FTC Staff Comment Before the Commodity Futures Trading Commission Concerning Dual Trading by Floor Brokers

V~UU030
COMMISSION AUTHORIZEO
Comment to the Commodity Futures Tradinl Commission
on Proposed Reaulation 155.5 from
the Staff or the Bureau of Economics
or the Federal Trade Commission!
July 20, 1990
I. 	Introduction and Summary
The staff of the Bureau of Economics of the Federal Trade Commission
appreciate this opportunity to comment on the Commodity Futures Trading
Commission's (CFTC) Proposed Re~ulation 155.5 on dual trading by floor
brokers on the commodity exchanges. 	Regulation 155.5 would ban dual trading
except in limited instances. 	The CFTC plans to implement the ban in phases
according 	to its 	proposed 	Dual 	Trading 	Restriction 	Implementation 	Plan
(DTRIP). By phasing in the ban, the CFTC wiJ] have the opportunity to co]]ect
and analyze information on the effects of a ban before deciding whether to
extend it to a]] exchanges and a]] commodities.
The CFTC has already co]]ected a significant amount of data. 	These
data are described and analyzed in a November 1989 report prepared by the
Division of Economic Analysis of the CFTC entitled, "Economic Analysis of
Dual Trading on Commodity Exchanges" (hereafter "Dual Trading Study"). We
commend the CFTC for gathering and analyzing evidence on the effects of a
dual trading ban before extending it to a]] markets. 	We agree that a thorough
appraisal of the likely effects of a ban should precede a decision to extend it
to a]] markets.
To support the proposed ban on dual trading, the CFTC cites to a finding
from the Dual Trading Study that dual trading is not more prevalent in low
volume markets and in more distant trading months, the types of markets in
which dual trading should provide the greatest benefits. Our analysis of some
of the CFTC's data leads us to suggest that the CFTC reconsider this finding.
In addition, we believe that some of the evidence contained in the Dual Trading
Study suggests that, on some exchanges, dual trading provides demonstrable
benefits. Thus, we suggest that the CFTC amend its proposed DTRIP to a]]ow
more intensive analysis of the exchanges where dual trading may provide
benefits.
Most importantly, we recommend that, to the extent possible, the DTRIP
place the initial round of bans on the exchanges and commodities that have
already been extensively analyzed.a If the commodities and exchanges for the
initial round of bans are selected in this manner, the CFTC would then be able
1 These comments represent the views of the Staff of the Bureau of
Economics of the Federal Trade Commission. 	They are not necessarily the
views of the Commission or any individual Commissioner. Questions concerning
these comments should be addressed to Timothy Daniel, Assistant Director for
Economic Policy Analysis, (202) 326·3520.
255 Fed. Reg. 1047 (January 11, 1990).
a In the Dual Trading Report, the CFTC staff analyzed in detail fourteen
particular futures from ten different exchanges.

to cxaminc in dctail how particular markcts operatc with and without a dual
trading ban. Wc believc that our suggestions might help insure that the CFTC
gathers sufficicnt information during the DTRIP to determine the likcly costs
and bcncfits from cxtending thc ban to all markcts.
Thc rcmainder of our commcnt is organized as follows. First. wc discuss
the intcrest and cxpericncc of the fcderal Trade Commission. 	Sccond. wc
summarizc bricfly thc Proposcd Regulation 155.5. 	Third. wc dcfinc dual
trading and document its prevalence in various exchanges. Fourth. we discuss
the potential costs and benefits associated with dual trading. Fifth. we discuss
some of the evidence contained in the Dual Trading Study that assesses the
prevalence of and potential benefits of dual trading. In particular. we discuss
areas where our interpretation of the results differs from the interpretation
provided in the Dual Trading Study. 	Sixth. based largely on the preceding
analysis. we provide suggestions regarding possible modifications to the DTRIP.
II. 	Interest and Experience of the Staff of the Federal Trade Commission
The Federal Trade Commission is charged with promoting competition
and protecting consumers from unfair and deceptive commercial practices.· In
fulfilling this mandate, the staff of the Federal Trade Commission often submit
comments, upon request. to federal, state. and local governmental bodies to help
assess the competitive and consumer welfare implications of pending policy
issues. Commission staff have gained considerable experience in analyzing the
impact of various private and government restraints on competition and the
costs and benefits to consumers of those restraints. More specifically, the staff
have commented on regulations, similar to proposed regulation 155.5, which
address potential conflict of interest issues.I 	In addition. the staff have
submitted comments on the securities industry.6 and conducted studies of
various aspects of investment markets.7
4 See 15 U.S.C. Sections 41 ~.
6 Sec, 	for 	example. 	comments 	from 	the 	staff 	of 	the 	Federal 	Trade
Commission to the Department of Housing and Urban Development on proposed
revisions to regulations pertaining to the payment of refcrral fces to rcal cstatc
brokers (July 15.1988). and letter from Thomas B. Carter. Director of the FTC's
Dallas Regional Office to Mr. Jack B. Carson, Executive Director of the
Virginia Board of Pharmacy on proposed regulations concerning the sale of
prescription drugs by physicians (November 27, 1989).
6 See. for instance. commcnts of the staff of the Bureau of Economics of
the Federal Trade Commission to the Securities and Exchange Commission on
proposed amendments to rule 12b-l of the Investment Company Act of 1940
(September 6. 1988).
7 Sec. Hilke. Minimum 	Ouality 	Versus 	Disclosure 	Regulations: 	State
Regulation of Interstate Open-end Investment Company and Common Stock
Issues. 	Bureau 	of 	Economics 	Staff 	Report 	(1987); 	and 	Schumann. 	~
Regulation of Takeovers and Shareholder Wealth: The Effects of New vork's
1985 Takeover Statutes. Bureau of Economics Staff Report (1987).
2

III. 	Proposed Reaulatlon 155.5
Proposed regulation 	ISS.S would severely restrict dual trading. 	In
essence, the regulation wouid ban dual trading on futures and options markets
with limited exceptions. 	Given the limited nature of the exceptions, and for
ease of exposition, we will refer to the proposal as one that bans dual trading.
The proposed ban would not take effect immediately in all markets on
all exchanges. 	Instead, the CFTC plans to impose dual trading bans in a few
markets on a few exchanges so that some evidence on the effects of a ban can
be obtained and analyzed before the ban is made more general. CFTC refers to
the limited ban on dual trading as its Dual Trading Restriction Implementation
Plan (DTRIP).
IV. 	Dual Tradina: 	Defined and Measured
In futures and options markets, trades are executed on the floors of the
various exchanges. 8 	Currently, floor 	participants fall 	into one of 	three
categories: 	exclusive personal traders; exclusive brokers; and dual traders.
Exclusive personal traders trade solely for their own accounts. 	Exclusive
brokers, by contrast, specialize in handling customer orders. 	Dual traders fill
both roles, that is, during a particular trading session a dual trader handles
customer orders and trades for his own account.
A dual trader's total trading volume can be divided into two categories:
personal 	trading 	and 	customer 	trading. 	The 	larger 	of these 	volumes 	is
considered his "primary volume" and the smaller his "secondary volume." Non-
dual traders, by definition, have no secondary volume.
The prevalence of dual trading varies across exchanges and across
commodities within a particular exchange. 	The following two tables provide
information on the prevalence of dual trading across various exchanges. Table
I lists the total number of floor participants on various exchanges and the
proportion of these participants who en\aged in dual trading. 	The Table
provides two definitions of dual trading. 	The first definition, labeled 0%
secondary volume, considers a floor participant a dual trader if his secondary
volume is nonzero. 	In other words, under the 0% definition, nondual traders
specialize completely either in providing brokerage services for clients or in
trading on their own accounts. 	The second definition. labeled 10% secondary
volume, considers a floor participant a dual trader only if his secondary volume
exceeds 10% of his total trading volume. More traders qualify as dual traders
under the first definition than the second.
8 The two most prominent exchanges are the Chicago Board of Trade (CBT)
and the Chicago Mercantile Exchange (CME). 	Together, these two exchanges
account for approximately 76 percent of total futures and options trading in the
United States. (See SS Fed. Reg. 1052 (January II, 1990»
e These were the two definitions used throughout the CFTC's November
1989 Report.
3

TABLE I
Number of Floor Participants and the Proportion
Who Qualify as Dual Traders. various exchanges
(data from fifteen trading days. fourth quarter of 1988)
Exchange lO 	Total number 	Percent dual 	Percent dual
of traders 	traders: 0%· 	traders: 10%·
CBT 	3635 	39.6% 	8.2%
CME 	l707 	38.2 	ILl
MCE 	1091 	22.8 	3.7
NYME 	496 	52.8 	29.1
CMX 	250 	63.2 	32.3
CSCE 	208 	69.7 	47.7
NYCE 	171 	54.4 	28.7
KCBT 	91 	52.8 	15.4
MGE 	43 	25.6 	13.9
CRCE 	19 	42.1 	10.5
TOTAL 	7639 	39.1 	12.1
• 	0% (10%) definition means that the secondary
volume of dual traders exceeds 0% (10%)
Looking first at the "0%" column, Table I indicates that 39% of all
traders on these ten exchanges were dual traders to some extent during this
period. 	Note the variation across exchanges: 	the proportion of dual traders
ranges from about 70% on the CSCE to 23% on the MCE. 	The "10%" column
indicates that only 12% of all traders have a secondary volume that exceed 10%
of their total volume. Still, substantial variation exists: dual traders represent
over one fourth of all traders on four of the exchanges under this definition.
Table II provides for each exchange information on the total volume of
trading and the proportion of that volume handled by dual and nondual
traders. 	As in Table I. two definitions of dual trading (the 0% and 10%
standards) arc used.
10 Exchanges: 	CBT: 	Chicago Board of Trade; CME: 	Chicago Mercantile
Exchange; MCE: 	MidAmerica Commodity Exchange; NYME: 	New York
Mercantile Exchange; CMX: 	Commodity Exchange, Inc.; CSCE: 	New York
Coffee, Sugar. and Cocoa Exchange; NYCE: 	New York Cotton Exchange;
KCBT: 	Kansas City Board of Trade; MGE: 	Minneapolis Grain Exchange;
CRCE: Chicago Rice and Cotton Exchange.
4

TABLE II
Average Daily Total Trading Volume, Percent of Total Volume
Accounted for by Dual Traders, and Percent of Total Volume
Accounted for by Secondary Volume
(data from fifteen trading days, fourth quarter of 1988)
Exchange
CBT
CME
MCE
NYME
CMX
CSCE
NYCE
KCBT
MGE
CRCE
TOTAL
Average daily 	Pct. acct. 	Pct. acct. 	Pct acct.
volume 	for by dual 	for by dual 	for by second-
(contracts) 	traders: 0%· 	traders: 10%· 	au vol: 0%·
923,716 	55.5% 	18.8% 	6.8%
487,179 	35.8 	16.3 	5.7
27,969 	36.5 	14.5 	4.7
262,721 	39.7 	26.5 	8.5
99,029 	54.7 	38.5 	13.1
61,339 	59.5 	44.4 	14.6
24,305 	55.9 	35.2 	11.1
11,382 	35.7 	14. I 	4.5
3,128 	27.9 	21.1 	7.3
361 	56.7 	26.2 	9.1
1,901,129 	47.9 	21.2 	7.4
• 0% (10%) definition means that the secondary
volume of dual traders exceeds 0% (10%)
Table II shows clearly the wide variation across exchanges in both the
volume of trading and the proportion of volume handled by dual traders. Note,
in particular, that two exchanges with relatively high proportions of total
volume handled by dual traders are the New York Coffee, Sugar and Cocoa
Exchange (CSCE) and the Commodity Exchange, Inc. (CMX).
Tables I and II indicate that dual traders exist on all exchanges to some
extent, and that these traders handle a significant percentage of total volume.
Comparing the 10% columns of Tables I and II indicates that many dual traders
have a secondary volume below 10%; that is, many traders who qualify as dual
traders under the 0% standard qualify as nondual traders under the 10%
standard. In other words, a significant proportion of dual traders concentrate,
but do not specialize, on either trading for customers or trading on their own
accounts.
The right-most column of Table II provides an alternative way to
consider the volume of dual trading on a particular exchange. 	The figures in
this column represent the proportion of total volume represented by "secondary
volume." 	If all traders split equally their 	trading between personal and
customer 	trading, then 	the 	volume of trading 	represented 	by "secondary
volume" would be 50%. By contrast, if all traders specialized in either customer
or personal trading, secondary volume would be 0%. 	Table II shows that, on
5

average. only 7.4% of all volume is "secondary volume. oo11 Note again. however.
that the proportions vary across exchanges. and that secondary volume is
highest on the CSCE and the CMX.
We believe that the ,Dual Trading Study provides convincing evidence
that exchanges differ markedly in terms of the average volume of daily
contracts traded. the number of traders on the floor. and the proportion of
traders who are dual traders. 	Presumably. the observed differences can be
attributed to some extent to variations in important features. such as customers'
demand for liquidity. the cost to traders of obtaining information. the volume
of contracts traded in a particular session. and the ability of new traders to
enter and exit the exchange. While we do not attempt to explain the variation
in the structures of the various exchanges. we would caution that the observed
variation argues against generalizing the findings from one exchange to others.
or basing policy on the average effects across all exchanges.
V. 	The Costs and Benefits or Dual Tradln&
The costs associated with dual trading arise from a conflict of interest
problem. 	For instance. suppose a dual trader receives an order to sell a block
of futures that is sufficiently large to affect the price of the particular future
involved. 	The dual trader. could personally benefit by selling from his own
account ahead of the client's order. and then repurchasing the future for his
own account after its price had fallen. 	By considering the potential gains on
his own account. the dual trader may fail to secure the best terms for his client.
In addition. behavior of this nature may not only impose significant efficiency
costs by distorting pricing signals. but also may lower investor confidence in
the futures and options markets overall. The Federal Register Notice provides
details on other abuses that either rely on dual trading or are facilitated by it.12
As to potential benefits, it has been argued that dual trading improves
the operation of futures and options markets in two ways. 	First. the gains
available from being able to trade both for clients and for one's own account
may increase the supply of traders on the floors of the exchanges. 	This
increase in supply of traders could make the markets more liquid (i.e., make it
easier for a buyer to find a seller), thereby lowering the costs of trading to all
participants. 	Second. it has been argued that dual traders, because they have
their own money at stake, tend to be the 'most skillful traders. which then
lowers the costs of executing trades for their clients. The next section discusses
evidence collected and analyzed by the CFTC which attempts to measure the
benefits of dual trading.
11 The CFTC's Federal Register Notice highlights this finding. Stt 55 Fed.
Reg., p. 1050.
U The possibilities for abuse are not hypothetical. In 1989 the FBI alleged
that several traders on the Chicago Mercantile Exchange and the Chicago Board
of Trade engaged in abusive trading practices. Pursuant to that investigation,
the Department of Justice has obtained several plea agreements which describe
the methods by which these traders received compensation for their trading
abuses. ~ 55 Fed. Reg .• pp. 1049-52.
6

VI. 	AD Appraisal of Some of the FIDdID&s cited ID the Dual TradlD& Study
In an effort to analyze the determinants of dual trading and to measure
dual trading's benefits (which are the same as the costs of banning it), the
CFTC gathered extensive data from all but one of the active commodity
exchanges. 1S Specifically, the CFTC randomly selected IS trading days from
the fourth quarter of 1988 and gathered information on every trade made on
those dates. 	Information on approximately 4,200,000 futures transactions and
400,000 options transactions was gathered and analyzed. 	The CFTC's data
collection and subsequent analysis is impressive, and the findings from this
extensive study were reported in the Dual Trading Study. 	The Dual Trading
Study represents the first systematic attempt to assess the economic benefits
associated with dual trading.
There are a few instances, however, where we would interpret the
results contained in the Dual Trading Study somewhat differently than did the
CFTC Division of Economics staff. We focus on two such differences: whether
dual trading is more prevalent in low volume markets and in distant trading
months, 	and 	whether 	dual 	trading 	contributes 	measurably 	to a 	market's
liquidity.
a. 	Prevalence of Dual Trading in Low Volume Markets and More Distant
Trading Months
One of the findings cited in the Dual Trading Study is that "The
incidence of dual trading is not higher in low volume markets or more distant
trading months. d4 	This finding is important because it provides some support
for the CFTC's proposal to ban dual trading in all markets. 	The analysis
presented in the Dual Trading Study, while consistent with the finding that the
incidence of dual trading is not higher in low volume markets or more distant
months, relies on the comparison of simple averages. In this section, we present
some results using multivariate statistical techniques which we believe provide
some potentially useful insights on whether dual traaing is more prevalent in
low volume markets and distant trading months. 	Our analysis leads us to
suggest that the CFTC consider further the factors that influence the degree to
which dual trading exists in certain markets.
With·respect to low volume markets, the data presented in the CFTC
Report indicate that, on average, the proportion of trades handled by dual
traders tends to decline as the size of the market declines. 16 It should be noted,
however, that the data indicate significant variation across exchanges, and that
in two exchanges (the Chicago Rice &. Cotton Exchange and the New York
Coffee Sugar &. Cocoa Exchange) the proportion of trades handled by dual
traders is actually highest in the lowest volume markets (fewer than 60
contracts per day.) 	We suggest that the CFTC consider whether the variation
across exchanges in the proportion of dual trading in low volume markets
argues in favor of analyzing these markets in more detail before banning dual
trading in all markets.
IS One small exchange was omitted because the exchange could not provide
the necessary data in machine readable form.
14 Dual Trading Study, p. 3.
16 Dual Trading Study, Tables 1.10 and 1.11.
7

With respect to distant trading months, the data in the CFTC Report
indicate no apparent relationship between the proportion of trades handled by
dual traders and the time period separating the contract's trading and maturity
dates. 16 	-
Our analysis uses data from the Dual Trading Study to examine how
various factors influence the extent to which dual trading exists in a particular
market. 17 	The 	CFTC dataset 	contains detailed 	information 	on 	fourteen
particular commodity markets for fifteen randomly selected trading days from
the fourth quarter of 1988. 18 In creating this dataset, the CFTC first divided
each of the fifteen trading days into half-hour trading brackets. 	As a result,
each observation in the dataset contains information on, among other things,
the volume of trading handled by dual and nondual traders during a particular
half-hour trading period, and the prices at which contracts traded. Our interest
here is to discover whether the use of a different approach supports the Dual
Trading Study's finding that dual trading was not more prevalent in lower
volume trading markets and in more distant trading months. For our analysis,
we estimated the following multivariate regression equation:
where,
PCTDUAL, 	•
•
•
BASPREAD, •
•
•
percentage of the total volume of trading handled by dual
traders during bracket t;
total number of contracts traded during bracket t;
time period (in months) betwccn the trading date and thc
maturity date for contracts traded in bracket t;
estimatc of the bid/ask spread in bracket t, calculated as
the averagc of the absolute value of each price change
that occurred in thc half-hour bracket divided by the last
price observed in the bracket;
estimate of thc price risk in bracket t, measured by thc
difference betwcen the maximum and the minimum prices
during thc bracket divided by the average price observed
during the bracket;
random error term; and
16 Dual Trading Study, Table 1.13.
17 The CFTC's Division of Economic Analysis kindly provided these data
to us.
18 The commodity markets are listed in Table 3.5 of the Dual Trading Study.
These data underlie the multivariate regression analysis discussed on pages 41-
47 and in Appendix A of the Uual Trading Study.
8

a o and bl-b. arc coefficients to be estimated.
This specification presumes that the prevalence of dual trading on a
particular market depends on the potential rewards to engaging in the practice.
Factors that should influence a trader's decision to dual trade include the
market's riskiness and liquidity. 	The variable RISK is included as a measure
of the market's riskiness, and the variables TOTVOL and BASPREAD arc
included as measures of the market's liquidity.ll~ The variable LENGTH is
included to test whether, after controlling for these other factors, dual traders
are more or less prevalent in more distant trading months.
If, contrary to the finding cited in the Dual Trading Study, dual trading
pro"ided important benefits in low volume markets and in distant trading
months, the estimated coefficient on TOT VOL would be negative and the
estimated coefficient on LENGTH would be positive. The coefficient on RISK
would be positive if relatively risky markets (i.e., those with greater price
fluctuations) 	provide 	greater 	opportunities 	for 	dual 	traders 	to 	use 	their
personal accounts to provide temporary liquidity in hopes of earning income.
Finally, if the rewards to dual trading decrease as the market becomes more
liquid, we would expect the coefficient on BASPREAD to be negative.
These data raise two econometric issues. First, each observation is based
on 	trading 	that 	occurred 	in 	a 	particular 	thirty 	minute 	trading 	bracket.
Consequently, the variables included in the analysis are averages derived from
the individual trades that occurred during the bracket. 	As noted earlier,
trading volume in a particular bracket varies widely both across exchanges and
across markets on a given exchange. 	Because our variables are based on
averages, they will be measured more reliably in higher volume brackets.'o The
procedure used to estimate the parameters of interest should account for these
diff erences.
Second, the dependent variable (PCTDUAL) cannot be less than zero or
greater than one hundred. 	In general, the estimation procedure should try to
account for this feature of the dependent variable.21
Due 	to 	these econometric issues, 	we estimated 	the 	equation 	three
different ways." 	First, we used the standard ordinary least squares (OLS)
19 The difference between the bid and ask prices (BASPREAD) is ~ommonly
used as an indicator of liquidity. 	Greater liquidity leads to lower bid-ask
spreads, which makes it easier for a buyer to locate a seller and vice versa.
20 In statistical terms, the variance of an average will decline as the number
of observations used to compute the average increases.
21 In econometric terms, the dependent variable is censored both from below
and from above.
22 Our specification is admittedly simplistic in that it assumes that the
independent variables cause changes in the dependent variable, PCTDUAL. As
discussed in the Dual Trading Study; causation could also run the other way:
the 	amount 	of dual 	trading could 	influence a 	market's 	liquidity. 	This
simultaneity suggests that econometric techniques more elaborate than those
used here or in the Dual Trading Study may be required to estimate the
(continued ... )
9

approach. 	This 	approach 	makes 	no 	adjustment 	for 	either 	of 	the 	two
econometric issues discussed above. Second, we used a weighted ordinary least
squares (OLSWT) approach, which accounts for the fact that the data are
increasingly reliable as the number of contracts traded in the bracket increases.
Third, we used the TOBIT estimation procedure, which accounts for the fact
that the dependent variable is bounded by zero from below and one hundred
from above.23
The following table provides the estimated coefficients from these three
estimation procedures. 	For each procedure, the equation was estimated first
using the 0% definition of dual trading, and then using the 10% definition. 24
22(... continued)
relationships between dual trading and other variables of interest. Our purpose
in presenting these results is merely to point out that the Dual Trading Study's
finding regarding the prevalence of dual trading in lower volume markets and
more distant trading months may not hold up under relatively simple statistical
scrutiny.
sa Unfortunately, when we tried to control for both econometric issues
simultaneously (by using a weighted Tobit) the estimation procedure failed to
. converge.
24 Recall that the 0% (10%) definition defines a trader as a dual trader if his
secondary volume exceeds 0% (10%). Table III indicates that the proportion of
trading volume handled by dual traders falls from approximately 63% to
approximately 31% when the definition shifts from 0% to 10%. 	For our
purposes, both definitions are somewhat flawed. The 0% definition defines too
many traders as dual traders because a trader is considered to be. dual trader
if he makes a single trade on his own account to correct an error he made while
handling a client's order. While the 10% definition corrects this by considering
such a trader to be a nondual trader, the 10% threshold may cause errors in the
other direction: traders with close to but less than 10% secondary volume may
be incorrectly classified as nondua) traders.
10

TABLE III
Estimated regression coefficients
(absolute value of t-statistics in parentheses)
Dependent variable: 	PCTDUAL
Number of observations: 	5266
0% Definition
~ 	OLSWT 	TOBIT
TOTVOL 	.0002· 	.0002·· 	.0002
(2.28) 	(18.32) 	(0.59)
LENGTH 	1.54·· 	2.13·· 	2.04··
(11.89) 	(12.55) 	(13.90)
BASPREAD 	-5.63·· 	33.72·· 	-5.53··
(2.92) 	(8.47) 	(3.27)
RISK 	7.07·· 	2.87·· 	6.81··
(10.88) 	(5.93) 	(8.65)
ADJ R' 	0.0445 	0.1264 	nla
Mean of 	62.76 	62.76 	62.76
PCTDUAL
10% Definition
Q.L£.. 	OLSWT 	TOBIT
TOTVOL 	-.0002· 	.00003·· 	-.0001
(2.51) 	(2.82) 	(0.23)
LENGTH 	0.58·· 	0.95·· 	0.33·
(4.84) 	(7.45) 	(2.46)
BASPREAD 	10.14·· 	54.79·· 	11.03··
(5.63) 	(18.41) 	(8.08)
RISK 	6.29·· 	1.62·· 	7.54··
(10.38) 	(4.48) 	(11.25)
ADJ R' 	0.0494 	0.1716 	nla
Mean of 	30.95 	30.95 	30.95
PCTDUAL
•• significant at 99% level
• 	lignificant at 95% level
11

These 	results 	with 	respect 	to 	TOTVOL 	and 	LENGTH 	are 	both
interesting and 	puzzling. 	First, note that the coefficient on TOTVOL is
positive and significant using the weighted least squares (OLSWTl approach but
insignificantly different from zero using the TOBIT procedure. 6 The results
from the OLSWT analysis 'support the Dual Trading Study's finding in that
they indicate that dual trading is actually more prevalent in higher volume
markets. 	Still, the failure of this result to carryover to the TOBIT analyses
suggests that the relationship between the prevalence of dual trading and the
volume of trading in the market may be more complicated than suggested in the
Dual Trading Study.
Second, 	the 	coefficient 	on 	LENGTH 	is 	consistently 	positive 	and
significant, suggesting (contrary to the finding cited in the Dual Tradirw
Study) that dual trading is more prevalent in more distant trading months.
These results lead us to suggest that the CFTC reconsider its conclusion that
dual trading is not more prevalent in more distant trading months.27
The results from these regressions do not provide a clear picture of the
determinants of dual trading.28 The sensitivity of 'the results depending on the
specification used leads us to suggest that the CFTC consider using the DTRIP
to examine these issues further. 	Perhaps the most straightforward way to do
this would be to compare a market's performance during a period when dual
trading is permitted with its performance during a period when dual trading is
prohibited. 	If the CFTC decides to proceed with the DTRIP and ban dual
trading in selected markets, we suggest that the CFTC consider the possible
benefits and costs of banning dual trading in some of the markets in which it
has already collected and analyzed extensive data, such as those that underlie
the Dual Trading Study's regression analysis.
b. 	Measuring the Effect of Dual Trading on Liquidity
The econometric section of the Dual Trading Study presents a model
capable of estimating the importance of dual trading to a market's liquidity.
26 Looking at the results from the TOBIT procedure, the coefficient on
TOTVOL is positive under the 0% definition and negative under the 10%
definition. 	However, the 	relatively low 	t-statistics indicate 	that neither
coefficient is significantly different from zero.
26 The coefficients on LENGTH range in value from 0.33 to 2.13. 	This
suggests that, holding other factors constant, increasing the contract length by
three months would be associated with an increase in the proportion of trades
handled by dual traders of between 1.0% and 6.4%.
27 In addition, in separate analyses we added dummy variables to account
for the three types of markets included in the data (agricultural, financial, and
international/metals). 	Inclusion of these dummies did not alter the results
reported in the text. 	Interestingly, the coefficients on the dummy variables
suggest 	that 	dual 	trading 	was 	significantly 	more 	prevalent 	in
international/metals markets and significantly less prevalent on financial
markets. 	These analyses lead us to suggest that the CFTC use the DTRIP to
conduct detailed analyses of the effects of dual trading in particular markets.
J8 Note, for instance, that the coefficient on BASPREAD is positive and
significant under most specifications but negative and significant under lome.
12

i'
Specifically, this model generates an estimate of a parameter that captures the
marginal increase in liquidity provided by the volume of trade handled by dual
traders. In this model, liquidity is measured by the difference between the bid
and ask prices -- the so-called "spread". When this model is estimated using the
data from all fourteen commodities, the parameter estimate suggests that the
marginal contribution to liquidity from the trading volume of dual traders is
negligible. 	After citing this finding, the Dual Trading Study concludes:29
A literal interpretation of (this result) implies that,
at a given level of nondual trading, dual trading
has no effect on liquidity. 	As will be shown in
subsequent analyses, this literal interpretation is
overly 	aggressive. 	It 	is 	more 	reasonable 	to
conclude that •.. nondual trading makes a greater
contribution to liquidity than dual trading.
Despite the qualification, the text's overall impression is that dual
traders provide little if any contribution to a market's liquidity. 	One of the
"subsequent analyses", however, suggests the opposite. When the same model is
estimated using only the data from commodities traded on the metals and
international commodities exchanges, the results suggest that dual traders are
far more important than nondual traders to the exchange's liquidity.so 	While
the Dual Trading Study notes that, in this case, "the relative influence of dual
trading is more important than for nondual trading,"Sl it nonetheless concludes
that dual traders generally provide nothing unique to futures markets.
We think the conclusion 	may have been drawn 	too broadly. 	The
econometric results clearly suggest that dual traders provide a significant
amount of liquidity to the metals and international agricultural markets. 	In
addition, Tables I and II indicate that these are the exchanges (the CSCE and
the CMX) where dual trading is most prevalent. 	Together, these findings
suggest 	to 	us 	that dual 	traders may 	provide 	important services to 	these
exchanges and that banning dual trading could impose some efficiency costs.
Whether these costs would be outweighed by other benefits, of course, is the
ultimate question. 	Given that extending a ban on dual trading to these
exchanges might entail significant costs, we recommend that the CFTC analyze
these exchanges in more detail than is currently contemplated by the proposed
DTRIP.
We have one further suggestion regarding the design of the DTRIP. The
Dual Trading Study indicates that the CFTC has analyzed extensively data
from 	the 	fourth 	quarter 	of 	1988 	from 	fourteen 	futures 	markets 	on 	ten
exchanges. If these same markets were chosen for the initial round of bans on
dual trading, the CFTC would have the opportunity to examine how these
markets operate with and without a dual trading ban. 	Given the variety in
market characteristics, it becomes all the more important to analyze particular
markets in detail and not extrapolate the results found for certain markets (say,
29 Dual Trading Study, p. A-S.
ao In this analysis, the metals and international commodity exchanges are
the New York Coffee, Sugar and Cocoa Exchange (CSCE), the Commodity
Exchange,lnc. (CMX), and the New York Cotton Exchange (NYCE).
31 Dual Trading Study, p. A-II.
13

the financial markets or proximate trading months) to others (say, the metals
markets or distant trading months) where similar results may not apply. If such
an approach is possible, we believe it would provide information helpful in
assessing the costs and benefits of dual trading in various settings.
VII. 	Conclusion
We suggest that the CFTC revise its DTRIP in three ways. 	First. we
suggest that low volume markets and distant trading months be analyzed in
detail before extending a dual trading ban to these types of markets. Second.
we suggest that more attention be given 	to the metals and 	international
commodity exchanges because the Dual Trading Study suggests that dual
trading may be important to the efficient operation of these exchanges. Third.
we suggest that the DTRIP focus the initial round of dual trading bans on
markets where substantial amounts of information have already been collected
and analyzed. 	Such a design would allow the CFTC to examine efficiently
whether dual trading provided demonstrable benefits in excess of costs in
particular markets. 	If net benefits are detected. the CFTC may wish to
reconsider its proposal to extend the dual trading ban to all markets.
We hope our comments are helpful. 	In closing. we would like to
commend the CFTC for its decision to collect information on the likely effects
of a dual trading ban in selected markets prior to deciding whether to extend
the ban to all markets.
14