Good morning, I am delighted to be with you today. Thank you, Megan, for the introduction. Indeed what I want to talk to you about is how we can use information from retail investor trading and cryptocurrency to understand more about asset pricing models for crypto. Before I show you some of the more detailed analysis that we have done with retail trader data, I want to take a step back to just remind us: typically in asset pricing how do we think about the value of an asset? It's typically the expected present value of cash flows; either cash flows that are currently accruing to a business or an institution. Here we of course know that cryptocurrencies like bitcoin, the tokens themselves do not accrue any cash flows. Then the two other dimensions of value are service flows that we are expecting to receive in the future, and generally also any change in beliefs about the future value of a particular ecosystem or in this case a system like cryptocurrencies. I want to walk you through these different dimensions of value for a cryptocurrency and end obviously with what can we learn about how investors update their beliefs about the value of cryptocurrencies from this retail trader data that I will show you. I first want to just quickly remind us about some of the debate that we have seen about the value and where the value of cryptocurrencies comes from. In particular if you remember very early on, people were very excited about the role of cryptocurrencies as a potential replacement of our traditional payment system such as credit card networks like Mastercard and Visa. There I would say the debate has been relatively quickly settled to show that this cannot be the value of cryptocurrencies. The reason being that two of the main functionalities that we typically expect from a payment system are just not satisfied in typical cryptocurrencies. What I mean with this is that especially in proof of work-type cryptocurrencies, the energy utilisation and the energy needs are just too high. For example, we have seen for bitcoin alone that currently bitcoin would place similarly to a medium-sized economy such as, say, Malaysia or Argentina in its energy use. If obviously this scales up further these energy costs will be very, very high. In addition, we have also seen that the functionality of the current slate of cryptocurrencies just doesn't allow for quick-enough settlement of transactions. So in bitcoin for example the typical settlement time is about ten minutes, but in peak time has gone as high as half-an-hour. This obviously is not a sustainable situation if you really wanted to use this as a payment system for quick and speedy transactions. Having settled this, if you want, and moved this driver of value out of the way, the debate currently really is about how much can cryptocurrencies like bitcoin for example be a store of value versus a purely speculative security? Here I want to show you a couple of findings that the literature has highlighted about the statistical properties of cryptocurrencies that at least at its current stage make it very unlikely that they will be a good vehicle for store value. In particular, let me first show you, we have seen in cryptocurrencies that even a cryptocurrency like bitcoin that had a remarkable increase in its valuation of course, but also in the depth of its market, the liquidity of the exchanges in which it is traded, and still we have seen two constants, if you want, that haven't changed. We see that even in the current stage there is very high and persistent price impact from order flow on cryptocurrency exchanges. I will show you a bit of deeper research on this in a second. We have also seen that despite the increase in valuation and the increase in trading volume on crypto exchanges, the volatility of a cryptocurrency like bitcoin and the same is true for Ripple and ETH, to just name some of the largest and most widely-traded ones, we have actually seen that the implied vol has stayed almost unchanged. The volatility is still as high as it was, say, five, six years ago. That's actually quite remarkable because typically when we see securities increasing in its volume and its depth of trading, you typically see this type of implied vol go down. I think this also speaks to the fact as I was saying before that because these type of cryptocurrencies currently don't have a true application or utilisation in society, a lot of its volume so driven by beliefs about its future adoption and future utilisation, where of course individual disagreement and heterogeneity in belief really very much matter. I want to also show you that it's not just vol that stayed high; as I said before, in a lot of research that I have done before, we have also seen that there is large and persistent price impact from order flows on cryptocurrencies. Here I just want to give you an example from bitcoin. What we did here is, using tick-level trading data across the twenty most liquid and, if you want, biggest exchanges we have looked at the impact of order flow on bitcoin price. So we extracted here the common component of bitcoin returns across these twenty most liquid exchanges and related it to the common component of order flow across these exchanges. What you see is that, say, even at a daily level when looking at the daily price impact, we see that we have about four per cent effect on daily returns from ten thousand bitcoins being bought. This is actually much bigger than what we see in, say, typical securities on US equity. Then what I am showing you here is that if you go to shorter time periods, say at the five-minute or hourly level, this price impact actually doubles. I can also tell you that the idiosyncratic price impact on individual exchanges is also quite high, but that's obviously less interesting from a trading perspective because the idiosyncratic vol might also be driven by other components or other factors on these different exchanges. This is just to tell you that as of now, despite the big growth in cryptocurrency trading and the increasing depth of these markets, we actually still see a lot of aspects and characteristics of cryptocurrencies that do not make them - that do not provide for very good basis as a store of value. I just want to also quickly remind you that we have also seen that some of the initial purported benefits of cryptocurrencies, in particular its lack of correlation with other asset classes - which originally was true for crypto returns - that correlation has actually recently increased. There was an interesting paper that some of my colleagues at State Street actually released this year that shows that if you look at the last couple of years, the correlation between crypto and general market and other asset classes has increased, especially has increased in a state-dependent way; meaning that when the market goes down, the correlation increases, while in times when there is no stress in the market, the correlation is still very low. Again this points to the fact that maybe some of the more beneficial properties of crypto returns were driven early on by the fact that we are still in a period of crypto adoption. This adoption obviously when more people allocate more of their wealth to cryptocurrencies, can induce price pressure and temporarily increase the price of cryptocurrencies, such as bitcoin. But that doesn't tell us much about the long-term and persistent statistical properties of crypto returns. This is in particular true because for at least the largest cryptocurrency bitcoin right now, there is no true economic use that would in the future explain, say, or allow it to be a good hedge against any of the important risks that we have in society. I wanted to show you this as a backdrop to then argue that given the situation we find ourselves currently in the development of large cryptocurrencies like bitcoin, actually a lot of the returns seem to be driven by the dynamics in which investors - in particular retail investors - update about their beliefs of the future value and utilisation of cryptocurrencies. In other words, as an economist we would say that really it is important to understand basically asset pricing models that are driven here by the type of investment strategies that retail investors adopt. In particular, say, Harrison-Kreps-type models of how bubbles form when investors have heterogeneous beliefs. With this, how can we get at these beliefs given that people don't run around with somehow their beliefs stamped on their forehead? What we need for this to approximate or even get an estimate of people's beliefs, we need basically granular data on how people in those markets trade. Not just tick and exchange-level data as I showed you before, because there I understand what the trades are, but I wouldn't know who was trading and how individual people update their holdings, change their positions as a function of what they observe in market prices. So this is why I have, together with some co-authors of mine at LSE and Annette Wharton, we have turned to data from a large international discount brokerage that has provided us with transaction-level data on their retail investor trading in cryptocurrencies, but also in other assets such as equities, commodities, fixed income. What is quite unique about this platform, it's an international platform that in particular has clients in the US and also in Europe. What is quite remarkable is that these retail investors as early as 2015 had access to a wide set of cryptocurrencies, for example and in particular bitcoin, ETH and Ripple, which are some of the most traded on this platform and also in general. But they provide actually even a larger set of cryptocurrencies, and I will show you in the analysis that we do now how people's trading behaviour and individuals' trading behaviour across these different assets varies as a function of past return. I finally want to say that for this analysis right now we have information from 2015 through to 2020 we are leaving out the last year-and-a-half of the pandemic because we have not had access to this updated data yet. Obviously over this time period, from 2015 on, we had several large run-ups in crypto prices, also in market prices and in particular in some of the largest cryptocurrencies we obviously have, as I told you before, seen a lot of volatility. So this gives us a lot of variation to understand how investors trade in this environment. Now, as I said, the market participants here are retail traders so this is not people's, say, 401(k) or retirement savings, and we can see that there is really a lot of heterogeneity in how people trade, just even the frequency with which they trade. We can see some people actually have typical day trader behaviour where they go in and out of a security on an intra-day basis, but even say a couple of days, Other people are actually holding securities for quite a long time. Then again what we see; everything from people who have $2,000 on this platform all the way to people who have tens of thousands. I think we want to think of this maybe as people's… Don't think of this as people's long-term savings accounts or retirement investment accounts, but more as their brokerage accounts in which they do more active trading. The other thing that is actually quite remarkable here that we see, and I just want to show you a few descriptive statistics to highlight this, is that people tend to focus at a given point in time on trading in just a few securities. So this is a snapshot about people's stock trading. I will show you crypto trading in a second. It's just to show you that even for equities and other asset classes, where obviously there is a wide slew of available securities and companies that you can invest in, we actually see that the typical investor focusses on a few assets. For example we see that in a given month, the average investor makes about seventy-nine trades, but as you see, at the ninetieth percentile and then if you go even higher, some people trade much, much more. The other thing however, as I am showing you here, is that the average person focusses on about twelve stocks in a given time period. But even here I can tell you that very often, people actually switch from investing in a couple of stocks and then shifting into a different stock that they invest in, and then coming sometimes back. Then if you look at crypto, the average investor here holds one point eight cryptos on average, or basically is trading in about one point eight. It was actually remarkable to us how skewed the trading is to a few securities, even say in equities. Here I am just showing you that on their equity portfolio, I am ranking the number of trades as a percentage of total trades on this platform. You can see that the largest twenty to maximum fifty trades absorb almost all the trading volume. Those are stocks, basically the type of very hyped but also, say, very prominent stocks all the way from, say, Amazon, Apple, Tesla, to some of the more - I don't want to say meme stock because as you saw, this time period is a pre-meme stock time period in stocks, but from Shopify for example are already in this data and are highly traded. The reason why I wanted to show you this is to make you aware of the fact that when I compare the trading of these retail investors on my platform to, say, how these types of investors actually are trading in cryptocurrencies, we can already see that even on the stock dimension, these people are not holding a diversified portfolio that is, if you want, as a finance professor I would say properly allocated for the long run. In some sense, you could've expected that the trading behaviour that retail traders show on the equity side should look relatively similar to how they update about trading on cryptocurrencies. Actually that is at all not the case, and so this is actually very interesting to me. What I mean with this, to give you the headline results first, we find that investors seem to follow a strong or let me say a contrarian strategy when they invest in stocks or even commodities. But they show very strong momentum strategy in crypto. There is a really sharp divide. I will also show you that this actually holds within a given person. This is not that particular people trade crypto and they have somehow different beliefs from any other investor who is trading in the retail space. What I want to argue - and I will show you now the data - is that actually even people who trade in stock and crypto show very different updating behaviour, it you want, about the future price dynamic of cryptos versus stock when trading in one versus the other. How can we show this? What I will show you now are some results from, as I said, this granular data that I have from this online discount broker. The objects that you have to keep in mind - and sorry to be a bit tedious but I do need to walk you through them - we want to understand how people change their allocations to stocks, to crypto, to these different asset classes as a function of the past observed returns. In particular, I want to understand how much investors seem to believe that, say, in crypto stock whether prices will follow a more momentum path, or a mean reverting path and, as a result, how then investors allocate their portfolio composition to these different assets. What I am calculating here - and that will be the portfolio allocation variable that is important for my analysis - is what I call the overall shared change. I want to understand: what is the share of somebody's wealth on this platform that is allocated to, say, either cryptocurrencies or individual stocks? The way we calculate this is that we look at the share that you own in a particular, say, cryptocurrency times the price of cost appreciation over that time period as a function of the person's wealth. Then look at the change of that object over different time periods. What you can already see here is that this overall share can be driven by two aspects or two dimensions. It is driven by just the passive price appreciation that happened over that time period, and then any active allocation that the individual made to, say, this particular cryptocurrency or this particular stock. This is obviously, if you think about it, the bottom line allocation because we would think that if somebody actively monitors his or her portfolio and sees the market prices going up or down, they should respond to the passive appreciation and adjust their active allocation in a way to ultimately end up with the allocation in crypto that they would like to be in, or in a given stock. To differentiate how much people's allocation to a given asset changes purely through passive appreciation, I will also show you how people's active rebalances look like in crypto versus assets. What I want to get at here is: some people, when the markets go up, might end up with a large allocation to a particular stock or a particular asset class, just because they don't pay attention and are not rebalancing actively. Then of course there might be people who are following very closely. Yes, when the market goes up or the return on a particular asset goes up, they actively pull market actually either into this asset class or out of this asset class. I want to show you both of these dimensions. As you will see, investors behave very differently in cryptocurrencies than they behave in regular stocks. One other thing I want to highlight here is that what we do is we form in the data cohorts by investor type and the starting months that somebody starts investing on this platform to smooth out some of the idiosyncrasies of individual people's behaviour. What I mean with this is that as I explained to you before, people often have these bursts of activity and then long periods of inactivity and then activity again and so on. This might be driven just by the fact that people get busy. They are not professional investors and so I don't want to, in the results that I am showing you today, include all these idiosyncrasies or heterogeneity that might be driven by things that have nothing to do with how people update about different asset classes, but more have to do with the fact that they get distracted and don't even look at their portfolio! Their lives become busy and these types of things. This is just to make sure that you understand how these data are created. I can tell you that even when we run these same results that I am showing you today, at the individual person level - meaning with millions and millions of individual trades, the results are very similar. Having said this, let me show you now how these results look. This is exactly the analysis I highlighted before. On my left, the dependent variable here that I am trying to explain first is the overall change in the share that retail investors on this platform are allocating either to cryptocurrencies or to stocks. On the left panel is cryptocurrencies and on the right I am looking at the same universe of investors. I show you their trading behaviour in stocks. What you see basically is on the left-hand side, where we look at crypto, this is: how do people's allocations to these different cryptocurrencies, in particular think about it. Again as I said, people here can hold many different ones, but the most traded ones indeed are the three big ones, at least over that time period; bitcoin, Ripple, ETH. What you see is that then on the right-hand side, going down, basically I am looking at how these allocations change as a function of contemporaneous returns, one-week lag, one-month lag, three-month lag, six months to really get a sense of how much investors believe that returns of these different assets either follow momentum strategies or if one should follow contrarian strategies. Here you see this, as I said, stark discrepancy so on the left what you see is that there is a strong and significant relationship especially of contemporaneous stock returns, with the allocation to cryptocurrency. You see the positive and very significant coefficient so basically, when prices are going up, investors are allocating… More of the portfolio of these investors is in cryptocurrencies. I am also in the second and third column breaking out for you here how this varies when returns go up versus returns go down. Interestingly, you see the momentum-ness of these crypto investors is there as much on the upside as on the downside, meaning they're riding out the momentum on the downside as much as when the price is going up. You see that a lot of this is actually in the same week so it's highly driven by contemporaneous returns. Now, look at the right-hand side where I show you the same analysis but now for the stocks that people hold, remember this is at the stock-by-stock level. This is why the number of observations here is much higher. But you see a very different picture. You see that here actually the coefficients are negative. There is contemporaneous rebalancing, meaning contrarian rebalancing, out of stocks when the price went up. There is also actually a little bit of lag behaviour especially if you look at the second column in the right-hand panel; you see that when the prices for a particular stock go up, there is actually contrarian rebalancing contemporaneously and also with a one-week lag. Now, to show you how this looks for the active rebalancing, remember this was the combination of prices going up and rebalancing happening, what you see is that actually in cryptocurrencies we don't see any attempt of contrarian rebalancing at all. People seem to just passively ride out the ups and the downs, while on the right in the stocks, you see actually that people really actively lean against, if you want, the price dynamic and therefore rebalance out of stocks when the prices have gone up in the past and into when they went down. As a last point, I want to show you - let me skip this; this is just robustness results - the last thing I want to show you is that what to me at least was extremely surprising is that when we found this result, I thought maybe this is due to the fact that different types of investors sought into crypto versus stock. People who are in a way more into a gambling mindset or just real enthusiasts into crypto are the ones that are behaving in such a momentum way. Then other investors are just in stock. That is just not the case. The way I can show you this is that I know in my dara people who are investing both in cryptos and in stock and then I see people who come on to this platform to either only invest in stock or only invest in cryptocurrencies. Here just to show you a snapshot, when I just even break out my portfolios into investors that are both in cryptocurrencies and in stocks - you'll see this is a replication of what I showed you before - the same discrepancy in their crypto trading with their stock trading as I showed you in the full sample. This is to me actually extremely interesting. Let me skip ahead and come to my last slide. This is really interesting because this seems to suggest that this momentum behaviour in cryptocurrency is not driven by just a very specialised or specific set of investors who have these overly enthusiastic or, if you want, have a gambling mentality, why they're investing in crypto. But they might be just very different people. That is not the case. In a way, our results strongly suggest that for the same person that invests in both crypto and stock, we see that in stocks, they typically have a much more contrarian mindset while in cryptocurrencies, they seem to be much more believing that there are strong momentum returns and make allocations in this way. Let me stop here. We also find a lot of heterogeneity across different… Sorry, we find that there is little heterogeneity across different people's demographic characteristics. This is a very strong constant in the data. Thank you very much
Well, thank you, Antoinette, and we'll open it up now to Q and A. Thanks to all of you who already submitted questions and as a reminder, you can continue to ask questions via the side bar. Antoinette, we have a couple of questions here. The first question is around the value of cryptocurrencies. The question is around whether there is an optionality component that is priced in, so with greater volatility that means the value of bitcoin or cryptocurrencies as an option should increase. What are your thoughts on that?
This is an interesting question! I think it really depends on how you want to think about your investments in, say, a cryptocurrency like bitcoin. As I showed you in this retail investor data, if you are interested, basically, in investing into it as a purely speculative asset, where on the downside in a way you are hoping for these very skewed returns without necessarily worrying too much on the downside, then you could say there is an optionality. I also want to remind you that this implied vol and this very high implied vol that I showed you for an asset class is obviously not a wholly good thing. In particular if you thought about it as a store of value, my colleague Bob Merton for example did a very nice analysis where he looked at: if you wanted to hedge $100,000 worth of bitcoin for a year at the current volatility, implied vol, and the cost of hedging this exposure it would cost you about $30,000. That just tells you that I think we should not really think about this here as an optionality. We are not really buying an option; to what? It is really more a speculative asset where, if you have preference for gambling, it can be interesting. Otherwise not so much!
We have another question here, I think also related to valuations, and so the question is around comparing bitcoin market cap to equity market cap. If you compare the two, bitcoin perhaps doesn't look too high, but is it not fair to really compare the two markets in that way?
In my work, I didn't compare the market cap between these two. I agree that at the current state, it's not clear that we want to compare the market cap in any way because even the function or the perceived functionality of cryptocurrencies in the future does not spin any of the functionalities that you might have in equity. So I agree that the comparison per se doesn't mean very much because right now, from what we have seen, I think cryptocurrency valuation right now - and this is what I was trying to show you in the talk - is still very much based on your beliefs about the future adoption of cryptocurrencies either as, say, broader store of value. As I said before, at least right now the statistical properties of at least the biggest cryptocurrencies right now don't look like a good store of value, but people obviously can have beliefs. Then I think obviously for some of the cryptocurrencies that have a bigger opportunity to embed smart contracts and actually wider functionality, that they might be the basis of other types of application in society and in particular in financial markets. That I think is obviously something where there is still a lot of debate about it.
Thank you. Around bitcoin adoption: have you done any studies or are you aware of any studies that have evaluated whether there's particular demographics or kinds of investors that are driving bitcoin adoption?
This is a very nice question. Let me break this out in two things. When we talk about adoption right now, we have actually seen that in general where the adoption is happening, so I have a recent paper that just came out as a NBER working paper where we're actually looking at the volume of block transactions in bitcoin on the blockchain. You can be tracing every transaction since, basically, the inception or at least in 2015, of bitcoin transactions on the blockchain. What you see is the vast majority of adoption right now is trading related, meaning it's flows of cryptocurrencies either say from individual wallets or individual accounts, addresses, or from institutions like you'd say other intermediaries like exchanges, payment process, to exchanges. Then volume from exchanges. So it's almost eighty per cent of volume on the blockchain right now, on the bitcoin blockchain, is trade related. We see that less than five per cent can be attributed to either, say, miners - cashing out the bitcoins they're getting from mining - or confirmed illegal trades or trades with payment processes. So this is to say that when we think about adoption right now, bitcoin really is adoption for trading. Now I come to the second part of the question, which is that for these types of transactions, mainly the trading transactions, we have actually seen that there are strong demographic trends or divisions, if you want. Those are that, not surprisingly, it's younger people so people below thirty-five, people who actually are more finance literate in the sense that they have more exposure to maybe financial markets before, other ones who are adopting. Then there is a very strong gender component, so young men are much more likely to trade in cryptocurrencies than young women or women in general. I want to say just one thing to this question as well: I didn't have time to show you in the presentation but despite the fact that there are these differences in who adopts - sorry, who trades - in cryptocurrencies, we find in our data that the way people update about cryptocurrency prices seems to be very similar between these groups. Conditional on trading in these markets, richer versus poorer, younger versus older, male versus female participants actually seem to all have a very strong momentum strategy in crypto.
Switching gears a little bit, so regulation is certainly a very hot topic with respect to cryptocurrencies. What are your thoughts on whether there should be regulation around retail trading in cryptocurrencies or just regulation more generally and the impact that it'll have on cryptocurrencies?
I think this is a huge topic! There are lots of nuances to slaughter this topic at very high level. I would say two things. One is, in this work that I have done with my co-author, Igor Makarov, that literally just came out, we actually find that one huge problem that we have currently in cryptocurrency trading is that the enforcement of KYC standards is extremely lax and it's very difficult. It's in particular difficult to do at an individual level. What I mean with this, and I would urge you to look at the paper - and I will put it up for you if there is interest - but what we are finding is, you take even say US cryptocurrency exchanges like Coinbase or Gemini, who are wanting to do KYC enforcement, what we however see is that given that they are allowing transfer from places like Binance or Huobi or exchanges outside the US, they do not enforce KYC norms at all. You actually have no ability to truly prevent money laundering, tax evasion. There we will need to really think very hard how we regulate even retail trading to avoid these very negative consequences for society as a whole. Then in general we obviously might want to think about consumer financial protection as a strong dimension where we want to make sure people don't take the wrong allocations.
Well, thank you. It looks like we're just at time, so thank you so much, Antionette. Your perspectives are really valuable. I think you helped cut through a lot of the hype with cryptocurrencies and that's really great, so thank you so much. We're also really excited to have you as a partner at State Street Associates, so thank you.