Futures market efficiency before and after a structural break in the market integration between Henry Hub and NBP

Abstract

The development in the natural gas markets throughout the past two decades is likely to have made an impact on the market characteristics. Commodity market financialization in the early 2000s and the shale gas revolution beginning around 2005 are the most significant developments worth mentioning. In this thesis we will investigate the impact on market integration between the US and the UK markets in the period February 1997 to March 2018. The US and UK markets are represented by Henry Hub and NBP prices, respectively. Furthermore, we will analyze the futures market efficiency in both markets, and how the efficiency have been affected by the development. We consider contract maturities of 1 through 8 months.

By using the Augmented Dickey-Fuller test, we found that most of the time series variables are non-stationary processes with order of integration equal to one. As a result we used the Gregory-Hansen test of cointegration with an unknown structural break and determined a statistically significant breakpoint in the cointegration relation between Henry Hub and NBP spot prices in October 2008. Further investigation with the Johansen test of cointegration revealed that the markets co-moved in the period February 1997 - October 2008, but that this co-movement has disappeared since. In our analysis of the futures market efficiencies we found that all futures contract prices are cointegrated with the respective spot price. In the Henry Hub futures market, the contract with 1 month maturity fulfilled the Law of One Price prior to the breakpoint and contracts with maturity 1-5, and 7 months did so after the breakpoint. No Henry Hub contracts were unbiased predictors of the future spot price. In the NBP futures market, there were no contracts fulfilling the Law of One Price prior to the breakpoint, but all contracts did so in the post period. NBP futures contracts with maturity 1 and 3 months were statistically significant in supporting the unbiasedness hypothesis in the period post breakpoint. As a result, both markets have improved their efficiency, especially the NBP market, but they are still to a great extent inefficient.