Drivers of agricultural commodity prices since 2000

Table of Content

ABSTRACT

ACKNOWLEDGMENTS

TABLE OF FIGURES

LIST OF TABLES

1 INTRODUCTION

2 LITERATURE REVIEW
2.1 Fundamental and political factors influencing food prices
2.1.1 Demand from developing countries
2.1.2 Increasing oil price
2.1.3 Expanded biofuel production
2.1.4 Low stocks in agricultural commodities
2.1.5 Adverse meteorological conditions
2.1.6 Food policies
2.1.7 Other fundamental factors
2.2 Financialization in the agricultural commodity market
2.2.1 Increasing financial speculation has no impact on agricultural commodity markets
2.2.2 Increasing financial speculation is beneficial
2.2.3 Financial speculation is harmful to agricultural commodity markets
2.3 Evidences for the impact of investor’s sentiment on asset prices

3 ANALYSIS OF POTENTIAL SENTIMENT PROXIES
3.1 AAII Investor Sentiment Survey
3.1.1 AAII data and limitations
3.2 Implied volatility of S&P500 index options
3.3 First-day returns on initial public offerings
3.3.1 First-day IPO returns data and limitations
3.4 Closed-end fund discount
3.4.1 Closed-end fund discount data and limitation
3.5 Open interest in the wheat futures market
3.5.1 Open interest data and limitations
3.6 Trading volume in the wheat futures market
3.6.1 Trading volume data and limitation
3.7 Wheat futures returns
3.7.1 Limitations of wheat futures returns

4 EMPIRICAL EVIDENCE
4.1 Descriptive Statistics of relevant variables
4.2 Construction of the composite sentiment index
4.2.1 Preliminary sentiment index
4.2.2 Controlling for macroeconomic influences
4.2.3 Controlling for the capital market
4.2.4 Comparison to other sentiment indices
4.2.5 Does the sentiment index capture mood swings in the wheat market?
4.3 Interdependence between investor sentiment and returns on wheat futures ..
4.3.1 Interdependence between SENT┴# and returns on wheat futures
4.4 Dependence between investor sentiment and returns on wheat futures
4.4.1 Simple linear time series regression analysis
4.4.2 Multiple linear time series regression analysis

5 CONCLUSION

6 REFERENCE LIST

7 APPENDIX A: FUTURES MARKET TERMINOLOGY

Abstract…

Since the prices in the agricultural commodity market surged in 2008 and 2011 many articles investigated these turbulences from different perspectives. However, neither fundamental factors nor increased financial speculation provides a completely satisfactory explanation on this complex topic. Our study therefore tries to capture all these factors in a single approach and investigates the impact of investor sentiment on wheat futures returns. We construct an investor sentiment index out of monthly data from the period 2000 to 2013 by conducting a principal component analysis (PCA) with a set of well-established sentiment proxies. In particular, we employ three equity market proxies suggested by Baker & Wurgler (2007) and three wheat market specific sentiment proxies, in order to obtain a tailored investor sentiment index.

The sentiment index is statistical significant at the 95 per cent significance level and predicts about 2.45 per cent of the total variation in the subsequent month wheat futures returns. The results are also robust after controlling for fundamental factors in an extended multiple linear regression analysis. Our sentiment index therefore proves that investor sentiment impacts wheat futures returns, although further research is needed to verify this relationship.

Acknowledgments

I would like to express my gratitude to my supervisor Dr. Peter Moles for the useful comments, remarks and engagement through the learning process of this master thesis. Furthermore I would like to thank Dr Maria Boutchkova for the support on the way and for the help in various statistical issues. Also, I like to thank my friends, particularly Ms. Spro, who helped me to lift me to lift the language style of this dissertation to an academic level. I would like to thank my family, who have supported me throughout entire process and who made my dissertation possible in the first place. I am very grateful for your support. Lastly, I want to thank the ‘Secret Room’ of the University of Edinburgh - Business School in all its inhabitants for spending the majority of the last year with me.

Table of Figures

Figure 1: Contango/Backwardation and spot prices in the wheat market 3 Figure 2: Open interest and volume in the wheat futures market 2000 - 2013 ..

Figure 3: Open interest and settlement price of wheat

Figure 4: Weekly Volume-to-Open Interest ratio for the wheat market

Figure 5: Masters (2008) arguments how speculation impacts food prices

Figure 6: Worldwide meat consumption by area 1975 - 2012

Figure 7: Oil and food price development since 2000

Figure 8: Biofuel transmission mechanism

Figure 9: Wheat production and beginning of the year stock 2000 - 2013

Figure 10: Worldwide production of corn and wheat, 2000 -2010

Figure 11: Proxies used by Gao & Süss (2012)

Figure 12: Comparison S&P500 index and AAII-Bullish investors 2000 - 2013

Figure 13: VIX for the period 2000 to 2013

Figure 14: VIX and Open interest in wheat futures market

Figure 15: Monthly IPO levels and first-day returns in the US 2000-2013

Figure 16: Closed-end fund discount in the US since 2000

Figure 17: NC_L-S and Wheat Settlement price 2004-2013

Figure 18: Trading volume of the front month wheat contract 2000-2013

Figure 19: Comparison of SENT┴# to the sentiment index of Baker & Wurgler

Figure 20: Comparison SENT┴ and SENT┴# 2000 - 2013

Figure 21: Bull-spread┴# 2000-2013

Figure 22: S&P500 index vs. SENT┴# 2000 - 2013

Figure 23: Wheat settlement prices vs. SENT┴# 2000 - 2013

Figure 24: SENT┴# and settlement prices of wheat futures 2006-2013

List of tables

Table 1: Average consumption growth of selected agricultural commodities

Table 2: ‘Bspread’ summary table

Table 3: Historical Performance AAII Sentiment Survey

Table 4: ‘VIX‘ summary table

Table 5: ‘IPOr‘ summary table

Table 6: First-day IPO returns in the US summary statistics

Table 7: CEFdis/CEFdif summary table

Table 8: NC_L-S summary table

Table 9: VOL / VOLstand summary table

Table 10: Wheat futures returns summary table

Table 11: Descriptive statistics of all relevant variables

Table 12: Component loadings of SENT

Table 13: Correlation of raw sentiment proxies

Table 14: Component loadings of SENT (differences)

Table 15: Component loadings for SENT┴

Table 16: Correlation of SENT┴ proxies

Table 17: Component loadings for SENT┴#

Table 18: Correlation of SENT┴# proxies

Table 19: Comparison table of scaled SENT┴# and scaled BW index

Table 20: Summary statistics SENT┴ and SENT┴#

Table 21: Correlation between wheat futures variables and SENT┴#

Table 22: Correlation matrix between SENT┴+ proxies and rt+1

Table 23: Simple linear regression analysis of rt+1 on SENT┴#

Table 24: Multiple regression analysis of rt+1 on macroeconomic factors and SENT┴#

Table 25: Comparison of AIC and BIC values in both extend models 61

‘ We have good analysis that speculation played a role in 2007 and 2008.

Speculation did matter and it did amplify, that debate can be put to rest. These spikes are not a nuisance, they kill. They ’ ve killed thousands of people. ’

Joachim von Braun -

Director of Germany’s Center for Development Research

‘ Speculators create the bubble which lies above everything. They increase prices with their expectations, with their bets on the future, and their activities distort prices, especially in the commodities sector. And that is just like secretly

hoarding food during a hunger crisis in order to make profits from increasing prices. ’

George Soros -

Chairman of Soros fund management

‘ I see so much focus on food, and it seems to be so trendy in the investment world (...). The underlying dilemma has been created by the wealth of the BRICs (Brazil, Russia, India, China) countries; but, for the past year or so, it ’ s also been a major theme for financial institutions. The markets seem to me to

have a bubble-like quality ’

Jim O’Neill -

Former chief economist at Goldman Sachs

1 Introduction

World market prices for several staple foods reached an all-time high in 2008 with price increases of up to 431 per cent (rice) compared to the price level in January 2002 (Trostle 2011). This enormous price surge particularly affected people in developing countries, who spend a high percentage of their total income on food. Afterwards food riots and demonstrations against rising food prices caused political instability and social unrest in some parts of Asia and Africa. For instance, in Bangladesh, 20 people were injured by the police while protesting against rising food prices (Al Jazeera 2008). High agricultural commodity prices are even thought to have played an important role in the emergence of the ‘Arab Spring’ in 2010, because many North African countries have to import a majority of their food at world market prices (The Economist 2012).

The global media drew attention to these protests and investigated the causes of soaring prices for staple crops. A large public debate emerged, attempting to explain the ‘food price crisis’. A shortcoming of the debate was that it gave primarily consideration to only two factors. First the growing demand from developing countries like China was thought to be the main fundamental driver of food prices. Second the ever increasing speculation by financial institutions was brought to light by the NGOs and media, accusing the speculators of distorting commodity prices. The reputation, and trustworthiness of financial institutions was already damaged due to the Financial Crisis of 2008 and after. Their reputation was further damaged when prominent people, working within the financial industry, agreed that speculation had boosted food prices. But this highly complex topic was often simplified down to easy, one- dimension, causalities, and lacked findings based on in-depth analysis. Resentment of financial speculators is not a new phenomenon, dating as far back as Hume (1888) and Smith (1893), who both already accused commodity traders of causing market distortions. This mistrust is often built on insufficient information and knowledge about the workings of the financial markets, and particularly the futures market.

In futures markets, participants can enter a futures contract to buy (or sell) a predetermined quantity of a commodity (underlying) at a specific date in the future. No cash is exchanged when the contract is opened and the value at inception is zero. For instance, farmers can enter a short position to sell their crop half a year later for a fixed price and therefore reduce the risk of falling spot prices. A futures contract is thus a bet on the future spot price of the underlying commodity (further terminology is illustrated in the Appendix A).

Speculators entering a contract in a long position gain a reward for bearing the risk of future price fluctuations. The theory of normal backwardation (Keynes 1930; Hicks 1939) states that today’s futures prices are set below expected future spot prices and speculators earn, on average, a premium for bearing additional risk. Thus, a speculator’s realised return consists of two components. One, an insurance premium received from the hedgers. Two, any unexpected deviation of the future spot price from the expected future spot price (Gorton & Rouwenhorst 2006). This form of active speculation, where speculators form their opinion based on future price developments and changes in the fundamentals, is important and well-accepted, in order to supply the market with liquidity.

New players started entering the futures markets for agricultural commodities following a different, more passive, investment strategy. These new investors are termed Commodity Index Traders (CITs), and they pursue a strategy that promotes holding long positions continuously and rolling them over regularly. CITs do not purchase the underlying commodity itself, instead they roll-over their contracts before they expire. Thus, these investors earn a positive ‘roll-return’ if the contract they are selling (front month contract) has a higher price than the contract they are buying (second nearest contract). This is the case when markets are in backwardation, meaning the spot price is higher than the futures price (not to be confused with ‘normal backwardation’). Backwardation occurs typically in markets distinguished by low inventory and where insecurity about future outcomes (e.g. harvests) is high. The owner of a consumption asset has therefore a benefit from holding the actual commodity instead of a futures contract (called convenience yield). When prices of a

futures contract rise above spot prices (contango), CITs generate a loss when they roll over their contracts (Domanski & Heath 2007). In agricultural futures markets such as that for wheat, both patterns appear regularly over time, but since 2008 the market has been persistently in contango (see Figure 1).

illustration not visible in this excerpt

Figure 1: Contango/Backwardation and spot prices in the wheat market

Since 2003, index funds and exchange-traded funds tracking commodity indices, such as the S&P Goldman Sachs commodity index, became very popular and attracted about $440 billion of capital inflows (Berthelsen 2013). One reason might be the returns of commodity index investments, which provide a very good diversification opportunity for institutional investors. There is evidence for a long and persistent negative correlation with bond and equity portfolios. Commodity futures returns are also positively correlated with inflation and unexpected inflation (Gorton & Rouwenhorst 2006). In times of high price inflation commodities futures might therefore act as insurance. The period of expanding CIT activities, accompanied with increasing trading volume and open interest, is called the ‘financialization’ of the commodity futures market.

During the financialization period, open interest in agricultural futures market began to build up. They reached a peak in 2008 before bottoming out in 2009, experiencing another peak in 2011 (Figure 2). Likewise, trading volume increased sharply, with remarkable spikes in 2008 and 2010.

illustration not visible in this excerpt

Source: Datastream Both variables are measured monthly for the closest to the 5th closest wheat contract

Figure 2: Open interest and volume in the wheat futures market 2000 - 2013

The growth in open interest coincided with a lagged surge of settlement prices for the nearest available future contract (Figure 3). This obvious pattern led many observers to believe there was a direct connection between increasing market activity and spiking food prices.

illustration not visible in this excerpt

Source: Datastream Open interest is measured monthly for the closest to the 5th closest wheat contract

Figure 3: Open interest and settlement price of wheat

According to Bessembinder & Seguin (1993), increasing trading activity amplifies price volatility, but price volatility might also be diminished by large open interest in the market. One measure of activity is the trading volume-to- open interest ratio. The ratio fluctuates around an average of 4.59 in a range of

1.52 - 8.25. With the entrance of CITs in the market the ratio increased, but before the prices began to drop in 2008, trading activities nearly came to a halt 4

and the ratio plumbed its lowest level (Figure 4). Afterwards, turnover recovered again and the ratio levelled out at its old average of about 4.5.

illustration not visible in this excerpt

Source: Datastream

Figure 4: Weekly Volume-to-Open Interest ratio for the wheat market

A further issue in the agricultural commodity market in recent years, concerns the failure of futures and spot prices to converge at expiry date. Convergence is an important requirement in futures contract pricing and might therefore been interpreted as a sign of market distortion (Meijerink et al. 2011).

After the first slump in food prices in the first quarter of 2008, politicians in the US drew an interest in the matter and portfolio manager Michael W. Masters was questioned before the US Senate in regards to the influence institutional investors had on food and energy price inflation. Based on facts such as presented above, Masters (2008) argued that due to the rapid inflow of money from institutional investors, commodity futures prices surged and the futures curve shifted from backwardation to contango. Masters further argues that spot market participants (e.g. farmers) might have misinterpreted this false price signal as the result of changes in commodity fundamentals - wanting to take advantage of prospective price increases, they stored their harvest in order to sell it at a later date at a higher price (hoarding). This in turn led to an increase in the spot market prices because supply was withdrawn from the spot market. Based on this line of argument, Masters (2008) postulates that financial speculation caused global hunger and should be regulated tightly.

illustration not visible in this excerpt

Figure 5: Masters (2008) arguments how speculation impacts food prices

Masters (2008) bases his arguments on simple correlation measurements which do not reveal anything about causality. Several of his arguments are not scientifically proven. For example, the transmission mechanism of prices from the futures to the spot market is not as simple and clear as he claims. Furthermore, rising demand in futures contracts does not necessarily increase prices, because futures contracts are not a scarce good and each additional long position (demand) is matched by a supplementary short position (supply).

The following literature review gives an overview of current research about the drivers of agricultural commodity prices since 2000 and tries to shed light on all the issues mentioned above.

‘ The factors driving current food price increases are complex. We make no attempt to calculate what percentage of price changes are attributable to the many disparate causes, and, indeed, think it is impossible to do so. ’

(Abbott, Hurt & Tyner 2008, p. 8)

2 Literature review

A variety of academic papers, civil organization reports and newspaper articles examine the turbulence in the agricultural commodity market in the years 2008 and 2011 from different perspectives and with mixed objectives. The reasoning of the causes of these events is along a broad spectrum which includes fundamental factors, external macroeconomic effects and political as well as financial aspects. All of these factors are presumed to interact.

Mostly it is not simply one factor which can be held responsible for increasing agricultural commodity prices and it is also difficult to unwind each of them unequivocally or pool them in one single model. Although the majority of commentators agree that the long-term underlying price trend is based on fundamental economic factors, several alternative explanations are mentioned where financial speculation takes a distinctive role (Timmer 2009; Meijerink et al. 2011).

In order to maintain a high-quality level of research the results of the papers incorporated in this literature review had to be based on a consistent theory as well as an accurate empirical analysis. Thus, several non-academic reports were not taken into consideration.

2.1 Fundamental and political factors influencing food prices

2.1.1 Demand from developing countries World China India

illustration not visible in this excerpt

Source: USDA, World Bank, Baffes & Haniotis (2010), Own calculation; Numbers are the average growth per period

Table 1: Average consumption growth of selected agricultural commodities

Relatively soon after the price surge in 2008, academics searched for explanations of what had actually caused the price spike. A key argument relates to the story of ‘the march of the meat-eating Chinese — that is, the growing number of people in emerging economies who are, for the first time, rich enough to start eating like Westerners.’ (Krugman 2008) (Table 1).

China European Union North America South America Rest of the World

illustration not visible in this excerpt

Figure 6: Worldwide meat consumption by area 1975 - 2012

According to this line of argument, the emerging middle-class in developing countries shifted their diet towards dairy products and, in particular, meat (Braun 2008), which increased the demand for grain for animal food (Trostle 2011). One impact of this change might be a higher demand for grains, because to produce 100 calories of red meat calls for an equivalent of approximately 700 calories of grain (Piesse & Thirtle 2009).

However, Baffes & Haniotis (2010) report that at the same time as the spot market price of rice and wheat rose exorbitantly, the consumption of these products flatlined in China. Net import of wheat into the Indian and Chinese market furthermore accounted for only 0.17 per cent of global production between 2007 and 2010 (Worthy 2011).

Although Wright (2011) questions the accuracy of the numbers provided by Baffes & Haniotis (2010), he admits that the demand of these emerging states rose only gradually over recent decades (Figure 6). Consequently, the information should have already been factored into the prices before the spike. This is the reason Worthy (2011) concludes that other factors must have triggered the dramatic rise and fall of wheat and other agricultural products in 2008 and 2011. Still, the additional demand by India and China, constituting approximately 37 per cent of the global population (United Nations 2013), might impact food prices indirectly, either through energy consumption or, more specifically, oil prices (Headey & Fan 2008; Piesse & Thirtle 2009; Baffes & Haniotis 2010).

2.1.2 Increasing oil price

illustration not visible in this excerpt

Figure 7: Oil and food price development since 2000

When looking at Figure 7 it is tempting to argue that there is a correlation between oil prices and food prices in general. Especially since oil products, such as fuel and fertilizer, account for a high percentage of farmers’ variable costs (Avalos 2013). According to Mitchell (2008), oil-related input costs increased production costs in the US agriculture industry by 15 to 20 per cent from 2002 to 2007. Furthermore, Baffes (2007) estimates that 17 per cent of the oil price increase is passed on to agricultural commodity prices. Nazlioglu (2011) proves a non-linear causality from oil to agricultural commodity prices which, in his opinion, explains the surge in the food prices in 2008, although he does not exactly quantify the effect.

Another possible transmission of rising oil prices to the agricultural markets might be due to a decline in the usage of fertilizer, which in turn led to a lower yield and therefore a lower supply. However, a cut in fertilizer use should normally cause a lower output in the following year. Yet after the high oil prices in 2008 the harvest was relatively good, contradicting this argument (Wright 2011).

An indirect relationship between oil and food prices might exist through the biofuel market. For instance, after the rollout of new ethanol promotion policies in the US in 2006, Du, Yu & Hayes (2011) detect higher volatility spillovers amongst the crude oil and the crops markets. They argue that the tighter interdependence of these markets is caused by the extensive ethanol production. The mechanism behind this is described as follows:

illustration not visible in this excerpt

Figure 8: Biofuel transmission mechanism

2.1.3 Expanded biofuel production

Additional policy implementations in the EU and the US increased the mandatory quota of biofuel in petrol which might have allocated the agents of production (i.e. labour, acreage, capital) away from the production of food- related items (Mitchell 2008). Braun & Tadesse (2012) forecast that if this trend continues the price of maize, oilseeds and sugar will increase by more than 11 per cent before 2020. Collins (2008) estimates as well that around 60 per cent of the growth in maize prices in the period 2006-2008 was due to the additional use of maize for biofuels.

However, Baffes and Haniotis (2010) doubt that biofuel production caused the rise in prices in 2008, basing their argument on official data indicating that only a small proportion of total acreage is actually used for biofuels.

Wright (2011) imbeds the discussion of the impact of biofuel in the theory of storage (Working 1949; Fama & French 1987) where biofuel usage causes declining inventories for crops, making the markets more prone to external shocks such as droughts.

2.1.4 Low stocks in agricultural commodities

illustration not visible in this excerpt

Figure 9: Wheat production and beginning of the year stock 2000 - 2013

Low levels of stocks might have influenced prices, making them more prone to external factors. Theory suggests however that inventories build up when a bubble is present, which was not the case before 2007/08 (Figure 9). Moreover, it is difficult to obtain reliable and accurate data on stock levels, in particular in markets such as rice, where mainly small farmers, processors and traders are involved in the production process so there is hardly any available data on inventory levels and meaningful analysis is therefore nearly impossible (Timmer 2009). Inventories might pile up at several points along the production chain without being reported in any official statistics. As a side effect, the convenience yield (i.e. the utility of owning a commodity now) becomes higher when stocks are lower (Arbeiterkammer Wien & UNCTAD 2011).

Wright (2011) argues that decreased stocks in agricultural commodities, caused by substantial biofuels mandates, made the markets uncommonly sensitive to various short-run disorders such as the Australian droughts.

2.1.5 Adverse meteorological conditions

illustration not visible in this excerpt

Figure 10: Worldwide production of corn and wheat, 2000 -2010

An external shock on the supply side for agricultural commodities might be extreme weather conditions, such as the three severe droughts in Australia between 2002 and 2008 and the very dry period in South America in 2009. According to an estimate by the World Bank, these may have lowered the world-wide wheat supply by 4 per cent (Mitchell 2008). The general aggregated production however follows an up-and-down pattern which can be seen in Figure 10. For instance, Headey & Fan (2008) state that various other production shortfalls in the last twenty years which were also severe (e.g. 2000/01), did not trigger a price peak. Although, the decrease in output in the affected countries was substantial, the harvest in other countries, such as Russia, proved decent and kept global demand stable. So Headey & Fan (2008) conclude that weather was not a significant key driver of the price surge in 2007/08 because it was not extraordinarily different from preceding years. They also acknowledge though that the low inventories during this time might have amplified the impact of adverse weather conditions. Furthermore, news about bad weather conditions might have led to an overreaction in the market for agricultural commodity prices (Barberis, Shleifer & Vishny 1998).

2.1.6 Food policies

In 2007/8, during price increases and adverse weather periods, countries, especially those that export rice, enforced export restrictions in order to keep supply stable in their domestic market (Trostle 2011) while some others imported more to provide enough food for their population. A study by Headey (2011) estimated the impact on world rice prices in 2007/2008, due to export restrictions and panic purchases, being about 140 per cent.

Likewise, Mitra & Josling (2009) researched the hypothetical price changes, excluding the export restrictions enforced by India, and suggest the price level would have been around 20 per cent lower. This is similar to Headey (2011), who estimated the impact of Indian export restrictions on rice price being about 27 per cent. Braun Torero, 2008 quantify the price increase of grain caused by the export bans during 2007/8 to approximately 30 per cent and a more recent study by Yu et al. (2011) estimates the impact of export restrictions on wheat and rice prices to be 14 and 24 per cent, respectively.

The literature more or less agrees that political actions influenced prices and that the shift towards more national self-sufficiency had important adverse effects on agricultural commodity prices (Headey & Fan 2008). Headey (2011) also analysed the situation in the context of game theory and concluded that the enforcement of export restrictions by trading partners was perfectly rational behaviour considering the theoretical argument that when one big trading country introduces trade bans, this is likely to have a ripple effect.

2.1.7 Other fundamental factors

Other factors contributing to food price spikes are, for instance, global central bank policies (Belke, Bordon & Volz 2013; Frankel 2006) and the depreciation of the US dollar (Gilbert 2010a; Mitchell 2008; Meijerink et al. 2011).

2.2 Financialization in the agricultural commodity market

In the large body of literature about the financialization of agricultural commodities, there are two comprehensive literature reviews available, namely Shutes & Meijerink (2012) and Glauben et al. (2013). The following overview of the current literature is inspired by the two papers mentioned above.

2.2.1 Increasing financial speculation has no impact on agricultural commodity markets

Glauben et al. (2013) analyse 35 empirical studies which were published in the period 2010 to 2012, their examination gives an overview of the current state of research. According to their findings, financial speculation has neither a significant influence on the levels nor on the volatility of the price of securities in the agricultural commodities market. Most of the studies examined by Glauben et al. (2013) find fundamental factors to be the cause of the price increases in 2008 and 2011. In particular, peer-reviewed and quantitative research papers find only a few connections between speculation and the spot price level or price volatility of agricultural commodities. On the other hand, non-published articles or the outputs of NGOs often support the ‘Masters hypothesis’.

2.2.1.1 Studies employing Granger causality tests

By applying Workings’ (1960) speculative ‘T-index’, Sanders, Irwin & Merrin (2010) ascertain that there was no ‘excessive’ speculation in the nine commodity futures markets studied in the period 2006-2008. Furthermore, a bivariate Granger causality test reveals that CITs follow a trend rather than setting one. The Granger Causality test set out to identify causalities between time series. This test was chosen since it has been argued that it is one of the most reliable bivariate causality tests.

In order to discover whether these CITs are a disruptive force in the commodity futures markets, Stoll & Whaley (2010) undertook six different tests for 2006-2009, whereof four are simple correlation measurements (e.g. co- movements of futures prices) but in addition they conducted Granger causality tests. Their analysis suggests that ‘the futures markets are deep and fully capable of absorbing commodity index investment rolls for most commodity futures markets.’ (Stoll & Whaley 2010, p. 66). Correlation measurements have only weak explanatory power however and reveal no causal links. In comparison, Gilbert (2012) performs a variety of Granger causality tests and refutes the ‘Masters hypothesis’ for a dataset of grains and vegetables oils future markets in the years 2006-2011. Likewise, Irwin & Sanders (2011b) are not able to identify any causality between increased activities of CITs and corn, soybean and wheat futures prices. Gilbert & Pfuderer (2012) have extended Irwin & Sanders’s study (2011b) to less liquid markets and find that the investments of commodity traders in these markets does affect the returns. They suspect as well that CITs have the same impact on more liquid markets but the common Granger causality test fails to find proof of this. The strength of this paper is that they explicitly mention the shortcomings of their analysis and accept that various potential causes might be the trigger of the recent price surges though common statistical tests fail to show the quantity of this impact.

Aulerich, Irwin & Garcia (2013) use private data to directly test the ‘Masters hypothesis’. They examine whether the increased activity of non- commercial players in the futures market for agricultural commodities had an influence on the volatility as well as the prices during the period January 2004 to September 2009. In particular, they find that the null hypothesis stating that aggregate positions of CIT have no impact on daily returns cannot be rejected for 9 out of 12 agricultural commodity markets. They conclude that ‘The Masters Hypothesis is simply not a valid characterization of reality.’ (Aulerich, Irwin & Garcia 2013). Capelle-Blancard & Coulibaly (2011) obtain comparable results using panel data for the period from 2006 to 2010. However, they see fundamental factors such as supply and demand as the main drivers for the surge in spot prices of these commodities during the first peak in 2008.

2.2.1.2 Studies using different econometric methods

The report of Irwin & Sanders (2011a), analysing theoretically as well as empirically the ‘Masters hypothesis’, gained a lot of public attention and reached a broad audience being published under the name of the OECD. The authors outline, based on the theory of Hieronymus (1977), the difference between supply and demand in the futures and in the spot market. They argue that futures markets are a zero-sum game because for every additional long position (demand) an extra short position (supply) is opened. Hence, by definition, the capital that flows into the market equals the money that flows out of it. Furthermore, most of the futures contracts are closed out before expiration and cause no additional demand in the spot market. This is because futures contracts are not a scarce good and could therefore be created in an unlimited number at a given price level. Consequently, Irwin & Sanders (2011a) conclude that there is no linkage between the increased CIT activities and the price levels in agricultural futures markets during the price spike in the year 2008.

Frenk (2010), an employee of Michael W. Masters at the lobby group Better Markets Inc., wrote a review on the OECD report wherein he criticizes the superficial analysis, the use of inappropriate statistical models, and the contradicting results compared to other studies using, according to Frenk (2010), more appropriate statistical models. However, Shutes & Meijerink (2012) refute all these allegations in their literature review.

Similarly, Irwin, Sanders & Merrin (2009) derived four arguments contradicting the view that speculative long positions by CITs fuelled a bubble in commodity prices. The first being the conceptually inadequate work of the bubble-proponents as well as their misunderstandings of the functionality of the futures market. The second, the majority of the peer-reviewed papers indicate no connection between increased speculation and futures price changes. The third, that facts contradict the bubble hypothesis (e.g. no accumulation of inventories). Lastly, throughout history speculation has been accused of being the disruptive force causing market turbulence and higher volatility (Shutes & Meijerink 2012). It should be considered, though, that the study derives its conclusion by using only theoretical arguments and does not conduct a separate empirical analysis to reinforce them.

Hamilton & Wu (2012) use regression analysis and find that lagged notional positions of CITs cannot predict excess returns of agricultural futures contracts in the period 2006-2011. Their findings suggest therefore that there is no increased buying pressure from CITs in this market that has influenced the risk premium or prices. Additionally, they point out that there is neither reliable empirical evidence nor a theoretical framework to explain the transmission from futures to spot markets.

2.2.2 Increasing financial speculation is beneficial

Gilbert & Morgan (2010) analyse the recent increase in food price volatility and compare it to past data from 1970 to 2009. In their study, in contrast to several public claims, they find that volatility has decreased rather than increased and in the past there have been other periods with similar high volatility. With data going back to 1900, Jacks (2007) examines the impact of the implementation of futures markets in various countries. His hypothesis, that futures markets generally decreased the volatility of commodity prices, is undermined by his analysis. Furthermore, Jacks (2007) draws the conclusion that in the past, extreme price movements in the agricultural commodity markets have repeatedly led to overregulation. Also analysing a long observation period (42 years) by making use of a SADF procedure, Etienne, Garcia & Irwin (2013) detect two bubble periods (1971-1976 and 2006-2011) in the agricultural commodity futures market. Furthermore, they show that there tend to be fewer bubbles in the latter half of the observation period and these bubbles tend to be shorter. Consequently, the authors conclude that the newly entered market participants enhanced market liquidity and increased the absorption of new information. No clear causality is proven however and a third force might be the actual reason for this phenomenon.

Likewise, Brunetti, Büyükşahin & Harris (2010) show that speculators (i.e. hedge funds and swap dealers) supply the markets with liquidity and therefore reduce price volatility in the futures market. Similar results, based on the increased liquidity argument, were obtained by Irwin & Sanders (2011c). Moreover, Bohl, Javed & Stephan (2012) conclude that increased levels of futures trading diminished the information asymmetries in the market and enhanced price discovery in the period 2006-2011.

At the end of 2012, Bohl & Stephan (2012) extend their previous study by expanding the dataset and dividing the sample into two equally long sub- periods (i.e. 1992-2002 and 2002-2012) in order to find out whether financialization increased the impact of speculation on conditional volatility.

[...]