Envisioning Capital Recycling in Decentralized Networks

The purpose of this piece is to explore how decentralized PoS networks may be able to utilize idle base layer assets in their ecosystem and potentially support themselves autonomously, even if just on a partial basis. In other words, how might decentralized PoS networks be able to enhance the productivity of their own native underlying assets which, after all, are the lifeblood of their incentive structures? The Ethereum blockchain and MakerDAO network will be used for reference in this piece for illustrative purposes.

What is capital recycling in traditional markets?

The concept of capital recycling first became popular in the real estate investment trust (“REIT”) sector of the traditional marketplace in the early 1990’s. If a REIT traded at a significant discount to net asset value (“NAV”), the implied course of action to (attempt to) create value would be to sell assets and buy back stock or pay down debt (or both) with the proceeds. Capital recycling has since grown to become a common practice among private equity funds, where rather than managers distributing promptly to limited partners the proceeds received from a portfolio company exit or refinancing, managers permitted themselves to retain, or distribute and then recall, these proceeds to make new or follow-on investments with the anticipation of creating additional (future) value for investors. Hence, the term “recycling”.

How might this tie in to decentralized PoS cryptonetworks?

Honest answer? I don’t know. But let’s get creative.

What follows below is hypothetical thought exercise that does not currently exist (yet). Readers are expected to have a basic understanding of Ethereum and the MakerDAO network built atop it. For those that are unfamiliar with MakerDAO, I recommend reading the first part of Vision Hill’s MakerDAO Case Study.

In the case of (single collateral) MakerDAO, ETH is locked up as collateral via smart contracts to enable users to open collateralized debt positions (“CDPs”) that are essentially Dai-denominated loans. The minimum collateralization ratio is 150%, meaning for every unit of Dai that users extract from the MakerDAO network, 1.5x that value must be pledged as collateral in ETH. This has led to quite a few interesting experimental developments in the DeFi ecosystem in last few months (see Dharma, Augur, Uniswap, UMA Protocol, and more), strengthening the Ethereum community in the process. In fact, just a couple days ago, the Set team announced their TokenSets launch, becoming the first player in this space to use the new multi-collateral Dai oracles in production.

Currently, a remarkable ~2% of the total Ether supply is locked up in the MakerDAO ecosystem. However, despite this DeFi fanfare, the ETH locked as collateral in the MakerDAO system currently sits there quietly. So long as the network’s collateralization ratios are properly managed for risk management purposes, this ETH essentially becomes idle, unproductive, and underutilized. Is there a way to concurrently utilize this ETH for other purposes, such as for staking/validation services to support the base layer, while it’s pledged as collateral?

Dan Elitzer recently proposed a similar thesis (Superfluid Collateral) to initiate thoughtful conversation around ways to potentially bring productivity to the underutilized ETH collateral in the MakerDAO network. While this piece proposes a similar outcome (productivity), and may somewhat be an extension of the Superfluid Collateral thesis, the key difference is there is no rehypothecation here (the assets will not be continuously passed around like a leverage baton with “house of cards” risk; instead they will productively be used to support their native networks, with a single specific path of action in the event of redemption - the holder receives the collateral back).

Suppose for illustrative sake that Ethereum’s beacon chain is live, and that a certain portion of the ETH collateral in MakerDAO concurrently gets automatically bonded to the beacon blockchain for validation. While protocols/smart contracts certainly can’t run their own nodes, for purposes of this exercise let’s assume there is an automated (codified) selection process run by MKR holders that enables this ETH collateral to be delegated to an existing validator. Let’s also assume MKR holders vote (for risk management purposes) what proportions of the ETH collateral can become bonded versus what proportions need to be held back in liquid collateral reserves to protect against any delinquent CDPs. The liquid collateral reserves (e.g. the amount of ETH collateral in this example that is not bonded) should be able to sufficiently meet a certain percentage of free-willed CDP redemptions (the amounts of which would likely be determined at the discretion of MKR holders). The incentive for users to enable collateral bonding is to earn more collateral via validation rewards. This can essentially become a margin hedge if done at the right volumes.

What risks get introduced from this?

First, the liquid collateral reserves may become insufficient if the price of ETH collapses, or a suddenly large slew of CDP holders decide to voluntarily close out their positions (or both). Given the beacon chain’s unbonding period may span several days until enough liquidity gets gathered again for distribution, CDP holders may find themselves in a temporary liquidity crunch. In such an event, CDPs can still be closed, but rather than CDP holders (or keepers) gaining possession of ETH collateral (as they do today in SCD), in such a scenario they could gain claims to soon-to-be-received collateral (presumably, user consent would be needed in advance). MCD will likely be adaptive and follow similar suit. Those claims could openly trade on the open market as IOUs at incrementally small discounts (I envision this being similar to the repo rate market). Thus, if one is able to sell that claim for enough Dai to repay the outstanding CDP loan, the CDP can successfully be closed and the ETH collateral will be distributed to the new claim holder. Otherwise, if such a user is unable to sell the claim, that user is left waiting until the beacon chain’s unbonding period lapses to receive the ETH collateral to repay the loan.

What happens to the MakerDAO network in this scenario? The network’s collateral value is not necessarily worth less than the amount of Dai it is contracted to back; rather, a portion of it is just temporarily illiquid. The CDP is closed, and there is essentially a “burn accrual outstanding” for when the Dai gets repaid (e.g., burned). Can the network still operate on burn accruals? The system already calculates stability fee accruals (e.g., “future burns”), so the answer is likely yes. The Dai USD peg in this case may be expected to break slightly to the downside in the interim until the claim is redeemed, and then (all else equal) it should return to its peg. The Ethereum codebase would have to know when to unbond collateral to satisfy any such outstanding debt, otherwise the system risks censorship.

While it may be easier (and far less complex) to instead tokenize the bonded beacon chain ETH, enabling top validators to issue tokens representing claims on bonded ETH (a similar approach to Compound V2) that can be used as collateral in the MakerDAO network (essentially reversing the above proposition), this approach does not enable productivity for the existing non-staked collateralized ETH sitting idle in the MakerDAO network.

What happens if there is a fork?

Does the ETH collateral in this case (whether liquid or bonded to the beacon chain) become artificially duplicated? In the ETH beacon chain validation example, does a duplicated phantom validator appear? Participation rate management can protect against this by limiting new validator demand in order to protect against existing validator dilution. However, there will still be an artificially created second ledger with phantom value (additional collateral) that can potentially grow over time. If there is no participation rate being managed, more ETH can be bonded to the beacon chain. While some may view this as a “velocity sink” catalyst that can potentially facilitate further ETH value accrual, the reality is the risk of alpha decay (declining performance over time) for the existing validators can potentially jeopardize the base layer’s security proposition over time. This is a tricky thing to forecast, primarily because while yields in percentage terms may decline over time, the absolute dollar value of the validator rewards may still increase over time.

In any event, to protect against the potential risk of weakened security, stake slashing as a result of “offline activity” can be used as a defense mechanism to devalue the forked network’s value and its corresponding artificial collateral. If the forked ETH isn’t bonded over time, it will lose enough value that the wider market consensus is to no longer support the forked network. This is because productive market participants will be able to proportionally grow their network stake over time by collecting a larger proportion of validator rewards at the expense of unproductive market participants.

How would this work in an interoperable world?

Suppose one wanted to collateralize Dai with say, BTC, instead of ETH. This user will need a way to have that BTC on the Ethereum blockchain. Enter Cosmos IBC (Polkadot would be another example). The user will need to take BTC on the Bitcoin blockchain, lock that BTC, and then prove the asset(s) are locked up / frozen on the Bitcoin side (ideally via some sort of smart contract). Then, the Ethereum blockchain will print a new token on the Ethereum chain that represents a claim to the underlying asset(s) on the Bitcoin side. Assuming that the asset(s) on Ethereum are 1:1 redeemable for the underlying asset(s) on Bitcoin, they should have the same value. Now, assuming this as an acceptable method of collateral, the user is able to use a tokenized representation of BTC to collateralize Dai. Upon CDP closure, the user will burn the asset(s) on Ethereum (which represent claims on the underlying locked BTC on the Bitcoin blockchain), submit a proof to the Bitcoin blockchain on the burn, enabling the Bitcoin smart contract to unfreeze the original BTC. While this enables greater cross-chain network effect growth (which can certainly enhance network value), it is unlikely (all else equal) that cross-chain claims to underlying assets in other networks (such as the above example with BTC) will be able to be used productively to support their native network from a security perspective (e.g., Ethereum in this case).

With the above risks (illiquidity of collateral and a network partition) potentially addressed, we can now envision a decentralized network that can autonomously support itself, even if just partially. This may spark a new generation of “capital recycling”; the base layer digital asset (ETH) flows up to enable applications (MakerDAO) built atop it, only to then flow back down (albeit partially) to contribute additional base layer support (network security). After all, regardless of whether a network introduces a “money-ness” element or not, the base layer (and the corresponding native digital asset) has to be “valuable enough” to support the applications built atop it. While there is no asset disposal per se (assuming, of course, that there is proper risk management and no flow disruption) as there is in the case of traditional market capital recycling, we can now envision a recycling mechanism whereby decentralized PoS networks are able to enhance not just the productivity of their own native underlying assets, but their overall network health and value as well.

Whether or not this kind of recycling initiative ever comes to fruition in decentralized networks remains to be seen, but enhancing network productivity to enable greater utility value is something that will help this asset class grow as there would be less reliance on speculation. Lastly, while the financial risks of the autonomous recycling mechanism above can likely be calculated with probable certainty, we should not ignore the magnified technical and systemic risks introduced as a result of the substantially larger coding requirements.

Grateful thanks to Dan Elitzer whose work has inspired this piece, as well as Michael Dunworth, Cyrus Younessi, Jason Choi, Tom Shaughnessy, and of course, Scott Army for the feedback and insights.

— — — — — — — — —

Disclosure: I personally hold positions in BTC and ETH, and indirectly through Vision Hill hold BTC, ETH and MKR. The content provided herein is of my own opinion and should not be considered investment advice, and is not a recommendation of, or an offer to sell or solicitation of an offer to buy, any particular security, strategy, or investment product. I have not received and will not be receiving any compensation as a result of this publication.