Great. Thank you, Michael, and great to see everybody. During the course of this session, we'll be talking about new research on dissecting FX investor demand, co-authored by my colleagues, Alex and Haoran. Many of you in the room will be familiar with our measures of investor flow and investor demand. Having published metrics for more than 25 years, we've really explored these indicators for 25 years or so now to gauge investor demand in different asset markets, and crucially, what that means for forward-looking implications for asset returns. This session will be no different in that context, but we do do something differently in this context for looking at the evolution of FX demand based on different demand drivers. The goal here for this research was to look underneath the hood. We know FX flow demand can move markets for currencies, but we recognise that investors have differing needs for currency, and they can be driven by different motivations. So what we'll explore during the course of the next 20 minutes or so will be to understand what are the different currency drivers that we see in our data, and importantly, what do they mean from the perspective of enhancing investor predictions around currency flows and currency returns? So just to set out the plans for what we'll cover today, we'll start off just with the theoretical motivation behind this research, where we take this more granular approach to quantifying investor currency demand. We'll explore through also a couple of polling questions to the audience, the degree to which we see differences in persistence and price impact based on the fact that there are different underlying currency demand drivers by investors. Then we'll go straight into some important implications with respect to how investors are thinking about hedging currency risk, and indeed, how these different dimensions of FX return drivers can impact models that are aiming at explaining and predicting both asset flows and returns across stocks, and bonds, and currencies. Let's start with the theoretical motivation. Just a brief recap for those that aren't familiar, when we think about institutional demand, we're really looking at this unique vantage point State Street has as a global custodian servicing $40 trillion in assets. Each day we measure in aggregate, how much demand global investors have for a particular market through their buying and selling activity in aggregate. Now, we know from our prior research that these investors trade large portfolios and typically exhibit a high degree of persistence in their behaviour. Michael mentioned a couple of examples of this earlier on in the day, but in addition, they tend to move prices in price impact terms. When they trade on average, they consume liquidity and therefore price can move. Now these two attributes are important because taken together, our historical transactions based on underlying cash flows tell us something about future asset prices and market impact. Now, in the context of today's session, we explore different demand drivers for FX, which are more risk-motivated than purely speculatively motivated, and we understand and decompose where we see these characteristics around persistence and price impact differ, to give our clients a bit of a toolkit around how to use some of these more granular metrics. So this slide really outlines the fact that we know investors can necessitate the need for FX for different reasons. Three of those we motivate on the slide here. The first being that investors are global investors. They purchase stocks and bonds in all sorts of currencies and that necessitates by construction a demand for the FX. We capture this in our cross-border asset-flow indicator. So some of you will recall we publish daily measures of cross-border equity flows and cross-border bond flows, which represent this demand by investors to invest overseas. The second of those is much more risk-driven based on the portfolios investors hold. We all hold global portfolios, and they inherently embed a lot of currency risk for investors. We can decompose and explore the degree to which investors are hedging those currency exposures, based on the fact that those underlying international assets are observing volatility in price and when those price changes, investors wish to maintain a desired level of FX hedging. And of course, there are pure speculative FX motivations. If a currency investor believes an FX rate is undervalued, or has some other hypothesis, there is also, of course, demand driven by FX speculation. So for today's session, we are focusing mostly on the former two of these demand drivers. We are generally ignoring the speculative-driven investor motive here, which of course are reflected in our aggregate measures of FX flow and are focusing on these more risk-driven basis for currency demand. So just to set the scene a bit, we'll just briefly highlight some definitions of some of the key metrics we look at to answer this question. First, investors naturally can gauge a view on currencies indirectly through their underlying hedging activity. So the first metric we look at you'll see on the slide here is our hedge ratio measure. This we estimate looking at the positions that our investors are holding for funds that have both currency and non-domestic asset underlying exposure. You can see that we actually look at hedging activity from two different perspectives. The first which we call foreign hedge ratios which is for any given currency pair, to what extent do investors express risk concern over that exposure in terms of hedging? So if we're looking at, say, the Australian dollar, how much of that Australian dollar exposure is hedged relative to the overall equity and fixed income Australian asset exposure that is held by investors globally outside of Australia? A very simple example would be a UK-based investor holding ¥100 million worth of Nikkei exposure. If they hedged that by a short position of 50 million Japanese yen, they would have hedged half of that Japanese yen exposure. But for a smaller subset of currencies, we also publish a domestic hedge ratio, where we look at really investors' aggregate appetite to hedge currency risk in aggregate. So here the equivalent example would be a UK-based investor holds a billion sterling's worth of assets denominated in euros, dollars, and many other currencies. If they had collectively held half-a-billion long sterling position against those currencies, that in turn would represent them hedging half of their overall currency risk. For this context, both of these represent the degree to which investors are hedging their overall currency exposure and represent demand for FX from a risk-management perspective. The second set of metrics we look at relate to capital flows. Really, you'll see here we're looking at this across assets, but also in a more granular level of detail when it comes to FX. So to start with we look at cross-border asset flows, investments into stocks and bonds overseas. This of course lines up to our first underlying FX return driver being that investors need to consider hedging when buying non-domestic assets. We also then start to decompose the underlying FX flows we observe. First, we differentiate between foreign investors and domestic investors. A UK investor for example buying sterling and selling the yen would represent a foreign sale of the Japanese yen, but in domestic context, a UK-based portfolio holding multiple currency exposures, when they transact in sterling against those currencies that would represent domestic transactions in the currencies. Second, we then differentiate based on the degree to which portfolios are driven by more equity features or fixed income features. Actually, in our dataset, and many of the indicators we make available, we hone in on what we consider to be these true multi-asset funds which hold both currency and other asset exposures. Here we can understand the degree to which equity portfolios, or fixed income portfolios specifically, are demonstrating hedging in the form of FX flows in the context of their underlying asset portfolio. So, for example, a UK-based international equity portfolio transacting in FX, we can capture as an equity FX hedging flow. So the point for this decomposition really is to answer the question, can investors benefit by differentiating on the need for FX based on the fact that investors use currency for different purposes? Bringing these differentiated measures back to the drivers of demand, this talk really follows on from a talk, for those of you that were here last year, where academic Robin Greenwood and one of my colleagues, Alex, talked about some research around recent hedging trends they'd seen. They uncovered some key themes around general levels of hedging activity by global investors. First, that investors typically hedge less equity risk than they do fixed income risk. No surprise, given the higher volatility contribution currencies bring in a fixed-income setting than they do in an equity setting. Two, we've seen notable trend evidence of increased hedging activity through time, particularly in the equity space. So we're seeing the gap closed by investors hedging more in general. Crucially, for this study, third, we found evidence that investors tend to stick to target hedge ratios and if those hedge ratios start to deviate meaningfully, we see evidence of investors rebalancing back in their currency positioning to maintain a target degree of hedging. Now for this study, this has implications in the sense that we should really care about the cross-correlations between FX features and other asset features, because this can give us a strong gauge on the degree of FX hedging demand, and we'll show you this shortly. Now, following on from last year's piece, the goal here is really to look at the degree to which we see differences in characteristics and implications of these different FX hedging, and different FX demand drivers for the use cases we described earlier in terms of supporting international asset transactions and hedging currency risk in general. So let's start with some very simple examples around how we would think about how these different FX demand drivers come together. You could think of aggregate FX demand being a composite of three underlying motivations. First, the fact that investors are purchasing assets, which necessitates the need for currency. That would result in a long currency demand, which is then offset by the degree to which they hedge that at the time of that transaction. So we subtract from that FX forward hedging, which you can represent across either equity or fixed income dimensions. Then on top of that, there will be, of course, direct FX speculation that can add to that demand. The simple example would be on the bottom of the slide there. If we're looking at Eurodollar investors, Eurodollar at parity, one investor hedges half their currency, one fully hedges. That leads to a net demand, in this context for euros, that we would consider to be an overall demand for FX based on pure hedging activity. Now, what we do next is first look at the degree of persistence and price impact of these different FX demand drivers and then, crucially, bring this back to what this means for forecasting forward-looking FX returns. Before we jump into those results, we'll just walk through a few of the different control variables that we look at for these regressions. First, we control for some important price effects. So you'll see that we control for prior 12-month momentum and prior month return. In this context, we're looking at FX returns as well as underlying asset returns here in the context of equities and fixed income. We also consider a variety of different behavioural features, some of which we've already discussed. We look at flow momentum itself, as well as the trailing one-month flows, again across both cross-border equity flows, cross-border bond flows, and the underlying decomposed FX flows we've just talked about in terms of differentiating between domestic and foreign investors, and between the FX hedging flows by equity and fixed income portfolios. We also look at excess holdings, and in this context, we actually focus on excess holdings from an asset perspective. So when investors naturally get quite extended overweight or underweight, that can trigger a need to rebalance, but we don't focus in this context on FX holdings, on the basis that that represents a more aggregate measure of behaviour, which we're trying to really get more granular on in this study. And of course, we focus on hedge ratio. Hedge ratio is being very crucial there to understand the degree to which fixed-income and equity investors have appetite for currency demand. So all of the regressions we'll walk through later really represent some combination of these features, as you'll see later. Of course, as much of you will have seen in prior research, we typically undertake similar transformations across these variables. So flows we're looking at on a trailing one-month window to really smooth the trend in capital transactions we're seeing. In all of our cross-sectional tests, we're constructing Z-scores so that we can compare like for like across our currency universe. So with the details out of the way, I'd like to jump to an illustration, which is actually one that's agnostic to the return features to serve as a benchmark, which is FX return reversal. Most of us are aware of this phenomenon paving out in currency returns, but by way of illustration, this example is showing the correlation between FX returns this month and FX returns in the prior month, for G10 currencies. The left-hand chart there shows you what that correlation has looked like over the full sample in the black bars and post-GFC, essentially from 2010 onwards in the blue bars. So quite compellingly negative returns, indeed evidence of short-term reversal effects in currency markets, meaning that recent winners tend to underperform in the subsequent month. We see that on the right-hand side, equally prevalent in our cross-sectional analysis, with a similar degree of magnitude. Now, this serves as a benchmark. This tells us that trailing return dynamics can tell us something about future currency performance, in this case in an inverse way, which we then look to build upon when we layer in these incremental demand drivers for FX. So this is a punchline preview, we'll come back to this later on the deck. Really the underlying motivation here is that it pays to pay attention to investor behaviour in the sense that predicting currency returns, flows, and lots of these indirect hedging demands can matter quite significantly. It also pays to capture these differentiations in different FX return drivers. So what you'll see on the slide here, there's quite a lot on this chart, is we're plotting the adjusted R-squared on average across G10 currencies when we're trying to forecast the next month's currency return. So for any given exchange rate we're predicting the next month's return based on this month's return features and behavioural features, and we analyse what the average adjusted R-squared is across these metrics. You'll see that we've broken this out into a collection of equity features, fixed income features, and a combination of the two. There's some very interesting dynamics that we see here. First, if you look at the dark blue bars there, we find that the behavioural metrics which you might remember were flow momentum, trailing one-month flow, as well as excess positioning, and hedging activity by these underlying different agents, can tell us a lot about the next month FX return. So can price. If you look at the bright blue bars there, that represents really our short-term reversal factor as well as price momentum. They are also incrementally useful in forecasting FX returns. Quite neatly, if you were to stack the dark blue bar on top of the light blue bar, you end up with something approximately equal to the green, meaning that both behavioural and price-based features are quite additive in thinking about forecasting forward-looking FX performance. Then the second summary I'd make here is if you were to look from left to right as we look at the equity features, the fixed income features, and on the right-hand side, when we bring those all together, again you'll see forecasting of FX returns becomes more strong when you consider the differentiation again in FX demand drivers between equity and fixed income investors. So we'll come back to some of these efficacy considerations, but now is the time to bring up a couple of opportunities for a poll for those of you in the audience. So a great time to please get your phone out. We'd like to explore the degree to which we see differentiation in the persistence and price impact of these different FX demand drivers by investors. So there'll be two back-to-back polls. The first example you'll see now on the slide here. It would be great to get your thoughts on which of these different FX demand drivers in the past have displayed greatest persistence. Option A will be essentially the investor that's buying stocks and bonds internationally and is choosing whether or not to hedge that transaction. That's FX demand to purchase or sell international securities. Option B you then have the FX demand driven by the necessity to hedge underlying price moves in international holdings. So you've got stocks and bonds from non-domestic markets in your portfolio and you're deciding whether to hedge that or not. Option C, if you think they're approximately the same. This is all about which of these effects do we believe may be more persistent through time. Already got a good number on the poll here, quite evenly split interestingly. I'll just give it another moment just to see if we swing one way or the other and then I'll give you some empirical evidence on this. Okay. Great. Interesting. Okay, so the borderline majority vote we have here is around the hedging flows. Actually, for those of you that stuck with the cross-border asset flows, those that are hedging those transactions were right. You'll see here we're showing on the left-hand side the autocorrelation in terms of one-month persistent metrics for flows for equity investors, on the right-hand side for fixed-income investors. The two different bars here in the light blue show you the persistence of cross-border asset flows. We see much greater persistence there, in the region of 30 per cent to 40 per cent, than we do for hedging flows, where we see much weaker persistence. You can see, in fact, those are much closer to zero in those dark blue bars. So great, thank you for that. Keep your polls ready because we have a second poll to pull up here as well, which is to look now at the price impact question. So again thinking back to those same underlying FX demand drivers. Same factors, but now we're concerned about the price impact for the currency. So which of these demand drivers do we think impact FX exchange rate pricing more? A) when investors are purchasing international assets. B) when they're trading FX to hedge the underlying assets. C) approximately the same. Okay. Much clearer view, I think, in the room on this one. Certainly, an overwhelming majority leaning towards the direct FX trading for hedges. Great. Thank you. Excellent responses. So we'll dive straight into the price impact slide here. The majority are definitely right on this one. So you'll look here at the correlation between flows this month with FX returns this month. Here we're looking on the left-hand side at those underlying FX hedging flows by equity and fixed-income portfolios. We can see those in the region of 10 per cent to 15 per cent, much stronger in terms of FX price impact than the underlying asset flows that one would expect to see on the right-hand side, where we do indeed see price impact, but of a much weaker magnitude. So in short, we see differences in persistence, differences in price impact based on the fact that investors require FX needs for different motivations and with varying degrees of urgency. So let's move into what drives hedge demand reflected by these cross-correlations. Here we're now looking at really three different dimensions to assess evidence of investor hedging. The first looking at the first driver of FX demand, which is when we're buying foreign assets and have a necessity to hedge at least some of that currency exposure. So here we're looking at the correlation of FX flows this month with cross-border asset flows this month. Now if we assume that we as perhaps UK investors are buying Japanese equities, if we were to sell the yen forward to hedge some of that exchange rate risk, we would expect to see a negative correlation between our asset flow and our FX flow. Indeed, these results show you essentially a panel regression result and a cross-sectional result across all currencies. We do indeed see strong evidence of negative correlations here and evidence that investors are hedging their underlying international asset purchases and sales. The clear trends you'll see here. On the left-hand side, we're looking at this across a full sample starting from the late '90s on the right-hand side since 2020. There is a distinction here between the evidence of hedging between fixed income and equity investors. For example, these lighter blue bars, we see much more pronounced negative correlations here, indicating a greater propensity for fixed-income investors to hedge those foreign asset purchases than their corresponding equity counterparts. It's interesting to note that on the right-hand side of that chart, since 2010 that gap has closed, as mentioned at the beginning, and we're seeing much greater evidence of equity investors too, choosing to hedge currency risk when they transact in international stocks. Next, we look at the degree to which investors are hedging the underlying currency risk when their underlying portfolio prices are drifting. So let's assume now, as a UK investor, we hold some exposure to Australian equities. When the ASX index is moving up or down, for us to maintain our target hedge, we have to transact in the Australian dollar to maintain our desired level of currency risk. Now, if that's the case, we would typically expect again to see negative correlations between the FX flows this month, which are essentially our assumption of FX hedging activity, with the underlying local asset returns this month. Again, we do see broad-based evidence of those correlations being quite strong and negative, and of similar magnitude to what we saw on the prior slide. Here again, you'll see much more similarity really, between both the full sample and more recent period, but again, evidence that in this case, equity investors typically are demanding much more liquidity when they transact in FX, on the basis that equity markets are, of course, expected to be more volatile than fixed income markets. We motivate that the maintenance of hedging is therefore a bigger demand driver for equity investors and a bigger consumer of liquidity in the context of this correlation. Then third, we'll look at hedge ratios. Now if investors are indeed targeting their hedge ratios on average, we should expect them to see rebalancing evidence. So imagine a scenario where investors will perhaps over-hedge. The hedge ratio is higher than it would ordinarily want to be. That means they've got a position in FX that is too short. So for them to rebalance they would need to buy the FX to bring that hedge ratio back in line. That would imply a positive correlation between hedge ratios and subsequent moves in FX measured by flows. Lo and behold, again here the correlations are consistent with that theory in the sense that we see strong correlations between target hedge ratios by investors and the same month flows in FX. So this gives us again strong conviction that investors are trading in concert with target hedging, and we do see evidence of rebalancing back to target levels. Now really all of the results so far have been focused on differentiation in the needs for investor demand for currency and the impact that has on persistence, price impact, and rebalancing. Now, the ultimate goal is to then think about the role these different FX demand drivers have in explaining and forecasting flows and returns for currencies. So this slide will typically follow a very similar format to the teaser slide we started off with. Here we're looking at explaining G10 currency returns. We're showing you the average adjusted R-squared for each of these currency regressions and you'll see on the left-hand side we're showing this for the entire sample. On the right-hand side for the more recent period. So in this context, we're explaining four different types of asset flows. Cross-border equity flows, cross-border fixed income flows, and the underlying FX hedging flows by equity and fixed income portfolios. There are some important differences here from this sort of analysis. First of all, asset flows can be explained far more significantly by this collection of features that we described at the earlier section of this presentation. These price-based features and these behavioural features can explain more about cross-border asset flows than the underlying FX hedging flows itself. We can also see if we look at the breakdown between the behavioural factors in black, the returns features in blue, and the collective features in green, that behavioural features also play a greater role than return features in explaining the underlying asset flows we've observed. But we do see again additive features in the sense that both price-based and behavioural features are both important. When we think about this in the context of explaining returns, we see something quite similar in the sense that if we look at the features that are more equity-focused versus the features that are more fixed-income focused, both help us explain, in this case, currency returns in the same month contemporaneously speaking. When we combine these equity and fixed income features together on the right-hand side of this, again, much more of those FX returns are explained by the fact that equity and fixed income investors have differing FX demand needs. In this context, when we're explaining returns though, the price-based features do matter more than the behavioural features at explaining FX returns. Now, crucially focusing on our forward-looking regressions, we now care about predicting flows and predicting returns from a currency context. When it comes to forecasting flows, again we're looking at these across the cross-border asset and FX dimensions that we saw previously. You'll see some similarities in the finding here, if we're looking on the left-hand side chart for example, when trying to forecast forward-looking cross-border equity and cross-border fixed income flows, we find both price-based features and behavioural features to both be important predictors. But actually, we do again find the behavioural features looking at things like flow momentum, positioning, hedge ratios, and the underlying demand for hedging by equity and fixed income investors to be more important than things like short-term reversal and price momentum for example. If we look at the third and fourth blocks on that chart where we're looking at forecasting FX returns through the use of these hedging portfolios, we see the importance of the return features to be there more relatively important. Those lighter blue bars there are a bigger chunk of the green if you like. We do find again additivity between the behavioural and price-based elements of forecasting next month's flow. When it comes to predicting returns, there are a couple of nuances here and this slide really does highlight the fact that we see this very strong short-term reversal effect in FX. Here we're looking at the correlation between flows in one month and flows the subsequent month, essentially coming back to some of that context we talked about at the beginning. The idea here is that there's an understandable difference in expectation between FX correlations and underlying asset correlations. For example, if we think about the fact that equity returns and FX returns are typically positively correlated, and investors are targeting hedging of that FX exposure, if the equity market rallies, they have a bigger position, they'll have a bigger FX exposure to hedge, which means they need to sell the currency, but they're selling a currency that's rising in price. So from an equity standpoint, we would reasonably expect to see, once short-term reversal effects have passed, a negative correlation between the FX hedging flow and the FX return. On the flip side, for fixed income, we'd expect to see the reverse because correlations between FX returns, and fixed-income returns are typically negative. So in that context when it comes to hedging flow, we would be selling fixed income when it's falling. Indeed if you look on the right-hand side there, we do see that differentiation between asset hedging in the context of these portfolios and the underlying FX returns. Now going back to how this all comes together. This was the slide we started with. The key point being here that when we try to forecast next month's FX returns, underlying FX return drivers by equity and fixed income features are both additive, they're both important. Behavioural and price features are also both irrespectively important and additive. The key difference being here though, you might recall from the earlier slide, when we looked at explaining FX returns, we found that return features mattered a lot more than behavioural features. In the context of forecasting FX returns, we see the price-based features still to be important, but less so than many of our measures around these different demand drivers from underlying hedging activity, for example. So all of these examples so far have really been on a pairwise basis across G10 against the dollar. The question we also wanted to ask was to what extent does this differ when we look at the Dollar Index or the DXY? So I'll make this the last slide that we describe here, where again the goal is to predict the DXY next month based on return features and behavioural features over the previous month. You'll see that actually there's similarities here, but two key perhaps differences. The first one being that behavioural features and return features both matter, as we saw in the previous examples on average, but certainly the elements around fixed-income portfolio hedging have mattered more in the past than what we've seen from equity investors. So I'll just wrap up with some key takeaways. First, we've laid out a framework for thinking about how to decompose FX demand by investors. Many of us in the room will be familiar with our flagship aggregate measures of FX flow, and the demand that that captures. The goal for this research was really to break that down into its components and recognise the fact that FX demand is driven by different needs with varying degrees of urgency, such as hedging foreign asset purchases and hedging the price moves of underlying portfolio holdings. We differentiate in our interpretation around the degree to which equity and bond funds necessitate need for FX, as well as the differences between domestic and foreign investors. We found evidence that fixed-income portfolios have had a greater tendency to maintain hedge ratios when buying foreign assets, whereas equity hedges have on average demanded greater liquidity when hedging their underlying portfolio price changes. Crucially, considering these different sources of FX demand together, we found evidence that forecasting FX returns can be improved by looking beneath the hood and capturing these different dimensions of FX demand and are additive to price-based factors. One of the things I hadn't focussed on too much in the earlier slides in the interest of time, many of our studies we've looked at on a currency-by-currency basis. We have found benefit also in tailoring approach for a particular currency pair over a panel approach, where we look at all currencies equally as well. So when thinking about forecasting currency returns one month ahead, it can pay to tailor when thinking about these characteristics. I'll stop there. I'm very happy to take any questions and thank you all for listening.

We have a bit of time, so we might as well sit down. So yes, we've got about four minutes I think, just before we head off to Milan to hear from Alberto. Fascinating talk. A couple of questions. So first one is of all the different bits of analysis you did, was there anything that came up that wasn't intuitive or was surprising to you?

Yes, good question. I think most of this lined up with our economic intuition and followed on from some of the studies from last year. I think some of the more surprising elements were perhaps the speed at which we've seen increased hedging demand from equity investors, particularly the price impact slide. We've done lots of research recently on currency liquidity, and it was perhaps a surprise to see how strong that effect has balanced. In the past, that was quite heavily dominated by fixed-income investors. Today, that's much more balanced. Equity investors are certainly showing much stronger evidence of rebalancing and hedging to target hedge ratios. I think that's probably the most key surprise that we found. Most of these we found lined up with trends that we've seen over multiple decades of hedging activity but that was one that certainly struck out.

Yes, because I mean, historically that is surprising. No offence to any equity managers in the room just to clear that, but I think we would have assumed that there would have been more information content in the change in hedge ratios of fixed-income investors who were managing the currency more directly.

So that's an interesting point. So there are differentiations to be had there. So first of all, we still, just to be clear, do find stronger evidence of fixed-income investors hedging their underlying portfolios. Hedge ratios are still considerably higher. If you look at the hedge ratio indicators on Insights, you'll see that those are much closer to, say, 100 per cent fully hedged positions, whereas equity hedging is rising, but it's maybe 20 per cent to 30 per cent depending on the currency. So it's certainly still a big gap. But in terms of their liquidity consumption, when they're then doing these hedging trades, we see that impact being much greater for the equity portfolio given the higher volatility of the asset.

Well, I mean, fixed income has been quite volatile this year as well but…

True.

Just a related question, have you done or do you plan to do similar work in EM, and would you expect any different behaviours?

Good question. Yes. We haven't yet expanded this to EM. That's definitely a further extension. We generally would expect to see I think, some differences on the basis that hedging there is obviously done differently. It's certainly done to a weaker extent. There are considerations around proxy hedging and other factors. We would expect some of the characteristics around, say, the price impact of those hedging trades to probably be greater, if you think about the correlation between the underlying hedging flow and price impact. We'd typically find stronger evidence of persistence in flow in EM too. So that could be another interesting effect of study. So we'd probably expect to see a longer lead time there on the autocorrelation between the hedging flow and subsequent flows and returns, but definitely an area for further study, yes.

Then there's a couple of questions that relate to regime stuff, which would be over time, when might we expect to see different FX demand drivers impact markets? Then specifically, do you think the rate environment would change that in the sense that we've just gone from a zero rate environment to a non-zero rate environment? Do we think that might have an impact going forward?

Yes, a good question. So this study is obviously done on a pretty lengthy period of time, over 20 or so years, and we did deliberately, quite intentionally focus on the post-GFC era, given the change in hedging costs since then. As you saw from many of those slides, the results were unequivocally very similar. So actually we didn't see a high degree of sensitivity between those two time periods. I think general regime shifts where we would expect this to differ would comprise of probably two things. One is we do generally find that institutional investors on average do demand liquidity through time, but there are times where that's weaker, and that can be times in general where other agents in the market are more heavy consumers of liquidity. In the context of hedging specifically, one or two examples could be, for example, if we enter periods where markets are characterised by particularly high volatility, we might expect to see bigger hedge adjustments there on the adjustment to the underlying portfolio hedges, for example. If investors are holding foreign assets and the volatility of the market is greater, we'd expect to see that effect perhaps come through even more strongly relative to the hedge of foreign asset purchases. I think both of those underlying components together could be more prominent also, when we're in a scenario where perhaps market uncertainty is greater, and there could be an overall higher degree of hedging by investors globally. So I think there could be an element of the market environment from a risk perspective as well as from an uncertainty standpoint. Yes, the rate environment would likely have a factor. That's going to pave into transaction costs and the opportunity cost for hedging. Given the differences we've seen in some of the characteristics between the fixed income and equity portfolios, we might expect to see that gap change also on the basis that the rate environment is perhaps quite different.

Okay, perfect. I would imagine, so everyone currently sees the aggregate flows on Insights, but this work will be the basis for, I'm sure, a little bit more detail, and it just highlights the granularity that we can go into on this. So please join me in thanking Neill for his talk. Thank you very much.

Thank you. Cheers.