Title: DETERMINANTS OF THE COMPONENTS OF THE BIDASK SPREAD ON THE LSE
1DETERMINANTS OF THE COMPONENTS OF THE BID-ASK
SPREAD ON THE LSE
- Dr. Evangelos GIOUVRIS
- School of Management
- Royal Holloway, University of London
2MOTIVATION
- Changes in trading systems
- Floor trading to electronic trading
- Batch trading (discrete time) to continuous
- Quote driven (obligatory market making) to order
driven e.g FTSE100 stocks - Quote driven to hybrid e.g FTSE250 stocks
3GENERIC RESEARCH QUESTION
- How do changes in trading systems affect
information dissemination, risk perception and as
a consequence the cost components of the bid-ask
spread?
4BID-ASK SPREAD THEORY (1)
- Spread has three components inventory, order
processing and asymmetric info - Stoll (1989) decomposes spread making the
following assumptions - The market is informationally efficient so that
expected price changes are independent of current
and past information. - The bid-ask spread is constant over time and all
transactions occur at the highest bid or the
lowest ask - The proportions of bid-ask spread components are
the same for all securities
5BID-ASK SPREAD THEORY (2)
- CRITICISM spread components are biased as a
result of the existence of positive
autocorrelation in returns which leads to a
downwards bias in the estimation of the realised
spread - George.J, Kaul.G Nimalendran.M (1991) correct
correct for this by estimating returns based on
transaction prices and bid to bid - Kim Ogden improve by using mid-quotes
6RESEARCH HYPOTHESES
- H1 the asymmetric component is higher under a
quote-driven trading regime and lower under an
order-driven trading regime - H2 the asymmetric information component of the
spread does not change between quote driven and
hybrid markets. - H3 volatility appears to have a stronger impact
on both cost components under a quote driven
regime but reduces significantly under an order
driven regime (FTSE100). - H4 volatility appears to have the same impact on
the cost components of the spread under a quote
driven and a hybrid market. (FTSE250)
7COST COMPONENT REGRESSIONS
- Si2-cov(RDi,tRDi,t-1)1/2
- Si?0?1SQi?t
- Si?0?1Di?2SQi?3(DiSQi)?t (GKN regression,
difference between transaction prices and
subsequent bid quotes) - Si?0?1Di?2SQi?3(DiSQi)?t (KO regression,
difference between transactions prices and
mid-quotes ) - Si?0?1Di?2SQi?3(Di(1/TSSQ2i)1/2)?t (KO
regression with KO estimator)
8COST COMPONENTS OF THE BID-ASK SPREAD FOR FTSE100
STOCKS
9COST COMPONENTS OF THE BID-ASK SPREAD FOR FTSE250
STOCKS
10DETERMINANTS OF COST COMPONENTS REGRESSIONS
- closing spreadit a0 a1Di ?0NTit ?1(DiNTit)
?0TVit ?1(DiTVit) ?0VOLit ?1(DiVOLit) ?it - spread(every trade) it a0 a1Di ?0TVit
?1(DiTVit) ?0VOLit ?1(DiVOLit) ?it - asymmetric informationit a0 a1Di ?0NTit
?1(DiNTit) ?0TVit ?1(DiTVit) ?0VOLit
?1(DiVOLit) ?it - order processingit a0 a1Di ?0NTit
?1(DiNTit) ?0TVit ?1(DiTVit) ?0VOLit
?1(DiVOLit) ?it
11REGRESSION OF FTSE100 ALL STOCKS ON NUMBER OF
TRADES (NT), TRADING VOLUME (TV) AND VOLATILITY
(VOL) UNDER SEAQ AND SETS
12REGRESSION OF FTSE250 ALL STOCKS ON NUMBER OF
TRADES (NT), TRADING VOLUME (TV) AND VOLATILITY
(VOL) UNDER SEAQ AND SETSMM