Quick overview of the ecosystem:

• Waves: base-layer chain

• WAVES: native token of Waves

• Neutrino: protocol that issues stablecoins

• USDN: stablecoin pegged to 1 USD issued by Neutrino

• NSBT: Neutrino’s token

• Vires Finance: money market on Waves

• VIRES: Vires Finance’s token

For more details on the above protocols or the previous incident, read my previous piece.

Current status of the ecosystem

The USDN stablecoin has almost been restored twice in the past 3 months, but that doesn’t make it a “stable” coin. Even though it didn’t crash like UST did, USDN fails to peg around $1 during market turmoils like a stablecoin should. As the team stated, it has almost been restored, meaning USDN hasn’t been restored to $1 or even ~$0.99 yet. Some updates on the Waves ecosystem: TVL is down 78% from ATH, WAVES is down 87%, USDN market cap down ~22%, Vires Finance TVL down ~47%, Vires stablecoin pools have > 90% utilization and liquidity cannot be withdrawn. The USDN/3Crv pool remains unbalanced, with ~70% in USDN, far from the ideal 50/50.

The 2 main problems waiting to be solved are 1.) fixing the peg and 2.) solving liquidity issues on the money market. The team has been taking steps and coming up with proposals since the first depeg situation happened in April, now they are rolling out a second phase of their revival plan to further fix the system.

The revival plan — fixing the peg

• less NSBT required to swap USDN<>WAVES

The arbitrage opportunity between USDN<>WAVES is limited to NSBT stakers (which are gNSBT holders), this limitation weakens the ability for arbitrageurs to fix the peg when price deviates from $1. The team altered the parameters for calculating the swap limit per 24 hours in early April, increasing the amounts users can swap per NSBT staked. This strengthens the peg naturally, but as seen in the UST/LUNA incident, arbitrageurs may not be able to absorb the large sell-off pressure under rapidly deteriorating confidence in the system. Under extreme circumstances, people will lose confidence in the entire system, leaving no incentive to hold either WAVES or USDN, therefore most would sell directly on the market rather than swap USDN<>WAVES to maintain the peg.

• introduce new recap token that recapitalizes when USDN is under-collateralized

USDN are minted 1:1 in value by locking WAVES, but currently the backing ratio is about 37.8% (~$515M in deficit), which is severely under-collateralized. In its original design, NSBT are auctioned for WAVES to fill the deficit in collateral. After the backing ratio is above 100%, these NSBT can be liquidated for more than the value they paid for. Similar to purchasing bonds from the treasury for their recapitalization, but the maturity depends on when the backing ratio returns to > 1. However, ~95% of NSBT are staked for the ~50% APR (the lock period is 45 months, you will lose some amount if unstaked earlier) and they couldn’t mint more tokens as it has almost reached the max supply (~2.8M).

They are introducing a new recap token to substitute the current NSBT for recapitalizing when the backing ratio is below 1. Once the new recap token launches, users would be able to exchange their locked NSBT for new recap tokens. They basically have the same utility, but difference in how they share revenue from the platform (not relevant here). However, I believe not many people will be willing to buy this *bond *from Neutrino when the market has lost faith in the system -> WAVES market cap < USDN market cap -> deficit exists so need to issue recap token to recapitalize. Since these “bondholders” would only be able to redeem if the backing ratio returns to 1, people would need to have a lot of faith in the system when the market doesn’t. The recap token has barely any difference from the original design, so I believe it wouldn’t have much effect on strengthening the peg. Even if most NSBT weren’t staked and could be auctioned, they would need almost 10x current NSBT market cap(~$50M) sold to fill the ~$515M in deficit.

The revival plan — solving liquidity issues

• set daily withdrawal limits on USDC, USDT

• introducing dynamic limits based on pool utilization rate

• liquidity position lock for 3–6–12 months

~$600M of liquidity has been withdrawn from Vires Finance since the alleged manipulation was exposed and stablecoin depegged in April. Currently there is still ~$563M debt in stablecoin pools (~$25M in USDN, ~$249M in USDT, ~$289M in USDC), in which the utilizations are above 90%. Depositors were and have been unable to withdraw their funds, even if liquidity is paid back they were mostly withdrawn immediately by bots. The following 3 plans are meant to restrain the withdrawal ability of bots, and prevent bank run situations like the previous ones. The third one would be locking liquidity positions to receive VIRES as rewards, this would decrease users’ ability to withdraw funds under extreme circumstances. The team mentioned $20M has been locked after a day of inception, this would account for merely ~1.6% of total value in the protocol. These are ways to mitigate sudden withdrawals based on fear or lack of confidence, but the fundamental issue may still lie in strengthening the USDN peg and restoring confidence in this ecosystem.

• Begin buying and locking CRV tokens with 45% of the WAVES staking profits from Neutrino, and vote to incentivize the USDN/3Crv pool

This would deepen USDN liquidity, which mitigates imbalance pool situations when large amounts of USDN are sold-off. The team stated the goal of locking 1–3% of CRV to have sufficient voting power in the distribution of rewards. However, I believe most people wouldn’t deposit into imbalanced pools even if the APY is high.

• Liquidate large accounts, Sasha taking control over their collateral and returning liquidity back

This proposal would allow Sasha Ivanov (founder of Waves) to take over the debt position of a few whale accounts with a total ~$400M. Sasha will then liquidate their collateral, sell USDN on the market and return liquidity back to Vires for users to withdraw. The large amounts of USDN have to be sold in stages without crashing the price, which would be handled by Sasha and his team. Although liquidity would be withdrawn once it returns to the protocol, this would probably bring Vires back to a more healthy state.

Evaluating USDN with Vitalik’s two thought experiments

Vitalik recently published a piece on evaluating automated stablecoins, perhaps as a reminder of the fundamentals of these “algorithmic stablecoins” after the LUNA crash. The following are two thought experiments that he would go through when evaluating whether an automated stablecoin is truly “stable” or not. In the article, he used UST and RAI as examples for the 2 experiments. (RAI is a non-pegged stablecoin with ETH as collateral)

1. Can the stablecoin, even in theory, safely “wind down” to zero users?

He mentioned traditional companies shutdown all the time without really hurting their customers, but in the crypto world users lose huge amounts of funds when a protocol/project collapses. I’m going to go through two scenarios, both causing demand to decrease in USDN, but one has a larger impact on WAVES, the other hits USDN more.

1.) If WAVES market cap drops more than USDN -> the deficit between USDN and WAVES market cap increases -> NSBT/new recap tokens are auctioned off for WAVES at a discount to fill the deficit -> less circulating WAVES

This circumstance is under the premise that people have enough confidence in the system to purchase *debt *from Neutrino. If not, there wouldn’t be a “death spiral” like the UST case, but USDN would stay undercollateralized like the current status (~37.8% backing ratio).

2.) If USDN market cap drops more than WAVES, and USDN < $1 -> smart contract detects an excess of reserves -> generates a corresponding amount of USDN to buy back the NSBT liquidation orders -> increase USDN supply -> USDN price drops further until no liquidation orders are in queue

If there are enough NSBT liquidation orders in queue, USDN would go down even further. USDN will be overcollateralized, but restoring the peg would still rely on USDN<>WAVES arbitrage opportunities.

2. What happens if you try to peg the stablecoin to an index that goes up 20% per year?

The second thought experiment is to evaluate the sustainability of these stablecoins, whether they would be able to track a hypothetical index that goes up 20% a year. The two possibilities that Vitalik laid out for this to happen:

1.) It charges some kind of negative interest rate on holders that equilibrates to basically cancel out the USD-denominated growth rate built into the index.

2.) It turns into a Ponzi, giving stablecoin holders amazing returns for some time until one day it suddenly collapses with a bang.

UST belongs to the second and RAI belongs to the first, USDN would belong to the second as well. Vitalik is suggesting a negative interest rate to be able to respond to zero interest rate situations.

I would say USDN doesn’t really pass these thought experiments, and as the previous two depeg incidents show, USDN isn’t stable during extreme events.

I think the Luna crash gave us a lesson, to examine algostables with extreme circumstances and of their fundamental design. The Waves team has put a lot of effort into this revival plan, but only faith of the market in the system could save the peg, rather than the stability mechanism (arbitrage) they designed. The overall plan could bring the ecosystem back to healthy and stable states, but I believe it’s not sufficient for USDN to survive another crisis without depegging. (None of this is financial advice)

In the beginning, I was trying to understand the sudden spike in Waves TVL, which at the highest point reached $4.75b, making Waves the 7th largest smart contract chain in terms of TVL. Then there was a lot of discussion and drama on twitter about how Waves is ponzi, how Alameda manipulated $WAVES prices and the proposal on Vires loan factors. I was initially only going to share with my colleagues, but I found this very interesting so I decided to write down my learnings.

I’m going to go through the following items, mainly focusing on the recent drama around Waves.

• The Waves ecosystem and applications

• how did Waves gain capital and users

• The fud — the ponzi accusation, Alameda, $USDN losing peg, the new proposal

• comparison between Luna/$UST and Waves/$USDN

The Waves ecosystem and applications

The TVL on the chain is mostly contributed by Neutrino and Vires Finance (around 90%), so I’m only going to briefly go through these two protocols.

Neutrino is the stablecoin protocol of this ecosystem. There are stablecoins ($USDN, $EURN), decentralized forex, synthetic assets, and their governance token ($NSBT) in their system. Stablecoins are their main product, and these $USDN are fully backed by the native $WAVES token. $WAVES are locked to mint new $USDN. The stability mechanism is that users can swap $WAVES<>$USDN whenever $USDN prices deviate from $1. When $USDN exceeds $1, arbitrageurs can swap $1 WAVES for USDN in the Neutrino dApp and sell on the market, this increases $USDN supply and would eventually bring $USDN back to its peg when no arbitrage opportunities exist. The same happens the other way around. Seems fair and similar to many stablecoin mechanisms. But on Neutrino, only $NSBT stakers ($gNSBT holders) can swap $WAVES<>$USDN. There’s even a max limit of how much you can swap depending on how many $NSBT tokens you stake, more on this later. Another thing to notice is that when the market cap of WAVES locked falls below the USDN market cap, the governance token $NSBT would be sold for WAVES to fill in the reserve gap. However, this is not the usual case. As $WAVES prices surge in the past weeks, $USDN is backed by at least 2.5x of $WAVES.

Vires Finance is the money market on Waves, which is very similar to Aave. Users can deposit any supported assets and use them as collateral against their borrowings. There’s also a bridge function, which users can bridge assets (mostly stablecoins) between Bitcoin, Ethereum, Polygon and BSC. At the time of writing (Apr, 7th), there are still over $1.4b locked on the platform.

How did Waves gain capital and users

After looking at the two major protocols that contribute to the TVL of Waves, it’s still not clear what is attracting so much capital. Some of the potential reasons are:

1. Russia Ukraine war brings attention to the “Russian Ethereum”

2. Waves upgrades to 2.0 , Allbridge and Gravity partnership announcement

3. high deposit rates on money market platform

4. $USDN staking rates

5. manipulation?

The first reason seems possible, as the invasion (Feb, 24th) matches the beginning of the surge of Waves token price and TVL (late February early March). People may be seeking potential alternatives to banks that have been restricted in some way. This may be the initial reason that drew attention to Waves, but in a month $WAVES ~7x and Waves TVL ~4x so I believe there are other drivers as well.

The second reason seems more fundamental. In early March, Waves announced an improved consensus mechanism, compatibility with EVM ,and other upgrades named together Waves 2.0. However, Waves has been around since 2016 so even though upgrades and partnerships are bullish, I think the reaction to these news wouldn’t be as crazy as it was.

The third has always been an effective way for new protocols to attract users and capital. Anchor has been attracting capital with its~19% fixed yield for $UST. The stablecoin yields on Vires are around 15%-19% if utilization stays under 80% (which are most cases), although these are much higher than most money markets (Aave: ~3%, Compound: ~2.5%), they aren’t attractive enough for users to deposit on Vires instead of Anchor.

The fourth would be for the $USDN staking rates, this serves as its main source of demand and primary use case. $USDN can be staked on Waves Exchange, KuCoin and a few other exchanges. The yield is calculated as $WAVES backing ratio * $WAVES staking yield(~3.9%), and is generally 8%-15% depending on the backing ratio. The rates are not bad for stablecoins staking but I’m not sure if it’s enough to attract so many users.

The last reason would be price manipulation. If we look at TVL on Waves in the past month in $WAVES, it hasn’t grown much, meaning most of the TVL growth comes from $WAVES price appreciation rather than organic user or activity growth. We will go through more about the possibilities of price manipulation in the next part.

The fud — the ponzi accusation, Alameda, $USDN losing peg, the new proposal

It all started with this thread on twitter about Waves being a massive ponzi scheme.

Is it ponzi?

0xHamZ accused Waves of folding leverage and manipulating $WAVES prices by: deposit USDN on Vires → borrow USDC on Vires → transfer USDC to Binance → buy WAVES with USDC → convert WAVES to USDN → start over. The main effects of this manipulation are

1. creating demand for $WAVES out of nowhere → drive prices higher

2. creating $USDC/$USDT borrowing demand → increase lending rates → attract more users to deposit

3. expand $USDN market cap

There are even transaction records provided as proof to support his saying. He also pointed out that the $USDN market cap has regular expansions every few days, this could be everytime $USDN is minted in Neutrino (the 5th step in the graph above).

After 2 days, the founder of Waves Sasha Ivanov fought back. They accused Alameda of manipulating $WAVES prices as well.

Waves fought back.

He provided evidence that Alameda was selling borrowed $WAVES on Vires using USDC/USDT as collateral, and longing $WAVES perpetuals for profits. Funding rates of $WAVES have long been negative (short pays long), so shorting spot + longing perps could earn funding rates while having zero exposure to $WAVES.

The two sides both offered proof to their side of the story, I guess maybe both of them are right? On Apr, 4th $USDN lost its peg, reaching $0.68 at its lowest point. I think the main reasons are

1. fud caused by the ponzi scheme accusation on twitter

2. the stability mechanism is too weak

The first reason is about the dispute mentioned above that begin with the accusation from 0xHamZ. As his post caused fud → users withdraw staked $USDN on Neutrino → sell $USDN on curve → no arbitrageurs fixing the prices on AMMs + no traders swapping $USDN to $WAVES → lose peg. The lack of faith in Waves would result in no arbitrageurs buying $USDN to fix the price, then as the selling pressure continues, $USDN loses its peg. As seen in the photo below, the 50/50 $USDN+3Crv pool becomes imbalanced as people dump $USDN in the market.

As users lose faith in the Waves ecosystem, they also withdrew $USDC/$USDT deposited on Vires. This immediately caused utilization in these pools to increase, causing lending rates to spike. Interest rates don’t increase proportionally, as utilization rates exceed 80% lending and borrowing rates spike to lower borrowing demand, then consequently lower utilization rates.

The second reason is that the stability mechanism of $USDN is too weak. As mentioned earlier, the $USDN peg relies on arbitrageurs to swap between $WAVES<>$USDN but only large $NSBT stakers ($gNSBT holders) can take this opportunity, the more $gNSBT you hold the more you can swap. The team has proposed new parameters to increase the swap limit per $gNSBT, which increases the limit from 2,240 to 63,096 per 10,000 $gNSBT. Although this strengthens the stability mechanism, having a limited number of arbitrageurs may still not be enough in extreme circumstances.

After $USDN lost peg, there was a new DAO proposal on Vires to 1) set maximum borrow rate to 40% 2) set $WAVES/$USDN/$EURN liquidation threshold to 0.1%. The proposal was voted among the community by $NSBT holders from Apr, 5th to Apr, 10th, at the time of writing around 55% voted against it.

There weren’t a lot of details, but I believe setting the maximum borrow rate for all assets would limit the utilization of all pools to around 86%. If the team had been borrowing large amounts of USDC/USDT from the pool, this would lower their risk of paying 80%-100% rates in fud circumstances like this.

The current liquidation threshold for $WAVES/$USDN/$EURN is 70%/95%80% respectively, this means the maximum debt borrowed for $100 collateral would be $70, $95 , and $80. Setting the liquidation threshold to 0.1% basically means liquidating almost the entire $WAVES/$USDN/$EURN debt position. On April 7th, there was about $20M, $30M, and $947k in debt for these assets, and I believe these numbers were slightly higher when the proposal was submitted. This would punish people selling $WAVES since $20M of $WAVES (about 1% of the current market cap) would have to be bought in the market to repay the outstanding debt, same for $USDN and $EURN.

Comparison between Luna and Waves

At first glance, $USDN and $UST may seem very similar, which is what I thought when I first learned about Waves and $USDN. But in fact, they are different in many ways. The following table is some of the major comparisons between the two networks, the top part is about stablecoins and the bottom part is about money market.

The major difference between them is the stablecoin stability mechanism. Every $USDN is at least backed by $1 WAVES locked in Neutrino, but no $LUNA is locked as a backing asset for $UST. $UST is an algorithmic stablecoin, meaning it doesn’t require collateral or reserve locked to be minted. Although LFG has been stacking BTC as reserves for $UST, which makes it similar to crypto-backed as users can swap $1 UST<> $0.98 BTC. This serves as a cushion and reduces the reflexivity of the system, but $UST can still only be minted/burned by burning/minting $LUNA (The BTC pool is only a reserve). As mentioned previously, the limited number of arbitrageurs makes the $USDN much weaker compared to $UST where anyone can exploit price deviations.

The other difference lies in the strategies of the money markets. Vires and Anchor both use high lending rates to attract capital, but on Vires borrowing rates > lending rates while on Anchor borrowing rates < lending rates. The Vires case is like normal profitable lending/borrowing protocols, but in order to attract capital with high lending rates, the borrowing rates have to be even higher. No one would borrow stablecoins at ~20% when they could borrow on other protocols for ~3%. The Anchor case is to lower borrowing rates to attract borrowers and increase lending rates to attract lenders. This makes sense to attract users but would drain the reserve very fast. Although Vires and Anchor have different approaches, their common issue is how to attract more borrowers.

Luna has been trying to expand use cases and access to liquidity of $UST to increase demand for $UST borrowing, to prevent 1.) expanding the difference between borrowing and lending on Anchor and 2.) draining the ~19% yield provided to attract lenders.

Starting with ponzi is a smart and effective way of attracting capital and users in the beginning, but ponzi isn’t sustainable. In the end, it still comes down to whether protocols can create real use cases and sufficient demand to support the growth of a project.

There are still things that I don’t quite understand:

1. Are there any more reasons to hold $USDN other than staking it? If not, is the staking yield high enough for users in DeFi to do so?

2. If the team is folding leverage to pump $WAVES prices, why did they do it now when they could much earlier?

I didn’t know Waves at all until last week so there must be details or more stories that I’m missing, so if there are any suggestions or comments please let me know!

In this article, I try to use existing valuation models on the FLOW token. FLOW is the native token for the Flow blockchain, which is built for NFTs, interactive and collectible crypto experiences. Flow is also the team behind one of the first NFT projects — CryptoKitties, and they’ve built the Flow blockchain to enable more consumer applications on blockchain. The most well-known project on Flow is NBA topshot, where users can purchase and collect packs with moments of NBA games.

According to the FLOW token economics, the token could serve as 1. payment for computation and validation services (transaction fees), 2. medium of exchange, 3. deposit for data storage, 4. collateral for secondary tokens and 5. participation in governance. In this model, I value FLOW as a currency for the first three use cases mentioned above. However, there is a gap between the intrinsic value and the price of a token so I tried to incorporate** investor value** in the final token price.

The framework would be using the equation of exchange to estimate the currency value of FLOW, then the nth order investor model to anticipate investor value. For velocity, I used the Baumol-Tobin model for money demand to estimate the holding period of FLOW as a medium of exchange. In the last section, more details and rationale behind the assumptions made are elaborated. Most of the existing theories and models have been used to value hypothetical tokens, here I included the characteristics of FLOW tokens in these valuation methods.

There are a lot of assumptions in this model, and they would vary depending on your view of the crypto economy, how successful the Flow blockchain would be, and how the use cases of FLOW token would evolve over time. Although the final product of the model wouldn’t be able to predict the price of the token, it gives us information about where the majority of value comes from. This process also demonstrates some difficulties one would encounter when applying valuation models to existing tokens. Feel free to download this model and make different assumptions to see how they affect the price.

Equation of Exchange

In monetary economics, the equation of exchange defines nominal expenditure of an economy (GDP) over a certain period of time as money supply times velocity, which is the frequency of a unit of money being spent. Both Chris Burniske and Vitalik Buterin applied this equation to cryptoasset valuations, here I’m also using this equation as the framework to value the medium of exchange function of FLOW. In traditional macroeconomics, M stands for money supply, V for velocity, P for price level and Q for real expenditures. The same concept applies to FLOW, where M is the token supply, V is the velocity of tokens, P is the price of utility in the Flow network and Q is the expenditures on Flow. The P*Q part of the equation could also be interpreted as the GDP of the Flow economy. One thing to mention is that every parameter in this equation is priced in USD, so that we could solve for the USD price of FLOW by simply dividing the token supply base (M) by the amount of tokens. More details of the inputs in this equation are provided in the following sections.

M*V=P*Q

M — token supply base, how staking rewards and locked-up investments affect the token supply

V — derived from the Baumol-Tobin model, how transaction costs and forgone return on store of value would affect the holding period of Flow

P — Flow use cases, how much utility costs, how much demand for the network would be converted to demand for the token

Q — market/industry fundamentals, how much demand there will be for the Flow network

Q — Market/Industry Fundamentals

First, we are going to estimate how much demand there will be for the Flow network by looking at the markets and Flow’s position in them. Three measures of the Flow economy are needed, number of transactions for transaction fees, users for new storage fees and transaction volume for the amount settled in FLOW. For markets, I separated the NFT and DeFi market since they are not comparable in terms of transaction volume. Then market growth and market share of Flow would be estimated for each market, and the product of market growth and market share growth would be the growth of the Flow economy. I assumed the three measures grow at the same rate the Flow economy grows in NFT and DeFi markets. This wouldn’t be an accurate estimation, since a user could transact more times and volume as more projects are built on Flow. However, since storage fees and transaction fees account for a small fraction of the total GDP of the economy (results are seen later), this imprecision should be acceptable.

For the NFT market, I estimated a high growth of 120% this year and it steadily declines to 10% in 10 years as the industry matures. For market share captured by Flow, I used an S-curve with a 35% market share at peak. For the DeFi market, 500% growth this year and declines to 10% as well, and a 10% market share at peak. A much lower estimated market share for DeFi since Flow is built for NFTs and I believe the success of NBA topshot would attract more NFT-related projects rather than DeFi to be built on Flow. More rationale is provided in the last section.

I used data of projects from dappradar for the first half of year 1 (Nov 16, 2020- May 15, 2021) as a benchmark for future estimations of the three measures, and categorize NBA topshot, VIV3 and MotoGP Ignition as NFT and Bloctoswap as DeFi. Number of transactions and transaction volume are available data, but users require an estimation. I estimated 800,000 users in NFT and 10,000 users in DeFi on Flow for the first half of year 1. To summarize, the following graphs depict my assumptions of number of transactions, transaction volume and users for Flow in the NFT and DeFi market.

P — Use cases of the FLOW token

The main use cases of FLOW tokens are for staking, medium of exchange, utility for the Flow network and voting for governance. I included both medium of exchange and payment for utility as the total GDP (or P*Q) of the Flow economy, since they offer FLOW functions similar to a currency and can be solved by the equation.

Medium of exchange accounts for transaction volume of projects settled in FLOW. For example, VIV3 is a NFT marketplace where all the artworks are priced and transacted in FLOW, so the token would be used as a medium of exchange on this marketplace, and users would have to own FLOW in order to purchase artwork on VIV3. On the other hand, NBA topshot accepts payments with credit card and other cryptocurrencies then settles in USDC, so FLOW wouldn’t be essential to users. Note that if the project exchanges currencies for users and settles transactions in FLOW, the token still accounts for a medium of exchange function since it increases the demand for the token likewise.

For demand from medium of exchange, I used an S-curve to estimate the percentage of transaction volume settled in FLOW. In the early days, many projects may accept multiple cryptocurrencies or even fiat for payment, but as the network and industry mature more of these transactions will be settled in the native currency. However, this percentage will largely depend on whether these NFT projects would accept multiple payment methods like NBA topshot or accept only FLOW like VIV3.

The two payments for utility on Flow are transaction fees and storage fees. Note that these utility fees could be paid by projects on behalf of their users, but who the actual payer is should be irrelevant to the demand or the value of the tokens. Transaction fees are paid to validators/stakers as rewards for contributing to the network. According to the team, transaction fees would start low at 0.001 FLOW in the early days, and emphasizes the importance of transaction and storage fees being** responsive to market demand**. Therefore, I estimated both transaction and storage fees in USD with their growth aligned with the growth of the Flow economy (the growth we discussed in the previous section). I used a $10 FLOW token as the reference price for estimating the current utility prices in USD. Using USD as the currency for these fees enables us to estimate the P*Q part of the equation without using the prices of FLOW tokens every year as an input in the model.

Storage fees would include 0.001 FLOW as a new user deposit for 10kb storage, additional fees are required for extra storage on the chain. Additional storage fees are not elaborated in the token economics paper, so I assume a 0.001 FLOW fee for a certain amount of storage. Demand for additional storage is also estimated from an S-curve of percentage of total users, assuming users require more storage as each owns more assets.

The assumptions on transaction fees and storage fees could seem unfounded, but estimating utility prices based on demand is the most reasonable method we have. In addition, from the graph below we could infer how minor the estimations for utility fees are. This graph shows the breakdown of total GDP (P * Q), suggesting transaction volume settled in FLOW accounts for the majority payment amount. We could therefore conclude that the demand for FLOW tokens is highly dependent on whether projects settle in FLOW for payments. Users are required to pay for utilities with FLOW tokens, but because Flow is built for consumer applications, the utility costs are low. Therefore, the value of FLOW tokens comes more from its function as a medium of exchange for different projects, and less from paying for the utility on Flow blockchain.

Another thing to consider is that these storage fees would be held out of circulation, and this would serve as a constraint on token supply. To avoid using FLOW token price as an input in the model, I wouldn’t be able to estimate the total storage fees held out of circulation (in FLOW) and deduct them from the amount of circulating tokens. This would result in an underestimation of token value, but accumulated storage fees held out of circulation accounts for a little fraction of circulating tokens so the underestimation should be negligible.

M — Token supply and distribution

The Flow team disclosed an indication of the circulating supply schedule as tokens unlock and new tokens are distributed to stakers/validators. All the investments are locked-up until the end of year 1 (Nov 15, 2021), so until then all the tokens in circulation are staking rewards. Inflation in the Flow economy is a combination of staking rewards and new issuance of tokens, but they only issue new tokens as necessary to make up the difference between transaction fees and guaranteed payment, excess amounts are held in escrow to offset future inflation.

From the schedule, we know that the team plans to cap inflation at about 3% starting from the third year. Therefore I extended this schedule to fit the 10-year timeframe of this model by setting a 3% inflation to the total supply fully distributed to stakers, as shown in the table below.

Something important to define here is the difference between circulating tokens, tokens outstanding and total supply. This concept is similar to how floating shares, outstanding shares and authorized shares are different. Circulating tokens are staking tokens available in the market, including staking rewards, unlocked investments and staked tokens that are not locked-up. These are the tokens that users can conduct transactions with, meaning the medium of exchange/currency function of FLOW is only available to the ones in circulation. Tokens outstanding are defined as total supply less dapper labs ownership, collateral reserve and foundation reserve (the green portion of the table) that would not enter the supply directly. These are the tokens that all the values of Flow network are distributed to. Even though the ones locked-up do not contribute medium of exchange value, they could be sold in the market at the same price when the lockup period is over so they share the entire value of FLOW tokens. Finally, total supply represents all tokens in existence.

V — Baumol-Tobin Model

Velocity is probably the most important but controversial input in the MV=PQ equation, in the VOLT model by Alex Evans he introduced using the Baumol-Tobin model to estimate the velocity of tokens. This model shows the tradeoff between having liquidity by holding a medium of exchange and earning return by holding a** store of value**. Here the return does not mean an economic return, but more of a nominal return that preserves one’s purchasing power. People should always keep their fortune in store of value assets, and only keep a minimal amount of medium of exchange as “working capital” for expenditures. In the crypto economy, I assume people will view Bitcoin and stablecoins as store of value as they view gold and cash in the traditional financial industry, stablecoins would be the major medium of exchange, FLOW and other utility tokens would be the medium of exchange for their native economy/network (similar to how different countries have their own currency). This model minimizes the money management costs between store of value and medium of exchange, including the transaction costs and forgone return(opportunity cost). Below shows inputs for this model, and in the following context are details about the selection of these inputs and the assumptions made.

Inputs for this model

• Opportunity cost: the forgone return of holding FLOW instead of a store of value

• Transaction cost: all costs involved in exchanging store of value to medium of exchange

• Total spending: annual spending or demand for medium of exchange to cover expenditures

The below graph demonstrates how users in this model would spend their tokens. Users exchange a fixed amount of their store of value assets to FLOW each time and spend them at an even rate. This fixed amount is the optimal cash balance a user should exchange each time to minimize money management costs. The optimal cash balance times transfers per year (4 in this graph) would be the annual spending of a user.

In the VOLT model, expected return on store of value was used as the forgone return/opportunity cost of holding a medium of exchange. However, the fact that you could stake these FLOW tokens and earn rewards should reduce the opportunity cost of holding a medium of exchange. So in this model I used expected return on store of value less expected return on FLOW as the opportunity cost. Expected return on store of value is a difficult estimation since it represents the average expectation of people in the cryptoeconomy. I assumed a 10% expected return from store of value assets, with a combination of Bitcoin and stablecoins. For FLOW, I used the inflation of the token supply as the expected return. This would not be accurate because if token holders consist of investors and speculators, the price would diverge from its currency value, increasing the return of holding FLOW. Other than that, the growth of the currency value may be much higher than the inflation rate.

This transaction cost should include all costs associated with exchanging a store of value asset to FLOW (medium of exchange), including network fees, exchange fees, spreads and other costs resulting from inconvenience. This represents the** friction** in the economy, the lower the friction is the less FLOW people hold, because they can always convert their store of value assets to FLOW when they need it. The downstream effect on velocity is that the less FLOW people hold, the higher the velocity would be, and the lower the token price would be. For simplicity, I used Bitcoin transaction fees as the transaction cost for this model, since it’s difficult to measure all the associated costs. Transaction fees on-chain are highly correlated to network demand, in high-demand times BTC transaction fees could be as high as $40 but in low-demand times it could be under $10. For this model I applied a $10 cost for each transaction, considering increasing demand could increase transaction fees, it increases to about $15 in 10 years. There could be a debate on this transaction cost estimation, and I’d like to hear any suggestions.

The total spending in this model represents annual expenditure of FLOW per user. For this input, I simply used the total payment amount of the network divided by total users.

The product of this model would be the average velocity of all tokens as medium of exchange, however there are other tokens outstanding that are staked or locked up as investment. Tokens staked or locked up have a velocity of 0, decreasing the velocity of the entire tokens outstanding pool. Introducing the mechanism of staking serves as an incentive for users to hold tokens, which decreases the velocity of tokens. Data from flowscan and our token supply schedule can give us information on what percentage of total token supply are staked (ranging from 43%-98%), but not the percentage of circulating tokens staked. I tried to assume that all the reserves and locked-up investments are staked, but according to the calculation that assumption is invalid. So I estimated 90% of tokens in circulation staked in 2022 and the percentage declines as more demand increases and more use cases become available.

Nth order investor

There is always a gap between valuing and pricing. There are many determinants of price that are not taken into consideration when we do valuations, so this nth order investor model tries to simulate how “groupthink” affects prices.

In How to Value a Crypto Asset by Brett Winton, he introduced the nth order investor method to estimate the investor value of a token. Different from users that hold tokens to transact or pay for storage, investors would hold tokens to anticipate future appreciation. This method assumes investors have a fixed hurdle rate and invest horizon, therefore the 1st investor would expect the current price to be the present value of the utility value at the end of his/her investment horizon. Other investors could do the same by discounting the price other investors anticipate, but the price of the token should be the maximum amount any investor would pay for.

In Improving the equation of exchange, they improved the model by setting various investment horizons, and I used their version of nth order investor model for FLOW. In the table below, the 0th investor value is the medium of exchange/currency value per token we derived from the equation of exchange. Each nth order investor would discount the (n-1)th order investor price one year from now to know his/her investor price. So the 3rd order investor in 2022 could be anticipating the currency value in 3 years, the 1st order investor price in 2 years, or the 2nd order investor price in a year. The final fair value of the token would be the maximum amount among all the investor value orders.

One could argue that if we extend the forecast horizon to over 10 years, these investor values would change. But it also makes sense that as the token and its economy matures, most token holders would be actual users instead of investors or speculators, therefore currency value and fair value of token would converge as shown in the bottom graph. Both values decrease in 2030 in this graph because by then inflation exceeds the growth of total token value, just as prices decrease when supply grows faster than demand. I wouldn’t say this would certainly happen, but under the assumptions made in this model where inflation stays at 3% while the growth of Flow economy continues to decrease, the value of token would eventually decrease.

For discounting, I applied a 40% discount rate for token investors. There is room for discussion on this selection just as for a VC discount rate for early-stage companies. Discount rates decrease throughout the years, assuming as blockchain industry matures and crypto becomes mainstream we would see diminishing returns in investing in crypto and therefore investors would require lower returns and less risk premium.

In conclusion, the currency value derived from MV=PQ equation would be the value floor of the token given its means of exchange function, and the nth order investor model helps us estimate the investor value by speculating on the currency value an investor would anticipate.

Discussion

• The value of a governance token

An important use case of the FLOW token is participation in governance. As the network expands, the value of voting rights or determining crucial matters of the Flow blockchain would definitely increase. However, this model does not include any value from its use case in governance, because there aren’t any valuation methods in merely valuing voting rights. There’s definitely room to discuss and I’d love to hear suggestions.

• Staking

I struggled whether staking rewards are similar to dividends and could be valued using a DCF model like in traditional finance. I did not include staking rewards because they are a form of inflation like how stock dividends don’t create additional value to stockholders. One could argue that not all token holders are stakers so the token value to stakers should be higher, and value to “pure users” (who don’t stake) should be lower. However, I believe in equilibrium all token holders would be both users and stakers since not staking would decrease the real value of your tokens. Applying this notion to the Baumol-Tobin model would mean that all users would stake their tokens when they hold them as working capital for future needs. In equilibrium, the percentage of tokens staked would not be a distribution between people who stake and don’t stake, but between times when tokens are staked and times when tokens are traded during transactions. In conclusion, the staking mechanism contributes value to the token by decreasing velocity and accepting FLOW as rewards for validating transactions.

• S-curve details

I applied S-curves to the percentage of market share of Flow, transaction volume settled in FLOW, demand for additional storage and tokens staked. These numbers depend on both internal and external factors such as the development of the Flow network, the adoption of FLOW tokens and the overall market and cryptoeconomy. For all the S-curves in this model, I assumed 2023 as the start of high-growth phase and the takeover time to be 7 years. As the Flow network grows, its market share, transaction volume settled in FLOW and demand for additional storage increases while percentage of tokens staked decreases.

Inputs for the S-curve

• Peak value: the highest value achieved

• Start of high-growth phase: when it reaches 10% of peak value and begins high growth phase

• Takeover time: amount of time it takes from 10% to 90% of the peak value

• NFT and DeFi market

More details and rationale behind estimates of the NFT and DeFi market and the position of Flow in the industries are discussed here. In the past two years, the growth of the NFT market is 246% and 139% respectively. NFTs are still in the early stage with very high growth, but like all industries the growth slows down and remains stable in mature stages. I applied a 120% growth rate for 2021 and assumed it declines to a modest 10% at the end of the forecast period, this would imply 54% CAGR over the next ten years.

If you look at the top NFT projects other than NBA topshot, most of them are built on Ethereum. On the positive side, Flow is built for consumer applications with low transaction fees and its success in monetizing existing IPs with NBA topshot could bring huge partnerships in the future. (Flow has already announced partnerships with UFC and Dr.Seuss to release collectibles.) On the negative side, solutions for lowering gas fees on Ethereum and the development of other blockchains could limit the growth of Flow. Considering both sides, I applied a 35% market share at peak for Flow.

If we measure the DeFi market size from total value locked, it has 22x in 2020 and has more than doubled in the first half of 2021. I applied a 500% growth rate for 2021 and assumed it declines to a modest 10% at the end of the forecast period as well. For the market share captured by Flow, I also applied an S-curve but with a much lower share (10%) at peak. First because currently the DeFi market is dominated by Ethereum-based projects. Second, Flow is built for NFTs and has proven success in NBA topshot so would probably attract much more NFT than DeFi projects to build on this blockchain.

• Including token price or return in the model

An ideal valuation model should not include inputs like FLOW token price or expected return on FLOW, which are products that this model is trying to estimate. I tried to avoid circular references like this but still couldn’t exclude all of them. In the P section, I included the current token price as a reference of current utility costs in USD. And in estimating velocity, I also included expected return on FLOW to know the net forgone return of holding a medium of exchange instead of store of value assets. I’d love to hear if there are better methods on avoiding circular references in token valuation.

• How this framework can be applied to other utility tokens

This valuation framework allows us to understand where the value of the token comes from by breaking down the functions of the token and estimate the demand for each of the use cases, which could be applied to other utility tokens as well. As model statistician George Box said, “All models are wrong, but some are useful”. Although there are a lot of assumptions and extrapolations involved in this model, it certainly offers a more methodical way to value a token.