Discussion of:
Competition and Exchange Data Fees

Paper by:             Jonathan Brogaard, James Brugler, and Dominik Roesch

Discussion by:  Andreas Park                     

May 14, 2021

8th Annual Conference on Financial Market Regulation, 2021

  • Data fees are contentious:
    • necessary for brokers to serve customers
       
    • exchanges are monopolists for their data
       
  • Question: is this market competitive? Can venues raise data fees at will?
     
  • Answer: this article tests whether investors change behavior due to fee change.

Data Fees and Exchange Competition

  1. Order protection rule (OPR) applies to the entire book
     
  2. Venues and brokers must ensure not to violate OPR.
     
  3. Data fees are regulated
  1. Since 2016: Regulation of market data fees for professional subscribers
     
  2. Venues (Exchanges and ATSs) need to file any fee changes (data and trading)
     
  3. Data fee methodology to estimate [acceptable] fee range based on contribution to price discovery and trading activity

Tales from Canada

Tales from Canada: The story of the TMX Group

And that's why people are upset:

  • trade about the same
  • but pay ever more 

When venue introduces data fees:
 

  1. competitive quotes \(\searrow\)
     
  2. market share \(\searrow\)
    • fewer ISO (strong) and non-ISO (less strong)
    • less sophisticated brokers route fewer orders
       
  3. e-spread \(\searrow\)
    r-spread \(\nearrow\)
    p-imp \(\searrow\)
     
  4. market liquidity decreases and price efficiency declines marketwide
     
  5. Part of the effect is due to some traders losing access to the order book depth information.

Data Fees and Exchange Competition

Methodology

Structure of Discussion

Measures & Interpretation

Story & explanation

Odds-and-ends

  • Methodology
    • Difference-in-difference of structural shift?

Structure of Discussion

  • Measures & Interpretation
    • Quoted spread
    • Quoted depth and %of NBBO depth
    • E-Spread?
  • Story/explanation
    • What exactly happens when data fees are changed?
    • Who makes which decision and what role does data play?
  • Odds-and-ends
  • Who consumes data?
    • trading desks (brokers)
    • smart order routers (brokers)
    • investment advisors 
    • investors (retail & institutions)
  • What exactly happens when data fees are changed?
  • Who makes which decision and what role does data play?

trading/investment

  • Should I trade at the prevailing market conditions?

routing

  • Where do I send an order?
     

One question: do they even have the free data?

Story & explanation

  • Who makes routing decisions?
    • (trading desks)
    • smart order routers
    • some customers (DMA institutions)

Constraints

  • cannot ignore venue easily (for marketable orders)
  • changing the logic in routers is expensive
  • cancelling user subscription \(\not=\) cancelling router data use
  • Analysis in Section 3.2 (broker routing) is the most important part of the paper (and my favourite)
  • I particularly like the non-sophisticated dealer part
  • also good argument for long post-event window

Story & explanation

  • Treatment: stock on the affected venue
  • Control: same stock on all other venues

Example: Volume

  • drop in volume because people don't trade for lack of cheap info
  • people shift volume elsewhere
    • Control is not a control
    • really just a market-wide event

Difference-in-difference vs marketwide structural shift?

Methodology

  • Alternative Methodology:
    • Mediation analysis
    • see: Imai, Keele, and Yamamoto (2010) for a causal mediation analysis
    • structural model of equations (a bit like IV)
    • Average Causal Mediation Effect (ACME)
      • the direct effect (of the fee introduction: less information for trading)  
      • and the indirect (mediated) effect (shifting volume between venues)

Methodology

  • Why not use:
    • Quoted depth
    • % of NBBO depth
  • used "quote quality"
    • % of time at NBBO
        • formulation not quite clear
        • is it best bid and ask or best bid/ask?

Measures & Interpretation

  • actually hard to interpret:
    • no trade-through requires that trades on treated happen only if it's at the best
    • driven by market as a whole/other venues not just treated
    • could easily see significant worsening of q-spread 90% of day but when trades happen, all is well
    • reverse can also be true: quoted better 90% of time but trades only happen when things are worse

E-Spread?

Measures & Interpretation

  • monthly vs daily TAQ? (\(\to\) Holden and Jacobsen 2014 necessary?)
     
  • check for major fee change on venues other than  ARCA, BATS, & DirectEdge
     
  • e-spread venue \(\searrow\) but e-spread market \(\nearrow\)?
     
  • explore more whether the change in price impact is driven by ISO decline
     
  • for market-wide: provide some summary stats that inform the reader about the cross-sectional composition of the above/below median groups
     
  • I did not understand the auction segment at all and I don't know what it would add.

Odds-and-ends

final thoughts

Well-argued paper with clear research question and credible answers ("it's not quite so easy to up the fees")

My suggestion: Focus on visible liquidity & broker routing and trim rest

@financeUTM

andreas.park@rotman.utoronto.ca

slides.com/ap248

sites.google.com/site/parkandreas/

youtube.com/user/andreaspark2812/

Discussion of Brogaard, Brugler, and Roesch "Competition and Exchange Data Fees"

By Andreas Park

Discussion of Brogaard, Brugler, and Roesch "Competition and Exchange Data Fees"

  • 459

More from Andreas Park