An Interest Rates Primer for Cryptocurrency Folks
I’ve banged on a little bit on Twitter about how digital asset markets are continually recapitulating discoveries from fiat markets (securitization, credit-default swaps, corporate actions oh wait no disregard that). But aside from an effort by Genesis Trading, there’s been a surprising lack of interest or development in crypto interest rate markets.
From what I’ve heard (and @ me if I’m wrong), crypto lending markets are not big. You can borrow bitcoin and other major cryptocurrencies for shorting or market-making purposes, but the rates are stiff (8-10% on cash-collateralised loans) and volumes are small.
This has always baffled me! Because crypto markets settle so quickly—near-immediate, rather than T+2 settlement in fiat FX markets—you’d think that market-makers would need to borrow a lot more, and a lot more often, to fund their day-to-day operations.
(Why is it so? If you post an offer to sell euros or South Pacific pesos or widgets or whatever, and someone comes in and pays your offer, you now have to deliver widgets to them. But generally banks don’t keep huge piles of widgets (or euros, or South Pacific pesos) on their balance sheets; instead, they borrow them from someone else, and deliver the borrowed widgets to the buyer. In fiat FX markets, you have two business days between trade date and settlement date to net your trades and borrow the required widgets (Canada, Turkey, and a few other countries give you one business day); in cryptocurrency markets, settlement usually happens near-immediately and there’s not so many opportunities for netting. That means that as a market-maker, you should need to borrow (and lend) more actively in order to manage your crypto balances.)
In practice, most digital-asset market-making desks just hodl titanic amounts of crypto on their balance sheet, and settle their trades from their balance sheet hodlings. This works great when prices are going up… when they’re going down, not so much.
The lending market is opaque, as well. It’s entirely OTC except for a few small retail-oriented operations (and the very-short-term funding markets on venues like BitMEX), and rates are pretty much entirely by negotiation. In a market like this, and when the demand to borrow or lend crypto is limited at best, you wouldn’t think there’d be a lot of interest-rate market activity.
But there’s one place where bitcoin interest rates are visible—and, intriguingly, the markets have started to move in a big way since crypto prices started to fall in mid-November. That rate market is the one embedded into the CME & CBOE futures market.
Here Comes The Science
(Is “Here Comes The Science” still a meme? Please tell me it’s still a meme and I’m not getting old.)
This bit is all Futures & Forwards 101, so if you’re a markets practitioner you can skip this bit.
Imagine that bitcoin has a higher interest rate than US dollars (it does, as we’ll see later). And imagine that you want to buy bitcoin for US dollars. You can choose one of two ways:
- “Value today”: you go to your bitcoin exchange of choice, hand over your USD, and receive bitcoin immediately; or,
- “Forward value”: you agree to buy the bitcoin today, and lock in the rate; but you also agree that you won’t pay for (or receive) the bitcoin until a month from today.
The difference between the two: if you buy the bitcoin “value today”, and receive it today, you can earn more interest in the intervening month than you would if you’d held on to the USD. So, all other things being equal, you would want to take delivery of your bitcoin as soon as possible. Equivalently, you’d pay less for bitcoin to be delivered in a month’s time than you would for bitcoin to be delivered today, because you’re giving up that extra interest.
And that shows up in the bitcoin futures market: futures that expire in a month, or three months, or six months, tend to trade at a cheaper price than the price for immediate delivery on a bitcoin exchange—and the discount gets larger the further out you go.
(If bitcoin’s interest rate was lower than the USD interest rate, then bitcoin futures would trade at a premium, because you’d rather take delivery of the low-interest bitcoin as late as possible. And if the two interest rates are the same, you’d be indifferent between owning bitcoin and owning USD, so the futures and the spot price would be the same.)
Let’s be more rigorous about this. A futures contract price can (and must, if the arbitrageurs are doing their job) be tied back to the spot price of the underlying commodity. In the example of a BTCUSD future: the price to buy a BTCUSD future should reflect the all-in cost of these three parts:
- Buying BTCUSD today;
- The interest cost of borrowing the USD from today to the delivery date of the future;
- The interest earned from lending out the BTC over the same period.
This is called the “cash-and-carry” arbitrage; it exists because those two trades—buying the future, and the buy-spot-lend-BTC-borrow-USD trifecta—are economically identical. In both cases, you end up long some BTC and short some USD on the delivery date of the future (give or take the fact that bitcoin futures are currently all cash-settled, but that doesn’t change the economics).
If you turn the handle backward on the above, you can back out an implied BTC interest rate from the spot price, the futures price, and a USD interest rate.
Let’s try it. These numbers are extremely fuzzy, but they’re close enough for a general idea (and the huge tick size of bitcoin futures means that you can’t be accurate to within more than about one percentage point on the interest rate anyway).
Here is the CME bitcoin futures chain at about 5:10am Pacific time on January 29th 2019. Cash bitcoin (CME’s Bitcoin Reference Rate, available here) was trading around 3399 at the time:
Let’s use the front month—the February future—as our example. The formula:
F = S * (1 + r2 * d/360) / (1 + r1 * d/365)
- F (the futures price) is 3360 mid.
- S (the spot price) is 3399.
- d (the number of calendar days to expiry) is 24.
- r2 (the USD interest rate for a 24-day run) takes a little thought. Market convention for this particular usage is to look at the USD LIBOR fixings for today, and then interpolate as your heart (or your quant desk) desires: let’s just linearly interpolate between yesterday’s 1wk and 1mth LIBOR (2.4065% and 2.50175%) and call it 2.475%.
Fill in all the variables, crank the handle, and the implied interest rate for BTC is about 20.2% annualised. That’s… a lot!
Even for longer runs, the rates are still very elevated: the implied BTC rate for the June futures (3310 mid, 150 days to expiry, r2 about 2.79%) is nearly 9.5%.
(For the nerds, yes, BTC’s daycount is ACT/365. And for the real nerds, yes, I’ve disregarded futures margin discounting, don’t @ me.)
But what does it mean?
There’s no Central Bank of Bitcoin conducting open-market operations to set the overnight interest rate for bitcoin; the rates spring entirely from the interplay between borrowers and lenders. If there’s an excess of people wanting to lend, then rates will fall; if there’s an excess of people wanting to borrow, then rates will rise.
The market for personal loans in cryptocurrency isn’t exactly huge, so the main source of demand to borrow bitcoin comes from people who want to short it. As the demand to short bitcoin through futures grows, the futures price will trade at a wider discount to spot, and the implied bitcoin interest rate will move higher.
And this is what’s been happening since mid-November, when bitcoin prices collapsed below $6000 and never looked back. The demand to short bitcoin in the futures market increased; so the futures started trading at a bigger discount to spot; and that increased the implied bitcoin interest rate.
Before the November drop, bitcoin futures traded at roughly the same as the spot price, implying that BTC and USD interest rates were about the same: a bit over 2%. After mid-November, the implied BTC interest rate spiked to 8% in a matter of days, and it’s stayed at those elevated levels ever since.
A side note: one of the other inputs to bitcoin’s implied interest rate is the storage costs. If it were hugely expensive to store bitcoin securely, that storage cost would have to be reflected in the cash-and-carry arb; it would show up as a reduced BTC interest rate, or equivalently a smaller discount (or larger premium) between futures and spot.
The spread between actual bitcoin lending rates (10-ish%) and futures-implied bitcoin lending rates (8-ish% for longer terms) might be at least partly due to storage costs. If it is, we should expect to see that spread compress over time as it becomes easier and cheaper to securely store bitcoin. The spread is about the same as the 2% management fee charged by GBTC, the one US-listed bitcoin exchange-traded product; if the management fee mostly reflects storage costs, then that’d suggest that our hypothesis is on track.
How do I take advantage of this?
There’s a way to capture the high BTC interest rates implied in the futures market, and still have exposure to the price of bitcoin: sell cash bitcoin, and buy bitcoin futures to replace the bitcoin.
You’ll need to keep the USD from selling the bitcoin around, in order to make margin payments on the futures if the bitcoin price goes down; but you’ll also be receiving interest on that USD (assuming your futures broker isn’t ripping you off).
The bitcoin futures price will converge with the price of cash bitcoin when the futures mature—so, the discount between the price you sell the cash bitcoin at and the price you buy the bitcoin futures at is the excess “interest” on the bitcoin.