Trading Truths

plexusry Building

Perf

Betler

otmdnGe

APRIL 2OO4

COMMENTARY 80

TRADING TRUTHS How Mis-Measurement of Trading Costs ls Leading lnvestors Astray Some recently published data argues that large cap stocks are more expensive to trade on the NYSE than small caps. In this Commentary, we debunk that theory by demonstrating that this conclusion results from incomplete data and faulty logic. We also show that: . lmplementation costs are much higher than those reported using VWAP measures; . So-called "pennying" only accounts for a small fraction of the cost of trading, and . Damning the specialist-" and floor brckers ie nn! an appropriate nor desen.,ed response.

A recent maqazine article' states in the sub-title that 'Blue chips now cost more to trade on the Big Board than

small caps

do." lt goes on to state "The higher cost of

trading giant-cap stocks at the NYSE, compared with smaller names, runs counter to the conventional wisdom that large, actively traded stocks have lower transaction costs than smaller, less liquid ones."

The author provides a table, (portions of which are

dangerous: it may lead to mis-identification of the magnitude, source and relevance of transaction costs and thus contribute to inferior oerformance. We will develop three important points: 1. Using pennies per share as the metric ignores the fact that small cap stocks trade at lower prices than large cap stocks. When expressed in economic terms as percentage of principal, this key finding is simply false:

liquid, large cap stocks trade at no higher impact cost than smaller stocks. 2. f he actual costs of implementing

investment ideas is in fact significantly higher than the figures nrrnted hcre Indced thc differences orroted here are Yvvrvv trivial, while true implementation costs are egregious and much larqer than can be seen through the narrow perspective of the Volume Weighted Average Price

reproduced on the following page), that shows "Micro Cap" stocks of less than $250 million capitalization costing 0.5 cents De.:r share on the NYSE whi!a "Giant Cap" stocks with capitalizations in excess of $25 billion cost 1 .9 cents per sha,re The arlicle gets fairly breathless

about how 'iransaction costs on the Big Board are actually higher for big, actively traded stocks than they

(VWAP) comparison.

3. The damning of specialists and floor brokers on the basis of thrs evidence is faulty logic based on faulty measurement.

are for small, lightly traded stocks."

The article attributes this phenomenon in part to pennying; .... NYSE specialists' and floor brokers

Measu ring Economic Costs

putative practice of bidding a stock up or down by one

cent and trading for their own accounts ahead of investors."

The author recognizes that the finding "runs counter to the conventional wisdom that large, actively traded

stocks have lower transaction costs." Indeed, it is counterintuitive; fortunately, the finding is false. And it is

Cents per share sounds like a simple metric, but it only works if the share prices are reasonably similar. For example, a dime cost on a $100 stock is 0.1oh; on a $10 stock, it represents 1.0% -- a big difference.

l

Schack, Justin: "Trading Places;" Institutional Investor Magazine, November 2003. Po B1-85.

The share price for NYSE listed stocks with capitalization less than $250 million is about $7; for stocks in excess of $25 billion the average price is around $35. The table below adjusts VWAP impact differences in stock price. lt shows the expected relationship: large, liquid stocks are traded with less impact than smaller capitalization stocks.

The picture here is one of impact as a percent of traded dollars rising as market cap falls. Showing costs as a

percent

VWAP lmpact

Price/shr

cap

f/shr

($)

Giant cap ($ 2SA+; Large cap ($se - $258) Midcap ($1B - $58)

1A

35

t.o

ZY

t.J

20

Small cap ($2SOw - $ta) Microcap $250M or less

u.o 0.5

14

makers, hedge funds, day traders, speculators, etc. enter

the market opportunistically solely to garner short term trading profits. In the process they provide liquidity to investors.

Institutional traders enter the market to implement investment decisions. They cannot be opportunistic traders; they have an agenda, a blotter of specific trades to complete. Therefore investors pay speculators for the provision of liquidity, leading to frictional costs in the range of five cents per hundred dollars traded (5 bp.)

principal traded shows

the

expected

declines.

VWAP lmpact Measures

VWAP defines impact as the difference between the average price in the trade being executed and the average price in the market during the day. Noie thet tho VWAP impacts in the above table do not net to zero, as they would be if the game were one that pitted institutional buyers against institutional sellers. The investors must be losing to someone else, specifically market makers and speculators. Specialists, market

of

relationship, even though the impact in cents per share

The last column provides a partial explanation of trade size effects: the average trade in the largest cap group is almost 19 times the dollar 66 size in the Microcap group. Other 5.8 things being equal, we expect large 6.5 trades to cost more than small 4.2 trades. Things are not equal, 6.9 however, since large cap stocks are far more liquid and therefore easier to trade. Thus the interplay of trade size and available li^!i/.,1;+.., /.r^+^-q;h^.,r.a +r2^llro .-.nci nl*eTences iliquidity are ignored by the VWAP benchmark, which uses the same benchmark irresoective of trade size. VWAP lmpact (basis points)

The Real Cosfs of Implementing lnvestment Needs and ldeas So far we've identified cents oer share measurement as leading to erroneous conclusions between large cap and small cap stocks. But there are hidden costs of trading beyond the impact. First there's the commission. The following table shows NYSE commission costs by capitalization group. The cents per share commissions are fairly steady. Showing the cost in basis points highlights the higher commission costs of trading lower priced stocks.

A more inclusive alternative to the VWAP measure is called lmplementation Shortfall. lt measures the change in price from the starting gate, where the trader receives the orcier, to ihe finish iine vyhen tne eniife iraoe is completed. lt answers the question: How much of the potential return did I lose to the cost of implementing the trades?

Giant cap ($ 25B+) Large cap ($SA - $258) Midcap ($1B - $5B) Small cap ($250M - $18 Microcao $250M or less

12

15

4.3 +.2

20 28

A4 -.

47

3.4

1

4.0

The costs below show figures for all buy and sell trades

for a sample of 89 investment managers during the second quarter of 2003.

Another hidden cost of trading is the cost of searching for liquidity. Liquidity search costs are real; they eat into the research advantage for stock purchases,

mplementation Shortfall lmpact Measures

lmplShort lmpact:

Price/shr

cap

d/shr

($)

Giant cap ($ 25B+) Large cap ($5B - $25B) Midcap ($18 - $5B) Small cap ($ZSOtttt - $1B) Microcap $250M or less

6.U

35

474

2,286

5.9 5.2

29

1,448 744

4.4 2.6

14

21.4 27.5 31.0 35.4

I

20 7

.

lmplShort lmpact:

(basis points)

Trade Size: ($thou;

119

and they cost real dollars on security sales when orices decline before finding the liquidity to

comolete the trade. They occur because institutional orders

are often too large to be simply presented to the market, and it is difficult to search for liquidity without influencing the price against your interests. When measured from the starting-gate price, the search costs easily dwarf the impact and commission costs, as the table shows. They also show the expected relationship: smaller cap stocks are more exoensive to trade. Search Costs

NYSE Search Cost (bp) Qlshr

cap Giant cap ($ ZSa+; Large cap ($58 - S25B)

42

tv

Pennying.' Does this Data Su

Now we turn to the assertion that the alleged extra cost

the specialists. The estimates we have seen for the cost of pennying are in the $150 million dollar range. The NYSE Fact Book for 2002 quotes a dollar volume of $10.278 trillion We double that figure to account for the fact that

r.^,^ LIIEIU

f

Note that searcl^ costs are nuch higher for the Microcap

stocks. The trading rn these stocks is describeo as by appointment." The term indicates that the brokers and specialists find the other sioe when no interest exists on the flnor A e-n.trv n.o.ess or searehino for the other Y' J Y side leads io h gn searci cosrs, Orce both buyer and

.- - J'e-araf geo quarr ry a'0

aooarent rncact s smai

pport the Accusation?

for large cap trades comes from pennyinE by

4.4 6.2 7.0 8.3 8.7

23 35 58

Midcap (S1B - S5B) Sma lcao rS250M - SlB) Microcap 5250\l or less

VWAP has its uses, but analyzing trade cost data is notoriously tricky. When the conclusions don't match the intuition. one needs to be extra careful.

srze.

.

t^-

^^r'^- IUI UVEIY ^,,^-,, ^ d^ h,,,/^r ^^rU d^ JUIIUI 5 UUYtrI d

Shafe

SOld.

Dividing $'150 million by this number comes to 0.0007 basis points or two hundredths of a cent on an average Q?6 chrro Thiq nrn^llnts tc :a! ta\ltr 2o/" Cf the \1\AAP cost and is not large enough to account for the difference in cost, Grven the current high sensitivity concerning specialist behav or, the offhand attribution to penryrng st'rkes us as careless and inflammatory. In summary, the key to successful trading is to

manage the speed of execution Trade too fast and your demands for liquidity will allow others -rades Ieno to be a much aroer noiron Fira , ," v J' \1 r^roran ' of the Ca '. vo rme. leadirg to a raL-,tology where the trade s -easu'ed aoa nst itself. lf an nstitution were the only bu5,er for the Oay tne VWAP costs show as zero, no matter ,',^at rtre orice act o.r o't"e stoc< drring the day Add inn

I rvqvr

u

I

comnnntr^-c a.q t-tr tn,n\!.r- r:h e nr tntgl 6631Total lmplementation Costs

I

mplementation Shoftfall

to profit from your urgency. Trade too slow and the information edge you are trying to capture may be discovered by othets, wiping out your advantage. Thus traders need to weigh the impact costs of urgently executed orders against the search costs of -ore passive trading techniques. VWAP, properly expressed in economic terms. gauges only the impact costs. T ming is de-emphasized. A trader judged solely

againsi a VWAP standard could avoid impact and transfer the costs into the search cost category. thus fooling his evaluator and Volume Weighted Averaqe Price

Total

voP

(bp)

f/shr

//chr

Giant cap (S 25B+; Large cap rS5B - 525B; Midcap (S1B - 55B) Small cap (S250M - S1B; Microcap 5250M or less

+J

14.8 to.+

o LO

to.o

LJ

117

to. /

201

14.6

0.6 0.5

60 83

possibly himself.

As Mark Twain said, there are three kinds ^{ li^^. u llub,

r;^^ ll9>,

l^-^^.J udllllluu

li^anal DLdlrJUUs. a+a+ia+ino llu>J dllu

More charitably, as was said in the movte State And Main, "lt s not a lie, it's a gift for fiction."

The relationships are as we expect: small cap stocks are more costly to trade when measured in economic terms as percent of principal. In cents per share, costs peak in the Midcap range, but this is an artifact of the share price falling faster than the costs are rising for sequentially smaller capitalization catagories. The VWAP - repeated from the first table - ignores the hidden search costs and understates the real costs of trading. lt also reverses the cost to capitalization relationship.

2 Readers

familiar with the Plexus Group "iceberg" of costs will notice that we exclude opportunity cost in this article,