DeFi Protocols Should Act More Like Fiduciaries

This week, I grapple with the concepts of financial centralization and decentralization, anchoring around custody, staking and DeFi. 

On the centralized side, we look at BitGo's acquisition of Lumina, Coinbase Custody and its similarity to Schwab and Betterment Institutional. 

On the decentralized side, we examine the recent $500 million increase in value within the Compound protocol, as well as the recursive loops that could pose a broader financial risk to the ecosystem.

It is a difficult one for me to untangle, in part because I am not sure for what audience I am writing. Ever since starting to work more deeply in the crypto ecosystem, I've come across a very different set of norms and expectations in the financial industry and the fintech startup community. Notably, crypto builders champion software that is "trustless," "decentralized" and "permissionless." This creates a worldview towards incumbent money and financial products that does not merely wish to reform them, but to abandon them all together. In turn, this community is also far more authentic in trying to change the world.

See also: Lex Sokolin - Software Ate the World, Here’s How It Eats Finance

Having seen most of fintech fail to accomplish structural transformation over the last decade, I tend to agree with the desire to cause fundamental change.

And yet, we need fundamental change not for the sake of change but for the sake of empowering people to live better financial lives. Most of my audience lives in a nation-state with strong legal and economic systems in place. There is no need for them to stand outside of society, like the crypto-Borg, to benefit from the inventions being created. Participation should not be all or nothing.

There will be a range of projects on offer in the new world. Some will be maximally decentralized, trustless and built for an adversarial environment (i.e., everyone is trying to steal your money). Others will extend the financial and economic activity that the financial industry performs today, and weave it into blockchain-based environments. The best outcome is to pull the financial share of global GDP into a digital chassis like the internet, but for value transfer and settlement.

Having held licensed operating roles inside of broker/dealers and registered investment advisers, I can attest to the impact licensing and the power of the state plays in the consumer adoption of primary financial apps (i.e., not toy money but the default bank or retirement account) and the size of money flows that a company can handle. A regular person wants to see that their money is FDIC-insured, not Lloyds of London- or Nexus Mutual-insured. A regular institutional investor allocating an endowment wants to deal with a qualified trust company custodian, not a trustless protocol. At least today.

This is not a normative statement – something I "wish" to be. It is a descriptive statement of how things are. 

Let's anchor in examples, starting with BitGo and Coinbase Custody. These companies remind me of the large institutional custodians in the U.S., which now hold over $3 trillion in assets for advisers, in addition to manufacturing large-scale exchange-traded funds and delivering broad technology suites for their users. A few months ago, BitGo acquired Lumina, the digital asset portfolio management company. Lumina was started by a team with experience at Addepar, the alternatives-focused family office performance reporting tool from Silicon Valley by way of Palantir. 

BitGo is following the playbook of the RIA custodians, which offer portfolio management and trading software as part of the custody package. Similarly, the wirehouses of Bank of America Merrill Lynch and Morgan Stanley also build this type of software for their 20,000+ institutional sales footprint. 

BitGo also has tax preparation and off-chain trading/settlement services for its crypto fund and exchange clients. This is a firm that grew up custody-first, focusing on how to store private keys to blockchain-based assets in a multi-signature environment, insuring assets up to $100 million, and providing a regulated trust company as an institutional counterparty. It is now integrating bits and pieces of technology to make those fund relationships stickier.

Would we be better off without the "centralization" that BitGo provides? Would we be better off without the crypto funds that brought assets into the space, funded new projects, generated trading activity and created investor returns all the while pioneering how companies tread the regulated path? I think not.

Coinbase Custody is an interesting counterpoint here. Coinbase grew up the way discount stock brokers grew up in the 1990s: providing retail investors (versus institutional investors) access to an asset class in a more convenient way. Whereas Schwab, Fidelity and TD Ameritrade built their online business on price and UX competition with phone brokers, Coinbase built its business on a demystified user experience for crypto assets.

Having served dozens of millions of customers, and de facto custody of more than $20 billion in digital assets, the natural play for Coinbase was to extend its services to institutional clients. This is how the institutional arms of Schwab, Fidelity as well as robo-advisers like Betterment have been built. 

See also: Lex Sokolin - Libra Wanted a Currency, All We Need Are DeFi’s Open Payment Rails

For context, the value of advised assets under management in the United States has grown to $9 trillion by 2018 from $4.4 trillion in 2011. About 30% of the pie sits with the large Wall Street banks, another 50% is with the discount and online brokers and the rest is with independents. Coinbase is a discount broker integrated with a custodian and an exchange all in one. The retail market is its cash flow engine, while the institutional markets are its scale engine. There is a lot of work left to do.

Various crypto protocols, including Ethereum, are upgrading to "staking" approaches to secure their networks (versus spending electricity-powered computing power on a Bitcoin puzzle). You can think of staking as capital collateral for a bank or a native savings account in some foreign economy or ownership of preferred stock yielding both voting governance and cash dividends. It is highly technical because it derives from mathematical work to be done in securing a blockchain. 

The result will be that institutional actors will find custody solutions even more attractive given they can earn a yield on parked assets. This is certainly the case with institutional share classes for money market funds provided by traditional custodians to their large clients. Nothing new under the sun!

Retail investors will also have access to this infrastructure through the consumer footprint at Coinbase, bridging millions into learning how to support blockchain networks through participation. While the dream is that everyone runs their own node, such an outcome is simply not practical today. Investors want to allocate assets, not learn how to run byzantine software or package financial products. 

This creates a centralization tendency for capital to accrue at custodians, the way it has in the traditional money management industry. Take for example Institutional Shareholder Services, a company founded in 1985 that aggregates voting power across fund products that hold equities to make it practical to participate in company governance. But, practically speaking, this is not a new problem.

We are better off teaching 30 million people about Ethereum staking through Coinbase, even if they do not hold their own keys. Because this is the on-ramp towards decentralization. It is the equivalent of financial literacy for the 21st century.

In many decentralized finance projects, the concepts of custody, regulation and trusting an intermediary are being actively minimized by the builders of those projects. To interact with this emerging sector, all you need is a software device with a key to the global crypto money network. Notably, however, even in this paradigm successful projects accrue massive network effects (both financial and social) and effectively intermediate on your behalf.

Last week, one of the core building blocks of the DeFi ecosystem underwent a profound change. Compound, a protocol to clear borrowing and lending of various digital assets, introduced a token that gives a holder governance voting rights. The token is earned when a user borrows assets from the platform. This set off a series of strange financial outcomes, the most clear of which has been to three times the token price and increase the balance sheet by over $500 million. 

The underlying reason is that the yield payback became self-reinforcing. Users can lend an asset to receive yields in the 10%-30% range, depending on the market. The borrowers must pay that interest with a mark-up, but they also earn back the Compound token through their activity. That token has been increasing in value, because the market is perceiving there to be economic activity on the Compound network. As a result, what you get paid as a reward is more than what you have to pay in interest. This creates a recursive loop.

Is this economic activity? Or is it arbitrage? Is there a difference? Remember that PayPal paid users $20 to sign up and refer in others, so multi-level marketing is perhaps at the core of financial entrepreneurship.

Naturally, I went looking at the other DeFi protocols and their tokens to see if similar loops are in the works. Exchanges that do "transaction mining" (i.e., trading for metrics sake) have largely been punished by the ecosystem and remind me of churning in the traditional markets, so I would be surprised to see that come back in vogue.

But there are indeed other DeFi protocols, such as Balancer (an automated market maker, or "AMM"), that have mechanisms to earn a return through providing liquidity into some pool of capital in which users trade. An example would be a PieDAO pool that takes multiple stablecoins and generates 20% returns in the tokens of the underlying AMM. It is valuable for the project to have assets, and it will therefore reward you for parking them there.

The more I stare at all these projects, the more clear something becomes. Here is the key chart I put together based on the data at DeFi Market Cap.

Within the top 100 assets, representing a bit over $6 billion in exposure, 11 are protocol tokens like the Compound one I described, but the rest are some combination of liquidity pools, constructed portfolios or structured notes (a "wrapped" coin in the crypto parlance). That suggests over 80% of the assets are either derivatives or structured products or some other packaging of underlying exposure. 

While that footprint translates to less than 20% of the overall value today, it is a notable explosion in complexity for a limited underlying financial activity. Further, speculation in these correlated assets opens up increasing capacity for implicit leverage. As the Compound token becomes more expensive, more people show up to borrow underlying exposure and earn capital gains, which in turn fuels the cycle.

The world saw defaults on billions of value in structured products in 2008. Let's not make the same mistake again by assuming technology protocols are immune from default risk and black swan events.

Going back to the beginning of this article, I started with the distinction between re-engineering the existing system and building a completely new permissionless one. Looking at the types of products emerging from DeFi, I would suggest these are advanced institutional capital markets machines creating sophisticated exposures. 

In the traditional context, the entities making them would be regulated, the people selling them would be licensed and the products themselves would be registered – all under some version of fiduciary duty to do no harm. In fact, it has never been easier to be a true fiduciary and steward of money, with digital scarcity, the authenticity of financial assets and a record of transactions baked into the blockchain protocol itself. 

I hope that in re-imagining the world of finance from scratch – whether or not there is licensing or everything is permissionless – we hold such an oath to heart.