[PPT] - Financialization of Commodity Markets: Non linear Consequences from PowerPoint Presentation

SLIDE 1

Nombre de la presentación en cuerpo 17

Financialization of Commodity Markets: Non linear Consequences from Heterogeneous Agent Behavior Non-linear Consequences from Heterogeneous Agent Behavior

Javier Ibarlucia Javier Ibarlucia

Diego Bastourre Diego Bastourre

Jorge Carrera Jorge Carrera

Central Bank of Argentina

Session I – Agricultural Profitability in the 21st Century Farming, Finance and the Global Marketplace - Regional Economic Symposium

1 The views expressed here are those of the authors and do not necessarily reflect the official position of the Central Bank of Argentina.

Kansas, June 8, 2010

SLIDE 2

Nombre de la presentación en cuerpo 17

Outline

1. Motivation

2 C dit P i R t d l t d l t d

2. Commodity Prices: Recent developments and long run trends
3. Long-run drivers of commodity prices
4. Financialization of commodity markets
5. A heterogeneous agent model
6. Econometric results
7. Open questions: Structural changes

2

8. Conclusions

SLIDE 3

Nombre de la presentación en cuerpo 17

A i l l i ifi h i l f l L i

1) Motivation

Agricultural exports

(as % of total exports)

Agricultural exports represent a very significant share in total exports for several Latin American countries.

(as % of total exports)

3

(2007)

0-5 5-10 10-30 30-50 50-70 70-100 No Data Source: FAO

SLIDE 4

Nombre de la presentación en cuerpo 17

d i ifi i f h GDP

1) Motivation

Exports of agricultural, fishery and forest products

19 1 P

… and a very significant portion of the GDP.

11.0 11.7 12.1 19.1 Argentina Chile Uruguay Paraguay 6.0 8.4 9.4 Bolivia Guatemala Ecuador g 2 9 3.8 3.8 4.4 C l bi Brazil Peru Latin America 0.1 1.5 2.0 2.9 Venezuela Mexico Dominican Republic Colombia % of total GDP

4

5 10 15 20 25 Source: FAO

SLIDE 5

Nombre de la presentación en cuerpo 17

2) Commodity Prices: recent developments and long run trends

600 350

Key commodity price indexes, 2002M1-2009M12 (2002=100)

Up to the intensification of the

400 500 250 300

Up to the intensification of the financial crisis, nominal commodity prices grew at a strong peace during the last 6 years

300 150 200 100 200 50 100

Strong reversal: Record values of June-July were cut at least

2M1 2M6 M11 3M4 3M9 4M2 4M7 M12 5M5 M10 6M3 6M8 7M1 7M6 M11 8M4 8M9 9M2 9M7 M12 50

by a 40% at the end of 2008.

5

2002 2002 2002M 2003 2003 2004 2004 2004M 2005 2005M 2006 2006 2007 2007 2007M 2008 2008 2009 2009 2009M

Crude Oil (LHS) Copper (LHS) Aluminium (RHS) Soybeans (RHS) Corn (RHS)

SLIDE 6

Nombre de la presentación en cuerpo 17

2) Commodity Prices: recent developments and long run trends

Food and metal price IMF indexes (nominal and real, 1960=100)

600 700 160 180 200

Contrary to the belief that prices reached historical high levels, long-run perspective shows

400 500 100 120 140

decaying prices if world inflation is incorporated

100 200 300 40 60 80 100 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 20 Food Metals Food (real) Metals (real)

6

Food Metals Food (real) Metals (real)

SLIDE 7

Nombre de la presentación en cuerpo 17

2) Commodity Prices: recent developments and long run trends

Oil (nominal and real, 1960=100)

The history of oil is different…

3000 3500 4000 400 450 500

y

2000 2500 3000 250 300 350 500 1000 1500 100 150 200 500 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 50 Oil Oil (real)

7

Oil Oil (real)

SLIDE 8

Nombre de la presentación en cuerpo 17

2) Commodity Prices: recent developments and long run trends

Real Price Volatility (1960-2009)

0 12 0 08 0.10 0.12 0 04 0.06 0.08 0 00 0.02 0.04 0.00 1960-01 1961-01 1962-01 1963-01 1964-01 1965-01 1966-01 1967-01 1968-01 1969-01 1970-01 1971-01 1972-01 1973-01 1974-01 1975-01 1976-01 1977-01 1978-01 1979-01 1980-01 1981-01 1982-01 1983-01 1984-01 1985-01 1986-01 1987-01 1988-01 1989-01 1990-01 1991-01 1992-01 1993-01 1994-01 1995-01 1996-01 1997-01 1998-01 1999-01 2000-01 2001-01 2002-01 2003-01 2004-01 2005-01 2006-01 2007-01 2008-01 2009-01 Food Metals

8

SLIDE 9

Nombre de la presentación en cuerpo 17

2) Commodity Prices: recent developments and long run trends

Real Price Volatility (1960-2009)

0 40 0.30 0.35 0.40 0 15 0.20 0.25 0.05 0.10 0.15 0.00 1961-01 1962-01 1963-01 1964-01 1965-01 1966-01 1967-01 1968-01 1969-01 1970-01 1971-01 1972-01 1973-01 1974-01 1975-01 1976-01 1977-01 1978-01 1979-01 1980-01 1981-01 1982-01 1983-01 1984-01 1985-01 1986-01 1987-01 1988-01 1989-01 1990-01 1991-01 1992-01 1993-01 1994-01 1995-01 1996-01 1997-01 1998-01 1999-01 2000-01 2001-01 2002-01 2003-01 2004-01 2005-01 2006-01 2007-01 2008-01 2009-01 Oil

9

SLIDE 10

Nombre de la presentación en cuerpo 17

3) Long-run drivers of commodity prices

Prebisch (1950) and Singer (1950) claimed that, contrary to the classical view,

commodity prices would fall relatively to those of the industry goods. The influence

f this hypothesis on empirical research has been significant: univariate models to

study long-run trends.

A different

approach for studying commodity prices starts asking which macroeconomic factors could act as determinants of them.

US Multilateral Real Exchange Rate:
Pioneering single-good model of Ridler and Yandle (1972) demonstrates that a real

exchange rate appreciation induces a fall in dollar commodity prices.

10

Commodity Price elasticity to Dollar Multilateral RER should lie between 0 and -1

(Dornbusch, 1985).

SLIDE 11

Nombre de la presentación en cuerpo 17

3) Long-run drivers of commodity prices

World demand:

i) Food:

Engel’s law is an accurate framework to predict the impact of income on food
Engel s law is an accurate framework to predict the impact of income on food

commodities (Houthakker, 1987; Hamilton, 2001).

Income-elasticity decreases as long as the transit to development is completed.

ii) Metals:

Inverse U-shape relationship between its use and income level. Consumption

increases up to 15,000 or 20,000 per capita GDP in PPP dollars (IMF, 2006).

Interest rate:

Interest rate increases reduce commodity prices through 3 channels (Frankel, 2006): i) by increasing the incentive for extraction (or production) today rather than tomorrow; ii) by decreasing the desire of firms to carry inventories; and iii) by encouraging speculators to shift out of commodity contracts into treasury bills

11

encouraging speculators to shift out of commodity contracts into treasury bills. The 3 transmission channels work to reduce spot prices.

SLIDE 12

Nombre de la presentación en cuerpo 17

4) Financialization of commodities

Intense debate regarding financialization in commodity markets. Some

authors have blamed financial markets as the only source of violent price ups y p p and downs. Others have neglected any influence on prices. B “fi i li ti f diti ” ll f t t li k d f t i)

By “financialization of commodities” we usually refer to two linked facts: i)

recent impressive growth of derivative market activity; ii) increasing participation of financial investors in commodity future markets participation of financial investors in commodity future markets.

Why to believe there is a connection between prices and speculative activity?

Casual empirical evidence: Commodity prices soared jointly with a rapid increase in turnover on commodity-linked instruments since 2002.

12

SLIDE 13

Nombre de la presentación en cuerpo 17

4) Financialization of commodities

Derivative commodity contracts and commodity prices, 2002Q1 to 2008Q2

30 500

I th li ht f thi id it

25 350 400 450

In the light of this evidence it results quite natural to associate this significant increase in financial commodity-market

15 20 200 250 300

y deepening with soaring prices

10 15 100 150 200 5

2002Q1 2003Q1 2004Q1 2005Q1 2006Q1 2007Q1 2008Q1

50

13

Number of Contracts in millions (lhs) IMF Non Fuel Commodities Index (rhs) IMF Energy Index (rhs)

SLIDE 14

Nombre de la presentación en cuerpo 17

C t if fi i li ti h l d f d t l l i b ti i ld t

4) Financialization of commodities

1200

Caveat: if financialization has played a fundamental role in boosting prices, we would expect

lower growth rates for those commodities that lack derivative markets. But…

Price changes in selected commodity markets from 2001 to 2008

1000

Non-exchange traded commodities Exchange traded commodities

600 800 400 600 200

14

Cadmium Molybdenum Rhodium Cobalt Ferrochrome Tungsten Rice Manganese Iron ore Ruthenium Steel Tin Copper WTI Lead Nickel Natural Gas Silver Gold Zinc Aluminium

Source: Deustche Bank (2008)

SLIDE 15

Nombre de la presentación en cuerpo 17

Mi t t f d i ti k t i l h d fi i l i t

4) Financialization of commodities

450 90

Microstructure of derivative markets: commercial hedgers vs. financial investors.
Strong increase in financial open positions in oil, gold and copper since 2002. But…

Non-commercial open positions in oil, gold and copper, 1986M1-2008M5

350 400 450 70 80 90

WTI (LHS)

250 300 350 50 60 70

Gold (RHS)

150 200 250 30 40 50

Copper (RHS)

50 100 150 10 20 30

15

1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008

SLIDE 16

Nombre de la presentación en cuerpo 17

Th i t i diff t f ft i lt l diti O fi i l iti f

4) Financialization of commodities

The picture is different for soft agricultural commodities. Open financial positions for

these commodities have increased recently but they are lower than those of the mid-

nineties. Financial investor activity is not totally a new phenomenon.

700

Soybean

Non-commercial open positions in soybeans, corn and wheat, 1986M1-2008M5

500 600

y Corn Wheat

300 400 100 200

16

100 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008

SLIDE 17

Nombre de la presentación en cuerpo 17

4) Financialization of commodities

120 400

Non-commercial net positions in

50 500

Non-commercial net positions in (th d t t ) lh

Oil Copper The impact of speculative activity on prices could depend on net positions…

60 80 100 250 300 350

p WTI (thousand contracts) lhs WTI index price rhs

20 30 40 350 400 450

copper (thousand contracts) lhs Copper index price rhs

Net long 20 40 150 200 250 Net long 10 20 200 250 300 g

40
20

50 100 Net short

20
10

50 100 150 Net short

60

1986 1987 1989 1991 1993 1995 1997 1998 2000 2002 2004 2006 2008 Net short

30

1986 1988 1991 1994 1997 1999 2002 2005 2008

Not all net long positions correspond to high prices.

17

It is not necessary to have neither commodity derivative markets nor strong net long

positions of financial participants to experience a commodity price boom or a bust.

SLIDE 18

Nombre de la presentación en cuerpo 17

5) A heterogeneous agent model

Commodity price change in next period is determined by the interaction of 3 agents: fundamentalists (F), chartists (C) and portfolio managers (PM) in accordance to:

           

PM 1 t 3 F 1 t 2 C 1 t 1 1 t

P E a P E a P E a P

   

      

Expectations of the fundamentalists are Expectations of the fundamentalists are based on the notion of commodity price reversion towards long run equilibrium

 

   

t t t F 1 t

X F P P E    





t

M

Chartists employ technical analysis and follow the current trends in prices

 

1 t t C 1 t

P P P E

 

   

Portfolio Managers have an information advantage and adjust their expectations

  

 

  

F C PM

P E P E 1 P E   

18

g j p employing a weighted average of Fundamentalist and Chartist expectations

  

 

  

F 1 t t C 1 t t PM 1 t

P E w P E w 1 P E

  

      1 w

t 



SLIDE 19

Nombre de la presentación en cuerpo 17

5) A heterogeneous agent model

wt plays a crucial role because it governs the weight given to F and C expectations by

PM at time t.

   

 

2

1 This variable adjusts endogenously according to the size of past misalignment. It is the f li i i h d l

   

 

2 d t d t d t t

X F P exp 1 w

   

   

Exponential specification source of non-linearity in the model.

Parameter

determines the speed in which portfolio managers adjust their expectations



expectations.

General expression of commodity price change is obtained by replacing equations and

rearranging terms. Thus, rearranging terms. Thus,

 

 

 

 

Price dynamics depends on several factors

19

 

 

 

 

t 2 d t 3 t 2 d t 3 t 2 t 3 1 1 t

M M exp 1 a P M exp 1 a M a P ) a a ( P

  

                 

SLIDE 20

Nombre de la presentación en cuerpo 17

5) A heterogeneous agent model

1) Terms that we find in a standard ECM: a purely autoregressive term and an error correction factor.

 

 

 

 

t 2 d t 3 t 2 d t 3 t 2 t 3 1 1 t

M M exp 1 a P M exp 1 a M a P ) a a ( P

  

                 

2) Terms which generate the non-linear adjustment pattern. We will focus the empirical analysis in the non-linear error correction factor.

 

 

 

 

t 2 d t 3 t 2 d t 3 t 2 t 3 1 1 t

M M exp 1 a P M exp 1 a M a P ) a a ( P

  

                 

Outcome of the model: Equilibrium price adjustment is non-linear. The higher the gap between actual and equilibrium price is, the faster the price adjustment will be.

20

STAR model is suitable for testing this hypothesis once the long run equilibrium is

estimated.

SLIDE 21

Nombre de la presentación en cuerpo 17

6) Econometric results

Operational steps to implement a STAR model:

1) Linear model: it serves as a benchmark to contrast the non-linearity hypothesis. The estimation follows common techniques of time series analysis. 2) Linearity test: Teräsvirta (1994) suggests using a first order Taylor expansion to

btain an auxiliary regression to contrast the null hypothesis of linearity.

3) E ti ti It i d b li th d Th f i iti l diti 3) Estimation: It is done by a non-linear method. The use of proper initial conditions will increase the probability of reaching a maximum in the likelihood function. 4) Non linear I R analysis: Useful tools in STAR systems are the generalized I R 4) Non-linear I-R analysis: Useful tools in STAR systems are the generalized I-R functions. Intuition: effects of shocks depend on the history, size and sign of disturbances.

21

SLIDE 22

Nombre de la presentación en cuerpo 17

6) Econometric results

Variables
All-Comm Index: Food and Metal indexes from IFS (IMF). USA CPI is the deflator

300

All-Comm Index (1973=100)

200 250 100 150 50 100

22

1973 1975 1978 1980 1983 1985 1988 1990 1993 1995 1998 2000 2003 2005 2008

Real Nominal

SLIDE 23

Nombre de la presentación en cuerpo 17

D i

6) Econometric results

Drivers
US Real Exchange Rate (RER): Broad multilateral version published by the

F d l R B k f N Y k Federal Reserve Bank of New York.

Real International Interest Rate (IR): one-year Treasury constant maturity rate

f h B d f G f h F d l R S from the Board of Governors of the Federal Reserve System.

Industrial Production Index (PRO): IPIs of developed countries plus 4 emerging

A i i (K I di M l i d Chi ) i h d i i d i l Asian economies (Korea, India, Malaysia and China) weighted using industrial value added of each country. h b i i l ff

Real Dow Jones Index (DOW): Are there substitution or complementary effects

between commodities and stock markets? Gorton and Rouwenhorst (2004) and Deutsche Bank (2005) find a negative non-conditional correlation between them.

23

Time trend

SLIDE 24

Nombre de la presentación en cuerpo 17

 

RER t p 1 j p t j , 2 1 t 2 2 t

X M RER          



  

6) Econometric results

     

P t 2 d t p p t j 12 1 t 12 12 p p t j 11 1 t 11 11 t

TV exp 1 X M X M P                            

    

 

Empirical model: ESTAR

Exponential TVF

     

t d t 1 j p t j , 12 1 t 12 12 1 j p t j , 11 1 t 11 11 t

p         

 

 

 

RER t p 1 j p t j , 2 1 t 2 2 t

X M RER          



  

Non-linear term

j

 

IR t p 1 j p t j , 3 1 t 3 3 t

X M IR          



  

 

PRO p

X M PRO    



 

PRO t 1 j p t j , 4 1 t 4 4 t

X M PRO          



  

 

DOW p

X M DOW         



 

t 1 j p t j , 5 1 t 5 5 t

X M DOW          



  

Where:

 

DOW PRO IR RER P X

24

 

DOW , PRO , IR , RER , P X  ) X ( F P M

t t t

 

d t

TV 

is the Transition Variable lagged d periods is the misalignment and

) X ( F

t

is the long-run equilibrium is the misalignment and

SLIDE 25

Nombre de la presentación en cuerpo 17

 

RER t p 1 j p t j , 2 1 t 2 2 t

X M RER          



  

6) Econometric results

Empirical model: ESTAR
Econometric strategy adopted is in line with the Engle and Granger proposal.

We estimate in a first stage the long run equation for commodity prices and test for

We estimate in a first stage the long run equation for commodity prices and test for

cointegration.

Then, if a cointegration relationship is found, the whole error correction system will be

estimated using as regressor the misalignment obtained in the first step.

25

SLIDE 26

Nombre de la presentación en cuerpo 17

 

RER t p 1 j p t j , 2 1 t 2 2 t

X M RER          



  

t p p j 4 p t 3 t 2 1 t

t X X P     



  

     

6) Econometric results

Estimation of the long-run equation (equilibrium prices)
We apply the Dynamic OLS (DOLS) (Stock and Watson, 1993). It deals with potential

simultaneity among the regressors and small sample bias by the inclusion of leads and lags of the explanatory variable differences.

p t p p j 4 p t 3 t 2 1 t

t X X P     



  

     

Monte Carlo experiments show that DOLS performs better, particularly in small

samples, compared to alternative estimators of long-run parameters as those proposed by Engle and Granger (1987), Johansen (1988), and Phillips and Hansen (1990).

Standard statistical inference remains valid when heteroskedastic and autocorrelation

26

Standard statistical inference remains valid when heteroskedastic and autocorrelation

consistent (HAC) standard errors are employed.

SLIDE 27

Nombre de la presentación en cuerpo 17

 

RER t p 1 j p t j , 2 1 t 2 2 t

X M RER          



  

t p p j 4 p t 3 t 2 1 t

t X X P     



  

     

6) Econometric results

Estimation of the long-run equation (equilibrium prices)

OLS DOLS Coefficient Coefficient p-value Industrial Production Index 3.6124 4.1368 0.0000 Variable US Real Exchange Rate

0.7270
0.5994

0.0000 Real International Interest Rate

1.0907
1.5452

0.0138 Real Down Jones Index 0 1299 0 1086 0 0163 Real Down Jones Index

0.1299
0.1086

0.0163 Time Trend

0.0097
0.0111

0.0000 Constant

4.1659
6.9366

0.0000

Few cointegration tests have been developed for DOLS: Shin (1994) and Choi et al (2008)
We find evidence of cointegration applying Shin Test at 2.5% statistical significance.

27

For robustness check, we carried out the standard ADF non-cointegration test based on OLS

residuals and we rejected the null hypothesis of unit root residuals.

SLIDE 28

Nombre de la presentación en cuerpo 17

 

RER t p 1 j p t j , 2 1 t 2 2 t

X M RER          



  

t p p j 4 p t 3 t 2 1 t

t X X P     



  

     

6) Econometric results

Estimation of the long-run equation (equilibrium prices)

6 25

All-Comm Index (logarithmic scale)

5.85 6.05 6.25 5.25 5.45 5.65 4 65 4.85 5.05 4.25 4.45 4.65 M01 M07 M01 M07 M01 M07 M01 M07 M01 M07 M01 M07 M01 M07 M01

28

1973M 1975M 1978M 1980M 1983M 1985M 1988M 1990M 1993M 1995M 1998M 2000M 2003M 2005M 2008M

Equilibrium Actual

SLIDE 29

Nombre de la presentación en cuerpo 17

 

RER t p 1 j p t j , 2 1 t 2 2 t

X M RER          



  

t p p j 4 p t 3 t 2 1 t

t X X P     



  

     

6) Econometric results

Estimation of the long-run equation (equilibrium prices)

All-Comm Index Misalignment

20% 30% 40% 0% 10% 20%

20%
10%

0%

40%
30%

1 7 1 7 1 7 1 7 1 7 1 7 1 7 1

29

1973M01 1975M07 1978M01 1980M07 1983M01 1985M07 1988M01 1990M07 1993M01 1995M07 1998M01 2000M07 2003M01 2005M07 2008M01

Misalignment

SLIDE 30

Nombre de la presentación en cuerpo 17

 

RER t p 1 j p t j , 2 1 t 2 2 t

X M RER          



  

t p p j 4 p t 3 t 2 1 t

t X X P     



  

     

t 2 d t 1 t 3 2 d t p t p , 2 d t 1 t 2 d t p t p , 1 1 t 1 p t p , t

TV M TV X TV M TV X M X P                     

         

6) Econometric results

Linearity F-test results

Transition V i bl Lag LM Statistic p-value Variable g p AV1 1 1.38647 0.04912 AV4 12 1.38309 0.05040 AV3 12 1 21178 0 16396 AV3 12 1.21178 0.16396 AV4 11 1.19485 0.18166 AV1 6 1.15964 0.22287 AV2 1 1 15512 0 22859 AV2 1 1.15512 0.22859 AV2 12 1.12511 0.26910 AV2 6 1.12423 0.27036 AV4 4 1 12396 0 27074 AV4 4 1.12396 0.27074 AV3 5 1.12106 0.27490

Potential Transition Variables

30

d t

M 

d t

AVj 

Average of the current misalignment and those of the previous j periods; j range from 1 to 4 Misalignment

SLIDE 31

Nombre de la presentación en cuerpo 17

 

RER t p 1 j p t j , 2 1 t 2 2 t

X M RER          



  

t p p j 4 p t 3 t 2 1 t

t X X P     



  

     

t 2 d t 1 t 3 2 d t p t p , 2 d t 1 t 2 d t p t p , 1 1 t 1 p t p , t

TV M TV X TV M TV X M X P                     

         

 

1 t t t 1 t 1 t

M w ) 109 . ( ) 035 . ( w * M ) 109 . ( M ) 035 . ( GEC

  

     

0.0.0292) value

(p

22 . 55   

6) Econometric results

   

P 2 p p

1         

 

Estimation results of non-linear ECM

   

P t 2 d t 1 j p t j , 12 1 t 12 12 1 j p t j , 11 1 t 11 11 t

TV exp 1 X M X M P                            

      

 

0.2914) value

(p

035 .

11

  

0.0418) value

(p

0.109 α12    0.0292) value

(p

22 . 55   

 M

) 109 ( ) 035 ( * M ) 109 ( M ) 035 ( GEC  

) (p

12

) (p 

Global Equilibrium Correction Factor (GEC) =-0.074

 

1 t t t 1 t 1 t

M w ) 109 . ( ) 035 . ( w * M ) 109 . ( M ) 035 . ( GEC

  

     

Our findings support non-linear adjustment hypothesis due to heterogeneous agents in commodity markets.

h i li i l h (l i ) d d d i h i i h i

When misalignment is low enough (low regime), wt tends to 0 and second term vanishes. PM mimic chartist

and initial misalignment is widened at a 3.5% monthly rate. In this case, there is not equilibrium correction.

When the misalignment reaches a sizable value (high regime), wt tends to 1 and the GEC adjustment

31

g ( g g )

t

j coefficient attains a maximum of -7.4% monthly rate. PM assign a larger weight to F expectations. Therefore, we will observe price reversion toward the equilibrium.

SLIDE 32

Nombre de la presentación en cuerpo 17

 

RER t p 1 j p t j , 2 1 t 2 2 t

X M RER          



  

t p p j 4 p t 3 t 2 1 t

t X X P     



  

     

t 2 d t 1 t 3 2 d t p t p , 2 d t 1 t 2 d t p t p , 1 1 t 1 p t p , t

TV M TV X TV M TV X M X P                     

         

 

1 t t t 1 t 1 t

M w ) 109 . ( ) 035 . ( w * M ) 109 . ( M ) 035 . ( GEC

  

     

0.0.0292) value

(p

22 . 55   

6) Econometric results

1 00

Estimation results of non-linear ECM

Misalignment, transition function and global adjustment coefficient

4% 0.70 0.80 0.90 1.00 tion 2% 4% nt

Misalignment = 8.5% TVF = 0.32

0.40 0.50 0.60 0.70

n Variable Funct
2%

0% ustment Coefficien 0.10 0.20 0.30

0. 0

Transitio

6%
4%

Global Adju 0.00 0% 5% 10% 15% 20% 25%

Misalignment

8%

0% 5% 10% 15% 20% 25%

Misalignment

32

Maximum reversion speed (-7.4% monthly) is attained when misalignment surpasses 25%.
Corollary: higher misalignments in the past involve higher values of the transition function and this will

indicate stronger price adjustments.

SLIDE 33

Nombre de la presentación en cuerpo 17

 

RER t p 1 j p t j , 2 1 t 2 2 t

X M RER          



  

t p p j 4 p t 3 t 2 1 t

t X X P     



  

     

t 2 d t 1 t 3 2 d t p t p , 2 d t 1 t 2 d t p t p , 1 1 t 1 p t p , t

TV M TV X TV M TV X M X P                     

         

 

1 t t t 1 t 1 t

M w ) 109 . ( ) 035 . ( w * M ) 109 . ( M ) 035 . ( GEC

  

     

0.0.0292) value

(p

22 . 55   

6) Econometric results

Estimation results of non-linear ECM

Transition function in the 1973-2008 period

0.80 0.90 1.00 0.80 0.90 1.00 0.40 0.50 0.60 0.70

ansition Function

0.40 0.50 0.60 0.70

nsition Function

0.00 0.10 0.20 0.30 73 75 77 79 80 82 84 86 88 90 91 93 95 97 99 01 02 04 06 08

Tra

0.00 0.10 0.20 0.30

Tran

We assign observations to low regime if transition function takes values lower than 0.30 while we

will consider they belong to high regime every time transition function exceeds 0.70.

73 75 77 79 80 82 84 86 88 90 91 93 95 97 99 01 02 04 06 08

40%
20%

0% 20% 40%

Misalignment

33

w co s de t ey be o g to g eg e eve y t e t a s t o u ct o e ceeds 0.70.

56% of the time the market is dominated by chartists while fundamentalists only prevail about

18.85%. The rest of the time (25%) corresponds to transition periods.

SLIDE 34

Nombre de la presentación en cuerpo 17

 

RER t p 1 j p t j , 2 1 t 2 2 t

X M RER          



  

t p p j 4 p t 3 t 2 1 t

t X X P     



  

     

t 2 d t 1 t 3 2 d t p t p , 2 d t 1 t 2 d t p t p , 1 1 t 1 p t p , t

TV M TV X TV M TV X M X P                     

         

 

1 t t t 1 t 1 t

M w ) 109 . ( ) 035 . ( w * M ) 109 . ( M ) 035 . ( GEC

  

     

0.0.0292) value

(p

22 . 55   

6) Econometric results

Estimation results of non-linear ECM

Summarizing, results support the hypothesis that high gaps between actual and ilib i i d b d l i l f hil ll i li equilibrium prices tend to be corrected relatively fast, while small misalignments tend to persist over time without any endogenous correcting force taking place.

34

SLIDE 35

Nombre de la presentación en cuerpo 17

 

RER t p 1 j p t j , 2 1 t 2 2 t

X M RER          



  

t p p j 4 p t 3 t 2 1 t

t X X P     



  

     

t 2 d t 1 t 3 2 d t p t p , 2 d t 1 t 2 d t p t p , 1 1 t 1 p t p , t

TV M TV X TV M TV X M X P                     

         

 

1 t t t 1 t 1 t

M w ) 109 . ( ) 035 . ( w * M ) 109 . ( M ) 035 . ( GEC

  

     

0.0.0292) value

(p

22 . 55   

6) Econometric results

Non-linear impulse-response analysis (Koop et al., 1996)

US effective real exchange rate shock (1-std deviation) g ( )

0.50%

0.00%

0.50%

0.00%

1.50%
1.00%
1.50%
1.00%
3 00%
2.50%
2.00%
3 00%
2.50%
2.00%
3.50%

3.00% 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25

Months

3.50%
3.00%

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25

Months

35

a) Low misalignment state b) High misalignment state

SLIDE 36

Nombre de la presentación en cuerpo 17

 

RER t p 1 j p t j , 2 1 t 2 2 t

X M RER          



  

t p p j 4 p t 3 t 2 1 t

t X X P     



  

     

t 2 d t 1 t 3 2 d t p t p , 2 d t 1 t 2 d t p t p , 1 1 t 1 p t p , t

TV M TV X TV M TV X M X P                     

         

 

1 t t t 1 t 1 t

M w ) 109 . ( ) 035 . ( w * M ) 109 . ( M ) 035 . ( GEC

  

     

0.0.0292) value

(p

22 . 55   

6) Econometric results

Non-linear impulse-response analysis (Koop et al., 1996)

Industrial production shock (1-std deviation) p ( )

3.00% 3.50% 3.00% 3.50% 1.50% 2.00% 2.50% 1.50% 2.00% 2.50% 0 00% 0.50% 1.00% 0 00% 0.50% 1.00%

0.50%

0.00% 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25

Months

0.50%

0.00% 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25

Months

36

a) Low misalignment state b) High misalignment state

SLIDE 37

Nombre de la presentación en cuerpo 17

 

RER t p 1 j p t j , 2 1 t 2 2 t

X M RER          



  

t p p j 4 p t 3 t 2 1 t

t X X P     



  

     

t 2 d t 1 t 3 2 d t p t p , 2 d t 1 t 2 d t p t p , 1 1 t 1 p t p , t

TV M TV X TV M TV X M X P                     

         

 

1 t t t 1 t 1 t

M w ) 109 . ( ) 035 . ( w * M ) 109 . ( M ) 035 . ( GEC

  

     

0.0.0292) value

(p

22 . 55   

6) Econometric results

Non-linear impulse-response analysis (Koop et al., 1996)

US real interest rate shock (1-std deviation) ( )

0 00% 0.50% 0 00% 0.50%

1.00%
0.50%

0.00%

1.00%
0.50%

0.00%

2.00%
1.50%
2.00%
1.50%
3.00%
2.50%

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25

Months

3.00%
2.50%

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25

Months

37

a) Low misalignment state b) High misalignment state

SLIDE 38

Nombre de la presentación en cuerpo 17

 

RER t p 1 j p t j , 2 1 t 2 2 t

X M RER          



  

t p p j 4 p t 3 t 2 1 t

t X X P     



  

     

t 2 d t 1 t 3 2 d t p t p , 2 d t 1 t 2 d t p t p , 1 1 t 1 p t p , t

TV M TV X TV M TV X M X P                     

         

 

1 t t t 1 t 1 t

M w ) 109 . ( ) 035 . ( w * M ) 109 . ( M ) 035 . ( GEC

  

     

0.0.0292) value

(p

22 . 55   

6) Econometric results

Non-linear impulse-response analysis (Koop et al., 1996)

Real Dow Jones index (1-std deviation) ( )

2.50% 3.00% 2.50% 3.00% 1.50% 2.00% 1.50% 2.00% 0.00% 0.50% 1.00% 0.00% 0.50% 1.00%

0.50%

0.00% 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25

Months

0.50%

0.00% 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25

Months

38

a) Low misalignment state b) High misalignment state

SLIDE 39

Nombre de la presentación en cuerpo 17

7) Open questions: Structural changes

There are two processes that have intensified from the 2000s and both should

be taken into account when we use our empirical results to configure future scenarios for commodity prices:

Biofuels
Changes in international demand pattern of commodities due to structural

transformations in big emerging economies such as China and India.

39

SLIDE 40

Nombre de la presentación en cuerpo 17

7) Open questions: Biofuels

Higher petroleum prices and extended support policies have induced a rapid growth in

biofuel production in recent years.

World biofuel production World biofuel production

(equivalent daily oil barrels)

1.8

million barrels

3.0

%

Brazil 1.2 1.5 2.0 2.5 U.S. Others Contribution in total fuel supply (right axis) 0.9 1.5 0.3 0.6 0.5 1.0

40

0.0 2005 2006 2007 2008 2009 (e) 0.0 Source: IEA (e): estimated

SLIDE 41

Nombre de la presentación en cuerpo 17

7) Open questions: Biofuels

United States and Brazil dominate ethanol production and also constitute the main

consumer markets.

World ethanol production

Year 2009

Brazil 32.7%

Producer Feedstock United States Corn

European Union 5.2% China 2.7%

Brazil Sugarcane European Union Sugar beet, wheat, corn

USA Thailand 2.2% Canada 1.4%

41

USA 53.4% Rest 2.4% Source: Renewable Fuels Association

SLIDE 42

Nombre de la presentación en cuerpo 17

7) Open questions: Biofuels

Biodiesel production has also experienced a pronounced expansion, with the

European Union as the main producer and market.

World biodiesel production

million Tons

World biodiesel production forecast 2010

France 11.5% Rest 17.0%

14.3 15.9 19.2 15 20 25 Argentina Brazil USA European Union Rest

Germany 14.2% Italy 5.1% Brazil 10.4% Thailand 3.4%

7.1 9.5 5 10 15

Rest EU 19.0% USA 11.0% Argentina 8.4% Source: Oil World

5 2006 2007 2008 2009 e 2010 p Source: Oil World

Producer Feedstock European Union Rapeseed United States Soybean

42

Argentina Soybean Brazil Soybean Thailand Palm oil

SLIDE 43

Nombre de la presentación en cuerpo 17

7) Open questions: Changes in demand patterns

Evidence shows a strong relationship between income per capita and the quantity and

quality of the diet. quality of the diet.

“Nutritional transition” (Popkins, 1993; Popkins, Horton and Kim, 2001).
First stage: income increases from very low levels elevate food consumption.
Second stage: When income continues going up not only calory consumption

increases but diet becomes more diversified: Cereals lose importance while meat, fish, fruits and vegetables gain shares.

43

SLIDE 44

Nombre de la presentación en cuerpo 17

7) Open questions: Changes in demand patterns

However, China and India are going thorough this process at a higher speed than
ther countries in the past. This has a big impact on commodity demand.

120 140

USA

Income per capita and meat consumption (2002)

80 100 120 apita (Kg)

Brazil Argentina Israel

40 60 80

nsumption per ca

China Japan Korea

20 40 Co

India

44

5,000

10,000 15,000 20,000 25,000 30,000 35,000 40,000 Income per capita (PPP)

Source: Elaborated by the authors elaboration based on FAO data.

SLIDE 45

Nombre de la presentación en cuerpo 17

7) Open questions: Changes in demand patterns

High growth performance of China and India had also a great effect on demand of metals
High growth performance of China and India had also a great effect on demand of metals.
There exists an inverse U-shape relationship between metal consumption and income level.

Consumption increases up to 15,000 or 20,000 per capita GDP in PPP dollars (IMF, 2006).

1 200 1,400 16 18 Korea

Income per capita and metals consumption (2005)

800 1,000 1,200 capita (Kg) 12 14 Copper Cons Japan 600 800 teel Consumption per 6 8 10 sumption per capita (k USA p 200 400 St 2 4 kg)

Steel Consumption per person (LHS) Copper Consumption per person (RHS)

India

USA China

45

5

10 15 20 25 30 35 40 45 50

Income per capita (USD current USD thousand)

Source: Elaborated by the authors based on data from IMF, International Copper Study Group and World Steel Association data.

SLIDE 46

Nombre de la presentación en cuerpo 17

7) Open questions: Changes in demand patterns

Structural changes were reflected in world trade.

Chi d I di h i ld t d

8% 9% 10%

China and India share in world trade

6% 7% 8% 3% 4% 5% 1% 2% 3%

46

0% 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008

Source: Elaborated by the authors based on UN Comtrade

SLIDE 47

Nombre de la presentación en cuerpo 17

7) Open questions: Changes in demand patterns

M i l i dit k t

Soya beans imports -China and India-

as % of world imports

Mainly in commodity markets…

50 60

%

40 20 30 10 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008

47

Source: Elaborated by the authors based on UN Comtrade

1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008

SLIDE 48

Nombre de la presentación en cuerpo 17

7) Open questions:

P

S0

1

S1

Demands of US ethanol program alone

account for over half the world´s unmet d f l (Th E i t D 6th

P0 1 P1 2 P2

need for cereals (The Economist, Dec 6th, 2007)

D0 D1

Prices will fall back if harvests increase.

Q Q0 Q1 Q2

But harvests rise if new land is brought

into cultivation or yields go up.

48

SLIDE 49

Nombre de la presentación en cuerpo 17

7) Open questions:

P

Short run:

R t d t b ti k 10% i i i

S0

1

S1

Response tends to be sticky: a 10% rise in prices yields a 1-2% increase in supply (IFPRI)

Long run:

P0 1 P1 2 P2

Long run:

New land is in remote parts of Brazil, Russia,

Kazakhstan…It would require big investments in

D0 D1

infrastructure. This could take a long time (and

environmental damage?)

GM

foods could increase yields strong. Their

Q Q0 Q1 Q2

introduction takes time too!

High oil prices would reduce the use of oil-based

fertilizers which have contributed to the sharply

49

fertilizers which have contributed to the sharply increase in yields.

SLIDE 50

Nombre de la presentación en cuerpo 17

7) Open questions:

P

So, price will probably remain high until supply

increases

S0

1

S1

increases.

In the meantime some people (and countries)

P0 1 P1 2 P2

suffer more than others.

Asymmetric cost-benefits distribution because

D0 D1

Asymmetric cost benefits distribution because

f Consumption Basket Composition Effect:

The higher the food share in consumption basket

Q Q0 Q1 Q2

(relative to oil share), the higher the transition cost.

50

SLIDE 51

Nombre de la presentación en cuerpo 17

8) Conclusions

For us it is appealing to think financialization as an amplifier factor of commodity

price cycles. In our framework speculative activity it is an element that affects short r n price d namics b t not the long r n eq ilibri m run price dynamics, but not the long run equilibrium.

It is not necessary to have neither commodity derivative markets nor strong net long

positions of financial participants to experience a commodity price boom or a bust positions of financial participants to experience a commodity price boom or a bust.

In the theoretical model, fundamentals continue to be the only real force that drives

long run prices However heterogeneity in expectations among market participants is long run prices. However, heterogeneity in expectations among market participants is important in determining the adjustment properties to equilibrium.

Regarding fundamentals we confirm roles for US real exchange rate industrial
Regarding fundamentals we confirm roles for US real exchange rate, industrial

production (world demand), and real interest rate. We verify a negative conditional correlation with the real return of stock markets.

51

SLIDE 52

Nombre de la presentación en cuerpo 17

8) Conclusions

Our findings support the idea that commodity prices tend to correct toward the

equilibrium, but this correction takes place if past misalignment is high enough. Thus, in the “low misalignment regime” correcting forces do not prevail and prices can , g g g p p move in any direction, depending on market sentiments.

Policy implications for emerging countries
For commodity-dependent emerging countries: Misalignment should be carefully
monitored. Price reversions tend to be abrupt when misalignment surpasses 20-25%.
Factors affecting commodity prices (real interest rates, real exchange rates, asset

returns) are similar to those that influence capital flows. This explains why it is hard for d l i i hi f d h k i h l fi Th developing countries to cushion terms of trade shocks with external finance. The same fundamentals that worsen terms of trade affect negatively the access to international credit market.

52

Structural changes (Biofuels and demand pattern changes) should be taken into

account when we configure future scenarios for commodity prices.

SLIDE 53

Nombre de la presentación en cuerpo 17

Thanks!

53

SLIDE 54

Nombre de la presentación en cuerpo 17

1) Motivation

Agricultural production for some countries of Latin America has an important contribution to the GDP.

Agricultural GDP

20.8 21.5 Paraguay Guatemala 6.9 8.0 8.5 10.6 Peru Costa Rica Colombia Ecuador 5.9 6.0 6.4 6.9 Dominican Republic Chile Uruguay Peru 3 7 4.7 5.2 5.6 V l Argentina Brazil Latin America

54 54

3.7 3.7 5 10 15 20 25 Mexico Venezuela % of total GDP Source: FAO

SLIDE 55

Nombre de la presentación en cuerpo 17

1) Motivation

In addition, several countries have a high share of agricultural population.

Agricultural population

37 40 52 Costa Rica Paraguay Guatemala 29 32 35 Peru Dominican Republic Ecuador 15 23 26 Brazil Mexico Colombia 8 8 12 Uruguay Argentina Chile % of total population

55 55

7 10 20 30 40 50 60 Venezuela p p Source: FAO

SLIDE 56

Nombre de la presentación en cuerpo 17

1) Motivation

Finally, the Latin America’s contribution to the world agricultural GDP is important

Agricultural GDP g

1.3 1.9 3.2 Argentina Mexico Brazil 0.4 0.5 0.8 Venezuela Chile Colombia 0.2 0.4 0.4 Ecuador Peru Guatemala 0.1 0.1 0.2 Costa Rica Paraguay Dominican Republic % of world agricultural GDP

56 56

0.1 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 Uruguay % of world agricultural GDP Source: FAO

SLIDE 57

Nombre de la presentación en cuerpo 17

2) Commodity Prices: recent developments and long run trends

Food and metal price IMF indexes (nominal and real, 1960=100)

Contrary to the belief that prices reached historical high levels, long-run perspective shows decaying prices if world inflation is incorporated

600 700

decaying prices if world inflation is incorporated

400 500 100 200 300 100 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 Food Food (real) Metals Metals (real)

57

( ) ( )

SLIDE 58

Nombre de la presentación en cuerpo 17

2) Long-run drivers of commodity prices

Prebisch (1950) and Singer (1950) claimed that, contrary to the classical view,

primary product prices would fall relatively to those of the industry. The influence of this hypothesis on empirical research has been significant.

A different

approach for studying commodity prices starts asking which A different approach for studying commodity prices starts asking which macroeconomic factors could act as determinants of them.

US Real Exchange Rate:
Pioneering single-good model of Ridler and Yandle (1972) demonstrates that an

increase a real exchange rate (appreciation) should result in a fall in dollar commodity prices.        * CPI P ln 

58

The implied elasticity is (Dornbusch, 1985):

                    * * * eCPI CPI ln CPI    

SLIDE 59

Nombre de la presentación en cuerpo 17

Motivación 3) Financialization of commodities

The direction of causality…

Net financial positions and commodity prices, correlations and causality, 1986-2008

COPPER GOLD SILVER WTI SUGAR SOYA MAIZE WHEAT T vs T-3 0.35 0.64 0.53 0.39 0.09 0.52 0.58 0.34 T vs T-2 0.37 0.59 0.50 0.39 0.07 0.51 0.56 0.37 T vs T-1 0.38 0.50 0.46 0.40 0.05 0.47 0.50 0.39 T vs T 0.39 0.42 0.44 0.41 0.02 0.40 0.42 0.37 T vs T+1 0.22 0.27 0.17 0.16 0.07 0.20 0.26 0.21 T vs T+2 0.19 0.29 0.13 0.11 0.15 0.15 0.24 0.20 T vs T+3 0.20 0.34 0.14 0.12 0.18 0.12 0.21 0.17 Granger Causality Test

Price Variantions do not

0.82225 1.57599*

Changes in Net Financial

1.77820** 0.67559 0.86989 0.80581 0.30402 0.98956 * **

Granger Cause Changes in Net Financial Positions

*** 59

Positions do not Granger Cause Price Variantions

1.56979* 1.29875 0.80934 2.09115** 1.21476 0.76235 2.785*** 1.05728

SLIDE 60

Nombre de la presentación en cuerpo 17

 

RER t p 1 j p t j , 2 1 t 2 2 t

X M RER          



  

t p p j 4 p t 3 t 2 1 t

t X X P     



  

     

t 2 d t 1 t 3 2 d t p t p , 2 d t 1 t 2 d t p t p , 1 1 t 1 p t p , t

TV M TV X TV M TV X M X P                     

         

5) Econometric results

Linearity F-test (Teräsvirta, 1994)

1) Estimate the model under the assumption of linearity and compute the RSS0 of the commodity price equation..

2 2

TV M TV X TV M TV X M X P                

2) Estimate the auxiliary regression and calculate the RSS1:

t 2 d t 1 t 3 2 d t p t p , 2 d t 1 t 2 d t p t p , 1 1 t 1 p t p , t

TV M TV X TV M TV X M X P                     

         

3) Compute the critical value of the F test ) p

 

1 1

RSS l RSS RSS F  

60 2 0 l

RSS

SLIDE 61

Nombre de la presentación en cuerpo 17

7) Open questions: Biofuels

The main liquid biofuel by volume is ethanol, which has quadrupled its production in

the last decade.

World ethanol production

90.4 90 100

million m3

70 80 90 +310% 40 50 60 29.2 10 20 30

61

10 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 Source: FO Licht

SLIDE 62

Nombre de la presentación en cuerpo 17

7) Open questions: Biofuels

Corn consumption for ethanol production accounts for almost 40% of corn demand in

the USA.

USA corn consumption

350 million Tons 50 % For ethanol production Other uses 210 280 30 40 Ethanol / total (right axis) 140 210 20 30 70 10

62

2004/2005 2006/2007 2008/2009 2010/2011* 2012/2013

Source: USDA and FAPRI Forecast

SLIDE 63

Nombre de la presentación en cuerpo 17

7) Open questions: Biofuels

Future targets and investment plans suggest strong gorwth will continue in the near

future.

Biodiesel production forecast

4,000 million gallons

Ethanol production forecast

20,000 million gallons 2,500 3,000 3,500 Euro zone United States Brazil Argentina 12,000 16,000 United States Euro zone Brazil 1,000 1,500 2,000 4,000 8,000 500 2005 2007 2009 2011 2013 2015 2017 2019 Source: FAPRI , 2005 2007 2009 2011 2013 2015 2017 2019 Source: FAPRI

63

SLIDE 64

Nombre de la presentación en cuerpo 17

7) Open questions: Biofuels

Several reports agree that biofuel production played a large role in the record increase

in food commodity prices from 2005 to mid-2008.

Source Estimate Commodity Time period World Bank (April 2008) 75% global food index January 2002–February 2008

Alternative estimates of the impact of biofuel production on market prices

IFPRI (May 2008) 39% 21-22% corn rice & wheat 2000–2007 2000–2007 CEA (May 2008) 35% corn March 2007–March 2008 CEA (May 2008) 3% global food index March 2007–March 2008 Collins (June 2008) 25-60% 19-26% corn US retail food 2006–2008 2006–2008 Glauber (June 2008) 23-31% 10% 4-5% commodities global food index US retail food April 2007–April 2008 April 2007–April 2008 January–April 2008 42% coarse grains 2008–2017

64

OECD-FAO (May 2008) 34% 24% vegetable oils wheat 2008–2017 2008–2017 Source: FAO Secretariat

SLIDE 65

Nombre de la presentación en cuerpo 17

7) Open questions: Biofuels

Future targets and investment plans suggest strong gorwth will continue in the near

future.

Biodiesel production forecast

4,000 million gallons

Ethanol production forecast

20,000 million gallons 2,500 3,000 3,500 Euro zone United States Brazil Argentina 12,000 16,000 United States Euro zone Brazil 1,000 1,500 2,000 4,000 8,000 500 2005 2007 2009 2011 2013 2015 2017 2019 Source: FAPRI , 2005 2007 2009 2011 2013 2015 2017 2019 Source: FAPRI

65

SLIDE 66

Nombre de la presentación en cuerpo 17

7) Open questions: Changes in demand patterns

China and India have been experimented very high growth rates.

GDP Growth Rates (1980 – 2009)

12% 16% China India Advanced economies Western Hemisphere World 8% World 4% 0%

66

4%

1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009e

SLIDE 67

Nombre de la presentación en cuerpo 17

7) Open questions: Changes in demand patterns

However, China and India are going thorough this process at a higher speed than
ther countries in the past. This has a big impact on commodity demand.

China India

Diet composition of China and India (1990 – 2005)

Producto 1990-92 1995-97 2003-05 1990-92 1995-97 2003-05 Alcoholic Beverages 3,0 3,8 4,0 0,3 0,4 0,4 Cereals 42,7 32,8 21,2 42,6 40,2 36,6 Eggs 1,5 2,3 2,6 0,3 0,4 0,4 Fish and seafood 2,5 3,7 3,4 1,1 1,1 1,2 Fruits 3,9 5,9 7,2 7,4 8,5 7,9 Meat 5,9 6,8 7,6 1,2 1,2 1,3 Milk 1,3 1,3 2,7 13,9 14,8 15,5 Oilcrops 1,1 1,2 0,9 1,9 1,9 1,7 Pulses 0,4 0,2 0,1 3,4 3,2 2,8 Starchy Roots 12 1 10 5 9 8 5 2 5 5 5 2 Starchy Roots 12,1 10,5 9,8 5,2 5,5 5,2 Sugar and Sweeteners 1,7 1,4 1,0 5,9 5,7 5,8 Vegetable Oils 1,2 1,2 1,4 1,7 1,9 2,4 Vegetables 21,5 27,2 35,9 13,8 13,7 17,2 Others 1,3 1,6 2,2 1,4 1,5 1,7 T t l 100 0 100 0 100 0 100 0 100 0 100 0

67

Total 100,0 100,0 100,0 100,0 100,0 100,0

Source: Elaborated by the authors based on FAO data.

SLIDE 68

Nombre de la presentación en cuerpo 17

7) Open questions: Changes in demand patterns

Iron ore imports -China and India-

as % of world imports

60 % 40 50 20 30 10 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008

68

1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008

Source: Elaborated by the authors based on UN Comtrade

SLIDE 69

Nombre de la presentación en cuerpo 17

7) Open questions: Changes in demand patterns

Crude Oil Imports -China and India-

as % of world imports

25 % 15 20 10 5

69

Source: Elaborated by the authors based on UN Comtrade

1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008