9rorrID

COMMENTARY 68 ocToBER 2001

DRIFTING ICEBERGS; PERCEPTIONS AND REALITY IN TRADI/VG COSIS
A recent survey concluded that market academics believed the markets were improved by regulatory
changes while traders felt the opposite. Who is more correct? Surprisingty, both, but in different dimensions.

Has Market Quality Deteriorated?
Late last year, Bob Schwartz and Dan Weavs surveyed both
buyside/sellside practitioners and academics for their cunent
perceptions of market quality. The key issue was whether regulatory
changes that reduced spreads and increased aress to markets had
improved market effciency. The academics viewed the changes
favorably;however, both buyside and sellside traders were less
enthusiastic. While the 'new market' may be friendlier to the retail
investor, the overwhelming perception was that regulations have led
to higher msts and more difficult trading conditions for Institutions.
ls perception reality? The trade/s views were shaped a by
combination of increased volatility, market fragmentation, and heavier
orderflow as assets exploded while the market soared and dipped.
But is more difficult trading the'fault'of the market? Moreover, is the
new market truly more expensive?
Reality: Lower Cosfs
Years ago Plexus introduced the use of an iceberg to describe
execution msts. At the tip was commission, easily seen but relatively
insignificant. Market lmpact (Brokerage)was the next layer; more
significant, but still small relative to what lay under the water. Plexus
measures lmpact as the difference between the release pnce and the
average execution pnce on the trade day. The hidden msts lie in the
multiday Timing and Opporlunity categories as institutional traders
Cap trades. lmpact costs include spreads; yet the total is rising even
though spreads are shrinking. The nse in lmpact was accompanied
by a fall in multiday delay msts. This anomaly hints that the nature ol
the trading has shifted the cost structure in ways unrelated to market
procedures. We shall see that these changes blur perceptions of
trading costs. In the next few sections we will show that: 'l) Trading
became more concentrated; 2) Momentum conditions lightened,
icadng io ic;v.vered rcsts; and 3) The ffight to safety shifted institutionai
trading focus from small cap stocks to lower-cost large caps.
More Concentrated Trading
The table shows that the increase in lmpact results from an increase
in very large daily trades exceeding 250,000 shares:
. The percentage of very large trades soared, accompanied by a
tendency to concentrate heavy trading under the control of a single
broker, especially in NASDAQ trades. Increasingly, traderc are using
brokers to search out liquidity wherever it can be found.
. lmpact mst for large trades increased even more signifcantly. Costs
for small trades may be shrinking, but that is a moot point for
Institutions : ov er 7 5o/o of the daily trades by ou r clients results from
orders exceeding 50,000 shares.
One Day Trades Are Larger - And More Expensive!
NYSE % Chanqe NASDAO %Change
% of Total Activitv
1Q2001 41 20 39 35
1 999 u 29
Averaqe lmpact

search for, and sometimes never find, liquidity.

1 Q2001 _?7 hh 42 228
1 999 -26 bp -29 bp

Tlischaft below shows iceberg data fur ieca fquarteis;Lsing iire
1999 average as a reference. Here we focus on execution costs and
omit Opportunity effect, which will be discussed later.
Large and Small Cap Trading Costs since 1999
Sizeand 1mp-act aione do not teJithe wl-oie stoi-y. Price i-nonrentum
during trading represents the greatest challenge for traders since it is
the most impoftant determinant of cost. The percentage of orders
encountering adverse momentum conditions dropped 3 percentage
l"o" points for both Large and Small Cap funds, to 25% and 31oh

250
Larse caps Smatt caps : IiJ' respectively. This three percentage point drop leads to a 15% drop in
momentum trading costs. This drop was partially offset by reduced
200
150
gains from trading orders in Favorable market conditions.
100
50
0
Gosf of Standardized Trades
.oe Cr""t.d" r"""."$ nouo ."tt rr"t
^d
We've shown that shifting trading emphasis accounts for most of the
""s"gs drop in msts, but the question remains: have costs dropped, or not?
Plexus data shows that costs remain flat for Large Cap and have Plexus reestimates its benchmak mst equations each quarter, using
recently fallen for Small Cap after peaking in the Summer of 2000. the previous six months of trading data. Thus any shifts in trade mst
This is contrary to perception, or is it? We note that Schwartz and structure will be reflected in changes in the mst equations.
Weaver conducted their survey shortly after costs, especially Small
Cap, rose to their all time high. The next two graphs compare quarterly benchmark costs for
standardized Large and Small Cap stocks, having the following
Brokerage msts (lmpact, shown in red) rose for both Large and Small characteristics typical of institutional trades:
 This reinfnrces or:r helieJthat clients are bemming more attuned to
the eost of delays and cancellations. De-emphasizing low broker
lmpaet in exchange for lower Timing and Opportunity costs has likely
led to higlrer eaptured returns as well as lower total costs.

The graphs show expected costs for 0% (red) and far 4ok Canclusions
momentum (yellow), The cost figures are the average cf
Traders rjisprute the claim that new market regulations improved
NYSflNASDAQ equations for both buy and sell o de s
market uuality" F:rorn the Institution's viewpoint, lower costs at the
Standardized Benchmark eosts - Large Cap i"etail !e.;ei ha.re led to mirch higher costs for institutional size.
250
204
Flexus ieebergs show fairly rnnstant Large Cap trading costs since
150
o 1899. and Snrall Cap stockstrending cheaper. Both of these
100
J
(9 findinps arc cr:ntrary to irader perceptions.
ul 5U
o-
0 This at-a-glarrce perception may be too facile. lt appears that
Pre 99 1999 1Q00 2Q00 3Q00 4Q00 1Q01 2Q01 resJrictror:-q !n Small Cap tracling costs primarily reflect changes in
order clrar;rctenstirs The icetrerg may not look so favorable had the
Large Cap costs initially fell in '1999, and did not start to nse until late
perrei rtaqe nf A.dverse Momenhrrn orders not dropped off.
in the year, After peaking in early 2000. costs fell hack and are nr.ncr
below the historic average.
A trend lnwards higher broker lmpact suppotls the traders'
Standardized Benchmark eosts - Small eap argrrment" Shi"inking spreads have ied to increased premiums for
250
siz,e. ln:nicaily, traders are responding by trading more aggressively,
204
leading ta larger intra-day volr.imes. They can fincl liquidity, and the
150
o o';er"all speed *f trading as measured by daily completion rates has
o
J
(9
100 not riropped. In exchange for controlling Timing costs, traders are
ul 50 willing to pay r:p fnr mpirl fill of lsvgs orelers.
o-
0
Pre 99 1999 1Q00 2Q00 3Q00 4000 1Q01 2Q0t Tt'ris is a su*t:essful strategy when urgency is high, and leads to
lower trtal eosts. But as Lrrgency drops, we do not see a
Similarly, Small Cap trade costs forAdverse momentum orders fell
cornnrensuiate drop in lmpact" Consequently, the total cost of
of later in 2000, but the cost for neutral momentum remains high
neutral Srnall Cap orders remains high. In contrast, neutral Large
Perhaps earlier high-momentum costs were distorted by the
eap orcjere show litlle drift from histonc averages
NASDAQ bubble and then reverted backto more normal levels. We
believe. however. that recommendations to Plexus clients to trade
Vlle believe that this reflects the changes in market regulations.
more aggressively into momentum to avoid delay costs also
Smalier cap stoeks have greater sensitivity to retail traders.
contributed to the shift. "Normal" momentLrm cersts outside of this
improved aricess to markets and lower spreads have provided small
self-selected universe may be much higher.
traders wifh an opportunity to crowcl in front of Institutional sized
orders, whieh pay lrigher premiums to access size. At least in this
Completion Rates Remain High sectar of the market, reducing costs for the little guy may truly be
shifting the i.'rrils tr,rthe hig; Institutiernal guy.
Because the PAEG/L and the iceber_os reflect only traded shares,
these results may be affected by traders and managers who cancel
orderc rather than pay up. We tested this hypothesis by compadng Plexus Graup Reschedules Conference
completion rates and unexecuted retums for those clients providing
ln response to the tragic evenfs of Sepfembe r 11th. Plexus Group
complete order information. rescheduled its canference at Amelia lsland to February 24-27, 2002.
We invite your aftendance.
Completion Rates Are Holding Up
Completion Rate Opportunity Cost bp ln honor af our clients, fiends and acquaintances whose lives were ',

tragicalty tost, Plexus Group has made a cantibutiontothe Fred

Alger Families Trust.

2000 1001 1999

Reprint any poftian with credit given to:

M
tr Large Cap tr Small eap

The left side of the graph shows that completion rates have held up.
Both the Large Cap and the Small Cap funds now show slightly
grorrp
higher mmpletion rates despite larger orders and increased volatility. 1 1 1 5A W. Olynpic Blvd., l#ffi Los Angeles, CA 9m64

One result of higher completion rates is lower Opportunity costs for PH: 31 0. 31 2. 5505 FAX: 31 0.31 2.55A0 vwwu.plexusgroup.com
.:..:.:l'..j:..:...::.:.:..:::.:::..:::.:]:].:.':']j:]]:':]:.::::]:

the Large Cap funds, and steady costs for the Small eaps.