Vous êtes sur la page 1sur 8

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/269687211

An analysis of the impacts of rainfall and rainfall variability on cash and food
crop production in the Gambia

Article · June 2014

CITATIONS READS

0 389

1 author:

Bukhari M S Sillah
Islamic Development Bank
20 PUBLICATIONS   27 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Islamic Banking and Finance View project

All content following this page was uploaded by Bukhari M S Sillah on 13 November 2016.

The user has requested enhancement of the downloaded file.


Global Journal of Quantitative Science
Vol. 1. No.2. June 2014 Pp 17-23 .

An analysis of the impacts of rainfall and rainfall variability on


cash and food crop production in the Gambia

Bukhari Sillah,

Economics Department, College of Business Administration; King Saud University, KSA,


bsillah73@yahoo.com

Abstract
This paper employs panel data econometrics method to analyze the impacts of climate change on
cash and food crop production in the Gambia. It finds that land size and rainfall ultimately
determine changes in the crop production; the crop production does not respond to changes in its
own price changes, and there is no evidence to support the claim that farmers substitute cash crop
cultivation for the food crop cultivation to reap high earnings from cash crop price increases. In
fact, the data present evidence that farmers consider the production of the two crops as
complements. Groundnut, rice, and maize productions are found to respond positively to new
variety introduction, whereas cotton and sorghum productions have fallen in response to new
variety innovations

Introduction
This paper attempts to investigate analytically how the climate changes are being translated into
the changes of cash and food crop production in the Gambia. Cash crop production has a long
standing tradition in the Gambia, and it has been the backbone of the local economy giving
employment and living to more than 70% of the population. This tradition has been perceived to
largely at the expense of food production that the Gambia has become a net food importer. This
has been the situation since 1904 as the feeling can be read from this excerpt, “It is a difficult
matter in a small colony like the Gambia to deal year after year with the question of agriculture.
The groundnut is the main product, and the export has so often been described that there is
nothing fresh to add beyond once again stating that the welfare of the colony depends on this crop.
The people devote almost their whole attention to it, only growing small quantities of millet, Koos,
1
and corn for food.” But as people stick to the tradition and resist the change, the climate never
stays the same. Here we define climate to consist of rainfall and the variability of rainfall. We define
the cash crop to be the groundnut and the food crops to be paddy rice, sorghum, millet and maize.
This definition is also in accordance with what obtains in the National Agricultural Statistics. The
paper develops a production function for the crops, and then analyzes how the crop production
responds to the climate variable changes. The study is driven by the desire to ascertain the
responses of these agricultural crops to climate changes, so that since climate change mitigation
takes both long time and global effort, the Gambia farmer, can be advised correctly based on
evidence on how to adopt their agricultural traditions to climate change. The paper will be the first
of its kind in the Gambian context that will provide such econometric evidence for the relationships
between agricultural crops and the climate change in the Gambia. The evidence can be put with
other evidence around Africa to draw country case studies that can inform the Africans of policies
of climate change adoption. the study is organized as section 2 gives a background review of the
variables in the national context, section 3 reviews some relevant literatures, section 4 presents the
models, section 5 presents the findings and the analysis, and section 6 concludes and derives the
policy implications. The research is constrained by data; we wish to start the analysis from around
1843 when the groundnut trade was introduced in the Gambia, but the data would not permit us.
The data on the factors of productions of the crops are not adequately available to allow us to
conduct this study; instead we have contented ourselves with the producer prices to capture the
information about factors of productions other land, which has been explicitly included in the
model.

1
The Colonial Report of the Gambia 1904, p.16
Bukhari Sillah, An analysis of the impacts of rainfall and rainfall variability on cash and food crop
production in the Gambia http://www.gjqs.org/ Page 18

Background
In this background review, we take stock of the crop production and the climate change from 1974
to 2007. The groundnut has more than a century observations; thus a good picture can be drawn
from it. The other crops have only recently been given attention; there are limited observations on
them. The climate change has been taken for granted, very recently the national statistics has
started documenting it; thus, the paper, to some extent, has to make use of the international
records on the climate variables.

Crop productions
For the period 1974 – 2007, groundnut recorded the highest output of 151400 tons and the highest
yield of 1.537 tons per hectare in 1982. The lowest output for the groundnut production was in
1996 at 45800 tons. Rice experienced the highest output in 1980 at 42700 tons and the highest
yield of 2.364 tons per hectare in 1995; and rice is the only crop that registered a yield of more 2
tons per hectare. Its production dropped to the lowest in 2007 at 11400 tons. Sorghum registered
the highest production and the highest yield in 2001 at 33400 tons and 1.275 tons per hectares
respectively. The highest millet production was seen in 2004 at 132500 tons, and its highest yield
was seen in 1983 and 2003 at 1.09 tons per hectares. Maize’s highest output was recorded in
2003 at 33400 tons and its highest yield was recorded in 1974 at 1.982 tons per hectare. In terms
of output variability, millet recorded the highest among the crops at 0.6312 standard deviations per
mean value; and rice recorded the lowest output variability at 0.263standard deviation per mean
value. Millet production is often forgone for the sorghum and maize production, and that could be
reasons for its high output variability. Rice production is relatively stable as the traditional swamp
rice fields have been fairly cultivated annually; it experiences shocks when upland and irrigation
schemes are added. Upland rice fields heavily depend on rainfall, and irrigation schemes rely on
the machines and fuels, which are often unreliable. Rainfalls and inputs other land do explain
variations in the output; but farmers also vary the land allocated to a specific crop resulting in the
crop output variations. In term of land use, Millet has the highest land use variability of
0.567standard deviation per mean value; and groundnut has the lowest land use variability of
0.202standard deviation per mean value. This confirms our initial observation that millet is a crop
that farmers easily forgo for other crops; its land use could be varied easily to satisfy the land
allocation of other crops. This is not the case with groundnut, which is the cash source for the
farmers; thus, the opportunity cost associated with groundnut production is higher than that of the
other crops. There is declining trends in the yields of the crops. The other observation is that no
single year has experienced highest outputs for all the crops or lowest outputs for all the crops.
Though there were climate variations in this study period, there was no one year that was best or
worst for all the crops. Thus crops could respond differently to climate changes, crop specific
effects could be present and farmers’ behaviours in response to crop price changes could all
explain these differing output variations in the crops. This strengthens our choice of panel data
econometrics to be able to identify particularly crop specific effects. The farmers’ behaviours are
studied in the time series model to be able to identify the substitution effects and cross price
elsatcities of the crops.

Climate Changes
The Gambia has a semi-arid climate with two seasons, rain and dry seasons. The rain season
starts from June to October, and the dry season goes from November to May. In December, the
temperature falls as low as 13 Celsius degrees; it peaks up in the months of April and May
exceeding 40 Celsius degrees in some climatic zones. On average, the temperature stands at 28
Celsius degrees. The country is 200km from the tropical forests and 300km from the Northern
Desert. It has three climatic zones namely Sahelian, Sudanian-Sahelian and Sudanian-Guinean
zones. According to Trolldalen(1991), the rainfalls vary from zone to zone. For the Sahelian zone,
the rainfall can exceed 600 millimetres per annum; it ranges from 600 – 900 millimetres for the
Sudanian-Sahelian zone and 900 -1200 millimetres for the Sudanian-Guinean zone. Similarly, the
crop duration varies as 79 days, 71 – 119 days, and 120 – 150 days for the three climatic zones
respectively, Trolldalen (1991). As the rainfall varies, the crop durations vary and consequently the
crop yields vary. The country has been experiencing decreasing rainfalls. Trolldalen(1991) finds
that Gambia has been experiencing an average annual decrease of 15.5 % in the rainfall from
1886 to 1977. For the period, 1990 – 2007, the highest annual rainfall was recorded in 1999 at
1174.76 millimetres, and the lowest was recorded in 2002 at 597.84 millimetres. In this period, the
annual rainfall declined eleven times from its previous year and increased only six times. If this fact
can be projected into the future, then we can say there is 64.7 per cent chance that the annual
rainfall will be lower in the next year than this year with annual average decline rate of 10.1 per
Bukhari Sillah, An analysis of the impacts of rainfall and rainfall variability on cash and food crop
production in the Gambia http://www.gjqs.org/ Page 19

cent, and only 35.3 per cent chance that it will be higher with an annual average increase rate of
27.9 per cent. The high increase rate results from sudden jumps in the annual rainfalls following a
series of consecutive declines. Thus, from these rainfall patterns, we can infer that there is a higher
chance for a rainfall decline in the future than for an increase, and when it increases, it is also likely
to cause flooding.

Literature
Climate has direct linkage with the plants. The type of climatic zones also determines the types of
the plants. For plants to grow, they require essentially four major resources, water, nutrition and
temperature and light, Saugier(1996). Farmers can, to a great extent, control the nutrition resource,
and in the case of the Gambia light resource is available all the year around, while temperature
averaging 28 Celsius degrees is suitable for the crops the farmers plant in the country. Thus, it
remains there only water resource that is beyond the control of the farmers, and which eventually
determines the crop yield in the country assuming the plant disease and pests constant. In fact a
major plant pest, locus, is dependent on the water (rainfall) availability in the sub-region of West
Africa. In a climatologically modelled study, Porter and Jamiesson (1996) find that causes of
variations in crop yields are largely due to temperature and rainfall. In an historical survey of the
agriculture in the Gambia from 1948 to 1983, Trolldalen (1991) find that “the food problem and
failed agricultural projects were explained as ecological degradation” and “drought is a constant
menace.” That is, it is rainfall that in fact determines the entire health of the country, if it does not
rain, it will be aid. The country will sustain itself if it rains, and it will need aid, if it does not rain.
Trolldalen then applies a geographical model to evolution of the environment and agricultural
production in the Gambia and finds a correlation coefficient of 0.47 between cash crop and the
precipitation, a correlation coefficient of 0.47 between upland cereal crops (sorghum, millet and
maize) and the precipitation, and a correlation coefficient of 0.1 between rice and the precipitation.
Rice is grown in three areas upland, swamp and irrigated fields. The irrigated fields depend less on
the rainfall, but the upland and swamp rice depend largely on the rainfall. However, in the long run,
, all rice production depends on the rainfall as the level of rainfall also determines the extent of salt
intrusion inland through the river flows from the Atlantic coast, Trolldalen(1991), causing damages
to the reverie rice fields of swamp and irrigation. The upland rice depends entirely for water on the
rainfall. In a cross country panel data for the Sub-Sahara Africa, Borris et. al (2008) finds that
rainfall and temperature have significant impacts on agricultural production. The study does not
spell out the responses of various country specific crops to the climate change. Our current paper
attempts, among other things, to fill in this gap for the Gambia. The paper estimates the
dependency relationships between the rainfall and the various crops produced in the Gambia; it
also measures the substitution coefficients among the various crops as the farmer optimally
chooses to allocate his scarce time endowment among the planting of the various crops. Since the
rainfalls can be forecasted with great accuracy, then knowing the dependency degrees of various
crops on the rainfall will help the farmers choose an optimal combination of the agricultural
planting. This paper attempts to contribute this valuable information.

Theoretical Framework and the Econometric Model


We assume the crop production follows the Cob-Web theory, where this year’s planting will depend
on the last year’s produce price. We treat the supply of crop output and the production to be the
same, what is planted and harvested (production) is also what is supplied as output; one is to be
offered for sale, which is the cash crop and the other to be supplied for food consumption and that
is the food crop. But at the beginning of planting, the decision about the size of area to plant and
the variety of the crops to plant can be influenced largely by prices obtained in the last trading
season. Thus, we generalize this functional relationship between output and price as,

Yit  f ( Pit 1 )

Where Y= crop output measured in tons

P=price per ton

i=1, 2… 5 (groundnut, cotton, rice and coarse grains: sorghum and maize).

If the last year’s price was high, this year’s planting will be more ambitious than that of the last
years and hence this year’s output will increase. As much as last year’s price determines this
year’s output, this year’s climate condition will affect directly this year’s output. In this model, we
reduce the climate change to be the rainfall. Given the climatic zone of the Gambia, rain is a major
Bukhari Sillah, An analysis of the impacts of rainfall and rainfall variability on cash and food crop
production in the Gambia http://www.gjqs.org/ Page 20

climatic variable that determines the growth of plants including the cash and food crops, see
section 3 for the details. Total rainfall is crucial for the crops, likewise the distribution of the annul
rainfall, for example scarce rains in the beginning of the planting stunt the growth and result in poor
yields, whereas late rains into the time of harvesting spoil the harvest and result in lower output.
However, in a whole, it is the change in the annual rainfall that this paper assumes ultimately
matters to the agricultural production in the Gambia. In addition, the technology of production is
slash and burn, which implies that more output is expected when more lands are cleared. Thus,
land inputs are relevant for explaining changes in the agricultural output in the Gambia. With the
rainfall, R, and land inputs, L, our functional relationship will be res-specified as,

Yit  f ( Pit 1 , Lit , Rit )

Where R=annual rainfall measured in millimetres, and L = land in thousand hectares. We specify
this relationship as a panel data econometric model, and use the natural logarithm values of the
variables to get the following:

ln Yit     i   ln Pit 1   2 ln Lit   2 ln Rt   it (1)

The Cob-Web theory states that the planting behaviour is adjusted based on the past information
of the market price, which is adaptive expectation. The farmers do also speculate about the future
market price, which may influence how much planting to undertake this year, but futures
commodity market does not exist for the Gambian farmers, and thus the paper cannot investigate
that aspect.

This panel data model will produce the responses of the crops to their prices and the rainfall. It will
give us individual effects on each crop type. But because cash crops (groundnut and rice) often
compete with the food crops over the same labour time, then the farmer, given his endowment of
time has to choose a best combination planting between these two crops, one cash crop, which is
groundnut, and three food crops, which are rice, sorghum, and maize. We then model the
functional relationship of the Cob-Web supply function as,

ln Fit    Ai  A1 ln Pf it 1  A2 ln Pcit 1  A3 ln Lit  A4 ln Rit  vit (2)

Where F=Food crops production, Pf = food price, Pc = cash crop price, and v=disturbance term,
which is assumed to satisfy the assumptions of classical linear regression model. The equation (2)
assumes that the production of food crop depends on the past price of the food crop, the past price
of the cash crops, the land inputs, and the rainfall. Because the data include longer time series
data, 1990 – 2007, than the number of cross sections, which are five; we include the lagged
dependent variable in both equations above. This transformation also serves to provide an
opportunity to analyze both the short run and long run influences of the rainfall on the crop
production.

Results
Panel data estimation with fixed individual effects and random time effects produced interesting
results. The land inputs and rainfall influence significantly the crop production in both the short run
and the long run; whereas, the information about the past selling prices of the produce is found
irrelevant, and has a wrong sign. No improvement has been attained by res-specifying the
functional form with either the contemporaneous or forward price (approximated by the next period
price). Thus, no other form is found to perform better than our Cob-Web functional form for the
agricultural production in the Gambia. We produce the estimation results for the equation (1) in
table 1. We interpret the individual effects as the crop specific effects in terms of new variety and
technological innovations introduced to influence the production of the crops. Groundnut, rice and
maize have positive individual effects, while cotton and sorghum have negative ones. It can be
inferred from these effects that groundnut, rice and maize have been responding positively to new
variety introduction and technological innovations, while cotton and sorghum have responded
negatively. This explains to some degree why the farmers have been shunning the cultivation of
cotton. Groundnut is the most responsive crop to the new variety introduction and to the
technological innovations, whereas the cotton is the least responsive; the response of the rice is
also not encouraging. Recently, new variety and technology have been introduced for rice
cultivation, and this is reflected by the estimation in terms of the positive cross section effects for
rice. However, the new variety and technology have not been more important for the agricultural
production than the land size and the rainfall. The estimation shows that farming is still largely land
Bukhari Sillah, An analysis of the impacts of rainfall and rainfall variability on cash and food crop
production in the Gambia http://www.gjqs.org/ Page 21

intensive and rain dependent. 1% increase in the annual rainfall, holding other things constant,
leads on average to 0.5498% increase in the crop production in the short run and a 0.782%
increase in the long run. On the other hand, a 1% increase in the land size, holding other things
constant, leads on average to a 0.507% increase in the crop production in the short run and a
0.721% increase in the long run. 70.3% of the rainfall impact is realized in the short run, and
similarly 70.4% of the land size impact is realized in the short run. These two important
determinants of the agricultural output in the Gambia are also the most fast diminishing factors of
productions. Land size for the land intensive crops is shrinking due to the encroachment of
industrial projects and housings on the fertile lands. Whereas, rainfalls are determined by the
global climatic conditions, which have been worsening, and as a result the country has been
experiencing declines in the rainfalls; the decline rate is estimated to be 15.5% per annum for the
period 1886 - 1977, Trolldalen (1991). The estimation results do not show any evidence for the
substitution effects between the food crop and the cash crop production, see table 2. In fact, the
data show some evidence for the complement effects between the food and the cash crop
cultivation. Thus, farmers do not substitute the cash crop cultivation for the food crop cultivation
due to an increase in the price of the cash crop; and crop-own prices are irrelevant to the farmers’
cultivation and planting decisions.

Table 1a: Short run crop production elasticity


Dependent Variable: LY

Variable Coefficient Std. Error t-Statistic Prob.

C -1.192737 1.233982 -0.966576 0.3368


LY(-1) 0.296619 0.070122 4.230022 0.0001
LP(-1) -0.063652 0.057960 -1.098201 0.2756
LL 0.507934 0.052500 9.674879 0.0000
LR 0.549849 0.163310 3.366898 0.0012

Effects Specification

Cross-section fixed (dummy variables)

R-squared 0.979484 Mean dependent var 9.569508


Adjusted R-squared 0.977324 S.D. dependent var 1.564099
Log likelihood 7.050043 F-statistic 453.5513
Durbin-Watson stat 1.801819 Prob(F-statistic) 0.000000

Table 1b: Cross-section effects

Groundnut 0.302787
Cotton -0.264543
Rice 0.052018
Sorghum -0.113739
Maize 0.023477

Table 2a: Short run cross price elasticity coefficients

Dependent Variable: LY

Variable Coefficient Std. Error t-Statistic Prob.

C -3.012524 1.601248 -1.881360 0.0667


LY(-1) 0.475671 0.107679 4.417502 0.0001
LPF(-1) -0.444280 0.151830 -2.926169 0.0055
LPC 0.377948 0.145038 2.605859 0.0125
LL 0.331076 0.085461 3.873991 0.0004
Bukhari Sillah, An analysis of the impacts of rainfall and rainfall variability on cash and food crop
production in the Gambia http://www.gjqs.org/ Page 22

LR 0.715036 0.205030 3.487467 0.0011

Effects Specification

Cross-section fixed (dummy variables)

R-squared 0.714305 Mean dependent var 9.870244


Adjusted R-squared 0.667796 S.D. dependent var 0.392155
Log likelihood 7.827248 F-statistic 15.35857
Durbin-Watson stat 2.267214 Prob(F-statistic) 0.000000

Table 2b: Cross-Section Effects

Rice 0.272773
Sorghum -0.192992
Maize -0.079781

Conclusions and Policy Implications


The production of groundnut and coarse grains is significantly determined by the land size devoted
to them and the annual rainfalls. The groundnut, rice and maize are found to positively respond to
new variety introduction and technological innovations, whereas the cotton and sorghum have
been negatively responding. The estimation results of the model find no evidence that farmers
substitute cash planting for the food cultivation; the results have in fact shown that the two crops
are complements. There are three implications from the research findings. One, the individual
effects, which are interpreted here as crop specific effects such as new variety and technological
innovations, should be intensified for the groundnut, rice and maize as they are found to positively
respond to these specific effects. The new variety introduction and technological innovations for
cotton and sorghum should be revisited because the current specific effects are found yielding
negative results. Second, the land size and rainfall ultimately determine the agricultural output. But
land size is fixed, the fertile land size in the country is estimated at 550,000 hectares, and in 2007,
an estimate of 34.95% of it was cultivated, and the cultivated area has been increasing on average
for the period of the study at a rate of 5.4% per annum. If this rate continues for the next twenty
years from 2007, the total arable land will be all cultivated. In other words, in the next twenty years,
assuming other things constant, the total arable land will be slashed and burned for the crop
cultivation of the land intensive crops. Thereafter the production of the land intensive crops will be
stunted unless the technological innovations are added to the land. These technological
innovations are not expected to take place in the next twenty years, since the slash and burn
technology has been around for more than a century with negligible technological improvements,
and we cannot expect this old habit to die in the next twenty years. The recommendation is that the
agriculturalist scientists and the policy makers should intensify the innovations and adoption of
labour intensive crops, such as vegetables and fruit plants that require relatively small land size to
produce high output to boost the total agricultural output of the country and minimize the slash and
burn impacts on the arable lands. This recommendation is further reinforced by the fact that rainfall
is also decreasing, and production of land intensive crops cannot be done by irrigation. The final
implications from this study are that the perception that farmers substitute cash crop cultivation for
the food crop cultivation is rejected, and that farmers do not respond to changes in the crop-own
prices. That is, the produce own prices have been distorted, and cannot function as a market
signal for the farmers to change their planting decisions. These distortions should be lifted to link
accurately the farmers to the market price changes, so that resources can be better allocated
between the agricultural sector and the other sectors of the economy.

References
Balestro, P., Nervlove, M. Pooling cross-section and time series data in the estimation of a
dynamic model: The demand for natural gas, Econometrica 1996, 34, 585 – 612
rd
Baltagi, B., Econometric Analysis of panel data, 3 ed., John Wiley and Sons, Chichester 2005
nd
Matyas, L. Svestre, P. the econometrics of panel data, Handbook of theory and applications, 2
ed., Kluwer Academic Publishers, Dordrecht 1996
Barrios, Salvador; Outtara, Bazoumana, & Stroble, Eric, 2008, the impact of climate change on
agricultural production: is Africa different?, Food Policy August 2008,33 (4), 287-298
Bukhari Sillah, An analysis of the impacts of rainfall and rainfall variability on cash and food crop
production in the Gambia http://www.gjqs.org/ Page 23

Trolldalen, John Martin on the fringe: a system approach to the evolution of the environment and
agricultural production in the Gambia, West Africa 1948 – 1983, NORAGRIC Occasional
paper series C, Agricultural University of Norway, Development and Environment No. 10,
1991
Curran, Sara R. et al. ed. the Global Governance of Food, Routledge Publication 2009
Peter, D.; Maracchi, G. & Ghazi, A. , ed., Course on Climate Change Impact on Agriculture and
Forestry, proceedings of the European School of Climatology and Natural Hazards Course
held in Volterra, Italy, March 16 – 23 1996
Ferguson, C.E. learning, expectations, and cobweb models, Journal of Economics 1960, 20, no.
3- 4, 297 – 315
Branch, William A. local convergence properties of cobweb model with rationally heterogeneous
expectations, Journal of Economic Dynamics and Control 2002, 27, Iss. 1, 63 – 85
Mitra, Sophie & Boussard, Jean-Mar a non-linear cobweb model of agricultural commodity price
fluctuations, Department of Economics, Fordham University, NY, U.S.A, discussion paper
2008- 11
Kaldor, Nicholas a classificatory note on the Determinateness of Equilibrium, the Review of
Economic Studies 1934, 1, No. 2, 122 – 136
Marc, Nerlove adaptive expectations and cobweb phenomena, Quarterly Journal of Economics
1958, 73, 227 - 240

View publication stats

Vous aimerez peut-être aussi