Originally published on Jan 16, 2018
As the fundamental analysis of token economies gains importance and more people come up with different models, the below offers my current thinking on some of the important aspects.
A utility token’s value (and price) depends on the investors’ time horizon, opportunity cost (discount rate) and velocity. I will address valuation models, discount rate and velocity.
KEY POINTS (TL;DR)
1/ Adding up discounted token values to derive current token value appears wrong, we should only discount one future value.
2/ Token economies (GDP) are not free cashflows, they are cash revenues. Cashflows accumulate to the investor every year, but revenues can only be monetized by selling the token to a natural buyer in the future.
3/ There will be a large gap in discount rates used by USD based funds and BTC/ETH based funds due to lack of (proper) risk-free rates. This will have a large impact on risk-adjusted returns and initial pricing of projects (including crowding out effects).
4/ Velocity will very likely not be the same every year, which complicates the analysis even further, we should use ranges.
UTILITY TOKEN VALUATION MODELS
1/ The currently prevailing valuation models use a mix of the medium of exchange equation (MV=PQ) to derive the token economy’s size based on adoption assumptions and then use a DCF-style (Discounted Cash Flow) forecast and discounting of the resulting monetary base.
2/ Traditionally, a DCF is used to value cashflow generating assets like stocks, real estate and other investment projects. Since we don’t deal with cashflows in tokens, one has to be careful how a DCF is applied. In a DCF, we use tangible cashflows, which, in the context of a token economy can be thought of as cash revenues (can also be viewed as the economy’s GDP). For example, the sum of all transactions in the file-sharing industry for 1 year.
3/ In a DCF, once the CF is earned by the investor, it belongs to him/her every year. It stays the same in USD terms. In crypto, the market cap is constantly “generated”, and captured in the value of the token (via natural supply and demand for the service). However, the token price may not be stable due to market forces, cross asset valuations etc. So, to monetize the token value, the GDP has to grow, the token has to capture the increased demand for you to sell to someone else (a user of the service for example).
4/ However, to sum all discounted future utility values as seen in some earlier models is incorrect as the generated volumes in a token economy do not add up like cashflows in a DCF, they can only grow or contract based on the size of the GDP. You should discount only one future token value and then break it up into current and future utility value (akin to the present value of cashflows and the terminal value in a DCF). The great
has a practical valuation model here. It only makes sense to add up the discounted token values if you are an equity investor in the respective protocol/dApp developer and receive cashflows in the form of fees or some other mechanism, which accumulate to the company.
5/ As a result, the token economy has to grow by at least the cost of capital for an investor to realize and monetize their investment in X years, which brings us to another very interesting aspect — the discount rate.
1/ In the absense of cashflows, and, given everyone has objective real-time data about a token economy, a major driver in the investment decisions will be the discount rate, which investors initially use.
2/ As token prices move, the implied discount rate relative to the fundamental token value will change. This may then trigger exits as the investor’s required return is reached and a position can be liquidated.
3/ For traditional fund mangers, the discount rate is the opportunity cost they can get on the global asset market, adjusted for the respective risks for the asset class. So far, the assumption seems to be that crypto will be viewed as an asset class similar to stocks, bonds or commodities, meaning they will likely use some version of a traditional CAPM + VC risk premium (to reflect early stage risk) + even more risk premium for the higher risk/volatility of crypto tokens. This will likely add up to a number easily reaching 30–40%.
3a/ Important note: this may not necessarily apply for Bitcoin. There are other assumptions that can be made if the narrative of bitcoin replacing gold as a store of value continues (which seems to be the case nowadays). The volatility of Bitcoin will actually reduce portfolio risk due to the very low correlation with other assets. That is, until everyone gets into Bitcoin which will lead to correlations increasing somewhat.
4/ A far more interesting question is what the discount rate should be for a crypto based investor, meaning a fund which raised money in crypto and returns money to LPs in crypto. Assuming that the funding happens in BTC or ETH, I would argue that such a fund would use a crypto based “CAPM”. This will be for people who decide to base (part of) their life on the new digital economies that Bitcoin and Ethereum have created. Arguably, this is not the case today, but may become the case in future.
4a/ An interesting question would be if people would keep their money in ETH instead of USD/EUR, because, by staking, they should be able to receive enough fees to cover the ETH annual inflation and hence have no reason to move back to USD/EUR. Same for BTC, where, being deflationary over time, the store of value function would keep up with the fiat inflation.
5/ In this case, evaluating a new token economy should happen relative to BTC/ETH and the implied discount rate would be much lower for such investors as they would only care about the specific token risk premium (similar to the equity risk premium concept in traditional investing). For comparison, a traditional investor who evaluates the same token will have to start from a much higher base discount rate for crypto. This is a huge gap which can have very interesting crowding out consequences.
6/ As a quick example, by using
’s sample model, if you have an investor who’s discount rate is 40% (which seems a fair assumption for institutional investors given that the private equity and venture capital asset classes typically enjoy ca. 20–25% expected returns), the INEA token’s utility value given a 5 year holding period is 0.32 USD. By halving the discount rate to 20% for a crypto-based investor, the implied discounted token value goes up to 0.69 USD. What this means is that, based on the initial token price, either a traditional fund will find it impossible to invest or crypto funds will enjoy significantly better risk-adjusted returns.
7/ This exists on the traditional financial markets. Let’s assume an investment fund has the USD as its base currency. Their required return would be the USD risk free rate plus a premium of some sort, depending on what assets they invest in. If they want to invest in a Brazilian stock for example, then they would take their own required return, add a premium for the Brazilian Real plus a premium for investing in the stock. Compare this to a local fund who’s base currency is the Brazilian Real and they would require only the company specific risk premium over their own required return (cost of capital). Now, the local player’s risk-free rate will be higher which should equalize the required returns somewhat. In the end it will come to two main factors — the ability and cost to hedge out any FX risk (say from USD to BTC — which will be hard as there are no large liquid longer-term forward markets) and the required return of the LP’s in the respective funds. All else equal, I would argue that BTC/ETH based investors would not really have a risk free rate. Why? Because BTC is deflationary in the long-term and in the case of ETH, longer term staking fees should equal the long-term ETH inflation. Hence, crypto-based investors would only have token-specific risk premium as their required return. Outside investors would have a BTC/ETH risk premium >0 and then would add any token-specific premium.
8/ There is an optimal holding period in a token economy based on the opportunity cost of the investor. Please note in Chris’ example above that even though the nominal token value grows over time, there is an optimal discounted token value, which is not the one furthest away in time.
The last major factor to consider is velocity.
1/ By using MV=PQ for the establishment of a value (not a price) of a utility token, it becomes apparent that one of the key variables is velocity (V). All example models posted online by the various analysts make the assumption of a constant velocity for the forecast period (usually for simplicity reasons).
2/ Much thought should go into the V number, which is often compared to the USD M1 velocity (for the case of bitcoin and other cryptocurrencies), the velocity of mobile money (M-Pesa in Kenya as one example) for fintech tokens etc. Another leg of the analysis is that the tokens held by hodl’ers have a velocity of 0 and the ones used for transactions can have a very high velocity. For some utility tokens, there are arguments that the velocity can be so high so that the underlying token value approaches zero.
3/ One obvious question is whether V should stay the same every year? This is objectively hard to assess, so a sensible thing would be to have a range of possible V’s for each year based on analysis of the various factors affecting V.
4/ Let’s look at some of them:
- Factors, known upfront ahead of the ICO/TGE, like % held by founders, advisors and early investors along with the vesting schedules, % held in a foundation (or a similar vehicle) and used as token inventory along with any release schedule. These are the obvious ones, which can be factored in easily.
- The next set of factors are about bonding/staking by nodes or similar actors in the token economy. Again, we might have upfront information about the mechanism design of the economy and the incentive structure. I would further argue that we can see two sub types of stakers: internal and external to the token economy.
- Internal are the stakeholders who are a natural part of the economy, for example, a transportation company, which holds tokens of some imaginary shipping blockchain in order to use it for their business.
- An external stakeholder would be an investor, who holds these tokens as a node because they receive some sort of reward for this — fees for example.
- The internal stakers would have a constant V=0 as long as they are using the blockchain. The external would have V=0 as long as the opportunity cost on some other blockchain is higher. For example, if the investor calculates that he can get a better return by being an ETH miner or a FileCoin node for example, their V will go from 0 to something else. The above depends on the price of the token vs. its value.
- Additionally, there will be different V’s for transactions vs. trading: users (very high V) vs. speculators (high, but lower) vs. investors (lower, but not 0).
- Final factor to mention is the token price itself and the reflexivity loops caused by market participants nudging the prices up or down (Vitalik’s thoughts on this are summarized here). Let’s explore:
> if prices are nudged upwards by speculators, then some 0 velocity volume will come to the market and sell the premium to current+future utility value. This will increase velocity, thus decreasing token prices, all else equal.
> if prices are driven down by shorting, then the natural demand will kick in to bring back the token price close to the current utility value, unless the demand is satisfied by increased selling from hodlers who cut losses and inject more velocity in the system. This negative loop can continue until the hodl only equilibrium price is reached where natural bonders’ demand is activated and/or the implied discount rate (expected return over X years) becomes attractive enough for new investors. As an example, think of Maker’s incentives for MKR holders to support the price (see here for more on Maker).
> stakers/bonders also have an opportunity cost when they are external to the natural token economy — so they will tend to go from 0 velocity to >0 as and when the opportunity cost between protocols/economies makes sense.
5/ It is evident that in reality there will be a range of current utility prices which make sense based on varying velocities and they are updated in real time with the market moves. Analyzing the incentive structure and the mechanism design of a token economy is paramount.
A few final thoughts for investors:
1/ Token economies offer the opportunity for investment funds to help grow / develop a token ecosystem and get a liquid stake which can be bought/sold at any time. Today, as an institutional investor, you can buy a small stake in a listed company and you can’t really help its development, you leave this to the management. Unless you are a private equity fund.
2/ PEs will suddenly find themselves with the opportunity to do what they know well —actively support their portfolio companies and entering the unknown waters of liquid markets, so far left to hedge funds. PEs will likely need to hire traders. Will we have PEs transition to HFs or vice versa?
3/ Several investors acting together can bring more value to a protocol by supporting its development and growth. This may work even better than being co-investors in a company where there are active and passive investors.
4/ Activist investors from the traditional finance world will find crypto very welcoming. It is liquid, offers vast opportunities and they can easily do their thing by forking.