2018 NYU Stern FinTech Conference: Research Session 2: ICOs and Crypto-assets

{Music} I'm really, really excited
to introduce our next panel which will be led by my esteemed
colleague, David Yermack, who's the head of the finance
department here at NYU Stern and who did some of
the pioneering research on blockchain and Bitcoin. So let me hand it
over to you, David. (audience applauding) – Okay, good morning. We have three speakers in
the panel this morning, and in addition to me, Yannis
Bakos and Katya Malinova. The first speaker was
intended to be my colleague, Sabrina Howell. Sabrina had a baby one
week ago, and we knew that this would happen but
we're not entirely sure when, but since it was last Friday,
she won't be speaking today. The paper, and I don't think
we have the right topic here, but the paper is
co-authored by me, so I will stand in for her. Okay, so I'll be speaking
about a paper I've co-authored with Sabrina and Marina
Niessner from Yale on initial coin offerings,
and I want to begin by speaking about the Token
Summit that was held in May of 2017, about a
year and a half ago, and it actually took
place right in this room.

The people who ran
this had no connection with the university, but
sort of acted like they did. They rented a room, they put
the university's name up, they tried very hard to get
faculty to come speak at this, including both Sabrina
and me, and to a person, we all refused because we
thought that these tokens were completely fraudulent. Up until that point, there
had been crypto assets that were mined assets,
that were airdropped, that were created through
a variety of ways, but this was really
something new to just create these and sell
them directly to the public. This is one of the few
things I regret in my career, not taking part in
the Token Summit, because not only did this
turn out to be very real, but it was the point of
departure, and we can look here at the monthly data, that if
you pick up the story in May of 2017, that was
really the first month that you saw crypto sales
of what we now call ICOs of any real volume. These grew exponentially
through the end of the year, to the point that about
six billion was raised, and now in the first
nine months of 2018, there's been triple
that, more than triple, 20 billion raised, and
a couple of big spikes representing some very
large offerings that closed at different points in time.

So we have an innovation here that has very quickly gotten
traction in the markets, and it is beginning to
look like a substitute or successor to venture capital. The worldwide venture
capital industry, in its entirety, is
about 150 billion, and when you've raised 20
billion year to date in ICOs that tend to be focused
really on a few industries, you see that venture capital
is giving way to this as a new type of fundraising
for entrepreneurs, and the design of these things
is really quite interesting, and this is really what
I wish to speak about in my time this morning. So here's a ranking of
the largest ICOs to date, and the amounts of capital
being raised by entrepreneurs is just very, very impressive. You never see initial public
offerings raise this much money unless it's
something like Conoco or a very well-established
company, but it's now possible for entrepreneurs to raise money
in the hundreds of millions of dollars, and typically,
these are quite liquid in that the people
who invest in them are able to turn around and
sell them, sometimes right away or often after very
short vesting periods, which is quite different than
traditional venture capital which tends to be
locked up for years.

Now the tokens have clear
antecedents, I think, in three different ways. They resemble in many
ways social media tokens, and I think the most
successful of these up to now has been the QQ coin that
was successfully launched in China a few years
ago by Tencent. They also resemble, in many
ways, video game tokens, so there is a whole
parallel universe, where probably none of you,
but your kids may spend a lot of time here,
playing fantasy games that have their own
internal economies, where you can win coins,
but most importantly, trade these coins, so there
are now investment platforms such as this thing called
MMO bucks, where external to the game, you can
win virtual money, and then speculate
in it and trade it with other people and so forth.

One of these was acquired a
few years ago by Goldman Sachs, who knows a lot about
foreign exchange and didn't see too
many differences in how you would run this
market versus a forex market. I think the closest
analog to these, though, is in the world of
professional sports. This is the Giants Stadium that is in the Meadowlands
not very far from here, and when it was built
a few years ago, several hundred million
dollars was raised by individually selling
the seats through a device that's called a
personal seat license, and about half the teams
in the NFL have done this in recent years, and what
you're buying is the right to be a customer for a
platform, in this case, a sports stadium where
there is a clear limit on the capacity,
the number of people who can go to the game, for
instance, and you have the right to trade those seats
in a secondary market, which is what's on the right.

This thing called PSL Source,
and what the team is doing is really capitalizing the value
of future consumer surplus. In other words, you're raising
money from your customers and they're willing to invest
in this on the proposition that the enjoyment for being
a customer into the future is basically worth more
than the team is charging, and I think in sports,
in particular, this
solves a problem where you can't jack
up ticket prices if the team makes the playoffs without risking public
blow back and condemnation in the press, but if you
can quietly sell this as a capital asset in
advance to speculators, it's probably a way for
the team to raise capital at a much lower cost than
going to a bank and so forth. So I think what ICOs are doing are using a very similar model.

They are raising money
from future customers and they're using
smart contracts to put
a cap on the number of tokens, the number of
customers, in the sense, who will be able to use
the platform in the future. So we've traced this
back in history, and skipping a few
steps such as Wimbledon and the Sydney Olympics, the
earliest example we found is Royal Albert Hall,
which about 150 years ago was built with the same model, with creating a scarce
allocation of seats to future customers and
then selling the right to be a customer in perpetuity
to people who still today are trading these in a secondary
market 150 years later. Apparently, there are even
medieval churches in Germany that have used this
business model. I don't know if you
can resell the pews, but I think the innovation
here is very interesting in the capital markets,
that you're raising money not from debt and equity,
but from future customers, but like many things in finance, it turns out it was
probably invented by the British two
or 300 years ago. Now, in our study,
we look at a sample of 453 ICOs, and we
require you not only to have successfully raised
money but to have traded for at least 90 days in
the secondary market.

So unapologetically, we are looking at the
top of the distribution, the people who have
raised most of the capital and gotten most of
the attention, so
this is quarterly data about investment in
blockchain startups, and you see venture
capital almost exclusively, which is the red
bar, up until 2016, when ICOs begin to peak out. The total raised each
quarter in light blue is the uptake of ICOs, but it's
really in the second quarter of last year, right after that
Token Summit in this room, that this market just exploded.

So with the 90-day
trading exclusion, we don't capture
every one of these but we get the large
majority, so the dark blue is the amount of capital
represented in our sample. It begins to tail off at
the end of the sample period simply because we cut
off the data collection at a certain date, but if
we ran the study forward, we would sweep in
all of those as well. So some of the data
that gives you a sense of what the market looks
like, many of these ICOs are actually split
into two pieces, where they have
the public auction, but not until there's
been a pre-sale, where selected customers, who are sometimes called
the Bitcoin whales, these tend to be well known
venture capital investors, but they will get often
a preferential price and agree to somewhat
different conditions than the public is
able to participate at. So this happens about
45% of the time, and 34%, about one in three, have what
is called dynamic pricing in the auctions, meaning that
the price changes constantly based either on what price
people were paying a minute ago or the total volume
raised, or really, any function that
you wish to use.

This is completely unregulated,
so you can do things in this market that you cannot
do in a US sale of securities in the regular markets. It's quite interesting
to see what people do. It's almost like an experiment, like what if you would
abolish the securities laws, what would people do instead? Mostly these are sold
for other crypto, and they add to more than 100% because you sometimes have
a choice of Ether, Bitcoin, relatively few except real
money, such as US dollars, which is taken about
10% of the time. I think our most interesting
results are connected to the disclosure that these
coin issuers are doing. There is great fear on
the part of regulators, that essentially,
these will exist without consumer protection
and that you won't benefit from the disclosure
that you would get in a typical
securities offering.

On the other hand, we've
had an intellectual argument going back, really,
to 1933 and 1934, that the securities laws
may be not only unneeded but even overkill,
that if you had a world of just voluntary
disclosure, first, people would disclose because
it's in their interest to communicate with investors, and you would get a greater
diversity of disclosure that is much more
closely connected to each individual company
based on its industry, and ownership, and
economics, and so forth, and we find exactly that. The first thing to note is that there's just a
huge amount of disclosure. Almost everybody in our sample
communicates with the market, and the variety of
channels is quite rich. There's a lot of social
media, Twitter and Telegram. About 81%, four out of five, file what they call white
papers that look a lot like a prospectus,
but the white papers disclose individually quite a
bit of different information for each one, and some of the
most interesting disclosure is to post the code
underlying the token itself.

Typically, these are smart
contracts on Ethereum, and to simply post
the code on GitHub, in terms of due diligence
for an investor, there's really no substitute
for actually reading the code, if you're going to
buy a crypto asset. So this is probably something
that the SEC would not require if it were regulating
this market. This is really quite
different, and we were struck not only by the detail
of the disclosure but the analysis that we do. These are regression
results that we are studying in terms of the after
market liquidity as a measure of the
success of the tokens. It's closely connected to
whether you have a white paper, whether you post
the code on GitHub, and also to more
traditional economic methods like do you have vesting
schedules for the insiders and that if they have
an allocation of tokens, and they usually do, are they
restricted from selling them into the market, at least
for a period of time? So I think the closer you look
at this market, first of all, it seems to be extremely
rational that if you take steps to reassure the public
and give guarantees of your own commitment
to the success, that clearly is reflected in
the after market liquidity and your ability to attract
investors, and arguably, there's more disclosure going on that's more informative here
than you get in the IPOs of equity securities, which
we were really struck by.

So the idea that this market
may be fraudulent and risky, I think every market has its
own baseline of fraud and risk, but in many ways, this
looks some quite compelling as a way to raise capital,
and very interesting for investors to participate in. So two other quick results. We looked at the background
of the founders themselves. We read LinkedIn for several
hundred coin issuers. Turned out 97% were male,
which is the biggest gender gap I have ever seen in any
kind of finance study, and even CEOs of public
companies and so forth don't skew quite this extreme,
but the ones who are best at launching these
tokens are the ones who are already
successful entrepreneurs. A good example is
the Brave browser, where the people
who brought Netscape to the market many years ago
are also behind this browser. And it's much better to have
entrepreneurship experience than to have experience
in finance or cryptography or computer science, at least
in terms of the response of investors, and we find that
the most successful tokens are the ones that are these
so-called utility tokens.

So these are not simply the
securitization of other assets such as Tether or some
of the gold tokens that you're seeing. These are tokens that are
really focused on customers getting access to a scarce
platform, and I think in informing the debate
where the SEC is obsessed with regulating
these as securities and saying that they
are not anything other than speculative
instruments in disguise, you would point to
a result like this where it's really the, the
utility value of the token seems to be what is
attracting the speculators and the investors. So the conclusion here is that really something
quite interesting is going on, and I think against all
expectations, this market is continuing to thrive,
it is a very innovative and new way to raise capital,
and I think it's something that is likely to be with
us well into the future and is going to demand
integration into the fabric of the capital markets
sooner rather than later. So, thank you very much, and I will now move
to the next paper. Okay, so the second paper will
be presented by my colleague, Yannis Bakos, and if you would.

(audience applauding) – Thank you. (audience applauding) Thank you very much, David. What I want to talk about is
about a particular application of this fascinating new
world that David described, and David saw the
slide in the beginning where he had the 10
more successful tokens, and for counting about six or
seven of them were platforms, and as he mentioned, one of the
most successful applications of tokens is issuing
them by platforms that typically are new
platforms that are trying to establish themselves. So what I will talk
about is some work that looks at how platforms
can use tokens as a way to promote adoption,
and this is joint work with my colleague,
Hanna Halaburda, and I have some motivation
but you got a whole 15 minutes of much better explanation
for how this environment has become popular, the
exponential growth that we have. One possibility is
what we're seeing is we're seeing a lot
of hype, one possibility is the reason why
IPOs are so successful is that they give you the
opportunity to do things without following all
of the regulations.

There's a possibility
we have a big bubble, but if that is the
case, at some point, this is going to deflate,
and then the question becomes how much real value do we have? Can ICOs, can tokens help
companies create value that they couldn't create
otherwise, and so the work that we'll talk about is
in this stream of research, and in particular, in the
context of a platform. Now, platforms create value by bringing together many
different types of users. It could be friends on
Facebook, it could be borrowers and lenders in the lending club, it could be drivers
and riders in Uber. The key thing here
is that, typically, platforms have very
strong network effects, which means that platforms
become much more valuable as they can attract per
team or participants. I mean, obviously, I don't
want to be in Facebook unless my friends
are in Facebook, and once my friends
are in Facebook, then Facebook becomes
much more valuable to me, and this network effect is
something that economists have studied for decades,
and one of the results that is very common when
you look in the literature and you analyze platforms is what we call the
coordination problem.

So essentially, the
chicken-and-egg problem. It's the challenge in
bootstrapping the platform. The fact that nobody wants
to be on the platform unless everybody else
is on the platform means that the platform
can have a problem when it is a new platform in
trying to establish itself, and this is actually especially
true if there is a cost in joining the platform. If you must incur a cost
in joining the platform, you are not going to do
it unless you believe that the platform is successful. Now typically, the way
platforms solve this problem is by subsidizing early users. Like I'm actually, I've
been around long enough to have received $20 to
open a PayPal account, plus $10 for every
colleague or every friend that I recommended to PayPal. That was quite
expensive for PayPal. I think I ended up
getting about 150, $200 back in 2001 or 1999.

I never got an incentive from
Uber but I did get a month of 30% off from Lyft and a
whole bunch of free rides, and this is typical. As a platform, you want to
offer incentives to the users and what you do is this
way you establish yourself and you make money
from the later users that come in the platform
as part of the second wave. Subsidies require capital,
and capital may be available to companies like Uber, and it was available back
in, for PayPal back in 1999, but for a lot of new
platforms, it may be difficult to raise capital and offer
subsidies, and especially, if they're perceived as
being risky, and what we find and what the question we
address is whether platforms can actually use
tokens, an ICO, as a way to incentivize users to
converse on the platform, and therefore is a way to
make the platform successful, because a lot of the
strategies in platform is trying to establish yourself as a self-fulfilling prophecy
that you are the place where everybody will converge, so if you can convince your
users that you're successful, you become successful.

So can you actually get
your users to join you by giving them a token
instead of a subsidy, and when would you
want to do that? And we find that this
is actually true, that if you issued
tokens, this allows you to either reduce the
subsidy that is necessary, or even better, instead
of subsidizing your users, you can make money from them and you can essentially
make it a lot easier to finance the development
and the establishment of the platform. Now, depending on what
is your cost of capital, tokens could be a
lot more attractive than other alternatives,
because your other alternatives are typically, they're
raising VC funding, basically, so in
equity, and more rarely, you can maybe access the
debt markets and get debt, so depending or what is
your cost of capital, it could be the tokens
are the way to go, and not surprisingly,
this is especially true when your cost of capital is
high, when your cost of capital is moderate or low,
actually turns out you could make more
money in the long run by issuing subsidies or
by attracting your users rather than giving them
tokens, and the reason for that is that when you issue
tokens, essentially, what you are doing is you
are, especially in the case of utility tokens
which is typical in what the platform
would do, is essentially, you are pre-selling
your future services, and you need to compensate
the users for the fact that they are taking a risk.

So you will typically need,
the more risky you are, the more you will need
to compensate your users, and in effect,
what you are doing is you are giving up future
benefits, future profits, for the return that you're
going to get upfront, which will allow you
to create the platform and launch the platform. And we, actually, we do this
with some economic modeling. We study this question, and
we have a two-period model, so that we can capture the
dynamics of the early adopters and the second wave
of later adopters, and so we have a platform. We look at early
users and late users, and we have a situation where
we have a coordination problem in both periods,
which basically means that we're looking
at a situation where
most of the value from the platform comes from
the network of other users, and therefore, the platform
is not going to be able to convince those users to join, unless everybody believes
that everybody else will join, or alternatively, unless
the users are bribed or subsidized in order to join.

So first period, users
arrive, we look at two, we compare two possibilities. One possibility is that
the platform subsidizes, or sets a price for users, and this could be
a negative price, which is what you would do if
you are offering a subsidy. Almost always what
happens in these cases is your early price
is discounted. That's how you get
your users to join, and you hope to make up the
discount from the later users. Once we've established yourself, you'll be able to raise
your price, and as I said, this discount
could actually mean that you go below zero,
you pay your users, as Uber or PayPal were
doing at the beginning.

The other possibility is that
the platform sells tokens, so you do an ICO, and you
let users buy your token, and then what you hope is
users have an incentive to make the platform successful. They're going to
converge on the platform because the tokens will have
more value in the future if the platform is successful,
and you can see what happens in the first period, and then
you have the second period, where the second wave
of adopters come in, some of the users from the
first period will leave, and what these users will do,
they will trade their tokens to the new arrivals, and
depending on the setting, we study either the
platform now has a new price for the second period.

In the second period, your
price will never be negative so you will always
try to make money, and you will make again
as much money as you can, depending on how
successful the platform was in the first period, and
depending on how much it can convince, the new
users that it will succeed in the second period as well. The other possibility
is that the platform still sells tokens. It sells tokens to the new
arrivals or anybody that arrived in the first period
and didn't buy a token, they can buy a token
in the second period. So this is the setting
that we study, and again, let me repeat that the
key feature, of course, is that users that bought the
token in the first period, they always have the option
of trading the token, and essentially, realizing the
return in the second period because token prices will go up if the platform is successful. So what we find is
by issuing tokens, the platform can make more
money in the first period, so when the platform
tries to launch itself, it needs to offer lower
subsidies and, actually, it can make money by selling
its tokens, essentially, by selling the pre-selling,
the future services, and the reason this
works is that users that buy the early token,
they know that their token or they expect that token to
appreciate, and so in a way, they're investing in the
success of the platform, so you can think of
them that even though this is a utility
token, there is a little of an equity component in
the sense that the users become vested in the
success of the platform, where the return is not going
to be a formal appreciation of equity, but this
could be an appreciation of the utility token.

Without the token, the platform
needs to subsidize users in the first period,
which means that you need to spend more money in the
first period, but you make it up in the second period. It turns out that once you
are successful, you don't need to compete with the users
that have bought our tokens in the first period. You can enjoy the full
value of your success. You can price,
basically, and extract as much value as you can. So our first result is that
the second period effect actually outweighs the
first period effect, which means that if you
face no cost of capital, if you can do either and
it won't cost you more to subsidize users
in the first period, you make more money
by offering a subsidy.

Hold off on the token. However, if you have a
high cost of capital, or if capital is not available
to you as it is not available to a lot of these
aspiring platforms, then you make more
money by issuing tokens, and the best strategy that you
can follow is to issue tokens in the first period and then
also issue tokens later, and it turns out actually
that the token strategy is more attractive, in a
way, the more risky you are, so the less you can establish
yourself as being inevitable, the more you want to issue
tokens, and that makes sense because if you can
establish yourself as being the inevitable winner, then users basically
are pretty sure that you will actually succeed
and they don't perceive you as risky, and you
can raise capital, but if you are a risky platform
upfront, then it turns out that, again, users
will need a big subsidy in order to decide to adopt
you, and in that case, you may be better off by
issuing tokens, and by the way, I've been telling you a
story, we do have a model with a lot of Greek
letters and diagrams.

I'm not going to bother you
with that, but to conclude, kind of will summarize
what we find, again, tokens provide the
mechanism for a platform to incentivize users to adopt
it, and a key aspect here is the fact that
tokens are tradable. The fact that early users,
that they buy the tokens, can trade them later,
and participate in the success of platform. If you have an option to do
either, what happens with tokens is you get more money
early, but in some ways, you're giving up future sales, you're selling your
future at a discount.

So in some situations, if
you have low cost of capital, it would make sense to
subsidize the early adopters, but if your cost
of capital is high, then you don't want
to offer subsidies. We just give them cheap tokens, and the key, actually,
finding out of this is that for a platform
that is facing a high cost of capital, it could not be
able perhaps to establish itself without tokens, so
what tokens enable, they enable the
creation of platforms that could not be
sustainable without them, so in that sense, they do what economies call
create more economic surplus, and that's why we think
the whole token ecosystem can actually create
a lot of value and it's a very
interesting innovation that we are staying on top
of and we keep studying.

So I will stop here. Thank you very much. (audience applauding) – Thank you, Yannis. I'm surprised to hear you
didn't get the Uber discount, and I'll invite you. I think everyone
else in the roo– – [Yannis] I started
using Uber way too early to get the discount. – Talk to me afterwards
'cause I can still get you in. Oh. (chuckles) Okay, our third presentation
will be by Katya Malinova of the University of
Toronto, and the title is Market Design with
Blockchain Technology. (audience applauding) – Well, thank you, David. Thank you, the
organizers, for having me. I should say that I
have moved (chuckles) to DeGroote School of Business
at McMaster, but indeed, I'm still on leave from
the University of Toronto, and my colleague, Andreas Park, is at the University of Toronto.

So as David, as Yannis
talked about, block-based, blockchain-based trading
is now a reality, and the question that
we ask in this paper is what's the impact
of blockchain design on blockchain-traded assets on trading
blockchain-based assets? And accumulated from my
research in market structure, my main question was,
well, what's different about trading
blockchain-based assets from the trading perspective,
not the valuation or investment perspective,
and one of them is that multiple trading
protocols are now possible. You can still trade them
in a limit or a book, you can trade them peer-to-peer. In peer-to-peer,
it's here differs from the over-the-counter
markets that many of us are used to when there is dealer and an intermediary involved, and I cannot just
go and find you.

With tokens, I can go and trade
with any of you right away, and I can find you
right away as well, so we don't have to go through
the brokerage to do this. The second feature
is that by default, this trading is
extremely transparent. If you go on Etherscan, you
could actually see the addresses for any transaction that
traded with each other. So you could be able,
you might be able to identify frequent
traders and you will see, you will see the IDs
that trade frequently, and it's not just
the transactions but you can also
aggregate these trades and see the holdings. So you can sort of see
the whales or traders that trade a lot, or rather,
hold large positions. And so this brought
us to two answers to our original question,
what's different? Well, first, we
get frictionless,
peer-to-peer trading, and this is, of course, the
nature of blockchain technology like Internet allowed us
to transfer information.

Now we could transfer ownership, including ownership
of contracts. And the second is that the
informational environment for trading these
assets, securities,
tokens, utility tokens is dramatically different
from that that we know from equity markets
or bond markets or any securities markets. Of course, here, there
is one distinction that when I say
wallets or addresses, they're not the same
as traders, and again, this would be a difference to traditional,
over-the-counter market where you probably would
know the counterparty that you're trading with. And that brought us to the
question that we try to address in this research paper, and
it's a theoretical paper. How does the design
of the technology, how does the design of
the ledger transparency and the usage of the identifiers with possible
peer-to-peer interaction, how does that affect
trading behavior, and how does that affect
economic outcomes? And we're agnostic in this paper as to whether we think
this is a utility token or security token, but it's
probably, what we had in mind when writing it
as security token more so than the utility token so that you actually
traded frequently and not just I sold it
to you and I'm done.

But there's now a lot of
literature on the topic that I will skip in
the interest of time, including by people
in this room, and in terms of the
model, again, as I said, it's a theoretical model
and it's a trading-based, theoretical model so
we have a risky assets, we're gonna have
two large investors. One is gonna be hit by liquidity
shock so they have to trade and the other has a possibility
of absorbing the shock of trading at zero cost, and
we have some small traders that are distributed
so the peers that you could also
trade with, again, differently to the
over counter market, and we have an intermediary
that is able to, that you're able to trade with, but the intermediary's
risk-averse, and so every time you
go to the intermediary, it's an inefficient
transfer of risk, and also, you pay the cost to the
intermediary, so the question that we have is, well,
how do the costs compare? Well, there are direct costs
for trading, and when we, we're assuming in the paper
that it's quite costly for you to trade with small investors,
small traders, as a trader, and the reason for that
is we say data processing but it doesn't just have
to be data processing.

For me to trade with you
with so many platforms, and for me to find you,
I have to actually go, and because there is no broker,
I can't just hold an account at a brokerage and trade in
multiple different platforms. I actually have to go
and open an account on every single one of
these platforms for me to trade with you,
and that's costly. Includes also capital transfer
often, and just generally, when I trade and I
have many transactions, you could also think of
this as a transaction cost because the more
transactions I have, the larger the validation
costs that I have, the mining cost, so it's
a cost per transaction, not so much per value,
and if I'm trading with a large trader, the
result's an indirect cost because the person
may front-run me, by going to the intermediary
and raising the public, raising the price. And so when we're
thinking about it, you're thinking of three
different possibilities of sort of designing
the addresses or designing the blockchain,
so one is our benchmark. When we presume that
everything is transparent, so as I showed you an Etherscan,
you can see the address that you're trading
with and you can see all the past transactions,
and however, I can't just go and generate multiple search
addresses, and why can't I? And one is because as a
private blockchain designer, you decided to prevent
me and you do the KYC, know your customer,
and you say that's it, you're allocated just
a single address.

Another possibility,
as a contract designer, a token designer,
that you decided that you're gonna
restrict who is able to trade your contracts. Why would you do that? Well, one example why
you might want to do this is if you want to restrict
trading your tokens just to created investors. You'd need to know which
addresses, which wallets these accredited investors have. So there is a possibility
outside of the public blockchain and outside of trading
Bitcoin and Ether that you are able to
actually restrict trading. So in the benchmark
model, again, everybody knows and
sees everything. We're not restricting
transparency, just
the number of IDs, and the options for the
large trader are two.

You can trade with
small investors or you can trade
with a large one. If you trade with
the small ones, you have this complexity cost. If you trade with a large
ones, they might front-run you, and if we only trade once,
then actually, I wouldn't trade with a large investor,
because if we only trade once, they are gonna front-run me and they're gonna extract
the entire surplus. I'm not gonna get
anything from it. However, for trade repeatedly,
because we're large traders and we know tomorrow, I'll come,
today, you'll front-run me, tomorrow I'll come back to you and I'll wanna trade
with you again, or you may have to trade, and you have to come
and trade with me, and I won't because
you've front-run me today, so in theoretical
terms, we're modeling it as a repeated game
and you're punished by the grim trigger strategy,
so there is a theory to that but the bottom line is when
we interact repeatedly, social norms are
gonna start to work, and the large traders will
trade with each other, so we will avoid trading
with an intermediary.

We will avoid their
inefficient risk transfer to the intermediary. We will avoid paying the
extra cost so this model is, the benchmark model is
actually best for welfare, pretty much by design. However, we recognize
that people may want to hide their trading intentions and they may want to hide
their IDs, and one way to do it is just to say we
have the same model but we don't show
you who trades. We don't show you the IDs. Well, in this case,
because I don't know which of you is a large trader.

If I just randomly approach
some of you, chances are, I'm not gonna find the large
trader, so in equilibrium, I trade with small investors and I trade with an
intermediary as a large trader, and turns out, this is the
most inefficient setting, so if you wanted to
restrict transparency, that shouldn't be the
way to go because this, even though it
seems very natural that you just take a normal
setting, a regular setting, and you put a privacy screen,
the problem is that may in, that disallows the large
investors to find each other, and that creates
problems for liquidity, and if it creates
problems for liquidity, that's gonna create problem
for welfare as well, so another way to think
of achieving privacy is what was suggested
by Vitalik Buterin, one discussed in Ethereum,
is just allow people to create as many
addresses as they want and try to hide by creating
multiple addresses.

I'm not talking whether it's
technologically possible, but even if you were to
just put a privacy screen, still, you should be
thinking of putting multiple, multiple addresses,
so in this case, if you have multiple
addresses, a lot of, a lot of the
results in our paper depend on the
probabilities of acceptance and the probability that if
I come to trade with you, what are the chances that I'm
gonna find a counterparty? The problem with putting a
privacy screen in a single ID is I cannot find
that counterparty. Now, when I have, allow
you to have multiple IDs, large trades, multiple IDs, I'm much more likely to
find that large trader, and whether or not
it's beneficial or not is gonna depend on whether
or not that large trader trades with me or not. Now, as I'm almost out of time, I'm gonna skip the
technical details, except that they're not
getting skipped. (chuckles) And I'm gonna proceed with
a couple observations.

One is that when we
have an intermediary, every time we have
an intermediary, the market is potentially
socially inefficient. When we have large
traders interact with small investors
peer-to-peer, that's gonna impose transaction
costs on large investors, so what we would like to do is we would like to design
a system that would ensure that our large traders
are able to find and trade with large traders, and if we're comparing their
pack designs, for this reason, when we have a privacy screen
with a single ID restriction that large traders can
just have a single ID, that's gonna be welfare
reducing, and in the system, when the large traders
don't interact, it doesn't really matter,
but another comment that I'd like to make
is often, in our model, in order to prevent large
traders from front-running, I'd have to pay concession
to the large trader, to say, well, if we
interact infrequently, you don't know when you're gonna
get hit by liquidity shock, so I'll pay you today
a little bit extra, still less than through
the intermediary to prevent you
from front-running.

Well, in a system
with multiple IDs, because I don't know
who the large one is, I'm gonna have to
pay this concession, not just to the large
trader, but also to the IDs of the small traders, and
so there are configurations when welfare efficient
is to do one thing and large traders are gonna
prefer something else. So in summary, back-office
designs do have front-office, do have trading implications. We should think
about it carefully, and what our model tells
us that the public solution to privacy might actually
be the most efficient one.

Thank you very much. (audience applauding).

You May Also Like