Risk frees and currencies: Part 2
A hot, giant pool of money.
I have no idea when this will publish, because I’m always so lackluster at finishing things.
Right now, crypto is a bastion of financialization and speculation, with limited true, practical use cases. Most real world cases tend towards the bizarre or superfluous (NFTs) or provide some mechanism to flout someone’s laws (pseudo-securities, tokenization of real world assets, cross-border lending). There is no basal economy in crypto so to speak — the morass of market makers, lenders, and financial products all belie the one primordial truth — there is a fixed supply of dollars converted into digital poker chips, and the object of most protocols is to attract those dollars for as long as possible. Some poker chips receive a certain level of favoritism above others, largely based on either substantial adoption or some proof of capture of real world value (e.g. VC money being pumped in). We can, for the most part, imagine the crypto universe much like real world currencies, existing in a spectrum between the hard monies and soft (or even fictitious monies).
Intuitively, we all know that dividing an object, save Banach-Tarski spaces, doesn’t increase the amount of the object available. Money is no exception, despite the mental gymnastics we often tell ourselves. Money, at its core, is a representation of the value accessible in the world it operates in. It is simply a future claim of value, and principally speaking, the creation or destruction of money itself does not change the total value in the system.
Many people who subscribe to methods like on-chain analysis try to decipher whether a holder is a speculator or a true believer/user. True fundamental, or no-trade, value occurs when actual utility is gained from buying an asset. When I purchase food to eat, that’s fundamental value. When a company creates a new widget for consumers, that’s also fundamental value. Conversely, if I purchase something with the idea that it will be worth more later (speculation), that is not fundamental value.
Fundamental value cannot be created from nothing. Technological revolutions unlock new sources of value; arguably, the things we hold dear to us would look alien to a person from 100 years ago. But the act of money creation is not an act of fundamental value itself — or more specifically, it has rapidly diminishing utility.
The initial act of monetary creation is a net benefit, allowing more divisibility, fungibility, and liquidity for economic exchange than a non-money-based system (e.g. barter). Similarly, infrastructure that provides means for the easy transmission of money (e.g. online payments) is clearly a source of economic value. Removing friction from transactions is a source of value - up to the point of redundancy, of course.
For crypto up to this point, it is hard with a straight face to argue that the primary driver of crypto market capitalization is new fundamental value creation. While I am a proponent of crypto, and think that certain innovations like decentralized finance hold immense promise, the new creation of a coin or protocol to more efficiently provision money for yield generation is not itself a source of fundamental value.
The way I personally view the current ecosystem in cryptocurrency is akin to a hot, giant pool of money. Regardless of inter-protocol operations, we can argue that behind the active facade sits some measure of true economic value, which primarily occurs from the exchange of real world assets (including fiat currencies) for crypto-assets. At any given point in time, there is some amount of real value (which we could, in theory, track via inflow/outflows from fiat onramps for example) which backs the total value of all crypto-assets. With few exceptions (including, for instance, economic value unlocked by NFTs or certain DeFi innovations), we can argue that this represents the total fundamental value in the system, with the rest being purely liquidity (speculative).
This is very similar in short timescales to real world currencies. We can view foreign exchange as a closed system, where most economic value on Earth is represented by the total value of all money and money-like instruments present in the world. Regardless of money creation, the economic value backing the money/money-like assets does not change. If I represented all the world’s wealth with $100 in USD, it would simply change the exchange rate between hard assets (like bread to eat) and money, but wouldn’t create or reduce wealth (it would actually marginally reduce it, due to reducing liquidity, fungibility, and divisibility).
With the exception of true world economic growth, the value of currencies in reference to one another is very much, in crypto terms “player versus player”. A currency devalued by monetary creation in isolation will lose value in relation to other currencies, which we can easily observe in cryptocurrency dilution as well. It makes little sense, with the exception of new inflows, to view fantastical yields such as those provided by a Wonderland or OlympusDAO as anything more than mathematical abstractions, if the currency itself is being devalued at a similar rate. The total value (e.g. a market capitalization) doesn’t change.
One of the most interesting innovations of cryptocurrency is, however, the immense liquidity it provides, partly by design, partly via flouting the norms of traditional financial regulations. Cryptocurrency provides many of the same benefits as online payments, but without the cruft and middlemen, who tend to act as guardians to keep money from escaping. In the normal economic paradigm - for example, a bank - there are structured checkpoints for money to preserve overall system liquidity. When these fails (e.g. a bank run), the central bank acts as a lender of last resort, backstopping liquidity and keeping the whole system from essentially imploding. This fundamentally occurs due to duration mismatch - banks are in the business of lending out deposits. Lending (loans) tend to be long in duration, while deposits, with the exception of certain products (e.g. certificates of deposits) are short in duration (deposits can be withdrawn at any time).
In crypto, there are usually no such gatekeepers. As I noted in my article on DeFi risk, most existing DeFi lending skips over this issue entirely by only providing overcollateralized lending, where the borrower and lender are essentially the same individual. This hand-waves the issue of duration mismatch — the deposited reserves backing the promissory token are present for the entire lifetime of the loan. However, this doesn’t capture most of the problem.
For protocols or even the underlying blockchains (L1s) themselves, the real fundamental value behind them is the critical driver of success. As Yogi Berra once said,
Nobody goes there anymore. It's too crowded.
On a protocol basis, attracting inflows is not just preferred in most cases, but the difference between life and (slow and painful) death. For a market maker analogue (e.g. an automated market maker, like Uniswap or Osmosis), liquidity in trading pools critically provides market depth, allowing realistic approximation of market spot prices (without liquidity, market impact on any substantial order would be extreme). For a lender, pseudo-custodialized reserves (e.g. a MakerDAO) are important for backing the value of the promissory token.
For a blockchain, attracting real fundamental value is needed to ensure security and stability. In proof-of-stake systems, this is intuitively obvious in how staking works. As token value increases in fundamental value terms (e.g. USD or another ‘real’ currency), it becomes prohibitively expensive to attack the network. Similarly, higher token prices incentivize continued staking and creation of validators, which supports network stability. For proof-of-work systems, this is a second-order relationship — given that hash rate is largely a function of real world expenditure (e.g. computing power, electricity) while the payout is in the token itself (new Bitcoins mined), high token prices provide incentive for increased mining participation and network security.
However, as previously touched on, cryptocurrencies provide unbridled liquidity, not checked by real world annoyances like capital controls, AML/KYC, or regulation (yet, obviously). Capital can flow incredibly quickly between protocols and blockchains, mostly encumbered by poor user experience or difficulties operating between blockchains (especially without invoking a centralized exchange intermediary, which may increase risk or have certain rules themselves).
This is a problem fundamentally, because the velocity and volatility of capital flows directly relates to risk, as noted above. It becomes in the most extreme case a game of musical chairs, especially in relation to staking risk and network security. While at high levels of staking or pool liquidity outflow may marginally impact stability, this accelerates exponentially at terminal levels. Similarly, this has a chilling effect on a protocol or blockchain’s future longevity — with few exceptions, a compromised chain or dead protocol stays dead forever (at least so far).
In the absence of true fundamental value, the primary mechanism driving inflows and outflows must be market prices, and the ability to generate returns in exchange for risk. As we noted, at any given time, there must be a fixed amount of ‘real value’ in the crypto ecosystem, and most inter-crypto inflows/outflows likely occur at the expense of other cryptocurrencies (both on a protocol and blockchain basis). Arguably, blockchain interoperability provides frictional barriers to capital flows at the current time, but the advent of bridging and “LayerZero” solutions aims to solve this.
This isn’t by definition a bad thing — all investments involve an exchange of risk for return. However, as liquidity becomes absolute - where any coin or token can be exchange for any other - we approach a theoretical equilibrium. For a rational actor, in the absence of capital constraints, we would anticipate selection of protocols/blockchains that at a given time provide the highest expectation of risk-adjusted return. The key here, however, is risk-adjusted. Let’s think about it.
The Risk Free in Crypto
Arguably, in USD terms, there is no trade in crypto that maps to a conventional understanding of risk-free. Holding Bitcoin itself, for example, has significant risk, which can easily be seen by the asset’s volatility versus conventional assets like real estate, equities, or fixed income. That said — we have a sleigh-of-hand here.
As noted in the previous article, we can invoke a soft understanding of risk-free for investments by considering currency-specific risk-free rates. In this vein, Treasuries act (or similarly, the overnight rates) can be viewed as the USD-specific risk-free rate - unless by choice, the US Treasury cannot default on its obligations, since they are in its own currency. Any solvency risk the Treasury has is essentially felt by all USD holders; arguably, excess risk from holding a Treasury to maturity (with the exception of mark-to-market risk due to yield changes) is negligible.
Many view the basis trade to be the fundamental risk-free rate in crypto. The basis trade traditionally is defined similar to the FX case — going long the spot of a given cryptocurrency (usually Bitcoin or Ether) and short either the matching future or perpetual future (perp, for short). Historically, this has been a fairly profitable trade; while there are periods where the spread turns negative, in general it provides fairly high returns for low risk (compared to other asset classes). We can see the implied premium looking at the Binance-ETH basis below (source: Coinglass):
Conventional understanding of this trade implies the premium is primarily due to demand for long leverage, which is consistent with increasing basis during periods of crypto bull runs. However, this trade arguably does bear risk, albeit less than non-market-neutral strategies. Intuitively, if this was due to market demand alone, increased basis rates would quickly attract new traders, leveling out bases fairly quickly. Similarly, we have a problem explaining cross-asset basis differentials at the same time and on the same exchange. While we could argue that stylized facts about an asset - including its volatility (and hence attractiveness to margin due to mark-to-market liquidation risk) and utility - may impact relative basis, it still seems to provide unsatisfactory explanations alone. Consider, for example, the basis (note: LUNA largely doesn’t have quarterly or monthly futures, so the primary ‘basis’ is via long spot, short perpetuals) provided by LUNA:
Compared to ETH:
We can see in ETH’s case, the funding rate largely seems to track spot price changes, allowing us to fit the idea of short or long-dominated leverage needs. However, for LUNA, we actually see the reverse - as price increases, the basis becomes sharply negative. This isn’t intuitive, unless we believe people just get really angry and demand to short and lose money. While many hand-wave this as a side effect of LUNA’s unstaking period (21 days), we see negligible spot/funding correlation with ATOM, which has an identical unstaking period to LUNA:
Surely, you must be joking, Mr. Feynman.
So, viewing the funding rate or implied futures basis as the fundamental currency-specific risk-free rate already has issues. Even in the case where spot/funding correlate, the fact that the basis rate can be negative should give us a clue that our math is probably not right here. This implies a risk-free rate with spot sensitivity — hardly risk-free.
Moreover and perhaps more insidiously, different exchanges provide different basis rates. This is also fairly intuitive - holding money on an exchange has risk (from hacks, counterparty risk, etc.)! It is not by itself risk-free. This is doubly so for cross-exchange basis trading — if one exchange decides to Quadriga (abscond) with your trade position, you become essentially unhedged. Hardly risk-free.
So if the basis trade is not the risk-free rate in crypto, what is?
As we implied, in most staking-based cryptocurrencies (protocol and blockchain), we do have a source of return that isn’t by itself more risky than hodling alone: staking.
Real Staking Yields
In the traditional proof-of-stake paradigm, currencies can be assigned to validating entities, which provide the staked currency as collateral. Validators are allowed to create new tokens in proportion to their holdings of the currency, with the possibility of loss (slashing) in cases of validator malfunction or malfeasance.
This intrinsically ties to the security of the network, as mentioned above, to the fundamental (real) value backing it. In an essentially worthless blockchain (where the token price may have many, many leading zeros), the cost for malicious action (including double spend) is marginal. The chain becomes insecure.
A principal problem to solve is bootstrapping initial liquidity, because, in essence, liquidity attracts liquidity. The current mechanism for this is largely from venture capital funding — by providing large development grants and airdrops (literal token rewards granted for holding or doing an action on a chain) to current or near-future builders and token-holders.
This hopes to accomplish two goals. Principally, the immediate reaction of increased funding is capital inflows, as liquidity is drawn from both intra-crypto and external (although likely more from other protocols and blockchains than from fiat itself) sources. Secondarily, by providing investment incentives to builders, chains can hope to achieve “value ex nihilo” - the creation of true, future fundamental value, helping provide stickiness to on-chain liquidity.
For both proof-of-work and proof-of-stake chains, the compensation to secure the chain (and help provide said on-chain liquidity/keep fundamental value) comes from two sources — transaction processing and inflation. For most chains, inflation is the primary mechanism today — stakers in PoS, for example, achieve return principally by taking a larger and larger piece of the true ‘fundamental value’ backing the blockchain, as their proportion of the market capitalization increases. This is not entirely but close to zero-sum — while stakers do provide network security to non-stakers, in the absence of inflows, their returns come at the expense of non-stakers.
In the future of course, most chains intend to provide most staking rewards as a function of transaction fees, rather than conjuring fictitious money. This relies on the idea of value ex nihilo - by conjuring money today, we can spur development for tomorrow, spurring actual transaction volume. Whether this will succeed remains to be seen.
However, for our purposes today, the actual source of yield in this case is a marginal concern (unless you have an extremely long time horizon). By invoking our definition of currency-specific risk-free rates, we can see that staking on PoS is perhaps the closest candidate to a true risk-free rate.
More specifically, when we talk about return, we must account for inflation. On PoS chains where the staking proportion is well under 100% (e.g. all coins are staked), we would expect that a staked coin will produce a real return after inflation, given that stakers bear the inflation rewards. If all coins are staked, we would expect the real staking yield to be nearly zero — while the supply of coins will increase, the relative proportions between actors will remain the same.
Real staking yield = Nominal Staking Yield - Inflation
Many coins starting out, for the purpose of attracting initial inflows from speculators, will promise extremely high yields. This is intuitive in the value of staking - the marginal value to network security is high for low levels of staking, and minimal for high levels of staking. If this presents a high risk-adjusted return opportunity, in a low friction market, we would anticipate this should attract inflows (at the expense of outflows elsewhere). This will increase in the short-term the exchange value of the coin relative to other currencies - identical to our understanding of covered interest rate parity! However, as inflows pick up and the proportion of stakers increases, the rewards drop precipitously. At high enough levels, this reward may start to become unattractive, leading to currency outflows (and an implied equilibrium state).
Notably, real staking yield can be observed to be nearly risk-free denominated in the currency itself. If I stake ATOM with a validator, my risks will be nearly identical to holding unstaked ATOM. Any compromise or loss of the network itself will impact staked and unstaked ATOM nearly equally. The exception, of course, is slashing.
While not all PoS networks implement slashing, slashing is a method of penalizing validators for misbehavior at the expense of their collateral. If I place my ATOM with a malevolent validator, I will be liable and incur loss (in ATOM) if it goes rogue. However, importantly, this is a diversifiable risk. We should not expect, prima facie, the market to reward stakers who stake with riskier validators. While riskier validators may provide lower commissions (and look like ‘better returns’), we can, in theory, provide equal or proportional stakes to all extant validators, fully diversifying our slashing risk.
Even if we provide our stake equally, however, there is still some basal implied risk of loss from slashing. We can call this the slashing risk premium. Are there any more risks compared to holding the currency itself? As far as this author can tell, not really. In certain networks that implement unstaking periods, we could arguably view the unstaking period as a risk factor as well, which can be neutralized by suitably shorting the forward to remain neutral to spot price changes. This implies that the real staking yield, accounting for diversified slashing risk, should be as close to a currency-specific risk free rate as possible.
In networks that do implement slashing (notably, Near Protocol currently does not), we can more formally write out the components of the currency-specific risk-free rate as:
Currency-specific RFR = Nominal Staking Yield - Inflation - Diversified Risk of Slashing
However, we have to remember that our currency-specific risk-free rate is, at its core, a sleigh of hand. Most of us cannot pay our liabilities in dog coins.
How do we compare it to our conventional understanding of risk-free rates?
In practice, we can consider changes in exchange rates over the lifetime of our investment. If I buy ATOM for $5 and stake in such a way that I double my holdings in ATOM terms but the spot price drops to $2.5, did I make money? As I live in USD, I did not.
However, we have to be mindful that the spot price of a currency reflects explicitly its inflation rate. If the supply of ATOM doubles globally with no inflows or outflows, we would expect the spot price in USD to halve. However, if in the same period the supply of USD also doubles with no inflows or outflows, we would expect the spot price of ATOMUSD to remain stable. If we compute the FX risk (changes over the investment lifetime to the spot prices of the two currencies), we have to account for relative inflation changes as well (or else we may double-count it). This can provide us a method to convert a crypto RFR into a cross-currency RFR (and becomes more practical if forwards exist, allowing us to lock in explicit exchange rates - the basis trade).
In the absence of capital constraints, we would expect some form of covered interest parity to emerge, even between fiat and cryptocurrency space. However, in practice, this is a long time away, if ever.
Since I’m both tired of writing and I’m sure you’re tired of reading, I’ll stop this article here. The main takeaway is this — the act of money creation itself is not an action that creates fundamental value. We can note that, at present, the fundamental value embedded in crypto is fairly small and fixed at any given point in time, owing in practice due to difficulty on-ramping/off-ramping fiat currencies. This implies that the primary mechanism of inflow/outflow attraction (and hence exchange rates) should be, in a frictionless environment (inter-blockchain is hardly frictionless at current, but may be in the future), the attractiveness of risk-adjusted yields available. Fundamentally, the risk-free rate in a given currency space should be the real staking yield, adjusting for FX risk.’
In the next post, I will talk about a method to neutralize FX risk (and hence provide a quantitative approach to measuring a ‘truer’ crypto RFR), as well as some ways to imply interesting risk premia values (specifically, at a network and protocol level).
Ciao for now.
Lily
You write so clearly on this stuff Lily
Awesome read. Thank you.