Market Competition in Crude Oil Futures. Elena Skouratova Department of Finance University of Texas at San Antonio

Market Competition in Crude Oil Futures Elena Skouratova Department of Finance University of Texas at San Antonio Yiuman Tse* Department of Finance University of Texas at San Antonio Valeria Martinez Department of Finance Fairfield University September 2008 JEL classification: G13; G14; G18 Keywords: Crude oil; ICE; NYMEX *Corresponding author. College of Business, 501 West Durango Blvd., San Antonio, Texas 78207. Tel.: 210-458-2503; fax: 210-458-2515; E-mail: yiuman.tse@utsa.edu

Market Competition in Crude Oil Futures Abstract On February 3, 2006 the ICE Futures introduced electronic trading of West Texas Intermediate (WTI) crude oil futures contracts that compete directly with the NYMEX benchmark light, sweet crude oil contracts traded on the floor. In response to ICE s electronic contract, on September 5, 2006 the New York Mercantile Exchange (NYMEX) launched electronic trading of their standard-size crude oil futures contracts on the Chicago Mercantile Exchange s (CME) Globex electronic trading platform. We analyze market quality conditions before and after the introduction of the NYMEX full-size crude oil contracts. We find price clustering of floor-traded regular-size NYMEX futures significantly improved after introduction of parallel electronic trading, indicating lower trading costs and better market quality. Price discovery analysis revealed that electronically traded futures contracts contribute similarly to fundamental value determination, open-outcry trading provides little price discovery. The research provides insights into the importance of regulatory parity between US exchanges and non-regulated markets.

1. Introduction In recent years, there has been a strong interest in regulation of futures market reporting systems, as well as their consolidation, fragmentation, and integrity. At the same time competition is intensifying and globalization is escalating. The question is whether increased competition among exchanges has a favorable effect on market quality, including decreased trading costs and improved price discovery; or a negative effect caused by fragmentation of order flow. Throughout modern history, energy products and energy markets have played significant roles in the economies of all industrialized countries. Energy producers, distributors, consumers, and financial institutions depend on efficient operation of spot and futures markets in hedging risks and performing price discovery. The London-based ICE Futures Exchange, a subsidiary of the Intercontinental Exchange (ICE), is a leading European energy exchange. It dominates the market for Brent Crude Oil. The New York Mercantile Exchange (NYMEX) is the world's largest trading market for energy and dominates the market for West Texas Intermediate (WTI) light, sweet crude oil futures contracts, a benchmark for US oil prices. NYMEX light, sweet crude oil futures (crude oil futures from here on) were introduced in 1983 and became the world s most liquid forum for crude oil. In 1999 the Commodity Futures Trading Commission permitted ICE members to trade electronically from the US without designating the exchange as a US futures exchange and without US oversight and regulation. On February 3, 2006 ICE Futures exchange introduced electronic trading of the WTI crude oil futures contract that competes directly with the NYMEX benchmark crude oil contract traded on the floor. The ICE s futures contract replicates 1

most of the terms of the NYMEX contract and is also based on the WTI crude oil. The contract is designed to meet the needs of international and domestic investors and trades almost 24 hours a day. This was the first time that US crude oil futures contracts were traded on an exchange outside the US. 1 By early September, trading of the ICE WTI contract had attracted nearly a 35% market share. NYMEX faced growing competition from ICE WTI contracts and pressure from its customers to offer electronic trading. Thus on September 5, 2006 NYMEX launched its physically-delivered electronic standard-size crude oil futures contracts on the CME Globex platform. Electronic trading of ICE WTI Crude Oil contracts fell to about a 30% market share by February 2007. NYMEX once again proved that it is well equipped to meet the competition. However, recent interest in the trading of crude oil products does not stop here. One of the latest additions to the energy market is the US Oil Fund (USO) that began trading on the American Stock Exchange (AMEX) in April 2006. The USO exchange traded fund (ETF) is primarily invested in WTI crude oil futures contracts traded on the NYMEX. Nonetheless, in contrast to the futures contracts that trade in units of 1,000 barrels, the cost of a USO ETF share represents a single barrel, making it more affordable to small investors. The USO ETF allows individual investors to take part in the growing oil commodities market. On the other hand, analysts argue that the US Oil Fund is an overspecialized security that contributes little to the market efficiency and only seems to create another revenue stream. The goal of this paper is to study the impact of the introduction of full-size electronically traded futures contracts and the USO ETF on market quality conditions 1 Levin,. C. and Coleman, N. (2006). 2

such as: volume, price clustering, effective spreads, and price discovery. The paper has three objectives: evaluate the effect on market quality of competition in the futures market between US and non-us regulated exchanges; compare the simultaneous trading of futures on the open outcry and electronictrading platforms. analyze the effect of multi-market NYMEX and ICE trading of Crude oil futures and the USO ETF. Our work should provide insights for regulators and lawmakers on the importance of regulatory parity between non-regulated markets and US exchanges, which are subject to Commodity Futures Trading Commission (CFTC) regulation, NYMEX is fully regulated by CFTC, while the ICE Futures exchange is exempt from regulation under the no-action process. 2 NYMEX is particularly concerned about the lack of parity in WTI crude oil futures contracts regulations with respect to largetrader reporting requirements and position limits. According to the NYMEX evenhanded regulation is crucial for preserving market efficiency, open and fair competition, and customer protection. A recent investigation of the hedge fund Amaranth Advisors' distortion of natural gas prices 3 concludes that traders regularly buy and sell natural gas contracts on both the NYMEX and the ICE, and that it is almost impossible for NYMEX to prevent price manipulation and excessive speculation if large traders choose to trade on the ICE. At the same time, such large trader behavior may distort NYMEX prices. NYMEX CEO, Dr. James Newsome, concludes that the regulatory disparity between 2 As of June 2008, the no action process has ceased. Nonetheless, during our period of study NYMEX oil futures were subject to CFTC regulation while ICE s WTI crude oil futures were not. 3 Levin, C. and Coleman, N. (2006). 3

the NYMEX and certain except commercial markets (ECMs), particularly the ICE, which are functionally equivalent to each other, has created serious challenges for the CFTC as well as for NYMEX in its capacity as a self-regulatory organization (SRO). 4 In general, market participants believe that regulation should not create incentives for traders to choose trading venues that are trading the same or similar products on the basis of differences in regulatory handling rather than on the basis of market quality or the intrinsic features of the products. 5 2. Literature review Several papers argue that the use of electronic trading systems decreases transaction costs and increases liquidity, claiming the superiority of screen-based trading. Examples of this work include: Martinez and Tse (2007), Frino, Hill, Jarnecic, and Aitken (2004), Domowitz (2002), Frino, McInish, and Toner (1998), and Pirrong (1996). They compare bid-ask spreads under the two trading environments and find significantly lower trading costs under electronic trading concluding electronic trading provides better market quality conditions. Nonetheless, other authors like Thiessen (2002), Tse and Zabotina (2001), and Martens (1998) find that floor and screen based systems complement each other and play important roles under different market conditions. Hence, arguments remain about which system is optimal for futures trading. Nonetheless, the introduction of a NYMEX electronic full-size crude oil contract not only poses the question of whether open outcry or electronic is the best type of trading venue, but also makes us wonder about the effects of competition and fragmentation generated in the crude oil futures market. 4 Testimony of Dr. J. Newsome, CEO, NYMEX Inc. (2007). 4

So far, the literature on market competition and fragmentation has been inconclusive. For example, according to Hamilton (1979) and McInish and Wood (1996), when individual stocks trade on multiple markets, they exhibit lower bid-ask spreads and lower price volatility or a competition effect. At the same time, Mendelson (1987) constructs a model arguing that individual stocks traded on multiple exchanges, reduce market liquidity and increase price volatility causing the market to fragment. Similar results are empirically obtained by Bessembinder and Kaufman (1997) and Davis and Lightfoot (1998). In general, they show that multimarket trading does not reduce, but sometimes increases bid-ask spreads and volatility. Hence in this work, fragmentation dominates competition. Yet a third angle to our story refers to market conditions under different regulatory settings. How does trading in a US regulated exchange compare to trading in a non US regulated exchange? Pirrong (1999) studies the complex nature of financial exchanges, taking into account the effect of members on competition and the view of exchanges as an entity. He finds that if exchanges exercise market power, or are protected from competitive entry, they may adopt inefficient rules that benefit members at the expense of customers. Eleswarapu, Thompson, and Venkataraman (2004), study the impact of regulation on trading costs and the degree of information asymmetry in equity markets. They find reduced trading costs and less information asymmetry as a result of regulation. Cihák and Podpiera (2008) analyze the effects of integrated supervision on financial institutions and securities markets and conclude that greater supervisory 5 Statement concerning recent CFTC hearing regarding foreign boards of trade. NYMEX.com., 7/26/2006. 5

integration is associated with higher quality supervision. Thus it seems that regulation may in fact increase market quality conditions. 3. Data Our research concentrates on the six-month period before the introduction of the electronically traded regular-size NYMEX crude oil futures contract (period 1: March to August 2006), and the six month period following this event (period 2: September 2006 to February 2007). We use tick-by-tick data for oil futures prices on regular-size contracts from the NYMEX (floor and electronic) and ICE (electronic). Data for NYMEX crude oil futures contracts are provided by Tick Data Inc. The tick-by-tick data for WTI ICE crude oil futures are obtained from CQG Data Inc. Daily prices and trading volumes for futures contracts are obtained from Commodity Systems Inc. For our analysis we use the nearby most actively traded futures contract and focus on the daily openoutcry trading period, the time during which the bulk of the market s trading takes place. The regular-size NYMEX crude oil futures contract trades in units of 1,000 barrels, and the delivery point is Cushing, Oklahoma. The open-outcry trading session is conducted from 9:00 a.m. until 2:30 p.m. Electronic trading is conducted from 6:00 p.m. until 5:15 p.m. EST on the CME Globex trading platform, Sunday through Friday. The minimum price fluctuation of the contract is $0.01 per barrel ($10.00 per contract). The ICE WTI crude oil futures contract trades in units of 1,000 barrels. The contract price is in US dollars and cents per barrel, with a minimum price fluctuation of $0.01 per barrel, ($10.00 per contract). The WTI ICE futures contract is cash 6

settled against the prevailing market price for the NYMEX s crude oil futures. It trades from 8:00 p.m. to 6:00 p.m. EST time. In 2002 NYMEX introduced a smaller size crude oil futures contract traded electronically (e-mini). E-mini futures were designed to attract smaller investors and individual speculators. The NYMEX e-mini crude oil futures contract is for 500 barrels of crude oil, 50% of the size of a standard futures contract. The contract is traded electronically on the CME Globex platform from 6:00 p.m. to 5:15 p.m. EST, Sundays through Fridays. The minimum price fluctuation is $0.025 per barrel ($12.50 per contract). Given the importance of small investor trading in current markets and for completeness of our analysis, we include the e-mini futures contract in our study of crude oil market quality conditions. We also incorporate information on the recently created US Oil ETF. Intraday data for the NYMEX e-mini oil contract comes from Tick Data Inc. and data for the US Oil ETF comes from the Trade and Quote database (TAQ). 4. Methodology and Results The focal point of the analysis is on the changes in market quality conditions including: volume, price clustering, trading costs (bid-ask spreads), and price discovery. 4.1. Volume Figure 1 presents the daily market share for the regular-size NYMEX openoutcry contract, the NYMEX e-mini electronic contract, and the ICE WTI electronic contract. Panel A shows market shares for the March-August 2008 period, prior to the 7

introduction of NYMEX regular-size electronic contracts. Here we observe the market share of the NYMEX regular-size contract gradually decreased from 80% at the beginning of March 2006 to 48% by the end of August 2006. The market share of the ICE WTI electronic contract progressively increased and reached 35% of daily trading volume in the crude oil futures market by August 2006. The NYMEX e-mini electronic contract held the least volume starting from 3.5% at the beginning of the period and increasing to 16.9% by the end of August 2006. During the September 2006-February 2007 period, (Panel B, Figure 1) we see the picture changes. The newly introduced regular-size NYMEX electronic contract was able to capture about 30% of the total daily volume by the end of January 2007, while the regular-size NYMEX open-outcry contract continued to lose its share of volume until it decreased to one third of the total traded volume by the end of February 2007. The ICE WTI electronic contract also receded to about one third of total daily volume. In addition, the share of the NYMEX e-mini futures contract slipped to 3.97%. Overall the NYMEX attempt to regain its market share was positive. 4.2. Price Clustering Price clustering refers to the fact that transaction prices are not evenly distributed among all possible ending values, but tend to cluster around even numbers. It seems price clustering helps traders simplify negotiations. Researchers have long studied this phenomenon as an indicator of market quality. Harris (1991) reports that stock prices cluster on round fractions. He argues that price clustering rises with price level and volatility and declines with capitalization and transaction frequency. Pirrong (1996) finds that price clustering is higher in open-outcry venues than in electronic trading, suggesting collusion between locals. Tse and Zabotina (2001) also propose 8

that clustering of spreads over even ticks in open-outcry markets implies possible collusion by locals, and means higher costs for investors. Figure 3 illustrates the frequency distribution of the last digit in the prices of regular-size floor and electronic NYMEX contracts and the ICE WTI electronic contract. The results reveal significant price clustering for regular-size NYMEX openoutcry contracts in both periods analyzed. Panel A shows that between March and August of 2006, 41.4% of the last digits of the prices for regular-size NYMEX openoutcry contract were zero; while 36.7% of the prices ended in the digit five. The ICE WTI futures contracts exhibit no significant price clustering. Panel B in Figure 3 shows price clustering considerably improves in regularsize NYMEX open-outcry futures contracts in the September 2006-February 2007 period, following the introduction of the NYMEX full-size screen-based contract. 6 The regular-size NYMEX electronic futures contract and the ICE WTI electronic futures contract do not show evidence of price clustering during this period. We can explain the improved situation regarding price clustering of NYMEX open-outcry futures by the increased competition from electronic trading of the same contract. Nonetheless, we must keep in mind that out of the three contracts analyzed, only the NYMEX open-outcry contract presents price clustering. This shows lower market quality conditions in open outcry than in electronic venues. 4.3. Effective Bid-Ask Spread To analyze liquidity in the crude oil futures market, we study bid-ask spreads for NYMEX and ICE futures contracts. Do to the lack of bid-ask spread data in the 6 We exclude the NYMEX e-mini contract from the price clustering analysis because this contract s minimum tick size is two and a half times the tick size of the NYMEX and ICE regular-size contracts. Thus, no appropriate comparisons can be made in terms of price clustering. 9

futures market, we estimate effective spreads using the Thompson and Waller (1988) estimator. Thompson and Waller estimate the bid-ask spread as the average absolute tickto-tick price change over a time interval: where 1 T TW = = p t 1 t (1) T T t= 1 p t I is the sum of the absolute non-zero price changes. Frino, McInish, and Toner (1998) discover that mean and median spreads are wider on LIFFE floor trading compared to spreads from electronic DTB trading. At the same time, they state that, despite higher liquidity and thus narrower bid-ask spreads of DTB in quiet markets, LIFFE offers more liquidity during highly volatile periods. Tse and Zabotina (2001) find that median spreads are higher in open-outcry markets than in electronic markets. Tse and Zabotina state that decreased bid-ask spreads mean lower transaction costs for investors. Martens (1998) finds that the adverse selection problem is smaller in open-outcry trading, leading to lower bid-ask spreads. Tse and Erenburg (2003) find that competition for order flow improves market quality and tends to narrow spreads in all markets. They conclude competition does not increase fragmentation of the Nasdaq 100 ETF. Tse and Xiang (2005) find a significantly reduced bid-ask spread for the regular-size NYMEX crude oil futures after introduction of e-mini futures in 2002. Table 1 presents estimated spreads for the floor-traded regular-size NYMEX contract, the e-mini NYMEX contract, and the ICE WTI contract for the period before and after the introduction of the NYMEX regular-size electronically traded contract. The table shows that spreads for the NYMEX open-outcry and the ICE contracts remained fairly constant from the first to the second period. As verified in Table 2, average spreads for these contracts did not change significantly form one subperiod to 10

another. Table 2 also shows that spreads for the NYMEX mini contract increased significantly from the first to the second period, showing signs of market quality deterioration. It is important to note that although the NYMEX e-mini contract has a minimum tick size which is two and a half times the tick size for the regular-size contracts, over the course of our study the e-mini spreads are only slightly higher than those for the ICE futures and about half the size of spreads for the NYMEX full-size open-outcry contract. This shows that although the e-mini market quality drops with the introduction of the NYMEX regular-size electronic contract, its liquidity is considerably better than that of the NYMEX open-outcry contract. Despite the fact that the ICE WTI electronic futures contract continues to gain daily volume, the effective bid-ask spread does not change significantly. Overall, we find that the average daily spread of the floor-traded regular-size NYMEX contract is three times larger than the spread of the electronically traded futures contracts (NYMEX and ICE Futures). 4.4. Price Discovery For related products trading in multiple venues, price discovery is essential in determining the informationally dominant market in terms of the development of the asset s implicit efficient price. Hence prices of regular-size NYMEX open-outcry and electronic futures, NYMEX e-mini futures, WTI ICE electronic futures, and the USO ETF are strongly related and driven by the same information. The prices of the crude oil futures and the USO ETF in different markets are kept from drifting apart because of intermarket arbitrage. These prices are therefore cointegrated and share one common factor, the implicit efficient price described by Hasbrouck (1995). He defines the concept of information shares (IS) as the relative contribution of each market to price discovery. Hasbrouck (2002) and Sapp (2002) 11

argue that the model does not conclude which market has the best price. Instead, the model indicates which market moves first in the price-correction process. We follow the Hasbrouck information share model to examine the pricediscovery process. The model starts with the examination of the vector error correction (VEC) representation of n cointegrated time series Y t ={y it }: k Y t = αβ'y t-1 + A j Y t-j + ε t (2) j=1 where α is the error correction term, β is the cointegrating vector, A j s are matrices of parameters, and ε t is a zero-mean vector of serially uncorrelated residuals with covariance matrix Ω = {σ ij }. The VEC model has two parts: the first, αβ'y t-1, represents the long-run relation between price series, the second, A j Y t-j + ε t represents the short-run dynamics induced by market frictions. The VEC model in Eq. (1) can also be represented in an integrated form of a vector moving average (VMA): t Y t = ιη ( ε τ ) + Ψ*(L) ε t (3) τ=1 where ι is a column vector of ones, η is a common row vector of the impact matrix, and Ψ* is a matrix polynomial in the lag operator L. The first part of Eq. (3) is the common factor or permanent component. The second part is the transitory component which does not have a permanent impact on Y t. Hasbrouck defines the information share of a trading market as the portion of the variance of the common factor that is attributable to innovations in that market. The information share of a market can then be represented as: 2 2 ηj σ j ISj = (4) ηωη' 12

where the numerator is the variance contribution of the j th market, and the denominator is the total variance of the common factor. If the covariance matrix, Ω, is not diagonal, the information share is not exactly identifiable, and the result depends on the ordering of the variables in a Cholesky factorization of the covariance matrix, Ω=FF', where F is a lower triangular matrix. Different ordering of the variables will produce lower and upper bounds of the markets information shares. In Hasbrouck s (1995) study the upper and lower bounds of information shares of NYSE and other trading centers are very close. Nonetheless, many other studies using lower frequency data, including Martens (1998), Huang (2002), and Booth et al (2002), present considerable differences between lower and upper bounds. As suggested by Baillie, et al (2002) the average of the information shares given by all orderings in the Cholesky factorization is a reasonable estimate of the market contribution to the price discovery. Similar to earlier research, we discover a wide range between lower and upper information share estimations. So following Baillie, et al (2002), we obtain the average daily information share of each market in our study. We use the Hasbrouck (1995) information share model to analyze price discovery for all futures contracts and the USO ETF from March 2006 to February 2007. We use second-by-second and minute-by-minute trade prices for futures contracts and the USO ETF, reporting only minute-by-minute results to save space. We estimate the lower and upper bounds of the information share for each contract using all ordering combinations. We then estimate the average information share for each contract during each trading day, using the average of the upper and lower bounds of the information shares for each market. Last, we average the daily results to present mean monthly information shares for each futures contract and the USO ETF. 13

The focus of the analysis is on the relative importance of three dimensions of price discovery: trading mechanism (open outcry vs. electronic trading), trading market (NYMEX vs. ICE Futures), and trading instrument (futures vs. ETF). Panel A of Table 3 shows information shares during the March to August 2006 period for the regular-size NYMEX open-outcry contract, the NYMEX e-mini contract, and the WTI ICE futures contract. The results indicate the NYMEX e-mini and ICE futures each provide roughly 40% of the price discovery with the NYMEX pit contract supplying the remainder. The NYMEX e-mini futures contribution to fundamental value during this period is considerably high, despite its low share market share of 9.33% during this time frame. Panel B of Table 3 shows information shares for the regular-size NYMEX electronic contract, the regular size NYMEX open-outcry contract, the NYMEX e- mini contract, and the WTI ICE futures contract for the September 2006-February 2007 period. We observe that the NYMEX regular and mini electronic contracts along with the ICE electronic contract each contribute roughly 30% of the price information with the NYMEX open-outcry contract contributing the remainder. For this time period the information share of the regular-size NYMEX electronic contract increases from 27% to 33% while that of the NYMEX e-mini futures fluctuates between 27% and 29%. However, the contribution of the regularsize NYMEX open-outcry contract declines from 14% to 9%. The ICE contract s contribution to price discovery which starts at 42% in March of 2006 steadily declines to 29% by February 2007. Overall, it seems that NYMEX is regaining its pricing power, and electronic trading is dominant over the open-outcry market. We repeat the price discovery analysis of the crude oil market using prices for the three NYMEX futures contracts and the USO ETF. We see that the USO ETF 14

provides between 13% and 24% of price discovery between September 2006 and February 2007. Table 4 illustrates, that even though the USO ETF contributes less to the fundamental value than NYMEX electronically traded regular-size and e-mini futures contracts, it surpasses the information share of the NYMEX pit-traded regularsize futures contract. Next we add ICE WTI futures contract to our analysis. Table 5 shows information shares for the three NYMEX futures contracts, the ICE futures contract and the USO ETF. We observe all three electronically traded futures contracts (NYMEX full size, NYMEX e-mini, and ICE) provide comparable contributions to price discovery in the crude oil market. The USO ETF s contribution to fundamental value increases from 9% in September 2006 to 18% in February 2007 and openoutcry NYMEX futures provide the lowest share of price discovery out of the five financial instruments analyzed. Overall, our results support previous research that conjectures the dominance of electronic trading. 5. Conclusions On September 5, 2006, the NYMEX launched electronic trading of its regularsize crude oil futures contract, in response to the launch of the WTI crude oil contract by its rival, ICE futures exchange. We analyze the impact of competition between two exchanges on crude oil futures market quality before and after the introduction of the NYMEX full-size electronic contract. We find that electronic trading of regular-size NYMEX crude oil futures gains more than a 30% share of the total trading volume by the end of February 2007. At the same time the market share of ICE WTI oil futures contracts declines and becomes stable at a 30% mark. Price clustering of floor-traded regular-size NYMEX futures improves after the introduction of a parallel regular-size NYMEX electronic 15

contract. This improvement suggests lower trading costs and better market quality on the NYMEX. Electronic trading of regular-size NYMEX futures and WTI ICE futures does not display price clustering. Effective bid-ask spreads increased in the electronically traded NYMEX e-mini futures contract as a result of order flow fragmentation. Finally, our price discovery analysis reveals that while all three electronically traded futures contracts contribute significantly to the fundamental value of crude oil, open-outcry trading provides very little price discovery. We find that, despite low trading volume, e-mini NYMEX futures provide an important contribution to crude oil s fundamental value. We also show the USO ETF provides an important portion of price discovery in the crude oil market, indicating this is not a redundant product. In sum, electronically traded futures present better market quality conditions than openoutcry futures; small investor oriented oil e-mini futures and Lefts also provide a significant contribution to price discovery; and regulatory differences between trading venues can help exchanges attract investors. Based on the above, it comes as no surprise that on June 27, 2008, the CFTC announced that Commission staff had amended the no-action relief letter in which ICE Futures is allowed direct access to U.S. markets. The amendment states that for ICE Futures to continue to have direct access to U.S. markets they must adopt equivalent U.S. position limits and accountability levels on its West Texas Intermediary (WTI) crude oil contract. ICE Futures will also have to follow similar U.S. hedge exemption requirements and report violations of these provisions to the CFTC. For surveillance purposes, the CFTC also requires that they provide detailed market information, equivalent to U.S. standards. 16

References Baillie, R., Booth, G., Tse, Y., & Zabotina, T. (2002). Price Discovery and Common Factors Models. Journal of Financial Markets, 5, 305-321. Bessembinder, H., & Kaufman, H. (1997). A comparison of trade execution costs for NYSE and Nasdaq listed stocks. Journal of Financial and Quantitative Analysis, 32, 287 310. Booth, G., Lin, J., Martikainen, T., & Tse, Y. (2002). Trading and pricing in upstairs and downstairs stock markets. Review of Financial Studies, 15, 1111-1135. CFTC Press release no. 5511-08. June 27, 2008. CFTC Conditions Foreign Access on Adoption of Position Limits on London Crude Oil Contract Cihak, M. & Podpiera R. (2008). Integrated financial supervision: Which model? North American Journal of Economics and Finance, 19, 135-152. Davis, J., & Lightfoot, L. (1998). Fragmentation versus consolidation of securities trading: Evidence from the operation of rule 19c-3. Journal of Law and Economics, 41, 209-238. Domowitz, I. (2002). Liquidity, transactions costs, and reintermediation in electronic markets. Journal of Financial Services Research, 22, 141-157. Eleswarapu, V., Thompson, R., & Venkataraman, K. (2004). The impact of regulation fair disclosure: Trading costs and information asymmetry. Journal of Financial and Quantitative Analysis, 39, 209-225 Frino, A., Hill, A., Jarnecic, E., & Aitken, M. (2004). The impact of electronic trading on bid-ask spreads: Evidence from futures markets in Hong Kong, London and Sidney. Journal of Futures Markets, 24, 675-696. 17

Frino, A., McInish, T., & Toner, M. (1998). The Liquidity of automated exchanges: New evidence from German bund futures. Journal of International Financial Markets, Institutions & Money, 8, 225-242. Hamilton, J. (1979). Marketplace fragmentation, competition, and the efficiency of the stock exchange. The Journal of Finance, 34, 171-187. Harris, L. E. (1991). Stock price clustering and discreteness. Review of Financial Studies, 4, 389-415. Hasbrouck, J. (1995). One security, many markets: Determining the contribution to price discovery. Journal of Finance, 50, 1175-1199. Hasbrouck, J. (2002). Stalking the "efficient price" in market microstructure specifications: an overview. Journal of Financial Markets, 5, 329-339 Huang, R. (2002). The quality of ECN and Nasdaq market maker quotes. Journal of Finance, 57, 1285-1319. Levin, C. & Coleman, N. (2006). United States Senate, Permanent Subcommittee on Investigations, Committee on Homeland Security and Governmental Affairs. The role of market speculation in rising oil and gas prices: A need to put the cop back on the beat. Staff report. Martens, M. (1998). Price discovery in high and low volatility periods: Open outcry versus electronic trading. Journal of International Financial Markets, Institutions and Money, 8, 243-260. Martinez, V., & Tse, Y. (2007). Multi-market trading of gold futures. Review of Futures Markets, 15, 239-261. McInish, Wood (1996). In A. W. Lo (Ed), Competition, fragmentation and market quality in the industrial organization and regulation of the security (pp. 63-74). Chicago: The University of Chicago press. 18

Mendelson, H. (1987). Consolidation, fragmentation and market performance. Journal of Financial and Quantitative Analysis, 22, 189-207. www.nymex.com 7/26/2006. Statement Concerning Recent CFTC Hearing Regarding Foreign Boards of Trade. Pirrong, C. (1999). The organization of financial exchange markets: Theory and evidence. Journal of Financial Markets 2, 329-357. Pirrong, C. (1996). Market liquidity and depth on computerized and open outcry trading systems: A comparison of DTB and LIFFE Bund contracts. Journal of Futures Markets, 16, 519-544. Sapp, S. (2002). Price leadership in the spot foreign exchange market. Journal of Financial and Quantitative Analysis, 37, 425-428. Testimony of Dr. J. Newsome, CEO NYMEX, Inc. (2007). Senate Committee on Homeland Security and Governmental Affairs, Permanent Subcommittee on Investigations concerning Excessive Speculation in the Natural Gas Market. Theissen, E., (2002). Price discovery in floor and screen trading systems. Journal of Empirical Finance, 9, 455-474. Thompson, S., & Waller, M. (1988). Determinants of liquidity costs in commodity futures markets. Review of Futures Markets, 7, 110-126. Tse, Y., & Erenburg, G. (2003). Competition for order flow, market quality and price discovery in the Nasdaq 100 index tracking stock. Journal of Financial Research, 26, 301-318. Tse, Y., & Xiang., J. (2005). Market quality and price discovery: Introduction of the E-mini energy futures. Global Finance Journal, 16, 164-179. 19

Tse, Y., & Zabotina, T. (2001). Transaction costs and market quality: Open outcry versus electronic trading. The Journal of Futures Markets, 21, 713-735.. 20

Figure 1 Daily Market Shares for Crude Oil Futures Panel A: First period - March to August 2006 90.00 80.00 70.00 60.00 % 50.00 40.00 30.00 20.00 10.00 0.00 20060301 20060308 20060315 20060323 20060330 20060406 20060413 20060421 20060428 20060505 20060512 20060519 20060526 20060605 20060612 20060619 20060626 20060705 20060712 20060719 20060726 20060802 20060809 20060816 20060823 20060830 date NYMEX pit NYMEX e-mini ICE WTI Panel B: Second period - September 2006 to February 2007 70.00 60.00 50.00 40.00 30.00 20.00 10.00 0.00 20060905 20060912 20060919 20060926 20061003 20061010 20061017 20061024 20061031 20061107 20061114 20061121 20061201 20061208 20061215 20061222 20070103 20070110 20070118 20070125 20070201 20070208 20070215 20070223 % date NYMEX electronic NYMEX pit NYMEX e-mini ICE WTI The figure shows daily market share of the NYMEX screen-based and floor regular-size contracts (NYMEX electronic and NYMEX pit), the NYMEX mini contract (NYMEX e-mini), and the ICE regular-size contract (ICE WTI). 21

Figure 2 Price Clustering in Crude Oil Futures Markets Panel A: First period - March to August 2006 45 40 35 30 25 % 20 15 10 5 0 0 1 2 3 4 5 6 7 8 9 last digit of futures prices ICE WTI NYMEX pit Panel B: Second period - September 2006 to February 2007 30 25 20 % 15 10 5 0 0 1 2 3 4 5 6 7 8 9 NYMEX electr NYMEX pit WTI ICE Frequency distribution of the last digit of crude oil futures contracts. Price clustering is compared during the open-outcry trading session: for the NYMEX screen-based and floor regular-size contracts (NYMEX electr and NYMEX pit) and the ICE regular-size contract (ICE WTI). 22

Table 1 Effective Spreads Period 1 Daily Average Spread Period 2 Daily Average Spread Panel A: Floor traded regular-size NYMEX crude oil futures Mar-06 0.000951 Sep-06 0.000795 Apr-06 0.000802 Oct-06 0.000790 May-06 0.000818 Nov-06 0.000656 Jun-06 0.000763 Dec-06 0.000719 Jul-06 0.000737 Jan-07 0.000891 Aug-06 0.000740 Feb-07 0.000818 Panel B: Electronically traded e-mini NYMEX crude oil futures Mar-06 0.000418 Sep-06 0.000407 Apr-06 0.000374 Oct-06 0.000441 May-06 0.000370 Nov-06 0.000454 Jun-06 0.000365 Dec-06 0.000429 Jul-06 0.000356 Jan-07 0.000503 Aug-06 0.000352 Feb-07 0.000455 Panel C: Electronically traded ICE futures WTI crude oil futures Mar-06 0.000363 Sep-06 0.000283 Apr-06 0.000313 Oct-06 0.000308 May-06 0.000334 Nov-06 0.000306 Jun-06 0.000311 Dec-06 0.000314 Jul-06 0.000294 Jan-07 0.000373 Aug-06 0.000243 Feb-07 0.000342 Estimated daily average effective bid-ask spreads of floor traded regular-size NYMEX crude oil futures, electronically traded e-mini NYMEX crude oil futures, and electronically traded ICE WTI crude oil futures. Effective spreads are estimated during the open-outcry trading period using the Thompson and Waller (1988) spread estimator. 23

Table 2 Comparison of Average Daily Effective Bid-Ask Spreads Period 1 Period 2 p-value Floor Traded Regular Sized NYMEX Crude Oil Futures 0.000925 0.000980 0.228 Electronically Traded e-mini NYMEX Crude Oil Futures 0.000373 0.000449 <0.0001 Electronically Traded ICE Futures WTI Crude Oil Futures 0.000310 0.000321 0.0734 The table compares mean daily effective bid-ask spreads from the first period (March to August 2006) to the second period (September 2006 to February 2007) of the analysis. Effective spreads are compared for floor-traded regular-size NYMEX crude oil futures, electronically traded e-mini NYMEX crude oil futures, and electronically traded ICE WTI crude oil futures. 24

Table 3 Price Discovery Analysis of NYMEX and ICE Crude Oil Futures NYMEX electr NYMEX pit ICE WTI NYMEX e-mini Panel A: First period - March to August 2006 Mar-06 0.18 0.40 0.42 Apr-06 0.19 0.42 0.39 May-06 0.21 0.39 0.40 Jun-06 0.19 0.38 0.43 Jul-06 0.19 0.39 0.42 Aug-06 0.18 0.41 0.40 Panel B: Second period - September 2006 to February 2007 Sep-06 0.27 0.14 0.31 0.28 Oct-06 0.30 0.12 0.31 0.27 Nov-06 0.31 0.11 0.29 0.29 Dec-06 0.33 0.09 0.30 0.28 Jan-07 0.32 0.11 0.29 0.28 Feb-07 0.33 0.09 0.29 0.29 The table reports the mean information shares of Hasbrouck (1995) price discovery model using minute-by-minute trade prices of the nearby contracts during the regular open-outcry hours. Price discovery is compared for the NYMEX screen-based and floor regular-size contracts (NYMEX electr and NYMEX pit), the ICE regular-size contract (ICE WTI), and the NYMEX mini contract (NYMEX e-mini). 25

Table 4 Price Discovery Analysis of NYMEX Crude Oil Futures and the USO ETF from September 2006 to February 2007 NYMEX electr NYMEX pit NYMEX e-mini USO Sep-06 0.33 0.18 0.36 0.13 Oct-06 0.37 0.15 0.34 0.14 Nov-06 0.37 0.12 0.33 0.18 Dec-06 0.40 0.10 0.32 0.18 Jan-07 0.35 0.12 0.30 0.24 Feb-07 0.36 0.10 0.31 0.24 The table reports the mean information shares of Hasbrouck (1995) price discovery model using minute-by-minute trade prices of the nearby contracts during the regular open-outcry hours from September 2006 to February 2007. Price discovery is compared for the NYMEX screen-based and floor regular-size contracts (NYMEX electr and NYMEX pit), the NYMEX mini contract (NYMEX e-mini), and the United States Oil ETF (USO). 26

Table 5 Price Discovery Analysis of NYMEX and ICE Crude Oil Futures and the USO ETF from September 2006 to February 2007 NYMEX electr NYMEX pit WTI ICE NYMEX e-mini USO Sep-06 0.24 0.13 0.29 0.25 0.09 Oct-06 0.28 0.09 0.29 0.26 0.09 Nov-06 0.28 0.12 0.25 0.28 0.07 Dec-06 0.30 0.07 0.26 0.24 0.13 Jan-07 0.26 0.09 0.23 0.23 0.19 Feb-07 0.27 0.07 0.24 0.24 0.18 The table reports the mean information shares of Hasbrouck (1995) price discovery model using minute-by-minute trade prices of the nearby contracts during the regular open-outcry hours from September 2006 to February 2007. Price discovery is compared for the NYMEX screen-based and floor regular-size contracts (NYMEX electr and NYMEX pit), the ICE regular-size contract (ICE WTI), the NYMEX mini contract (NYMEX e-mini), and the United States Oil ETF (USO). 27