Transcript
Informed Trading in Parallel Auction and Dealer Markets: An Analysis on The London Stock Exchange Pankaj K. Jain Christine Jiang Thomas McInish and Nareerat Taechapiroontong Department of Finance, Insurance, and Real Estate Fogelman College of Business and Economics The University of Memphis
Objectives and Contributions • Test whether intensity of trader anonymity is correlated with trading with informed traders (adverse selection) – Use permanent price impact (PPI) of trades to gauge information content of orders in 2 parallel markets • Provide evidence based on unique structure of the LSE. – Compare adverse selection problem between parallel anonymous Auction market and non-anonymous voluntarily Dealer market. – Fully time-synchronized markets and no firm-specific differences (same firms) 2
Fragmented vs consolidated markets • Market Dominance Hypothesis – Chowdry and Nanda (1991 RFS) • Winner takes all. Migration by both uninformed and informed to the most liquid market.
– Glosten (1994 JF) • Is the electronic limit order book inevitable?
• Market Co-existence Hypothesis – Madhavan (1995 RFS) • Consolidation with full disclosure of trading information • Fragmentation without disclosure of trading information 3
Previous Studies on Trader Anonymity • Survey: Institutional investors prefer to trade in anonymous automated execution systems that provide low disclosure of identity of the company submitting the orders. Economides and Schwartz (1995) and Schwartz and Steil (1996)
• Theory: Negotiated dealer market serves as a screening device to eliminate informed trades. Seppi (1990) and Pagano and Roell (1992)
• Professional non-anonymous relationship between specialist and brokers reduces the adverse selection problem. Benveniste et al. (1992)
• Off-exchange dealers are likely to “cream skim” order flow and divert informed orders to on-exchange market. Easley, Kiefer and O’Hara (1996)
• Upstairs dealer market facilitates searching and matching of order flow. Seppi (1990), Burdett and O’Hara (1987) and Grossman (1992)
4
Previous Studies on LSE & other ║markets • Friederich and Payne (2007 EJ) – execution and information risks govern the choice of order execution venue between SETS and dealer markets – market wide liquidity shocks generate commonality – off book liquidity suppliers perform stabilization like specialists
• Ellul, Shin, and Tonks (2005 JFQA) – Opening and closing call auctions – call market dominates the dealership system in terms of price discovery – call suffers from a high failure rate to open and close trading, especially on days characterized by difficult trading conditions – call's trading costs increase significantly when (a) asymmetric information is high, (b) trading is expected to be slow, (c) order flow is unbalanced, and (d) uncertainty is high
• Other parallel market studies – – – –
Heidle and Huang (2002) NYSE/AMEX vs Nasdaq Gramming, Schiereck, and Theissen (2001) ║ Frankfurt systems Booth et al. (2002) upstairs price impact lower in Helsinki Smith et al. (2001) upstairs price impact lower in Toronto
5
Order flow of SETS stocks on the London Stock Exchange Customer submits order to member firms with/without trading venues
Member firm handle order in one of three ways according to customer’s instructions
Dealership system executes entire order against his own inventory (principal cross) or matches order with other customer’s order (agency cross)
Mix partially executed in dealer systems and work the rest in limit order system.
Member must report all trades from dealer systems within 3 minutes, except “Work Principal Agreement” orders.
SETS limit order book system submits as market order that executes immediately or as a new limit order
All orders executed in SETS limit order system are automatically reported. 6
Comparison of SETS
and
Dealer markets in 2000
Trading mechanism
Order-driven electronic limit order book market
Quote-driven multiple dealer telephone market
Liquidity provided by
Public limit orders and voluntarily dealers
Dealers
Access
Members only
Members only
Trader Anonymity
Pre-trade but not post-trade
Non-anonymous
Pre-trade transparency
All outstanding limit order book prices and sizes are available to member firms. A member firm can observe the entire limit order book and the ID number of the broker placing the limit order.
No pre-trade information is available to public. Quotes are provided based on bilateral inquiry.
Post-trade transparency
Immediate trade report. Identity of the counterparties are fully revealed when transaction is confirmed.
Trade report delay up to 3 minutes and incomplete for Work Principal Agreements
Minimum order size
No minimum
No minimum; Smaller orders are generally routed to retail service providers (RSPs) for immediate execution
Settlement period
T+5
No standard settlement
7
Data selection and processing • Main data source: London transaction data service • Compustat global file used for Market capitalization
• SETS stocks: FTSE 100 or FTSE 250 in 2000 • • • •
Trading days >=80 days in 2000 Sample stocks: 177 Delete 28 stocks for methodological problem Final sample: 149
• Trade Reports File contains: – Trade direction (buy or sell) – Trade location (SETS or Dealer) – Code that identifies each counterparty, but there no information concerning their actual identity
• Standard trade and quote filters are applied
8
Methodology • Keim and Madhavan (1996) and Booth et al. (2002)’s price impact measures: – Permanent price impact = BS*ln (PA/PB) =inform. content – Temporary price impact= BS*ln (PT/PA) =liquidity cost – Total price impact = BS*ln (PT/PB) Note: BS is buy/sell indicator; PB,PA,PT are before, after, and trade prices
9
0.0003
Percentage Cumulative Returns
0.0002
0.0001
0
-0.0001
-0.0002
-0.0003 -20
-18
-16
-14
-12
-10
-8
-6
-4
-2
0
2
4
6
8
10
12
14
16
18
20
Trade re lative to trade at time 0
SETS buy
Dealer buy
SETS sell
Dealer sell
Fig. 1. Cumulative average returns around large GBP trades. We identify the 5% of trades that have the greatest GBP value. We label each of these trades, in turn, as trade 0. For each trade 0, we identify the twenty previous trades, trades -1 through -21, and the subsequent 21 trades, trades +1 through +21. We calculate the return for each trade from -20 to +20 as the difference in the log of the trade price minus the log of the previous trade price. These returns are averaged and 10 cumulated beginning with trade -20. Mean values of cumulative average returns are plotted above.
Table 4. Information Differences on SETS and Dealer Permanent Price Impact Independent variables Intercept SETS Cap Price Volatility Freq Size Adj. R2 F-value
Coefficient -0.2224 0.2849* 0.0008 0.0274* 0.1031* -0.0399* 0.0311* 0.7062 119.97
t-statistics -1.74 21.62 0.08 2.00 4.90 -4.77 2.28
Temporary Price Impact Coefficient t-statistics 1.3699* 12.19 -0.1956* -16.92 -0.0024 -0.26 -0.1060* -8.82 0.1176* 6.37 -0.0180* -2.45 -0.0948* -7.93 0.7836 175.58
11
Conclusions •
• • • •
•
Regulators of the London Stock Exchange accomplished their goals of providing efficient markets by offering alternative trading venues. Dealers compete effectively with SETS More number of trades on SETS Larger trade size on dealer market Price impact measures suggest that SETs trades have larger information content Dealers effectively screen out informed traders or charge them more for providing liquidity 12
