Blockchain Disruption and Smart Contracts

paper by William Cong and Zhiguo He

discussion by Andreas Park

2018 American Economic Association Meeting in Philadelphia

Research question:

How do
blockchain-registered
smart contracts
affect economic interactions?

Basic Setup

  • suppliers post prices via smart contracts
  • customers arrive and transact/place order
  • blockchain verifiers confirm delivery and record transaction
  1. smart contract: delivery-contingent payment => trust for unknown entrant
  2. blockchain: observing competitors activities => collusion easier 

Results

Economic Insights/Mechanism

  1. blockchain = escrow/trusted party
  2. blockchain changes information environment

Model Mechanism Review (simplified):
no blockchain = benchmark

  1. no verification => no entry
  2. incumbents play a la Green & Porter (ECTA 1984)
    • dynamic game with imperfect monitoring
    • tacit collusion on price-gauging
    • deviations from price-gauging get punished
    • sometimes punishment unnecessary/inefficient
      •  A had no customers not because B cheated
      • but because there were no customers

Model with Blockchain and smart contract

Idea

  • supplier is unknown/new
  • may not deliver a product
    • => pay for nothing
  • if consumers believe that new entrant may be unreliable:
    • => no entry.

Role of blockchain: escrow account that releases payment upon delivery confirmation

Model Mechanism Review (simplified):

with blockchain

  1. with verification => entry happens (if entrant is competitive)
  2. activities/transactions are recorded on/visible through the blockchain and they are visible to suppliers
    • improved monitoring
      • more tacit collusion
      • no inefficient punishments

Welfare Results (simplified):
with blockchain

  1. entry => average quality higher
  2. 2 incumbents only: tacit  collusion space expands
  3. 2 incumbents + entrant: when collusion possible
    • any collusion equilibrium has surplus < any eq. from traditional world
    • exist equilibria with surplus >,=,< any eq. from traditional world

Comment: What is the right benchmark?

But: there are other established market solutions!

  • intermediaries or escrows
  • e.g., marketplaces as certifiers such as Amazon, Apple's App store

 

Currently:

  • w/o blockchain = no verification 
  • w blockchain = verification

Why not:

  • intermediary verifies vs.
  • blockchain verifies

Why important?

  • getting info from outside on the blockchain is a hard problem
  • requires essentially trusted third party

Comment: is entry vs collusion the key trade-off when thinking about blockchain-registered smart contracts?

  • Main message: blockchain presents trade-off
    • verification => trust => entry (good)
    • visible activities => better monitoring => more collusion (bad).
  • Applicability and assessment of surplus trade-offs depend on choice of benchmark.
  • Delivery verification can obtain with intermediaries 
    • => Key feature of blockchain is non-intermediated decentralized interaction

Comment 1: Key question reg. smart contracts in this setting is/should be the trade-off that non-intermediated decentralized interactions bring.

Model Ingredient Review (simplified): product market

q_A,q_B,q_C
qA,qB,qCq_A,q_B,q_C

what are the buyers' decision rules?

what's the role of qualities? why needed?

how do I interpret this in finance?

for most of the paper: truthful verification

Model Ingredient Review (simplified): verifiers

  • Four items of interest:
    • the truth: w
    • the info about w: x
    • the vote on w: y
    • the bias in the vote
  • Q: How do we interpret a bias?
    • ​​For price quotes that are fed by an oracle?
    • For UPS delivery information?

Comment 2 (part 1): the main results in the paper don't use the verifier and instead assume perfect revelation. 

Review of model extensions

  • imperfect consensus may still lead to entry (unless bias is large) 
  • privately observed quality can be included
  • only these two extensions exploit the lengthy consensus modelling and the qualities

Comment 2 (part 2): These results have an appendix-like/robustness-check feel. Why not simplify the model and shift all the complex material to an online appendix? 

Broader Thoughts

  • Collusion (or accusation thereof) is a genuine concern
    • highlighted in Bank of Canada's Jasper II experiment.
  • In practice, tacit collusion is difficult to establish, in particular with many players
    • competitors have an interest to be able to plausibly deny collusion
    • => can they take active steps to create plausible deniability?
      • state channels, multi-IDs, or Zk-snarks?
  • Paper does talk about the design: permissioned vs. public (Sec 4.3 vs 4.4), regulatory nodes, who sees what (Sec 5.1), but the discussion is dispersed throughout the paper.

Comment 3: Does the paper address a question on blockchain design question? If so, organize the paper as such.

Conclusion

  • Blockchain:
    • decentralized ledger that allows peer-to-peer transfers of value
  • smart contracts on blockchain
    • automated execution of contracts with verifiable parameters
  • This paper:
    • smart contracts expand the contracting space relative to world without any verification
    • blockchains change the information environment and affect behavior 
      • follows Yermack (2016) and Malinova & Park (2016).

Summary Main Comments

  1. Welfare/surplus results depend on benchmark
    • IMO,more suitable benchmark would be setting with intermediary
    • should bring out trade-off from core blockchain feature (= enabling non-intermediated transactions)
  2. Many model ingredients aren't used except for results with appendix-feel
  3. If there is a market design question, need to bring it out clearer.

Congrats on getting into the
Review of Financial Studies!