Content uploaded by Jeffery D. Connor
Author content
All content in this area was uploaded by Jeffery D. Connor on May 30, 2022
Content may be subject to copyright.
Analysis
Mitigating economic risk from climate variability in rain-fed agriculture through
enterprise mix diversification
John M. Kandulu ⁎, Brett A. Bryan, Darran King, Jeffery D. Connor
CSIRO Ecosystem Sciences, Waite Campus, Waite Rd., Urrbrae, South Australia, 5064, Australia
abstractarticle info
Article history:
Received 23 January 2012
Received in revised form 27 March 2012
Accepted 30 April 2012
Available online 21 May 2012
Keywords:
Climate variability
Adaptation
Yield uncertainty
Economic net returns
Agricultural enterprise
Finance
Monte Carlo
Climate variability, and its increase with climate change, pose substantial economic risks to agriculturalists and
hence, limittheir ability to respond to globalchallenges such as food security.Enterprise mix diversification is the
most common, and is widely regarded as the most effective, strategy for mitigating multiple sources of short-
term economic risk to agricultural enterprises. However, assessments of enterprise mix diversification as a strat-
egy for mitigating climate risks to ensure long term viability of agricultural enterprises are sparse. Using the
Lower Murray region in southern Australia as a case study, we combined APSIM modelling with Monte Carlo
simulation, probability theory, and finance techniques, to assess the extent to which enterprise mix diversifica-
tion can mitigate climate-induced variability in long term net returns from rain-fed agriculture. We found that
diversification can reduce the standard deviation by up to A$200 ha
−1
, or 52% of mean net returns; increase
the probability of breakingeven by up to 20%, and increase the mean of 10% of worst probable annual netreturns
(Conditional Value at Risk) by up to A$100 ha
−1
. We conclude that enterprise mix diversification can also be an
effective strategy for hedging against climate-induced economic risk for agriculturalists in marginal areas.
© 2012 Elsevier B.V. All rights reserved.
1. Introduction
Many of the world's major agricultural regions are characterised
by uncertain and variable climatic conditions including temperature
and rainfall (Azam-Ali, 2007; Furuya and Kobayashi, 2009; Naylor et
al., 2007; Power et al., 1999; E. Wang et al., 2009). Notwithstanding
variance in market input costs and commodity prices (Hazell et al.,
1990; Ramaswami et al., 2003), climate variability is the principal
source of risk affecting long term economic viability of rain-fed agri-
cultural systems both in industrial (Iglesias and Quiroga, 2007;
Lotze-Campen and Schellnhuber, 2009; Marton et al., 2007) and
smallholder farming systems (Kurukulasuriya and Ajwad, 2007;
Magombeyi and Taigbenu, 2008). Climate models predict an increase
in future climate variability and a significant increase in the frequency
of below-average rainfalls and above-average temperatures in many
agricultural regions (IPCC, 2007; Naylor et al., 2007; Suppiah et al.,
2006). All else being equal, this is likely to increase the uncertainty
and variability in agricultural yields and net returns, and increase
the frequency with which these are below average (John et al.,
2005; E.L. Wang et al., 2009). Consequently, the viability of agricultur-
al enterprises will become increasingly threatened in the long run.
To manage the severity of the impact of climate variability on net
returns, agriculturalists routinely adopt mitigation strategies involving
various adjustments in enterprise mix, and production technologies
and techniques (Bryant et al., 2000; Kelkar et al., 2008; van Ittersum et
al., 2003). The diversification of agricultural enterprise mixes consisting
of several different crops and livestock (hereafter, diversification), is
widely regarded as the most common and effective strategy for mitigat-
ing climate-induced variability in net returns from rain-fed agriculture
(Amita, 2006; Azam-Ali, 2007; Correal et al., 2006). Diversification can
also reduce the magnitude and frequency of below-average net returns
under climate uncertainty (Berhanu et al., 2007).
The benefits of diversification are premised on the utilization of
imperfectly correlated net returns from multiple agricultural enter-
prises. Most of the benefit of diversification comes from hedging
against market input and commodity price fluctuations (Bhende
and Venkataram, 1994; Ramaswami et al., 2003; World-Bank,
2004). However, here we propose that diversification may also be
beneficial for hedging against climatic variability. When the impacts
of climatic variability differ between multiple agricultural enterprises,
losses from investments in some enterprises are offset by gains, or
moderated by less severe losses, in other enterprises thereby reduc-
ing the impact on overall net returns (Fraser, 2007; Fraser et al.,
2005). Conversely, the benefits of diversification typically come at a
cost of reduced expected short-term net returns (Chan et al., 1998;
Markowitz, 1952a, 1952b, 1994). This is because diversification in-
volves investing in multiple enterprises to mitigate long term uncer-
tainty and variability even when investments in alternative non-
diversified enterprises may offer higher expected net returns in the
short term (Cooper et al., 2008). As such, it is necessary to quantify
the tradeoffs between the benefits and costs of diversification when
assessing the benefits of agricultural diversification. Further, the
Ecological Economics 79 (2012) 105–112
⁎Corresponding author. Tel.: + 61 8 8303 8678; fax: +61 8 8303 8582.
E-mail address: john.kandulu@csiro.au (J.M. Kandulu).
0921-8009/$ –see front matter © 2012 Elsevier B.V. All rights reserved.
doi:10.1016/j.ecolecon.2012.04.025
Contents lists available at SciVerse ScienceDirect
Ecological Economics
journal homepage: www.elsevier.com/locate/ecolecon
nature and strength of correlated yields across alternative agricultural
enterprises need to be fully understood and quantified when
assessing the benefits of agricultural diversification. There is a general
consensus from the finance literature that not considering the nature
and strength of correlated yields may under- or over-estimate the
benefit of diversification (Bangun et al., 2006; Chan et al., 1998,
1999; Markowitz, 1952a, 1952b, 1994).
Several studies have assessed various factors affecting the poten-
tial for enterprise mix diversification to mitigate multiple sources of
risk to agricultural enterprises with the objective of aiding agricultur-
alists' short term decision making (Barkley et al., 2008; Cooper et al.,
2008; Hardaker et al., 2004; Kingwell, 1994; Pannell et al., 2000).
Major sources of short term risk and uncertainty assessed in these
studies broadly consist of price and production risks including
changes in commodity and input prices, and variability in yields due
to weather, pests and diseases. To achieve this, these studies applied
various methods including crop modelling (Cooper et al., 2008), sto-
chastic risk modelling (Kingwell, 1994), utility and probability theory
(Kingwell, 1994; Ladanyi, 2008; Pannell et al., 2000), and portfolio
theory (Barkley et al., 2008). Luo et al. (2005) assessed the impact
of climate change in increasing yield risk. However, few other studies
have considered long term climatic sources of uncertainty and risk
(Lien et al., 2009). Lien et al. (2009) speculate that this is because rel-
evant historical data necessary for long term analyses are usually
sparse and that most studies rely on a few observations of net returns.
However, long-term risk and uncertainty analyses are important for
informing risk management strategies that ensure long-run economic
viability of agricultural enterprises (Lien et al., 2009; Meza and Silva,
2009). In the context of increasingly frequent droughts in many of the
worlds agricultural regions (Furuya and Kobayashi, 2009; Howden et
al., 2007; IPCC, 2007; Lotze-Campen and Schellnhuber, 2009) and
growing threats to global food security (Fraser, 2007; Fraser et al.,
2005; Yang, 2009), the effectiveness of diversification at mitigating
the risk of crop failure bears significant relevance. Further, emerging
markets for ecosystem services (Yang et al., 2010) provide alternative
enterprises with returns that may be even less correlated with agri-
cultural returns, thus broadening the scope for diversification as an
effective strategy for mitigating the risk of low incomes.
In this study, we assessed the ability of enterprise mix diversifica-
tion to mitigate climate-induced variability in long-term economic
net returns from rain-fed agriculture. We leave the application and
operationalization of diversification to future work. Variability was
assessed based on historical data. Using a case study in the
11.8 million hectare Lower Murray region in southern Australia, we
fitted probability density functions to modelled long term crop and
livestock yield data. We considered four alternative agricultural en-
terprise types consisting of three non-diversified enterprises and
one diversified enterprise comprised of a mix of rain-fed agricultural
enterprises. We used Monte Carlo simulation to quantify the variabil-
ity in yields and, via a profit function, net returns. We quantified the
benefits and costs of enterprise mix diversification using techniques
from finance theory including the probability of break-even and Con-
ditional Value at Risk (CVaR). We quantified the trade-off between
the reduced variability in returns, measured using the value of stan-
dard deviation, and reduced expected net returns, and assessed the
spatial heterogeneity in these effects across the region. We discuss
the implications of diversification as an adaptation strategy for agri-
culturalists to cope with increasing climatic variability.
2. Methods
2.1. Study Area
The Lower Murray region (Fig. 1) in southern Australia covers a total
area of 11,871,363 ha. Mean annual rainfall ranges from 200 mm yr
−1
in the drier northern areas of the SAMDB to 1400 mm yr
−1
in the
southern Wimmera. Rain-fed agriculture is the dominant land use cov-
ering over 50% of the region and is an important component of the
regional economy (Bryan et al., 2011). The average size of agricultural
land used for rain-fed agriculture in the study area is around 1000 ha.
Agricultural systems vary greatly across the region depending on
climate and soil types. The cropping of cereals (wheat, barley), pulses
(lupins, beans, peas), and sheep (pasture, forage shrubs) grazing are
typical agricultural enterprises.
2.2. Modelled Agricultural Systems
We modelled and compared yield and economic outcomes for
three non-diversified agricultural systems and one diversified agri-
cultural system in the study area. The three non-diversified agricul-
tural systems were defined as continuous single-crop agricultural
systems of wheat, lupins, and sheep grazing on modified pastures
(hereafter, sheep). The diversified agricultural system was defined as
a mixed enterprise comprising continuous cropping (and grazing) of
wheat, lupins, and sheep in equal proportions of available agricultural
land in any one year production horizon. In considering yields under
continuous cropping and grazing agricultural systems only, we con-
trolled for effects of land management on yields by assuming land
management to be constant across soil/climatic zones thereby ensur-
ing that variability in yields can be largely attributed to variability in
climate.
2.3. Crop Yield Modelling
We used the Agricultural Production Simulator (APSIM, Keating et
al., 2003) to predict crop yields. APSIM has been widely used and val-
idated for Australia (Luo et al., 2005, 2007; E. Wang et al., 2009; E.L.
Wang et al., 2009). Annual yields were modelled for wheat, lupins,
and sheep for 138 unique soil/climate zones over 116 years. Soil/cli-
mate zones were mapped by overlaying a layer defining 15 soil
types and a layer defining 16 climate zones (Bryan et al., 2007). Soil
types were defined based on state government soil mapping. Climate
zones were calculated using k-means classification of mean annual
rainfall, mean annual temperature, and annual moisture index layers
modelled using ANUCLIM. Soil/climate zones (hereafter, zones) were
assumed to have homogeneous production potential for the purposes
of this study. APSIM soil profile parameters were specified for each of
the 15 soil types using field-derived soil survey data. Historical daily
climate records were acquired for the 116-year period from 1889 to
2005 from the SILO data base. Typical land management regimes
(sowing windows, fertiliser application rates) were defined for the
study area based on expert opinion. For full details and other applica-
tions of this modelling we refer readers to Bryan et al. (2007, 2010).
Of the 138 zones modelled across the entire region, we selected
nine to illustrate the results of our analysis (Fig. 1). The nine illustra-
tive zones reflect the 10th, 20th, 30th, 40th, 50th, 60th, 70th, 80th,
and 90th yield percentiles.
2.4. Quantifying Climate-Induced Yield Variability
To assess the benefits of diversification, we treated annual net
returns as stochastic. This is premised on the assumption that climate,
the key driver of yield variability and the focus of our study, is gener-
ally assumed to be stochastic (Furuya and Kobayashi, 2009; Hardaker
and Lien, 2010; Iglesias and Quiroga, 2007). Probability theory pro-
vides a suitable framework for the quantification of climate-driven
uncertainty and variability in net returns over a given time horizon
(Hardaker and Lien, 2010; Lien et al., 2009).
In each of the 138 zones, we generated frequency distributions for
yields Q1
i
, for each of the three enterprises i, where Q1 is the yield of a
primary product, and iis an element of I{wheat, lupins, sheep}. For
sheep, we generated frequency distributions for yields of both a
106 J.M. Kandulu et al. / Ecological Economics 79 (2012) 105–112
primary product (meat), Q1, and a secondary product (wool), Q2.We
then fitted probability density functions to the frequency distribu-
tions to characterise climate-induced variability in yield outputs
using @RISK software. We fitted probability density functions of var-
ious forms including normal, gamma, exponential, triangular and beta
distributions to the yield distributions using the chi-squared statis-
tics, χ
2
, to measure the goodness of fit of each distribution (Iglesias
and Quiroga, 2007) as in Eq. (1).
χ2¼X
k
j¼1
Nj−Ej
Ej
2
ð1Þ
Where kis the number of discrete intervals in a histogram derived
from 117 years of simulated yield time-series data; N
j
is the frequency
of observations in each interval; and E
j
is the expected (theoretical)
frequency captured by the probability density function. The probabil-
ity density function with the best fit as measured by the chi-squared
statistic was selected for use in Monte Carlo simulation of net eco-
nomic returns.
2.5. Quantifying Variability in Economic Net Returns
To fully account for the effect of climate variability on economic
net returns from rain-fed agriculture in the study area, we quantified
variability in long term average annual profit per hectare (Bryan et al.,
2009, 2011; Deressa and Hassan, 2009; Kurukulasuriya and Ajwad,
2007). After Benhin (2008), we controlled all other economic factors
including costs of production and commodity prices. We defined eco-
nomic net returns as revenues from sale of commodities produced
less the fixed and variable cost incurred in the production of agricul-
tural commodities. Following Bryan et al. (2009, 2011), we used a
profit function to calculate net economic returns for wheat, lupins
and sheep such that:
NRi¼P1iQ1iTRNi
ðÞ
þP2iQ2iQ1i
ðÞ−QCiQ1i
ðÞþACiþFDCiþFOCiþFLCi
ðÞðÞ
ð2Þ
where P1
i
($/tonne or $/DSE) is the farm gate price of primary prod-
ucts; P2
i
($/kg) is the farm gate price of secondary products (wool in
the case of sheep); Q1
i
(tonnes/ha or DSE/ha —Dry Sheep Equivalent
Fig. 1. Location and land use in the Lower Murray study area.
107J.M. Kandulu et al. / Ecological Economics 79 (2012) 105–112
after McLaren (1997) is used to measure sheep production to stan-
dardise for the energy requirements of the animals where 1 head of
sheep=1.5 DSE.) Q1
i
is a measure of yield of a primary product; Q2
i
(kg of wool) is a measure of yield of a secondary product (only rele-
vant for sheep); TRN
i
is the proportion sheep sold; QC
i
($/tonne or
$/DSE) is variable costs for example harvest, and storage; AC
i
($/ha)
is variable cost that varies with farming area for example cost of
seeding, fertiliser and pesticide treatment; FOC
i
($/ha) is fixed operat-
ing costs for example, energy, maintenance, and administrative over-
heads; FDC
i
($/ha) is fixed depreciation costs on farm machinery and
infrastructure, and; FLC
i
($/ha) is fixed costs of labour for example,
farmer wages.
Average annual per hectare net returns to the diversified agricul-
tural enterprise system, NR
d
. were calculated as:
NRd¼∑NRi
ðÞ
3∈iwheat;lupins;sheep
fg
:ð3Þ
Table 1 outlines notation descriptions and values used in Eq. (2).
The profit function integrates a variety of production, price, and cost
data in Table 1 to calculate agricultural profit at full equity (PFE) in
$/ha (Bryan et al., 2011). The profit function has been found to pro-
vide a reasonable estimate of economic returns to agriculture in the
study area (Bryan et al., 2011).
The benefits of diversification in relation to climatic variability rely
on imperfect correlation between yields of crops and grazing systems
(Correal et al., 2006; Iglesias and Quiroga, 2007). Hence, it is impor-
tant to quantify yield correlations and include these in simulation of
net returns. We calculated pair-wise Pearson correlation coefficients
for yields ρ
i,i
between wheat and lupins, wheat and sheep, and lupins
and sheep from the modelled yield data from coincident years.
To quantify climate-induced variability in net returns for each en-
terprise, NR
i
, we generated 1000 Monte Carlo simulations of net
returns using Eq. (2) with random samples for the yield parameter
Q1
i
, drawn from the modelled probability density functions for yields.
To quantify climate-induced variability in net returns for the diversi-
fied agricultural enterprise system, NR
d
, we generated 1000 Monte
Carlo simulations (Hardaker and Lien, 2010) of net returns using
Eq. (2) with random samples for the yield parameter Q1
i
drawn
from the modelled probability density functions for yields using
@RISK, and considering yield correlations ρ
i,i
. Frequency distributions
were then developed for the average of net returns under the three
enterprises (see Eq. (3)). In the same way as for yield, we fitted prob-
ability density functions to them and selected the best using goodness
of fit and Chi-square test (see Eq. (1)).
2.6. Quantifying Benefits from Diversification
To assess the benefits of diversification, we considered agricultur-
alists in the study area as investors faced with the challenge of choos-
ing among four alternative agricultural enterprises with uncertain net
returns. The financial risk management literature offers various mea-
sures for assessing trade-offs between expected net returns and over-
all variability in net returns. Specifically, the concept of Conditional
Value at Risk or CVaR (Uryasev and Rockafellar, 2001) has been
used to assess variability of net returns and probabilities of low-end
net returns from alternative investments. One way to apply CVaR is
to calculate the average expected return of the lowest 10% of possible
outcomes, CVaR
0.1
(Rockafellar and Uryasev, 2002).
We used four indicators including mean net returns, standard de-
viation of net returns, the probability of break even, P(NR≥0), and
CVaR
0.1
to quantify the expected net returns and variability of net
returns from each of the four alternative investment options.
3. Results
We present quantitative results on climate-induced yield variability,
Q1
I
, yield correlations, ρ
i,i
, variability in net economic returns, NR
i,d
,and
assess impacts of switching to diversification. First, we summarise
results on yield variability, and correlations broadly across the region,
but we refer the reader to the online supporting material for more
detailed results for the nine illustrative zones. Second, we summarise
results on variability in net economic returns, and impacts of diversifi-
cation broadly across the region. Third, we summarise results on vari-
ability in net economic returns, and assess impacts of switching to
diversification in further detail based on results from the nine illustra-
tive zones.
3.1. Climate-Induced Yield, Variability, and Correlation
Wheat yields ranged from 0.00 tonnes ha
−1
in the drier northern
parts of the region to 6.20 tonnes ha
−1
in the wetter southern parts,
with a mean of 1.60 tonnes ha
−1
. Standard deviations for wheat
yields ranged from 28% of mean yields, mostly in the wetter southern
parts, to 70% of mean yields, mostly in the drier northern parts. Yields
for lupins ranged from 0.00 tonnes ha
−1
in the drier northern parts of
the region to 3.11 tonnes ha
−1
in the wetter southern parts with a
mean of 1.11 tonnes ha
−1
. Standard deviations for lupin yields
ranged from 21% of mean yields, mostly in the wetter southern
parts, to 100% of mean yields, mostly in the drier northern parts. Un-
like wheat and lupins, yields for sheep did not necessarily follow spa-
tial distributions of rainfall across the region. Yields for sheep ranged
from 0.00 DSE
1
ha
−1
to 15.00 DSE ha
−1
, with mean of 3.40 DSE ha
−1
.
Standard deviation values for sheep yields ranged from 89% to 274%
of mean yields.
Correlation coefficients between yields for wheat and lupins
ranged from 0.40 to 0.95, with strong correlations occurring mostly
in the drier northern parts. Correlation coefficients between wheat
and sheep yields ranged from −0.05 to 0.95 with weak correlation
coefficients estimated mostly in the wetter southern parts. Correla-
tion coefficients between lupins and sheep ranged from −0.13 to
0.90 with weak correlation coefficients also estimated mostly in the
wetter southern parts.
3.2. Variability in Economic Net Returns
Fig. 2 maps the four economic indicators including the four agri-
cultural enterprise systems, wheat, lupins, sheep, and a diversified ag-
ricultural enterprise system. The four economic indicators mean,
Table 1
Notation descriptions and values for NR
I
calculations (See Eq. (2)).
Notation Definition Value
Wheat Lupins Sheep
P1 Price of primary commodity farmed
(A$tonne
−1
or A$DSE
−1
)
257 211 22
Q1 Quantity of the primary product
(t ha
−1
, DSE ha
−1
)
TRN Turn-off rate (number of sheep sold as
portion of total herd, =1 for cropping)
1 1 0.31
P2 Price of secondary commodities
(A$kg
−1
of wool, only applies to sheep)
0 0 4.0
Q2 Quantity of secondary commodity
(kg of wool ha
−1
)
0 0 2.73
QC Quantity costs (A$ tonne
−1
or A$DSE
−1
) 0 0 4.0
AC Area costs (A$ ha
−1
) 149 96 3
FDC Fixed depreciation costs (A$ ha
−1
)19132
FOC Fixed operating costs (A$ ha
−1
)48314
FLC Fixed labour costs (A$ ha
−1
)35233
1
Note that sheep production is measured in Dry Sheep Equivalent (DSE) terms after
McLaren (1997), where 1 head of sheep = 1.5 DSE.
108 J.M. Kandulu et al. / Ecological Economics 79 (2012) 105–112
standard deviation as a proportion of the mean NR, the probability of
break even P(NR≥0), and CVaR
0.1
vary across the landscape, and
across the four agricultural systems (Fig. 2).
Overall, there is significant spatial variation in mean NR ranging
from −A$121 ha
−1
to A$672 ha
−1
. Sheep has the lowest expected
NR of all enterprises, followed by lupins, and wheat has the highest
mean net returns. Lowest mean NR are estimated for wheat and lu-
pins particularly in the drier northern parts of the study area. Median
NR expected values occur in the central parts of the study area. The
highest mean NR occur in the wetter southern parts (Fig. 2).
We found a strong association between the magnitude and vari-
ance in expected net returns. Overall, the highest standard deviations
in expected net returns occur in the more productive, higher rainfall
areas in the south, whilst lower standard deviation values occurred
in the less productive, lower rainfall areas in the north (Fig. 2). Lowest
standard deviations were calculated for sheep and the diversified ag-
ricultural enterprise (A$0–A$110 ha
−1
) followed by lupins, with the
highest standard deviation is estimated for wheat (A$110 ha
−1
to A
$320 ha
−1
), particularly in the south and southeast moderate to
very high rainfall region.
The probability of breaking even P(NR ≥0), was relatively low,
below 20%, in locations with low expected NR typically in the drier
northern areas, and high, 60% or higher, in locations with high
mean expected NR mostly in the south. Overall, across the region, P
(NR≥0) was the highest for sheep, followed by the diversified agri-
cultural system, and the lowest for wheat and lupins, particularly in
the drier northern areas.
Values for CVaR
0.1
were associated with mean expected NR.CVaR
0.1
values, up to −A$236 ha
−1
in some specific cases, were calculated for
wheat and lupins across most of the study area. Positive CVaR
0.1
values were calculated for sheep and the diversified agricultural sys-
tem for most of the region. Higher CVaR
0.1
values, up to A$441 ha
−1
,
were calculated for the diversified agricultural system particularly in
the more productive southern areas.
3.3. Benefits of Diversification
Fig. 3 shows frequency histograms for net returns, and values for
the four economic indicators for wheat, lupins and sheep, and the di-
versified agricultural system in the nine illustrative locations selected
across the region.
To assess benefits from diversification, we consider a decision to
switch from a single enterprise with the highest expected NR to the
diversified agricultural system. The highest returning non-diversified
Fig. 2. Measures of potential net economic returns ($/ha) under alternative non-diversified and diversified enterprise farm systems across the study area.
109J.M. Kandulu et al. / Ecological Economics 79 (2012) 105–112
agricultural enterprise system in zones 23, 1, 46 and 70 is sheep. The
highest returning non-diversified agricultural system in zones 88, 96,
123, 138, and 134 is wheat (Fig. 3).
Fig. 3 shows that there is limited scope for beneficial diversifica-
tion in low to moderate rainfall locations such as zones 23, 1, 46
and 70. Switching from sheep grazing to the diversified agricultural
system in these locations would reduce expected NR. This decision
will reduce expected NR from A$47 ha
−1
to −A$64 ha
−1
in zone
23, A$25 ha
−1
to −A$70 ha
−1
in zone 1, A$26 ha
−1
to −A
$33 ha
−1
in zone 46, and A$44 ha
−1
to A$19 ha
−1
in zone 70. Fur-
ther, switching from sheep grazing to the diversified agricultural sys-
tem in these locations would increase standard deviations of NR by
14–57%, reduce break even probabilities, P(NR ≥0) by 40–59%, and
decrease CVaR
0.1
by about A$126 ha
−1
. Thus the overall effect of
such a decision would be an increase in the frequency and likelihood
of below average NR.
In Fig. 3, the benefit from diversification is most appreciable if
switching from wheat to the diversified agricultural enterprise sys-
tem in zones 88, 96, and 123 in moderate to high rainfall areas. Whilst
wheat has the highest expected NR in these locations at A$77 ha
−1
,A
$204 ha
−1
and A$181 ha
−1
respectively, wheat also has the most var-
iable NR with standard deviations of 300%, 99%, and 170% of mean, re-
spectively. The decision to switch to the diversified agricultural
system results in lower NR than wheat at A$56 ha
−1
, A$109 ha
−1
and A$122 ha
−1
, respectively. However, the variability in NR would
also be lower with standard deviations of 273%, 94%, and 151% of
the mean, respectively. In switching to a diversified agricultural sys-
tem in zones 88, 96, and 123, expected NR would be reduced by
27%, 47%, and 33% respectively, but the standard deviations, as a per-
centage of mean of NR would also be reduced by 34%, 52%, and 40%,
respectively. The probability of break even is estimated to increase
by 3 and 4% in zones 96 and 123 respectively. Further, the value of
CVaR
0.1
, the mean of 10% of the worst probable annual NR, is estimat-
ed to increase by 46% in zone 88, 61% in zone 96, and 50% in zone 123.
Fig. 3 also shows that there is limited scope for beneficial diversifica-
tion in moderate to very high rainfall locations particularly in zones 138
and 134. Wheat in zone138, and wheat and lupins in 134, are estimated
to have higher mean NR, lower standard deviations, higher P(NR ≥0),
and lower CVaR
0.1
values than the diversified enterprise. Switching to
adiversified enterprise in these locations would reduce mean NR by
up to 70% in zone 138, largely because low-price enterprises including
sheep, and to a lesser extent lupins, would replace high-price wheat
(Table 1). In effect, diversification in these zones would result in a 70%
decrease in mean NR,and50–75% decrease in CVaR
0.1
.
4. Discussion
4.1. Economic Impacts of Diversification
We have demonstrated the ability of diversification to mitigate the
impacts of climate-driven variability in net returns from rain-fed ag-
riculture in the Lower Murray region in southern Australia. In dryer
areas with more variable rainfall, diversification was not beneficial
because it introduced more water-intensive crops —wheat and lu-
pins, which did not do well. In wetter areas, diversification intro-
duced low-price crops —sheep and lupins, into areas that were
highly suitable for an alternative high-price enterprise —wheat. Rath-
er, the greatest benefit of diversification was found in marginal agri-
cultural areas where there was a trade-off between the benefitof
reduced variability and the cost of reduced expected net returns.
Fig. 3. Comparison of measures of potential net economic returns ($/ha) under the most profitable non-diversified alternat ive (white) and the diversified (green) agricultural sys-
tem for a selection of 9 locations across the study area.
110 J.M. Kandulu et al. / Ecological Economics 79 (2012) 105–112
Correlation analysis showed that yields between any two farm en-
terprises were not perfectly correlated (i.e. are less than 1) indicating
that changes in climate variables cause non-proportional impacts on
yield production (Ludwig et al., 2009). We can deduce, therefore, that
there are benefits from diversification in the region (Markowitz,
1952a, 1952b, 1994). There is however, less potential for beneficial
diversification in relatively dry regions. In these regions, diversification
would introduce water-intensive and rainfall-sensitive cropping
enterprises, particularly wheat and lupins, in unsuitable locations with
low, and highly variable rainfall. In these areas, sheep, the least water-
intensive enterprise, is superior to diversification as it is estimated to
have higher mean and less variable net returns. Diversification is most
beneficial in moderate to high rainfall regions. In these locations, the
diversified enterprise benefits from a combination of risk-reducing
characteristics of sheep, and high-return, and moderate-variability char-
acteristics of lupins. Together these characteristics hedge against extreme
losses in years with unfavourable climate and reduce the likelihood and
magnitude of extremely low net returns while benefiting from high-
return (yet high-variability) properties of wheat (Kurukulasuriya and
Ajwad, 2007; Magombeyi and Taigbenu, 2008). Wheat is a superior enter-
prise to diversification in high rainfall areas where mean wheat yields are
high and though highly variable, extremely low yields are highly unlikely
to occur. Further, wheat has the highest net returns per tonne produced
compared to lupins and sheep.
4.2. Consistency of Findings
Findings from our studies (Fig. 2) are self-validating as they mirror
the land use practices observed in the study area to the extent that
pasture grazing sheep is the main enterprise in the drier northern
areas, and cropping enterprises lupins and wheat can be largely ob-
served in the central moderate rainfall regions and high rainfall
areas in the south (Fig. 1).
Our study findings are also consistent with previous studies that
have showed that there are benefits from diversification where invest-
ments in multiple agricultural enterprises respond differently to vari-
ability in climate (Iglesias and Quiroga, 2007). Yields are imperfectly
correlated as different enterprises respond differently to variability in
climate in the study area. In addition, correlations are varied spatially
across zones. This is consistent with the finding that diversification
can mitigate risk from climatic variability, but the benefits vary spatially
depending on, among other things, the nature and strength of correlat-
ed yields. Our findings are also consistent with the expectation that the
benefit of reduced variability from diversification comes at a cost of re-
duced expected net returns when alternative non-diversified enter-
prises offer higher expected net returns (Bangun et al., 2006; Chan et
al., 1998; Markowitz, 1952a, 1952b, 1994).
Our findings are consistent with previous studies that highlight
the need to quantify uncertainty characteristics of net returns from
diversification options for specific enterprises and spatial contexts
to determine whether or not and to what extent diversification is
beneficial (Howden et al., 2007; Thomas, 2008; Thomas et al.,
2007). This exercise is useful for working out targeted diversification
strategies for each location and the transfer of this information
through farm extension services. Our results provide evidence for
the benefits of diversification under climate variability in marginal
agricultural lands. Judicious use of diversification strategies can help
agriculturalists adapt to increasing climatic variability associated
with climatic change, and help agriculturalists respond to the global
challenge of increasing food demand.
4.3. Limitations and Future Directions
There are a number of limitations in this study. Only equal propor-
tions of combinations of three investments with equal allocations are
considered in the diversified investment option. Future studies could
systematically determine optimal diversification strategies taking
into account risk profiles of various agriculturalists, a broader range
of agricultural enterprises (Hardaker, 2000; Hardaker et al., 2004;
Ranjan and Shogren, 2009), and short term variability in input and
commodity price (Cooper et al., 2008; Lien et al., 2009). These
methods require a level of detail that is beyond what is necessary
for the comparative purposes of this study.
Our profit function does not account for long term average capital
investment costs and the effects of (dis)economies of scale and scope.
Considering the nature of the three enterprises under consideration,
long run additional capital investment costs of switching between pas-
ture grazing sheep, wheat and lupin can reasonably be assumed to be
insignificant as most capital investments are either already made, or
are highly transferrable across the three enterprises. This is generally
true as most dryland agriculturalists in the study area, and in other re-
gions, typically practice crop and animal rotation agriculture systems
for other objectives including nitrogen fixation by legumes,disease con-
trol, weed control, soil structure and soil organic matter content (Bryan
et al., 2009; E. Wang et al., 2009). Accounting for long term capital in-
vestment cost may significantly change relative net returns between di-
versified and non-diversified enterprises in other contexts.
Using historic climate data, we have demonstrated that farm en-
terprise mix diversification can mitigate financial risk from climate
variability. Transferring the results of this study in understanding
the effectiveness of diversification as an adaptive response to future
climate will depend upon the magnitude and direction of changes
in climatic variables (e.g. rainfall and temperature), and changes in
variability. If we consider an overall warming and drying trend with
increasing variability then we can make some general comments
about the effectiveness of diversification. For example, for the drier
parts of the study area the negative effects of adding water-
intensive crops to pasture systems as a diversification strategy may
be intensified. Conversely, for the higher rainfall areas, the benefits
of diversification of current continuous cropping systems with live-
stock and legumes may be greater. Understanding the complex
spatio-temporal effects of climate change on financial risk (i.e. how
it influences the shape and relative position of the histograms in
Fig. 3) requires additional modelling.
5. Conclusion
Agriculturalists already practice diversification for various traditional
reasons including but not limited to management of short-term risk
due to variance in market input costs and commodity prices, and disease
and weed control, but diversification is not widely used as a strategy for
mitigating long-term climate risks. Current levels of diversification to
meet other objectives would inevitably yield incidental benefits including
mitigating long-term climate risks however, we propose strategic diversi-
fication for specifically mitigating long-term climate risks.
We found that the greatest benefit of diversification for mitigating
climate-induced variability occurs in the marginal cropping/grazing,
moderate-rainfall parts of the study area. Diversified agricultural sys-
tems in these areas offer agriculturalists a strategy for hedging against
climatic risk in economic returns. There is limited scope for beneficial
diversification in high- and low-rainfall areas.
In the context of increasing climate variability and frequency of
droughts in many of the world's agricultural regions, growing threats
to global food security due to climate change, and emerging markets
for agricultural ecosystem services, diversification may grow in signif-
icance and relevance as a strategy for avoiding high cost of crop fail-
ure and managing long term agricultural enterprise income risk.
Appendix A. Supplementary Data
Supplementary data to this article can be found online at http://
dx.doi.org/10.1016/j.ecolecon.2012.04.025.
111J.M. Kandulu et al. / Ecological Economics 79 (2012) 105–112
References
Amita, S., 2006. Changing interface between agriculture and livestocks: a study of live-
lihood options under dry land systems in Gujarat. Working Paper. Gujarat Institute
of Development Research, Ahmedabad. 25 pp.
Azam-Ali, S., 2007. Agricultural diversification: the potential for underutilised crops in
Africa's changing climates. Rivista Di Biologia-Biology Forum 100, 27–37.
Bangun, P., Nasution, A.H., Mawengkang, H., 2006. Evaluating covariance and selecting
the risk model for asset allocation. IMT-GT Regional Conference on Mathematics,
Statistics and Applications. Universiti Sains Malaysia, Penang.
Barkley, A., Peterson, H.H., Shroyer, J., 2008. Wheat variety selection to maximize
returns and minimize risk: an application of portfolio theory. Journal of Agricultur-
al and Applied Economics 42, 39–55.
Benhin, J.K.A., 2008. South African crop farming and climate change: an economic as-
sessment of impacts. Global Environmental Change: Human and Policy Dimen-
sions 18, 666–678.
Berhanu, W., Colman, D., Fayissa, B., 2007. Diversification and livelihood sustainability
in a semi-arid environment: a case study from Southern Ethiopia. Journal of Devel-
opment Studies 43, 871–889.
Bhende, M.J., Venkataram, J.V., 1994. Impact of diversification on household income
and risk —a whole-farm modeling approach. Agricultural Systems 44, 301–312.
Bryan, B.A., Crossman, N.D., McNeill, J., Wang, E., Barrett, G., M.M., F., Morrison, J.B.,
Pettit, C., Freudenberger, D., O'Leary, G., Fawcett, J., Meyer, W., 2007. Lower Murray
Landscape Futures Dryland Component: Volume 3 —Preliminary Analysis and
Modelling. CSIRO Water for a Healthy Country.
Bryan, B.A., Hajkowicz, S., Marvanek, S., Young, M.D., 2009. Mapping economic returns
to agriculture for informing environmental policy in the Murray–Darling Basin,
Australia. Environmental Modeling and Assessment 14, 375–390.
Bryan, B.A., King, D., Wang, E.L., 2010. Biofuels agriculture: landscape-scale trade-offs
between fuel, economics, carbon, energy, food, and fiber. Global Change Biology
Bioenergy 2, 330–345.
Bryan, B.A., King, D., Ward, J.R., 2011. Modelling and mapping agricultural opportunity
costs to guide landscape planning for natural resource management. Ecological In-
dicators 11, 199–208.
Bryant, C.R., Smit, B., Brklacich, M., Johnston, T.R., Smithers, J., Chiotti, Q., Singh, B.,
2000. Adaptation in Canadian agriculture to climatic variability and change. Cli-
mate Change 45, 181–201.
Chan, L.K.C., Karceski, J., Lakonishok, J., 1998. The risk and return from factors. Journal
of Financial and Quantitative Analysis 33, 159–188.
Chan, L.K.C., Karceski, J., Lakonishok, J., 1999. On portfolio optimization: forecasting co-
variances and choosing the risk model. Review of Financial Studies 12, 937–974.
Cooper, P.J.M., Dimes, J., Rao, K.P.C., Shapiro, B., Shiferaw, B., Twomlow, S., 2008. Coping
better with current climatic variability in the rain-fed farming systems of sub-
Saharan Africa: an essential first step in adapting to future climate change? Agri-
culture, Ecosystems & Environment 126, 24–35.
Correal, E., Robledo, A., Rios, S., Rivera, D., 2006. Mediterranean Dryland Mixed Sheep-
Cereal Systems. Sociedad Espanola para el Estudio de los Pastos (SEEP), pp. 14–26.
Deressa, T.T., Hassan, R.M., 2009. Economic impact of climate change on crop produc-
tion in Ethiopia: evidence from cross-section measures. Journal of African Econo-
mies 18, 529–554.
Fraser, E.D.G., 2007. Travelling in antique lands: using past famines to develop an
adaptability/resilience framework to identify food systems vulnerable to climate
change. Climate Change 83, 495–514.
Fraser, E.D.G., Mabee, W., Figge, F., 2005. A framework for assessing the vulnerability of
food systems to future shocks. Futures 37, 465–479.
Furuya, J., Kobayashi, S., 2009. Impact of global warming on agricultural product mar-
kets: stochastic world food model analysis. Sustainability Science 4, 71–79.
Hardaker, J.B., Lien, G., 2010. Probabilities for decision analysis in agriculture and rural
resource economics: the need for a paradigm change. Agricultural Systems 103,
345–350.
Hardaker, J.B., Huirne, R.B.M., Anderson, J.R., Lien, G., 2004. Coping with Risk in Agricul-
ture. xii +332 pp.
Hazell, P.B.R., Jaramillo, M., Williamson, A., 1990. The relationship between world price
instability and the prices farmers receive in developing-countries. Journal of Agri-
cultural Economics 41, 227–240.
Howden, S.M., Soussana, J.F., Tubiello, F.N., Chhetri, N., Dunlop, M., Meinke, H., 2007.
Adapting agriculture to climate change. Proceedings of the National Academy of
Sciences of the United States of America 104, 19691–19696.
Iglesias, A., Quiroga, S., 2007. Measuring the risk of climate variability to cereal produc-
tion at five sites in Spain. Climate Research 34, 47–57.
IPCC, 2007. Climate change 2007: synthesis. Summary for policy makers. Fourth As-
sessment Report of the Intergovernmental Panel on Climate Change. Cambridge
University Press, Cambridge.
John, M., Pannell, D., Kingwell, R., 2005. Climate change and the economics of farm
management in the face of land degradation: dryland salinity in western Australia.
Canadian Journal of Agricultural Economics/Revue Canadienne d'Agroeconomie
53, 443–459.
Keating, B.A., Carberry, P.S., Hammer, G.L., Probert, M.E., Robertson, M.J., Holzworth, D.,
Huth, N.I., Hargreaves, J.N.G., Meinke, H., Hochman, Z., McLean, G., Verburg, K.,
Snow, V., Dimes, J.P., Silburn, M., Wang, E., Brown, S., Bristow, K.L., Asseng, S.,
Chapman, S., McCown, R.L., Freebairn, D.M., Smith, C.J., 2003. An overview of
APSIM, a model designed for farming systems simulation. European Journal of
Agronomy 18, 267–288.
Kelkar, U., Narula, K.K., Sharma, V.P., Chandna, U., 2008. Vulnerability and adaptation to
climate variability and water stress in Uttarakhand State, India. Global Environ-
mental Change: Human and Policy Dimensions 18, 564–574.
Kingwell, R.S., 1994. Risk attitude and dryland farm-management. Agricultural Systems
45, 191–202.
Kurukulasuriya, P., Ajwad, M.I., 2007. Application of the Ricardian technique to esti-
mate the impact of climate change on smallholder farming in Sri Lanka. Climate
Change 81, 39–59.
Ladanyi, M., 2008. Risk methods and their applications in agriculture. Applied Ecology
and Environmental Research 6, 147–164.
Lien, G., Hardaker, J.B., van Asseldonk, M., Richardson, J.W., 2009. Risk programming
and sparse data: how to get more reliable results. Agricultural Systems 101, 42–48.
Lotze-Campen, H., Schellnhuber, H.J., 2009. Climate impacts and adaptation options in
agriculture: what we know and what we don't know. Journal fur Verbraucherschutz
und Lebensmittelsicherheit-Journal of Consumer Protection and Food Safety 4,
145–150.
Ludwig, F., Milroy, S.P., Asseng, S., 2009. Impacts of recent climate change on wheat
production systems in Western Australia. Climate Change 92, 495–517.
Luo, Q.Y., Bryan, B., Bellotti, W., Williams, M., 2005. Spatial analysis of environmental
change impacts on wheat production in Mid–Lower North, South Australia. Cli-
mate Change 72, 213–228.
Luo, Q.Y., Bellotti, W., Williams, M., Cooper, I., Bryan, B., 2007. Risk analysis of possible
impacts of climate change on South Australian wheat production. Climate Change
85, 89–101.
Magombeyi, M.S., Taigbenu, A.E., 2008. Crop yield risk analysis and mitigation of small-
holder farmers at quaternary catchment level: case study of B72A in Olifants river
basin, South Africa. Physics and Chemistry of the Earth 33, 744–756.
Markowitz, H.M., 1952a. Portfolio selection. Journal of Finance 77–91.
Markowitz, H.M., 1952b. Portfolio Selection: Efficient Diversification of Investments.
Yale University Press, New Haven, CT.
Markowitz, H.M., 1994. The general mean-variance portfolio selection problem. Philo-
sophical Transactions of the Royal Society of London, Series A: Mathematical, Phys-
ical and Engineering Sciences 347, 543–549.
Marton, L., Pilar, P.M., Grewal, M.S., 2007. Long-term studies of crop yields with chang-
ing rainfall and fertilization. Agrartechnische Forschung-Agricultural Engineering
Research 13, 37–47.
McLaren, C., 1997. Dry Sheep Equivalents for comparing different classes of livestock.
Agriculture Notes AG05901329-8062.
Meza, F.J., Silva, D., 2009. Dynamic adaptation of maize and wheat production to cli-
mate change. Climate Change 94, 143–156.
Naylor, R.L., Battisti, D.S., Vimont, D.J., Falcon, W.P., Burke, M.B., 2007. Assessing risks of
climate variability and climate change for Indonesian rice agriculture. Proceedings
of the National Academy of Sciences of the United States of America 104,
7752–7757.
Pannell, D.J., Malcolm, B., Kingwell, R.S., 2000. Are we risking too much? Perspectives
on risk in farm modelling. Agricultural Economics 23, 69–78.
Power, S., Tseitkin, F., Mehta, V., Lavery, B., Torok, S., Holbrook, N., 1999. Decadal cli-
mate variability in Australia during the twentieth century. International Journal
of Climatology 19, 169–184.
Ramaswami, B., Ravi, S., Chopra, S.D., 2003. Risk Management in Agriculture, Discus-
sion Papers in Economics 03-08. Indian Statistical Institute, Delhi Planning
Unitistical Institute, Delhi Planning Unit.
Ranjan, R., Shogren, J.F., 2009. Dynamic endogenous risks & non-expected utility be-
havior. The Korean Economic Review 215–240.
Rockafellar, R.T., Uryasev, S., 2002. Conditional value-at-risk for general loss distribu-
tions. Journal of Banking & Finance 26, 1443–1471.
Suppiah, R., Preston, B., Whetton, P.H., McInnes, K.L., Jones, R.N., Macadam, I., Bathols, J.,
Kirono, D., 2006. Climate Change under Enhanced Greenhouse Conditions in South
Australia: An Updated Report on: Assessment of Climate Change, Impacts and Risk
Management Strategies Relevant to South Australia, CSIRO Regional Climate
Change Report for SA Gov.
Thomas, R.J., 2008. Opportunities to reduce the vulnerability of dryland farmers in Cen-
tral and West Asia and North Africa to climate change. Agriculture, Ecosystems &
Environment 126, 36–45.
Thomas, R.J., Pauw, E.D., Qadir, M., Amri, A., Pala, M., Yahyaoui, A., El-Bouhssini, M.,
Baum, M., Iniguez, L., Shideed, K., 2007. Increasing the resilience of dryland agro-
ecosystems to climate change. Journal of SAT Agricultural Research 4, 1–37.
Uryasev, S., Rockafellar, R.T., 2001. Conditional value-at-risk: optimization approach.
In: Uryasev, S.P., Pardalos, P.M. (Eds.), Stochastic Optimization: Algorithms and Ap-
plications. Springer, Dordrecht, pp. 411–435.
van Ittersum, M.K., Howden, S.M., Asseng, S., 2003. Sensitivity of productivity and deep
drainage of wheat cropping systems in a Mediterranean environment to changes
in CO
2
, temperature and precipitation. Agriculture, Ecosystems & Environment
97, 255–273.
Wang, E., Cresswell, H., Bryan, B., Glover, M., King, D., 2009. Modelling farming systems
performance at catchment and regional scales to support natural resource man-
agement. NJAS - Wageningen Journal of Life Sciences 57, 101–108.
Wang, E.L., McIntosh, P., Jiang, Q., Xu, J., 2009. Quantifying the value of historical climate
knowledge and climate forecasts using agricultural systems modelling. Climate
Change 96, 45–61.
World-Bank, 2004. Agricultural Diversification for the Poor Guidelines for Practitioners.
Yang, C., 2009. Climate change and food security. Crop, Environment & Bioinformatics
6, 134–140.
Yang, W.H., Bryan, B.A., MacDonald, D.H., Ward, J.R., Wells, G., Crossman, N.D., Connor,
J.D., 2010. A conservation industry for sustaining natural capital and ecosystem
services in agricultural landscapes. Ecological Economics 69, 680–689.
112 J.M. Kandulu et al. / Ecological Economics 79 (2012) 105–112