Lifting the veil: An analysis of pretrade transparency at the NYSE - PowerPoint PPT Presentation


PPT – Lifting the veil: An analysis of pretrade transparency at the NYSE PowerPoint presentation | free to view - id: 307d4-MDRjM


The Adobe Flash plugin is needed to view this content

Get the plugin now

View by Category
About This Presentation

Lifting the veil: An analysis of pretrade transparency at the NYSE


Revenue generator for NYSE: subscribers increased from 2,700 to 6,000 between Jan and May 2002 ... Results in 1332 NYSE-listed securities ... – PowerPoint PPT presentation

Number of Views:87
Avg rating:3.0/5.0
Slides: 28
Provided by: ekkehart


Write a Comment
User Comments (0)
Transcript and Presenter's Notes

Title: Lifting the veil: An analysis of pretrade transparency at the NYSE

Lifting the veilAn analysis of pre-trade
transparency at the NYSE
  • Ekkehart Boehmer, Texas AM
  • Gideon Saar, NYU
  • Lei Yu, NYU

  • The ongoing proliferation of trading platforms
    raises important market design issues
  • We look at market transparency
  • Great regulatory interest SECs Market 2000
    study recommends greater pre-trade transparency
    (display of customer limit orders)
  • Limited empirical and theoretical evidence on the
    effects of transparency in auction markets
  • This paper analyze economic consequences of an
    increase in pre-trade transparency

  • Implemented by NYSE on January 24, 2002
  • Before OpenBook, only specialist could see LOB
  • Reveals limit order volume at all price points
    for all NYSE stocks, refreshed every 10 seconds,
    730 am 430 pm
  • Exogenous to issuers
  • Revenue generator for NYSE subscribers increased
    from 2,700 to 6,000 between Jan and May 2002
  • Represents significant supply of liquidity
  • 99 of all orders and 75 of all volume is
    submitted electronically
  • About 2/3 of these are limit orders

Our goals
  • Investigate how a change in pre-trade
    transparency affects
  • Trading strategies of investors
  • Behavior of specialists
  • Informational efficiency
  • Liquidity

Predictions for increase in pre-trade
transparency Trading strategies
  • Limit order traders face two risks that are
    amplified with more transparency (Harris 1996)
  • Information leakage
  • Front-running
  • Remedies
  • Break up orders
  • Cancel and resubmit more often
  • Use floor brokers to selectively disclose trading

Predictions for increase in pre-trade
transparency Market quality
  • Greater informational efficiency and greater
  • Baruch 2002, Glosten 1999
  • Less liquidity and wider spreads
  • Madhavan, Porter, Weaver 2000
  • Supported by empirical results from TSE change in

Predictions for increase in pre-trade
transparency Specialists
  • May trade less because they lose part of their
    informational advantage
  • May trade more as limit-order traders withdraw

Research design to identify permanent effects of
the change in transparency
  • Compare pre- and post-OpenBook periods
  • Choose the two most recent full trading weeks
    before Jan 24, 2002
  • Learning about changes in (others) trading
    strategies takes time expect gradual adjustment
    to new equilibrium
  • Use four 2-week post-event periods February,
    March, April, and May

Sample construction
  • All common stocks of domestic issuers that are
    continuously traded Jan to May 2002
  • Exclude trusts, funds, firms with multiple share
  • Results in 1332 NYSE-listed securities
  • Choose 400 of these, stratified by median dollar
    trading volume during 2001Q4
  • Standard data from TAQ, Factset, using typical
  • Proprietary data from SOD, CAUD, Lofopen

Descriptive statistics over time
Descriptive statistics over volume quartiles
Changes in order cancellation rates
January median 61
  • Columns represent median pairwise changes from
    pre-event to four different post-event periods
  • Orange indicates significance at 5 or better
    (Wilcoxon Test) grey indicates no significance

Traders cancel orders faster and more often
(consistent with Harris)
Cancellation rate (Jan 61)
Time to cancel (Jan 290 sec)
Cox - Cancellation rate
Weibull - Time to cancel
  • Duration models to control for
  • Censoring
  • Distance from
  • quote

Orders become smaller (consistent with Harris)
Limit order size (January median 543 shares)
but less intermediated (inconsistent with
Floor-to-limit ratio decreases (based on shares
Visibility effect?
Specialists become less active
Specialist participation rate in share volume
(Jan median 18)
Quoted depth () added to the book by specialists
and floor (January 60,000)
  • Increase in risk of proprietary trading?
  • Crowding out effect?
  • Shift from floor to book
  • Note reduced activity does not necessarily imply
    lower profits

How did OpenBook affect informational efficiency
of prices?
  • Hasbroucks (1993) measure
  • Decompose variance of (log) transaction prices
    into an efficient-price and a transitory
  • Compute ratio of transitory to total price
  • Deviations from a random walk
  • Compute 30 and 60 minute autocorrelation of
    quote-midpoint returns
  • Larger (absolute) autocorrelation indicates less
    resemblance to a random walk
  • A decline in either measure would suggest
    improvement in info efficiency

Informational efficiency improves
Deviation from efficient prices
  • Rather weak results
  • But direction of changes is consistent with
    Baruch 2002, Glosten 1999

30-minute autocorrelation
60-minute autocorrelation
Changes in liquidity
  • How does cumulative depth displayed in the book
  • Construct 5-minute snapshots of the book
  • Compute depth at different intervals from the
    relevant quote
  • Average across snapshots
  • Do traders pay more for execution?
  • Compute effective spreads
  • Control for changes in volume, volatility, and

Conditional book depth increases
Cross-sectional regression (N400)
Pooled TS-CS regression (N8000)
  • Different specifications yield virtually
    identical results
  • Results inconsistent with Madhavan, Porter, and

Conditional effective spreads decline
Cross-sectional regression (N400)
Pooled TS-CS regression (N8000)
  • Similar results using orders (as opposed to
  • Different specifications yield virtually
    identical results
  • Results inconsistent with Madhavan, Porter, and

Do the results just reflect a trend?
  • Markets generally became more liquid and order
    size declined during recent years
  • Are we picking up this trend?
  • Not likely
  • We measure changes in our variables from pre 9/11
    to January 2002 (OpenBook)
  • For all variables, changes are very small
    relative to OpenBook effects and often in the
    opposite direction

  • Regulatory interest in greater pre-trade
    disclosure faces academic debate without
  • We find, contrary to evidence from the TSE, that
    opening the book
  • Improves market quality
  • Encourages active order management
  • Changes the role of specialists and floor brokers
  • Our analysis reemphasizes the importance of
    market design for innovation and as an instrument
    for regulatory changes

Why do results differ between NYSE and TSE?
  • Network technology (TSE change preceded
    internet) traders may not have been able to use
    the book effectively for technological reasons
  • TSE displayed only a few ticks beyond the inside
  • Simultaneous other changes on TSE (e.g., reserve
    order display requirements)
  • TSE had two-tiered system, one was electronic

Cumulative shares on the book weak increase
Categories 0.16, 0.83, 3.3, 16.7 from quote,
entire book Based on five-minute snapshots
Effective spreads decline
Group 1 most active stocks
Group 2
Group 4 least active stocks
Group 3
Descriptive statistics time dimension
Daily medians for all 400 sample firms
Descriptive statistics cross-sectional dimension
January pre-event period medians
Price volatility
  • Liquidity increases for every trader?
  • Previous measures may not appropriately reflect
    cost of trading large positions that are broken
    up and executed over time
  • Higher intraday volatility means greater price
  • On the other hand, higher volatility may make
    limit orders more profitable
  • Measure volatility as daily price range

Intraday volatility increases
Based on daily price ranges (January 0.60)
Effects on share valuation partial analysis of
welfare consequences
  • CARs and sensitivity to liquidity changes

Welfare consequences
  • Focus on share price responses to change in
    market structure
  • Liquidity affects required returns (Amihud and
    Mendelson, 1986)
  • Gradual change in trader strategies and liquidity
    makes it difficult to isolate valuation effects
  • Underlying assumption liquidity effects are not
  • Estimate market model of daily returns during
  • Compute CAR from Jan 24 to the end of each
    post-event period
  • One-factor model using value-weighted NNM/AMEX
  • Three-factor model adding JPM commodity futures
    and US T-Bond indexes
  • Estimate relation between CAR and changes in

OpenBook is followed by a period of positive
abnormal returns
Standard event-study test Wilcoxon test on
median CAR
Alternative test Pooled regression, Wilcoxon
test on median daily post-event dummy
  • Estimate pooled regression of returns on three
    factors and daily dummies, one for each day of
    the post-event period
  • Uses only time-series variation in coefficients
    to construct test
  • Alleviates correlation problem due to calendar
    time clustering

CARs increase with declining relative effective
  • Regress CARs cross-sectionally (from beginning of
    pre-event to the end of each post period) on
    corresponding changes in relative effective
  • Coefficients measure responsiveness of CAR to
    changes in spreads, and medians are all negative
    and significant

Each month shows the median coefficient on
changes in relative effective spreads for each of
the four volume groups (ordered from most active
to least active)