Building Beller

grolfP,
Perf ormdnGe

COMMENTARY 82 DECEMBER 2OO4

MELTING ICEBERGS
ln Commentary 68 of October 2001, we noted that transaction cost icebergs
were drifting, but the advent of decimalization and a changing market
environment made the direction a tough call. Early indications were towards
lower costs, but we feared higher costs. We were wrong!

The obvious question is'what happened?' We r'*rg*' t:

identified three oossible sources of lornrer costs:
, u o
---"*l1l',1!t 19::'9:91
oo
. changing market technology - making buyside
-6
(,o
desks more efficient and reducing brokerage costs, o
g
. changing market structures - decimalization has p
g
reduced impact without increasing liquidity search
costs, and * Timlng . lmpact Comm
. changing buyside behavior and composition
orders have much less momentum and are more The obvious question is: "what happened?" We
liquid, and despite the rise in hedge funds, volatility identified three possible sources of lower costs: 1l
is at an 8 year low. changing market technology, 21 changing market
structures, and 3l changing buyside behavior (and
All three have contributed to lower costs and, composition). We will discuss each of these along with
consequently, to lower PAEG/L's@. But we see signs their possible contribution to the overall cost drop. We
that costs are flattening. As money flows back into will then discuss what we are currently seeing - and try
equities and starts moving from sector to sector the to provide some 'auguring' into what costs are likely to do
trend may change once again. Stay tuned. in the near future.

The chart below shows the dollar weighted trading costs Changing Technology
for US Large Cap trading between 2000 and ZVUJ.
\4/e recently sent al! clients a copr" of
the Plexus-sponsored report by the
Tabb Grouo entitled "lnstitutional
Equity Trading in America: A Buy-
Side Perspective." (Available on
www.plexusgroup.com.) The report
reviewed the changes occurring in
Costs fell from a high of 122bp in 2000 down to 58 bp in technology and how those changes are impacting buy-
2003 (a 64 bp or 52o/o drop). More importantly, traded side traders. The key points:
returns net of costs rose nearly +50 bp, showing that
reducing costs had a positive impact on returns. The 1. OMS systems have reduced the delay between
cost drops were similar in percentage terms for all Market the time that orders hit the desk and the time that
Caps, across Large and Small desks, and across other traders can actually start to trade.
Regions. 2. FIX and OMS connectivity has improved routing of
smaller or very liquid orders for automated
execution - allowino traders to focus on the laroer
 orders where they can add value. advantage to the buyside from ECN's.
3. Aggregation across markets/ECN's has improved
However, the ECN numbers are understated. Many
access to liquidity - and more than offset the
reduction in dealer liquidity due to decimalization. traditional brokers either have their own ECN or use
other ECN routers. Full service brokers have been the
Further, capturing a montage on markets may
actually improve information that was lost when early adapters of ECN routers, and their use between
2000 and 2003 grew from 50% to 59% while their costs
markets became fragmented.
fellfrom -114bp to -59 bp. Direct ECN impact is muted,
4. Algorithms- disciplined tools - ensure that traders but ECN's continue to influence buyside behavior and
work orders in an efficient and relativelv
force traditional brokers to keep costs competitive.
inexpensive manner.
Algorithmic trading holds great promise, but appears to Commissions, surprisingly, are the one cost that has
be used primarily by hedge funds, brokers, and a small risen since 2000. This is a consequence of the shift
handful of traditional buyside firms. Most buy-side firms towards agency trading in NASDAQ plus lower
continue to use MOC, MOO, or VWAP strategies; with the underlying equity prices (the average dropped from $43
primary emphasis on accessing liquidity rather than cost in 2000 to $26 in 2003).
minimization Surprisingly, we don't see cost differences
yet between clients using sophisticated algorithms and
those who trade more naively. Small desks routing Gommissions Rise, Yet Erckerage Ccsts F:ll
orders to brokers had just as great a cost drop as those q
o
with sophisticated strategies. While this may change 50

over time, the biggest advantage of technology appears o 40i

O so-L
to be making buy-side traders and brokers more efficient. 0
o
0 ro l-
o 0
co
2000
Changing Market Structure 2001
s lmpact
2002
* Comm

The biggest change is

decimalization. We looked at ri.'i': i.E i ?rjil:ii:i t',+1 ;ri;i;rFi' t+;l' t$i+.'ii:J'i:F',':..ii i,:--.= $. i:,i,i
the early impact in 2001, and
found that despite fears to the i'r=i",:, ;:,*; .-g11;j,{.f 9, {}fl}u.{r.:,T '...X,j#f .i1;+,.#i:-;:, .T :i".};E ti:l..Z
contrary, trading costs were
already coming down. We
i=+;ii
attributed the drop to much
But even as explicit commissions rose, increased agency
lower intraday volatility, with the caveat that costs would
trading led to broker impact costs falling at an even faster
likely rebound once markets became more active. And
we were right - but for only a short period of time.
rate. On a combined basis, brokerage (impact +
bp. For small caps, the
commission) fell from 55 bp to 35
lmmediately following 9111, and again during the lraq
drop was even greater (84 bp to 54 bp), but for
invasion, trading costs did rise, but they fell again in short
MicroCaps (. $SOO MM), the drop is only from 77 bp to
order.
68 bp. In these cases, penny commissions add up.
The second significant change was the 1997 Order
Handling Rule and the ATS Act. This opened up the Changing Buyside Behavior
NASDAQ to both ECN's and to Alternative Trading
Systems (ATS's). The result was an agency alternative
for the retail investor, which in turn attracted greater flow
Perhaps the biggest contributor to lower costs is the
change in buyside behavior. One of the most striking is
from the institutions.
the reduction in manager and analyst expectations. As a
lronically, ECN's and ATS's had a smaller direct result, we see less chasing of new market favorites and,
contribution to the drop in costs than we initially assumed. more importantly, less panicked flight due to bad
Their percentage of dollars traded grew from about 8% in. earnings. In addition, the influence of retail day traders
2000 to 12o/o in 2003, lower than many would expect. has been greatly reduced. While Hedge funds, the new
Total trading costs for this class of brokers (including the wave of investors, have become a significant market
delay cost of waiting foriseeking liquidity) fell from -52 bp participant, their quant orientation has not led to
to -48 bp. These numbers are slightly better than the increased volatility (volatility is at an 8 year low).
overall average of -55 bp, but eliminating the commission Managers have also pared back illiquid holdings, and
advantage (2( vs 4-51) results in no additional cost have again embraced the logic of diversification.
 Changes in Desk Order Characteristics also bottomed in 3Q03, but have remained stable, but
with a greater divergence of individual client results. The
Avg. Shares % Daily Vol 2 day l\/lomentum
momentum/size/cost relationship will bear continued
2000 68,400 39o/o 2.O1o/o
monitoring.
2001 64,300 360/o 1.35o/o

2002 64,000 3OVo 1.12o/o

. , : :'

2003 60,400 31o/o 0.794/o What Abtaut Other Markets? t',," i,- ',
,,

While reducing the liquidity demands is helpful, the Most of the discussion above has focused on US large
biggest change and contributor to lower costs cap stocks. However, we saw similar cost and
appears to be the drop in momentum. As the table characteristic trends in Europe and in smaller cap stocks.
below reveals, the percentage of momentum outliers The first set of charts looks at US small caps.
fell dramatically since 2000 (Adverse Momentum fell
to 8.8% of 1Q04 orders). While there has been little
change in the cost of Adverse momentum orders, the
-c
1 8 percentage point reduction accounted for 47 of the
a
64 bp drop in toial costs since 2000. On the flip side, tt
O
the buy-side also gave up some of the captured gains
c
for favorable orders. but the effect was minimal. E
o
F
2000 2003 2000 2003 2000 2003
Changes in Order Momentum
* Timing : lmpact Comm
Adve rse N eutral Favorable
2000 2003 2000 2003 2000 2003 Costs dropped across the small cap spectrum, down
7o orders 31% 134/a 55o/o eml 140 6%Y
50% in both the Mid/Small cap range, and a 42o/o drop for
I Micro caps. Like the Large Caps, the MidCap ($1 - 108)
-261 bp -234 bp -85 bp -48 bp +4 bp
Cost i cost droo reflects a combination of reduced momentum
and improved relative liquidity.
But only lower momentum
': ':..: '1,: ';".
i 'i,
'.:;" : -:.,.-i.: : .:": :: helped the Small and
'; :j:t: ,.:.:: ..:,;..: i
..,... ,,:1 ,: ; :- i. i.i.,i,::,
MicroCaps. And the MicroCap
drop in momentum is smaller
than the other cuts, consistent
with the smaller reduction in
trading costs.
Equally impressive is the drop in the cost of Neutral
Changes in Small Cap Characteristics
orders. The next table provides some more perspective:
$1-108 <.5u
Avg ok Dail,1 Avg. % Dail;, % Daily
Ghanges in Neuiral Momentum Order Size & Costs Mom. Vol. Mom. Vol. Mom. Vol.
+250K 50 - 250K < 50K 2000 2.O4 OZ-/o Z. JJ 109% 2.22 198o/o

2000 2003 2000 2003 2000 2003

2003 0.89 44o/o 1.38 1O5o/o t.o/ 217o/o
% orders O+'/a 30 -/o lo70 1A% 20o/o zo-/o
The net conclusion is that Mid and Small Caps have
Cost -168 bp -84 bp -93 bp -50 bp -49 bp -29 bp
benefited as much as Large Caps in both the changing
marketplace, technology, and manager behavior. But the
Two changes stand out; a higher percentage of small
MicroCap response has been more muted - although
orders and lower costs across all sizes. We don't have
costs are down 42o/o - reflecting the reality that these
momentum distributions for each size cut, but the
stocks continue to march to different drummers.
average momentum fell 55% (to 1.22oh) in the +250K set,..
by 610/o (to .66%) in the medium size and by 62% (to 2003 European Equity Costs / Characteristics
.25o/o) in the small trades. These numbers go a long way Size 2 day o/o Daily
Timing lmpacr Comm Total
towards explaining the cost drops. ($,000's) Mom. Volume

EU 2.O4 62Yo Z.JY 1O9o/o 2.22 198o/o 460/o

We found that the percentage of large orders bottomed in US 0.89 44o/o 1.38 105% 1.67 217o/o 31
2Q03 and have risen each quarter since. Overall costs
The table below provides additional insights. When we
comoare the 2003 European measures in the table to US More Global Warming?
large caps, we find: Or Wiil the lcebergs Grow Again?
Consistent percentages of momentum outliers.
Europe also has a higher percentage of very large Obviously, the last four years have deflated both
or0ers. market expectations and trading costs from the
European costs are lower in each momentum 3Q2000 oeak. but is 2003 a cost bottom? The real
category, but only in the largest order category. This question remains: what do we see as the future trend
is a consistent pattern. in costs?
The difference in large Order costs is all broker impact
- 7 bp for European stocks vs. 27 bp for large US 1Q04 Large Cap and European costs were slightly
orqers. lower than the 2003 average, but US Small Cap costs
started to rise in 4Q03. The confluence of change in
Changes in European Characteristics the exchange structure, trader skills and tools, and
Momentum Order Size investor behavior has resulted in radically different
Adverse Neutral Favorable +250 K 50-250 < 50K markets in a short period of time. As the investment
% Orders 11o/o 51Vo 5"/o oc'/o 1 4Vo 20%
environment evolves, changes in market structures
and trading skills/tools will have a lasting inflr-renee
Cost 177 bp +67 bp -57 bp -3i hn
But we are not as certain about behavior; market
Focusing on Large orders, the first big difference is lower participants have a knack for repeating mistakes, and
momentum for European orders (.9% vs. 1.3%). Butwe a move towards less disciplined stock selection at
suspect that the real difference is the combination of a some point is guaranteed.
smaller stock universe for large orders, less underlying
momentum, and greater emphasis on Full Service We guessed once that higher costs were in the offing.
brokers. Reducing broker's risk allows for better This time, with costs at a much lower level, we make
bids/offers for sizable trades - and leads to lower overall the same (albeit safer) call. But we doubt that trading
costs. costs will approach the 2000 level any time soon.

Plexus Neurs ri
;i
Put this on your calendar! The Plexus Group 9th Client Conference is scheduled for September 18 - 20,2005 at fne .lji
,.t!
Fairmont Turnberry /s/e Resorf & Club (http:/lwww.fairmont.com/turnberryisle/)
:i
t::-i:: .:r:-.::'::l"r +ir.n::iii=r-:.rr:-l::t:a,:j:.:.r,aj- i;il{

Reprint any portion with credit given to: il

g roltP. ij.
.:::

*
11150 W. Olympic Blvd., #1000 Los Angeles, CA 90064
PH: 310.235.3700 FAX: 31 0.31 2.5506 www.plexusgroup.com H
Plexus Group rs a subsidrary of JPMorgan Chase, N.A. E

@ 2005 Plexus Group, Inc. #

tr
E+

Eq

+ln::1-:::::i-.j,a::.:=i+i-jii,:+:.::.rili:l:':i,+:r:r_ljr{tijri.f*r-;t:rt ii-::i:=--:.:-i::=.::-..i-:=:iY:i::i;1ii l::r.+l:!:.:'jltr?:Fl:;::.$