dealing type of transaction that's going to alleviate that balance sheet constraint. So far, the segments of the sponsor repo in the overall repo market remains to be a very small share. For many different reasons, I mean, it's more costly. I mean, from these participation perspective, if you adhere all these CCP rules, there's also an element that so far only the overnight repos are allowed on the sponsor platform. So if you want to fund term, you cannot use the sponsor platform but I do think it is a possibility for us to think about to see economize the balance sheet usage to promote more sponsored usage. And finally, going back to Will's questions about how about the other currencies. So if you look at the BS over-the-counter derivative survey, about 95% of the FX forwards and swaps have a US dollar linked legs. And the direction of CIP deviation often for most currencies vis-a-vis the US dollar except for the Australian and New Zealand tend to be in a way that it is more costly for market participants to borrow a dollar in the FX swap markets compared to like, say the dollar cash markets repo and CDCP or LIBOR. So this suggests, I mean, this market is very US dollar centric and people who do not have access to direct dollar funding would have access to their local currency funding and swap back into US dollar. So this is largely being primarily used as a market to get US dollar funding, not the other way around with the exception of Australian New Zealand dollar where the deviation goes the other way around which you can imagine a similar type of relative scarcity of the Aussie Kiwi dollar funding relative to US dollar funding. So we can imagine scenarios where in fact some of the Australian banks found in the US dollar markets and swap back into their local currency to meet their local currency funding demand. Perfect, many thanks. Just to remind those which are online that you can use the chat function in order to post questions and we're monitoring the chat and I can read out the questions if you have any. Other than that, I would also, yes, Arancia? So sorry, Arancia, do you have any views on the rule to mandate central clearing through treasuries basically? There's a debate around that and what it means for the overall size for the repo market, the pricing. I mean, there's a lot of debate whether that would take leverage down the system. I mean, what's your thoughts around that? Yeah, so it's somewhat related to the sponsor repo question. So my personal view is central clearing, the repo market seems to be more first-order than central clearing the cash treasury but one can argue perhaps these two things are quite related. It's hard to do one without the other and the reason is, right? I mean, if you look at dealer balance sheet it's the gross repo that's basically crowding out their total balance sheet usage. It's not their outright treasury holdings, right? I mean, they are treasury primary dealers but they don't really hold much of an outright position. We're talking about $120 billion, considered large for the entire collection of primary dealers where for repo we're talking more than a trillion dollar. So if you can net out that a trillion dollar is going to be a lot more efficient in terms of compressing the balance sheet. Any further questions in the room? Yes, please. Thank you. Do you, I mean, Wenxin, as you spoke very well, both the FX swap plan as well as the interventions on the intermediate balance sheets will make this market work better, compress things, provide lender flashes are in different ways. But could I get you to speculate a little more about whether we should worry about moral hazard, especially as we think about the development of these markets and how they, if governments intervene or central banks intervene, to what extent are we shrinking the ability of these markets to intermediate or not? Meaning, if we're very aggressive in doing so, are we destroying the markets? And somewhat linking also to our discussion this morning with the first paper of the extent to which we are leading to collateral scarcity, but just generally scarcity in the ability of markets to intermediate this, or do you see it as really a one way? You kind of, in your talk was very much of a one way. It seems like the intervention is going to be positive. But if you could think a little bit, if you could speak out a little bit about some counters of moral hazard, I would find that useful. Great. And then that testing has a question. Okay. From the yield spread between the FX swap costs and the repo costs at the end of the quarter, what is the implied perceived cost of equity capital? And we have a third question, maybe, over that. Thanks, Matthias Dreyman. I have a question. You linked the developments of this steady state size of the balance sheets, indicating that there's a link between leverage and investors. But obviously, there's a lot of talk there's too many leverage investors. So should we encourage it? And that's one question. The second question, is this a steady state? Should we think about it in the steady state? Or is it really only mattering when we change there? You show these charts where QT, so it's the change rather than 11. Thanks. Okay, great. Thanks a lot for another round of terrific questions. Maybe do it in the reverse order, because the third question is also somewhat related to the moral hazard question. So what we have shown you is that there is a strong negative relationship in the data between the slope of the yield curve and these levered investors positions, either measured from primary dealer's own outright position or from these implied levered investors running a relative value trade, basically going along cash treasuries and hedging out the duration risk using treasury futures. So it's particularly pronounced during the tightening cycle when you have a lot of movements in the slope of yield curve, which is basically what we have. And on top of that, you have the Federal Reserve not rolling over its existing holdings and the overall holdings of the treasury holdings have declined as in more than $800 billion. So I view this as a particularly important phenomena for the current stage. If you think about long-run steady state, someone has to absorb that duration and it's not the levered investor. It is not the primary dealer's. They're not in the job to hold those duration on a steady state. We should instead have the long-term institutional investors of pension insurance in particular to absorb more duration, but in order to incentivize the real money, there has to be a correction in the term premium. I guess in recent weeks, we've seen a modest rise in the term premium sort of going on in the right direction. You can also think about the size of dealer balance sheets or balance sheet capacity is a function of the speed of adjustments. Basically, if dealers are not able to accommodate more leveraged demand from the basis trades or themselves not being willing to increase their outright holdings more then the market pricing for the term premium needs to be adjusted more quickly to really attract the real monies to fill the vacuum. So going back to the moral hazard questions like whether we should make these markets as efficiently as possible to discourage, I guess, leverage and risk-taking. It's a great question. I think the answer is nuanced. So one way to think about it is all these spreads like either in terms of the GCF tri-party repo spreads or in terms of the CIP deviations are a form of insurance premium that we're paying just to ensure that the whole system is not overly leveraged because our struggle is not be able to really exactly monitor and pinpoint the level of risk-taking and instead we have a more blunt tool which is essentially the boss or three leverage ratio saying that nobody can be leveraged even if you're holding reserves more than X amount. So that is one way to perhaps prevent moral hazard and over excessive leverage-taking but from an economist's perspective this is always a little dissatisfying because we're not really making penalty of activities based on the risk. So another argument is to think perhaps the way that we are regulating the system by imposing these very bland total leverage constraint is creating some distortion. Perhaps we don't have to compromise as much of financial resilience if we were to increase the risk-weighted capital regulation but relaxing some of this leverage ratio. So overall we have a little less that we lost the whole system runs a little more smoother when it comes to these risk-free money dealing activities but when it comes to the actual risk-taking we have to be a bit more strict. So I sort of see benefits in both type of arguments. Going back to Annette's question on the cost of capital typically the bank, the back of envelope calculation is if you think about the expected return on capital is like 10% and with like a 3% boss or three leverage for the continental European banks we're talking about 30 basis points opportunity costs of putting on something that's balance sheet intensive that's largely in line with the overall size of the CIP deviation. So the back of envelope works out but on quarter end obviously these things go up much higher but it's only like four days a year and I think the Basel committee is actually encouraging more jurisdiction to move into a daily average regime like the UK has done that many years ago. So perhaps like it's not as relevant for banks to think about raising costly equity just to cover those days. Perfect, many thanks. So we do have two questions from our audience online. One is from Jens Peterson and I think it's very much related. How would it change in the calculation of the US G-CIP surcharge from year end to an average over the year change the year end spike in funding rates? And then the second question is from Dimitri's Steve Fanides who would like to thank you for your presentation and from experience the pickup and US dollar premium paid for US dollar funding via FX swaps has become smaller. The years gone by from 2017 to 2022. Is there a reason behind this? Yeah, great. Also doing in the reverse order like, so why is the FX swap basis become smaller? Like we talk about the supply side constraints from dealer balance sheet. There's also demand sector, demand side factor, right? Think about, we talk about levered investors these relative value hedge funds need repo financing to finance their treasury holdings. The parallel of that is the non-US based institutional investors essentially funding their dollar purchases using FX swaps. They don't use the word funding, they use the word hedging. Think about the Japanese life insurance companies holding large chunk of US trade rates and corporate bonds maintain the hydration on average which reported around 50% and that is a lot of hedging or short-term funding demand if you like and that's coming from FX swaps. So again, very similar to the lack of incentives to investing longer data duration when the curve is very inverted or flat for the FX hedged investors, if you think about earning the long-term yield and paying the short rate as the funding rate when the curve is quite inverted the demand is quite lukewarm. So one way to rationalize the relative unremarkable trends of FX swaps in recent days or recent years is that the curve is just not attractive for the FX hedged investors to come into this market which dampens the demands and therefore mitigates some of the basis. Okay, so going back to the G-sub surcharge if we were to move to a more daily averaging or like even months and calculation of the surcharge score, this relative dynamics that we have seen that the European banks basically switched out from the repo funding into FX swap market funding that pattern may not hold because if you add additional cakes especially for the largest most important banks and the way that the G-sub is calculated is highly non-linear you really don't want to go to the next bucket. So there could be a lot more frictions to make the FX swap markets like a reliable markets to get residual funding on the quarter turn as well. Perfect, many thanks Mingsi for this inspiring speech and for your great answers to all these questions. Many thanks. Great, thank you.