Discussion of "Stablecoin Runs and the Centralization of Arbitrage"

Paper by Ma, Zeng, and Zhang

Discussion by: Andreas Park

 

 

What is a stablecoin?

digital representation of a unit of a fiat currency on a blockchain

pulled from Nick Carter's talk on "Will stablecoins serve or subvert U.S. interests?"

stablecoins are crypto's/blockchain's first "killer" use case

BTC, ETH

HQLA: USD, EUR

asset (gold)

fee-backed

Seigniorage

Crypto

Traditional

Algorithmic

Collateral-Backed

Taxonomy of Stablecoins

DEPOSITS

other assets

JPM coin

USDC

USDT

UST, Basis, Neutrino

DAI, FEI

Collateral Backed Stablecoins: USDT & USDC

\(\Rightarrow\) 5% over-collateralized

primary market acces: 6 entities only

Collateral Backed Stablecoins: USDT & USDC

  • "Cash at Reserve Banks" once was SVB
  • Reserve fund = short-date US treasuries & overnight repos

primary market acces: 560+ entities

What makes a Stablecoin stable?

USD-USDT (6 months)

\(\Rightarrow\) need a primary/reference market mechanism to allow for forces of arbitrage to align prices

  • stablecoins are issued 
    • by a single entity or
    • a blockchain-based algorithm (smart contract)
  • they trade in a secondary market
    • on crypto-exchanges against fiat
    • on crypto-exchanges against cryptos
    • on-chain against other tokens
  • \(\Rightarrow\) stablecoin price fluctuates

Arbitrage when price(stablecoin)>$1

arbitrageur

issuer/ primary market

secondary market

What makes a Stablecoin stable?

Arbitrage when price(stablecoin)<$1

arbitrageur

issuer/ primary market

secondary market

This paper: Runs!

Key Model Ingredients and what they do

1. stablecoin with primary and secondary market

2. liquidity transformation by stablecoin issuer \(\to\) Diamond-Dybvig bank run model

3. Morris-Shin-style global game

  • secondary market trading creates price dislocations
  • requires arbitrage related liquidations in the primary market
  • arbitrage related primary market liquidations require early sale of illiquid asset
  • prospect of early sale creates run risk
  • primary assets subject to fundamental risk
  • global games creates unique equilibrium for run
  • run threshold is the key variable of interest

What's the Key Tension

  1. price drop stablecoin to USD prompts arbitrageurs to
    • buy stablecoins for USD in secondary market
    • sell stablecoins for USD at parity in primary market

       
  2. to exchange stablecoins for USD at parity issuer needs to

    • find cash or

    • sell illiquid assets

Issues in the Secondary Market 

Issues in the Primary Market 

What's the Key Tension

What's the Key Mechanism

  • running has costs
    • you lose future convenience yield
    • you sell when the price is below parity
    • you have price impact
  • (fire-)sale of illiquid assets is costly and reduces value of stablecoin backing
  • highly illiquid backing makes costly sale more likely
  • increases run risk

key implication:
less liquid backing \(\to\) more run risk

  1. price drop stablecoin to USD prompts arbitrageurs to
    • buy stablecoins for USD in secondary market
    • sell stablecoins for USD at parity in primary market

       
  2. to exchange stablecoins for USD at parity issuer needs to

    • find cash or

    • sell illiquid assets

Issues in the Secondary Market 

Issues in the Primary Market 

key implication:
more liquidity \(\to\) more run risk

Intuition for Key Result

key implication:


more liquidity \(\to\) less price impact

less price impact \(\to\) more run risk

more abitrageurs \(\to\) more secondary market liquidity

Model Implications

  1. less liquid backing \(\to\) more run risk
  2. more arbitrageurs \(\to\) more run risk
  3. less liquid backing \(\to\) choose fewer arbitrageurs
  4. more liquid backing \(\to\) may have higher run risk (than expected)

Comments

Comment 0: all OK with the theory?

  • theory model is fine (and well thought out)
     
  • big question: is it true and does it capture what's going on?

Comment 1: What brings prices back to parity?

1. primary market arbitrage

2. secondary market arbitrage

Arbitrage when price(stablecoin)<$1

arbitrageur

issuer/ primary market

secondary market

arbitrageur

issuer/ primary market

secondary market

Cost on Kraken: .1 bps

Source: Liao (2023) "How to preserve the singleness of money for tokenised forms of money? "

Does secondary market activity justify primary market flows?

Comment 2: Price impact argument for infinitesimal investors

  • secondary market liquidity does a lot of heavy lifting in the model
  • concern about price impact is the cited reason for run aversion
  • but: investors (prospective runners) are infinitesimal and have no price impact.
  • So ... ?

Comment 3: cranky referee #2

censored ;-)

Comment 4: issuers optimize over #of arbitrageurs?

  • key variable: # of arbitrageurs for primary market.
    • Tether: 6
    • Circle: 500+
  • In the grand scheme of things, the choice of \(\phi\), the assets to back with seems much more salient.
  • Appendix D is not convincing.
  • alternative explanation: Tether has bad access to banking network
     
  • \(\to\) cannot enable more arbitrageurs

Comments 5: odds and ends

  • concentration measure for arbitrage
    • fraction of top 5?
    • why not  Hirshman-Herfindahl Index (or 1/HHI) 
  • institutions of Tether and Circle operations
    • Tether:
      • large discrete chunks traditionally issued via Bitfinex
      • banking connection in the Carribean \(\to\) cannot redeem for \(r\)
    • Circle:
      • strong connection banking network
      • high redeem-issuance frequency \(\to\) redeem USDC for \(r\)
  • Calibration
    • CDS spread \(\not=\) CDS price (since 2009), so Appendix G requires some work (see ISDA's pricing model or Augustin, Saleh, Xu (JEDC 2020))

Comment 3: cranky referee #2

  • The formulation of theorems requires work. 
    • Proposition 2: "The run threshold, that is, run risk, is increasing in \(\phi\) if and only if \(g(\phi) > K\), where \(g(\phi)\) is continuous and strictly decreasing in \(\phi\) , and satisfies \(\lim_{\phi \to 0} g(\phi)>0\)."

    • Proposition 3: "When the stablecoin engages in a higher level of liquidity transformation, the stablecoin issuer optimally designs a more concentrated arbitrageur sector, that is, \(n\) decreases in \(\phi\) when \(\phi\) is not too large and the cumulative distribution function \(G\) is close to linear."
    • same for other propositions
  • comparative statics results are static but are described in text like they are dynamic