Episode Transcript
Michael Brandt: I wanted to take a micro approach and start from the bottom thinking about changes and beliefs as being an accumulation of daily information. And that's daily information that may be public, such as public economic news, or it might be just observations of markets and market activity.
Indrani De: Hello everyone, I'm Indrani De, Head of Global Investment Research at FTSE Russell. And today I'm delighted to be joined by Michael Brandt. He's a professor at Duke University Fuqua School of Business. Welcome Michael, I’m so glad that you're joining us in this session. Why don't we start off by talking about your career journey, what brought you to this moment, and your key research interests at this point?
Michael Brandt: Thank you for having me. Yes, I started my career back in the late nineties on strategic and tactical asset allocation problems. And my work had been very methodological, looking at techniques for circumventing some of the issues that we had documented with mean variance optimisers. So realising that if you have noisy inputs, you end up with even more noisy portfolio weights that then need to be constrained and perhaps shrunk to some extent.
So over the years, as I was working with investors on their processes, my research interest shifted from trying to help with methodology to trying to understand what kind of information leads an investor to wanting to shift their strategic or tactical asset allocation.
And rather than some of my colleagues who take sort of a top-down approach, trying to think about how portfolio weights and returns might vary from one quarter to the next or one month to the next, I wanted to take a micro approach and start from the bottom thinking about changes and beliefs as being an accumulation of daily information. And that's daily information that may be public, such as public economic news, or it might be just observations of markets and market activity.
And so that led me to two distinct research agendas. In one, I'm looking at nowcasting, where I'm collecting the daily flow of information about the macroeconomy and construct point-in-time forecasts of where we are in terms of output, inflation, labour, and macro-sentiment.
On the other hand, I've been studying the relationship between order flow imbalances and specifically inflows and outflows into broad asset markets and the impact of those on returns and thinking about whether or not that we're really looking at more of a microstructure problem or a much broader phenomenon where inflows into asset classes actually have large significant price impacts.
Finally, I'm combining the two. So, I've been working on trying to think about how does the information about the economy evolve and can we simultaneously observe within asset markets rebalancings that reflect that change in information. So the classic example might be something like a portfolio rebalance around a sector rotation strategy, whereas the information shifts from being a strong economy with strong growth to being perhaps weaker growth. Investors shift their portfolios from more aggressive stances to lower beta, more defensive sectors. And that's exactly what we see in the data, that order flow imbalances reflect that sort of rebalancing activity.
Indrani De: That's fascinating. So, you know, given the current market environment, people are obviously very concerned about financial market imbalances, what's leading to those imbalances and impact, and you have done so much research on this. So let's talk about some key imbalances and how you see them really impacting markets. Where should people be really concerned to?
Michael Brandt: From the basics of trying to think about what's different about an asset class versus a single stock. So in a single stock, we understand that if investors are buying or selling the stock, that activity may embed information. And so therefore prices move for a reason. The price, that information is being incorporated into the prices.
I'm thinking more about asset classes, and I think of those as more public information sort of assets, that most of us can see all the information that's relevant for pricing and asset broadly, what the interest rate might be, what the currency value might be. It's hard to think about individuals as having private information about broad asset classes. And so what I've been studying is the role of order flow imbalances in these broad asset classes. And what we find is that it is just as strong as when you look at individual stocks.
So that if you have large order flow imbalances, large extra capital flowing into a given asset class, we start seeing the market capitalisation of those asset classes appreciating by orders of maybe a factor of, let's say, eight, which means that if I have hundred million dollars of capital flowing into the S&P index, the S&P index appreciates by eight hundred million dollars in market capitalisation. That's one of the empirical findings from the paper I will present tomorrow.
Indrani De: That's very interesting. So talking about the time frame, like, do all these things have an impact immediately? Is it a medium term? Is it long term? And has that changed now that nowcasting is becoming much more common? Many of the Federal Reserve Banks are doing it. So because even if people don't have the actual readings but they have a nowcast reading, does the time horizon of the impact change?
Michael Brandt: What you would expect to observe from the microstructure view that we sort of think about from the single stock is that these effects are first short-lived, strong at a very short horizon. So at the immediate trade horizon, when the trader comes to the market maker, that's when we expect to see prices move the most. And then dissipate, or maybe even revert because in a classical market microstructure model, its short-term mean reversion and that is the compensation for providing liquidity.
We see exactly the opposite in these broad macro markets. The effect gets stronger the longer the horizon is. So we accumulate order flow imbalances not just over hours, but over days, weeks, and ultimately months. And the correlation between order flow imbalances and contemporaneous returns becomes stronger the longer the horizon is that we're looking at. And the effect is permanent. So even if you look out as far as a couple of quarters, it does not mean revert. So it's really just an inflow of capital that leads to a permanent change in capitalisation of an asset class.
One of the contributions of our paper is that we look very broadly across different asset classes. So equity indices, bonds, and rate futures. We looked at commodities, currencies, and even looked at VIX and Bitcoin. And regardless of which asset class you look at, the effect is very similar.
To this question of timing, what we observe is not so much of a time trend, but rather cyclical behaviour. And in fact, there is a very common structure across all these different asset classes to the cyclical behaviour of how sensitive markets are to inflows and outflows. There are years in our historical data where the sensitivity is very low and other years where markets have become very sensitive. But there's a strong common component. Roughly 60% of covariation is explained by just one principal component.
So there's something going on. We don't quite know what it is yet and what's driving it. But it will be important to understand going forward.
Indrani De: It's super interesting that the impact at the asset class level is actually stronger at the longer horizon. That's fascinating. So let's tie back to one of our indices in the Russell family for the US capitalisation. So the Russell 1000 for the large cap, Russell 2000 for the small cap. Any insights on financial market imbalances that might have had a permanent or a long-term impact on the performance of these two indices in recent years?
Michael Brandt: It’s sort of a question of volatility, and market cap, and ultimately the volatility of order form balances that you see from day to day. And there are some markets that we look at that are very balanced over time and then others that see very systematic inflows and outflows, particularly over courses of quarters and business cycles. So there's sort of two pieces to the order flow imbalance story. One is seasonal rebalancing. So this is kind of the classic story of an investor who holds a 60/40 benchmark portfolio. And every quarter, every year, every so often has to rebalance back to policy weights because relative performance has sort of driven away from equilibrium. That generates very predictable order flow imbalances at certain calendar points.
The second is more seasonal and related to business cycles. And that very much would affect your small caps much differently than it would just a broad-based large cap index, where you might see very systematic inflows and outflows at different turning points of what I think of as people changing their minds about where we're heading in the near term, or the medium term in terms of the economy.
Indrani De: What implications does it have for the optimal frequency and timing for portfolio construction, portfolio rebalancing?
Michael Brandt: Part of it is avoiding the big flows. So if I am rebalancing a portfolio and I'm just balancing back to my policy weights, I probably don't want to do it at the same time as I might know the aggregate industry does. I might want to stagger it through a longer period as opposed to waiting for my ultimate decision period at the end of a quarter or end of a year.
So just being aware of the fact that there is a broader set of market participants who might follow very similar strategies. In this case a very passive uninformed strategy. We're just rebalancing the market caps, that yet has an important impact on prices.
The other is trying to understand transaction costs because this is ultimately a contributor to a transaction cost. And so knowing that whether the markets that you're trading in are currently in a state in which an inflow and an outflow has a more significant impact than other times is going to be important to modelling transaction costs over time.
Indrani De: Absolutely. So taking a slightly longer view, all these issues that you talked about, would you say for asset allocation, from the asset allocation perspective, are they more important in the tactical asset allocation or in the strategic asset allocation space?
Michael Brandt: I think they're important for both. They're important for the tactical asset allocators because transaction cost is critical to harvesting returns. And they're important to the strategic asset allocators because they're exactly the type of investors that are generating that seasonal, sort of almost predictable rebalancing.
You just watch the financial news and inevitably at the end of every quarter you'll see news stories come up where somebody predicts how much an asset class is going to move because there's going to be a predictable inflow outflow from a particular type of investor. And pension plans rebalancing to their strategic benchmarks is a very common example of a manager that might do that.
Indrani De: Can you touch upon the issue of correlated strategies and where capital flows are happening right now and what does it tie into these imbalances?
Michael Brandt: Yeah. The thing about transaction costs that I've always found fascinating is that when we model the transaction costs, we think about it as the cost of rebalancing my portfolio. But if my portfolio behaves very similarly to other managers' portfolios at the same time, which can easily happen because we're processing the same signals. Really I need to start thinking more about the portfolio, the price impact of that broad set of transactions which all are going in the same direction. And so correlated strategies, understanding how correlated my strategy is with the broader strategies is very important because it tells me how is my flow going to be amplified into a much broader flow that then has a much larger price impact, much larger than what my individual transaction cost model might suggest.
So again, going back to kind of the popular financial news, you might see very frequently stories about trend followers, for example, getting to a point where they're flipping their portfolios from being long equities to being short equities, thereby generating a very large flow. It's not that any of the managers is going to produce a very large flow, but as a group using highly correlated signals, their transactions end up generating a very large flow and thereby contemporaneous price impact.
Indrani De: You raised a very interesting point about being aware of how much one's own strategy is correlated with others in the market, but any deep-in-the-weeds information on how to best get that estimate of how correlated somebody's strategy is to everything else happening in the market?
Michael Brandt: I don't have one. But I have a reverse argument, which is that if you see that your trades have price impacts, which are significantly larger than what your price impact model might predict, it most likely is going to be the outcome of correlated strategies trading at the same time as you're trading as well.
One way that you might see that happening in real time is you have a trade that you start at the beginning of the day and as you to continue to execute it. Use a VWAP strategy, for example. The price just keeps moving against you all day long.
Most likely it's not a market maker detecting that you're in there trading as much as everybody else in the market trading the same correlated signals in the same direction, and thereby creating a market impact that is much larger than what you would anticipate at ex ante your market impact to be.
Indrani De: That was super insightful, Michael. Thank you so much for sharing your insights with us.
Michael Brandt: Thank you very much.
