8/8/2019 Risk Analysis and Management in Rainfed Rice Systems
1/213
8/8/2019 Risk Analysis and Management in Rainfed Rice Systems
2/213
8/8/2019 Risk Analysis and Management in Rainfed Rice Systems
3/213
Risk Analysisin Rainfed
and ManagementRice Systems
S. Pandey, B.C. Barah, R.A. Villano, and S. Pal, Editors
IRRI
8/8/2019 Risk Analysis and Management in Rainfed Rice Systems
4/213
8/8/2019 Risk Analysis and Management in Rainfed Rice Systems
5/213
Contents
Introduction
Units of measurement
Risk and rainfed rice: some conceptual and methodological issues
S. Pandey
Risk and rainfed rice in India: an overview
C. Ramasamy and K. Uma
Decomposition of income variability in rainfed areas: the case of rice
in eastern India
B.C. Barah
Labor use and employment pattern in rainfed rice-producing states of India
G.K. Chadha
Crop insurance: a policy perspective
P.K. Mishra
The nature and causes of changes in variability of rice
production in eastern India: a district-level analysis
S. Pandey and S. Pal
Growth and variability in agriculture revisited: district-level evidence of
rice production in eastern India
S. Pal, S. Pandey, and Abedullah
Rainfed rice and risk-coping strategies: some microeconomic
evidence from eastern Uttar Pradesh
S. Pandex H.N. Singh, and R.A. Villano
Risk and the value of rainfall forecast for rainfed rice in the Philippines
Abedullah and S. Pandey
Characterizing risk and strategies for managing risk in
flood-prone rice cultivation in Assam
B.C. Bhowmick, S. Pandey, R.A. Villano, and J.K. Gogoi
Risk and its management in the rainfed rice ecosystem of Bihar
J. Thakur
Risk and its management in rainfed rice ecosystem of West Bengal
N.K. Saha, S.K. Bardhan Roy, and U.S. Aich
Risk and rice production in Orissa, eastern India
D. Naik, S. Pandey, D. Behura, and R.A. Villano
Risk and rice technology design
L.J. Wade
Participants
v
xi
1
21
29
39
63
73
97
115
135
143
157
165
173
187
195
8/8/2019 Risk Analysis and Management in Rainfed Rice Systems
6/213
8/8/2019 Risk Analysis and Management in Rainfed Rice Systems
7/213
8/8/2019 Risk Analysis and Management in Rainfed Rice Systems
8/213
components of farming systems (not just that of
rice), and that help stabilize area (not just yield)
are seen as important for reducing risk. As such
technologies will require farmers to use
information on weather and other factors that
condition crop performance, provision of such
information is seen as an important risk-
reducing strategy. Similarly, dissemination of
such risk-reducing technologies that tend to be
somewhat information-intensive will require
reform of the extension system that is designed,
in most countries, for delivering simple
technology packages.
risk-related literature in the context of rainfed
rice farming in India. While recognizing the
importance of climatic risk, they emphasize risks
associated with the timely supply of inputs and
with prices. The review also indicates that most
of the literature on the study of farm level risk in
rice production in India is somewhat dated.
Perhaps a rapid growth in productivity realized
through the adoption of modem varieties and
other related technologies in irrigated areas
during the Green Revolution period diverted
attention from problems facing rainfed areas.
However, with the increasing attention now
being paid to rainfed rice systems of easternIndia, interest in issues of risk and its
management has come once again to the
forefront. Ramasamy and Uma identify
important areas that require increased research
attention to develop risk-reducing interventions.
variability can be an important source of revenue
variability of rice farmers. Using the method of
variance decomposition, Barah finds that price
variability is more important than yield
variability in irrigated environments, but that theopposite holds true in the rainfed environment.
The implication is that the design of price policy
should vary according to the environmental
conditions including basic infrastructure. Of
course, this poses the challenge of how to design
a differential price policy in an environment
where spatial economic linkages are growing
stronger. Yield-stabilizing and -enhancing
measures such as biotic and abiotic stress-
tolerant varieties and insurance against
Ramasamy and Uma provide an overview of
The paper by Barah shows that price
vi
calamities are preferred policies particularly in
the poorly endowed regions.
employment in income diversification and,
consequently, in risk management of farm
households in India. Using rural employment
data for India, the author shows that expansion
of rural nonfarm employment has helped
farmers manage risk better. In addition,
expansion of nonfarm rural industry will directly
promote economic growth by better use of
forward and backward linkages associated with
agricultural growth. Improvements in
infrastructure and agrarian reforms are seen as
important policy interventions needed to
stimulate the growth of nonfarm employment in
rural areas.
Mishra discusses the issues related to crop
insurance as a policy response for stabilization
of crop income. Although the current wisdom is
that publicly funded crop insurance programs
are financially unviable, the author finds the
total social benefit of the comprehensive crop
insurance scheme in India to be higher than the
total cost. Thus, from the social point of view,
the author shows that crop insurance programs
can be desirable, even though they may not be
financially viable without the subsidy. Nevertheless, the author suggests that
opportunities for improving the financial
performance of crop insurance schemes should
be exploited as much as possible to improve
their financial viability. The problems of
adverse selection and moral hazard, however,
continue to erode the viability of crop insurance
schemes.
Two papers (Pandey and Pal, and Pal et al)
have focused attention on assessing the pattern
of changes in productivity and variability ineastern India using district-level data. The
authors report a diverse pattern of change with
variability increasing in some districts,
decreasing in others, but remaining more or less
unchanged in most. Eastern Uttar Pradesh and
West Bengal are the two states where the change
in variability (defined by the CV of production)
has been stabilizing. This is attributed mainly to
the availability of supplemental irrigation that
reduced risk and encouraged farmers to adopt
Chada discusses the role of nonfarm rural
8/8/2019 Risk Analysis and Management in Rainfed Rice Systems
9/213
yield-increasing modem varieties. Parts of
Bihar and Orissa, on the other hand, experienced
an increase in production variability. The
coefficient of variation of district-level yield was
found to be positively related to the coefficient
of rainfall and negatively related to the quantity
of fertilizer used. As the latter is an indicator of
the extent of adoption of improved technologies,
the stability consequences of the adoption of
improved technologies appear to have been
favorable in eastern India. The findings of these
two studies support the view that productivity
gains in parts of eastern India have been
achieved without increasing instability. This
indicates that growth and stability are not
necessarily incompatible goals. The expansion
of irrigation and development of rice varieties
suitable to the environment of eastern India are
likely to be the causal factors that have reduced
instability and increased growth simultaneously.
However, there is an underlying trend toward an
increasing correlation in production across
districts that, if unchecked, could have a
destabilizing effect.
The Pandey et al paper analyzes risk
management strategies using panel data from
two villages with contrasting risk profiles in
eastern Uttar Pradesh. Diversification andmaintenance of flexibility are seen as two major
strategies for reducing risk. The analysis of
panel data permitted the authors to document
changes in cropping patterns, varieties of rice
grown, methods of crop establishment, and input
use over time and relate these changes to
rainfall. The paper shows that area variability is
an important component of variability in rice
production. Most of the biological research on
rice ignores area variability and focuses on yield
variability. One important contribution of this paper is that it shows that the risk benefits of
stabilization of rice yield in the study villages
are quite small. This is mainly due to a very
small share of rice in the total household
income. As a result, stabilization of rice income
will not necessarily translate into stabilization of
total household income. As farmer income
sources are already diversified away from rice,
rice research can have more impact by focusing
on yield improvement rather than on yield
stabilization per se. However, in other areas
where income diversification opportunities are
constrained by infrastructure and biophysical
factors, stabilization of rice yield can result in
substantial income gains.
The paper by Abedullah and Pandey
provides an estimate of the economic value of
rainfall forecast to rainfed rice farmers in the
Philippines. This is the only paper in the
volume that includes data from outside India.
Using a decision-theoretical approach, the
authors estimate the value of three types of
seasonal rainfall forecast (average, below
average, and above average). The economic
value of forecasts arises from farmers being able
to alter crop management practices if they have
access to the forecasts. To get around the
problems related to forecast accuracy, the
authors estimate the economic value of a perfect
forecast as such estimates provide the upper
limit to the value of a forecast. The estimated
value of such a forecast was found to be 1% of
the net returns from rice. For the rainfed rice
area of the Philippines, the total value was
estimated to be $6.6 million per year.
et al, and Naik et al) analyze rice production
and instability in Assam, Bihar, West Bengal,
and Orissa, respectively, the major rice- producing states of eastern India. Although rice
is grown in three different time periods in
eastern India, the papers show that rainfed rice is
grown mainly in July to November and the
variability of total rice production is determined
mainly by the variability during this period. The
Bhowmick et al paper highlights the importance
of flood risk in Assam. Overall, the variability
of rice production in Assam has changed very
little. The results from Bihar show that its
variability of rice production and yield is thehighest of all eastern Indian states. The
interaction between modem varieties and
complex hydrology in Bihar is probably the
main reason for increased production variability
in this state. Highly variable environmental
conditions could also be a reason for the
shrinkage of rice area in this state. The Saha et
al paper on West Bengal provides a more
detailed description of changes in productivity
patterns in West Bengal and how farmers
manage risk by adjusting crop management.
Four papers (Bhowmick et al, Thakur, Saha
vii
8/8/2019 Risk Analysis and Management in Rainfed Rice Systems
10/213
West Bengal is the only state with the least
variation in productivity over time. The paper
shows that low variability has resulted mainly
from stabilization of rainy-season rice
production even though the importance ofsummer rice has grown over time. The Naik et
al paper analyzes the instability of rice
production in Orissa. The paper shows that the
variability in rice yield across districts is not
related to the adoption of modern varieties but
mainly to soil/climatic factors. In the case of
Orissa, opportunities to reduce risk by
manipulating crop management practices of rice
seem circumscribed by hydrological factors,
especially in the coastal belt.
From a biological perspective, the Wade paper brings out clearly opportunities to reduce
risk through a better understanding of the
genotype by environment interactions.
Experimental data and crop simulation are seen
as important in understanding the nature of risk
and its management through manipulation of
varieties and crop management. The paper also
provides some insights into the more
downstream aspects of technology adaptation
and dissemination through farmer participatory
methods and involving nontraditional extension
agencies such as farmer organizations and
nongovernment organizations.
Summary and synthesis
The papers presented during the workshop and
the discussions that ensued covered many issues
related to agricultural growth in eastern India
and risk management. While most of the papers
used the concepts and methods that were
popularized when the study of interaction
between agricultural risk and technologyadoption was popular during the late 1970s and
1980s, the findings reported during the
workshop provide new insights into conceptual
and research issues. Some of these major issues
are summarized below.
Yield stabilization
Rice production is an important economic
activity in eastern India. Despite its importance,
the share of rice in the total household income
viii
may not be as high as often believed. If rice
contributes to only a small proportion of the
total income of farm households, interventions
that stabilize rice income are not necessarily
effective in stabilizing total household income.Thus, yield-stabilizing technologies for a single
crop are not likely to be effective in reducing
risk, even for an important crop such as rice.
However, in areas where income diversification
is limited due to limited infrastructure or less
favorable agroclimatic conditions, the economic
cost associated with instability in rice production
can be substantial. Thus, there is a need to
delineate rainfed rice environments in terms of
the current extent of income diversification and
the possibilities for diversification in the future.Interventions designed mainly for stabilizing
yield and incomes (such as technologies with
higher yield stability and crop insurance) are
likely to be less useful in environments with
ample opportunities for diversification.
Naturally, technologies that improve the average
yield of rice are always important, irrespective
of the nature of the environment.
Data needs
One of the major gaps in the current empirical
work on risk analysis in the context of rainfed
rice systems is the lack of information on the
relative importance of various risk-coping
mechanisms and how they change with
increasing commercialization of agriculture.
Very little information is available on the
determinants of various strategies that farmers
employ to cope with risk and their associated
cost. One of the difficulties has been the lack of
panel data to study the correlation between
climatic fluctuations and farmers responses.Village-level studies such as the ones conducted
by the international Crops Research Institute for
the Semi-Arid Tropics (ICRISAT) can be
important in bridging the information gap for
rainfed rice systems also.
Risk and externality
Most of the current studies on risk for rainfed
rice areas focus on the farm or household as the
unit of analysis. Little information is available
8/8/2019 Risk Analysis and Management in Rainfed Rice Systems
11/213
at a higher level of aggregation such as the
community, district, or state. Often, the source
of increased risk in downstream areas may be
the result of inappropriate land use in the
upstream areas. For example, deforestation andthe use of soil-eroding practices in the upper
parts of a watershed can increase the flood risk
in downstream areas. Such risks are better
managed through the use of more sustainable
land use systems in the upper slopes than by
other means. In this regard, collective
institutions can play an important role in the
management of overall risk for the whole
watershed. However, the study of interactions
between collective institutions and risk
management remains a relatively unchartedterritory.
Macroeconomic instability and risk
Another area of research that deserves adequate
attention is the effect of macroeconomic
instability on risk in agriculture. As agriculture
changes from subsistence to commercial
orientation, the agricultural sector becomes more
prone to macroeconomic shocks of fluctuations
in foreign exchange rates and interest rates.With the anticipated trend toward globalization
of trade following the WTO agreement, the
macroeconomic shocks are likely to be
transmitted more easily across countries.
Policies and institutional mechanisms needed to
stabilize food production and farmers income
under such conditions are yet to be adequately
scrutinized.
Upscaling and extrapolation
A methodological issue relates to the
quantification of the impacts (production losses,
income, and welfare) of unpredictable shocks at
different geographic scales. Farm-level losses
can be quantified through a sample survey and
other traditional methods. However,
geographically referenced spatial information on
factors that affect production losses are needed
to extrapolate such information to the regional
and subnational levels. Such databases are now
becoming increasingly available with the
growing popularity of geographic information
systems (GIS). Nevertheless, methods and
approaches are needed to upscale farm level
effects of risk to a higher level of aggregation.
Economic value of forecasts
Opportunities for reducing the economic cost of
risk by providing of forecast information have
not been adequately addressed in the agricultural
sector worldwide, except for some specific high-
value crops. In most developing countries,
climatic forecasts are rarely available in a form
useful to farmers for planning agricultural
operations. Similarly, opportunities for reducing
risk through the better use of information of
forecast prices are rarely available.Policymakers and farmers alike have not
adequately appreciated the importance of
information acquisition and use for risk
management. Perhaps the value of information
is low in traditional subsistence-oriented
agriculture. But with increasing
commercialization, the use of forecast
information can be an important strategy for
reducing price and weather risk.
Risk analysis of rainfed rice systems
In rainfed rice areas, the overall theme of risk
analysis and management has not been
adequately studied. Parallels are drawn from
earlier work conducted in irrigated areas where
the interaction between risk and technology
adoption was widely discussed. While
theoretical and conceptual advances made in
such studies are relevant to rainfed environments
also, empirical applications for analyzing
farmers decisions regarding the choice of
technology, their risk-coping strategies, and the
overall effect of farmers risk management
strategies on production instability at the
aggregate level have been far too few. Due to
the high degree of heterogeneity in rainfed
environments, the domain of a specific
technology is likely to be much narrower. A
careful delineation of rice production
environments, based on risk profile and farmers
socioeconomic conditions, is needed to target
technology development. With the availability
of more powerful computer technology, many
8/8/2019 Risk Analysis and Management in Rainfed Rice Systems
12/213
powerful tools ranging from crop simulation to
GIS are now in the hands of analysts. More
studies that use such tools to quantify effects of
risk at the production systems level are needed.
Increasing the adoption of modem varieties ofrice even in rainfed areas and the growth in
income of farm households observed during the
1990s in eastern India indicate that farm
households have been able to mitigate the effect
of risk to a certain extent.
S. Pandey*
B.C. Barah
R.A. Villano
S. Pal
*Sushil Pandey and Renato Villano are agricultural economist and assistant scientist, respectively, at the
International Rice Research Institute, Los Baos, Laguna, Philippines; B.C. Barah and S. Pal are principal
scientist and senior scientist, respectively, at the National Centre for Agricultural Economics and Policy Research
(NCAP), New Delhi, India. The editors acknowledge support and encouragement from Dr. Mahabub Hossain, head,
Social Sciences Division, IRRI; Dr. D. Jha, national professor and exdirector, NCAP and Dr. Mruthyunjaya, director,
NCAP. Editorial assistance provided by Dr. Bill Hardy, Ms. Teresita Rola, Ms. Millet Magsino, Ms. Erlie Putungan,
Mr. Juan Lazaro IV, and Mr. George Reyes of the Communication and Publications Services, IRRI, is gratefully
acknowledged.
8/8/2019 Risk Analysis and Management in Rainfed Rice Systems
13/213
Units of measurement
All data on rice and production in this report are expressed in terms of rough rice. The conversion
factor used is 1 kg of rough rice = 0.66 kg of milled rice.
All monetary values for studies in India are expressed in Indian rupees. In 1997, the exchange
rate was 1US$ = Rs 36.31.
xi
8/8/2019 Risk Analysis and Management in Rainfed Rice Systems
14/213
8/8/2019 Risk Analysis and Management in Rainfed Rice Systems
15/213
Risk and rainfed rice: some conceptual andmethodological issues
S. Pandey
The study of risk and its interaction with technology is an important topic in agricultural
development. The paper provides a review and synthesis of conceptual and empirical issues
in risk analysis in the context of rainfed rice farming. Various strategies employed by farmers
for managing risk are discussed and implications of these strategies for designing and
disseminating technologies are derived. Methodological and measurement issues that require
further development are highlighted.
Rainfed rice farmers, like farmers everywhere,
have to carry out production activities in an
inherently uncertain environment. Production is
affected by drought, flood, and pests and
diseases, which occur in an unpredictable way.
In addition, farmers income and welfare also
depend on uncertainty related to economic
parameters such as price and marketing.
Efficient management of risk is hence the
essence of rainfed agriculture. For poor farmers,risk considerations may loom large in their
choices of crops and the method of production.
Hence, a study of how farmers are likely to
respond to technological and policy
interventions in the face of risk is critical in
designing these interventions.
Definition and measurement of risk
Although the word risk is used in all walks of
life to describe the chances of some undesirable
outcome, defining it precisely and
unambiguously is not easy. This is reflected in
the following statement: Risk is like love; we
have a good idea of what it is, but we cant
define it precisely (Stiglitz as quoted in
Roumasset et a1 1979). The Macquarie
Dictionary defines risk as exposure to the
chance of injury or loss. As injury or loss is
a subjective concept with its consequence
depending on the person as well as the
circumstance, what is considered to be risky by
one individual may not be seen to be so by
another person. Risk, hence, is subjective.
It is essential to draw a clear distinction
between risk and variability. The latter term
merely implies that a variable of interest is not
fixed but has different values. No risk is
involved if the value of the variable can be
known with certainty. For example, farm size
may vary from farmer to farmer but can be
known with certainty. Similarly, soil type withina farm can vary from paddock to paddock but
can be known with a fair degree of certainty.
Uncertainty about the likely values, not the
variability per se, is the source of risk.
A notion such as risk that is intrinsically
subjective obviously cannot be measured by an
objective indicator. Subjective probability
distribution of an uncertain outcome of concern
to the decision maker, hence, is considered a
suitable indicator of risk. Under this definition,
risk can be measured by (1) the chance of an
undesirable outcome, (2) the variability of
outcome (or the converse of stability), and (3)
the probability distribution of outcome. The first
measure implies that a situation in which the
chance of an undesirable outcome is greater is
riskier. Although intuitively appealing, the
measure is problematic because it is not clear
when an outcome is unacceptable.
distribution-suchas the variance and the
coefficient of variationhave found common
Statistical descriptors of the probability
1
8/8/2019 Risk Analysis and Management in Rainfed Rice Systems
16/213
use as a measure of risk. However, Rothschild
and Stiglitz (1970) showed that none of the
statistical descriptors adequately measure risk.
They contend that it is impossible to devise a
universally valid statistical descriptor of risk
without simultaneously considering both the
probability distribution of outcomes and the risk
attitude of the decision maker.
Given these difficulties in devising the
adequate measure of risk, it has been argued that
the ambiguous terms more risky or less risky
should be avoided. If no satisfactory measuring
scale exists, then it is not possible to consider
risk as being more or less. What is
theoretically appealing is to view a decision as
being risk efficient or risk inefficient. Such
decisions may lead to an increase in the mean
income and/or a reduction in the dispersion of
income around the mean. Risk efficiency can be
best ascertained by comparing the whole
probability distributions of the uncertain
outcomes that correspond to different decisions.
Risk and its impact on technologyadoption
The impact of risk and risk aversion on the
choice of agricultural production techniques andinput use has been a topic of extensive
investigation (Feder et al 1985, Anderson and
Hazel1 1994). Theoretical studies on farmer
behavior under risk indicate that, in the absence
of a perfect market for insurance, resource
allocation for risk-averse farmers differs from
that for risk-neutral farmers (Sandmo 1971,
Anderson et al 1977).
The effect of risk is considered to depend on
risk perceptions and risk attitudes. Farmers may
be reluctant to adopt technologies that they perceive to be riskier. Risk perception depends
on the quality of the information they have and
their information- processing capabilities. To the
extent that farmers perceive a technology to be
riskier than it actually is, activities such as on-
farm research to generate more accurate
information and investment in educating farmers
are warranted.
Assuming farmers perceptions of risk
associated with a technology to be reasonably
accurate, whether adoption occurs also depends
2
on risk attitudes. Variability of income is
irrelevant to risk-neutral farmers. A technology
that generates a higher level of mean income
would be preferred by such farmers. However,
risk-averse farmers are likely to consider
simultaneously both the level of income and risk
and to reject a technology that they consider too
risky.
Empirical evidence indicates that farmers in
developing countries are generally risk-averse
(Binswanger 1980, Walker and Ryan 1990). If
poorer farmers are more averse to risk, rainfed
rice farmers who are mostly poor are likely to be
reluctant to adopt technologies that increase risk.
In addition to this direct effect, risk aversion also
indirectly affects technology adoption through
its impact on the credit market (Binswanger and
Sillers 1983). Risk-averse lenders may demand
greater collateral and may charge a higher
interest rate, depressing credit use by poorer
farmers. Similarly, more risk-averse farmers are
less likely to demand credit. To the extent that
credit use is essential for adoption of
technologies that require purchased inputs (such
as fertilizers), risk aversion discourages
technology adoption. This indirect effect of risk
aversion is often considered to be more
important than the direct effect (Binswanger andSillers 1983).
The study of risk basically consists of two
aspects: risk analysis and risk management
analysis. Risk analysis consists of the study of
the nature, magnitude, and sources of risk and
how technology affects these characteristics.
Risk management, on the other hand, involves
the use of methods that reduce risk and its
impact. Even if a technology is risky, farmers
may adopt it if adequate means for diffusing risk
are available.Risk could be studied at the micro (or farm)
level or at the macro (region or nation) level.
The purpose of farm-level analysis is to study
adoption decisions in the face of risk.
Macroeconomic parameters are assumed to be
given and farmers responses to risk are studied.
In the case of macro analysis. the purpose is to
study the implications of fluctuating production
for food security at the regional or national level.
Although farmers may adopt improved
technologies because they are profitable, the
8/8/2019 Risk Analysis and Management in Rainfed Rice Systems
17/213
instability of production at the aggregate level
may increase as a consequence. Appropriate
technological and policy interventions are
required to reduce such adverse effects on food
security.
Sources of risk
Income of farmers from agricultural production
can fluctuate as a result of fluctuations in yield,
price of output, area planted, price of input, and
input supply. Agricultural scientists are mainly
concerned with yield risk as it is often a major
component of risk, especially under rainfed
conditions. If a farmer is ultimately interested in
reducing the uncertainty of income (as income
and consumption in rural societies are highlycorrelated), other components of risk can also be
important. For example, a negative correlation
between the price of rice and yield tends to
stabilize the income from rice compared with a
situation when these two variables are positively
correlated. Hence, if the interest is in stabilizing
farmers incomes, it is necessary to evaluate the
consequence of price instability and how it
affects income stability. Evaluation of
technology in terms of instability of yield alone
will not be adequate. The importance of price
uncertainty is likely to increase as rice
production systems become more
commercialized.
income risk, especially in commercialized
systems. When the use of purchased inputs is
minimal and output is mainly for subsistence,
market prices are less relevant for resource
allocation by farmers. However, in
commercialized systems where traded inputs are
substituted for nontraded inputs and output is
mainly for the market, fluctuations in prices of both inputs and outputs can have a major impact
on farmer welfare.
A negative correlation between price and
yield is an important feature of agricultural
production, which helps in stabilizing farm
income. Prices tend to be high when production
is low and they tend to be low when production
is high. Stabilizing prices in this situation can
actually raise farm income instability. Market-
level analyses and the study of price policy are
Price risk is an important component of
required to help design price policies that
enhance farm income stability.
can lead to fluctuations in output as farmers
adjust input levels to prevailing conditions.
Marketing infrastructure is important indetermining input supply risks. Similarly,
government policies on production and
marketing of agricultural inputs determine input
supply and price risks.
Variability in input supply and input prices
Risk at the aggregate level
Even if aggregate food production is increasing,
wide fluctuations in total supply can seriously
affect food security at the household level,
especially that of the poor. The nature of public-sector interventions in the food market depends
on the instability of aggregate production. The
economic costs of maintaining a bigger stock of
food grain to deal with higher instability can be
substantial. In addition, the effect of instability
at the national level also spills over to
international markets and can cause wide swings
in prices, thus affecting food security in other
countries also. Analysis of the patterns of
instability in food grain production is hence
relevant in the context of food security.
of improved technology consisting of high-
yielding varieties (HYVs) and associated crop
management practices has increased food
production in Asia over the last 20 years. What
is still debatable is the effect on variability of
production. The adoption of modern varieties
and improved crop management techniques can
make aggregate production more variable by
increasing interregional correlation. When
farmers grow similar varieties and use similar
management practices, adverse climaticconditions over a large area can lead to a large
drop in production. Similarly, when the supply
of major inputs is unreliable and/or input prices
change, farmers are likely to adjust their input
use in the same direction, leading to covariate
movement in output. This economic response
can lead to increased production instability even
if the yield of modem varieties is more stable
than that of traditional varieties.
It is now widely accepted that the adoption
3
8/8/2019 Risk Analysis and Management in Rainfed Rice Systems
18/213
Empirical evidence shows that production
variability in the aggregate has increased with
the adoption of improved varieties in India
(Hazell 1982). Much of the increase in
production variability in food grains has been
attributed not to the adoption of improvedvarieties per se but to fluctuations in input
supply such as irrigation (due to power outages)
and fertilizers (Hazell 1982). On the other hand,
Walker (1989) found the adoption of HYVs of
sorghum and pearl millet to be a major factor
contributing to increased production variability
of these crops. Using district-level data from
India, Singh and Byerlee (1990) found that
variability in wheat yield, measured by the
coefficient of variation, has decreased over time,
mainly as a result of expansion in irrigated area.Rao (1968), Mehra (1981), and Pandey (1989)
have also discussed the effect of irrigation on
production variability.
In a more recent study covering three major
cropsrice, wheat, and maizeNaylor et al
(1997) found that the variability in global output
of rice and wheat, measured by the average
percentage deviation from the trend, increased
initially with the adoption of modem varieties
but then declined subsequently. The probability
of a significant shortfall below the trend also
decreased between the pre-
and post-Green
Revolution periods for these two crops. In the
case of maize, production variability was higher
during 1980-94 than during 1950-64. The
dominance of U.S. maize production in global
output and a greater downside sensitivity of
yield to climatic conditions when yields are
close to the ceiling have been attributed to the
increase in variability in global maize
production. Although the rapid growth in yield
of rice and wheat may have swamped the
increase in Variability during the GreenRevolution period, instability may be more
pronounced in the future as yield ceilings for
these crops are also approached. More evidence
on the aggregate effects of technology adoption
is contained in papers by Pal et al (2000) and
Pandey and Pal (2000).
4
Coping mechanisms of farmers
As a result of their exposure to risk, farmers
have developed various strategies over time to
avoid the negative consequences of
unpredictable variations in agricultural output. Agood understanding of these strategies is needed
to assess the likely responses of farmers to new
technologies or policies. Uptake of technologies
that complement and reinforce farmers coping
strategies is likely to be quite rapid. On the other
hand, interventions that undermine key
components of risk management strategies are
likely to be rejected.
Farmers risk-coping strategies can be
classified into ex ante and ex post, depending on
whether they help reduce risk or reduce theimpact of risk after a production shortfall has
occurred. Because of the lack of efficient
market- based mechanisms for diffusing risk,
farmers modify their production practices to
provide self-insurance so that the chances of
negative consequences are reduced to an
acceptable level. Ex ante strategies help reduce
fluctuations in income and are also referred to as
income-smoothing strategies. These strategies
can be costly, however, in terms of forgone
opportunities for income gains as farmers select
safer but low-return activities.
categories: those that reduce risk by
diversification and those that do so through
greater flexibility. Diversification is simply
captured in the principle of not putting all eggs
in one basket. The risk of income shortfall is
reduced by growing several crops that have
negatively or weakly correlated returns. This
principle is used in different types of
diversification common in rural societies.
Examples are spatial diversification of farms,diversification of agricultural enterprises, and
diversification from farm to nonfarm activities.
Maintaining flexibility is an adaptive
strategy that allows farmers to switch between
activities as the situation demands. Flexibility in
decision making permits farmers not only to
Ex ante strategies can be grouped into two
8/8/2019 Risk Analysis and Management in Rainfed Rice Systems
19/213
reduce the chances of low incomes but also to
capture income-increasing opportunities when
they do arise. Examples are using split doses of
fertilizers, temporally adjusting input use to crop
conditions, and adjusting the area allocated to a
crop depending on climatic conditions. While postponing agricultural decisions until
uncertainties are reduced can help lower
potential losses, such a strategy can also be
costly in terms of income forgone if operations
are delayed beyond the optimal biological
window.
shortfall in consumption when family income
drops below what is necessary for maintaining
consumption at its normal level. They are also
referred to as consumption-
smoothing strategiesas they help reduce fluctuations in consumption
even when income is fluctuating. These include
migration, consumption loans, asset liquidation,
and charity. A consumption shortfall can occur
despite these ex post strategies if the drop in
income is substantial.
strategies in different combinations to ensure
survival. Over a long period of time, some of
these strategies are incorporated into the nature
of the farming system and are often not easilyidentifiable as risk-coping mechanisms. Others
are employed only under certain risky situations
and are easier to identify as responses to risk.
Ex ante coping mechanisms
Ex ante coping mechanisms are designed to
exploit low correlation among activity returns
for stabilization of total income. These operate
through various types of diversification that
characterize traditional agriculture.Diversification may be considered as horizontal
or vertical. The former refers to scattering of
agricultural fields, growing of several crops,
growing of several varieties of the same crop,
and engaging in different income-generating
activities. The latter relates to spreading
agricultural operations over time. This refers to
strategies such as staggered planting, spreading
input use over a period of time, planting many
seeds per hill, and temporally diverse planting.
Ex post strategies are designed to prevent a
Farmers who are exposed to risk use these
Vertical diversification is a way of maintaining
flexibility to adjust agricultural operations to the
evolving uncertainty. Similarly, share cropping
is viewed as a way of reducing risk through
sharing of risk between the landlord and the
tenant.Spatial diversification of fields. Agricultural
fields vary from location to location in attributes
such as soil moisture retention and fertility. In
rainfed areas, these soil characteristics can vary
widely even across fields. Similarly, rainfall
distribution can also vary among fields in
different locations. These variations in soil
characteristics and rainfall across locations
create an opportunity for fanners to stabilize
agricultural output through spatial scattering of
fields. Although output from fields in onelocation can decrease because of poor rainfall, it
can increase in fields in other locations that
receive higher rainfall. Weakly or negatively
correlated crop yields across fields result in
these compensating movements so that total
farm output is more stable than output from
individual fields. Spatial scattering of fields is a
way of exploiting this stabilizing effect. In
addition, this strategy may also help farmers to
better exploit specific niches of different
microenvironments to enhance productivityenhancement. In spite of these potential gains,
spatial diversification of fields can cause an
efficiency loss because of the increased costs of
moving inputs across and marketing outputs
from widely separated fields. Whether or not
farmers use spatial scattering depends on the net
effect of these factors. In addition, local
institutions such as the inheritance law may
condition the prevalence of such a practice.
In rice-growing regions of Asia, it is not
uncommon to find a farm household operatingseveral parcels of land that are either spatially
scattered or differ in their location along the
toposequence. While risk considerations may
have played a role in determining the extent of
land fragmentation, casual observation indicates
that land fragmentation is driven mainly by the
desire to exploit different environmental niches
that are suitable for different crops. In parts of
eastern India, ail parcels of land are divided
among legal heirs so that everybody gets an
5
8/8/2019 Risk Analysis and Management in Rainfed Rice Systems
20/213
equal share of all types of environmental niches.
The desire for an equitable distribution of land
of different quality among heirs is often
considered to be a factor constraining efforts at
land consolidation.
If land fragmentation is an effective way ofreducing risk, one would expect to observe a
greater degree of fragmentation in areas where
environmental conditions are less stable.
However, such a pattern may not be observed
due to other counteracting factors. For example,
the extent of fragmentation in the more risky
Sahel region of Africa has been less than in the
more favorable Sudan region (Matlon 1991).
This is attributed to the differences in
environmental factors in these two regions. In
the Sahel, low rainfall prevents farmers from
cultivating a wider range of field types. As a
result, cropping is restricted to only certain field
types where crop success is more assured. In
Sudan zone, higher rainfall and generally better
soil conditions enable farmers to use a range of
field types. In this example, the lack of feasible
alternatives in the highly constraining
environment of the Sahel reduced the value of
spatial diversification as a risk management tool.
Even if the inheritance law may have a big
role in determining farm size and extent of
fragmentation, farmers can and do alter their
land resource base through land rental markets.
Field experience from eastern India indicates
that tenants with a given endowment of land
types prefer to rent a different land type. Renting
a better quality land increases average income. It
may also simultaneously achieve the objective
of risk reduction.
diversification, farm output can be stabilized by
growing several crops with poorly or negatively
correlated yields. Environmental conditions lessfavorable to some crops may be more favorable
to others, so that compensating variations in
yields of different crops would impart stability
to total output. In addition to risk reduction,
there are several other potential benefits of crop
diversification, such as a better exploitation of
environmental niches, staggering of labor
demand, and meeting the demand for a range of
outputs. Mixed cropping and intercropping,
Crop diversification. As with spatial
6
which are a common feature of traditional
agriculture in Asia, are a form of crop
diversification that reduces output variability
(Walker and Jodha 1986, Siddiq and Kundu
1993). Crop diversification, however, can also
be costly in terms of income gain forgone asfarm households include crops with lower but
more stable yields in their cropping pattern. In
addition, economies of size that may result from
specialization are also lost as production is
diversified.
Crop diversification is a feature of
traditional farming systems in Asia. The role of
crop diversification in risk reduction has been
analyzed extensively in the context of farming in
the semiarid tropics where farmers grow a range
of intercrops and mixed crops. Crop
diversification has been greater in the more risky
environments in the semiarid tropics of India
(Walker and Jodha 1986). In the rainfed rice
environments of eastern India, crop
diversification is greater in areas with a less
assured supply of irrigation (Pandey et al, this
volume). Crop diversification in flood-prone
areas in a village in eastern India declined after
dikes for protection from flood were constructed
(Ballabh and Pandey 1999).
Although diversification may reduce
instability, whether or not farmers are able to
diversify land use also depends on the
environmental conditions. Again taking the
example from Africa, low and unstable rainfall
and poor soils in the Sahel have constrained
opportunities for diversification, with the millet-
based cropping pattern being the dominant one.
In comparison, in the relatively favorable
environments of the northern Guinea zone, the
cropping pattern is more diversified (Matlon
1991). In addition, the more limited cropping
opportunities in the Sahel also mean that cropyields are likely to be highly correlated, thus
reducing the benefits from crop diversification.
In the humid environments of Asia, drainage
constraints in the submergence- prone bottom
land similarly limit opportunities for crop
diversification during the rainy season.
Varietal diversification. Growing several
varieties of a crop is a form of diversification
that can stabilize the total output of the crop if
8/8/2019 Risk Analysis and Management in Rainfed Rice Systems
21/213
yields of different varieties are poorly correlated.
Varieties with different duration can reduce risk
by avoiding period-specific risk, For example,
short-duration varieties can escape terminal
drought that can severely affect the yield of a
longer duration variety. Similarly, varieties withdifferent degrees of tolerance for pests and
diseases also help reduce losses.
invariably grow several varieties for different
reasons, including possible risk reduction. In a
rainfed rice village in Orissa, more than 70% of
the farmers grow two to five varieties, with 20%
of the farmers growing six to eight varieties
(Kshirsagar et a1 1997). Similarly, in the rainfed
lowland of Lao PDR, 60% of the farmers grow
four or more rice varieties (Pandey and
Sanamongkhoun 1998). As with crop
diversification, other advantages of varietal
diversification are niche matching, staggering
labor demand, and generating a range of product
characteristics. These latter motives are not
directly related to risk management and may
condition the extent of varietal diversification
practiced by farmers in a given area.
Income diversification. Like crop
diversification that uses weak correlation among
activity returns to stabilize farm income,
diversification of income from farm to nonfarmsources is another way of stabilizing income. If
fluctuations in nonfarm incomes are independent
of fluctuations in farm output, income
diversification through one or more members of
the family working in the nonfarm sector can
stabilize total family income. The extent of
income diversification may depend on factors
such as rural education, transportation
infrastructure, access to institutional credit, and
availability of local resources for nonfarm
activities. These factors may constrainopportunities for income diversification even
when agricultural risk is high. In areas with
environmental conditions conducive to a strong
agricultural base, income-generating activities
that take advantage of agricultures forward and
backward linkages expand. On the other hand,
income diversification in agriculturally poor
areas tends to be outward-looking, with
households diversifying their income
Rainfed rice farmers in eastern India almost
geographically (Reardon et a1 1988, 1992).
on risk and efficiency implications of share
cropping exists (Newbery and Stiglitz 1979,
Otsuka et a1 1992). At their very basic, share
cropping arrangements that lead to sharing ofinput and output also lead to sharing of risk
between the landlord and the tenant. However,
the existence of share cropping depends on
many other factors in addition to risk benefits
(Otsuka et a1 1992).
Temporal adjustments. Crop growth is a
biological process that occurs over a period of
time. The economic output is obtained upon
maturity when the crop is finally harvested. The
crop is exposed to various factors during the
intervening period between planting and harvest.
Some of these factors are known with a fair
degree of certainty, wheras others are highly
uncertain. These factors, together with
management interventions by farmers, determine
the ultimate economic value of the crop output.
Uncertainties are highest at planting time as
future values of uncertain events are known very
imprecisely. As uncertainties are resolved with
the passage of time, farmers can gain by making
decisions conditional on the occurrence of
uncertain events up to that time and the revised
expectation about the future occurrence ofuncertain events. Such a sequential decision-
making process imparts flexibility and allows
farmers to exploit favorable events for income
gains while reducing potential losses.
To assess the value of sequential decision
making, it may be useful to divide the cropping
season into early, middle, and late stages. The
early stage can be considered to include
preplanting and the period immediately after
planting. The major decisions to be made at this
stage are the crops, the variety, the timing of planting, and the method of establishment. The
middle stage is considered to be the period
between successful crop establishment and
flowering. Major decisions here are weeding,
fertilization, control of pests, and irrigation. The
final stage is the period after flowering until
harvest.
may determine the choice of crops. If rains are
Share cropping. A large volume of literature
The rainfall pattern during the early stage
7
8/8/2019 Risk Analysis and Management in Rainfed Rice Systems
22/213
low or delayed during this period, farmers may
forgo rice completely and expand the area under
crops that require less water. Similarly, if too
much water is received, farmers may expand the
area under rice at the expense of other crops. In
eastern India, sown area of rice contracts inyears with low and unstable early-season rainfall
(Pandey et al, 2000a). If the crop fails to
establish itself because of too much or too little
rain, farmers may decide to replant. Farmers
similarly may engage in gap filling and thinning
to reduce risk (Singh et al 1995).
The choice of what rice variety to grow also
depends partly on the nature of rainfall during
this early period. Farm-level data from eastern
India indicate that, in years with late rains,
farmers expand the area under short-
durationvarieties as a mechanism for escaping terminal
drought. Expanding the proportionate area under
traditional varieties and resorting more to dry
seeding as opposed to transplanting are other
responses exhibited by farmers in eastern India.
Once the crop is successfully established,
farmers may adapt the level of input they use,
depending on their assessment of crop health. If
the crop potential appears to be low, farmers
may leave some fields unweeded and apply
lower than normal quantities of fertilizer.
Surplus resources may be used for other crops in
the same or the following season. Farmers even
replant the area with some other crops if they
anticipate the rice yield being too low and if the
season has not advanced too far (Singh et al
1995).
During the third stage, most of the
uncertainties would have been resolved and few
decisions would remain to be made. If rice fails
during this stage, farmers may go for salvage
operations to obtain at least the byproduct
(straw, in the case of rice). Another responseobserved in eastern India is to establish post
rainy-season crops early in the rice field if soil
moisture conditions are favorable.
The temporal adjustments described above
are farmers mechanisms for reducing losses in
poorer years and increasing gains in more
favorable years. Relative to committing all
resources at the beginning of the cropping
season or on the basis of a fixed calendar, the
8
average farm income will always be higher
when flexible methods are adopted. However,
opportunities for using flexibility may be
constrained by farmers ability to process the
necessary information about crop status and the
likely future occurrence of uncertain events. Inaddition, in poorer and harsher environments,
flexibility may be so circumscribed that it cannot
be relied upon as an effective risk-coping
mechanism.
Ex post coping mechanisms
How do farmers cope with losses that do occur
despite the various risk-reducing mechanisms
adopted? The shortfall in agricultural
production will reduce consumption if farmersare not able to meet the deficit through some
other means. Depleting food and cash savings,
earning more wage income, borrowing,
liquidating assets, reducing consumption,
relying on charity, and permanent migration are
some of the mechanisms used for coping with a
production shortfall. The economic burden and
the long-term productivity impacts of these
mechanisms differ.
If farmers are able to save during better-
than-normal years and use the savings to meet
consumption deficits during drought years, they
may be able to maintain their consumption level
over time despite short-term fluctuations in
agricultural output. Savings in agricultural
societies may take various forms. They could be
held in the form of food grains, cash, and
jewelry. They could also be held in the form of
productive assets such as bullocks, farm
implements, and land. Even if own savings are
not enough to meet the consumption deficit,
village-level institutions may permit sharing of
risk across individuals such that individualconsumption fluctuates much less than
individual production.
developing countries indicates that consumption
smoothing is a common practice among farmers
(Townsend 1994, 1995). Based on data from the
International Crops Research Institute for the
Semi-Arid Tropics (ICRISAT), crop inventory
and cash reserves play major roles in smoothing
Empirical evidence from several studies in
8/8/2019 Risk Analysis and Management in Rainfed Rice Systems
23/213
consumption in the semiarid tropics of India
(Lim and Townsend 1994, Paxson and
Chaudhuri 1994). The importance of these two
mechanisms varies by farm size, with large
farmers relying more on crop inventory and
small farmers relying more on currency. The useof credit was another important mechanism.
Results for Thailand were also similar
(Townsend 1995).
. The effectiveness of these mechanisms
depends on the seventy of risk such as drought
and crop output in the preceding year. Problems
are less severe in a year with mild drought that
follows a good year, and these mechanisms may
be adequate to meet the shortfall. These internal
reserves, however, may be grossly inadequate
when drought years are consecutive or if droughtis severe. In such situations, farmers may be
forced to reduce consumption, with small
farmers and landless labor suffering the most.
Based on farm-level data from arid and
semiarid areas of India, the decline in cereal
consumption in a drought year relative to a
normal year varied between 12% and 22%
(Jodha 1978). In addition, there were drastic cuts
in the expenditure on protective food such as
milk, sugar, vegetables, fruits, meat, and others.
Pandey et al (2000b) made similar observationsfor eastern India. Such shortfalls in consumption
point to the inadequacy of consumption-
smoothing mechanisms, especially among small
farmers.
Livestock, in addition to being useful for
agricultural production, are also an important
store of wealth in rural societies. They serve an
important role in consumption smoothing.
During drought years, livestock are sold and
proceeds are used to meet a consumption
shortfall. Disposal of livestock can also help
reduce carrying costs, which tend to be high,
especially during drought years (Kinsey et al
1998). In the Sahel zone of Africa, where poor
environmental conditions constrain the efficacy
of ex ante mechanisms, manipulation of
livestock inventory is an important ex post
mechanism (Matlon 1991). Farmers in India
similarly use the livestock inventory to reduce
consumption shortfall (Jodha 1978).
A problem with the use of livestock for
consumption smoothing is that this coping
mechanism, while helping farmers to survive
during drought years, can reduce the long-term
production potential. Where livestock are simply
a store of wealth, this will not create a problem.Disposing of livestock in this case would be
similar to withdrawing cash from the bank. In
fact, disposal of small animals such as goats and
sheep, which tend to be good stores of value, is
generally the initial response to income
shortfalls. However, livestock are also the major
source of draft power needed for several farm
operations such as tillage, pumping irrigation
water, threshing rice, and hauling farm inputs
and outputs. Faced with the prospect of a severe
shortage in consumption in a severe or prolonged drought, farmers may sell productive
livestock such as cattle, buffaloes, and horses.
Once these productive livestock assets are
depleted, it takes a long time for them to be
replenished. Thus, even after the drought is over
and rainfall returns to normal, it may take
several years for farmers to rebuild their stock of
livestock. A typical feature of the livestock
depletion-replenishment cycle is that livestock
are sold when their prices are falling due to
excess supply during drought years (Jodha1978). Increased demand during the
replenishment phase pushes the prices up,
making it more difficult for farmers to reacquire
the livestock. If several drought years occur in a
row, the livestock asset may be depleted so
severely that several years of normal conditions
would be needed for full replenishment of the
livestock. The effect of drought can thus linger
on for several years until productive assets are
fully replaced. As the mortality of livestock is
higher in drought years due to poor nutrition, the
asset base can be depleted dramatically during a
run of drought years. Thus, this coping
mechanism could be costly in terms of future
production potential forgone. The impact is
likely to be greater for small farmers than for
large farmers as small farmers often need a
longer time to replenish the depleted stock.
As with the depletion of livestock, severe
droughts can lead to excessive exploitation of
common property resources (CPR) that are a
9
8/8/2019 Risk Analysis and Management in Rainfed Rice Systems
24/213
critical component of village livelihood systems
(Jodha 1986). The CPR are resources owned in
common by village residents. These include
community forests, pasture/waste land, ponds,
river banks and river beds, and groundwater. The
poorer segments of the rural population areespecially dependent on CPR, even in normal
times, to generate food, fiber, and income.
During drought periods, these resources become
even more important. For example, the reduced
supply of fodder during drought years increases
the reliance on forest and community grazing
areas for sustaining the livestock. Similarly,
additional incomes are generated by selling
timber, fuelwood, and other forest products.
Collection of edible forest products such as
fruits, nuts, and bamboo shoots also increases asfarmers attempt to meet the shortfall in
production. If these CPR are depleted
excessively during drought years, the
productivity of agriculture and livelihood of the
poor can be adversely affected for many years
even after the meteorological drought ends.
Short-term or permanent migration to earn
income from cities or far-away places is another
coping mechanism. Migration to nearby places
is likely to be less effective due to covariate
movements in income within a small geographic
area. Prospects for earning income within the
locality affected by drought are limited due to a
reduction in demand for labor in the agricultural
as well as nonagricultural sector. Employment in
far-away places or in sectors unlikely to be
affected by drought will have a stabilizing effect
as such income is less covariate with income in
drought-affected areas. In addition to seasonal
migration during drought periods, diversification
of earning with some family members working
permanently in cities helps smooth consumption.
A variant of this coping mechanism is the
marital relationship with families in far-away
places. Income transfers through this mechanism
have helped farmers in the semiarid tropics of
India to stabilize consumption during drought
years (Rosenzweig and Stark 1989). Similarly,
diversification of income from the farm to
nonfarm sector is a way of exploiting the low
covariance for income and consumption
stabilization. For example, the proportion of
10
income derived from nonfarm employment
outside the region has been higher in the riskier
Sahel zone than in the less risky Sudan zone of
Africa (Matlon 1991).
in smoothing consumption. Credit permits borrowing against future income potential to
meet a current consumption shortfall. In a
perfectly competitive market, the opportunity
cost of credit is equal to the interest on savings.
Hence, long-run consumption will not depend on
whether savings are used or credit is taken to
meet a shortfall in consumption in poor years. In
reality, credit markets are imperfect, with the
effective interest rate on credit being higher than
the interest on savings. Risk aversion among
lenders, the high transaction cost of serving alarge number of small farmers, and information
asymmetry between borrowers and lenders are
the major reasons for capital market failure in
developing countries (Binswanger and
Rosenzweig 1986). As a result, the use of credit
for consumption smoothing in developing
countries is limited, more so among small
farmers who are considered as high-risk
borrowers by formal credit institutions.
Despite a poorly developed formal market
for credit, the available evidence on the extent of
consumption smoothing indicates the presence
of informal institutional arrangements for risk
sharing in rural areas. These may be village-
level rice banks, local money lenders, mutual
self-help groups, interlocked credit and labor
markets, and social and family networks.
Income transfers (in cash or in kind) through
these informal arrangements can provide very
effective insurance, especially if the risk affects
only a few households (Jodha 1978, Ben-Porath
1980, Platteau 1991, Fafchamps 1992,
Townsend 1995). The provision of suchinsurance is believed to be one of the critical
functions of the family as an institution
(Rosenzweig 1988). Although very effective in
insuring poor households against a consumption
shortfall caused by life-cycle events such as
death or illness in the family, these mechanisms
are less effective in dealing with covariate risks
that affect everybody within the community.
Historical records of mass migration, starvation,
Credit can potentially play an important role
8/8/2019 Risk Analysis and Management in Rainfed Rice Systems
25/213
and death attest to the failure of these informal
mechanisms when droughts are severe and
widespread. These informal arrangements that
characterize traditional rural societies also seem
to weaken considerably in the face of
commercialization and greater exposure to the
outside world (Jodha 1978).
Publicly sponsored relief programs are used
to deal with the failure of these ex post
consumption-smoothing mechanisms in the face
of large covariate risk. To the extent that food
insecurity is due to the lack of exchange
entitlements, these relief programs are designed
to transfer income to farmers in affected areas to
reduce consumption deficits and prevent
excessive asset depletion. The relief programs
generally take the form of income transfer/employment generation although direct food
distribution may also be a component when
drought is severe. Several authors (Corbett 1988,
Hay 1988, Dev 1996) have discussed the
strengths and weaknesses of various types of
relief programs.
Methods for risk analysis
One of the most widely applied models for
studying decision making under uncertainty isthe expected utility model (Anderson et a1
1977). Under risky situations, decision makers
are assumed to select options that maximize the
expected utility of probabilistic consequences.
To implement the model, it is essential to know
the decision makers attitudes toward risk and
the probability of various outcomes resulting
from an action.
Attitudes toward risk are captured in the
utility function that transforms monetary gains
and losses into utility. Risk analysis consists of
combining the subjective probability of
outcomes and the associated utility to identify
risk-efficient decisions.
Different methods are available for
estimating the utility function and the implied
risk aversion coefficient (Binswanger 1980,
Binswanger and Sillers 1983,Antle 1983). Two
popular specifications used in applied work are
the utility function with constant partial risk
aversion coefficient and the utility function with
constant absolute risk aversion coefficient.
Estimates of both types of risk aversion
coefficients have been derived for several
farming systems.
derived in at least two ways: using historical
data and using a predictive simulation model.
Both approaches have advantages and
disadvantages. While the use of historical data is
based on the assumption that the future will be
similar to the past, the use of a predictive
simulation model requires that the model
adequately mimic the real production system.
The use of a simulation model to predict the
consequences of changes in technical parameters
is becoming more popular (Muchow and
Bellamy 1991, Lansigan et a1 1997). Usingstochastic weather input to drive a suitably
validated simulation model, the probability
distribution of yield for a specific technical
intervention can be obtained. The distribution of
yield can then be transformed into the
distribution of economic variable (for example,
profit), which is then used for economic risk
analysis. Anderson and Hazell (I 994, Lansigan
et a1 (1997) have discussed the advantages and
disadvantages of using simulation models to
identify risk-efficient technologies.Once the utility function and probability
distribution of income are obtained, several
approaches could be used to identify risk-
efficient decisions. A popular integrative
approach is the use of whole-farm planning
models. Variants ranging from simple stochastic
budgets to discrete stochastic programming
models are available (Hardaker et a1 1991).
Other approaches include stochastic dominance
analysis, nonoptimizing simulation models, and
variants thereof.
The required probability distributions can be
Objective function specification
The consequences of an action must be assessed
in relation to the objective function or what
farmers would like to achieve from their fanning
activities. While the objective function may
include aspects such as quality of life, childrens
education, and level of leisure, analysts often
focus on economic criteria such as farm income
11
8/8/2019 Risk Analysis and Management in Rainfed Rice Systems
26/213
and consumption. If the objective of a farmer is
to maintain a given level of consumption, the
utility function should be defined in terms of
consumption. However, as income and
consumption are highly correlated, income is
often used as a proxy for consumption. Farmers
derive income not only from rice but also from
several activities that include growing other
crops and nonfarm employment. Income from
rice is often a smaller component of their total
income, especially in the more marginal
environments. Even if rice production is low,
farmers may be able to maintain their
consumption level by obtaining additional
income from other sources. Thus, it is not
enough to evaluate rice technologies in terms of
how much additional income they can generatefrom rice. Income from all sources should be
considered simultaneously.
Risk benefit and rice research
What opportunities exist for rice research to
reduce fluctuations in income and consumption
of farmers? What is the size of the economic
benefit if rice yield and production could be
stabilized? Answers to these questions are
critical for designing suitable technological and policy interventions to reduce &he cost of risk.
define what we mean by cost of risk and
develop a device to measure it quantitatively.
For this, we use the expected utility model of
decision making. The model postulates that,
under risky situations, decision makers evaluate
decisions in terms of expected utility of income
and choose the action that maximizes the
expected utility. For risk-neutral decision
makers, the decision that maximizes the
expected utility is also the decision that
maximizes the expected income gain. A risk-
averse decision maker, on the other hand, would
be willing to sacrifice some income to avoid
taking risk. The cost of risk is the amount of
income sacrificed to protect or insure against
risk. Using the expected utility theory, the cost
of risk can be approximated as (Pandey et al
1999).
Before proceeding further, it is essential to
P= 0.5R [a2 Cr2 + 2 a (1 - a) g Cr Cy]
12
where P is the cost of risk (or risk deduction)
expressed as a proportion of mean income, R is
the coefficient of relative risk aversion, Cr is the
coefficient of variation (CV) of rice income, a is
the share of rice income in total income, Cy is
the CV of nonrice income, and g is the
correlation coefficient between rice and nonrice
income. The proportional risk premium
measured in this equation provides an estimate
of the cost of risk currently borne by farmers
relative to the situation in which the variability
of rice income is completely eliminated. As
there will always be some variability of rice
income that cannot be eliminated, the estimate
obtained from this equation can be considered an
upper bound value.
This indicates several ways through whichthe economic cost of risk can be reduced:
lowering the CV of income from rice, lowering
the ratio of rice income to nonrice income, and
reducing the correlation of rice income with
nonrice income. Stabilization of rice yield
through breeding and better crop management
can be an important research intervention. The
lowering of the share of rice to nonrice income
implies crop and income diversification. The
scope of technical intervention to achieve this
may be somewhat limited in the case of rainfedrice, as waterlogged conditions of the fields limit
other cropping options during the rainy season.
Farmers have other cropping alternatives only
during the postrainy season, provided moisture
is nonlimiting. Development of shorter duration
varieties in areas where the success of a
postrainy- season crop depends on how early it
is established can facilitate diversification of
crop income. Other options for encouraging crop
and income diversification are related to policy
interventions such as the development of road,
transport, and marketing infrastructure. These
policy interventions can also help reduce the
correlation between rice and nonrice income by
broadening the income base of rural households.
Technology and yield risk
Technical research can basically be classified
into two types: plant improvement and crop
management. Two risk-related issues involve
8/8/2019 Risk Analysis and Management in Rainfed Rice Systems
27/213
plant improvement research. The first is the
issue of the extent to which improved varieties
are more or less risky than traditional varieties.
Plant breeders use stability analysis and other
forms of genotype x environment (G x E)analysis to assess the stability and adaptability of
alternative varieties versus the traditional check.
The analysis of G x E interactions has been a
topic of interest among plant breeders and
powerful tools and methods have been
developed (Cooper and Hammer 1996).
However, most of these analyses use some
statistical notion of stability for discriminating
among cultivars. Such analysis could be usefully
complemented by decision analytical tools such
as stochastic dominance analysis to explicitly
account for farmers risk aversion (Binswanger
and Barah 1980, Witcombe 1989).
The second is the issue of the extent to
which a combination of several varieties reduces
risk. Ample evidence shows that farmers grow
several varieties of rice simultaneously in
rainfed areas (Kshirsagar and Pandey 1995,
Pandey and Sanamongkhoun 1998). Although
there may be several reasons for doing so, risk
reduction through varietal diversification
appears to be an important one (Smale et al
1994). The strategy of varietal diversification
could potentially be used to reduce the overall
risk, even if modem varieties are less stable than
their traditional counterparts.
Crop management research, on the other
hand, is concerned with altering yield risk by
manipulating agronomic practices. Agronomic
manipulation can reduce the yield risk
associated with stress conditions such as
drought, flood, and pests. For example,
improved nutrient management may help reduce
risk by making plants more tolerant of stresssuch as drought as well as by helping them to
recover faster when the stress is relieved (Wade
et al 1999). Similarly, options may exist for
reducing risk by manipulating timing, placement
and quantities of inputs.
The study of the quantitative effects of input
management on risk remains a major field of
inquiry by agricultural economists, among
others. Production function specifications that
permit estimation of marginal risk effects have
been developed (Just and Pope 1979, Antle
1983). Such production functions have been
applied to derive the optimal allocation of
several inputs under risky situations. Although
attempts have been made to quantify marginal
risk effects of several inputs such as fertilizers,
irrigation, and pesticides in a range of
environments using such a framework, the
empirical analyses have often produced
somewhat inconsistent results (Roumasset et al
1989, Pandey 1989).
An important area of research in the context
of crop management technology is the effect of
uncertainty on input use in a dynamic context.
Instead of committing all inputs at the beginning
of the crop season, inputs are used sequentially,
with farmers revising the level of input use
depending on crop conditions and their
expectations regarding stochastic variables such
as prices and weather. Possibilities for such
dynamic adjustments of input use provide
flexibility for efficient risk management. In
addition, reliable forecasts of stochastic
variables such as weather can improve the
efficiency of resource allocation by reducing the
level of uncertainty (Byerlee and Anderson
1982, Abedullah and Pandey 2000).
Data needs
Farmers are concerned about risk that manifests
itself in the form of unpredictable fluctuations in
yield over time. To analyze risk and risk-coping
mechanisms, temporal data are hence required.
The generation of temporal data, however, is
expensive and time-consuming. Plant breeders
have partially got around this constraint in their
selection program by including several testing
locations to capture different environmentalconditions. Fortunately, this approach has
worked well in the past. However, this kind of
spatial substitute for temporal data is less useful
when analyzing farmers coping mechanisms
and in studying how risk influences resource
allocation over time through its effect on assets
of farm households. Spatial data are not of much
help in studying these dynamic elements. The
only extensive panel data that have been used
widely to study risk and many other aspects of
13
8/8/2019 Risk Analysis and Management in Rainfed Rice Systems
28/213
the village economy is the village-level study
database generated by ICRISAT (Walker and
Ryan 1990). A similar kind of database covering
major cropping systems would certainly be very
useful for studying responses to risk in other
rainfed environments. As indicated in the paper
by Pandey et a1 (2000a), some progress is being
made in this direction.
Strategies for developing anddisseminating risk-reducingtechnologies
What are the implications of the above
discussion for developing and disseminating
risk-reducing technologies? Are the strategies
and institutional mechanisms likely to bedifferent from those now in place? Space
limitations preclude me from going into much
in-depth discussion on this topic, which by itself,
is very broad. Nevertheless, some comments on
this important topic are in order. Based on the
above discussion, it can be deduced that the
following features of technology help reduce
risk:
less input demanding;
lower degree of prior commitments of
inputs; technologies that improve flexibility of
decision making;
technologies that use information about
conditioning factors;
technologies that help stabilize area, not just
yield, as area variability can be an important
source of production variability; and
technologies that raise income by improving
the productivity of other components of the
farming systems (e.g., those that facilitate
crop diversification and intensification).
These considerations suggest the following
strategies for technology development,
adaptation, and dissemination for reducing risk:
Emphasis on technologies that reduce yield
losses in unfavorable years rather than those
that increase yield in favorable years only
(downside risk). However, trade-off between
yield gain and instability may be inevitable.
More emphasis on developing durable
14
resistance to pests/diseases and abiotic
stresses. Molecular techniques may have a
major role to play, especially when dealing
with polygenic traits.
complementary options in which each
component can also stand on its own.
Although productivity improvement can be
high when several components are
combined in the form of a package, such
packages also tend to increase risk. By
allowing farmers to pick and choose from a
complementary basket of options, such an
approach makes sequential adoption of the
most profitable (and least risky) components
possible.
More adaptive research and decentralized
regional testing for specific adaptation.
Specific technologies for each region
developed through adaptive research will
improve the suitability of such technologies
to their target domain and reduce risk. T