Introduction of Crude Oil Futures on the Shanghai International Energy Exchange: From a Market Quality Pespective¶
Background¶
In 2017, China surpassed the US to become the world's largest crude oil importer, importing over 8 million barrels a month. The main drivers behind the increase include China's industrial and population growth, along with a decline in domestic oil production. Further, China has increased its petroleum reserves and lowered trade barriers in the refining sector (Reuters and Bloomberg, 2018).
Producers and other oil-consuming industries typically hedge their oil price risk exposure with futures contracts. Currently, the main exchanges for oil contracts are located in the U.S. (Chicago Mercantile Exchange, CME) and Europe (Intercontinental Exchange Europe, ICE), with the major contracts being the U.S. West Texas Intermediate (WTI), Light Sweet Crude Oil (traded on CME) and Europe’s Brent crude (traded on ICE Europe).
On March 26, 2018, China launched its own crude oil futures contract on the Shanghai International Energy Exchange (IEE). Previously, trading of oil in Asia was limited to Tokyo (TOCOM), Singapore (SGX) and Dubai (DGX), however trading volumes and liquidity were not competitive relative to volume traded on CME and ICE. In addition to the IEE launch of the new oil futures, several adjustments were included in the contracts to account for freight and currency factors which distinguish it from other contracts traded on the CME and ICE (Reuters, 2018). Further, China's crude oil futures are based on seven grades of medium-sour crudes that are found in China and Arabian countries. By contrast, CME contracts are based on light sweet crude oil found in the U.S., Middle East and Far Eastern countries and ICE contracts are priced on North Sea oil.
The point of this analysis is to measure the market quality of China's first crude oil futures market accepting that the market is in its early stages. We find, after three months of trading, volumes are persistently increasing and the market's enthusiasm is rampant. Liquidity and volumes appear to be competing with those established in the West. The IEE also reported over 6,000 newly opened trading accounts with around 150 brokerages and international intermediaries like J.P. Morgan (Reuters, 2018).
import pandas as pd
from IPython.display import HTML
import datetime
from highcharts import Highstock
from highcharts import highcharts
from highcharts import Highchart
import csv
OPTIONS_PIE = {
'chart': {
'plotBackgroundColor': None,
'plotBorderWidth': None,
'plotShadow': False
},
'tooltip': {
'pointFormat': '{series.name}: <b>{point.percentage:.1f}%</b>'
},
}
OPTIONS_LINE = {
'tooltip': {'valueDecimals': 2, 'crosshairs': [True, True]},
'legend': {'enabled': True},
'chart': {'zoomType': 'x'},
}
OPTIONS_SPLINE = {
# 'tooltip': {'valueDecimals': 2, 'crosshairs': [True, True]},
'chart': {'zoomType': 'x'},
'rangeSelector': {'enabled': False},
'navigator': {'enabled': False},
'scrollbar': {'enabled': False}
}
Contract Specifications¶
The following table summarises the contract specifications of the three futures contracts. All contracts have equal contract sizes and delivery methods. Further, differences include the denomination in Yuan, a higher tick size and fewer trading hours in Shanghai. CME and ICE futures are denominated in USD and trading is open 23 hours per day.
import pandas as pd
import csv
# 1) Read CSV
Specreader = pd.read_csv('Shanghai_specs.csv', index_col=[0])
Specreader
Data¶
In this case study, we use data obtained via the Market Quality Dashboard, which generates summary metrics derived from underlying Thomson Reuters Tick History Data (TRTH). TRTH reports timestamps to the nearest millisecond for trades and quotes on the first level of the order book. We sample data for three months following the launch on the most actively traded contracts, given multiple contract expiries are available for trade at any given time.
Descriptive Statistics¶
Our first step compares the fundamental trading statistics for each crude oil futures market. The table below reports descriptive statistics for absolute quoted spreads (in ticks), trade value (in RMB), trade count, quote-to-trade ratio and average daily return.
Quoted spreads, which measure trading costs per contract traded, indicate that Shanghai crude oil futures typically range between 1 and 2 ticks, and are slightly higher on average relative to CME and ICE, with 1.77 ticks and 1.42 ticks, respectively.
Although Shanghai futures trade volumes are below those of the CME and ICE, we find that Shanghai futures experience the highest average return and lowest volatility (0.34% and 1.42%, respectively). CME, by contrast, shows the lowest average return with 0.03%. As we expected, CME has the highest trading volume, with a daily average of 657,788 contracts traded. ICE has about half that, with 326,854 contracts traded per day. Although Shanghai started trading only three months earlier, the average daily value reached half of ICE's trading within that period.
The trading values are notably different between the new and established contracts. We find that the average daily trading value is more than RMB168mn in CME and over RMB302mn in ICE. By contrast, Shanghai shows a relatively low daily average trade count, indicating that trades, in general, are larger than in the U.S. and Europe. CME and ICE show 211,369 and 141,376 trades per day on average, while Shanghai has 17,131.
import pandas as pd
import csv
table = pd.read_csv('Shanghai_descr3.csv', index_col=[0]).fillna("")
table
Variance¶
An important aspect of each market is its price efficiency, which can be measured by the variance ratio. The variance ratio measures the linearity of the variance of mid-point price returns in terms of a particular data interval. In this case study, we use intervals of 30 and 5 minutes.
In an efficient market, the variance of mid-point price returns over 30 minutes is expected to be close to six times the variance of returns over five minutes, which is equivalent to a 1:t relation. We then deduct 1 from our ratio, with the results that a variance ratio of approximately 0 indicates efficient prices.
The chart below shows that all three price series have variance ratios of around 0.2 and oscillate rather close together. CME shows ratios closer to 0, indicating more efficient prices. However, the graph shows that the three markets have comparable levels of variance and, therefore, comparable price efficiency.
# Variance Ratio
import pandas as pd
variance = pd.read_csv('shanghai_futures.csv')
variance['Date'] = pd.to_datetime(variance['Date'], format='%d/%m/%Y')
variance = variance[['Date',
'Shanghai Crude Oil futures (IEE) - Variance Ratio',
'Brent Crude futures (ICE) - Variance Ratio',
'Light Sweet Crude Oil (CME) - Variance Ratio']]
VARIANCE_CHART = Highstock()
options = OPTIONS_SPLINE
options['title'] = {'text': 'Variance Ratio'}
options['yAxis'] = {'opposite': False, 'title': { 'text': 'Variance Ratio' } }
options['xAxis'] = {'type': 'datetime'}
options['legend'] = {'enabled': True}
VARIANCE_CHART.set_dict_options(options)
VARIANCE_CHART.add_data_set(variance.iloc[:,[0,1]].values.tolist(), name='Shanghai Crude Oil futures (IEE)')
VARIANCE_CHART.add_data_set(variance.iloc[:,[0,2]].values.tolist(), name='Brent Crude futures (ICE)')
VARIANCE_CHART.add_data_set(variance.iloc[:,[0,3]].values.tolist(), name='Light Sweet Crude Oil (CME)')
VARIANCE_CHART
Quote-to-Trade Ratio¶
A useful measure for analysing the efficiency of a marketplace is its quote-to-trade ratio (QTT). This ratio measures the relative quoting activity with respect to the actual number of trades completed. A higher quote-to-trade ratio has been used as a proxy for high-frequency or algorithmic trading activity. The graphs below reveal that CME relative quoting activity is about 10 times higher than on IEE and ICE. Specifically, CME futures have an average of 12.6 quotes per trade in our sample period, while Shanghai and ICE contracts witness QTT ratios of 1.8 and 1.5, respectively, and are therefore more comparable.
The main reason for the discrepancy across markets is the regulation of high-frequency trading (HFT) in each region. While HFT firms are prominent players in U.S. markets, taxes and other regulatory controls are in place across Europe to curb the dominance of such traders in the market. Regulation in China goes even further, banning day trading entirely, leading to lower QTT ratios. Academic literature reports that high HFT activity accompanies elevated levels of so-called 'quote stuffing' which is thought to increase the probability of market manipulation (Egginton et al., 2016).
# QTT
import pandas as pd
qtt = pd.read_csv('shanghai_futures.csv')
qtt['Date'] = pd.to_datetime(qtt['Date'], format='%d/%m/%Y')
qtt = qtt[['Date',
'Shanghai Crude Oil futures (IEE) - Quote-to-Trade Ratio',
'Brent Crude futures (ICE) - Quote-to-Trade Ratio',
'Light Sweet Crude Oil (CME) - Quote-to-Trade Ratio']]
QTTRATIO_CHART = Highstock()
options = OPTIONS_SPLINE
options['title'] = {'text': 'Quote-to-Trade Ratio'}
options['yAxis'] = {'opposite': False, 'title': { 'text': 'Quote-to-Trade Ratio' } }
options['xAxis'] = {'type': 'datetime'}
options['legend'] = {'enabled': True}
QTTRATIO_CHART.set_dict_options(options)
QTTRATIO_CHART.add_data_set(qtt.iloc[:,[0,1]].values.tolist(), name='Shanghai Crude Oil futures (IEE)')
QTTRATIO_CHART.add_data_set(qtt.iloc[:,[0,2]].values.tolist(), name='Brent Crude futures (ICE)')
QTTRATIO_CHART.add_data_set(qtt.iloc[:,[0,3]].values.tolist(), name='Light Sweet Crude Oil (CME)')
QTTRATIO_CHART
Price Discovery¶
Finally, we assess the relative importance of the new oil contract in China, with respect to its counterparts in the West, by measuring the extent to which it leads or sets prices in the marketplace. We do this by tracking its price discovery score on an intraday basis. We implement a price discovery measure that combines two widely used approaches: information share (IS) and component share (CS). Both IS and CS decompose price changes into two components: i) permanent component (innovations in the fundamental value) and a ii) temporary component (noise). Prior studies use IS and CS to measure which market is the first to reflect new information (i.e., the permanent component of price changes). We sample data at one-second intervals and model the necessary vector error correction model with 200 lags as per best practice in the academic literature (see Putniņš, 2013).
The table below shows the average price discovery shares for the three month sample period for each oil contract. Values over 0.5 indicate the leading market. The results show that CME and ICE both are leading IEE in terms of speed in the impounding of new information. For the period during which all three markets are open, we report that Shanghai prices lead CME in terms of reflecting information on average 7.5%, however, when compared with ICE, this increases to approximately 30%.
import pandas as pd
import csv
table = pd.read_csv('shanghai_pd.csv', index_col=[0]).fillna("")
table
The following charts depict the ILS price discovery shares on a daily basis for Shanghai oil futures versus CME oil futures as well as ICE oil futures. It shows that CME persistently dominates Shanghai in terms of price discovery during the period. However, this is not the case with respect to ICE. On several occasions, Shanghai appears to dominate, with instances of leadership reaching in excess of 60%. The comparison between ICE and IEE is perhaps more meaningful given the quality of oil which underpins both contracts is similar, vis-a-vis WTI.
# Price Discovery CME IEE
import pandas as pd
pricedisc = pd.read_csv('shanghai_pd_weekly.csv')
pricedisc['Date'] = pd.to_datetime(pricedisc['Date'], format='%d/%m/%Y')
pricedisc = pricedisc[['Date',
'ILS (CME)',
'ILS (IEE)',
'ILS (ICE)',
'ILS (IEE)',
'ILS (CME)']]
PRICEDISC_CHART = Highstock()
options = OPTIONS_SPLINE
options['title'] = {'text': 'Price Discovery Shares of Shanghai vs. CME'}
options['yAxis'] = {'opposite': False, 'title': { 'text': 'Price Discovery Share Shanghai' } }
options['xAxis'] = {'type': 'datetime'}
options['legend'] = {'enabled': True}
PRICEDISC_CHART.set_dict_options(options)
#PRICEDISC_CHART.add_data_set(pricedisc.iloc[:,[0,1]].values.tolist(), name='Information Share (CME)')
#PRICEDISC_CHART.add_data_set(pricedisc.iloc[:,[0,2]].values.tolist(), name='Information Share (IEE)')
#PRICEDISC_CHART.add_data_set(pricedisc.iloc[:,[0,3]].values.tolist(), name='Component Share (CME)')
#PRICEDISC_CHART.add_data_set(pricedisc.iloc[:,[0,4]].values.tolist(), name='Component Share (IEE)')
#PRICEDISC_CHART.add_data_set(pricedisc.iloc[:,[0,1]].values.tolist(), name='Information Leadership Share (CME)')
PRICEDISC_CHART.add_data_set(pricedisc.iloc[:,[0,2]].values.tolist(), name='Information Leadership Share (IEE vs. CME)')
#PRICEDISC_CHART.add_data_set(pricedisc.iloc[:,[0,3]].values.tolist(), name='Information Leadership Share (CME)')
PRICEDISC_CHART.add_data_set(pricedisc.iloc[:,[0,4]].values.tolist(), name='Information Leadership Share (IEE vs. ICE)')
#PRICEDISC_CHART.add_data_set(pricedisc.iloc[:,[0,5]].values.tolist(), name='Information Leadership Share (CME vs. ICE)')
PRICEDISC_CHART
Conclusion¶
In this case study, we compare the newly established crude oil futures contract on the Shanghai International Energy Exchange with the incumbent crude oil contracts traded on CME and ICE. This contract is the first of its kind in China and witnessed more interest than previous Asian crude oil contracts traded in Tokyo and Singapore. Soon after its introduction, crude oil futures trading in China reached a quarter of the traded value on CME and half of the traded value on ICE with an increasing trend.
Surprisingly, we find that crude oil trading on IEE has only slightly higher quoted spreads compared to CME and ICE, while reporting similar price variance ratios compared to its US and European counterparts. However, the CME crude oil futures show slightly better price efficiency. Corresponding variance ratio measures support the conclusion that prices are as efficient as the long-established markets. Also, while CME and ICE mostly lead IEE in terms of price discovery their are some encouraging signs for Shanghai's crude oil futures despite their short trading history.
All in all, we see Shanghai's introduction of crude oil futures as successful given their comparable quotes, strong and increasing volumes, as well as relatively efficient prices.
References¶
Bloomberg: 'Forget Iran. The Real Oil Action Is in China', May 9th, 2018, LINK
Egginton, J.F., Van Ness, B.F. and Van Ness, R.A., 2016. Quote stuffing. Financial Management, 45(3), pp.583-608.
Frijns, B., Gilbert, A. and Tourani-Rad, A., 2015. The determinants of price discovery: Evidence from US-Canadian cross-listed shares. Journal of Banking & Finance, 59, pp.457-468.
Reuters: 'Shanghai crude oil futures look successful, but issues lurk: Russell', May 2nd, 2018, LINK
The Economist: 'China wants to reshape the global oil market', March 28th, 2018, LINK