- Email: [email protected]
Empirical analysis of optimal strategic petroleum reserve in China
Available online at www.sciencedirect.com
Energy Economics 30 (2008) 290 – 302 www.elsevier.com/locate/eneco
Empirical analysis of optimal strategic petroleum reserve in China Yi-Ming Wei a,b,c,⁎, Gang Wu a,b , Ying Fan a,c , Lan-Cui Liu a,b a b
Institute of Policy and Management, Chinese Academy of Science, Beijing, 100080, China School of Business, University of Science and Technology of China, Hefei, 230026, China c Center of Forecasting Sciences, Chinese Academy of Sciences, Beijing, 100080, China
Received 2 September 2005; received in revised form 12 July 2006; accepted 15 July 2006 Available online 23 August 2006
Abstract The Chinese government began to prepare for the establishment of strategic petroleum reserve in March 2004. Therefore, answering the question of what level of strategic petroleum reserve would be suitable for China's future economic development becomes essential. Using a decision tree model based on a cost function, this paper quantifies China's optimal strategic petroleum reserve for the period 2005–2020. This approach provides a methodology reference for further quantified discussion on China's SPR. Our results show that: for economic development and security of the energy supply, the strategic petroleum reserve should be the equivalent of 30–60 days of net oil import for an optimal solution, when the oil price is $ 50/bbl; with a reserve of an equivalent of 60–90 days of net oil import to have an optimal solution when oil price is $ 20–35/bbl. © 2006 Elsevier B.V. All rights reserved. Keywords: Strategic petroleum reserve (SPR); Loss of GDP; Decision tree model
1. Introduction With continuous development of China's economy, the oil demand is increasing rapidly. According to the 2005 China's Oil and Gas Industry Annual Report (State Information Center, 2005), it is estimated that China's oil demand in 2010 will be approximately 350 million tons, while domestic production is only 180 million tons. This implies that China will import 170 million tons, giving an import dependence of nearly 50%. However, most of the oil import came from the unstable Middle East, of which 80% has to get across the perilous Malacca Strait. ⁎ Corresponding author. Institute of Policy and Management (IPM), Chinese Academy of Sciences (CAS) P.O. Box 8712, Beijing 100080, China. Tel./fax: +86 10 62650861. E-mail addresses: [email protected], [email protected] (Y.-M. Wei). 0140-9883/$ - see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.eneco.2006.07.001
Y.-M. Wei et al. / Energy Economics 30 (2008) 290–302
291
China's oil imports are negotiated in cash (other than futures); so there are some strange trading phenomenon: such as buying when the price is high rather when the price is low; and when the oil price is high, more oil is imported. Therefore, China's oil supply security has many risks that can directly influence economic security and social stability. The strategic petroleum reserve (SPR) is an emergency petroleum store maintained by the government to decrease economic and social reparation resulting from a shortage. It is important to establish the SPR for China and to help deal with petroleum use crisis and ensure energy and economic security. Hence the Chinese government began to prepare for the establishment of SPR in March 2004. The first project included four stockpile bases, and the largest was completed in December 2006 — a reserve of 5.2 million cubic meters oil. Others will be completed at the end of 2008. If China had its own SPR, a petroleum crisis would not have occurred in China's Guangzhou and Shenzhen in August 2005. This event caused significant negative effects on economic development and social stability. It is important to distinguish between strategic reserve and operational and speculative reserve; although strategic, operational and speculative reserves constitute physically the same inventory. The former constitutes that part of the petroleum inventory not used unless an emergency has occurred, and is deployed mainly in order to provide lead-time between the occurrence of the emergency and measures required to resolve it. It also provides a bargaining position or prevents hostile actions, and thus ensures short-term protection and short-term demand (Samouilidis and Berahas, 1982). Industrialized countries suffered greatly during the two oil crises in the 1970s. The US Department of Energy (1988) estimated that the lost economic growth from the two oil shocks of the 1970s were $ 1.2 trillion (evidently estimated in 1987 or 1988). Post 1973–74, some industrialized countries set up the IEA (International Energy Agency), which required that each member be committed to maintain a reserve that is equivalent to 90 days of net oil imports. In order to be able to decrease the negative effects of oil shortages, strategic petroleum reserves are essential for dampening shocks to the economic system and providing an interval during which efforts can be made to mediate disputes. Thus, the larger the SPR, the lesser the economy is affected in a negative way. Thus the two fundamental questions are: How many days of net oil imports will be China's SPR? Will the SPR of 90 days of net oil imports be suitable for China? In fact, security of energy supply is a relative and uncertain problem. The degree of oil supply security is decided upon not only by the level of SPR, but also by the possibility of short-term oil supply disruption and the possible loss of the economy. Currently many countries would like to establish their SPR as large as possible to prevent any oil supply disruption through minimizing the total cost and making energy demand and cost to the economy an equilibrium state. Clearly there exists a trade-off between the stockpile and the vulnerability to supply disruptions. The larger the inventories held, the lower the vulnerability, but at the same time the higher the inventory holding costs. Therefore, the loss of GDP may be different for each of the countries. Even if oil supply disruption and dependence on oil imports could be the same, the SPR in different countries could be different. Some industrialized countries or regions have their own optimal SPR to meet their needs (see Table 1). Table 1 shows that the stockpile for the US and Japan is larger than those of other countries and regions. Chinese Taipei is the smallest. Then for China, what level of SPR can be optimal? Since the 1980s, there have been studies discussing the optimal SPR for industrialized countries. Such work provides valuable information for studying China's SPR issues. Teisberg (1981) developed a stochastic dynamic programming model that allows explicit consideration of such uncertainty. The model has been used to determine both the size of the US SPR and the
292
Y.-M. Wei et al. / Energy Economics 30 (2008) 290–302
Table 1 The SPR of some industrialized countries or regions in 1999 (Paik et al., 1999) Countries/regions
SPR controlled by government (million barrels)
US Japan South Korea Chinese Taipei Europe Total
563 315 43 12 325 1258
optimal fill up and drawdown rate contingent upon the supply conditions, time and available inventory. Chao and Manne (1982) developed a multi-period dynamic programming model for determining optimal stockpiles and petroleum usage rates based on their analysis of US petroleum supply policies. Samouilidis and Berahas (1982) established a cost function which includes the inventory procurement and maintenance cost and the shortage cost inflicted by a petroleum shortfall, and evaluated each scenario based on the cost function and the decision tree model. Samouilidis and Magirou (1985) presented a simple analysis for the optimal selection of the level of petroleum reserve for a small country based on the study of Samouilidis and Berahas (1982). Zweifel and Bonomo (1995) developed an optimal reserve model that takes into account multiple risks for oil and gas, and pointed out that one-dimensional rules such as “oil reserve for 90 days” turn out to be not only suboptimal but also to suggest adjustments exacerbating sub-optimality. Because the impact of the oil crisis in the 1970s on China's energy use and economic development is relatively small, (and China became a net crude oil import country in 1996), the domestic studies of China's SPR are mainly qualitative. Zhou (2001) analyzed the role of national SPR and indicated that the size of SPR should be suitable for China's need. An (2003) argued that it is necessary, and urgent, to establish China's national SPR and consider that 90 days of net oil imports were suitable. Therefore, based on a decision tree model, this paper quantifies China's optimal SPR level in different scenarios in 2010 and 2020. The plan of this paper is as follows. First, we portray different scenarios based on a decision tree model, and then develop a cost function of SPR which includes the cost of procurement, maintenance of SPR and expected shortage cost. Next, we analyze the empirical results. The last section discusses conclusions. 2. Methods and data 2.1. The decision tree model An alternative way to structure a decision problem pictorially is by using a decision tree. A decision tree depicts chronologically the sequence of actions and outcomes as they unfold. For a terminal decision problem based on prior information, the first fork (square node) corresponds to the action chosen by the decision maker, and the second fork (round node) corresponds to the event. The numbers at the end of these terminal branches are the corresponding payoffs (or losses) (Ravindran et al., 1987). The decision tree has been widely used as a modeling approach. However, very little is known from the literature on how the decision tree performs in predicting an optimal stockpile reserve. Decision trees are a non-parametric modeling approach, which recursively splits the multidimensional space defined by the independent variables into zones that are as homogenous as possible in terms of the response of the dependent variables (Vayssieres et al., 2000). This relates the decision variables and parameters with selected possible events; as illustrated in Fig. 1.
Y.-M. Wei et al. / Energy Economics 30 (2008) 290–302
293
Fig. 1. The decision tree model structure.
Each decision (S) concerning the stockpile of different periods (for simplicity, in Fig. 1 only three alternatives are presented) are associated with several oil prices and loss of GDP caused by shortages due to oil supply emergencies (again only three are presented). Any branch of this decision tree model expresses a scenario to be evaluated by the decision maker. 2.2. Methods and variables explanation In this paper, the economic factor considered is the shortage cost expressed as a function of GDP loss due to oil supply disruptions and the holding costs is expressed as a function of the level of strategic stockpile. This is a function by which the decision maker can evaluate each decision, i.e. each branch of the decision tree. Its general form is the following: TC ¼ SC þ EGL
ð1Þ
Where: TC SC EGL
total expected cost. cost of procurement and maintenance of the SPR. expected shortage cost.
(1) Expected shortage cost The sudden rising of oil prices creates three types of economic loss to the economy for oil imported countries: (1) transfer of wealth from oil imported countries to oil exported countries; (2) loss of the potential GDP; (3) macroeconomic adjustment loss. The transfer of wealth is a transfer of payment, and is equal to the quantity of oil import multiplied by the difference between the monopoly price and the competitive market price of oil. The GDP loss considered in this paper refers to the sum of the previously discussed three losses. EGL ¼
3 X
qi ⁎GDP⁎Q
ð2Þ
i¼l
Q ¼ r⁎e⁎c
ð3Þ
e ¼ bp þ by roil
ð4Þ
bp ¼ bd þ bs
ð5Þ
294
Y.-M. Wei et al. / Energy Economics 30 (2008) 290–302
Concentrating on the economic aspects of vulnerability of the oil price increase, it can be argued that these basically involve the decreasing GDP caused by oil shortages. The loss of potential GDP is equal to producers' and consumers' surplus losses with the economy in all markets affected by the oil price increase (Greene et al., 1998). So the total elasticity of oil price changes is equal to the elasticity of price with respect to demand and the elasticity of price with respect to supply surplus. (2) Cost of procurement and maintenance of the SPR SC ¼ IC þ RC þ OC
ð6Þ
IC ¼ Pj ⁎R⁎D
ð7Þ
RC ¼ Pj ⁎R⁎D⁎m
ð8Þ
OC ¼ Pj ⁎R⁎D⁎n
ð9Þ
The definition of all variables in this cost function can be seen in Table 2. (3) Decision analysis for the optimal SPR level Because decision-makers do not know exactly what oil supply disruption will occur and what oil prices will be in the future, they can assume or reckon the possibility of possible disruptions and prices based on experience and the history of disruptions, then make decisions. The optimal SPR is a decision with risk. There are many risks for decision-makers to consider prior to establishing SPR: the GDP loss will be too large if long-term oil supply disruption occurs and the size of SPR is too small in the future; the expensive reserve investment and maintenance cost for Table 2 The definition of variables and parameters Variable/ parameter
Explanation
IC RC OC Pj R D m n Q r e c
Oil import cost. Investment cost (building reserve infrastructure). Operation and maintenance cost. The oil price, refers to 20$/bbl, 35$/bbl, 50$/bbl respectively. Everyday net oil import. The days of net oil import of SPR. The assumptive ratio between investment cost and import cost. The assumptive ratio between operation and maintenance cost and import cost. The GDP loss ratio during oil supply disruption. The oil price increase when oil supply disruption occurs. The loss of GDP caused by the fluctuation of oil price (Greene et al., 1998). Constant, refers to the GDP loss rate when SPR is released, which is estimated based on Greene et al. (1998) and Teisberg (1981). The elasticity of oil price with respect to GDP (Greene et al., 1998). The elasticity of oil consumption with respect to GDP (Greene et al., 1998). The elasticity of GDP with respect to oil price (Greene et al., 1998). The elasticity of oil price with respect to demand (Greene et al., 1998). The elasticity of oil price with respect to supply (Greene et al., 1998). The possibility of oil supply disruption, i = 1,2,3 refers to short-term shortage, medium-term shortage, long-term shortage respectively.
βp βy σoil βd βs ρI
Y.-M. Wei et al. / Energy Economics 30 (2008) 290–302
295
establishing SPR will be wasted if oil supply disruption will not occur or the short-term disruption will occur in the future. Therefore, this paper analyzes the optimal SPR within scenarios of different oil prices using decision tree analysis. The decision tree is the figure form of expected value (EV) method. The decision dots are presented in small panes; the arcs linked with decision dots refer to possible SPR decisions, that are 30, 60, and 90 days of net oil imports. The remaining ends of arcs are condition dots presented in circles; the arcs linked with condition dots are possibility arcs above which show possible oil supply disruptions. The other ends are result dots presented by small triangles; there is a figure behind each which refers to the cost of every decision under different conditions (Figs. 2–5). When the possibility of oil supply disruption occurring is ρ, then the expected cost of every decision is E(α,ρ) and total expected cost of national SPR is E(α). Because the possible condition is discrete, the possibility of condition i is ρi. Then X EðaÞ ¼ qi Pða; iÞ; 8aaSl ð10Þ iaS2
Where S1 S2
is the set of every decisions; is the set of different conditions.
Fig. 2. The decision tree of SPR analysis for China in 2010 in scenario I. Note: The other units are always in 100 million dollars, besides possibilities.
296
Y.-M. Wei et al. / Energy Economics 30 (2008) 290–302
Fig. 3. The decision tree of SPR analysis for China in 2020 in scenario I.
When minimizing total cost, then decision α⁎ is optimal. min EðaÞ ¼ Eða⁎ Þ
aaS1
ð11Þ
2.3. The assumptions The empirical analysis in this paper is based on the following assumptions: (1) China's SPR base is establishing and has no stockpile. So we do not know how much the exact maintenance and investment costs are. We assume that the proportion among oil import cost, investment cost and maintenance cost in China's SPR investment is 75.4%, 22.9%, 1.7% respectively, which refers to the experience of the US' SPR. And the estimated (or assumed) value of the other parameters in cost function can be seen in Table 3. Greene et al. (1998) posited that although the SPR can be effective against a short-lived, random shock, it is relatively ineffectual against a strategic shock of two years or more. It is quite clear that SPR alone cannot resolve a major supply emergency. So the SPR cannot by itself provide long-term security. But Samouilidis and Berahas (1982) believed that the amount of the SPR held is a function not only of the holding costs and petroleum prices, but also of the displaced shortage costs (either monetary or social) and the other damages incurred by an energy emergency. So we consider both opinions.
Y.-M. Wei et al. / Energy Economics 30 (2008) 290–302
297
Fig. 4. The decision tree of SPR analysis for China in 2010 in scenario II.
(2) We assume that the impact of the SPR of 60 and 90 days of net oil import on the decreasing GDP loss are the same if long-term oil supply disruption occurs. (3) We assume that the impact of different SPR on the decreasing GDP loss is different whatever oil supply disruption occurs. (4) Based on some forecasts for China's future oil demand and production, we assumed that China's net import of crude oil is 130 million tons in 2005, 150 million tons in 2010, and 200 million tons in 2020 respectively. 2.4. The source of data The world oil prices, China's oil consumption and oil production for each year are taken from BP Statistical Review of World Energy 2004 (BP, 2004) and China Statistics Yearbook 2003 (State Statistical Bureau, 2004); the GDP of China is taken from China Statistics Yearbook 2003. The GDP and the oil demand of China in 2005, 2010 and 2020 are taken from Chen (2003). The historical information of oil supply disruption is taken from EIA (Energy Information Agency) of US Department of Energy. 3. Results and discussion The Chinese government began to establish national SPR in 2004. The biggest reserve will be finished in December 2006. So it is not possible to own national SPR now. The optimal SPR of
298
Y.-M. Wei et al. / Energy Economics 30 (2008) 290–302
Fig. 5. The decision tree of SPR analysis for China in 2020 in scenario II.
2005 calculated in this paper is theoretical (and is not discussed). The optimal SPR under different assumptions and oil prices in 2005, 2010, and 2020 can be found in Table 4. 3.1. Scenario I analysis For scenario I, the assumption that the impacts of the SPR of 60 and 90 days of net oil import on the decreasing GDP loss are the same if long-term oil supply disruption occurs. The results in 2010 show that: (1) The optimal SPR is 60 days of China's net oil imports when oil price is $ 20/bbl and the economy growth is around 7%; (2) The optimal SPR is also 60 days of China's net oil imports when oil price is $ 35/bbl and the economy growth is around 7%; (3) The optimal SPR is also 30 days of China's net oil imports when oil price is $ 50/bbl and the economy growth is around 7%. When oil price is $ 20/bbl and $ 35/bbl, the reserve cost per barrel is relatively low, the contribution of SPR of 60 days for decreasing the GDP loss caused by oil supply disruption is great compared with the SPR of 30 days, so the total cost for SPR of 60 days is less than that of 30 days. Because we assume that the contribution of SPR of 60 days and 90 days for decreasing the GDP
Y.-M. Wei et al. / Energy Economics 30 (2008) 290–302
299
Table 3 The estimated (or assumed) value of the parameters Parameter
Value
Scope
Notes
m
0.30
n βy
0.02 0.49
σoil βd
− 0.05 − 0.037
20$/bbl
− 0.068 − 0.099 0.023 0.032 0.056 62.6%
35$/bbl 50$/bbl 20$/bbl 35$/bbl 50$/bbl Short-term
18.7% 18.7%
Medium-term Long-term
βs
ρI
For the proportion of IC, RC, OC in the US SPR is 75.4%, 22.9%, 1.7% respectively. It is our calculation based on China's GDP, oil product, consumption, and international oil price from 1985–2003. The results from the US Department of Energy Annual Energy Outlook (1995) forecast (US Department of Energy, 1995)
The data from the US Department of Energy indicates that there are sixteen oil supply disruptions or shortages from the 1950s. According to the fluctuation of oil price, we calculate the probability of shortage.
loss caused by long-term oil supply disruption are the same, the reserve investment of 90 days is far more than that of 60 days. Therefore, the expected total cost of SPR of 60 days is lesser. When oil price is $ 50/bbl, the reserve cost per barrel is relatively high, and the GDP loss caused by oil supply disruption is less than the increased reserve investment for 60 and 90 days of SPR, although the contribution of SPR of 60 days for decreasing the GDP loss caused by oil supply disruption is great compared with the SPR of 30 days, so the total cost for establishing SPR of 60 and 90 days is larger than that of 30 days. Therefore, the optimal SPR in 2010 for China is 60 days of net oil import if the fluctuation of world oil price is $ 35/bbl or less; the optimal SPR for China should be 30 days of net oil import if the fluctuation of world oil price is $ 50/bbl or higher (Fig. 2). The results in 2020 show that the optimal SPR for China is 60 days of net oil import, when oil prices are $ 20/bbl, $ 35/bbl, or $ 50/bbl and the economic growth is around 7% (Fig. 3). It is estimated that China's oil demand will reach 450 million tons in 2020; and the domestic production will be 200 million tons. So the dependence on oil imports will be more than 55%, and the impact of the fluctuation of world oil prices on economic development will be greater than now; which makes the establishment of a larger SPR necessary. Theoretically, the greater the size of SPR in 2020 is, the lesser the GDP loss is. But the possibility of long-term oil supply disruption
Table 4 The optimal SPR in different assumptions and oil prices in 2005, 2010, and 2020 Year
2005 2010 2020
$20/bbl
$35/bbl
$50/bbl
Scenario I
Scenario II
Scenario I
Scenario II
Scenario I
Scenario II
60 days 60 days 60 days
60 days 90 days 90 days
30 days 60 days 60 days
30 days 90 days 90 days
30 days 30 days 60 days
30 days 30 days 90 days
300
Y.-M. Wei et al. / Energy Economics 30 (2008) 290–302
occurring is only 18.7% and the contribution of SPR of 60 days and 90 days for decreasing the GDP loss caused by long-term oil supply disruption are the same. The possibility of short-term oil supply disruption occurring is 62.6% and the reserve investment of 90 days is more than that of 60 days, although the contribution of SPR of 90 days for decreasing the GDP loss caused by oil supply disruption is greater than that of 60 days. So the total cost for establishing SPR of 90 days of net oil import is higher than that of 60 days. Therefore, the optimal SPR for China will be 60 days of net oil import in 2020 whatever the world oil price is. 3.2. Scenario II analysis For scenario II, the assumption that the impact of different SPR on the decreasing GDP loss is different when various oil supply disruption occurs. The results in 2010 show that (Fig. 4): (1) The optimal SPR is 90 days of China's net oil imports when oil price is $ 20/bbl and the economy growth is around 7%; (2) The optimal SPR is also 90 days of China's net oil imports when oil price is $ 35/bbl and the economy growth is around 7%; (3) The optimal SPR is also 30 days of China's net oil imports when oil price is $ 50/bbl and the economy growth is around 7%. We assume that the contribution of alternative SPR for the decreasing GDP loss is different when different oil supply disruption occurs. When oil price is $ 20/bbl and $ 35/bbl, the reserve cost per barrel is relatively low, the contribution of SPR of 90 days for decreasing the GDP loss caused by oil supply disruption is greater compared with the SPR of 60 and 30 days. So the total cost for SPR of 90 days is less than that of 60 and 30 days. Therefore, the expected least total cost of SPR is 90 days. When oil price is $ 50/bbl, the reserve cost per barrel is relatively high, and the GDP loss caused by oil supply disruption is less than the increased reserve investment of 60 and 90 days SPR, although the contribution of SPR of 60 days and 90 days for decreasing the GDP loss caused by oil supply disruption is greater compared with the SPR of 30 days. So the total cost for establishing SPR of 60 and 90 days is larger than that of 30 days. Therefore, the optimal SPR in 2010 for China is 90 days net oil import if the fluctuation of world oil price is $ 35/bbl or less; the optimal SPR for China should be 30 days' net oil import if the fluctuation of world oil price is $ 50/bbl or higher. The results in 2020 show that the optimal SPR for China is 90 days of net oil imports, when oil prices are $ 20/bbl, $ 35/bbl, or $ 50/bbl and the economy growth is around 7% (Fig. 5). Because the oil demand of China will be greater, the dependence on oil imports and the share of oil in the primary fuel mix will increase, and the impact of the fluctuation of world oil prices on economic development will be greater than now. This supports the argument for establishing a larger size of SPR. Therefore, the optimal SPR for China will be 90 days of net oil imports in 2002, whatever world oil price is. 4. Conclusions and further work The security value of strategic petroleum reserve is a very complex problem. In order to estimate the optimal SPR level, all factors related to SPR must be quantified. But this is difficult to accomplish. So we make a lot of assumptions. Although the approach omits many features of the
Y.-M. Wei et al. / Energy Economics 30 (2008) 290–302
301
problem, it considers the most important decision variables that greatly influence the optimal stockpile. Thus, it gives the decision maker basic information such as storage and shortage costs. For the same period, the optimal SPR of the lower oil price scenario is always larger than or equal to that of the higher oil prices scenario; which is in complete accord with the rule “buy low and sell high”. The optimal SPR will increase from 30 days to 90 days of net oil imports with greater requirements for energy and high dependence on oil imports, which is necessary for economic development and energy security. Our results show that the optimal SPR level is sometimes not to have the largest strategic stockpile. When oil prices are low, 60 days and 90 days of net oil imports are suitable for China's economic development and security in 2010 and 2020. When oil prices are high, the optimal SPR should be 30 and 60 days of net oil import for 2010 and 2020. Therefore, the oil prices can influence the optimal SPR. This paper discusses the impact of different world oil prices on China's SPR because the change of world oil price can influence import cost and, importantly, as the second biggest oil consuming country in the world, the establishment of China's SPR may influence world oil prices. This is one of the reasons why we considered the change of world oil prices in our model. But the decision model in this paper could not identify how much the world oil price will fluctuate. However, we might simulate the correlation between world oil price and China's strategic stockpile in further work. Acknowledgements The authors gratefully acknowledge the financial support from the National Natural Science Foundation of China (NSFC) under the grants Nos.70425001 and 70371064, the Key Projects from the Ministry of Science and Technology of China (grants 2001-BA608B-15, 2001-BA60501).Yi-Ming Wei truly appreciates the supports from Prof. Michael B. McElroy and Mr. Chris P. Nielsen at Harvard University. We also would like to thank Professor R.S.J. Tol and the anonymous referees for their helpful suggestions and corrections on the earlier draft of our paper according to which we improved the content. References An, Q.Y., 2003. It is no time to delay for establishing strategic petroleum reserve. China Youth Science and Technology 4, 18–20 (In Chinese). BP, 2004. BP Statistical review of world energy. British Petroleum (BP), London. Chao, H., Manne, A.S., 1982. Oil stockpiles and import reductions: a dynamic programming approach. Operations Research 31, 632–651. Chen, W.Y., 2003. Carbon quota price and CDM potentials after Marrakesh. Energy Policy 31, 709–719. Greene, D.L., Jones, D.W., Leiby, P.N., 1998. The outlook for US oil dependence. Energy Policy 26, 55–69. Paik, I., Leiby, B., Jones, D., et al., 1999. Strategic oil stocks in the APEC region. Proceedings of the 22nd IAEE Annual International Conference. International Association for Energy Economists. Ravindran, A., Phillips, D.T., Solberg, J.J., 1987. Operations Research: Principles and Practice, 2nd ed. John Wiley and Sons, Inc., New York. Samouilidis, J.E., Berahas, S.A., 1982. A methodological approach to strategic petroleum reserves. The International Journal of Management Science 10, 565–574. Samouilidis, J.E., Magirou, V.F., 1985. On the optimal level of a small country's strategic petroleum reserve. European Journal of Operational Research 20, 190–197. State Information Center, 2005. 2005 China's Oil and Gas Industry Annual Report. China Economic Publishing House, Beijing.
302
Y.-M. Wei et al. / Energy Economics 30 (2008) 290–302
State Statistical Bureau, 2004. China Statistical Yearbook 2003. China Statistical Press, Beijing. Teisberg, T.J., 1981. A dynamic programming model of the U.S. strategic petroleum reserve. The Bell Journal of Economics 12, 526–546. US Department of Energy, 1988. United States Energy Policy 1980–1988 DOE/S-0068. Washington DC, October. US Department of Energy, Energy Information Administration, 1995. Annual Energy Outlook 1995 DOE/EIA-0383. Washington DC, January. Vayssieres, M.P., Plant, R.E., Allen-diaz, B.H., 2000. Classification tree: an alternative non-parametric approach for predicting species distribution. Journal of Vegetation Science 11, 679–694. Zhou, D.D., 2001. Strategic petroleum reserve linked with national strategies-the roles of China's strategic petroleum reserve. China Petroleum 5, 15–16 (In Chinese). Zweifel, P., Bonomo, S., 1995. Energy security coping with multiple supply risks. Energy Economics 17, 179–183. Dr. Yi-Ming Wei is a Professor at the Institute of Policy and Management of the Chinese Academy of Sciences; and is currently a visiting scholar at Harvard University in the United States. Mr. Gang Wu is a Ph.D. candidate in Management Science at the Institute of Policy and Management, Chinese Academy of Sciences, China. Dr. Ying Fan is a Professor at the Institute of Policy and Management, Chinese Academy of Sciences, China. Her research field is energy policy and system engineering. In 2004, she was a visiting scholar at Cornell University, USA. Ms. Lan-Cui Liu is a Ph.D. candidate in Management Science at the Institute of Policy and Management, Chinese Academy of Sciences, China.
Energy Economics 30 (2008) 290 – 302 www.elsevier.com/locate/eneco
Empirical analysis of optimal strategic petroleum reserve in China Yi-Ming Wei a,b,c,⁎, Gang Wu a,b , Ying Fan a,c , Lan-Cui Liu a,b a b
Institute of Policy and Management, Chinese Academy of Science, Beijing, 100080, China School of Business, University of Science and Technology of China, Hefei, 230026, China c Center of Forecasting Sciences, Chinese Academy of Sciences, Beijing, 100080, China
Received 2 September 2005; received in revised form 12 July 2006; accepted 15 July 2006 Available online 23 August 2006
Abstract The Chinese government began to prepare for the establishment of strategic petroleum reserve in March 2004. Therefore, answering the question of what level of strategic petroleum reserve would be suitable for China's future economic development becomes essential. Using a decision tree model based on a cost function, this paper quantifies China's optimal strategic petroleum reserve for the period 2005–2020. This approach provides a methodology reference for further quantified discussion on China's SPR. Our results show that: for economic development and security of the energy supply, the strategic petroleum reserve should be the equivalent of 30–60 days of net oil import for an optimal solution, when the oil price is $ 50/bbl; with a reserve of an equivalent of 60–90 days of net oil import to have an optimal solution when oil price is $ 20–35/bbl. © 2006 Elsevier B.V. All rights reserved. Keywords: Strategic petroleum reserve (SPR); Loss of GDP; Decision tree model
1. Introduction With continuous development of China's economy, the oil demand is increasing rapidly. According to the 2005 China's Oil and Gas Industry Annual Report (State Information Center, 2005), it is estimated that China's oil demand in 2010 will be approximately 350 million tons, while domestic production is only 180 million tons. This implies that China will import 170 million tons, giving an import dependence of nearly 50%. However, most of the oil import came from the unstable Middle East, of which 80% has to get across the perilous Malacca Strait. ⁎ Corresponding author. Institute of Policy and Management (IPM), Chinese Academy of Sciences (CAS) P.O. Box 8712, Beijing 100080, China. Tel./fax: +86 10 62650861. E-mail addresses: [email protected], [email protected] (Y.-M. Wei). 0140-9883/$ - see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.eneco.2006.07.001
Y.-M. Wei et al. / Energy Economics 30 (2008) 290–302
291
China's oil imports are negotiated in cash (other than futures); so there are some strange trading phenomenon: such as buying when the price is high rather when the price is low; and when the oil price is high, more oil is imported. Therefore, China's oil supply security has many risks that can directly influence economic security and social stability. The strategic petroleum reserve (SPR) is an emergency petroleum store maintained by the government to decrease economic and social reparation resulting from a shortage. It is important to establish the SPR for China and to help deal with petroleum use crisis and ensure energy and economic security. Hence the Chinese government began to prepare for the establishment of SPR in March 2004. The first project included four stockpile bases, and the largest was completed in December 2006 — a reserve of 5.2 million cubic meters oil. Others will be completed at the end of 2008. If China had its own SPR, a petroleum crisis would not have occurred in China's Guangzhou and Shenzhen in August 2005. This event caused significant negative effects on economic development and social stability. It is important to distinguish between strategic reserve and operational and speculative reserve; although strategic, operational and speculative reserves constitute physically the same inventory. The former constitutes that part of the petroleum inventory not used unless an emergency has occurred, and is deployed mainly in order to provide lead-time between the occurrence of the emergency and measures required to resolve it. It also provides a bargaining position or prevents hostile actions, and thus ensures short-term protection and short-term demand (Samouilidis and Berahas, 1982). Industrialized countries suffered greatly during the two oil crises in the 1970s. The US Department of Energy (1988) estimated that the lost economic growth from the two oil shocks of the 1970s were $ 1.2 trillion (evidently estimated in 1987 or 1988). Post 1973–74, some industrialized countries set up the IEA (International Energy Agency), which required that each member be committed to maintain a reserve that is equivalent to 90 days of net oil imports. In order to be able to decrease the negative effects of oil shortages, strategic petroleum reserves are essential for dampening shocks to the economic system and providing an interval during which efforts can be made to mediate disputes. Thus, the larger the SPR, the lesser the economy is affected in a negative way. Thus the two fundamental questions are: How many days of net oil imports will be China's SPR? Will the SPR of 90 days of net oil imports be suitable for China? In fact, security of energy supply is a relative and uncertain problem. The degree of oil supply security is decided upon not only by the level of SPR, but also by the possibility of short-term oil supply disruption and the possible loss of the economy. Currently many countries would like to establish their SPR as large as possible to prevent any oil supply disruption through minimizing the total cost and making energy demand and cost to the economy an equilibrium state. Clearly there exists a trade-off between the stockpile and the vulnerability to supply disruptions. The larger the inventories held, the lower the vulnerability, but at the same time the higher the inventory holding costs. Therefore, the loss of GDP may be different for each of the countries. Even if oil supply disruption and dependence on oil imports could be the same, the SPR in different countries could be different. Some industrialized countries or regions have their own optimal SPR to meet their needs (see Table 1). Table 1 shows that the stockpile for the US and Japan is larger than those of other countries and regions. Chinese Taipei is the smallest. Then for China, what level of SPR can be optimal? Since the 1980s, there have been studies discussing the optimal SPR for industrialized countries. Such work provides valuable information for studying China's SPR issues. Teisberg (1981) developed a stochastic dynamic programming model that allows explicit consideration of such uncertainty. The model has been used to determine both the size of the US SPR and the
292
Y.-M. Wei et al. / Energy Economics 30 (2008) 290–302
Table 1 The SPR of some industrialized countries or regions in 1999 (Paik et al., 1999) Countries/regions
SPR controlled by government (million barrels)
US Japan South Korea Chinese Taipei Europe Total
563 315 43 12 325 1258
optimal fill up and drawdown rate contingent upon the supply conditions, time and available inventory. Chao and Manne (1982) developed a multi-period dynamic programming model for determining optimal stockpiles and petroleum usage rates based on their analysis of US petroleum supply policies. Samouilidis and Berahas (1982) established a cost function which includes the inventory procurement and maintenance cost and the shortage cost inflicted by a petroleum shortfall, and evaluated each scenario based on the cost function and the decision tree model. Samouilidis and Magirou (1985) presented a simple analysis for the optimal selection of the level of petroleum reserve for a small country based on the study of Samouilidis and Berahas (1982). Zweifel and Bonomo (1995) developed an optimal reserve model that takes into account multiple risks for oil and gas, and pointed out that one-dimensional rules such as “oil reserve for 90 days” turn out to be not only suboptimal but also to suggest adjustments exacerbating sub-optimality. Because the impact of the oil crisis in the 1970s on China's energy use and economic development is relatively small, (and China became a net crude oil import country in 1996), the domestic studies of China's SPR are mainly qualitative. Zhou (2001) analyzed the role of national SPR and indicated that the size of SPR should be suitable for China's need. An (2003) argued that it is necessary, and urgent, to establish China's national SPR and consider that 90 days of net oil imports were suitable. Therefore, based on a decision tree model, this paper quantifies China's optimal SPR level in different scenarios in 2010 and 2020. The plan of this paper is as follows. First, we portray different scenarios based on a decision tree model, and then develop a cost function of SPR which includes the cost of procurement, maintenance of SPR and expected shortage cost. Next, we analyze the empirical results. The last section discusses conclusions. 2. Methods and data 2.1. The decision tree model An alternative way to structure a decision problem pictorially is by using a decision tree. A decision tree depicts chronologically the sequence of actions and outcomes as they unfold. For a terminal decision problem based on prior information, the first fork (square node) corresponds to the action chosen by the decision maker, and the second fork (round node) corresponds to the event. The numbers at the end of these terminal branches are the corresponding payoffs (or losses) (Ravindran et al., 1987). The decision tree has been widely used as a modeling approach. However, very little is known from the literature on how the decision tree performs in predicting an optimal stockpile reserve. Decision trees are a non-parametric modeling approach, which recursively splits the multidimensional space defined by the independent variables into zones that are as homogenous as possible in terms of the response of the dependent variables (Vayssieres et al., 2000). This relates the decision variables and parameters with selected possible events; as illustrated in Fig. 1.
Y.-M. Wei et al. / Energy Economics 30 (2008) 290–302
293
Fig. 1. The decision tree model structure.
Each decision (S) concerning the stockpile of different periods (for simplicity, in Fig. 1 only three alternatives are presented) are associated with several oil prices and loss of GDP caused by shortages due to oil supply emergencies (again only three are presented). Any branch of this decision tree model expresses a scenario to be evaluated by the decision maker. 2.2. Methods and variables explanation In this paper, the economic factor considered is the shortage cost expressed as a function of GDP loss due to oil supply disruptions and the holding costs is expressed as a function of the level of strategic stockpile. This is a function by which the decision maker can evaluate each decision, i.e. each branch of the decision tree. Its general form is the following: TC ¼ SC þ EGL
ð1Þ
Where: TC SC EGL
total expected cost. cost of procurement and maintenance of the SPR. expected shortage cost.
(1) Expected shortage cost The sudden rising of oil prices creates three types of economic loss to the economy for oil imported countries: (1) transfer of wealth from oil imported countries to oil exported countries; (2) loss of the potential GDP; (3) macroeconomic adjustment loss. The transfer of wealth is a transfer of payment, and is equal to the quantity of oil import multiplied by the difference between the monopoly price and the competitive market price of oil. The GDP loss considered in this paper refers to the sum of the previously discussed three losses. EGL ¼
3 X
qi ⁎GDP⁎Q
ð2Þ
i¼l
Q ¼ r⁎e⁎c
ð3Þ
e ¼ bp þ by roil
ð4Þ
bp ¼ bd þ bs
ð5Þ
294
Y.-M. Wei et al. / Energy Economics 30 (2008) 290–302
Concentrating on the economic aspects of vulnerability of the oil price increase, it can be argued that these basically involve the decreasing GDP caused by oil shortages. The loss of potential GDP is equal to producers' and consumers' surplus losses with the economy in all markets affected by the oil price increase (Greene et al., 1998). So the total elasticity of oil price changes is equal to the elasticity of price with respect to demand and the elasticity of price with respect to supply surplus. (2) Cost of procurement and maintenance of the SPR SC ¼ IC þ RC þ OC
ð6Þ
IC ¼ Pj ⁎R⁎D
ð7Þ
RC ¼ Pj ⁎R⁎D⁎m
ð8Þ
OC ¼ Pj ⁎R⁎D⁎n
ð9Þ
The definition of all variables in this cost function can be seen in Table 2. (3) Decision analysis for the optimal SPR level Because decision-makers do not know exactly what oil supply disruption will occur and what oil prices will be in the future, they can assume or reckon the possibility of possible disruptions and prices based on experience and the history of disruptions, then make decisions. The optimal SPR is a decision with risk. There are many risks for decision-makers to consider prior to establishing SPR: the GDP loss will be too large if long-term oil supply disruption occurs and the size of SPR is too small in the future; the expensive reserve investment and maintenance cost for Table 2 The definition of variables and parameters Variable/ parameter
Explanation
IC RC OC Pj R D m n Q r e c
Oil import cost. Investment cost (building reserve infrastructure). Operation and maintenance cost. The oil price, refers to 20$/bbl, 35$/bbl, 50$/bbl respectively. Everyday net oil import. The days of net oil import of SPR. The assumptive ratio between investment cost and import cost. The assumptive ratio between operation and maintenance cost and import cost. The GDP loss ratio during oil supply disruption. The oil price increase when oil supply disruption occurs. The loss of GDP caused by the fluctuation of oil price (Greene et al., 1998). Constant, refers to the GDP loss rate when SPR is released, which is estimated based on Greene et al. (1998) and Teisberg (1981). The elasticity of oil price with respect to GDP (Greene et al., 1998). The elasticity of oil consumption with respect to GDP (Greene et al., 1998). The elasticity of GDP with respect to oil price (Greene et al., 1998). The elasticity of oil price with respect to demand (Greene et al., 1998). The elasticity of oil price with respect to supply (Greene et al., 1998). The possibility of oil supply disruption, i = 1,2,3 refers to short-term shortage, medium-term shortage, long-term shortage respectively.
βp βy σoil βd βs ρI
Y.-M. Wei et al. / Energy Economics 30 (2008) 290–302
295
establishing SPR will be wasted if oil supply disruption will not occur or the short-term disruption will occur in the future. Therefore, this paper analyzes the optimal SPR within scenarios of different oil prices using decision tree analysis. The decision tree is the figure form of expected value (EV) method. The decision dots are presented in small panes; the arcs linked with decision dots refer to possible SPR decisions, that are 30, 60, and 90 days of net oil imports. The remaining ends of arcs are condition dots presented in circles; the arcs linked with condition dots are possibility arcs above which show possible oil supply disruptions. The other ends are result dots presented by small triangles; there is a figure behind each which refers to the cost of every decision under different conditions (Figs. 2–5). When the possibility of oil supply disruption occurring is ρ, then the expected cost of every decision is E(α,ρ) and total expected cost of national SPR is E(α). Because the possible condition is discrete, the possibility of condition i is ρi. Then X EðaÞ ¼ qi Pða; iÞ; 8aaSl ð10Þ iaS2
Where S1 S2
is the set of every decisions; is the set of different conditions.
Fig. 2. The decision tree of SPR analysis for China in 2010 in scenario I. Note: The other units are always in 100 million dollars, besides possibilities.
296
Y.-M. Wei et al. / Energy Economics 30 (2008) 290–302
Fig. 3. The decision tree of SPR analysis for China in 2020 in scenario I.
When minimizing total cost, then decision α⁎ is optimal. min EðaÞ ¼ Eða⁎ Þ
aaS1
ð11Þ
2.3. The assumptions The empirical analysis in this paper is based on the following assumptions: (1) China's SPR base is establishing and has no stockpile. So we do not know how much the exact maintenance and investment costs are. We assume that the proportion among oil import cost, investment cost and maintenance cost in China's SPR investment is 75.4%, 22.9%, 1.7% respectively, which refers to the experience of the US' SPR. And the estimated (or assumed) value of the other parameters in cost function can be seen in Table 3. Greene et al. (1998) posited that although the SPR can be effective against a short-lived, random shock, it is relatively ineffectual against a strategic shock of two years or more. It is quite clear that SPR alone cannot resolve a major supply emergency. So the SPR cannot by itself provide long-term security. But Samouilidis and Berahas (1982) believed that the amount of the SPR held is a function not only of the holding costs and petroleum prices, but also of the displaced shortage costs (either monetary or social) and the other damages incurred by an energy emergency. So we consider both opinions.
Y.-M. Wei et al. / Energy Economics 30 (2008) 290–302
297
Fig. 4. The decision tree of SPR analysis for China in 2010 in scenario II.
(2) We assume that the impact of the SPR of 60 and 90 days of net oil import on the decreasing GDP loss are the same if long-term oil supply disruption occurs. (3) We assume that the impact of different SPR on the decreasing GDP loss is different whatever oil supply disruption occurs. (4) Based on some forecasts for China's future oil demand and production, we assumed that China's net import of crude oil is 130 million tons in 2005, 150 million tons in 2010, and 200 million tons in 2020 respectively. 2.4. The source of data The world oil prices, China's oil consumption and oil production for each year are taken from BP Statistical Review of World Energy 2004 (BP, 2004) and China Statistics Yearbook 2003 (State Statistical Bureau, 2004); the GDP of China is taken from China Statistics Yearbook 2003. The GDP and the oil demand of China in 2005, 2010 and 2020 are taken from Chen (2003). The historical information of oil supply disruption is taken from EIA (Energy Information Agency) of US Department of Energy. 3. Results and discussion The Chinese government began to establish national SPR in 2004. The biggest reserve will be finished in December 2006. So it is not possible to own national SPR now. The optimal SPR of
298
Y.-M. Wei et al. / Energy Economics 30 (2008) 290–302
Fig. 5. The decision tree of SPR analysis for China in 2020 in scenario II.
2005 calculated in this paper is theoretical (and is not discussed). The optimal SPR under different assumptions and oil prices in 2005, 2010, and 2020 can be found in Table 4. 3.1. Scenario I analysis For scenario I, the assumption that the impacts of the SPR of 60 and 90 days of net oil import on the decreasing GDP loss are the same if long-term oil supply disruption occurs. The results in 2010 show that: (1) The optimal SPR is 60 days of China's net oil imports when oil price is $ 20/bbl and the economy growth is around 7%; (2) The optimal SPR is also 60 days of China's net oil imports when oil price is $ 35/bbl and the economy growth is around 7%; (3) The optimal SPR is also 30 days of China's net oil imports when oil price is $ 50/bbl and the economy growth is around 7%. When oil price is $ 20/bbl and $ 35/bbl, the reserve cost per barrel is relatively low, the contribution of SPR of 60 days for decreasing the GDP loss caused by oil supply disruption is great compared with the SPR of 30 days, so the total cost for SPR of 60 days is less than that of 30 days. Because we assume that the contribution of SPR of 60 days and 90 days for decreasing the GDP
Y.-M. Wei et al. / Energy Economics 30 (2008) 290–302
299
Table 3 The estimated (or assumed) value of the parameters Parameter
Value
Scope
Notes
m
0.30
n βy
0.02 0.49
σoil βd
− 0.05 − 0.037
20$/bbl
− 0.068 − 0.099 0.023 0.032 0.056 62.6%
35$/bbl 50$/bbl 20$/bbl 35$/bbl 50$/bbl Short-term
18.7% 18.7%
Medium-term Long-term
βs
ρI
For the proportion of IC, RC, OC in the US SPR is 75.4%, 22.9%, 1.7% respectively. It is our calculation based on China's GDP, oil product, consumption, and international oil price from 1985–2003. The results from the US Department of Energy Annual Energy Outlook (1995) forecast (US Department of Energy, 1995)
The data from the US Department of Energy indicates that there are sixteen oil supply disruptions or shortages from the 1950s. According to the fluctuation of oil price, we calculate the probability of shortage.
loss caused by long-term oil supply disruption are the same, the reserve investment of 90 days is far more than that of 60 days. Therefore, the expected total cost of SPR of 60 days is lesser. When oil price is $ 50/bbl, the reserve cost per barrel is relatively high, and the GDP loss caused by oil supply disruption is less than the increased reserve investment for 60 and 90 days of SPR, although the contribution of SPR of 60 days for decreasing the GDP loss caused by oil supply disruption is great compared with the SPR of 30 days, so the total cost for establishing SPR of 60 and 90 days is larger than that of 30 days. Therefore, the optimal SPR in 2010 for China is 60 days of net oil import if the fluctuation of world oil price is $ 35/bbl or less; the optimal SPR for China should be 30 days of net oil import if the fluctuation of world oil price is $ 50/bbl or higher (Fig. 2). The results in 2020 show that the optimal SPR for China is 60 days of net oil import, when oil prices are $ 20/bbl, $ 35/bbl, or $ 50/bbl and the economic growth is around 7% (Fig. 3). It is estimated that China's oil demand will reach 450 million tons in 2020; and the domestic production will be 200 million tons. So the dependence on oil imports will be more than 55%, and the impact of the fluctuation of world oil prices on economic development will be greater than now; which makes the establishment of a larger SPR necessary. Theoretically, the greater the size of SPR in 2020 is, the lesser the GDP loss is. But the possibility of long-term oil supply disruption
Table 4 The optimal SPR in different assumptions and oil prices in 2005, 2010, and 2020 Year
2005 2010 2020
$20/bbl
$35/bbl
$50/bbl
Scenario I
Scenario II
Scenario I
Scenario II
Scenario I
Scenario II
60 days 60 days 60 days
60 days 90 days 90 days
30 days 60 days 60 days
30 days 90 days 90 days
30 days 30 days 60 days
30 days 30 days 90 days
300
Y.-M. Wei et al. / Energy Economics 30 (2008) 290–302
occurring is only 18.7% and the contribution of SPR of 60 days and 90 days for decreasing the GDP loss caused by long-term oil supply disruption are the same. The possibility of short-term oil supply disruption occurring is 62.6% and the reserve investment of 90 days is more than that of 60 days, although the contribution of SPR of 90 days for decreasing the GDP loss caused by oil supply disruption is greater than that of 60 days. So the total cost for establishing SPR of 90 days of net oil import is higher than that of 60 days. Therefore, the optimal SPR for China will be 60 days of net oil import in 2020 whatever the world oil price is. 3.2. Scenario II analysis For scenario II, the assumption that the impact of different SPR on the decreasing GDP loss is different when various oil supply disruption occurs. The results in 2010 show that (Fig. 4): (1) The optimal SPR is 90 days of China's net oil imports when oil price is $ 20/bbl and the economy growth is around 7%; (2) The optimal SPR is also 90 days of China's net oil imports when oil price is $ 35/bbl and the economy growth is around 7%; (3) The optimal SPR is also 30 days of China's net oil imports when oil price is $ 50/bbl and the economy growth is around 7%. We assume that the contribution of alternative SPR for the decreasing GDP loss is different when different oil supply disruption occurs. When oil price is $ 20/bbl and $ 35/bbl, the reserve cost per barrel is relatively low, the contribution of SPR of 90 days for decreasing the GDP loss caused by oil supply disruption is greater compared with the SPR of 60 and 30 days. So the total cost for SPR of 90 days is less than that of 60 and 30 days. Therefore, the expected least total cost of SPR is 90 days. When oil price is $ 50/bbl, the reserve cost per barrel is relatively high, and the GDP loss caused by oil supply disruption is less than the increased reserve investment of 60 and 90 days SPR, although the contribution of SPR of 60 days and 90 days for decreasing the GDP loss caused by oil supply disruption is greater compared with the SPR of 30 days. So the total cost for establishing SPR of 60 and 90 days is larger than that of 30 days. Therefore, the optimal SPR in 2010 for China is 90 days net oil import if the fluctuation of world oil price is $ 35/bbl or less; the optimal SPR for China should be 30 days' net oil import if the fluctuation of world oil price is $ 50/bbl or higher. The results in 2020 show that the optimal SPR for China is 90 days of net oil imports, when oil prices are $ 20/bbl, $ 35/bbl, or $ 50/bbl and the economy growth is around 7% (Fig. 5). Because the oil demand of China will be greater, the dependence on oil imports and the share of oil in the primary fuel mix will increase, and the impact of the fluctuation of world oil prices on economic development will be greater than now. This supports the argument for establishing a larger size of SPR. Therefore, the optimal SPR for China will be 90 days of net oil imports in 2002, whatever world oil price is. 4. Conclusions and further work The security value of strategic petroleum reserve is a very complex problem. In order to estimate the optimal SPR level, all factors related to SPR must be quantified. But this is difficult to accomplish. So we make a lot of assumptions. Although the approach omits many features of the
Y.-M. Wei et al. / Energy Economics 30 (2008) 290–302
301
problem, it considers the most important decision variables that greatly influence the optimal stockpile. Thus, it gives the decision maker basic information such as storage and shortage costs. For the same period, the optimal SPR of the lower oil price scenario is always larger than or equal to that of the higher oil prices scenario; which is in complete accord with the rule “buy low and sell high”. The optimal SPR will increase from 30 days to 90 days of net oil imports with greater requirements for energy and high dependence on oil imports, which is necessary for economic development and energy security. Our results show that the optimal SPR level is sometimes not to have the largest strategic stockpile. When oil prices are low, 60 days and 90 days of net oil imports are suitable for China's economic development and security in 2010 and 2020. When oil prices are high, the optimal SPR should be 30 and 60 days of net oil import for 2010 and 2020. Therefore, the oil prices can influence the optimal SPR. This paper discusses the impact of different world oil prices on China's SPR because the change of world oil price can influence import cost and, importantly, as the second biggest oil consuming country in the world, the establishment of China's SPR may influence world oil prices. This is one of the reasons why we considered the change of world oil prices in our model. But the decision model in this paper could not identify how much the world oil price will fluctuate. However, we might simulate the correlation between world oil price and China's strategic stockpile in further work. Acknowledgements The authors gratefully acknowledge the financial support from the National Natural Science Foundation of China (NSFC) under the grants Nos.70425001 and 70371064, the Key Projects from the Ministry of Science and Technology of China (grants 2001-BA608B-15, 2001-BA60501).Yi-Ming Wei truly appreciates the supports from Prof. Michael B. McElroy and Mr. Chris P. Nielsen at Harvard University. We also would like to thank Professor R.S.J. Tol and the anonymous referees for their helpful suggestions and corrections on the earlier draft of our paper according to which we improved the content. References An, Q.Y., 2003. It is no time to delay for establishing strategic petroleum reserve. China Youth Science and Technology 4, 18–20 (In Chinese). BP, 2004. BP Statistical review of world energy. British Petroleum (BP), London. Chao, H., Manne, A.S., 1982. Oil stockpiles and import reductions: a dynamic programming approach. Operations Research 31, 632–651. Chen, W.Y., 2003. Carbon quota price and CDM potentials after Marrakesh. Energy Policy 31, 709–719. Greene, D.L., Jones, D.W., Leiby, P.N., 1998. The outlook for US oil dependence. Energy Policy 26, 55–69. Paik, I., Leiby, B., Jones, D., et al., 1999. Strategic oil stocks in the APEC region. Proceedings of the 22nd IAEE Annual International Conference. International Association for Energy Economists. Ravindran, A., Phillips, D.T., Solberg, J.J., 1987. Operations Research: Principles and Practice, 2nd ed. John Wiley and Sons, Inc., New York. Samouilidis, J.E., Berahas, S.A., 1982. A methodological approach to strategic petroleum reserves. The International Journal of Management Science 10, 565–574. Samouilidis, J.E., Magirou, V.F., 1985. On the optimal level of a small country's strategic petroleum reserve. European Journal of Operational Research 20, 190–197. State Information Center, 2005. 2005 China's Oil and Gas Industry Annual Report. China Economic Publishing House, Beijing.
302
Y.-M. Wei et al. / Energy Economics 30 (2008) 290–302
State Statistical Bureau, 2004. China Statistical Yearbook 2003. China Statistical Press, Beijing. Teisberg, T.J., 1981. A dynamic programming model of the U.S. strategic petroleum reserve. The Bell Journal of Economics 12, 526–546. US Department of Energy, 1988. United States Energy Policy 1980–1988 DOE/S-0068. Washington DC, October. US Department of Energy, Energy Information Administration, 1995. Annual Energy Outlook 1995 DOE/EIA-0383. Washington DC, January. Vayssieres, M.P., Plant, R.E., Allen-diaz, B.H., 2000. Classification tree: an alternative non-parametric approach for predicting species distribution. Journal of Vegetation Science 11, 679–694. Zhou, D.D., 2001. Strategic petroleum reserve linked with national strategies-the roles of China's strategic petroleum reserve. China Petroleum 5, 15–16 (In Chinese). Zweifel, P., Bonomo, S., 1995. Energy security coping with multiple supply risks. Energy Economics 17, 179–183. Dr. Yi-Ming Wei is a Professor at the Institute of Policy and Management of the Chinese Academy of Sciences; and is currently a visiting scholar at Harvard University in the United States. Mr. Gang Wu is a Ph.D. candidate in Management Science at the Institute of Policy and Management, Chinese Academy of Sciences, China. Dr. Ying Fan is a Professor at the Institute of Policy and Management, Chinese Academy of Sciences, China. Her research field is energy policy and system engineering. In 2004, she was a visiting scholar at Cornell University, USA. Ms. Lan-Cui Liu is a Ph.D. candidate in Management Science at the Institute of Policy and Management, Chinese Academy of Sciences, China.