The Long Term Trend (?) in Cocoa Prices

The Long Term Trend (?) in Cocoa Prices
Christopher L. Gilbert
(University of Trento, Italy)
[email protected]
Presentation prepared for the First Conference on the
Economics and Politics of Chocolate, University of Leuven,
Belgium, 16-18 September 2012.
• Starting from Singer (1950) and Prebisch (1952, 1962), there has been a
concern that the purchasing power of primary commodity prices has been
declining and that that this decline might be expected to continue.
• The Prebisch-Singer debate has focussed almost exclusively on indices of
overall primary commodity prices or sub-indices thereof (agricultural
prices, metals and minerals etc.). Much less attention has been given to
the specific components of these indices even though producers or
countries are interested in specific products and not the prices of baskets
of commodities. (The exception is Kim et al, 2003).
• There is no reason to expect a trend in an aggregate if there is no trend in
the component prices.
• Furthermore, if trends in at least some individual prices are stochastic,
that property will be inherited by the aggregate index unless there is
cointegration between those prices with stochastic trends.
• It therefore makes sense to look at the prices of individual commodities
and not aggregates both from a practical and a theoretical perspective.
Cocoa prices, 1850-2011
I use a number of different sources with the objective of obtaining a
consistent annual price series extending from 1850 to 2011.
ICCO, Quarterly Bulletin of Cocoa Statistics (from 1990, but I now have
from 1957).
IMF, International Financial Statistics (1948-56, source stated as ICCO but
this is incorrect).
Weymar (1968), 1948-63, Accra beans in New York
Grilli and Yang (GY, 1988), 1900-47. This is an index set at 1977-79=100.
The source is a 1987 working paper for which no copy appears to exist.
Clarence-Smith (2000) reports the average price of cocoa imports
through Hamburg from 1850-1896 and reproduced from Mitchell (1992).
There is a problem in joining the GY and Hamburg series. I have used a
series from Clarence-Smith (2000) for the highest price paid by Cadbury
for São Tomé cocoa, yearly, from 1880-1911.
Real cocoa prices, 1850-2011
I deflate by the US Producer Price
Index (all items) which is available
from the start of the C19.
The result is a (cif) price for cocoa
relative to a basket of all wholesale
The trend, if any, in “real” cocoa
prices will depend in part on the
choice of deflator.
Grilli and Yang (1988), in common
with most others, deflate by
developed country export unit
values to give a terms of trade
Cocoa producers will be interested
in other prices - such as those of
oil and grains as well as of
manufactures – and in any case,
that deflator is not available back
to 1850.
Six cocoa epochs
We can visually distinguish six cocoa epochs.
Mixed low and high prices from 1850-80.
25 years of high prices from the mid 1880’s until 1910.
A sharply declining trend with high volatility from 1910-20
20 years of low prices from the mid 1920s to the end of
25 years of moderately high prices from the end of the WW2
until 1980 (punctuated by a briefer period of low prices in
the 1960s).
30 years of low and non-volatile prices from 1980 to the
Estimates of the cocoa price trend are likely to have varied
according to the epoch.
The cocoa price is non-stationary
Standard ADF tests show that the real cocoa price is neither stationary nor
Stationarity tests
Constant + trend
ADF statistic
ADF(2) = -2.53
ADF(2) = -2.89
ADF(1) = -11.56
5% critical value
Sample: 1855-2011. The test lag length was selected using the AIC.
The failure to reject trend non-stationarity, i.e. the conclusion that the price
is non-stationary even allowing for a deterministic trend, implies that any
estimated trend will be non-constant.
Trend estimates in different periods
The graph shows the recursive least
squares estimate of the cocoa price
trend from 1865-1900.
These are the estimates of the
slope from the regression of the log
real cocoa price on a deterministic
time trend from 1850 to the year in
The broken line gives the
boundaries of the 95% confidence
A significant negative trend estimate would have resulted for samples
terminating in 1937-58 and 1985-2011.
A significant positive trend estimate would have resulted for samples
terminating in 1881-1922.
If no trend, what does explain long term price
• There is no trend in cocoa prices – simply a shifting
equilibrium between production and consumption.
• The missing element in the story is the cocoa investment
function. Investment in the stock of trees determines the
position of the cocoa supply curve over the decades that
• High cocoa prices are certainly one factor which induce new
planting, but decisions are also made by governments, in
some cases motivated by more general developmental
The “Great Chocolate Boom”, 1880-1910
• Clarence-Smith (2000) defines the period 1880-1914 as “the
great chocolate boom”. He states, “The explosive growth of a
mass market for chocolate from the 1880s transformed world
consumption more radically than at any other time in history.
• Over this period, cocoa was primarily a Latin American and
Caribbean crop. The Portuguese colonies of Angola and São
Tomé and Prìncipe were the sole significant African exporters.
• Reported exports from cocoa-exporting territories grew at an
average annual rate of 4.85% over the 25 years 1880-1914 as
compared with only 1.9% over the 25 years 1850-74. I
interpret this as a rightward shift in the demand curve with an
upward sloping supply curve with only limited investment in
new planting.
What caused the boom to end?
It is tempting to believe that high
prices ended through a decline in
demand brought about by the First
World War. This is a possible factor
but the price decline set in well
before 1914. 1907 was the last high
price year. By 1910 the price was only
slightly more than half that of 1907.
Clarence-Smith (2000) makes much
of 1907 cartel action led by
producers in São Tomé who withheld
supplies from the market. I interpret
the cartel action as a (temporarily
successful) attempt to hold up prices
in a falling market.
I argue that the price falls in the first
decade of the C20 resulted from new
planting mainly in British colonial
territories, and in particular in the Gold
Coast (Ghana)., but also Nigeria,
Kameroun and Equatorial Guinea.
The chart shows African (light blue) and non-African cocoa exports from 1850-1914.
Low prices, 1910-45
• The cocoa trees planted in the 1890-1910 decades will have
been productive up to the Second World War. The cocoa
supply curve had therefore shifted sharply right. African, not
Caribbean, producers became the marginal suppliers.
• At the same time demand growth was slow and possibly
negative in the 1930s. This substantial supply overhang set
the stage for the political economy of supply management.
• The colonial authorities, particularly the British, established
marketing boards with the objective of smoothing prices over
time but also, by restricting exports, of avoiding “burdensome
surpluses” and hence maintaining prices to producers (and
margins in the supply chain).
The post-WW2 period
• The post-1945 European recovery led to renewed growth in chocolate
consumption and hence of cocoa grindings. The 1953 ending of
confectionary rationing in Britain was particularly important since Britain
was responsible for around 15% or world cocoa imports.
• African supply was now unresponsive, particularly in Gold Coast-Ghana,
because of the tight grip of marketing boards and the high wedge they
drove between fob and producer prices. Prices rose strongly in the early
1950s and, although volatile, were not low again until the 1980s.
• Côte d’Ivoire, under Houphoüet-Boigny, emphasized agriculture as the
route to development. Government allowed families, often Burkinabé, to
farm tropical forest where population densities were low. Substantial
intra- and international migration resulted in Côte d’Ivoire overtaking
Ghana as the world’s major cocoa exporter. The resulting downward
pressure on the cocoa price was offset by the simultaneous collapse in
Ghanaian exports caused by an excessive wedge between producer and
fob prices.
The 1980s and 1990s saw two important developments, both of which
put downward pressure on the cocoa prices.
Ghanaian cocoa production recovered.
Marketing chains were liberalized. Marketing boards and caisses de
stabilisation were dismantled or substantially reformed. These reforms
reduced costs in the marketing chain, allowing farmers to obtain a higher
share of the fob price. However, there was a fallacy of composition in
supposing that all countries could benefit in this way – world-wide
reform shifted the supply curve down reducing the fob price. Farmers
obtained a higher share of a lower price (Gilbert and Varangis, 2004).
What are the origins of cocoa investment?
• Cocoa is a smallholder crop. It is relatively easy for farmers to plant new
trees in a zone where cocoa production is already established. It is much
more difficult to establish cocoa in a new area where there are no supply
chain intermediaries and where farmers have no experience of drying and
• The major increases in cocoa investment in the periods 1890-1910
(primarily Gold Coast) and 1960-80 (primarily Côte d’Ivoire) were only
partially driven by high prices. Expansions resulted from policy decisions
by the governments or colonial authorities to develop cocoa.
• Similarly, market liberalization was politically driven.
• Once these new developments have been undertaken, the trees are
productive for 40 years ore more. It is seldom profitable to grub up trees
when prices are low – they are simply left and merge into the forest but
can be harvested again when prices justify this.
• I contend that it is these major developmental and policy decisions which
drive long term shifts in cocoa prices – not some inexorable trend.
• There is no trend in cocoa prices – simply a shifting equilibrium between
production and consumption.
• Once planted, cocoa trees remain productive for over 40 years. The result
is that over-supply can persist for up to 20 years.
• Major expansions of cocoa production are seldom simply a response to
high prices, although high prices may be a necessary precondition.
Governments are also, and perhaps necessarily, involved. Absent such
developmental decisions, periods of high prices can also be highly
• Cocoa consumption is likely to grow in the future as in the past,
particularly if Asian consumers develop a taste for chocolate. A major
expansion of demand will require a supply response. The future path of
cocoa prices will depend on whether the required expansion of cocoa
production follows the demand growth or whether governments instead
attempt to anticipate this growth by developing their own cocoa
production and chocolate manufacturing capabilities.
Thank you
for your

similar documents