Korean J Financ Stud Search

CLOSE

 Korean J Financ Stud > Volume 50(2); 2021 > Article
원유관련 ETP의 투자성과에 관한 연구*

### Abstract

The unprecedented COVID-19 pandemic at the beginning of 2020 jeopardized the entire world. Meanwhile, the Russia-Saudi Arabia oil war led a crude oil market going out of the frying pan into the fire. This research has two main purposes. First, we investigate structures of the crude oil related exchange-traded products (ETPs) in terms of operation and cost. Second, we analyze the correlation and investment performance of the crude oil related ETPs and West Texas Intermediate spot market. The major findings are as follows: (1) there is a positive correlation between the crude oil spot and the crude oil producing firm, (2) the investment performance of the crude oil related ETPs is inferior to that of the crude oil spot due to the rollover cost, (3) the investment performance of the crude oil related leverage ETNs (Exchange-Traded Notes) or inverse ETPs has deteriorated because their managing structure tracks the daily return which leads to the compounding effects. These empirical results show the characteristics of the gain and loss of the crude oil-related ETPs, which enhance conceptual understanding and offer implications to policymakers and authorities for the efficiency of the alternative investment strategy.

### 요약

2020년 초부터 COVID-19로 원유 수요 감소 및 산유국 간 원유가격 분쟁으로 인해 국제유가는 크게 하락하였고 시장 변동성이 확대되었다. 그리하여 원유 관련 상장지수상품(ETP: Exchange-Traded Products)의 손익 및 비용 구조를 이해하지 못한 많은 투자자들이 손실을 경험하였다. 본 연구는 한국에 상장된 원유 관련 ETP의 운용방식 및 비용구조를 조사하고 각 ETP와 서부텍사스중질유(WTI) 시장과의 상관성 및 투자성과를 분석하였다. 또한, 레버리지 ETN의 운용구조와 투자성과를 함께 살펴보았다. 실증분석 결과 원유생산기업 ETF와 WTI 및 선물 ETF 사이에는 통계적으로 유의한 양(+)의 상관관계가 확인되었다. 투자성과와 관련하여 원유생산기업 ETF가 WTI 및 선물 ETF 대비 열위하다는 결과를 도출하였다. 본 연구는 연기금 및 일반투자자의 원유 관련 ETP 손익의 특징을 활용한 투자전략을 제언하고, 정책당국에 관련된 함의를 제시할 수 있다.

### 1. Introduction

Crude oil price significantly fell due to collapse in demand caused by the unprecedented COVID-19 outbreak with the Russia-Saudi Arabia oil price war in the early 2020. Benchmark U.S. Crude oil WTI Cushing Spot dived into negative territory on Monday May 11, 2020; the U.S. West Texas Intermediate (WTI) was traded at $25.78 per barrel, a decline of 58% since May 13, 2019, a year ago at$61.04 per barrel. In the first half of 2020, the crude oil prices largely fell and a level of its volatility expanded. These led to an unusual phenomenon that a level of disparity for crude oil related leverage Exchanged-Traded Notes (ETNs) in the Korea Exchange (KRX) was magnified from 100% up to 300%. The investors alleged that crude oil related Exchanged-Traded Products (ETPs) could not trace the price of crude oil spot reasonably due to a rollover cost.
##### <Table 4>
Summary Statistics
WTIt is the daily yield of crude oil spot on t-day, and ETPat, ETPbt, and ETPct are open price return of KB Crude Company ETF, TIGER Crude Oil Futures, and KODEX Crude Oil Futures ETF on t-day, respectively. The unit of average, maximum, and minimum is percent (%). The sampling period is from December 28, 2016 to September 22, 2020.
WTIt ETPat ETPbt ETPct
Avg. 0.033 0.016 -0.053 -0.147
Max. 28.257 10.256 13.718 24.801
Min. -29.208 -15.229 -17.414 -29.708
S.D. 3.629 1.934 1.964 2.582
Skewness 0.493 -0.482 -0.741 -1.278
Kurtosis 27.843 11.557 17.214 35.284
<Table 5> displays the result of the correlation analysis between the daily yield of crude oil spot on t-1 (WTIt-1) and the open price return of the KB Crude Company ETF (ETPat), TIGER Crude Oil Futures ETF (ETPbt), and KODEX Crude Oil Futures ETF (ETPct) on t-day. It was confirmed that WTIt-1 had the high correlation with ETPbt, ETPct of 0.629 and 0.672, respectively. In addition, it was discovered that WTIt-1 had the correlation of 0.389 with ETPat, where as ETPat was correlated with ETPbt, ETPct of 0.564 and 0.460, successively. Through this correlation analysis, it is possible to examine a certain level of the tendency among variables, but additional statistical estimation is essential since the mutual influence of the variables cannot be corroborated.
##### <Table 5>
Correlation Analysis
WTIt is the daily yield of crude oil spot on t-day, and ETPat, ETPbt, and ETPct are open price return of KB Crude Company ETF, TIGER Crude Oil Futures, and KODEX Crude Oil Futures ETF on t-day, respectively. The unit of average, maximum, and minimum is percent (%). The sampling period is from December 28, 2016 to September 22, 2020.
WTIt-1 ETPat ETPbt ETPct
WTIt-1 1 0.389 0.629 0.672
ETPat 1 0.564 0.460
ETPbt 1 0.873
ETPct 1
The autocorrelation coefficients of the four variables were listed in <Table 6>. If the variables have high autocorrelation, the residuals estimated in the regression model will also show autocorrelation, which implies that the reliability of t-statistics, significance level and R2 would be lowered, even if a consistent estimate is obtained in the regression model. The most of the four variables showed low autocorrelation as shown in <Table 6>. There was a tendency of low autocorrelation when using yield data instead of price data, and all four variables are based on the yield data.
##### <Table 6>
Analysis of Autocorrelation
WTIt is the daily yield of crude oil spot on t-day, and ETPat, ETPbt, and ETPct are open price return of KB Crude Company ETF, TIGER Crude Oil Futures, and KODEX Crude Oil Futures ETF on t-day, respectively. The sampling period is from December 28, 2016 to September 22, 2020.
Period AC (Autocorrelation) PAC (Partial Autocorrelation)

WTIt ETPat ETPbt ETPct WTIt ETPat ETPbt ETPct
1 0.070 0.056 -0.062 0.019 0.070 0.056 -0.062 0.019
2 -0.136 -0.046 0.018 0.070 -0.142 -0.050 0.014 0.070
3 -0.098 0.045 0.025 0.081 -0.079 0.051 0.027 0.079
4 0.118 0.013 0.030 0.040 0.115 0.004 0.034 0.033
5 0.279 0.048 0.070 0.092 0.249 0.052 0.074 0.081
6 -0.038 -0.007 -0.034 -0.063 -0.056 -0.015 -0.027 -0.077
7 -0.042 -0.018 0.085 0.113 0.045 -0.012 0.078 0.100
8 -0.047 0.031 0.062 0.006 -0.031 0.027 0.070 -0.003
9 0.067 0.063 0.045 0.034 0.011 0.059 0.049 0.027
10 -0.013 0.052 0.032 -0.059 -0.096 0.047 0.030 -0.082
11 -0.102 -0.015 0.008 -0.016 -0.076 -0.016 0.007 -0.012
12 -0.078 -0.011 -0.016 -0.080 -0.078 -0.009 -0.035 -0.104
13 0.045 -0.015 0.014 -0.011 0.048 -0.025 0.002 0.020
14 0.093 0.017 0.036 0.051 0.056 0.014 0.027 0.052
In <Table 7>, the Augmented Dicky-Fuller test (ADF test) examined whether the variables used in the regression analysis follow a random walk. A unit root exists if the variables satisfy the random walk, and even if there is no correlation, spurious regression can be appeared in which the regression analysis result is significant. The ADF test in general assumes the existence of a unit root as a null hypothesis, which was rejected at the significance level of 1% or less for all four variables; therefore, WTIt, ETPat, ETPbt and ETPct did not have a unit root.
##### <Table 7>
Unit Root Test
WTIt is the daily yield of crude oil spot on t-day, and ETPat, ETPbt, and ETPct are open price return of KB Crude Company ETF, TIGER Crude Oil Futures, and KODEX Crude Oil Futures ETF on t-day, respectively. The lowest SIC value in the maximum of 21 lagged values was standardized for the time lag of the ADF test. The sampling period is from December 28, 2016 to September 22, 2020.

WTIt ETPat ETPbt ETPct
Lagged Value 4 0 0 0
t-Statistic -10.485 -28.909 -32.587 -30.052
P-Value <0.001 <0.001 <0.001 <0.001
Dozens of the Crude Company ETFs are listed and traded on the US stock market, but ETPa by KB Asset Management is the only Crude Company ETF listed on the KRX. Usually, the Crude Company ETF invest in crude oil-related producers, and it is estimated that there will be a correlation with crude oil spot. This study statistically analyzed the correlation between the market yield of the only Crude Company ETF listed and traded on the KRX, and the daily rate of return on crude oil spot. The ETPa traces the S&P Oil & Gas Exploration & Production Select Industry Index consisting of 74 oil and gas exploration and production companies listed in the United States. Due to the difference in trading hours between the crude oil spot market and the KRX, the daily yield on the t-1 will affect the open rate of return on the ETPa on t-day. The daily rate of return on crude oil spot at t-1 was calculated by dividing the spot price of crude oil on t-1 by that on t-2, and then, subtracting one. In addition, the ETPa’s open rate of return on t-day was calculated by dividing the opening price on t-day by the closing price of t-1, and then subtracting one. Apart from looking into the correlation between the daily rate of return on crude oil spot at t-1 and the open return rate of the ETPa by applying the VAR model, the two other ETFs listed on the KRX, ETPb and ETPc at t-1 were added to the variables in order to investigate a correlation with the daily return rate of crude oil spot.
The Schwarz Criterion (SC) was applied to determine the time lag of the VAR model estimation in <Table 8>. The time lag of Phase one, two, three, and four in the SC values were 16.038, 15.987, 15.945, and 15.957, respectively, which demonstrated that the minimum SC value was found on the time lag of Phase 3; thus, it was eventually estimated based on the value of m in <Table 8>. As a result of the VAR model estimation, the daily yield of crude oil spot on t-1 had a significant correlation of 1% with the open price return of the ETPb and ETPc on t-day. The daily yield of crude oil spot on t-1 correlated with each of the open return rate on the ETPb and ETPc at the significant level of 1%. Each of t-statistics were 28.166 and 32.755, which were relatively high. In addition, the daily yield of crude oil spot on t-1 showed the coefficient value of 0.219** with a significance level of 1% to the open rate of return on the ETPa on t-day, and the t-statistic was 13.423. It is possible to invest in the Crude Company ETF as well as crude oil futures ETFs for investors who expect the crude oil price rises. Therefore, an interpretation of these results support the Hypothesis I (There is correlation between the Crude Company ETFs and crude oil spot; however, a level of its correlation is lower than that between crude oil futures ETFs and crude oil spot).
##### <Table 8>
Analysis of Unrestricted VAR
WTIt is the daily yield of crude oil spot on t-day, and ETPat, ETPbt, and ETPct are open price return of KB Crude Company ETF, TIGER Crude Oil Futures, and KODEX Crude Oil Futures ETF on t-day, respectively. The lowest SIC value in the maximum of 21 lagged values was standardized for the time lag of the ADF test. ** means statistically significant at the level of 1%. The sampling period is from December 28, 2016 to September 22, 2020.
WTIt ETPat ETPbt ETPct
WTIt-1 Coefficient 0.055 0.219 ** 0.376 ** 0.529 **
S.D. 0.033 0.016 0.013 0.016
t-Statistic 1.660 13.423 28.166 32.785
WTIt-2 Coefficient -0.143 ** -0.062 ** -0.048 ** 0.087 **
S.D. 0.046 0.023 0.019 0.023
t-Statistic -3.070 -2.71 -2.58 3.827
WTIt-3 Coefficient -0.128 ** 0.079 ** 0.103 ** 0.131 **
S.D. 0.047 0.023 0.019 0.023
t-Statistic -2.740 3.447 5.489 5.748

F-statistic: 0.976
R2: 0.0022
Before performing variance decomposition and impulse response analysis using the VAR model, it is important to note that certain restrictions are required on the interaction between variables at the same time point. If there are no restrictions, a problem arises that the structural model used in the reduced form1) for estimation cannot be determined. If there is a clear theory about the relationship between variables, it can be used as a constraint; however, there is no clear theory about the correlation among crude oil spot, crude oil futures ETF and the Crude Company ETF. Hence, exogeneity was estimated, and the Cholesky ordering method by using the level of this exogeneity was applied. The degree of exogeneity between variables was identified by Granger (1969, 1980) causality test, and the result are shown in <Table 9>.
##### <Table 9>
Granger Causality Test
WTIt is the daily yield of crude oil spot on t-day, and ETPat, ETPbt, and ETPct are open price return of KB Crude Company ETF, TIGER Crude Oil Futures, and KODEX Crude Oil Futures ETF on t-day, respectively. The sampling period is from December 28, 2016 to September 22, 2020.
Null Hypothesis F-Stat α
ETPat does not cause WTIt 4.946 0.002
WTIt does not cause ETPat 65.546 0.000
ETPct does not cause WTIt 0.908 0.437
WTIt does not cause ETPct 301.059 0.000
ETPbt does not cause WTIt 3.623 0.013
WTIt does not cause ETPbt 258.307 0.000
ETPct does not cause ETPat 2.057 0.104
ETPat does not cause ETPct 5.350 0.001
ETPbt does not cause ETPat 0.625 0.599
ETPat does not cause ETPbt 0.918 0.432
ETPbt does not cause ETPct 3.176 0.024
ETPct does not cause ETPbt 1.963 0.118
WTIt was causal to all other variables at a significance level of 1% or less as a result of the Granger causality test in <Table 9>. Contrarily, the degree to which other variables are causal to WTIt is lower or the other variables were not causal to WTIt. According to the Granger causality test, ETPat was causal of ETPct; however, ETPct was not causative of ETPat. The Granger causality was found in ETPbt against ETPct at a significant level of below 5%, but ETPct did not cause ETPbt. Although ETPat and ETPbt did not have a mutual Granger causal relationship, the degree of exogeneity was measured by using the significance probability level. Finally, the variance decomposition and impulse reaction analysis were performed by implementing the Cholesky ordering as shown in <Table 10> and <Table 11>, after assuming that the exogeneity was high in the order of WTIt, ETPat, ETPbt, and ETPct.
<Table 10> portraits the variance decomposition result for the forecast error variance of the VAR model. Each number represents how much weight the upper variables influence the forecast error variance in the left variables in percentage. In the analysis of the variance decomposition, it is worth noting that 96.943% of the forecast error variance of WTIt was explained by itself at the Period 5, and the weight of WTIt on the top contribute to each of the variables on the left. The forecast error variance of ETPat at the Period 5 was elucidated by 82.123% itself, and by WTIt with a constant weight of 16.904%. At the Period 5, 41.823% delineated the forecast error variance of ETPbt by itself, and 47.403% of WTIt showed the highest explanatory power to ETPbt. The forecast error variance of ETPct was explicated by itself and WTIt of 19.302%, 51.136%, consecutively. That is, the daily rate of return on crude oil spot deciphered the forecast error variances for the daily rate of return on the crude oil futures ETFs and Crude Company ETF at a consistent explanatory power. The variance decomposition result expounded the Hypothesis I (There is correlation between the Crude Company ETFs and crude oil spot; however, a level of its correlation is lower than that between crude oil futures ETFs and crude oil spot) is well supported.
##### <Table 10>
Analysis of Variance Decomposition
WTIt is the daily yield of crude oil spot on t-day, and ETPat, ETPbt, and ETPct are open price return of KB Crude Company ETF, TIGER Crude Oil Futures, and KODEX Crude Oil Futures ETF on t-day, respectively. The sampling period is from December 28, 2016 to September 22, 2020.
Period WTIt ETPat ETPbt ETPct
WTIt 2 99.589 0.047 0.168 0.196
5 96.943 1.811 0.806 0.440
10 96.852 1.897 0.810 0.441
ETPat 2 16.196 83.008 0.006 0.789
5 16.904 82.123 0.167 0.805
10 16.916 82.102 0.170 0.812
ETPbt 2 46.631 10.054 43.085 0.230
5 47.403 10.123 41.823 0.651
10 47.390 10.118 41.806 0.686
ETPct 2 52.607 4.967 24.535 17.891
5 51.136 5.986 23.576 19.302
10 51.055 5.971 23.531 19.444
The reaction of the open return rate on both the Crude Company ETF and crude oil futures ETF against the impulse of crude oil spot was analyzed in <Table 11>. It was confirmed how ETPat, ETPbt, and ETPct react when an impact of one standard error is given to the daily yield of crude oil spot (WTIt). The positive coefficient values of ETPat, ETPbt, and ETPct were large at the Period 2 in the impulse reaction analysis on the daily yield of crude oil spot in <Table 11> due to the time lag. The reason for the time lag is the daily return rate of crude oil spot on t-1 is calculated at the dawn of t-day in Korea. In other words, the effect in the Period 2 was greater than that in the Period 1 after the shock. The response of the two other crude oil futures ETFs, ETPbt, and ETPct, against WTIt are 1.309, 1.826, respectively, which is bigger than that of the open return for ETPat against WTIt, recorded 0.778. This result confirmed the Hypothesis I (There is correlation between the Crude Company ETFs and crude oil spot; however, a level of its correlation is lower than that between crude oil futures ETFs and crude oil spot).
##### <Table 11>
Impulse Response Analysis
WTIt is the daily yield of crude oil spot on t-day, and ETPat, ETPbt, and ETPct are open price return of KB Crude Company ETF, TIGER Crude Oil Futures, and KODEX Crude Oil Futures ETF on t-day, respectively. The sampling period is from December 28, 2016 to September 22, 2020.
Period WTIt

ETPat ETPbt ETPct
1 0.032 0.230 0.253
2 0.778 1.309 1.826
3 -0.188 -0.258 0.053
4 -0.033 0.152 0.228
5 0.015 -0.017 0.004
6 -0.025 0.009 0.043
7 0.001 0.026 0.043
8 -0.004 0.005 0.036
9 0.000 0.011 0.027
10 -0.003 -0.002 -0.002
The impulse response analysis of <Table 11> are illustrated by the graph of <Figure 1>. Each of graphs demonstrated that ETPat, ETPbt, ETPct have a large response for the Period 2 when the impact of one standard error is given to the daily yield of crude oil spot, WTIt.
##### <Figure 1>
Impulse Response Analysis Graph
WTIt is the daily yield of crude oil spot on t-day, and ETPat, ETPbt, and ETPct are open price return of KB Crude Company ETF, TIGER Crude Oil Futures, and KODEX Crude Oil Futures ETF on t-day, respectively. The sampling period is from December 28, 2016 to September 22, 2020.

### 4.2 Comparison of Investment Performance of Crude Company ETF and Crude oil Futures ETF

Not only crude oil futures ETFs operated as crude oil futures but also Crude Company ETF investing in crude oil production companies had a significant correlation with crude oil spot as proven in the 4.1 previously. In Korea, the only one Crude Company ETF is listed and traded, but dozens of crude oil producing company ETFs are traded in the US, dividing into the three sectors: upstream, midstream, and downstream. The upstream refers to the exploration and production of oil and natural gas, while the midstream is the production of everything needed to transport and store before refining. The downstream is the field of converting crude oil and natural gas into finished products. The ETPa listed in Korea is based on the index2) composed of the US upstream companies. An increase in investment efficiency is expected by reducing investment costs owing to the fact that the Crude Company ETF invests in the stock market. The Crude Company ETFs do not incur the rollover cost, which is a different aspect compared to the crude oil futures ETFs. The Crude Company ETFs listed in the US could expect dividend income which increases the ETF base price, contrarily, the E of Korea does not generate dividend income due to swap costs. It was confirmed in <Table 12> whether the Crude Company ETFs, whose correlation with crude oil spot was confirmed, had higher risk-adjusted investment performance than the crude oil futures ETFs. The periodic yield of WTI, ETPa, ETPb, ETPc recorded the loss of -26.75%, -75.29%, -57.90%, and -69.61%, consecutively during the sampling term from December 28, 2016 to September 22, 2020, and all of the periodic return rate of the crude oil-related ETFs were lower than that of crude oil spot. The ETPa, especially, had the biggest loss of -75.29% among all others. It was interpreted that when oil prices decline, profitability of crude oil producers deteriorates. This led to a financial risk aggravates the company’s losses, which eventually caused a decrease in stock price. Furthermore, the periodic yield of two other crude oil futures ETFs during the same period of time put on record of the huge losses, -57.90% and -69.61%, which was larger than the loss of crude oil spot periodic return. This was conveyed as due to investment costs such as rollover costs, fund management fees, and currency hedging costs incurred in the process of operating crude oil futures ETFs. In the international financial market, crude oil spot prices use nearby month prices of crude oil futures. In the course of a sharp decline in oil prices, contango tends to expand as the price of nearby month futures falls significantly compared to the deferred month futures.
##### <Table 12>
Comparison of Investment Performance of Crude Oil-Related ETFs
WTI is crude oil spot, ETPa, ETPb, and ETPc are KB Crude Oil Enterprise ETF, KODEX Crude Oil Futures ETF and TIGER Crude Oil Futures ETF, respectively.
Panel A: Dec 28, 2016 ~ Sep 22, 2020
WTI ETPa ETPb ETPc
Risk-free AR 7.51% -31.54% -15.70% -21.09%
Volatility 57.61% 36.49% 39.16% 46.38%
Sharpe index 0.130 -0.864 -0.401 -0.455
Periodic Return -26.75% -75.29% -57.90% -69.61%

Panel B: Dec 28, 2016 ~ Oct 03, 2018
WTI ETPa ETPb ETPc

Risk-free AR 21.92% 2.80% 14.69% 19.20%
Volatility 24.82% 22.45% 20.79% 23.44%
Sharpe index 0.883 0.125 0.706 0.819
Periodic Return 41.34% 1.88% 25.32% 34.39%

Panel C: Oct 4, 2018 ~ Sep 22, 2020

WTI ETPa ETPb ETPc

Risk-free AR -5.41% -62.35% -42.96% -57.24%
Volatility 75.82% 45.48% 50.18% 59.85%
Sharpe index -0.071 -1.371 -0.856 -0.956
Periodic Return -46.72% -76.40% -66.58% -77.50%
The risk-free abnormal return of crude oil spot is 7.51%, which was higher than that of crude oil-related ETFs; -31.54%, -15.70%, -21.09%. The volatility of crude oil spot was 57.61%, which was also higher than that of crude oil-related ETFs; 36.49%, 39.16%, 46.38% as well. Thus, it was necessary to compare the Sharpe index, which measures a risk-adjusted investment performance. The Sharpe index of crude oil spot was 0.130, which was the only positive value, whilst the Sharpe index for the remaining crude oil-related ETFs was -0.864, -0.401 and -0.455, respectively. The Sharpe index of the ETPa was exceptionally low out of all other sample data. This result does not support the Hypothesis II (The investment performance of Crude Company ETF is higher than that of ETFs operated with crude oil futures).
The whole sample, Panel A in <Table 12>, dated from December 28, 2016 to September 22, 2020 was divided into the two different time periods in order to verify the robustness of the result derived previously; the former sample, Panel B, was used as an estimated sample, and the latter sample, Panel C, was used as a verification sample to compare with the result of the whole sample.
The estimated sampling period was set from December 28, 2016 to October 3, 2018. The reason is that the spot price of crude oil peaked at $76.41 per barrel on October 3, 2018 during the entire sampling period as shown in <Figure 2>, afterwards the oil price has declined. The verification sample period is from October 4, 2018 to September 22, 2020. Another practical benefit of dividing the sample is that it is possible to check whether the result of the entire sampling period is due to the sharp fluctuation of the oil prices in the first half of 2020, which is included in the verification sample period. ##### <Figure 2> Crude Oil Spot Price Trend The sampling period is from December 28, 2016 to September 22, 2020. As of October 3, 2018, when the peak during the same period was recorded, additional analysis was performed by dividing into estimated and verified samples. The crude oil prices showed a solid rise during the estimated sample period (Dec 28, 2016 ~ Oct 3, 2018) of Panel B in <Table 12>. In the same period when the spot price of crude oil increased steadily, the Sharpe index of the ETPa was 0.125, which was significantly lower than that of the crude oil spot, 0.883, and the two crude oil futures ETFs, 0.706, 0.819. This result does not support the Hypothesis II (The investment performance of Crude Company ETF is higher than that of ETFs operated with crude oil futures). The nearby month price of crude oil futures used as the spot prices of crude oil rise, causing the backwardation in which the nearby month price is higher than the deferred month price. As rollover costs are eliminated and rollover revenues might be generated in the situation of the backwardation; thus, the crude oil futures ETF’s performance can be relatively improved. As a result, there was no significant difference between the Sharpe index of crude oil spot (0.883), and of crude oil futures ETFs (0.706, 0.819), which was dissimilar to the whole sample period. For this reason, the difference between the Sharpe index of the ETPa (0.125), and of Crude Oil Futures ETFs (0.706, 0.819) during the same period was enlarged in comparison to the whole sample period (-0.864 vs. -0.401, -0.455) and the verification sample period when the spot price of crude oil declined (-1.371 vs. -0.856, -0.956). The verification sample period (Oct 4, 2018 ~ Sep 22, 2020) of Panel C in <Table 12> was the time when oil price plunged, unlikely the estimated sample period of Panel B. In particular, the situation was contrary to Panel B, including the first half of 2020 when oil prices plummeted and volatility escalated. During the same period, the Sharpe index of the Crude Company ETF (-1.371) was lower than that of the two crude oil futures ETFs (-0.856, -0.956), but the difference was smaller than that of Panel B where oil prices rose. Moreover, the variance between the Sharpe index of the crude oil spot (-0.071), and of crude oil futures ETFs (-0.856, -0.956) was widened more than that of Panel B when the oil price increased (0.883 vs. 0.706, 0.819). The reason could be found in the verification sample period of Panel C, where the crude oil spot price dropped. The price of nearby month futures of which was used for the crude oil spot price declined, whose price difference with a deferred month product was enlarged; that is, the rollover costs increased, led to a fall in base price of ETFs as a result of the contango expansion. The Hypothesis II (The investment performance of Crude Company ETF is higher than that of ETFs operated with crude oil futures) was, therefore, rejected. In other words, it was confirmed that the risk-adjusted investment performance of the Crude Company ETF was lower than that of crude oil futures ETF in all sections of the whole sample; the estimated sample and the verification sample. Also, the Hypothesis III (When the spot price of crude oil rises, an investment performance of crude oil futures ETF is higher than that of the Crude Company ETF. And when the spot price of crude oil falls, an investment performance of the Crude Company ETF is higher than that of crude oil futures ETF.) could not be supported. However, it could be construed that the performance of crude oil futures ETF, which is operated as crude oil futures during the rising spot price of crude oil, is advantageous in terms of investment costs such as rollover costs compared to the period of the crude oil spot price fallen. As a result, the periodic yield and Sharpe index of the Crude Company ETF were lower than that of crude oil futures ETF in all sections, notwithstanding there was a difference in degree. ### 4.3 Analysis of Investment Performance of Leverage ETN Generally speaking, structural features of leverage ETP and inverse ETP have different investment risks from a usual ETP, which affect return on investment. Leverage ETF is a high-risk product that can earn more profits when the price of investment assets rises, but also increases a size of loss when the price of investment assets falls. Leverage ETP and inverse ETP do not simply follow the 2x, -1x, or -2x of the underlying assets price in the certain period such as six months, one year, or so. They are operated with a goal of 2x, -1x, or -2x of a daily rate of return. Additionally, it has an operating structure that can generate 2x, -1x, and -2x of the daily return rate based on the closing price since each investor has different investment timing. For this reason, the cumulative rate of return of the leverage ETP and the inverse ETP for a certain period of time may differ from the sum of the daily return rate at which compounding effects occur. If volatility enlarges and fluctuates as in the first half of 2020, the compounding effect for leverage ETP and inverse ETP may be lower than the underlying asset price return. In order to corroborate that the investment performance of the leverage ETP and inverse ETP, including periods of expanding volatility, will be low; thus, the Hypothesis IV (Crude oil leverage ETP has lower investment performance over a certain period of time compared to 1x crude oil related ETP) was established. The term yield and Sharpe index of crude oil spot (WTI), KODEX Crude Oil Futures ETF (ETPc), Shinhan Crude Oil Leverage ETN (ETPd), Mirae Asset Crude Oil Leverage ETN (ETPf) were examined as the sample for an empirical analysis. The whole sampling period was from December 28, 2016 to September 22, 2020, and was divided into an uptrend sample (Dec 28, 2016 ~ Oct 3, 2018) and a downtrend sample (Oct 4, 2018 ~ Sep 22, 2020). In the entire sample period of Panel A in <Table 13>, the periodic return of the two leverage ETNs were -98.25% and -91.92%, respectively. During the same period in the first half of 2020, the yield of the leverage ETN showed a sharp decline in the course of crude oil prices oscillated. For example, the crude oil price was$41.28 per barrel as of March 6, 2020, and it recorded \$41.11 per barrel on September 18, 2020. Its periodic return rate of crude oil spot was almost flat at -0.41% for circa six months. The prices for the two leverage ETNs, however, fell steeply from ₩7,345 and ₩13,345 to ₩305 and ₩2,080, consecutively in the same period of time. Besides, the periodic returns of them were -95.85% and -84.41%, respectively. The yield of the leverage ETNs with the compounding effect bespoke a big disparity from the periodic return of the underlying assets. This overall result, therefore, supported the Hypothesis IV (Crude oil leverage ETP has lower investment performance over a certain period of time compared to 1x crude oil related ETP).
In Panel A, the Sharpe indices of the two leverage ETNs (ETPd: -0.859, ETPf: -0.380) were lower than that of crude oil spot (0.130). However, the ETPd’s Sharpe index, -0.859, was even lower in comparison with the 1x crude oil futures ETF’s Sharpe index, -0.455, whereas, the Sharpe index of ETPf, -0.380, was rather higher. This result is due to the difference between the underlying indices of the two leverage ETPs; the ETPd is a product to trace the lead month return of WTI Futures by two times, and the ETPf is based on an index invested in Brent crude oil futures as well as WTI Futures. However, the WTI Futures are subject to real acquisitions, on the other hand, the Brent crude oil futures are not. In 2020, due to the decrease in demand for crude oil due to the CIVID-19, storage costs became an issue in the international crude oil market, and the volatility of the WTI Futures was greater than that of the Brent crude oil futures as a result. Therefore, each Sharpe index of the variables in <Table 13> could not be simply compared. Panel B, the sampling period from December 28, 2016 to October 3, 2018, in <Table 13> was a section where the crude oil spot prices inclined, that is, there is a high possibility that backwardation will form as the nearby month futures price rises, which is relatively larger than the deferred month. In the rising period of crude oil price, rollover costs may occur less, or rollover income could rather happen. The periodic rate of return and the Sharpe index of the two leverage ETNs were relatively higher than the whole period or the downturn period compared to the crude oil spot. In other words, there was no remarkable difference between the Sharpe indices for crude oil spot (0.883), crude oil futures ETF (0.819), and crude oil leverage ETNs (0.847, 0.916).
##### <Table 13>
Investment Performance of Leverage Crude oil ETP
WTI is crude oil spot, and ETPc, ETPd and ETPf are KODEX Crude Futures ETF, Shinhan Crude Oil leverage ETN, and Mirae Asset Crude Oil leverage ETN, respectively.
Panel A: Dec 28, 2016 ~ Sep 22, 2020
WTI ETPc ETPd ETPf
Risk-free AR 7.51% -21.09% -70.93% -31.85%
Volatility 57.61% 46.38% 82.56% 83.87%
Sharpe index 0.130 -0.455 -0.859 -0.380
Periodic Return -26.75% -69.61% -98.25% -91.92%

Panel B: Dec 28, 2016 ~ Oct 03, 2018

WTI ETPc ETPd ETPf
Risk-free AR 21.92% 19.20% 39.36% 48.85%
Volatility 24.82% 23.44% 46.46% 53.34%
Sharpe index 0.883 0.819 0.847 0.916
Periodic Return 41.34% 34.39% 63.71% 81.48%

Panel C: Oct 4, 2018 ~ Sep 22, 2020

WTI ETPc ETPd ETPf

Risk-free AR -5.41% -57.24% -169.88% -104.25%
Volatility 75.82% 59.85% 104.53% 103.73%
Sharpe index -0.071 -0.956 -1.625 -1.005
Periodic Return -46.72% -77.50% -98.94% -95.60%
Panel C of the sampling period from October 4, 2018 to September 22, 2020 in <Table 13> was a section where the spot price of crude oil dwindled. There is a high possibility of transpiring contango because the crude oil futures price of nearby month decreases even more than that of deferred month, which could be the section where rollover costs become greater. Reflecting this situation, the Sharpe index of ETPc (-0.956) and the two leverage ETNs (-1.625, -1.005) were consequentially lower than that of crude oil spot (-0.0071). The difference in the Sharpe index of the two leverage ETNs as -1.625 and -1.005, respectively, was confirmed by that the operating method and the underlying index of the two ETNs are dissimilar as described above.

### 5. Conclusions

This study examined the types and cost structure of crude oil-related ETPs, and correlation between each ETP and WTI spot market with the investment performance. The ETPs related to crude oil listed on the KRX were sampled and analyzed together with the WTI spot for the purpose of this investigation.
At first, as a result of the empirical analysis, there was the significant positive (+) correlation with the crude oil spot in the Crude Company ETF, which is listed only one in the KRX. The second, it was confirmed that since the crude oil futures ETP is operated as crude oil futures, the yield may be inferior to the crude oil spot due to investment expenses such as rollover costs incur. However, there is not a rollover cost because the Crude Company ETF invests in the shares of the crude oil producing companies in the stock market, which is obviously the spot market. Nonetheless, these companies are put at their intrinsic risks, hence, the return of the Crude Company ETFs is rather vulnerable. The Crude Company ETF performed worse than the crude oil futures ETF as a result of examining the periodic rate of return and risk-adjusted performance from December 28, 2016 to September 22, 2020. The result of dividing the entire sample period into an increase and a decrease in crude oil prices confirmed the similar result. Meanwhile, there is a high probability of a backwardation happened in the rising spot price of crude oil. In this situation, the price of crude oil futures for the nearby month which is applied as the crude oil spot price rises over the deferred month futures. As a result, it was confirmed that the investment performance of the crude oil futures ETPs is relatively higher in the period of rising crude oil prices than in the falling crude oil prices. Third, the investment performance of the crude oil leverage ETNs was deteriorated in the first half of 2020 because of the intensified volatility of the crude oil market. The leverage ETP and inverse ETP do not follow fluctuations of the underlying asset prices for a certain period of time. It could result in huge losses at a steady tone in the periodic return when the volatility is high. It is because their managing structure tracks the daily return of 2x, -1x, -2x, which leads to the compounding effects.
The highlight and creativity of this study are as follows: the first, KRX has formed a long box range around 2,000 points for more than ten years since 2007; thus, the return of direct and indirect investors for risky assets has been lower than the expectations. Because of fatigue from such disappointment, the investors rather pay attention to the alternative investment in order to augment the return of the investment. The raw material investment has recently become of interest by the investors; however, it leads to losses due to misunderstanding and confusion. In this study, it is conceivable to directly help investors in time to intuitively understand the performance and features of each ETP by looking into the operation structure and investment cost related to the Crude ETPs in comparison to the operation performance of each ETP based on the WTI spot market.
The second, studies on ETPs related to crude oil in the existing literature are rare, and there are few studies particularly investigated ETPs for the crude oil producing companies. The crude oil futures ETPs which incur rollover costs while the Crude Company ETPs have no rollover costs but are exposed to the inherent risks of the company. This is creative that the correlation between them and the performance for both were investigated in this paper, which were not found in the previous literature.
The third, the investment strategy using the crude oil-related ETPs is presented based on the results of the empirical analysis on the correlation between the crude oil spot market and investment performance of the crude oil-related ETPs. This approach will not only enhance the creativity of the research, but it will be of significance that it directly contributes to fund managers and management such as general investors and pension funds. In addition, it will assist the financial authorities to establish policies that have already taken a series of measures on the leverage ETP investment.

### Notes

1) The unrestricted model used for empirical analysis in this study is also in reduced form.

2) S&P Oil & Gas Exploration & Production Select Industry Index.

### References

1. Baek, J. S, M Kim, and M Yeom, A Study on the Investment Strategy using Correlations between Oil Fund and Russian Fund, Working paper, Hankuk University of Foreign Studies, 2020.

2. Baek, J. S, M Jeong, and M Yeom, 2018, The Lead-Lag Relationships between Russian Financial Market and International Oil Price, Korean Corporation Management Review, Vol. 78, pp. 59-78.

3. Chang, C. P, and C. C Lee, 2015, Do Oil Spot and Futures Prices Move Together?, Energy Economics, Vol. 50, pp. 379-390.

4. Choi, D. H, and J Yi, 2016, An Empirical Study on Price and Return of Commodity ETFs in Korea, Derivatives and Quantitative Studies, Vol. 24 (4), pp. 525-555.

5. Choi, H, D. J Leatham, and K Sukcharoen, 2015, Oil Price Forecasting using Crack Spread Futures and Oil Exchange Traded Funds, Contemporary Economics, Vol. 9 (1), pp. 29-44.

6. DiLellio, J. A, R Hessea, and D. J Stanley, 2014, Portfolio Performance with Inverse and Leveraged ETFs, Financial Services Review, Vol. 23 (2), pp. 123-149.

7. Dulaney, T, T Husson, and C. M McCann, 2012, Leveraged, Inverse, and Futures-Based ETFs, PIABA Bar Journal, Vol. 19 (1), pp. 83-108.

8. Financial Services Commission, Measures To Improve ETF And ETN Markets, 2020, http://www.fsc.go.kr/downManager?bbsid=BBS0048&no=152661 Downloaded 30 September, 2020.

9. Flores, I. M, Do Leveraged ETFs Induce Volatility on their Underlying Indices?The German Case, European Central Bank, Working Paper, 2015.

10. Garbade, K. D, and W. L Silber, 1983, Price movements and price discovery in futures and cash markets, The Review of Economics and Statistics, pp. 289-297.

11. Granger, C. W, 1969, Investigating Causal Relations by Econometric Models and Cross-Spectral Methods, Econometrica:journal of the Econometric Society, pp. 424-438.

12. Granger, C. W, 1980, Testing for Causality:A Personal Viewpoint, Journal of Economic Dynamics and Control, Vol. 2, pp. 329-352.

13. Hammoudeh, S, and H Li, 2005, Oil Sensitivity and Systematic Risk in Oil-Sensitive Stock Indices, Journal of Economics and Business, Vol. 57 (1), pp. 1-21.

14. Jones, C. M, and G Kaul, 1996, Oil and the Stock Markets, The Journal of Finance, Vol. 51 (2), pp. 463-491.

15. Liebi, L. J, 2020, The Effect of ETFs on Financial Markets:A Literature Review, Financial Markets and Portfolio Management, pp. 1-14.

16. Naeem, M, Z Umar, S Ahmed, and E. M Ferrouhi, 2020, Dynamic Dependence between ETFs and Crude Oil Prices by using EGARCH-Copula Approach, Physica A:Statistical Mechanics and its Applications, Vol. 557, pp. 124885.

17. Ozdurak, C, and V Ulusoy, 2020, Price Discovery in Crude Oil Markets:Intraday Volatility Interactions between Crude Oil Futures and Energy Exchange Traded Funds, International Journal of Energy Economics and Policy, Vol. 10 (3), pp. 402-413.

18. Pessina, C, and R. E Whaley, 2021, Levered and Inverse Exchange-Traded Products:Blessing or Curse?, Financial Analysts Journal, Vol. 77 (1), pp. 10-29.

19. Peterson, P. E, 2015, Contango and Backwardation as Predictors of Commodity Price Direction, Proceedings of the NCCC-134 Conference on Applied Commodity Price Analysis, Forecasting, and Market Risk Management, Forecasting, and Market Risk Management, St. Louis, MO:

20. Polanco-Martínez, J. M, and L. M Abadie, 2016, Analyzing Crude Oil Spot Price Dynamics versus Long Term Future Prices:A Wavelet Analysis Approach, Energies, Vol. 9 (12), pp. 1089.

21. Silvapulle, P, and I. A Moosa, 1999, The Relationship between Spot and Futures Prices:Evidence from the Crude Oil Market, Journal of Futures Markets:Futures, Options, and Other Derivative Products, Vol. 19 (2), pp. 175-193.

22. Yoon, J. Y, 2015, The Misunderstanding and Usages about Commodity Investing:Focusing on Crude Oil Futures ETFs, Financial Planning Review, Vol. 8 (2), pp. 73-88.