Thread by @tarunchitra

Thesis: Crypto needs LIBOR if it is going to be a Eurodollar substitute

Antithesis: People trust centralized lenders/exchanges with fiat backed stablecoins

Synthesis: @Libra_'s new ≋LBR design *is* LIBOR

But can they do better than copying IMF SDR weights?

Yes! 👇🏾

First: What is 'Eurodollar'?

Eurodollars represent dollar deposits held in foreign banks.

There are a number of reasons people outside of the US want dollars (e.g. to buy oil).

Foreign banks charge extra interest for procuring & holding USD for you — an interest spread!

Americans referring to 'Eurodollars' usually mean rate spread (e.g. excess interest a foreigner pays for USD vs. a US national).

And eurodollar futures are the most traded product on CME

(Note: @ChiangRei and I used to trade the Eurodollar curve)

More:
unexpected-values.com/crypto-dollars/

Why does the 'stablecoins replace eurodollars' narrative need something like LIBOR?

@Libra_ users end up holding basis risk — they buy ≋LBR with local currency and subject to differing lock-up windows and liquidation conditions that depend on the currency they buy ≋LBR with

Primary dealers — banks, custodians, payment processors — will need, at times, to borrow basket currencies against their ≋LBR positions to service customer liquidity demands.

LIBOR = IR that banks charge each other for borrowing from each other

∴ ≋LBR ~ LIBOR + Collateral

Why?

The SDR-esque construction in Libra defines [I chose random weights]

≋LBR = 0.5 * USD + 0.3 * GBP + 0.2 EUR

Redeeming USD for ≋LBR is 'safest' and will have the lowest rate, but the GBP rate should be 1.6x the USD rate.

Control of Weights = Control of LIBOR!

The association currently states that they'll welcome 'transparent oversight' from the IMF, Swiss regulators, et. al.

This should give you pause (but not for the reason you might expect!)

Eurodollars and LIBOR are priced via market mechs. — why does ≋LBR need fixed weights?

Answer: It doesn't! In fact, the fixed weights, while they might temporarily placate governments, will likely cause headaches down the road.

Can we make this more fair and market-driven, avoiding the politics of weight selection?

Yes — by, surprisingly, turning to DeFi!

One of the main benefits of having a BFT database as the core of @Libra_ is that implementing SDR weights becomes a short smart contract.

OTOH: Building a CME quality matching engine is impossible to do on a blockchain?

Nope — const. function market makers are *good* for rates!

Evidence: Early success of @CurveFinance — borrow and lend baskets of fiat/alg. stablecoins.

Each basket of stablecoins mints a synthetic asset — basically ≋LBR — and the IR charged for borrowing this pool is like LIBOR

But there are no human weight choices here!

With CFMMs like Curve, supply (lenders who want interest on stablecoins) and demand (people who borrow stablecoins for crypto) are algorithmically matched+priced implying rates and weights.

Can we just use the @CurveFinance curve, implemented in a @LibraDev Move contract? No!

≋LBR is tricky!

Unlike Curve, which only has USD coins that have the same mean ($1), ≋LBR has other risks:

1. Basis risk — Can I redeem for fiat?
2. FX risk — EUR went from $1.50 to $1, take a loss
3. Quanto risk — Should I charge more interest paid in ≋LBR vs. USD?

You could design an automated market maker that accounts for these risks in the following ways:

1. AMM curve change with FX prices
2. Fees change with prices
3. Map 'impermanence loss' to Basis + FX + Quanto and hedge it out

Does this look like existing AMMs? No!

Idea 1: Use @BalancerLabs, which has a programmable AMM curve.

Unfortunately, Libra's risks don't vary geometrically w/ lending supply — can be 'sharp' default to 0 regime when redemption ≋LBR is banned/censored.

Math: sums of x^a y^b are bad at approximating max(ε, x^a y^b)

Does this idea have legs or are we stuck with a single entity choosing weights?

Nice thing about Libra's CFMM constraints: They're convex!

Lots of ReLU constraints — and as @GuilleAngeris and I showed, you can construct curves that respect these.

P.S. @CeloOrg has a CFMM!

tl;dr: @Libra_'s ≋LBR is basically pricing borrowing/lending rates for users and association members. Fixing SDR weights is very 20th century Bretton Woods, but some of the automated lending mechanisms in DeFi could provide fairer 21st century market-driven mechanisms.

P.P.S. The @CMEGroup's pro-rata order book for the Eurodollar pack (remember: $1T/liquidity per side is not uncommon!) is sort of a halfway house between an AMM and a FIFO limit order book — CFMMs actually could work well here to split the difference

also, p.s. Nice article, @max_bronstein