There has been a lot of talk about Bitcoin this year. Bitcoin saw a meteoric rise from $1,000 at the beginning of 2017 to over $20,000 in December. That’s 20x: 2,000%. The most astonishing part is that it wasn’t unprecedented. Bitcoin does this.
Accusations of ‘bubbles,’ ‘scams,’ and talk of tulips has kept the world outside the crypto space jaded enough to put off investing. Some brave souls have dipped their toes in, but nearly all of the activity has been speculation. Even in crypto circles, the common understanding is that Bitcoin has 1) no intrinsic value and 2) no fundamental basis for valuation. The case I’d like to make is that these two properties are not synonymous.
There is a vacuum in the crypto space: very few people are trying to fundamentally value cryptocurrencies. I want to prove that it’s not only possible to determine the value fundamentally, but rational to do so. Therefore the goal of this model is not to create a specific prediction, but to enable those who would like to justify their investment to do so according to the assumptions they see fit. Reasonable experts may disagree about the assumptions which underlie a valuation model, but that is to be expected. To be clear, this article poses a sound, rational framework for valuing cryptocurrency. This is not intended as financial advice.
Background
This article is the first of several to come in the Investor Series. These include concepts from economics, finance, and a few specific to cryptocurrency. I do my best to provide links to resources about these concepts, but if you are unfamiliar with any of them, please leave a comment and I’ll respond directly and by updating the article itself.
Additionally, it will be helpful to readers to have read some of the thought leaders’ work in this space, specifically Cryptoasset Valuations by Chris Burniske, whose work I’ve found unparalleled in the space. I’m also a big fan of Kyle Samani and Ari Paul.
In addition to this article, I’d also recommend these for context.
John Pfeffer’s Institutional Investor’s Take on Cryptoassets
Friedrich Hayek – The Denationalization of Money
Lou Kerner, Chris Burniske, Ari Paul – Cryptocurrency Conference Call
Valuing Cryptoassets
Cryptoasset valuation is a new field. When I’ve broached the subject with friends working in traditional finance, they’re a little quizzical.
So, do you just use a DCF? Comparable private transactions? Public multiples of some kind?
Not exactly. Cryptoasset valuation is a new field, and cryptoassets are a new asset class.
However, traditional equations are still at play. We’re not quite at a point where ideas are widespread enough to consider entirely unique valuation formulas for cryptoassets (but Willy Woo’s NVT ratio is a great start). As Chris points out though, a Black Scholes Model for crypto would be a sort of holy grail.
Of these traditional formulas, the most helpful equation in understanding cryptoassets is the Equation of Exchange, traditionally used to value currencies. Until now, currency creation has been the near-exclusive domain of central banks and governments. The difference with Bitcoin is that it’s not centrally created, planned, or controlled. Hence the term: decentralization.
While cryptocurrencies are new types of currencies, we are still able to use traditional metrics and formulas to measure them. One such formula is the Equation of Exchange, which is used to calculate aspects of a currency’s supply and demand.
MV = PQ
In essence, this means that the amount of currency in an economy multiplied by the number of times each unit is spent is equal to the amount of purchases, times the average purchase amount.
We call M the monetary base, or the amount of currency in an economy. M is most often calculated using the other factors which are more directly measurable. In solving for M, we usually break up the equation and move some variables around. Mathematically, the equation used to solve for M looks like M = PQ/V.
We’ll be solving for M in the model in order to estimate the necessary supply of Bitcoin to support a growing number of transactions over time. In solving for M, the monetary base, we’ll start from the right explaining each of these components beforehand.
P stands for the average purchase amount made with the currency.
Q stands for quantity – the quantity of “average” purchases.
V stands for velocity. The velocity of money measures how often a unit of currency is transacted, typically per year or per quarter.
The other equation we’ll be using in this model is a Net Present Value, which has 3 basic premises:
- The current price of an asset reflects expected future value.
- Future value is based on expected returns.
- Returns are proportional to risk.
Valuing Bitcoin
I’d like to restate that the assumptions included in this article are not meant to justify any particular valuation of Bitcoin. They will, hopefully, prompt healthy discussion of the drivers of value in the Bitcoin economy, and serve as a launchpad for future discussions of cryptocurrency’s value.
I took notes from Burniske’s model of the fictional INET protocol and included adjustable assumptions in my own. I think it’s important to justify assumptions wherever possible, and for this reason, the assumptions in my model have links in the comments of each cell, tracing them back to the strongest justifications I could find.
Of course, these assumptions are adjustable for a reason. Please include your own assumptions for any exploratory analysis. To change the assumptions, you’ll want to download or copy the Eat Sleep Crypto Bitcoin Valuation from Google Sheets, and revise that copy, as this one is view-only.
Overview
Models for cryptoassets are different than Discounted Cash Flows. With cryptoassets, there are no cash flows, or dividends; the asset itself appreciates.
This model is broken up into inputs, outputs, graphs, and backend calculations. As shown in the picture above, it is also color coded for traceability.
In the module labeled BTC Supply Inputs, the inputs for rows 2-7 are related to the supply of Bitcoin. I’ve also included a separate supply schedule sheet entitled ‘Bitcoin Supply Table,’ where cells are referenced in row 4.
% Hodl’d is the percentage of BTC held as investments (as opposed to being used for payments). “Hodl” is a widely used term in the crypto space. It’s a misspelling of hold, coming from this legendary post on a Bitcoin forum in 2013. % Hodl Liquidated is the divestment rate for invested coins. Assumptions are informed by data from a 2016 Coinbase survey.
Lost Coins is an estimate of how many bitcoins are inaccessible. The default assumption of 3 million coins is based on a 2018 study by Chainalysis.
The Economic Inputs module (in blue) includes inputs about economic factors , including worldwide GDP growth, and inflation rates. It also includes data about Bitcoin’s addressable market. For this example, we use the SWIFT payments network.
Bitcoin Valuation: Competing Theories
There have been a few theories about Bitcoin’s proper role in the world economy, and many debates even within the cryptocurrency community about this. I imagine the choice of the SWIFT network as a proxy for Bitcoin’s demand will be somewhat contentious, and to address that, I’d like to look at competing theories.
The first was the asset rotation thesis – that some percentage of another asset, namely gold will be displaced by investment in bitcoin.
A quick back of the envelope calculation:
190,000 tons of gold at $1,200 per ounce equals $7.3 trillion of gold total
Replacing 10% of investment in gold, total bitcoin supply equals $730 billion
$730 billion divided by 21 million bitcoin equals $34,700 per BTC
To the asset rotation thesis’s credit, we are starting to see evidence that bitcoin is replacing gold as an investment. However, this disregards Bitcoin’s network effect, and treats BTC as a commodity instead of a currency.
Side note: There has been much debate on Bitcoin’s properties as a medium of exchange versus as a store of value. It’s a topic I really enjoy, and one I can’t wait to post about in the future. For this article though, I’ve tried to stay out of that debate and create the best framework with assumptions most will agree on.
Another compelling argument for Bitcoin in the world economy is its use as a reserve currency. This argument is given by John Pfeffer in his paper An Institutional Investor’s Take on Cryptoassets.
As a cryptocurrency, Bitcoin is best valued with the medium of exchange equation. As a medium of exchange, Bitcoin has several addressable markets. While Bitcoin’s addressable markets are theoretically infinite, “large electronic payments” is sufficiently encompassing for this model. As a proxy for large electronic payments, SWIFT transactions are the most direct.
SWIFT
SWIFT, the Society for Worldwide Interbank Financial Telecommunication is the current system banks use for cross-border payments. Banks send messages back and forth to keep a system of credits. Bitcoin is a significant improvement from SWIFT. SWIFT operates only five days a week, takes days to settle, and users’ funds can be frozen by banks and governments. Bitcoin’s on-chain transactions are uncensorable, and faster than SWIFT settlement times. To replace SWIFT entirely, Bitcoin needs to be cheaper to transact.
SWIFT inputs are found in the Economic Inputs module, and in the SWIFT Payments Data sheet. The data comes directly from the SWIFT website. Predicted figures are based on historic growth of the network.
Also included in the Economic Inputs module is data on GDP growth, and inflation.
Adoption Curve Inputs
“A great technology company should have proprietary technology an order of magnitude better than its nearest substitute.”
― Zero to One: Notes on Start Ups, or How to Build the Future
The most subjective inputs in this model will be in the Adoption Curve Inputs module. These inputs represent your personal take on the scope of Bitcoin’s impact, and the amount of time that will take. For example, if you believe Bitcoin is a marginally disruptive technology, you may put 10% in for Market Share in cell B19. If you are a Bitcoin maximalist, you’ll likely put 100%, as you believe Bitcoin is an order of magnitude better than the existing payments infrastructure.
To model adoption of Bitcoin over time, I’ve used a logistic S-Curve function. The formula is used in row 19, columns E-Q. The Start of Fast Growth input is for the year you believe Bitcoin will have 10% adoption, which approximates the “tipping point” of adoption. After a tipping point, rapid growth follows. The take over time is the time it takes for adoption to go from 10% to 90%.
The specifics of the Adoption Inputs reflect in the Cumulative Adoption table, which factors into Bitcoin’s Current Utility Value as we’ll see next.
Synthesis
Now it’s time to calculate the intrinsic value of Bitcoin according to the Equation of Exchange. MV = PQ, if you’ll recall from earlier in the article.
These calculations are done in the spreadsheet in rows 18-28, but we’ll do them here in tables for simplicity, starting with values from 2018.
Our goal in using the Equation of Exchange is to solve for M, the value of the Bitcoin monetary base. An important catch here is that we’re solving for the actively used portion of Bitcoin’s supply. Coins held in paper wallets, for example, don’t explicitly affect the price of BTC.
To solve for M, we first want to input P into the equation. P is the average purchase amount of a currency. In this case, where Bitcoin is set to replace SWIFT transactions, we can take the Average Expected Message for 2018.
P = $558,018.69
Rounding that, we get:
M x V = 550,000 x Q
We’ll find Q next. Q stands for quantity – the quantity of average purchases. We also have the expected SWIFT messages per year, calculated in the spreadsheet using SWIFT’s data https://www.swift.com/about-us/swift-fin-traffic-figures. In 2018, Q = 3,485,852,902. We’re also assuming that Bitcoin is only taking a percentage of SWIFT transactions – according to default assumptions, 0.14% for 2018. We apply that to the number of SWIFT messages for a new (rounded) Q of 4,880,000.
Our equation now reads:
M x V = 550,000 x 4,880,000
We’re almost ready to solve for M, but we need V, velocity. To make things easier, we’ll move M to one side by dividing both sides by V.
M = 550,000 x 4,880,000 / V
You’ll recall V is a metric tracked by traditional economists, and published on websites including the St. Louis Fed. Bitcoin has turned a few heads in economist circles, and there are a few websites http://charts.woobull.com/bitcoin-velocity/ like this for Bitcoin metrics as well.
Bitcoin’s velocity fluxuates, but hovers around 5.5. Remember that we’re solving for the circulating and available supply of Bitcoin, so rather than using what Burniske calls hybrid velocity, we want to use transaction velocity. This is the number of times a particular unit of bitcoin is used in a year. To get this, we’ll divide the hybrid velocity of 5.5 by the percentage of bitcoin in circulation. Taking 1 minus our HODL % of 60% in 2018, we get 40%.
5.5 / .4 = 13.75
So our transaction velocity in 2018 is 13.75. Putting that into the equation, we have:
M = 550,000 x 4,880,000 / 13.75
We can now solve for M, the circulating portion of the bitcoin monetary base.
Multiplying and dividing these numbers out, we get:
M = 195,200,000,000
This means that to effectively process 0.14% of 2018 transactions from the SWIFT network, the circulating and available supply of Bitcoin would need to be worth 1.7 trillion dollars. Taking from our spreadsheet the circulating supply of Bitcoin in 2018, 5,362,500, we divide M by it.
195,200,000,000 / 5,362,500 = $36,400
Because of rounding, the model will be off by about 1% from our example, but the point stands. A per-Bitcoin value according to our assumptions would be $36,400.
With this same set of assumptions, the intrinsic value of one bitcoin continues to increase, roughly doubling for the next 8 years until adoption starts to level off in 2026. In 2030, after replacing 90% of SWIFT transactions, Bitcoin would be worth over $50 million per coin.
This is the intrinsic, or Current Utility Value (CUV) of Bitcoin. Currently, Bitcoin is trading around $7000, the market value. The difference between CUV and market value is analogy to the difference between book value and market value of a stock.
As I mentioned earlier, markets price assets according to future expectations. The Discounted Cash Flows method is used to price assets with foreseeable incomes. Bitcoin, however, has no cash flows. Instead, bitcoin itself is the appreciable asset.
Chris Burniske uses the term “Discounted Expected Utility Value”, abbreviated DEUV. We’ll use that. Bear in mind that most people aren’t valuing Bitcoin with any type of framework, so we should expect price discrepancies between DEUV and the market value as well.
To get the Discounted Expected Utility Value, we take a hypothetical End Year for our investment. In this case, we’ll say 2028, which gives us a 10 year holding period. Similar to a DCF model, we’ll also be using a discount rate. Discount rates are the average rate of return between the starting period of an investment horizon and the end period. Because risk is proportional to return, discount rates reflect the riskiness of an asset. Bitcoin is viewed as extraordinarily risky, so we’ll use a very high discount rate. In this case, 100%.
To get the DEUV for Bitcoin 10 years out, we take the Current Utility Value in 2028 and put in the discount rate to a Net Present Value formula.
NPV = 32,000,000 / (1 + 100%) ^ 10
Simplified
32,000,000 / 1024 = 31,250
This tells us that if the market held the same assumptions, Bitcoin would be trading around $31,250 in 2018.
Conclusion
The model accompanying this article was created with adjustable assumptions. However, there are many more factors surrounding the the ongoing development that should be considered. These include regulatory measures, the scaling debate, and the general perception of cryptocurrencies as scams, thanks to a few shady ICOs. Additionally, the lack of utility and intrinsic value of most tokens has left new entrants to crypto markets disenchanted with many 95% losses. Each of these problems have their root in a lack of economic understanding. For this reason, it is the goal of this series to highlight the importance of economic factors in cryptocurrency valuations. These factors are found at:
- The development level, where token economics must be taken into consideration by developers;
- The macroeconomic level, where macroeconomic trends in the traditional financial world must be accounted for;
- And to a lesser extent, the regulatory level, where regulation tends to (at least temporarily) affect the value and use-cases of particular cryptocurrencies. Being that blockchain development tends to outpace regulations and route around them, this is less of a factor than the previous two.
The ways in which these factors affect the prices will be explored for each currency in future posts. As for Bitcoin specifically, we have seen the public perception change and the use-cases for Bitcoin limited. The main issues facing Bitcoin are adoption, the ability to scale, and fading anonymity. The decision by the Bitcoin Core development team led to the blocks frequently becoming full in December 2017. This caused the fee market to become an auction system, where users had to bid for their transactions to be included by miners. At their peak, transaction fees reached an average of $41.
Obviously, Bitcoin is a less suitable medium of exchange with high fees, especially for smaller transactions. This is by design, to create demand for second-layer protocols.
The next article in this series will be published at the beginning of next week. Eat Sleep Crypto Investor Series #2 examines the effects of such protocols on Bitcoin’s price and adoption.