Circle's Implications of the Fed's Repo Facility and Raised Interest Rates
Date: January 20th, 2023
In this writing, I’ll be forecasting Circle’s net operating income starting Q3 2022 through 2023, discussing the financial impacts of the Circle Reserve Fund’s potential access to the Fed’s overnight reverse repo (ON RRP) facility, and how raised interest rates are impacting Circle's revenue.
Circle and BlackRock
In April 2022, Circle raised $400 million in a funding round from BlackRock, Fidelity, Marshall Wace, and Fin Capital. Soon after, in May 2022, BlackRock filed with the SEC for the creation of the “Circle Reserve Fund” which went live in November 2022. Circle Reserve Fund is a bespoke investment vehicle created and managed by BlackRock for the investment of Circle’s USDC deposits. As stated in the SEC filing, the investments are limited to short-duration U.S. Treasuries. The reserve fund is a big win for Circle:
It partners Circle with BlackRock, the world’s largest asset manager by AUM at $10 trillion… with a “T”
Publicly binds BlackRock, and therefore USDC reserves in the Circle Reserve Fund, to only be invested in short-duration US Treasuries
BlackRock could apply for the fund’s access to the Fed’s ON RRP facility
For purposes of this writing, I’ll mainly be focused on the impact of #3.
What is an ON RRP and why does the Fed have a facility for them?
How does an ON RRP work?
Repurchase agreements (repos) can be called repos or reverse repos depending on the perspective. For the borrower, it’s a reverse repo. For the lender, it’s a repo. In a nutshell, the graphic to the right explains how repos work with the Fed. The graphic is showcasing a bilateral repo but in Circle's case, they would use a tri-party repo. The main caveat is that there's a tri-party agent (essentially a middleman) involved in the transaction. The tri-party agent (Bank of NY Mellon) handles the back-office pricing, matching, collateral margins, etc so the transaction is simplified for the borrower and lender.
The Fed’s use case
Prior to the “Great Financial Crisis” of 08’-09’ timeframe, the Fed used its “limited reserve” framework where its main tool to control the Federal Funds Rate (FFR) was the buying and selling of Treasuries. Following the GFC, the Fed injected a bunch of capital into the economy, reserves rose, and now they operate in their “ample reserve” framework. In the Fed’s ample reserve framework, the 2 main tools the Fed uses to control the FFR are Interest on Reserve Balances (IORB) and ON RRPs. The IORB allows institutions to deposit reserves at the Fed and earn interest. In theory, this sets a floor lending rate for those institutions because any other borrower would pose a greater risk. The problem is that not all major institutions have the opportunity to deposit at the Fed and earn IORB. The excess demand pressure can cause interest rates to go lower than IORB. The solution has been to allow a broader set of institutions access to the Fed’s ON RRP facility which, for the past few years, offers a rate that’s 10bps below the IORB.
If you’re wondering what happened to open market operations with treasury securities or want more info on these tools, see the following link (The Fed's New Monetary Policy Tools).
Circle and the Fed’s ON RRP Facility
Currently, Circle has 3 options for cash liquidity: hold cash (no liquidation needed), sell securities prior to maturity, or wait until the treasury matures (weighted avg 30 days). With these options, the management team has already found their desired ratios for invested reserves vs cash reserves (~80% and 20% total). In a liquidity crisis (a reaction scenario), selling securities prior to maturity would be the only option to boost cash liquidity. Due to its overnight duration, the inclusion of the ON RRP facility would add a 4th option for cash liquidity. The repo agreement would have inherently less credit risk because the borrower on the other end of the transaction is the Federal Reserve. With that, it provides an extremely low-duration, low-risk way for Circle to earn a yield on its reserves. This should mean that Circle could hold a lower percentage of reserves as cash while maintaining a similar risk profile. As a result, generating additional interest income with minimal marginal OpEx.
Access to the Fed’s repo facility requires admittance and BlackRock intends to apply for access to the facility for the Circle Reserve Fund. According to Ledger Insights, they can’t apply until November 2023 (1 year after the start of the Circle Reserve Fund). If they’re approved it would:
Mark a big step in the government’s acceptance of crypto markets
Add credibility to USDC’s backing
Enable more interest revenue for Circle through means of a lower cash percentage of reserves
Circle's financial outlook in a higher interest rate environment
Circle’s main revenue drivers are interest rates and the amount of USDC they’ve issued. USDC in circulation is ~43 billion and around 75% is attributable to Circle (they have a pro-rata revenue agreement with Coinbase, they both issue USDC). Since their most recent S-4 financials ending June 2022, USDC in circulation dropped from its peak at 55 billion to a stabilized 43 billion. The decline in circulating supply was caused by the broader crypto sell-off and Binance’s auto-conversion for USDC pairs. However, the biggest financial change for Circle has been the increase in interest rates. The figure below is showing the FFR (nearly equivalent to 1-month t-bill yield) since January 2020. The second half of 2022, is the first time Circle’s been able to achieve an attractive interest rate on invested reserves since they amassed scale. From June 2022 to the start of 2023, the FFR increased from 0.80% to 4.33%. In that same time period, the invested USDC reserves amount to ~43 billion. These are reserves that have minimal marginal cost as they increase and that don’t currently pay out interest to regular depositors (USDC holders). So, as interest rates increase or invested reserves increase it pads revenue with little marginal OpEx expense attributable.
Financial projections
Highlights:
2nd half of 2022: first time sustainably turning a profit
2023: $1.6B interest revenue and $770M of net operating income. This is the first full year of profitability.
NOI margin hovering around 45%
At the current USDC supply, the breakeven interest rate on invested reserves is ~1.75%
At the current USDC supply and interest rates, every additional 2% of reserves moved from cash to investments adds a marginal $20-30M to net interest revenue
Below are my financial projections for Circle from 2nd half of 2022 through 2023.
Notes/assumptions on the financials
In general, I'm trying to showcase that:
Raised interest rates will have a massive positive impact on Circle's net revenue
The ON RRP could lead to an additional tens of millions in incremental profit through a lower cash reserve %
OpEx should look cleaner moving forward
On that note, the forecasts of TTS revenue and OpEx could be improved but I think are generally in line. Additionally, the impact of the margin of error for these items is meager at current interest rates.
Lastly, with the wave of profit coming down the pipeline and the changing industry, all of Circle's OpEx is very discretionary. OpEx could materially change based on an expansion in headcount, executive compensation/bonuses, acquisitions, etc.
Historic data
The bulk of the historic Circle financial data is from their past S-4 financials, monthly attestations, and their NAV from the “Circle Reserve Fund”.
In accordance with the proposed Concord and Circle merger, Circle released S-4 financials for recent years, ending June 30, 2022. Circle stopped releasing public financials due to the cancellation of the ongoing merger discussion with Concord. It was only releasing these public financials because it was legally obligated to due to Concord being a public SPAC. Circle is continuing to release monthly reserve attestations. Additionally, the BlackRock-managed Circle Reserve Fund now contains the majority of Circle’s invested reserves and has a NAV that is updated daily.
Excluded line items
For historic and future data I excluded: SeedInvest revenue, depreciation & amortization, digital asset impairments, and any discontinued operations. This will essentially net out to a clean EBITDA. Why each is excluded:
SeedInvest: entered into a contract to sell, isn’t guaranteed yet, either party can back out if obligations aren’t met by end of March 2023. Either way, it’s a negligible part of the business and not Circle’s core focus.
D&A: non-cash
Digital asset impairment: non-cash and shouldn’t play as big of a role moving forward now that the adults understand that the blue-chip tokens and alt-coins aren’t viable reserve assets
Discontinued operations: non-core business
Reserve Interest Income:
Formula: reserve interest income = USDC in circulation * % reserves invested * yield on investment
USDC in circulation: used the data from the monthly attestations to determine supply for the months November 2022 and prior. To determine the current USDC in circulation, I looked at the current invested amount in the Circle Reserve Fund and then backed it into the total circulation based on the prior months' average cash reserve %. From there I kept the circulating supply constant through 2023 but supplied a sensitized table to see the impacts if supply changes.
% of reserves invested: the average for the monthly reserve attestations that 1) provided the data and 2 ) where the investments were in cash equivalents (July 2022 - November 2022), which came out to 21%. This is in line with where Circle has publicly said they want to be (20% cash).
Yield on investment: from the more recent attestations, Circle’s invested reserves are only in Treasury securities and their weighted average maturity is ~30 days. When comparing the annualized yield of the historical reserve interest income (based on USDC in circulation) to that of the daily average 1-month treasury, Circle’s yield on invested reserves tended to be 12bps higher than the 1-month treasury yield. This margin was fairly steady for historical data, ranging from 4bps to 16bps. So, I kept projections 12bps higher than the 1-month treasury yield (to be more accurate you could model their maturity structure). To determine the appropriate treasury yield for future months, I used data from the treasury.gov website. The data contained nearly daily rates for maturities of 1-month, 2-month, 3-month, 6-month, and 1-year. From there I calculated the implied forward rates for the 1month1month, 2m1m, 3m3m, and 6m6m. Which came out to 4.60%, 4.83%, 4.87%, and 4.61% respectively. I used the 1-month rates where possible and then assumed a constant 1-month yield equal to the 3-month rate and 6-month rate for their time periods of the projection.
BlackRock fee for Circle Reserve Fund:
As stated in the fund’s SEC filing, there is a total 0.17% expense fee based on NAV. The forecasted figures assume that all of the invested reserves are in the Circle Reserve Fund.
Transaction and Treasury Services:
From the first 2 quarters of 2022, calculated the average ratio of USDC in circulation to TTS revenue and then applied that constant ratio of 4,694:1 for all the projected periods. This ratio was lower in quarters prior (ranging from 775:1 to 3,839:1), but it’s a minor component of revenue in a high-interest rate environment so I’m not too concerned.
USDC income sharing and transaction costs:
The large majority of this expense should be the pro-rata revenue-sharing agreement with Coinbase. Going through historical S-4 filings, you can nearly account for all of the USDC outstanding from the issuance/redemption breakdowns provided. Over time the % of USDC circulation attributable to Circle has increased, for the first 2 quarters of 2022 it was 73% and 78%. We would expect that the following formula is approximately true: (USDC costs / USDC revenue) = (Coinbase USDC / Total USDC). When you compare this for the quarters from Q1 2021 through Q2 2022, the cost ratio averages out to be 9% below the percentage of Coinbase USDC. There are a couple of quarters in 2021 skewing the data with variances of -19%, and -25%. I ended up choosing 25% as the income sharing/transaction costs % of interest revenue. This adds a +3% margin to Coinbase’s Q2 2022 supply of 22%.
Compensation Expenses
Formula for compensation “c” at period “n”: c_n=(c_(n-1) * wage growth rate) + (change in headcount * cost per new employee)
Wage growth rate: set at 7%, right around national average
Cost per new employee: was determined by using the above formula for historical compensation data and LinkedIn headcount for the change in headcount. From historicals, the cost per new employee was a little wobbly (ranging from 68k to 110k) but I think on average it will be a decent proxy. The average cost per employee of 82k was applied for the forecast. A 5% quarterly growth rate was applied to headcount, slowing from the prior quarter of ~25% but I think this is appropriate given their current size and the economic conditions.
Other OpEx: G&A, IT, Marketing & Advertising
A 15% annual growth rate was applied. This is slower than the 2021 rates because they swelled in 2021 and then stabilized.