economic impacts of el niño: the case of the the colombian ......economic impacts of el nino:˜ the...

Economic impacts of El Nino: the case ofthe the Colombian coffee market

Andrea Bastianin 1 Alessandro Lanza 2 Matteo Manera 3

1Universita degli Studi di Milano and Fondazione Eni Enrico Mattei (FEEM)

2Centro Euro-Mediterraneo sui Cambiamenti Climatici (CMCC)

3Universita degli Studi di Milano-Bicocca and Fondazione Eni Enrico Mattei (FEEM)

November 19, 2015 - FEEM-IEFE Joint Seminar

Introduction

Research questions & Results

What is ENSO and why do economists care?

ENSO and the Colombian coffee industry

A Structural VAR model for the coffee market

Results

Conclusions

IntroductionENSO: Current Status & Predictions

IntroductionENSO: Current Status & Predictions

ENSO Forecast in October 2014

ENSO Forecast in October 2015

Source: The International Research Institute for Climate and Society (IRI) - The Earth Institute, Columbia University.

IntroductionSea Surface Temperature Anomalies in 2015 & 1997 (Strong El Nino)

Sep 2015

120˚E 150˚E 180˚ 150˚W 120˚W 90˚W 60˚WLongitude

30˚S

0˚30

˚NLa

titud

e-0.5

-0.5 -0.5

-0.5

-0.5

0.5

0.5

0.5

0.5

0.5

0.5

0.5

0.5

0.5

0.5

0.5

1

1

1

1

1

1.5

1.5

2

2

2.5

2.5

2.5

3 3.5

Sep 1997


30˚S

0˚30

˚NLa

titud

e

-1

-1

-1

-0.5

-0.5

-0.5

-0.5

-0.5

-0.5

0.5

0.5

1

1

1.5

1.5

2

2.5 3

3.54


IntroductionSea Surface Temperature Anomalies in 2014 (neutral) & 2007 (La Nina)

Sep 2014


30˚S

0˚30

˚NLa

titud

e

-0.5

-0.50.5

0.5

0.5

0.5

0.5

0.5

0.5

0.5

1

1

1

1

1

1.51.5

2

Sep 2007


30˚S

0˚30

˚NLa

titud

e

-2

-1.5

-1

-1

-0.5

-0.5

0.5

0.5

0.5

0.5

0.5

0.50.5

0.5

0.50.5

1

1


Introduction





Results

Conclusions

Research questions

1. Economic impacts of El Nino Southern Oscillation (ENSO) onthe Colombian coffee market

2. Develop an econometric model of the coffee marketI Structural Vector Autoregressive (VAR) model to describe how

coffee export, production and price react to ENSO, demandand supply shocks

Main Results• The impact of ENSO shocks is small compared with other supplyand demand-side innovations

I El Nino shocks increase production and reduce price;I La Nina shocks reduce production and increase price.

Introduction


What is ENSO and why do economists care?What is ENSO?How do we measure ENSO?How does ENSO affect world economies?Why do economists care about ENSO?Literature



Results

Conclusions

What is ENSO?• El Nino Southern Oscillation (ENSO) refers to theinteraction between the atmosphere and ocean in the tropicalPacific.

◦ El Nino refers to the ocean component of ENSO.

◦ The Southern Oscillation component identifies swings inatmospheric pressure between the eastern and western tropical Pacific.

• Peruvian fishermen observed the warmer water during theChristmas season: El Nino means the Little Boy, or Christ Child inSpanish.

What is ENSO?• El Nino and La Nina are opposite phases of the ENSO cycle.

El Nino La NinaENSO Warm Phase ENSO Cold Phase

Temperature Warmer than usual Cooler than usualSST1 anomalies Positive NegativeAir pressure below-normal at Tahiti above-normal at Tahiti

above-normal at Darwin below-normal at DarwinRainfall droughts more likely droughts more likelyanomalies throughout the tropics throughout much

of the mid-latitudes

• Effects: periodic changes in sea-level pressure, sea-surfacetemperature, precipitation and winds all around the world.

• Effects of El Nino and La Nina on weather and hence oneconomies vary through time and space.

1SST = Sea Surface Temperature

El Nino and Rainfall

La Nina and Rainfall

El Nino in Colombia

La Nina in Colombia

How do we measure ENSO?1. Sea Surface Temperature anomalies (SST): El Nino (La Nina)corresponds to five consecutive 3-month running mean of SSTanomalies in the Nino 3.4 region that is above (below) thethreshold of +0.5◦ C (-0.5◦ C).2

1950 1960 1970 1980 1990 2000 2010−2

−1.5

−1

−0.5

0

0.5

1

1.5

2

SS

T a

nom

alie

s

El NinoLa Nina

2SST is also known as the Oceanic Nino Index (ONI).

How do we measure ENSO?2. Southern Oscillation Index (SOI): a standardized index based onthe observed sea level pressure differences between Tahiti andDarwin (AUS).

1950 1960 1970 1980 1990 2000 2010

−3

−2

−1

0

1

2

3

SO

I

El NinoLa Nina

How do we measure ENSO?

SST and -SOI

1950 1960 1970 1980 1990 2000 2010

−2

−1

0

1

2

3

−SOISST

• corr(SST,-SOI) = 0.7• Results based on SST;• Results based on -SOI are qualitatively identical.

How does ENSO affect world economies?

How does ENSO affect world economies?

I Direct and indirect effects on growth and inflation:

I Commodity price inflation due to production shortfalls;

I Warmer water damages fishing industries;

I More rain is detrimental for coffee production (e.g. plantdiseases);

I Flooded mines;

I Droughts affecting hydroelectric power generation andwaterway transportation;

I Health effects;

I State budget used to recovery from weather disasters.

Why do economists care about ENSO?

I Hot topic for policy makersI Sizable economic consequencesI Weather risk management

Understanding El Niño

Commodity Markets Outlook

A World Bank

Quarterly Report

Q4Q3Q2Q1

OCTOBER 2015

Why do economists care about ENSO?Sizable economic consequences

Source: Changhon S.A. (1999) “Impacts of 1997–98 El Niño– Generated Weather in the United States” Bulletin of the American Meteorological Society , 1819-27.

It means: $6.2 - $6.4 billion (in 2015 dollars)

Why do economists care about ENSO?Sizable economic consequences

How much is 6 billion dollars?• The cost of CERN’s particle accelerator (Large Hadron Collider)is 6.1 billion dollars.3

• Berlusconi’s estimated net-worth is $7.4 billion4

3Source: CERN Ask an Expert service4Source: Forbes World’s Billionaires list, 2015

http://askanexpert.web.cern.ch/AskAnExpert/en/Accelerators/LHCgeneral-en.html

Why do economists care about ENSO?Weather risk management

Source: CME group

Weather derivatives at the Chicago Mercantile Exchange:I Temperature-based (HDD, CDD, Avg. Temp.)I Hurricane Index based derivativesI Frost and snowfall-based options and futures

What do economists say about ENSO?• ENSO - via its effects on cereal production, yield and agriculture value added -explainins the fact that wealth and growth are systematically lower in the tropics (Hsiangand Meng, 2015, panel of 78 countries);

• Non-oil commodity price inflation increases by 3.5% following ENSO shocks (Brunner,2002, quarterly structural VAR, 1968-1998);

• Berry and Okulicz-Kozaryn (2008), with annual data (1894-1999), fail to discover anylong-run or short-run co-movements of ENSO and U.S. growth and inflation.

• Ubilava (2012), relying on regime switching models, shows that the impact of ENSOshocks on the price of coffee depends on the variety of coffee, as well as on the sign ofthe shock to ENSO. He uses monthly data for the 1989-2010 period.

I A positive shock to ENSO (El Nino) reduces the price of Arabica coffee, whileincreases the price of Robusta coffee;

I A negative shock to ENSO (La Nina) rises the price of Arabica coffee, whiledepresses the price of Robusta coffee;

I The magnitude of the effects of positive and negative shocks is similar;I Impact of ENSO depends on geography: Arabica is mainly produced in South

America, Robusta in Southeast Asia.

Introduction





Results

Conclusions

The impact of ENSO in ColombiaWhat did Colombia export in 2013?

Source: Center for International Development at Harvard University, “The Atlas ofEconomic Complexity”

http://www.atlas.cid.harvard.edu


The impact of ENSO in ColombiaWhat did Colombia export in 1995?

Source: Center for International Development at Harvard University, “The Atlas ofEconomic Complexity”



The impact of ENSO in ColombiaWhere did Colombia export green and roasted coffee in 2013?

Source: Center for International Development at Harvard University, “The Atlas ofEconomic Complexity”.



The impact of ENSO in ColombiaArabica & Robusta coffee production 2014-15

Arabica & Robusta coffee production (thousand 60-Kg bags)

Arabica

Robusta

Source: USDA-FAS, Production, Supply and Distribution Online.

http://apps.fas.usda.gov/psdonline/

The impact of ENSO in ColombiaArabica coffee production 2014-15

Arabica coffee production (thousand 60-Kg bags)

Arabica

Source: USDA-FAS, Production, Supply and Distribution Online.

http://apps.fas.usda.gov/psdonline/

The impact of ENSO in ColombiaThe geography of El Nino impacts in Colombia

El Nino La NinaDrier weather Higher rainfallHigher temperatures Cooler weatherHigher productivity Lower productivityPeriods of drought and erosion Rain and floodsBroca bug La Roya fungusSource: National Federation of Coffee Growers of Colombia

“In principle La Nina is more harmful toColombian coffee growing than El Nino.”

Source: Carlos Armando Uribe, Technical Manager of the

Colombian Coffee Growers Federation (FNC).

http://www.cafedecolombia.com/bb-fnc-en/index.php/comments/how_el_nino_la_nina_affect_production_of_cafe_de_colombia

http://www.cafedecolombia.com/bb-fnc-en/index.php/comments/how_el_nino_la_nina_affect_production_of_cafe_de_colombia

Introduction




A Structural VAR model for the coffee marketA theoretical model for the coffee market

Results

Conclusions

A structural VAR model for the coffee marketData and model

We consider a VAR (i.e. multivariate) model for:1. sstt (or -soit);2. ∆prodt : (first difference of log) Colombian production of

coffee (bags);3. cexpt : (log) Volume of export of Colombian coffee (bags);4. rpct : (log) External price of Colombian coffee (US $ cents per

pound) - adjusted for inflation using US CPI.- Monthly data: 1990:1 - 2015:8.

- Coffee price was regulated by international agreements in the1962-1989 period: target price and export quotas for producingcountries.


1990 1995 2000 2005 2010 2015

5001000150020002500

Coffee Production (Bags of green coffee equivalent, 1000s)

1990 1995 2000 2005 2010 2015

−100

0

100

200First difference of log Coffee Production (%)

1990 1995 2000 2005 2010 2015

500

1000

1500

Coffee Export (Bags of green coffee equivalent, 1000s)

1990 1995 2000 2005 2010 2015

−50

0

50

Log Coffee Export (% deviation from mean)

1990 1995 2000 2005 2010 2015

2

4

6

Real price of coffee (dollar per pound)

1990 1995 2000 2005 2010 2015−100

0

100

Log Real Price of Coffee (% deviation from mean)


Let yt ≡ [sstt , ∆prodt , cexpt , rpct ]′, then a Structural VAR modelcan be written as:

A0yt = µt +24∑

j=1Ajyt−j + εt

µt includes a constant and seasonal dummy variables;

εt are structural shocks;

reduced form errors et are obtained as follows:

et = A−10 εt

A structural VAR model for the coffee marketIdentification

esst

te∆prod

tecexp

terpc

t

=

a11 0 0 0a21 a22 0 0a31 a32 a33 0a41 a42 a43 a44

εEl Nino shockt

εcoffee supply shockt

εcoffee demand shockt

εresidual shockt

Structural VAR models in a nutshell: use (economic) theory toimpose restrictions - zero restrictions, in this case - that help providingan economic interpretation of shocks hitting the variables in thesystem.


• An ENSO shock is an unpredictable change in the SST index.

• Positive (Negative) ENSO shocks identify El Nino (La Nina) events

• Exclusion restrictions imply ENSO shocks affect Colombian coffeeproduction, export and price within the same month, but notvice-versa.

• Without further restrictions, ENSO shocks can move the supply aswell as the demand for coffee


Response of coffee price to ENSO shocks

Source: Ubilava (2012)

• The model implicitly assumes that responses of coffee price, productionand export to positive and negative ENSO shocks are symmetric (i.e.responses are of opposite sign but of the same magnitude).

A theoretical model for the coffee marketVertical short-run supply

Co

ffe

e S

up

ply

Pri

ce

QuantityQ

1. Coffee production involveslong lags between planting,cropping, harvesting, andmarketing

I At least 2 years before acoffee tree begins to bearfruits

I Several other years to reachfull productivity

2. Consistent with the literatureI Price elasticity of supply is

very low (Ponte, 2002);I Theoretical models of coffee

production use verticalsupply functions (Wickensand Greenfield, 1973).

A theoretical model for the coffee marketShort-run demand

Co

ffe

e S

up

ply

Pri

ce

QuantityQ*

P*E

I Standard downward slopingdemand

A theoretical model for the coffee marketSupply shocks

S

Pri

ce

QuantityQ*

P*E

S’

Q*

P*’ E’

• Coffee supply shocks are innovations toColombian coffee production that cannotbe explained based on El Nino shocks;

• Coffee supply is assumed not torespond to shocks to the demand forcoffee or to residual shocks on impact(i.e. within the same month);

• This is plausible since adjusting coffeeproduction takes time and is costly;

• Coffee production adjusts only in themedium/long-run to expected trendgrowth in demand.

A theoretical model for the coffee marketDemand shocks

S

Pri

ce

QuantityQ*

P*E

Q*

P*’E’

• Innovations to Colombian coffeeexports that cannot be explained basedon El Nino or supply shocks are calledcoffee demand shocks.

• These shocks are specific to theexternal demand for Colombian coffee.

• Increases in the real price of coffee dueto the fourth shock affect coffee exportswith a delay of at least one month.

A theoretical model for the coffee marketDemand shocks

S

Pri

ce

QuantityQ*

P*E

Q*

P*’E’

• Innovations to the real price of coffeethat cannot be explained based on ElNino, supply, or external demand shocksare called residual shocks.

• Without further information it is notpossible to exactly define the nature ofthe fourth shock.

• It captures a variety of coffeedemand-side shocks such as changes inpreferences, shocks to internal coffeedemand and fluctuations in precautionarydemand for coffee driven by uncertaintyabout future coffee supply shortfalls.

Introduction





Results

Conclusions

Impulse response functions for the coffee market block

0 10 20

−20

−10

0

Coffee supply shockC

offe

e pr

oduc

tion

0 10 20

−20

−10

0

Coffee demand shock

0 10 20

−20

−10

0

Residual shock

0 10 20

−5

0

5

10

15Coffee supply shock

Cof

fee

expo

rt

0 10 20

−5

0

5

10

15Coffee demand shock

0 10 20

−5

0

5

10

15Residual shock

0 10 20

0

5

10

Coffee supply shock

Rea

l pric

e of

cof

fee

Months0 10 20

0

5

10

Coffee demand shock

Months0 10 20

0

5

10

Residual shock

Months

Impulse response functions: (positive) ENSO shock

0 5 10 15 20

−202468

ENSO shockC

offe

e pr

oduc

tion

0 5 10 15 20−2

0

2

4

6

ENSO shock

Cof

fee

expo

rt

0 5 10 15 20−5

0

5ENSO shock

Rea

l Pric

e of

Cof

fee

Historical Decomposition of the real price coffee


Shock to supply - La Roya fungus


Shock to precautionary demand - Drought in Brazil - Contribution of other shocks zero or negative

Variance Decomposition

Variance Decomposition of Coffee Production Variance Decomposition of Coffee Exportt ENSO Supply Demand Residual t ENSO Supply Demand Residual1 0.75 99.25 0.00 0.00 1 0.44 7.46 92.10 0.0012 7.73 84.30 3.58 4.39 12 10.33 24.68 59.82 5.17∞ 11.58 69.76 8.71 9.95 ∞ 23.26 25.83 41.53 9.38Variance Decomposition of Real Price of Coffeet ENSO Supply Demand Residual1 1.24 0.98 0.60 97.1812 2.09 4.95 6.80 86.17∞ 4.43 6.96 16.35 72.26

Introduction





Results

Conclusions

Conclusions

1. The impact of ENSO shocks is small compared with othersupply and demand-side innovations

I El Nino shocks increase production and reduce price;I La Nina shocks reduce production and increase price;

2. In the short-run, ENSO shocks explain 0.7% of thefluctuations of coffee production and 1% of the variability ofthe real price of coffee.

3. In the long-run these percentages rise to 12% and 4%,respectively.

4. Determinants of the price: demand-side shocks moreimportant than suply-side shocks.

Conclusions• When studying the impact of ENSO on commodities it isimportant to control for other fundamentals.

• The impact of a coffee price shocks on the economy of Colombiadepends on the source of the price shocke.g. demand-side price shocks more important than supply-sideprice shock?

ReferencesBerry, B. J. L. and Okulicz-Kozaryn, A. (2008). Are there ENSO signals in the macroeconomy? Ecological

Economics, 64:625–633.Brunner, A. D. (2002). El Nino and world primary commodity prices: warm water or hot air? The Review of

Economics and Statistics, 84(1):176–183.Hsiang, S. M. and Meng, K. C. (2015). Tropical economics. American Economic Review,, 1105(5):257–261.Ponte, S. (2002). The ‘Latte Revolution’? Regulation, markets and consumption in the global coffee chain. World

Development, 30(7):1099 – 1122.Ubilava, D. (2012). El Nino, La Nina, and world coffee price dynamics. Agricultural Economics, 43(1):17–26.Wickens, M. R. and Greenfield, J. N. (1973). The econometrics of agricultural supply: an application to the world

coffee market. The Review of Economics and Statistics, 55(4):433–40.

economic impacts of el niño: the case of the the colombian ......economic impacts of el nino:˜ the...

Documents