油品裂解價差之期貨避險部位設定

Abstract

摘要 原油與其包括如:無鉛汽油、燃油等產製類之油品，是世界上最重要的能源來源。由於其供應世界能源需求所佔之比例相當高，其他能源對油品全面的替代性並不大。國際間原油的供應大部份掌握於中東政局不穩定的國家，以致原油價格的政治敏感性強。另一方面，原油產製品的消耗具季節性現象，因此部份國家政經情勢對於原油價格的影響，未必可以同時且全然地反應於原油產製品的價格上。此種現象同時也反應於原油與其產製品價格，短期內未必能夠維持穩定的裂解價差(crack spread)關係。 在自由市場之油品裂解廠商，其生產毛利潤受國際間油品裂解價差值所影響，因此油品業者可藉由特定的油品期貨(oil futures)進行避險交易，選擇適當時點規避廠商裂解毛利潤波動的風險。本文就裂解價差之期貨避險交易進行研究，希望能針對價格關係發現與適當裂解價差避險比例等部份，提供需注意要項及對實際期貨操作者之建議。 在相關油品市場間價格關係發現方面，本研究運用供需結構與油品價格間的套利機制，發展相對模型以觀察與說明裂解價差交易的行為特性，並據以作為檢定市場是否出現特定價格行為的工具。在完全市場且無相關交易成本的假設下，將可推導等同於相關油品期貨價格存在市場關連性(market linkage)虛無假說之裂解價差數列單根行為假說。由於裂解價差值可衡量，此一模式也就提供了市場關連性可驗証的假說，並可應用裂解價差值的單根行為檢定模式，說明市場關連性程度的強弱。在實證資料方面，1985~1995年間，各油品期貨市場間價格，顯著性存在關連性與長期穩定線性關係。亦即表示長期而言，裂解價差時間序列為一穩定的隨機變數。此種行為特性不但隱含油品期貨裂解價差交易的可行性，也可運用於油品期貨裂解價差的避險交易過程，作為避險部位設定的調整因子。 在適當裂解價差避險比例的估算方面，裂解價差避險策略所交易的商品期貨價格的波動，受商品現貨價格、利率水準和商品耐儲特性的影響。一般而言，導源於商品現貨價格與利率水準所產生的投資組合報酬的不確定性，可以藉由同時反向購買商品期貨/現貨，以及利率期貨，予以對沖以減少其相對應的不確定程度。而商品的耐儲特性，乃指持有商品現貨到未來，所必需額外支出的儲存相關成本等特性，為商品本身的獨特因素，一般而言，無法透過直接期貨避險工具轉嫁相對應的波動程度。 本研究考慮相關油品價格的變異性、裂解價差時間序列行為特性，以及油品特殊的耐儲性等特性，運用隨機漫步行程(random walk processes)發展裂解價差避險比例(hedge ratio)模型。本研究並進行實證分析以說明所提模型的避險效果。運用此一裂解價差避險比例模型，廠商可依據其擁有之現貨部位與裂解價差避險比例模型，訂定適當的期貨避險部位。 在裂解價差避險部位設定方面，為改善固定值價差避險比例策略無法適用之限制，本研究運用相關油品反向部位以沖銷便利殖利率(convenience yield)變動的風險，代入特性方程式中，求取油品期貨的特性解。此種估計方法可以獲得解的完全形態，也可縮小期貨價格實際值與估計值間的誤差，以確保估計解本身的有效性，改進了Gibson and Schwartz對商品期貨價格的估計方法，並可據以求得裂解價差交易中適當的期貨部位。實証上，採用1985~1995年間紐約商品交易所之交易價格資料。在發展相關檢定方法，以確定本文模式對於實際交易價格的可適用性後發現，便利殖利率被檢定顯著性具有平均-反轉行為(mean-reverting behaviors)，其中成交量較大的商品，其表現出的套利活動明顯、調整速度值也較大。本文構築的隨機避險模式，相較於Ederington的靜態模式而言，在所探討的實証資料區間及相關商品中，顯著性具有較高的避險效果。在實証中也發現，隨機性避險模式，不論是在強關連性或弱關連性情況下，均能提供高的避險績效。在弱關連性的情境中，本研究所發展的避險策略，具體提供了相對於傳統價差策略更佳的避險方式。

Abstract Organization of Petroleum Exporting Countries (OPEC) dominates the supply of the oil all over the world. This makes crude oil prices very sensitive to the political situations in the Middle East. Furthermore, since there is seasonal effect in the consumption of refined oil product mixes, the reaction of new information might not be immediate and efficient as that for the crude oil. Crack spread, a kind of commodity spread strategy, permits traders to take positions both on refining profit margins and refined product mixes. It is an important tool to energy complex hedgers and has become popular among sophisticated speculators. Much of the published works has focused on basis hedging, while theoretical analysis of spread hedging strategies has only received very limited attention. Although most textbooks only deal with one-to-one positions in spreading, more sophisticated approaches such as Shutz (1984) and Poitras (1989) have discussed some fixed-ratio spread strategies. As no convergence has been reached in spread hedge, no strong market linkage between related commodities in inter-commodity spread trading seems to exist. This research will explore the issue of low hedging effectiveness for traditional spreading under weak market-linkage environment. To explore the market linkage mechanism and crack spread hedging positions in oil futures trading, this paper extends the market linkage model (Garbade and Silber, 1983) using crack spread behaviors with tri-commodity hedging strategies. Using the concepts of gross refining cost and crack ratios, a trading-price market linkage model is developed to describe the behaviors of oil futures market linkage. Since trading prices of related futures are considered, our model is not only testable but can also be applied to observe the linked degree of related futures markets and the influential directions of tuning structures. Crack spread series is proved to provide a unit root-based approach to test market linkage hypothesis in a perfect market. Its first-order autoregressive coefficient can be employed to measure the degree of linkage. The cointegration test (Engle and Granger, 1987) helps to set up a kind of stable relationship. It is inferred theoretically that the increase in last-period crack spread should leads to an increase in contemporary crude oil futures prices and a simultaneous decrease in refined oil futures prices as well. The empirical results show that the influential directions due to variation in crack spread are consistent with our model. Additionally, it is confirmed that a stable relationship exists and some dynamic tuning structures are established. In order to improve the hedging effectiveness in spread trading, the behavior of crack spread series is also examined, tested and modeled. Confirmed stationarity for crack spread series is indispensable for building a stochastic spread hedging model. The crack spread series are assumed to follow random walk processes with approaching zero growth rates. After modeling the stochastic processes of concerned variables, we form a zero net investment portfolio to derive the characteristic equation of oil contingent claims. The minimum-variance crack spread hedge positions is then determined as a function of the values of crack spread. The empirical results, which is consistent to our theoretical model, show negative crack-spread hedge ratios. The convenience yields evidently exhibit mean-reverting behaviors. Besides, higher crack spread values always correspond to smaller absolute values of minimum-variance positions. This empirical analysis also shows that our stochastic crack spread hedging model outperforms the traditional one in terms of hedge effectiveness. The proper values are not equal to one, but locate in the range of -0.8~-0.9. It is suggested that the refiners should hold the crack spread portfolios according to our estimated hedge ratios to further ensure the production profit margins and reduce the variance of portfolio's return.