Analysis of the Link between Crude Oil and Staple Food Prices and Its Implications on Developing Countries

Table of Contents

1. INTRODUCTION AND OBJECTIVE OF THE STUDY
1.1 Introduction to a controversial question: Does biofuel production link food to crude oil prices?
1.2 Objectives and approach of the study

PART I

FRAMEWORK ON PRICE LINKS IN COMMODITY MARKETS

2. INTRODUCTION
2.1 The role of prices
2.2 Factors resulting in price distortions

3. METHODOLOGY OF PART I
3.1 Research questions
3.2 Selection of research studies included in the literature review
3.3 The concept of cointegration

4. PRICE LINKS BETWEEN SELECTED COMMODITY MARKETS - A LITERATURE REVIEW
4.1 Cointegration of natural gas and crude oil markets
4.2 Cointegration of natural and synthetic rubbers
4.3 Cointegration of cotton and polyester markets
4.4 Price links between competing crops
4.5 Linked commodity markets without cointegration

5. GENERAL FRAMEWORK OF PRICE LINKS BETWEEN DIFFERENT COMMODITY MARKETS
5.1 Factors driving cointegration
5.2 Factors limiting cointegration
5.3 Price setting mechanisms in cointegrated markets

PART II

FACTORS LINKING THE CRUDE AND THE PLANT OIL MARKET

6. METHODOLOGY OF PART II
6.1 The approach
6.2 Illustration of the approach
6.3 Strength and limitations compared to other methodologies

7. THE STRUCTURE OF THE CRUDE AND THE PLANT OIL MARKET
7.1 Definition of the relevant markets
7.2 Size of the markets

8. SUBSTITUTION EFFECTS
8.1 Overview of all areas of substitution between crude and plant oil
8.2 Demand side substitution in the transport sector
8.3 Demand side substitution in heating and electricity
8.4 Demand side substitution in the chemical sector
8.5 Substitution effects on the supply side

9. CONCLUSION ON THE LINK BETWEEN CRUDE AND PLANT OIL PRICES
9.1 Effect of the structure of crude and plant oil markets on price setting mechanisms
9.2 Lower price boundaries of plant oil defined by crude oil prices
9.3 Factors limiting cointegration
9.4 Comparison with empirical data and the results of other studies

PART III

THE EFFECTS OF A CO-MOVEMENT OF CRUDE OIL

AND AGRICULTURAL COMMODITY PRICES

10. THE EFFECTS OF A LINK BETWEEN CRUDE OIL AND FOOD PRICES
10.1 Impact on Food Security
10.2 Other macroeconomic impacts

11. IMPACT ON INDUSTRIALIZED COUNTRIES

12. IMPACT ON BIOFUEL INDUSTRIES

13. IMPACT ON THE CRUDE OIL MARKET

14. POLICIES TO MITIGATE NEGATIVE EFFECTS ON FOOD SECURITY
14.1 Measures to reduce the lower price boundaries
14.2 Measures to reduce immediate negative impacts in developing countries
14.3 Case Studies: The impact of a link between crude oil and food markets on selected developing countries and policy responses

15. SUMMARY AND CONCLUSION
15.1 The link between plant and crude oil prices
15.2 The impact of a link between plant and crude oil prices and potential counter measures
15.3 Limitations of the results
15.4 Further research required

ANNEXE
CALCULATIONS
i. Pure plant oil
ii. Biodiesel
iii. Heating oil
iv. Substitution on the supply side

LIST OF FIGURES

I. Development of agricultural prices 2005-2008

II. OECD-FAO prognosis of Feedstock and Oil Prices 2007-2017

III. FAPRI prognosis of Feedstock and Oil Prices 2007-2017

IV. Supply/demand based approach of the impact of biofuel production

V. Cointegration of natural gas and crude oil prices according to Ghouri

VI. Crude oil and natural gas prices according to Hartley et al. (2007)

VII. FAPRI prognosis of soybean and corn prices 2006-2016

VIII. Relation between world oilseed and maize prices outcomes of stochastic simulations according to the OECD-FAO

IX. Projection of HFCS substitution for sugar

X. Substitution between sugar and HFCS in the U.S. market

XI. Lower and upper price boundaries as a result of asymmetric substitution

XII. Global vegetable oil production 2002-2007

XIII. Overview on the substitution potential in the energy sector

XIV. Overview of the substitution potential in the chemical sector

XV. Marginal rates of substitutions between plant oil and crude oil

XVI. Development of diesel and crude oil prices 1997-2008

XVII. Linear regression of diesel and crude oil prices

XVIII. Substitution rate between soy/canola oil and crude oil with tax reduction in Germany and Ireland

XIX. Biodiesel production process (basic transesterification reaction)

XX. Other cost components of biofuel production

XXI. Substitution of fuel oil

XXII. Development of residual fuel oil, light fuel oil and crude oil prices 1997-2008

XXIII. Marginal rates of substitution in the electricity sector

XXIV. Oil demand in electricity generation 2005/2020

XXV. Oil consumption in electricity generation in European states

XXVI. Global biofuel production 2006

XXVII. Corn and crude prices according to Tyner and Taheripour

XXVIII. Indirect marginal rates of substitution

XXIX. Development of crude oil and plant oil prices

XXX. Development of crude oil, residual fuel and palm oil prices

XXXI. Share of net buyers and net sellers of food among people living below the poverty line

XXXII. Effects of a sustained US$ 10 crude oil price increase on selected developing countries and regions

XXXIII. Food Balance of Least-Developed Countries

XXXIV. Agricultural Trade Balance of Least-Developed Countries

XXXV. Break- even points of biofuels calculated as a function of crude oil prices

XXXVI. Schematic presentation of variable costs of oil resources at a given oil price level

XXXVII. Countries releasing food stocks in 2009

XXXVIII. Estimated food security conditions in Nigeria, September 2009

XXXIX. Estimated food security conditions in Uganda

XL. Increasing fuel tax on plant oil in Germany

XLI. Methanol and crude oil prices

XLII. Actual and anticipated crude glycerine prices

XLIII. Linear regression of heavy fuel oil and crude oil prices (1997-2008)

ABBREVIATIONS

illustration not visible in this excerpt

1. INTRODUCTION AND OBJECTIVE OF THE STUDY

1.1 Introduction to a controversial question: Does biofuel production link food to crude oil prices?

Food prices - particularly prices of agricultural commodities used as a feedstock for biofuel production - reached record highs in 2008. Within a period of slightly more than two years prices for staple food such as corn, soy, wheat, and vegetable oils have more than doubled (International Monetary Fund (2008a), p. 1).

Development of selected agricultural prices 2005-2008 according to the International Monetary Fund

illustration not visible in this excerpt

I. Development of agricultural prices 2005-2008

Source: International Monetary Fund, 2008, p. 1, own illustration

This price acceleration has occurred at a time of surging crude oil prices and a rapid expansion of biofuel production, which relies mainly on feedstock from food crops (UN Energy Department (2007), p. 31-35). Consequently, the market development has triggered a controversial debate on the question whether the increase of agricultural prices in line with crude oil prices is a mere coincidence, due to stock market speculation, or result of a lasting integration of the agricultural and the energy sector.

The debate has a strong impact on the global perception of biofuels. Once considered as a major chance for developing countries, biofuels were at the heart of policy agendas in 2005/2006. Studies and reports published by international institutions like the World Bank, the United Nations, or the EU Commission accentuate manifold advantages of the locally produced fuels. The main risk of biofuels highlighted in these studies is the need for ongoing subsidies to the industry. Environmental effects and negative impacts on food security are also mentioned, yet not the main focus in most of these reports (Kojima and Johnson (2005), p. 17ff; Dufey (2007), p. 1; Coelho (2005), p. 7ff; European Commission (2006a), p. 6-7; European Commission (2006b), p. 30-32; UN Energy Department (2007), p. 31-35).

In the light of escalating food prices, the negative impacts on food markets have become the focal point of many research studies on biofuels. Some papers published by large international institutions, such as an OECD study that dates back to September 2007, even openly address the question whether biofuel programs might be worse than any disease they are supposed to cure (Doornbosch and Steenblik (2007), p. 5ff; Sachs (2007), p. 5-7; Zilberman and Rajagopal (2007), p. 51f).

A potential integration of crude oil and food markets is assumed to increase the level as well as the volatility of food prices and thus threaten food security, which is defined as the ability of individual households to sustain an adequate level of calories. In most cases undernourishment does not result from a lack of available food - but from a lack of income to buy it. High and volatile food prices reduce the ability of net-food-purchasers in developing countries to prevent under nourishment (Thomas et al. (2006), p. 53). The effect of biofuel production on food prices is therefore of high importance to policy makers around the globe. The issue has been analysed by various universities and institutions - yet the results of their analyses strongly diverge as summarized in the following paragraphs.

1.1.1 The controversy over the current impact of biofuel production on food prices

Among the advocates of the hypothesis that the recent food price increase is driven by various effects and not mainly due to biofuel production are the authors of a joint report published by the OECD and the Food and Agriculture Organisation of the United Nations (FAO). The two institutions find that the observed price changes for agricultural products are mainly due to

- weather-related shortfalls in production,
- low stocks,
- and an increase in demand

According to the report the increase in demand for agricultural products has been caused by a confluence of factors such as economic growth, an expansion of meat consumption and the conversion of agricultural feedstock into ethanol and biodiesel. Consequently, biofuel production is seen as one factor contributing to the price surge - yet not the most important one. The report illustrates the impact of weather effects and biofuel production by describing their respective impacts on cereal markets:

“For cereals, weather-related shortfalls in production have occurred in a number of producing countries most notably in Australia, where production fell by more than 50 percent. In a global context of low global cereal stocks in recent years, these lower supplies have been a strong factor underpinning world prices. Reduced global stocks and production are confronted with stronger than expected demand for cereals for biofuel production (…). It is noteworthy, however, that the combined cereal supply shortfall in North America, Europe and Australia in 2006 of over 60 Mt was nearly four times larger than the 17 Mt increase in cereal use for ethanol in these countries.”

(OECD-FAO (2007), p. 28)

Supporting the results of the joint OECD-FAO analysis, a report published by the German Federal Ministry of Finance also states that the acceleration in agricultural commodity prices has been due to a combination of factors such as increasing demand for dairy products and meat in traditionally rice consuming countries, failure of crops, reduction of stock levels, and rising transport prices. Biofuels are seen as an additional factor contributing to the upward trend in an already tight commodity market (German Federal Ministry of Finance (2007), p. 79-80).

In a 2008 statement, the administration of the United States equally does not consider biofuel production to be the most relevant factor contributing to the food price surge. Instead, the growing demand in emerging countries is identified as main driver behind the market development. Biofuel production is found to contribute less than three percent to the acceleration in staple food prices (Sen (2008), p. 17; Chakrabortty (2008), p. 1).

In sharp contrast to these analyses, the World Bank states that 70-75 percent of the increase in food prices within the period 2002 to 2008 was due to biofuel production. A report published by the institution in August 2008 concludes that the food price development has indeed been caused by several factors - yet, the rapid expansion of biofuel production from grains and oilseeds in the United States and Europe is assumed to be by far the most important one (The World Bank (2008d), p. 2; Mitchell (2008), p. 17).

The diverging opinions on the issue demonstrate that the impact of biofuel production on staple food prices is highly debatable and approaches to calculate the effects might be influenced by political factors. While it is important to identify the drivers of the past acceleration in agricultural commodity prices it is, however, even more relevant to understand their future development in the light of a continuous expansion of biofuel capacities. The following paragraph therefore summarizes the most important studies on the expected future developments of agricultural prices.

1.1.2 Expected future developments

Like the analysis of the current market situation, the outlooks of the future development of food markets diverge markedly. Most forecasts include global biofuel production as a factor driving demand. Yet, the overall effect on anticipated agricultural price levels largely depends on expected supply-side responses, i.e. on the question to what extent the supply side can match the increasing demand through an expansion of planted areas and a more intensive use of available arable land.

Therefore, forecasts by the U.S. Food and Agriculture Policy Research Institute, the European Commission, the OECD-FAO, the Purdue University, or the International Food Policy Research Institute include diverging scenarios of future food prices depending on the underlying assumption regarding demand as well as supply side responses

The following graphs show two forecasts of food price developments: The first one is taken from a report published jointly by the Organisation for Economic Co-operation and Development (OECD) and the Food and Agriculture Organization of the United Nations (FAO) in 2007. The second one was published by the Food and Agriculture Policy Research Institute of the United States in the same year.

Development of Feedstock and Crude Oil Prices 2007-2017 according to the OECD-FAO

illustration not visible in this excerpt

II. OECD-FAO prognosis of Feedstock and Oil Prices 2007-2017 Source: OECD-FAO (2007), p. 49-50, own calculation/illustration

Development of Feedstock and Crude Oil Prices 2007-2017 according to the FAPRI

illustration not visible in this excerpt

III. FAPRI prognosis of Feedstock and Oil Prices 2007-2017 Source: Food and Agricultural Policy Research Institute (2007), p. 55-299

The authors of the OECD-FAO report forecast prices of most of the main agricultural commodities to decrease slightly in the next decade due to increased production levels. Only the prices of plant oil are expected to increase marginally. Based on different assumptions regarding land availability and demand the Food and Agriculture Policy Research Institute, on the other hand, expects soybean oil and palm oil as well as sugar prices to continue to rise substantially while corn and canola oil prices are expected to drop.¹

In spite of the differences in projected price levels of specific commodities, the forecasts cited above share a common aspect with respect to market integration: None of the studies expects agricultural and crude oil markets to link up within the next decade. Feedstock and crude oil prices are anticipated to develop independently in the long-run. The analyses define agricultural price levels as the balance between production - which is a function of arable land and yields - and demand arising from the food sector and industrial market segments. Biofuel production is consequently seen as a factor shifting the demand function and thereby changing equilibrium prices as illustrated on the following graph.

Calculation of the impact of biofuel production on food prices using a supply and demand based approach

illustration not visible in this excerpt

IV. Supply/demand based approach of the impact of biofuel production

Source: Own illustration

Other authors and institutions, for example the European Commission, the United Nations or the German Agency for Renewable Resources share this perspective. In recent reports these organisations calculate break-even points of different types of biofuels as a function of oil price levels.

Such a calculation is only possible if prices for agricultural commodities - which represent up to 80-90 percent of biofuel production costs - do not rise and fall in line with crude oil prices but remain constant. Hence, the underlying assumption of many reports on biofuel production published in the last years is that crude oil price movements will not affect agricultural commodity prices significantly and that the two markets will not integrate in the long run (European Commission (2007a), p. 9 ; European Commission (2007b), p. 11; Schmitz et al. (2006), p. 23, 25; Larson (2007), p. 23).

Other studies, however, indicate that agricultural commodity markets and crude oil markets might be linked in the future. Research papers published by the Purdue University and the Stanford University’s Program on Food Security and the Environment, for example, conclude that biofuel production links agricultural and energy markets. Consequently, the authors find that the world enters into an era with a close long-term connection between crude oil and agricultural commodities prices. An independent movement of energy and food prices is thus considered as being impossible (Tyner and Taheripour (2008), p. 2-4; Naylor et al. (2007), p. 30-43).

These papers support the hypothesis of Schmidhuber, an economist of the Food and Agriculture Organization of the United Nations, who concludes that energy prices define the floor and ceiling prices of agricultural feedstock. As prices for fossil energy reach or exceed the energy equivalent of agricultural products, the energy market creates demand for agricultural products linking prices in the long run (Schmidhuber (2007), p. 10-15).

Under this assumption the acceleration in agricultural prices in line with crude oil prices is neither a coincidence nor simply due to a shift of the demand function. Moreover, an expansion of agricultural production in the coming years will not be able to compensate the increase in demand and will not result in lower food prices either - unless crude oil prices decrease as well. If agricultural commodity and energy markets are integrated, food price levels will be defined by energy prices rather than depending on land availability and yields.

1.2 Objectives and approach of the study

In the light of these controversial opinions on an issue that could have a far reaching impact on global producers and consumers of food - particularly those in developing countries - the objective of this study is to analyse whether agricultural commodity and crude oil markets might be linked in the future and how an integration of markets would affect developing countries.

The study is structured as follows:

- Part I will analyse under which conditions prices in commodities markets might follow the same trend. This part of the dissertation will be based on a comprehensive literature review.

A structural link between prices of two or more commodities is often referred to as market cointegration. In cointegrated markets prices may deviate in the short term and correlation coefficients can be low at times. However, the markets return to equilibrium in the long run so that prices follow the same trend. This equilibrium or structural relationship can be empirically evaluated in order to prove the link between the markets (Alexander (1999), p. 24; Hendry and Juselius (2001), p. 75; Granger and Newbold (1974), p. 111-120; Jones and Nesmith (2006), p. 2f).

So far, no general framework on cointegration in commodity markets has been developed that explains why and under which conditions prices of two or more commodities are characterized by a stable long run relationship. However, drivers for cointegration effects have been identified in various studies focussing on specific commodities such as crude oil/ natural gas, cotton/ polyester, and natural rubber/ synthetic elastomers.

The results of the most relevant studies in this field of research will be summarized in order to develop a general framework on cointegration in commodity markets. This framework will be the conclusion of the first part of this dissertation.

- Part II will investigate whether a co-movement of staple food and crude oil prices is technically possible given the structural relationship of both markets. The analysis included in this part will be based on the framework developed in the first part of the dissertation. Sources used for the analysis, will comprise both, desk research as well as interviews with industry experts.

The focal point of the analysis will be plant oil which is the main input factor for biodiesel production, but can also substitute for petroleum derivates in industrial applications such as paints, surface coatings and plastics. As cointegration effects are mainly due to the substitution potential between two or more commodities (see Part I), the objective of the second part of the dissertation is to analyse all relevant market segments in which plant oil could replace crude oil. The approach will be to estimate the overall substitution potential and to calculate the marginal rates of substitution between crude and plant oil in each market segment. These estimates will be included in a model that identifies

- how much of the global vegetable oil market could be absorbed within each market segment,
- at what price ratio switching is economically attractive,
- whether the substitution potential between crude and plant oil could result in a co-movement of prices as observed in other commodity markets.
- Part III will summarize the potential effects of an integration of plant and crude oil markets on developing countries. When plant oil prices are linked to crude oil prices other staple food prices would be equally affected by rising or falling crude oil prices. Agricultural commodities compete for arable land so that any change in the prices of one type of crop results - ceteris paribus - in price changes of most others.

As mentioned above, high and volatile staple food prices reduce the food security of net food purchasers in developing countries. How severely consumers in these countries are affected depends on the degree of price transmission from international to domestic markets which differs by regions as demonstrated in previous research studies (Blein and Longo (2009), p. 1-2).

In addition to the effects on household level, a co-movement of staple food and crude oil prices will also affect the countries’ balance of trade either mitigating or worsening the effects of rising oil prices on the economic development.

Besides summarizing the potential effects of a co-movement of crude oil and staple food prices, the third, and last part of the dissertation, will analyse measures to mitigate the negative effects of a market coupling. This analysis will be based on the main findings of Part II and draw attention to the role of direct subsidies and tax relief programmes in various market segments.

PART I FRAMEWORK ON PRICE LINKS IN COMMODITY MARKETS

2. INTRODUCTION

The purpose of the first part of this dissertation is to analyse under which conditions prices in commodities markets might follow the same trend. As presented in the next chapter, the methodology used in this part is a systematic review of the literature on price links between two or more commodities. Based on key findings of this review a general framework on price links in commodity markets will be developed.

Part I starts with an introduction to the role of prices, how they are meant to reflect the social value of commodities and why they frequently fail to do so. While the subsequent literature review focuses on the reasons why prices in two or more commodity markets might be interlinked, this chapter introduces the theoretical context allowing for an interpretation of the results.

Factors that result in price distortions as presented in the following tend to have an influence on price levels in most commodity markets. Consequently, these factors have to be borne in mind when interpreting the results of the literature review.

2.1 The role of prices

As stated in the efficient market theory, prices have two main roles: firstly, they are meant to convey information (informative function) and secondly, they facilitate the efficient allocation of scarce resources (allocative function). As conveyors of information, prices inform about the willingness of buyers to acquire goods at a given price based on their subjective values and utilities. These prices are defined as buyers’ reservation prices and indicate the largest amount any specific buyer would be willing to pay for a good or service. Likewise, prices indicate the willingness of sellers or providers to sell at a given price based on the assessment of their production costs and profit expectations (Bernanke and Frank (2004), p. 62; Wienert (2008), p. 85ff; Friedman (2007), p. 8ff).

This rapid, two-way transmission of information enables an efficient coordination of markets, which would be difficult to organize otherwise. In perfectly competitive markets a price mechanism is thus the most efficient way to transmit information between economic units in the sense that a minimum number of variables is used (Thomsen (1992), p. 31).

Closely linked to the informational role of prices is the allocational role: From a consumer’s perspective, market prices ration goods and resources, allocating them to the individuals with the highest willingness to pay. From the point of view of sellers or providers, prices guide production and the allocation of input factors. Whenever the market price exceeds the cost of production of one or more suppliers, it creates an incentive to produce more. With increasing prices for a given product, it is not only current producers who are likely to boost outputs, but new providers will also have an incentive to enter the industry. Thus, additional production factors i.e. labour and capital will be provided precisely where the consumers’ willingness to pay is highest. This allocation principle has been called “the invisible hand” by Adam Smith in 1759. According to Smith’s theory, self-interested motivations of individuals still have positive results for the entire society because these motivations contribute to an efficient allocation of resources (Smith (1966), p. 288; Hirshleifer et al. (2005), p. 18-20; Landsburg (2008), p. 252). However, this is only possible when prices fully reflect the costs and benefits of supply and consumption and markets are not distorted by the factors presented in the following.

2.2 Factors resulting in price distortions

There are several factors which result in market and price distortions:

a) Transaction costs
b) Externalties
c) Uncompetitive markets
d) Imperfect information
e) Irrational behaviour
f) Government intervention

a) Transaction costs are defined as the costs which arise whenever market actors exchange ownership rights for any type of economic asset (Eggertsson (1990), p. 14). They include the costs for negotiation, the gathering of information, monitoring, co-ordination as well as costs for the enforcement of contracts. If these costs reach or exceed the benefits from trade, the transaction will not take place (Starr (2002), p. 2; Okoye et al. (2010), p. 3).

b) Externalities describe any type of costs or benefits that are neither borne by the producer nor by a consumer of a given product but passed on to a third party. Externalities arise when the social or economic activities of any group of individuals have an impact on other groups while that impact is not fully accounted or compensated for by the first group. Any underinternalization significantly reduces the efficiency of price-setting mechanisms (Roson and Small (1998), p. 118; Energy Information Administration p. 43; Sands (2003), p. 354; Schmidtchen et al. (2009), p. 13, 19, 36).

c) Uncompetitive markets result in price levels which do not convey information about production costs and reservation prices of producers. While a competitive firm is a price taker in its market, market actors in monopolies or oligopolies will set prices significantly above free market prices in order to maximise total profits. Consequently, prices will not result in an efficient allocation of resources (Mankiw (2008), p. 315-320; Hirshleifer et al. (2005), p. 253; Friedman (2007), p. 22-24).

d) Imperfect information is an additional factor reducing the effectiveness of price mechanisms. Prices can only convey information and allocate resources when sufficient information is available. However, economist such as Stiglitz point out that this is never the case. Consumers have to take decisions within a limited period of time. In addition, gathering information is costly so that both, producers and consumers, take decisions without being fully informed (Stiglitz et al. (2009), p. ixff - xxxvi). Moreover, market actors are often confronted with an information overload and diverging opinion making it impossible to handle all information. The widely observed phenomenon of price distributions - which could not exist if consumers were well informed - is seen as an indicator for imperfect information (Loasby (1974), p. 184; Stiglitz (1979), p. 340).

e) Irrational behaviour such as “crowd effects” can result from imperfect information or information overload. This behaviour arises whenever individuals confronted with imperfect information, mimic the behaviour of others in the belief that they might have additional knowledge (Hirshleifer et al. (2005), p. 327; Takayasu et al. p. 102). In this case prices do not convey available information, but rather reflect information market participants believe others to have (Altman (2006), p. 92; Sornette (2003), p. 22, 94-97, 99). In addition, the interpretation of information by market actors is often biased and influenced by framing effects and emotions which can also result in irrational behaviour. Both factors strongly influence the efficiency of prices (Baker et al. (2010), p. 3-6; Shleifer (2000), p. 5, (Montier (2002), p. 15ff; Weir (2005), p. 96).

f) Moreover, government intervention cause massive market distortions, potentially changing price levels to such a degree that they no longer reflect production costs or market fundamentals. Governments can influence markets by setting floor/ ceiling prices, by creating barriers of trade such as import and export tariffs / quotas, and by subsidizing/ taxing goods (Bernanke and Frank (2004), p. 173).

Among all factors resulting in price distortions, government interventions tend to be most relevant for the markets presented in the following. Agricultural commodity markets, which are the focal point of the analysis in Part II, are considered to be the most distorted sector of the world economy. Subsidies, import tariffs, quantitative restrictions and preferential trade agreements strongly influence these markets today and are likely to continue doing so in the future. According to the OECD, agriculture is the sector with the highest subsidy figures worldwide. Total support for agriculture amounted to US$311 billion, in other words approximately 1.3 percent of the total Gross Domestic Product in OECD countries in 2001 (Lloyd et al. (2009), p. 1ff; Organisation for Economic Co- operation and Development (2002), p. 4).

In addition, biofuel markets as well as renewable energy markets are also strongly influenced by government intervention schemes such as subsidies and mandates. Due to their strong impact on equilibrium prices, subsidy schemes have to be taken in consideration when analysing the link between crude and plant oil prices.

Part II will therefore include calculations of equilibrium prices with and without government intervention in order to identify the effect of mandates, subsidies, and other types of interventions schemes.

3. METHODOLOGY OF PART I

As mentioned above, the following analysis is based on a systematic review summarising literature relevant to the main research questions. These research questions, as well as the methodology on how the studies were chosen, are presented in the following paragraphs.

3.1 Research questions

The key questions for the following analysis are:

Under which conditions do prices in commodity markets follow the same trend?

Are there factors that limit price links between two or more commodities?

What type of price-setting mechanisms exist in linked commodity markets: − Are there constant ratios?

− Which factors define the price ratios between the prices of commodities and how do they develop over time?

− At which speed are prices transmitted and what determines this speed?

Based on the answers to these questions a general framework on price links in commodity markets will be developed.

3.2 Selection of research studies included in the literature review

An analysis of price links between two or more commodity markets is only possible when price data is publicly available. This is the case for all commodities traded at stock markets. The following table provides an overview of the main commodities traded internationally, based on Bloomberg and Financial times listings as well as on Labys and Pollak (1984), p. 4 and Rowe (1965), p. 6ff.

- Energy
- Crude oil (WTI, Brent, etc.)
- Heating oil / diesel
- Coal
- Gasoline

- Ethanol
- Natural gas
- Coal
- Uranium

- Propane

Agriculture

-Canola (oil /meal)

-Cocoa

-Coffee

-Corn

-Wheat

-Oats

-Rice

-Sorghum

-Palm Oil

-Palm Kernels and oil

-Coconut oils

-Inedible fats

-Soybean (oil /meal)

-Sugar Lean
- Orange Juice
-Lumber
-Cotton
-Wool
-Rubber
-Sisal
-Timber
-Jute
-Abaca
-Burlap
-Cattle
-Feeder Cattle
-Frozen Pork Bellies
-Lean Hogs

Metals

- Copper

- Gold

- Silver

- Platinum

- Titanium

- Magnesium

- Nickel
- Palladium
- Ore
- Zinc
- Silicon
- Lead
- Iron
- Aluminium

Based on this list of commodities, peer-reviewed journals, conference proceedings, and books have been searched for eligible articles on price links between one of these commodities and any second one. In addition, the term “commodity” itself in combination with terms such as “price links”, “price effects”, “market integration”, “market coupling” and “cointegration” has been used to systematically search databases and publications for relevant literature.

As a result of the screening process the following commodities were identified as being interlinked:

- Crude oil and natural gas markets
- Rubber and synthetic elastomers
- Cotton and polyester markets
- Competing crops

In addition to the interlinked markets, publications on the entire energy market as well as on the relationship between high fructose corn syrup and sugar have been included in the literature review. These studies contain information on barriers of market integration.

The following literature review synthesizes relevant aspects from eligible publications, indicating the main drivers and potential barriers for price links between commodities, as well as providing information on price transmission and price-setting mechanisms. The following two paragraphs will present a brief introduction to the concept of cointegration to facilitate the understanding of the research papers presented thereafter.

3.3 The concept of cointegration

The cointegrating vector approach was developed by Granger and Newbold as an improved method of analysing the relationship between variables that are characterised by a trend. In 1974, the authors state that standard regression methods very often appear to reveal a relationship whereas in fact there is none. This type of relationship has been termed “spurious regression” - a regression with no economic meaning (Granger (2003), p. 2-4; Granger and Newbold (1974), p. 111-120).

Spurious regression frequently occurs when applying classical linear regression models to economic data. These regression models rely on the assumption that the data is stationary, which means that it evolves independently of time. A random shock will thus not have any lasting effect on the evolution of a stationary time series. Therefore, these series of variables have a constant and unconditional mean and variance over time. When examining time series in economics, these conditions are hardly ever satisfied - even after removing deterministic terms such as drifts and trends (Dolado et al. (1999), p. 1).

To analyse trended or non-stationary time-series Granger thus developed the concept of cointegration. Time series are defined as cointegrated if they are individually characterised by a trend - which means they are non-stationary - but a linear combination of them is stationary. This linear combination is referred to as a cointegrating vector. Therefore the concept of cointegration of economic time series suggests that the variables have a longrun structural relationship that can be empirically evaluated.

Before Granger developed the cointegration approach, the main method to analyse the relationship between non-stationary variables had been to run an Ordinary Least Squares regression (OLS) on the data which was initially differenced. This method is assumed to be correct in large samples, yet problems occur when time series are limited in terms of length.

The cointegration approach, on the other hand, allows for a more detailed analysis of the structural relationship between two time series and also works with smaller samples. Applying cointegration techniques to distinct markets usually comprises three steps:

- an analysis of stationarity / non-stationarity of the times series
- a cointegration analysis examining whether a long-run equilibrium relationship between the variables exists, and
- and an estimate of an error correction model. This model reveals dynamic causalities that must be present in a cointegrated system, so-called because “errors” or short term deviations from equilibrium are corrected (Jones and Nesmith (2006), p. 2f; Alexander (1999), p. 3).

Most papers on cointegration in commodity markets summarized in the following chapter apply cointegration techniques to identify common long term trends as well as interdependencies in specific markets. The Ordinary Least Squares method is only used in two studies, which date back to 1972 and 1998, respectively.

4. PRICE LINKS BETWEEN SELECTED COMMODITY MARKETS - A LITERATURE REVIEW

4.1 Cointegration of natural gas and crude oil markets

The studies on the crude oil and the natural gas market presented in the following sections agree on the fact that gas and oil prices are cointegrated. The first paper, published by Ghouri in 2006, focuses exclusively on crude oil and natural gas prices. The studies published by Villar and Joutz, Brown and Yücel, and Hartley et al. include additional factors explaining the price movements in the short term. These studies conclude that natural gas prices are strongly influenced by weather variables and storage levels. Such transitory factors can result in significant price differences between crude oil and natural gas, which at times produce the appearance that the two prices are decoupling. Nonetheless, prices return to equilibrium in the long run and the cointegration of the two price series remains stable.

All four papers agree on the fact that causality goes from crude oil to natural gas prices. Long term natural gas prices are consequently determined by crude oil prices and not the other way around. The reason for the asymmetric price transmission is the difference in size of the two markets: Regionally segmented natural gas markets are significantly smaller than crude oil markets. Shocks originated in one of these markets are unlikely to influence global crude oil price levels.

Opinions on the drivers explaining the cointegration, however, diverge. Ghouri finds that the main reason for the stable long run relationship between crude oil and natural gas prices is the oil-indexing of gas contracts. Brown and Yücel, on the other hand, find that substitution effects on the demand side provoke the co-movement of the price series.

Villar and Joutz find that both, substitution effects on the demand side as well as the oilindexing of gas contracts contribute to the integration of the two commodity markets. In addition, the authors state that substitution effects on the supply side add to the stable long term relationship of crude oil and natural gas prices.

The study published by Hartley et al., again, highlights the importance of substitution effects on the demand side. The authors find that - while substitution is possible in various market segments - interfuel substitution in one specific sector explains why and at which ratio natural gas prices stay anchored in a long-term relationship with crude oil prices.

Details of the four studies are presented in the following sections.

1. Ghouri (2006)

Ghouri examines the relationship between several crude oil and natural gas prices as denoted at some of the main gas hubs. The paper concludes that there is a cointegration between each regional natural gas price series and the respective crude oil prices. Despite short-term drift and fluctuations, the prices tend to converge to their respective equilibrium in the long run. The author expects the price relationship between crude oil and natural gas to remain constant and thus uses the cointegration technique to forecast future natural gas prices (Ghouri (2006), p. 249ff).

As mentioned in the introduction, the author finds that the main driver for the integration of crude oil and natural gas markets is the fact that most gas trade contracts are linked to crude oil prices through specific indices. Prices for LNG contracts, for example, frequently comprise the following formula:

illustration not visible in this excerpt

where

PLNG is the price of LNG in US$ per barrel

α is non-oil price related cost, denoted in US$, which may include a constant, shipping cost or inflation indices

β is a fraction

PCrude is the relevant price of crude oil denoted in US$ per barrel

Historical relationship between natural gas prices and crude oil prices according to Ghouri (2006)

illustration not visible in this excerpt

V. Cointegration of natural gas and crude oil prices according to Ghouri Source: Ghouri (2006), p. 255

As presented on the graph, Ghouri also finds that the mean price of natural gas expressed in million British thermal units (mBtu) tends to be significantly lower than the respective crude oil price. The mean price of gas traded at Henry Hub² calculated in U.S. dollars per mBtu is about 32 percent lower than West Texas Intermediate (WTI)³ prices. European and British natural gas prices are, respectively, 24 percent and 32 percent lower than Brent.

2. Villar and Joutz (2006)

Villar and Joutz of the Energy Information Agency (EIA) and the George Washington University analyse the oil and gas relationship in 2006. The study focuses on the linkage between natural gas price denoted at the Henry Hub and the WTI crude oil price. Villar and Joutz develop a vector autoregression and an error correction model which include exogenous variables explaining short term price movements of natural gas. These variables comprise weather factors, such as heating degree days, as well as working inventories of natural gas.

Supporting the findings of Ghouri, the authors find a stable cointegrating relationship between the price series over the period 1989 to 2005 in spite of short term deviation from the common trend (Villar and Joutz (2006), p. 37- 39).

The authors state that this relationship is asymmetric: In the observation period, crude oil price movements provoke changes in the natural gas price, but the converse does not occur. According to Villar and Joutz, the main driver for this effect is the relative size of each market. Crude oil prices are determined on the world market while natural gas markets tend to be regionally segmented. Consequently, the domestic natural gas market is significantly smaller than the global crude oil market. Shocks, which have an impact on the U.S. natural gas market, are unlikely to influence the world market price of crude oil (Villar and Joutz (2006), p. 2,40).

Another key finding of Villar and Joutz is a statistically significant trend term suggesting that natural gas prices tend to increase at a slightly faster rate than crude oil prices. Due to this trend, the gap between the price levels of the two commodities expressed in US$/mBtu is narrowing over time (Villar and Joutz (2006), p. 40).

In contrast to Ghouri, Villar and Joutz find several reasons explaining the cointegration of crude oil and natural gas markets. The oil-indexing of gas contracts - identified as the main factor for market integration by Ghouri - is seen as one aspect reinforcing the price linkage, but not the most important one. According to the authors, several economic factors affecting the demand side as well as the supply side link natural gas and crude oil prices. On the demand side, oil and gas are linked by substitution effects. Reports quoted by the authors state that approximately 18 percent of U.S. natural gas usage could be switched to crude oil or petroleum products⁴. Moreover, a significant share of power generation capacity is dual-fired using either oil or gas to generate electricity. In the power sector, substitution of crude oil and natural gas is also possible through merit order effects. The relative prices of oil and gas determine the dispatching decisions of electricity companies and thus the amount of oil and gas derivatives consumed in the electricity generation sector.

In addition to these effects regarding the demand side, substitution on the supply side is seen as a factor driving market integration. Natural gas and crude oil compete for similar economic resources - e.g. skilled labour and drilling rigs - for both production and development of new capacities. Furthermore, exploring activities substantially depend on the relative long term price expectations of crude oil and natural gas and would concentrate on one of these commodities if long-term price differences were to persist (Villar and Joutz (2006), p. 4-5).

3. Brown and Yücel (2007)

A more recent study on the price cointegration of crude oil and natural gas has been conducted by Brown and Yücel in 2007. The focus of the paper is the development of the prices of both fuels against the background of a declining number of dual fuel capacities in the United States. The authors find that the cointegration of crude oil and natural gas prices has been driven by substitution effects on the demand side for many years. The U.S. power sector and the industry have switched back and forth between residual fuel oil and natural gas, using whichever energy source has been less expensive. Over the last decade, however, the number of facilities able to use both fuels has declined. U.S. natural gas prices have shown significant independent movement. In addition, natural prices have risen above levels formerly seen as their historical relationship with Brent prices (Brown and Yücel (2007), p. 1).

To analyse whether natural gas and crude oil prices are still cointegrated and which drivers explain the short term divergence of prices, Brown and Yücel analyse Henry Hub gas and WTI prices covering the period 1994-2006. Splitting the price data into two samples, the authors find a cointegrating relationship between oil and natural gas prices in the 1994- 2006 sample, but not in the 1997-2006 sample. Like Villar and Joutz, the authors then include additional variables such as data on heating degree and cooling degree days, as well as on natural gas storage levels. Supporting the findings of Villar and Joutz, the analysis demonstrates that the cointegration of the price series remains significant for both samples when these transitory factors are taken into account (Brown and Yücel (2007), p. 13ff).

According to the paper, the main drivers for the market integration are substitution effects on the demand side. While dual fuel capacities in electric power generation are declining, merit order effects persist. In addition, Brown and Yücel consider the large substitution potential in the petrochemical industry to strengthen the competition between crude oil and natural gas. The petrochemical industry in the U.S. relies on natural gas as feedstock, yet could easily switch to crude oil products - the main feedstock of this industry in other countries (Brown and Yücel (2007), p. 12ff).

4. Hartley et al.(2007)

Hartley et al., from the James Baker Institute for Public Policy, analyse the cointegration of crude oil and natural gas in a study published in November 2007. Focus of the study is the narrowing in the relative long-term price relationship between prices of the two fuels. As the prior studies, the results of their analysis indicate that the two commodities are cointegrated.

According to Hartley et al., the main driver for market integration is fuel switching in electricity generation. Like Villar and Joutz as well as Brown and Yücel, the authors find that both plant and grid level switching, i.e. direct substitution and merit order effects, drive interfuel substitution within the electricity generation sector. Generators can arbitrage the costs of producing electricity expressed in US$/MWh, which is equivalent to the price of fuel (US$/Btu) multiplied by the heat rate (Btu/MWh).

The widespread adoption of combined cycle gas turbines has therefore changed the relative long term price ratio of the two fuels by increasing the efficiency of gas based electricity generation. Over the last decades the price of a million British thermal units (Btu) of natural gas at Henry Hub has been roughly equivalent to one-tenth of the West Texas Intermediate per barrel. This 1:10 price relationship has declined to 1:6. Given that each barrel of crude oil (WTI) contains approximately 5.8 million Btu, the 1:6 ratio is close to thermal parity. The price relationship of the two fuels is thus determined by the technologies used for electricity generation (Hartley et al. (2007), p. 8ff).

Crude oil and natural gas prices according to Hartley et al. (2007)

illustration not visible in this excerpt

VI. Crude oil and natural gas prices according to Hartley et al. (2007) Source: Hartley et al. (2007), p. 10

In addition, the analysis conducted by Hartley et al. demonstrates that the long run relationship between crude oil and natural gas prices acts through the prices of residual fuel oil. Consequently, the relationship between crude oil and natural gas prices changes, when the price difference between residual fuel oil and crude oil declines or accelerates.

Like Brown and Yücel, the authors find that seasonal fluctuations, weather conditions, inventory levels, and supply disruptions affect short term movements of natural gas prices and can result in significant deviations from equilibrium price levels (Hartley et al. (2007), p. 9ff).

4.2 Cointegration of natural and synthetic rubbers

Another example of cointegration in commodity markets is the link between natural rubber and synthetic elastomers. The co-movement of prices in these two markets has been a key issue in several research papers. One of the reasons for the interest in the link between both markets is synthetic elastomers have continuously gained in market share putting a major source of income of some developing countries at stake. Natural rubber is produced manually by cutting in the bark of rubber trees to harvest latex. Production costs therefore depend on labour, land lease and capital costs. Synthetic rubber, on the other hand, is a petrochemical product based on either crude oil or natural gas (Rodgers et al. (2005), p. 42- 45).

The studies on the rubber market presented in the following paragraphs agree on the fact that natural and synthetic rubber prices are cointegrated and that the strong link between the markets is due to substitution effects on the demand side. The area of substitution is described as variable by Haque and by Bormann and Fertig: Haque states it increases whenever a price gap persists over an extended period. Bormann and Fertig, on the other hand, find that the area of substitution depends on the degree of the price difference.

Another key finding of the studies is that the cointegrating relationship is as asymmetric as in the natural gas and the crude oil markets. In this case, the asymmetry is not due to a difference in the actual size of markets but rather due to a difference in flexibility: The natural rubber production is price inelastic and can only be expanded in the long term because trees require time to grow. Synthetic elastomer production, on the other hand, is price elastic and relatively flexible as it can be expanded in the short run.

Further details on the studies are summarized in the following sections.

1. Haque (1972)

One of the most comprehensive studies on the global rubber market dates back to 1972. In this publication, Haque analyses the price development of natural and synthetic rubber using the ordinary least squares (OLS) method. The author finds that the price of natural rubber moves in line with synthetic elastomers. While prices frequently diverge in the short run; they return to equilibrium in the long run.

Haque states that natural rubber producers are price-takers. The long-run prices of rubber are set by synthetic elastomer producers. The asymmetry of the relationship is due to the fact that the supply of natural rubber cannot be expanded in the short run because rubber trees only start producing significant amounts of latex seven years after planting. The supply of synthetic rubber, which accounts for more than 50 percent of the total market, is flexible. Synthetic rubber production is therefore price elastic and hence price setting. As a result, long-run natural rubber prices can be expressed as a function of synthetic rubber prices. The analysis indicates that any one percent acceleration in the synthetic rubber price will ceteris paribus result in an increase of 0.9 percent in natural rubber prices.

Short term price movements of natural rubber, on the other hand, are strongly influenced by its supply and demand balance. This balance, in turn, depends on current harvests and stock ratios of natural rubber (Haque (1972), p. 12-14). The transitory factors influencing short term price levels in the natural rubber market are comparable to the variables explaining price movements in the natural gas market (see Hartley et al. (2007), p. 9; Brown and Yücel (2007), p. 13; Villar and Joutz (2006), p. 24).

According to Haque, the main drivers of market integration are substitution effects on the demand side. The two commodities are close, yet not perfect substitutes: In some applications natural rubber cannot be replaced by synthetic elastomers due to its superiority regarding tear strength and abrasion resistance. Specific synthetic rubbers are technically superior for other applications. Haque’s analysis indicates that there is no general consensus as to the actual area of competition between the raw materials. The degree of substitutability between natural rubber and synthetic rubber depends on the length of the time period for which a given price gap persists. Producers are likely to adapt their production plants in order to use the cheaper raw material - even if the process is explicitly designed for the more expensive one - whenever prices are anticipated to diverge for a longer period of time (Haque (1972), p. 8).

2. Pisanwanich (1998)

A second research study on the price relationship of natural and synthetic rubber has been conducted by Pisanwanich. Supporting the findings of Haque, Pisanwanich finds natural rubber prices to increase in line with synthetic elastomers prices. Consequently, natural rubber prices are indirectly linked to crude oil prices, which account for a large share of the production costs of synthetic rubber (Pisanwanich (1998), p. 63f).

Compared to Haque, Pisanwanich finds a significantly lower adjustment rate of natural rubber prices as a result of shocks in the synthetic elastomer market. The author also used the Ordinary Least Square (OLS) method to examine the relationship between natural rubber and synthetic elastomer prices in the period 1970 to 1994. The results of his analysis indicate that any one percent rise in prices of synthetic elastomers will only cause an increase of about 0.54 percent in the price of the natural rubber. The relationship of prices is estimated as follows:

illustration not visible in this excerpt

where

NR is natural rubber price in the period t denoted in US$ per ton

SR is the price of synthetic elastomers in the same period equally denoted in US$ per ton

Like Haque, Pisanwanich concludes that the main drivers for the integration of the two commodity markets are substitution effects on the demand side (Pisanwanich (1998), p. 66ff).

[...]

---

¹ The two forecasts comprise marketing years for the respective agricultural commodities. The marketing year 2006/2007 is denoted as 2007 in the diagrams. These marketing years are, however, not homogenous. While the U.S. marketing year for corn comprises the period September 1 to August 31, the marketing year for soybean oil covers October 1 to September 30. For crude oil prices, on the other hand, the years included in the forecast are calendar years. This lack of homogeneity represents a limiting factor of the diagrams

² The Henry Hub is the most important natural gas hub in the United States

³ West Texas Intermediate is a type of crude oil used as a benchmark in oil pricing and the underlying commodity of New York Mercantile Exchange’s crude oil future contracts

⁴ Source quoted by Villar and Joutz: 2002 Manufacturing Energy Consumption Survey (MECS)