Price Clustering of Shorts for NYSE Securities

1
Price Clustering of Shortsfor NYSE Securities

• A Class Project
• for FIN 631 Spring 2006
• Chris Riley
• University of Mississippi

2
Short Sales

• Short sellers are informed traders
• Christophe, Ferri, Angel (2004)
• Abnormal short selling is significantly linked to
  post earnings announcement stock returns
• Boehmer, Jones, Zhang (2005)
• Heavily shorted stocks underperform lightly
  shorted stocks by 1.07 (NYSE, mainly due to
  larger trades, more than 500 shares)
• Cohen, Diether, Malloy (2005)
• Shorting demand is a predictor of future stock
  returns shorting market is mechanism for private
  information revelation
• Diether, Lee, Werner (2005)
• Increasing short-sales predict future negative
  returns (NASDAQ, mainly due to small trades)
• Short sellers are not informed traders
• Daske, Richardson, Tuna (2005)
• Find no evidence that short sales precede bad
  news events (including earnings announcements)

3
Price Clustering

• Ball, Torous, and Tschoegl (1985)
• Price resolution hypothesis if value is not
  well known, then prices will cluster
• Harris (1991)
• Stock price clustering is persistent(time,
  stocks, market structures)
• Negotiation hypothesis traders use discrete
  price sets (that are coarser than the minimum
  tick) to lower costs of negotiating
• Nguyen, Van Ness, and Van Ness (2004)
• Find clustering on nickels, dimes, quarters on
  Cincinnati
• Harris (1991), Grossman, et al. (1997), and
  Nguyen, et al. (2004)
• Degree of price clustering varies across markets

4
Price Clustering of Short Sales

• Do short trades cluster less than non-short
  trades?
• Price resolution hypothesis if short sellers
  are more informed, value is better known, less
  price clustering of shorts
• Negotiation hypothesis price clustering less if
  negotiating costs (with the buy side) are smaller
  than expected gains (for the short seller) due to
  informed trading

5
Data Sample Description

• NYSE Trade Data
• REG SHO Short Trade
• From all exchanges where NYSE stocks trade
• NYSE-listed Stocks
• July, August, September 2005
• 930A.M. 400P.M. (Eastern)
• Trade on each of the 64 trading days
• Trade at least 100 shares a day
• Average price is at least 10
• 1544 Stocks
• Criteria is similar to Daske, Richardson, Tuna
  (2005)

6
Methodology

• Last Penny
• 10 categories xx.x0 xx.x9
• Cents
• 100 Categories xx.00 xx.99
• Matching
• Shorts are matched to the trade data
• Shorts that do not match are excluded
• Trades with both short and non-short portions are
  counted as both a non-short trade and a short

7
Price Clustering on Last Penny Non-short Trades
and Shorts
8
Price Clustering on CentsNon-short Trades and
Shorts
9
Relative Difference
10
Remarks

• Use the Lee Ready trade classification
  algorithm to further investigate the straddling
  of shorts around nickels, dimes, and quarters.
  Are these shorts more often providing or taking
  liquidity?
• To what extent are the last penny results for
  x.x4 and x.x9 driven by trades exempt from the
  tick test?

11
Misc. Summary Stats

• shorts non-shorts (matched) 175,182,525
  
• non-shorts 127,410,692
• shorts 47,771,833
• of trades with a short portion
• 0.272697479
• short / non-shorts proportion
• 0.374943674