An analysis of the role of financialisation, China and stockholding in agricultural commodity price movements

A wide range of commodity prices experienced a large peak in 2007/08, including many agricultural commodities. Since then agricultural commodity prices have remained at relatively high levels compared to the recent past and several agricultural commodities (e.g. maize and sugar) have experienced further, although smaller peaks. Studies of the recent commodity price movements have focused on financialisation, China, low stocks and biofuels. This thesis explores the role of three of these factors in agricultural commodity price formation, namely, financialisation, China and stockholding. Biofuels were analysed in work for my previous employer. The second chapter uses Granger-causality methods to assess whether index investor positions influence agricultural futures prices. Four extensions are explored that might overcome the limitations of Granger-causality tests in this context. Firstly, the analysis is extended to less liquid markets. Liquid markets tend to be relatively efficient and for those markets the semi-strong form of the Efficient Market Hypothesis suggests that prices should not be forecastable as new information is swiftly impounded in prices. Even if index investment had price impact finding Granger-causality based on weekly data would be surprising. In less liquid markets prices are likely to adjust more slowly to new information. Three further extensions are examined: long-horizon tests, using relative returns instead of returns and an index of positions instead of positions in individual markets. The analysis supports the conclusion that no impacts are discernible for liquid markets. However, Granger-causality is established in the less liquid soybean oil and livestock markets. Thus, there is clear evidence that index investment has been a factor influencing the level of grains and livestock prices in illiquid markets over the five years 2006-11.These results lead to the conjecture that index investment does also have price impact in liquid markets but that market efficiency prevents the detection of this impact using Granger-causality tests. Chapter 3 investigates the potential impacts of changes in Chinese stockholding and self-sufficiency policies on world wheat prices. Whether or not the decrease in world grain stocks around the middle of the 2000s has had a significant impact on world prices in the last decade is controversial. Since most of the decrease in stocks was due to decreases in China, the main issue is the role that China has played on world markets. So far most studies that address this question have relied on informal arguments rather than formal modelling of the stocks-trade-price relationships. More recently the impact of possible changes to China’s self-sufficiency policy on world markets has become the focus of attention. Trade and other grains policies are closely linked. Any impact of changes in grain policy in China will be transmitted to the world market through changing trade patterns. The model shows that a move away from autarky reduces stock levels in China and in the rest of the world resulting in lower global stock levels. In the two scenarios where China imports but does not export global stocks decline by 23 and 32 per cent, respectively, compared to autarky. In the two free trade scenarios, global stocks reduce by 38 and 44 per cent, respectively, compared to autarky. These reductions in stock levels when a country moves from autarky to trading with the rest of the world do not lead to an increase in price variation and, more importantly for policy-makers, do not lead to increases in extreme price movements. In the free trade scenarios, extreme price movements are even reduced despite much lower stock levels. Chapter 4 introduces a new approach to testing the competitive storage model which has been the main workhorse of the analysis of the role of storage in commodity price formation over the last decades. The relationship between storage and prices is complex. Stock levels depend on current prices and expected future prices. At the same time, the current prices and the futures expected price depend on the level of stocks carried forward from the current period to future periods. The main approach to testing the competitive storage model has been the comparison of the characteristics of the predicted price series with actual commodity price series. In this study a different approach is taken. A relatively simple model is taken to the experimental laboratory. Participants in the experiment are asked to make storage decisions within a competitive storage model framework. Participants’ behaviour in the experiment deviated from the behaviour predicted by the competitive storage model in a number of ways. The predicted relationship between the amount of wheat available and storage is non-linear in the model but is linear in the experiment. In addition, storage is more sensitive to “wheat” in storage than “wheat” harvested when the model suggests that the effect of wheat in storage and wheat from harvest should the same. Furthermore, average storage tends below the optimal level and storage does not vary as much as predicted by the competitive storage model. The resulting price series tend to be more variable than would be the case if stockholders behaved according to the competitive storage model.