Payment for order flow: not new ... 

• PFOF/off-exchange siphoning of retail is Not Great (for liquidity and uninformed): 
  • E.g., Easley, Kiefer, and O’Hara 1996
  • Bessembinder and Kaufman 1997
  • Comerton-Forde, Malinova, Park 2018
  • Hu and Murphy, 2022
• No Harm:
  • e.g., Battalio, 1997
  • Battalio, Greene,and Jennings, 1997
  • The SEC’s 1997 “Report on the Practice of Preferencing”

Payment for order flow: does not grow old ... 

Payment for order flow: remains controversial 

• June 8 2022, SEC Chair Gary Gensler:

• • "Payment for order flow can raise real issues around conflicts of interest. [...] distort routing decisions [...] incentivize broker-dealers to use gamification [...]

  • The European Union is actively considering banning payment for order flow, [...] joining [...] the United Kingdom, Canada, and Australia. [...]

  • I’ve asked staff to make recommendations"

• Bloomberg Sept 22, 2022: 
  • Sources: "The SEC will stop short of banning payment for order flow, a controversial way to process retail stock trades."
  • But not without the "buts" ...

Payment for order flow 

• Traditional view on issues:
  • siphoning the "dumb" flow off-exchange
  • high on-exchange transition costs 
    • \(\to\) is price improvement on high costs actually an improvement?

• SEC "most recently voiced" concerns:
  • tradeoff: PFOF vs. price improvement for customers
  • Echoed in Ernst and Spatt (2022) for options!
    • ​DMMs that pay offer worse terms for retail
    • add'l concern: "push" of retail to options trading (?)

• For more on internalization practices in options, stay tuned for Hendershott, Khan, and Riordan (2022) @ the 15:40 session!

This paper: cool new perspective!

• Zoom onto market makers/internalizers and their quotes
  • MM can trade
    • against only retail, or
    • against all (on an exchange)
• No brokers; no payments/fees/rebates
  • no conflicts of interest/routing conflicts
• No asymmetric information
  • Instead:
    • Retail and institutions may trade in different directions
    • \(\to\) retail orders help "diversify"/"hedge" against the institutional orders

Comment #1: Sheep vs. Wolves?

Define a sheep?

Comment #1: Retail vs. Institutional

• Retail = traders that can use PFOF (& an exchange)
• Institutional = can only trade on an exchange
  • No other ex ante distinction (?)

• \(\to\) The MM don't care which orders to buy vs. trade on an exchange
  • Buy retail by design only, not b/c of specific characteristics.
  • Can you imagine a market where institutional orders are bought instead?

Comment #1: Retail vs. Institutional

• Why the choice to define traders by venue access?
  • almost makes retail "more sophisticated" ...
• Ideally:
  • retail flow characteristics
  • institutional flow characteristics
  • which flow (if any) would be bought? 

Comment #1: Retail vs. Institutional

• At the very least, take a stronger stand?
• What do we know about
  • correlation b/n retail vs institutional?
  • sizes?
  • MM sector?
  • other parameters?

\(\to\) sharper predictions?!

Comment #1.2: want to know more about the true parameters!

Comment #1.2: want to know more about the true parameters!

Comment #3: key differences to the info-based models?

• Different trading motives, yet similar tradeoffs for MM
  • Retail offset the costs of trading against one-dimensional flow
  • A tad challenging to understand all the model parameters but PFOF arises, e.g.
    • institutional = small children of a large parent order
    • retail: random noise (buy and sell)
    • \(\to\) similar order flow as with info!
  • Also: "Informed" is stylized
    • A large order \(\to\) price impact & a trader knows this
    • \(\to\)they are arguably informed  ... 

Comment #3: key differences to the info-based models?

• Results that are different to info-based models arise 
  • when institutional & retail flows are negatively correlated
  • possibly b/c of differences in model setup
    • zero-profit pricing typical for info-based models (tractability?)
      • here: market clearing
    • inelastic demands of noise traders

How to empirically identify which channel is key for the wholesalers' liquidity provision choices?