The impact of trading party on the execution spread: evidence from futures markets.

Introduction

The bid-ask spread is an important component of the overall transaction costs. Several studies have found that the size of the spread is a function of several factors. For example, Klock and McCormick (1999) find that competition among NASDAQ market makers affects the

In this study, we examine the execution spread incurred by off-exchange customers (hereafter customers) when they initiate a trade to buy or sell futures contracts. Using the transaction records of four futures contracts traded on the Chicago Mercantile Exchange (CME), we measure the customer execution spread (hereafter execution spread) relative to the principal behind the trade counter party. Specifically, we measure the execution spread when customers trade against floor traders and when they trade against other customers and look into the differences of these measures and their determinants.

We find that on average customers pay a larger premium when they trade against other customers rather than against floor traders. This finding might reflect the location advantage of futures floor traders who stand ready in the futures pits and are able to observe important market activities, such as order flows. We also find that trading volume, market competition, and price volatility affect the execution spread. Competition among floor traders, however, does not influence the size of the execution spread when customers trade among themselves.

Literature Review

The behavior and the determinants of the bid-ask spread in equity markets have been researched extensively during the last two decades. For example, Glosten and Milgrom (1985), and Chung, Van Ness, and Van Ness (1999), examine the spread behavior of New York Stock Exchange (NYSE) stocks; Stoll (1989) and Kandel and Marx (1997), study the determinants of spread for NASDAQ. The focus also has been extended to foreign equity markets; for example, Menyah and Paudyal (2000) find a marginal level of inventory holding costs for stocks traded on the London Stock Exchange (LSE).

One of the characteristics of the futures markets is that there are no bid and ask prices available to the public. This creates a challenge for researchers examining and analyzing the spread behavior in futures markets. Several studies rely on bid-ask spread estimators, e.g., Roll (1984), Smith and Whaley (1994), and Bhattacharya (1983), to examine the spread. Wang, Michalski, Jordan, and Moriarty (1994) examine the impact of competition among market makers, price volatility, and trade size on the realized spread of the S&P 500 index futures around the October 19, 1987 crash. They find that volatility increases the realized bid-ask spread around the crash.

On the other hand, the number of market makers reduces the spread. Wang, Yau, and Baptiste (1997) examine the impact of trading volume on transaction costs in futures markets. Using daily observations covering the period between January 1990 and April 1994, they uncover two main findings: first, the bid-ask spread and trading volume are jointly determined, and, second, there is a positive relationship between trading volume and price volatility and an inverse relationship between trading volume and the bid-ask spread. Laux and Senchack (1992) examine the bid-ask spread for four financial futures contracts: Deutschemark, Swiss franc, Treasury bill, and Eurodollar loan deposits between 1982 and 1986. They find a positive relationship between the spread and volatility and a negative relationship between the spread and trading volume. The study also finds that the spread follows a U-shaped pattern over the life of a contract.

With the availability of high-frequency data, researchers also focus on the intraday behavior of the bid-ask spread. Ding (1999) examines the factors determining the intraday and daily bid-ask spreads in the foreign exchange futures market. The dataset includes the Deutschemark and Japanese yen futures traded on the CME during 1990. The results show that bid-ask spread is negatively related to the number of transactions, but positively related to price volatility. Ma, Peterson, and Sears (1992) examine the intraday bid-ask spread for Treasury bond, silver, soy beans, and corn futures during 1980 and 1986. Their findings show the U-shaped pattern in the bid-ask spread during the beginning and the end of the trading day. They suggest that the larger bid-ask spread at the beginning and end of the trading day is inconsistent with higher trading volume because higher trading volume should reduce the spread rather than increase it. Ferguson and Mann (2001) examine the execution spread in futures markets. The data set includes 14 futures contracts traded during the first six months of 1992 on the CME. They find that the customer execution spread in general is small and below the tick size regardless of the counter party. The study also identifies the U-shaped spread pattern after controlling for the announcement effects.

Data and Methodology Data

We use the dataset of Locke and Mann (in press). The dataset contains transaction records of four commodities traded on the CME in 1995: Deutschemark, Swiss franc, live cattle, and pork bellies. Each transaction is recorded closest to the minute, and a record contains commodity type, date, delivery month, buy-sell indicator, quantity, and price. In addition, the dataset also contains an indicator of the principal behind the transaction. The indicator is assigned by the Commodity Futures Trading Commission (CFTC). A transaction may be for the floor traders' personal account (CTI 1), in which they realize their own profits (losses), or for one of the three types of brokerage transactions, in which the floor traders receive a fixed amount of commissions per trade. A brokerage transaction could be for the floor trader's house account (CTI 2), for another floor trader's account (CTI 3), or for customer account (CTI 4). In this study, we focus on the floor traders' personal account trades (CTI 1) and customers' transactions (CTI 4). The distribution of trading across these trade categories is presented in Table 1. For all four markets, there are over 3 million transactions. The percentage of these transactions involving customers on both sides of the trade ranges from 4.11 percent for Deutschemark to 12.96 percent for live cattle. The transactions between customers and floor traders represent a larger proportion of the total transactions, ranging from 23.48 percent in Deutschemark to 30.37 percent in pork bellies. The customer-floor trader transactions account for 60.34 percent of overall trading volume (number of contracts) for pork bellies and for 39.81 percent for Deutschemark. The average prices per contract for pork bellies, live cattle, Deutschemark, and Swiss franc are $18,919, $26,404, $87,546, and $106,197, respectively. The last row of Table 1 presents the number of futures floor traders who trade for their personal account at least one time during 1995. For the four commodities, the number of floor traders ranges from 101 for pork bellies to as many as 281 traders for Deutschemark.

Methodology

Customers lacking the privilege to make trades on the floor have to communicate their transactions to floor traders who execute them. A floor trader charges a fixed brokerage commission for the service provided. The counter party could be another floor trader picking up the trade for his personal account or it could be another customer having a different opinion about the market movement.

In our analysis, for every trade we identify an initiating party who demands liquidity to trade and thus pays a premium in return. Our goal is to examine whether there are any differences in terms of the premium paid by liquidity demanders relative to the principal behind the counter party. We define the execution spread as the difference between the customer-initiated buy and customer-initiated sell prices. Intuitively, the execution spread is the premium paid for an initiated round-trip transaction. We use the execution spread rather than other spread estimators such as Roll's estimator because Locke and Venkatesh (1997) demonstrate that the bid-ask spread estimators do not accurately represent the transaction costs in futures markets. (1)

We use the quote rule to determine whether a buyer or a seller initiates a transaction. The quote rule has been used for this purpose in equity and derivatives markets. Recently Savickas and Wilson (2003) show that the quote rule correctly predicts the initiating party of the transaction 83 percent of the time in options markets. The quote rule defines a trade as a buyer- (seller-) initiated trade if the transaction price is above (below) the mid-spread. (2) In other words, the initiating buyer (seller) would pay a premium by buying (selling) at a price higher (lower) than the average pit price. We then calculate the execution spread for transactions initiated by customers when they trade against floor traders, denoted by SPD41, and also when they trade against other customers, denoted by SPD44. We ignore the transactions involving house accounts (CTI2) and another floor trader's account (CTI3) because these transactions do not contribute substantially to trading volume.

We also examine the relationship between the execution spread and market wide factors. The following regression is estimated:

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.] (1)

where

SPREA[D.sub.t] = SPD41 for transactions initiated by customers
                 against floor traders;
SPREA[D.sub.t] = SPD44 for transactions initiated by customers
                 against other customers;
   VO[L.sub.t] = Customer volume;
    F[T.sub.t] = Number of floor traders;
   SI[G.sub.t] = Standard deviation of futures prices;
            D1 = Dummy variable equal to 1 for the first 30 minutes
                 of trading and 0 otherwise;
            D2 = Dummy variable equal to 1 for the last 30 minutes
                 of trading and 0 otherwise; and
             t = Time subscript of a five-minute interval.

We analyze the determinants of the execution spread as a function of customer volume, competition among floor traders, price volatility, and time of the trade. We measure the customer volume as the number of contracts traded by customers during a five-minute interval. The sign of customer volume coefficient relative to the execution spread depends on the purpose of the transaction. For customer-initiated trades against floor traders, we expect the sign to be positive to reflect the larger adverse selection premium required by floor traders. On the other hand, when customers trade with each other, the volume may reflect the market liquidity and may have a negative impact on the execution spread.

We also use the number of active floor traders to proxy for market competition in the trading pits. We expect that the competition would reduce the execution spread when a floor trader is on one side of the transaction. For transactions between customers, however, the number of floor traders should have no effect on the execution spread because floor traders are acting as agents rather than as principals.

We compute the standard deviation of contract prices during five-minute intervals to measure market volatility. We anticipate a positive relationship between both measures of execution spread and the volatility because higher volatility implies a potentially greater adverse selection problem.

The regression also includes dummy variables for the first and last 30 minutes of the trading day to control for the time of trade effect which suggests that the execution spread is higher during the opening and closing periods.

Results Execution Spread Decomposition

Table 2 reports the execution spread behavior by commodity and initiation criteria. For pork bellies, the execution spread is in the range of $44.03 and $61.31. For live cattle, the execution spread is consistently smaller than the pork bellies execution spread, ranging between $20.22 and $29.66. The execution spread is in the range of $41.24 and $57.48 and $31.78 and $47.60 for the Swiss franc and for the Deutschemark, respectively. Columns 2, 3, and 4 of Table 2 present the intra-day pattern of the execution spread. For agricultural markets, the execution spread exhibits a U-shaped pattern of all measures of trade initiation. We do not observe the U-shaped pattern with the currency markets. Figure 1 shows the intra-day pattern of the execution spread across contracts. The graphs clearly indicate a U-shaped pattern for pork bellies and live cattle futures, but not for the currencies.

Table 3 shows the results of differences in means tests of execution spread with different trade initiation measures across futures contracts. We find negative and significant differences in means between SPD41 and SPD44 in all cases, except live cattle. The results yield an interesting observation that there is an order of execution costs in the market and that the counter party is an important factor for the magnitude of the execution cost. If customers initiate a trade, the execution spread is larger when they trade against another customer than when they trade against a floor trader. An explanation of the small execution spread for floor traders is that they have a competitive advantage over the customers in that they are present on the trading floor and are able to observe the market activities. Thus, when they are making market and responding to a customer-initiated trade, they require a smaller premium to protect themselves from trading against an informed trader. Customers, on the other hand, require a larger premium when they trade against another customer than when they trade against a floor trader because customers do not have the floor information that is available to floor traders.

Regression Analysis

In this section we examine, in a multivariate context, the relationships between the execution spread and market factors, i.e. customer volume, competition among floor traders, market volatility, and time of trade. Table 4 presents the characteristics of the variables examined in the regression analysis. For the agricultural markets, trading volume and the number of floor traders in the pits are higher for live cattle than for pork bellies. For the currency markets, the Deutschemark has higher trading volume as well as the number of floor traders than does the Swiss franc.

We estimate equation (1) for each type of trade initiations of the execution spread. The regression results are presented in Table 5. Adjusted R-squares range from 0.62 to 0.83. For the execution spread measure SPD41, the execution cost for customer-initiated trades against floor traders, we find an inconclusive result with the customer volume. Only agricultural futures contracts have significant coefficient estimates of customer volume (where one is positive and the other is negative). Although we find weak results with the customer volume, we observe that market competition, measured by the number of active floor traders, plays an important role in determining the magnitude of the execution spread. The coefficient estimates of the number of floor traders are significantly negative for most cases. This implies that floor traders compete for order flow and that reduces the execution cost for customers. This result is consistent with Wahal (1997), and Klock and McCormick (1999), who find that the number of market makers has a negative and significant impact on spreads for Nasdaq stocks.

The regression results for the other measure of the execution spread, SPD44 where customers initiate trades against other customers, yield an interesting finding. We observe that the coefficient estimates of the number of floor traders are statistically insignificant for all cases. This might imply that when customers trade among themselves with floor traders acting only as brokers, the competition among floor traders does not have an impact on the size of the execution spread.

For the two execution spread measures, we find that the coefficient estimates for volatility are positive and statistically significant in all cases. This result is consistent with Wang and Yau (2000), who find a positive relationship between spread and volatility in financial and metal futures markets.

We observe several interesting characteristics of transaction costs for customers. When customer initiate transactions with floor traders, the execution spread decreases with increasing number of traders in the pits which suggests that the competition among floor traders reduces the cost of trading for customers. The impact of customer volume on the execution spread is mixed. The results regarding the execution spread for transactions between customers generally are not affected by the customer volume and the competition among floor traders. This finding is consistent with the expectations because floor traders do not take sides in these transactions.

Conclusions

A futures pit is an interesting trading environment where participants trade simultaneously with multiple prices and quantities. In this study, we compute execution spread for customers who initiate trades relative to the counter party. We find that customers are better off when they initiate transactions with floor traders. These results are consistent with the notion that floor traders have an information advantage regarding the order flow. We also examine the determinants of the execution spread and find that trading volume reduces the execution spread, while volatility increases it. The impact of number of traders in the pit varies depending on the execution spread measures. The number of floor traders reduces the spread for transactions between customers and floor traders. It does not affect the execution spread when customers trade with each other.

Table 1--Descriptive Statistics

The table is based on the distribution of trading across trade
categories of Type 1 and Type 4. Type 1 and Type 4 refer to floor
traders' personal account trades and customers' transactions,
respectively. For all four markets, there are over 3 million
transactions. The last row of the table presents the number of
futures floor traders who trade for their personal account at
least one time during 1995

                                               Futures Contracts

                                           Pork Bellies   Live Cattle

% of Transactions between Type 1
  and Type 4                                  30.37%        25.87%
% of Transactions between Type 4
  and Type 4                                  11.38%        12.96%
Total Number of Transactions                 272,768        900,724
% of Total Volume of Type 1 Transactions      60.34%        58.42%
% of Total Volume of Type 4 Transactions      24.93%        24.78%
Average Price Level                           18,919        26,404
Number of Floor Traders                        101            219

                                               Futures Contracts

                                           Swiss Franc   Deutschemark

% of Transactions between Type 1
  and Type 4                                  25.95%         23.48%
% of Transactions between Type 4
  and Type 4                                   6.13%         4.11%
Total Number of Transactions                 972,421      1,124,080
% of Total Volume of Type 1 Transactions     48.12%         39.81%
% of Total Volume of Type 4 Transactions     15.21%         14.94%
Average Price Level                          106,197        87,546
Number of Floor Traders                        224           281

Table 2--Execution Spread by Initiation Criteria and
Futures Contracts

The execution spread is calculated as the cost of a transaction
for the initiating party using each five-minute interval of
trading. In this setting, the execution spread is the difference
between the initiated buy and the initiated sell prices. The
quote rule is used to identify the trade initiator. Quote rule
defines a trade as a buyer- (seller-)initiated trade if the
transaction price is above (below) the mid-spread

                                   Intra-day Spread

               Initiation   Mean    Open    Mid-day   Close
Contracts       Criteria

Pork Bellies   SPD41 (a)    47.54   61.31    44.03    53.71
               SPD44 (b)    48.49   61.22    45.64    51.93
Live Cattle    SPD41 (a)    22.57   29.66    21.01    24.58
               SPD44 (b)    21.68   27.97    20.22    23.87
Swiss Franc    SPD41 (a)    54.88   57.09    55.62    44.26
               SPD44 (b)    56.08   57.48    57.27    41.24
Deutschemark   SPD41 (a)    40.47   40.64    41.22    31.78
               SPD44 (b)    46.58   44.03    47.60    37.63

(a) SPD41 refers to the execution spread for transactions
between customers and dealers

(b) SPD44 refers to the execution spread for transactions
between customers

Table 3--Differences in Mean Execution Spreads

The table reports the differences in means tests of execution
spread with different trade initiation measuresacross futures
contracts

Commodity       Difference     Mean Difference   t-statistics

Pork Bellies   SPD41 - SPD44        -2.93          -3.33 *
Live Cattle    SPD41 - SPD44         0.78           2.68 *
Swiss Franc    SPD41 - SPD44        -1.17          -1.94 **
Deutschemark   SPD41 - SPD44        -5.75          -9.52 *

SPD41 refers to the execution spread for transactions between
customers and dealers

SPD44 refers to the execution spread for transactions between
customers

* Significant at the 5 percent level

** Significant at the 10 percent level

Table 4--Descriptive Statistics of Regression Variables

VOL refers to the customer volume. We measure the customer
volume as the number of contracts traded by customers during
a five-minute interval. FT is the number of floor traders in
the pits during a five-minute interval. SIG is a measure of
volatility. We compute the standard deviation of price during
five-minute intervals

Agricultural

                             Pork Bellies

                Mean    Min.    Median    Max.      STD

SPD41 (z)       49.20   10.00    40.00   1000.00    39.32
SPD44 (b)       52.13   10.00    40.00    585.00    39.73
  VOL           41.68    1.00    18.00   1178.00    68.92
  FT             9.68    1.00     9.00     29.00     4.81
  SIG           21.07    0.00    17.89    441.72    18.30

Currencies

                              Swiss Franc

SPD41 (a)       55.40   12.50    43.75   1664.41    49.73
SPD44 (b)       56.57   12.50    43.75    987.50    50.55
  VOL          173.34    1.00   110.00   2268.00   199.52
  FT            27.22    2.00    27.00     64.00    11.76
  SIG           29.14    0.00    23.16    537.70    24.53

Agricultural

                             Live Cattle

                Mean    Min.    Median    Max.      STD

SPD41 (z)       22.88   10.00    19.96    226.00    14.33
SPD44 (b)       22.10   10.00    19.00    313.73    14.68
  VOL          191.57    1.00   111.00   3430.00   241.41
  FT            18.99    2.00    18.00     49.00     7.96
  SIG           11.55    0.00     9.77    188.24     7.79

Currencies

                             Deutschemark

SPD41 (a)       40.72   12.50    30.40   1978.55    44.07
SPD44 (b)       46.47   12.50    33.49   1035.71    49.65
  VOL          256.85    1.00   147.00   5806.00   327.98
  FT            34.60    2.00    33.00     89.00    16.31
  SIG           21.54    0.00    16.37    366.19    19.74

(a) SPD41 refers to the execution spread for transactions
between customers and dealers

(b) SPD44 refers to the execution spread for transactions
between customers

Table 5--Determinants of the Execution Spread
The table is based on the following regression:

SPREA[D.sub.t] = [[alpha].sub.0] + [[alpha].sub.1] VO[L.sub.t]
+ [[alpha].sub.2] F[T.sub.t] + [[alpha].sub.3] SI[G.sub.t]
+ [[alpha].sub.4] D1 + [[alpha].sub.5] D2 + [[epsilon.sub.t] (1)

where, SPREA[D.sub.t] is the execution spread, VO[L.sub.t], is
the customer volume, F[T.sub.t] is number of floor traders,
SI[G.sub.t] is standard deviation of futures prices, D1 is
the dummy variable and the value is equal to 1 for the first
30 minutes of trading, and 0 otherwise, D2 is the dummy
variable and the value is equal to 1 for the last 30 minutes
of trading, and 0 otherwise, and t is the time subscript of
a five-minute interval

                                        Initiation Criteria

Commodity      Variables                SPD41 (a)   p-value

Pork Bellies   Intercept                1.153       0.0986
               VOL                     -0.018       0.0001
               FT                      -0.377       0.0001
               SIG                      2.105       0.0001
               D1                      -0.212       0.7338
               D2                      -0.084       0.8995
               Adjusted [R.sup.2]   0.83

Live Cattle    Intercept                3.344       0.0001
               VOL                      0.004       0.0001
               FT                      -0.037       0.0789
               SIG                      1.578       0.0001
               D1                       0.369       0.2506
               D2                      -0.391       0.2014
               Adjusted [R.sup.2]   0.73

Swiss Franc    Intercept                3.403       0.0001
               VOL                      0.001       0.3175
               FT                      -0.141       0.0001
               SIG                      1.842       0.0001
               D1                       1.232       0.0823
               D2                      -0.126       0.8670
               Adjusted [R.sup.2]   0.75

Deutschemark   Intercept                2.505       0.0001
               VOL                     -0.001       0.1514
               FT                      -0.080       0.0001
               SIG                      1.852       0.0001
               D1                       1.367       0.0723
               D2                      -1.992       0.0118
               Adjusted [R.sup.2]   0.64

                                        Initiation Criteria

Commodity      Variables                SPD44 (b)   p-value

Pork Bellies   Intercept                3.652       0.0052
               VOL                     -0.008       0.1670
               FT                      -0.098       0.3915
               SIG                      1.695       0.0001
               D1                       2.040       0.0441
               D2                       0.020       0.9850
               Adjusted [R.sup.2]   0.71

Live Cattle    Intercept                1.278       0.0011
               VOL                      0.000       0.4970
               FT                       0.030       0.1798
               SIG                      1.574       0.0001
               D1                       0.322       0.3503
               D2                      -0.197       0.5621
               Adjusted [R.sup.2]   0.73

Swiss Franc    Intercept                1.350       0.1503
               VOL                     -0.003       0.0533
               FT                       0.027       0.4420
               SIG                      1.635       0.0001
               D1                       0.788       0.3846
               D2                      -2.521       0.0120
               Adjusted [R.sup.2]   0.69

Deutschemark   Intercept                1.677       0.1346
               VOL                     -0.002       0.0406
               FT                      -0.016       0.5857
               SIG                      1.707       0.0001
               D1                       0.025       0.9821
               D2                       0.451       0.7185
               Adjusted [R.sup.2]   0.62

(a) SPD41 refers to the execution spread for transactions
between customers and dealers

(b) SPD44 refers to the execution spread for transactions
between customers

(1) Locke and Venkatesh (1997) find that Roll's estimator tends to underestimate the futures markets transaction costs, while the method of moments estimator tends to overestimate the futures markets transaction costs.

(2) Trades executed at spread mid-point are not classified as either buyer- or seller- initiated trades.

References

(1.) Bhattacharya, M., "Transactions Data Tests of Efficiency of the Chicago Options Exchange," Journal of Financial Economics, 12, no. 2 (August 1983), pp. 161-185.

(2.) Chung, K.H., B.F. Van Ness, and R.A. Van Ness, "Limit Orders and the Bid-Ask Spread," Journal of Financial Economics, 53, no. 2 (August 1999), pp. 255-287.

(3.) Ding, D.K., "The Determinants of Bid-Ask Spreads in the Foreign Exchange Futures Market: A Microstructure Analysis," The Journal of Futures Markets, 19, no. 3 (May 1999), pp. 307-324.

(4.) Ferguson, M. F., and S. C. Mann, "Execution Costs and their Intraday Variation in Futures Markets," Journal of Business (January 2001), Volume 74, Issue 1, pp. 125-160.

(5.) Glosten, L., and P. Milgrom, "Bid, Ask and Transaction Prices in a Specialist Market with Heterogeneously Informed Traders," Journal of Financial Economics, 14, no. 1 (March 1985), pp. 71-100.

(6.) Kandel, E., and L.M. Marx, "Nasdaq Market Structure and Spread Patterns," Journal of Financial Economics, 45, no. 1 (July 1997), pp. 61-89.

(7.) Klock, M.D., and T. McCormick, "The Impact of Market Maker Competition on Nasdaq Spreads," The Financial Review, 34, no. 4 (November 1999), pp. 55-74.

(8.) Laux, P.A., and A.J. Senchack, Jr., "Bid-Ask Spread in Financial Futures," The Journal of Futures Markets, 12, no. 6 (December 1992), pp. 621-634.

(9.) Locke, P.R., and S.C. Mann, "Professional Trader Discipline and Trade Disposition" Journal of Financial Economics (in press).

(10.) Locke, P.R., and P.C. Venkatesh, "Futures Market Transaction Costs," The Journal of Futures Market, 17, no. 2 (April 1997), pp. 229-245.

(11.) Ma, C.K., R.L. Peterson, and R.S. Sears, "Trading Noise, Adverse Selection, and Intraday Bid-Ask Spreads in Futures Markets," The Journal of Futures Markets, 12, no. 5 (October 1992), pp. 519-538.

(12.) Menyah, K., and K. Paudyal, "The Components of Bid-Ask Spreads on the London Stock Exchange," Journal of Banking & Finance, 24, no. 11 (November 2000), pp. 1767-1785.

(13.) Roll, R., "A Simple Implicit Measure of the Effective Bid-Ask Spread in an Efficient Market," Journal of Finance, 39, no. 4 (September 1984), pp. 1127-1139.

(14.) Stoll, H., "Inferring the Components of Bid-Ask Spread: Theory and Empirical Tests," Journal of Finance, 44, no. 1 (March 1989), pp. 115-134.

(15.) Savickas, R., and A.J. Wilson, "On Inferring the Direction of Option Trades," Journal of Financial and Quantitative Analysis, 38, no. 4 (December 2003), pp. 881-902.

(16.) Smith, T., and R.E. Whaley, "Estimating the Effective Bid/Ask Spread from Time and Sales Data," The Journal Futures Markets, 14, no. 4 (June 1994), pp. 437-456.

(17.) Wahal, S., "Entry, Exit, Market, Makers and the Bid-Ask Spread," Review of Financial Studies, 10, no. 3 (Fall 1997), pp. 871-901.

(18.) Wang, G.H.K., and J. Yau, "Trading Volume, Bid-Ask Spread, and Price Volatility in Futures Markets," The Journal of Futures Markets, 20, no. 10 (November 2000), pp. 943-970.

(19.) Wang, G.H.K., J. Yau, and T. Baptiste, "Trading Volume and Transaction Costs in Futures Markets," The Journal of Futures Markets, 17, no. 7 (October 1997), pp. 757-780.

(20.) Wang, G.H.K., R.J. Michalski, J.V. Jordan, and E.J. Moriarty, "An Intraday Analysis of Bid-Ask Spreads and Price Volatility in the S&P 500 Index Futures Market," The Journal of Futures Markets, 14, no. 7 (October 1994), pp. 837-859.

Olgun Fuat Sahin

Minnesota State University Moorhead

Pattarake Sarajoti

Sasin GIBA of Chulalongkorn University