African Journal of Economic and Management Studies
Smallholder farmers’ risk perceptions and risk management responses: Evidence
from the semi-arid region of Ghana
Richard Asravor,
Article information:
To cite this document:
Richard Asravor, (2018) "Smallholder farmers’ risk perceptions and risk management responses:
Evidence from the semi-arid region of Ghana", African Journal of Economic and Management
Studies, Vol. 9 Issue: 3, pp.367-387, https://doi.org/10.1108/AJEMS-10-2017-0250
Permanent link to this document:
https://doi.org/10.1108/AJEMS-10-2017-0250
Downloaded on: 27 June 2019, At: 08:06 (PT)
References: this document contains references to 37 other documents.
To copy this document: permissions@emeraldinsight.com
The fulltext of this document has been downloaded 138 times since 2018*
Users who downloaded this article also downloaded:
(2018),"Does FDI moderate the role of public R&D in accelerating agricultural production
in Africa?", African Journal of Economic and Management Studies, Vol. 9 Iss 3 pp. 290-304 <a
href="https://doi.org/10.1108/AJEMS-07-2017-0153">https://doi.org/10.1108/AJEMS-07-2017-0153</
a>
(2018),"The nexus between external debt, corruption and economic growth: evidence from five
SSA countries", African Journal of Economic and Management Studies, Vol. 9 Iss 3 pp. 319-334 <a
href="https://doi.org/10.1108/AJEMS-07-2017-0171">https://doi.org/10.1108/AJEMS-07-2017-0171</
a>
Access to this document was granted through an Emerald subscription provided by emerald-
srm:534301 []
For Authors
If you would like to write for this, or any other Emerald publication, then please use our Emerald
for Authors service information about how to choose which publication to write for and submission
guidelines are available for all. Please visit www.emeraldinsight.com/authors for more information.
About Emerald www.emeraldinsight.com
Emerald is a global publisher linking research and practice to the benefit of society. The company
manages a portfolio of more than 290 journals and over 2,350 books and book series volumes, as
well as providing an extensive range of online products and additional customer resources and
services.
Emerald is both COUNTER 4 and TRANSFER compliant. The organization is a partner of the
Committee on Publication Ethics (COPE) and also works with Portico and the LOCKSS initiative for
digital archive preservation.
*Related content and download information correct at time of download.
Downloaded by University of Ghana At 08:06 27 June 2019 (PT)
The current issue and full text archive of this journal is available on Emerald Insight at:
www.emeraldinsight.com/2040-0705.htm
Smallholder farmers’ risk Farmers’ riskperceptions
perceptions and risk and risk
management responses management
Evidence from the semi-arid region of Ghana 367
Richard Asravor Received 9 October 2017
Revised 2 February 2018
Department of Agricultural Economics and Agribusiness, 19 March 2018
College of Agriculture and Consumer Sciences, Accepted 29 April 2018
University of Ghana, Legon, Ghana
Abstract
Purpose – The purpose of this paper is to identify the perceptions of farmers on the major sources of risk and
to examine the effectiveness of the risk management responses of rural smallholder farm households in the
semi-arid region of Northern Ghana from the socioeconomic perspective.
Design/methodology/approach – Both descriptive statistics and exploratory factor analysis were used on
a Likert scale question to rank and identify the important risk perceptions and management strategies of the
farmers. The linear regression model was used to highlight the significant factors that affect the farmers’ risk
perception and management responses.
Findings – The effects of the variations in crop yield, fertiliser prices and crop price on household income
were perceived as the three most relevant sources of risk. Stabilising household income by growing different
crops, storing feed/seed reserves and spreading sales were the most effective risk management strategies.
Factor analysis identified market risk, production risk and human risk as major risk factors whereas
diversification, financial strategy, and off-farm employment were perceived as the most effective risk
management strategies. Farm and farmer characteristics were found to be significantly associated with risk
perceptions and risk management strategies. Risk perceptions significantly increase the risk management
strategy adopted by the smallholder rural farmers.
Practical implications – The findings of the paper call for the integration of farmers’ risk perceptions and
management strategies in the development of agricultural policies for the semi-arid regions of Ghana.
Originality/value – This paper deviates from the traditional technology adoption studies by modelling rural
household perceptions and management strategies using, using descriptive, factor analyses, and linear regression.
Keywords Risk management, Factor analysis, Risk perception, Semi-arid region
Paper type Research paper
1. Introduction
Undeniably, the agriculture sector of most countries faces more risk than the industrial and
service sectors (Maurer, 2014; Deressa et al., 2009). In developing countries, the performance
of the agriculture sector is generally uncertain (Aditto et al., 2012; Di Falco and Veronesi,
2014), due to the exposure of the sector to a variety of risks and biotic and abiotic factors.
Risk and uncertainty do not only impact households’ production decisions but also their
consumption and welfare decisions. Though numerous normative analyses have been done
on how farmers should behave under risk and uncertainties (Canagarajaha et al., 2001;
Bergfjord, 2009), less empirical studies have been conducted on how farmers perceive these
risks and how these perceptions influence their management strategies (Meraner and
Finger, 2017; Bishu et al., 2016).
According to Flaten et al. (2005), farmers’ perceptions of and responses to risks are vital
towards understanding their risk behaviour. The risk perceptions of farmers influence their
risk-taking behaviour (Koesling et al., 2004). Hence a comprehension of farmers risk African Journal of Economic and’ Management Studies
perceptions is important in helping farmers make better and informed decisions relating to Vol. 9 No. 3, 2018pp. 367-387
the risky agriculture businesses. Added to this, an understanding of farmers’ risk © Emerald Publishing Limited
2040-0705
perceptions and risk management strategies is essential for the proper formulation of DOI 10.1108/AJEMS-10-2017-0250
Downloaded by University of Ghana At 08:06 27 June 2019 (PT)
AJEMS appropriate policy to help overcome the perennial risks associated with the farming
9,3 business. Most empirical studies on risk perceptions and management responses of farmers
have either focussed on livestock farmers (Bishu et al., 2016; Flaten et al., 2005; Meuwissen
et al., 2001) or aquaculture (Bergfjord, 2009; Ahsan and Roth, 2010). Little studies have been
conducted on rainfall dependent smallholder crop farmers (Aditto et al., 2012; Koesling et al.,
2004) in developing countries, though these groups of farmers dominate the agricultural
368 sector of developing countries and contribute significantly towards the GDP of their
respective countries.
For instance, in Ghana, agriculture contributed 21.5 per cent to the nation’s GDP in 2014
(MoFA, 2015) and it is the mainstay of the rural Ghanaian economy (GSS, 2014). In many
rural communities of Ghanaian rain-dependent crop farming is the most important and only
source of employment and household income (MoFA, 2015). Within the rural settings of
Ghana, the semi-arid region of Ghana has almost 90 per cent of its rural population engaged
in crop farming (GSS, 2014). It is noted to be the hub of legumes and cereal production, and
fundamentally covers about 40 per cent of the total agricultural land of the country. The
semi-arid region of Ghana is among the poorest regions, the hardest hit by climate related
production risk (MoFA, 2012) and the most food insecure region.
Like other agri-businesses, crop farming in the rural semi-arid communities is
associated with production, marketing, human, institutional and financial risks. Most
of these risks are unknown to these rain-dependent crop farmers before the farming
season but are anticipated based on their perceptions and past experiences. These
perceptions influence the perceived risk management strategies and the frantic effort
made to manage risk. As reiterated by Gebreegziabher and Tadesse (2014) and
Legesse and Drake (2005), farmers make decisions based on what they think is likely to
occur, and sometimes based on what they fear or hope is possible. Thus, the actual risk
management strategy of a farmer begins with the perception of risk and the perceived risk
management strategy.
In order to understand the risk behaviour of rural households, it is essential to
investigate smallholder farmers’ perceptions and responses to risks. As indicated by Ahsan
and Roth (2010) and vanWinsen et al. (2016), an individual’s perception of risk influences his
or her investment and business decisions on the farm. Despite the importance of risk
perceptions and management strategies to the farming decisions of rural rain-fed
smallholder in the semi-arid regions of Ghana, empirical studies are lacking. To the very
best of my knowledge, no empirical studies have been explicitly conducted on the risk
perceptions and risk management strategies of smallholder farmers in Ghana and more
specifically the risk prone area of rural semi-arid regions of Ghana. An understanding of the
risk perception and risk management strategies of smallholder rain-dependent crop farmers
are essential preconditions for devising risk reducing strategies (Legesse and Drake, 2005;
Hardaker et al., 2004). Therefore, ignoring the risk perceptions of smallholder farmers can
lead to a significant overestimation of the input demand and the output supply elasticities
(Meraner and Finger, 2017; Menapace et al., 2016).
Based on this gap and the relevance of risk perceptions and management strategies to
smallholder households, the paper contributes to the extant empirical literature by
providing empirical insight into, farmers’ perceptions of the most important sources of risk
and risk management through the use of an exploratory and descriptive analysis; and
examine the interrelationship between farm and farmer characteristics, and household risk
perceptions and risk management strategies, using the linear regression model. The
succeeding sections focus on the extant literature on the risk perception and risk
management strategies (Section 2), the material and methods (Section 3), analysis and
methods of estimation (Section 4), results and discussions (Section 5), and the summary,
conclusion and policy recommendations (Section 6).
Downloaded by University of Ghana At 08:06 27 June 2019 (PT)
2. Literature review Farmers’ risk
Several theories have been adopted to explain risk perception and the subsequent decisions perceptions
taken to manage these perceived risks. Common theories relating to risk perceptions and and risk
management strategies are the protection motivation theory (PMT), risk compensation/risk
homeostasis theory and the situated rationality theory. The PMT is one of the most cited management
theories in the academic literature. The main arguments of this theory is that individuals are
very likely to protect themselves from any form of risks when they anticipate adverse 369
consequences, have the yearning of avoiding these negative consequences and are capable
of taking preventive measures. Thus, the perception of risk and the management of risk
increases when there is a reason for concern (Gucer et al., 2003; DeJoy, 1996). On the other
hand, the risk compensation/risk homeostasis theory indicates that individuals take risk
when they feel a greater sense of security. Consequently, safety measures or the
management strategies adopted depends on the risk perceptions and risk-taking behaviour
of such individuals (Wilde, 1994). The situated rationality theory also suggests that it is a
flaw to assume that safe behaviours are fundamentally rational and high-risk behaviours
are intrinsically irrational. In other words, there is a possible rational explanation for
choosing to take risks that is more expounding than assuming that a risk-taker is simply
thrill-seeking (Cafri et al., 2009).
In this study, the researcher assumes that anticipated risk influences the management
strategies adopted by the farm household, hence the PMT theory is adopted by this study.
Empirically, the issue of risk perceptions and management strategies have been shown to
have strong industry and regional connotations (Flaten et al., 2005; Legesse and Drake,
2005). Other studies have identified the farm type, the operational environments of farmers,
the geographical locations, institutional structures as prime determinants of risk
perceptions and risk management strategies of farmers (Meuwissen et al., 2001;
Bishu et al., 2016).
Numerous studies exist in the developed countries that have examined the risk perception
and management strategies of farmers. For instance in the mid-1990s, Martin (1996) opined
that price risk and rainfall variability was ranked highest by dairy farmers in New Zealand,
and routing spreading, drenching and maintaining feed reserves were the strategies used to
manage these risk. Dutch livestock farmers ranked price and production risks as the two most
important risk sources met by the production at lowest possible costs and insurance
(Meuwissen et al., 2001). According to Sonkkila (2002), Finnish farmers considered changes in
agricultural policy as the most important risk factor, whereas maintaining enough liquidity
and solidity was ranked as the most important management response.
Ahsan and Roth (2010) empirically investigated the risk perception and management
strategies of mussel farmers in Denmark using the exploratory factor analysis. Future prices,
changes in public regulations and demand for mussels were highly-ranked as the perceived
risks while bad weather, harmful algal blooms, oxygen depletion, E. coli, public views towards
mussel culture and change in government regulations were considered as important risk
factors in mussel farming. Producing at the lowest possible cost, maintaining good relations
with the government, cooperative marketing, liquidity, experience sharing and adoption of
new technology were perceived as the most important risk management strategies.
According to Le and Cheong (2010), Vietnamese catfish farmers perceive price and
production risks as the most significant risks while farm management and technical
measures were perceived to be the most effective risk management strategies for risk
reduction. Despite price risk being top rated, price risk management strategies were not
perceived as important measures for risk mitigation. Meraner and Finger (2017) found that
higher level of risk perception, age, subjective numeracy, farm succession, farm size and the
proportion of rented land significantly impact the risk management strategies of livestock
farmers’ in the German region of North Rhine-Westphalia.
Downloaded by University of Ghana At 08:06 27 June 2019 (PT)
AJEMS In Sub-Saharan Africa, quite a number of studies have been conducted on risk
9,3 perceptions and risk management strategies with each yielding varied results. A bulk of
these studies have been conducted in Ethiopia and on livestock farming (Bishu et al., 2016;
Gebreegziabher and Tadesse, 2014; Legesse and Drake, 2005). Legesse and Drake (2005)
found human capital, household characteristics, orientation, infrastructure, access to
resources, information and environmental factors to be influential in shaping the
370 perceptions of risks among smallholder farmers in Eastern Highlands of Ethiopia.
Additionally, the authors found from the logistics analysis that locational settings,
livelihood diversification strategies and asset endowments were important in determining
the perceived risk management strategies of these smallholder farmers.
Bishu et al.’s (2016) analysis of cattle farmers in the Tigray area of Northern Ethiopia
showed that the shortage of family labour, the high price of fodder and limited farm income
were perceived as the most important risks. The use of veterinary services, parasite control
and loan utilisation were perceived as the most important strategies for managing risks.
Gebreegziabher and Tadesse (2014) also focussed on livestock farmers in the Tigray area of
Northern Ethiopia. From the factor analysis, the researchers found technological,
production, financial, price/market, human and institutional factor were the major sources
of risks whereas diversification, disease reduction, financial management and market
network are perceived as the most effective risk management strategies in Northern
Ethiopia. Surprisingly, most dairy farmers considered themselves risk takers towards farm
decision and hence have a positive impact on technology adoption.
In their study of the risk perception and risk management strategies of 200 smallholder
farmers in the KwaZulu-Natal Province of South Africa, Kisaka-Lwayo and Ajuruchkwu
(2012) found that production, price or market, and financial risks were perceived as the most
important sources of risk by these smallholder farmers. The most important traditional risk
management strategies used by these farmers were crop diversification, precautionary
savings and participating in social networks.
According to Aditto et al. (2012), the increasingly worrying sources of risk to Thai
farmers are the uncertainty of product prices, input prices (market risk), diseases, pests
affecting plants and animals, natural disasters and excess rainfall (production risks). The
Thai farmers ranked the following as important strategies to manage risk: the storing feed
and/or seed reserves, having a farm reservoir for water supplies in dry the season, holding
cash and easily converted cash assets and working off-farm to supplement household
income. Thus, production , off-farm and financial strategies were viewed as important
strategies than marketing strategies.
3. Materials and methods
3.1 Study area and sampling technique
The research involved face to face interviews with sampled cereals and legumes smallholder
farmers from Northern Ghana. The choice of the semi-arid region of Ghana was because it is
the most vulnerable to climate risk and the poorest regions in Ghana (GSS, 2014). The
semi-arid region of Ghana is made up of the Northern region, Upper East and Upper West
region. It is the regions with the most rain-fed farmers yet the regions with the unimodal
rainfall as well as the least amount of rain (900 mm).
Themultistage sampling technique was used to sample 500 farmers from 12 districts and 5
communities within each of the districts. The first stage involved purposively selecting the
semi-arid regions of Ghana since it is the most risk prone region in Ghana. The second stage
involved using the simple random sampling technique to sample cereals and legumes farmers
in the various communities. To ensure homogeneity and heterogeneity within the sample, the
final stage involved using the clustering sampling technique whereby farmers were put into
three clusters, that is, the upper, the middle and the lower cereals and legume cluster.
Downloaded by University of Ghana At 08:06 27 June 2019 (PT)
3.2 Data description Farmers’ risk
Household and plot level data on the various farmer and farm characteristics as well as the perceptions
management and agronomic practices were collected as part of the larger questionnaire on and risk
the climate change adaptation strategies of smallholder farmers in Northern Ghana. As part
of the main questionnaire, a five-page questionnaire on farmers’ perceptions of risk and risk management
management strategies was added. Almost all the questions were of the closed-ended type
and were in the form of a five-point Likert scale. The 14 questions were asked on the 371
perceptions of the sources of risk, 7 of the questions focussed on the variability associated
with the various forms of risk identified and the remaining 7 was on the effect of the
variability on household income. The scoring of the Likert scale was from 1(no variation/no
effect) to 5 (very large variation/very large effect). Likewise, 24 questions were asked on the
various management strategies. The first 12 questions were on the effort it took to manage
these risks and the remaining 12 asked questions on the income stabilisation effect of these
effort. These 24 questions were scored using the Likert scale from 1(no effort/no effect) to 5
(very large effort/very large effect). The questionnaire was administered between June 2015
and September 2015 to the smallholder farmers in the semi-arid regions of Ghana.
4. Analysis
4.1 Method of estimation
Three estimation methods were applied in this study. The first was the use of the descriptive
statistics to rank the various sources of risk and risk management strategies (Flaten et al., 2005;
Bishu et al., 2016). This was done to identify the most important risk perceptions and strategies
of the rural households in the semi-arid regions of Ghana. The second estimation was the
conduct of the exploratory factor analysis. The exploratory factor analysis was employed to
identify the major sources of perceived risk factors and the effective risk management strategies
(Ahsan and Roth, 2010; Gebreegziabher and Tadesse, 2014). Exploratory factor analysis helped
in the generation of factor scores. The generated factor scores measure the deviation of an
individual’s score from the sample mean (Kim and Mueller, 1978). Factor scores are unique
composite variables which provide reliable information about an individual’s placement on the
factor. It aids in the separate allotment of scores to each latent variable based on the estimates.
The third stage involved the use of the standardised factor scores to run the linear
regression (Aditto et al., 2012; Bishu et al., 2016). The Probit model and its marginal effect
were also estimated using the direct scores given by the respondents as the dependent. In
order to use the direct scores from the responses gathered from the Likert scale it was
recoded. The direct response was recoded to 0 where the response was 1(no variation/no
effect) and all other response was recoded 1. This made the dependent variables a dummy
variable, hence the application of the limited dependent estimation technique. The detailed
diagnostic checks conducted before the performance of the exploratory factor analysis is
reported in the next subsections.
4.2 Diagnostics
Before conducting the factor analysis, the diagnostic test was carried out to assess the
adequacy of the data set. The results of these diagnostic tests are reported in Tables II and
IV. The overall KMO value of 0.86 for the perceived risk sources and 0.94 for the risk
management strategies indicate that the pattern of correlations was comparatively compact
and factor analysis was suitable (Ahsan, 2011). Additionally, the Bartlett test of Sphericity
was significant at the 1 per cent for both the perceived risk sources (χ2¼ 5,446.42, po0.001)
and risk management strategies (χ2¼ 10,230.33, po0.001), implying that the data come
from a multivariate normal distribution with zero covariance. The overall Cronbach’s α
values of 0.91 and 0.96 for the perceived risk sources and risk management strategies,
Downloaded by University of Ghana At 08:06 27 June 2019 (PT)
AJEMS respectively, were above the minimum required level of 0.6 hence was acceptable (Hair et al.,
9,3 2006; Field, 2005). Subsequent, the eigenvalues were computed and the scree plot generated
(see appendix 1 for the scree plot). The eigenvalues (eigenvalueW1) were detailed for three
factors out of the 14 sources of risk variables and 22 risk management variables for the farm
household in the semi-arid region of Ghana. It is important to state here that the initial
analysis using the Orthogonal (varimax) rotation was employed to ensure that the factors
372 were independent and could be used for the linear regression (Aditto et al., 2012). In
accordance with the empirical literature, factor loadings greater than 0.4 were retained and
interpreted (Bishu et al., 2016; Gebreegziabher and Tadesse, 2014; Legesse and Drake, 2005).
The value was chosen because it was considered to be above the minimum level required for
the interpretation of the structure (Hair et al., 2010; Bishu et al., 2016). Before conducting the
linear regression, the variance inflation factor was used to check for multicollinearity
(see appendix 2) while the Breusch-Pagan test (bptest) was used to check for
heteroscedasticity. The variance inflation factor indicated that there was no
multicollinearity among the predictor variables since all the predictor variables were
below 2, whilst the heteroscedasticity test using the Breusch-Pagan test (bptest) indicated
that the model was not homoscedastic at 1 per cent significant level. The robust standard
errors were, therefore, reported for such models since it helped avoid the heteroscedasticity
problems (Cameron and Trivedi, 2005).
5. Results and discussion
5.1 Farmer and farm characteristics
Table I shows that the semi-arid region of Ghana is a patriarchy region (96.2 per cent) with
older household heads (aged 48 years and above). Added to the above, majority of the
household heads hardly completed the nine years compulsory basic to junior high school.
Furthermore, the result confirms that farmers in the semi-arid region of Ghana are
subsistence farm households cultivating an average farm size of 3.21 hectare. The average
family size of 6.36 members confirmed the national statistics for the semi-arid region of
Ghana (Ghana Statistical Service, 2014). On average 31.2 per cent of the households in the
semi-arid region engaged in off-farm employment while the average reported annual farm
income (both crop and animal income) was GHc9154 ($2,773.66)[1].
5.2 Perceptions of risk sources
The summarised results of the average scores, standard deviations and the factor loadings of
farmers’ perception of each source of risk are presented in Table II. The results show that
production risk in the form of the effect of the high variation of crop yield on income was
perceived to be the most important sources of risk. This result is not surprising as most rural
households in the semi-arid region of Ghana are into cereals and legumes production.
Variable Measurement Mean SD Min Max
Age of head Years 48.43 14.38 19 90
Gender of head Dummy(1¼Male, 0¼ otherwise 96.2 0.19 0 1
Education head Years 5.71 1.96 1 7
Family size Number of members 6.36 2.12 1 10
Table I. Farm size In hectares 3.21 2.4 0.2 20.29
Distribution of TLU Number of livestock 2.73 9.09 0 176.3
Socioeconomic Off-farm Dummy (1¼ if engage, 0¼Otherwise) 0.312 0.46 0 1
characteristics of Farm income Ghana Cedis (GHc) 9,154 10,743 0 115,403
respondents Source: Field Survey (2015)
Downloaded by University of Ghana At 08:06 27 June 2019 (PT)
Farmers’ risk
perceptions
and risk
management
373
Table II.
Varimax rotated
factor loadings of
perceived sources of
risk for Northern
Ghana
Downloaded by University of Ghana At 08:06 27 June 2019 (PT)
Most important risk factors
Sources of risk Mean SD Rank Market risk Prod. risk Human risk Communality Cronbach’s α KMO
Unexpected variability related to crop yields 3.85 1.1 4 0.71 0.27 −0.08 0.58 0.9 0.86
Unexpected variability related to crop prices 3.65 1.17 6 0.79 0.34 0.05 0.74 0.9 0.91
Unexpected variability related to fertilizer price 3.71 1.13 5 0.69 0.10 0.09 0.5 0.9 0.86
Unexpected variability related crop pests 2.27 1.12 14 −0.08 0.38 0.69 0.62 0.91 0.78
Variability related wages hired Labour 2.65 1.17 12 0.12 0.05 0.83 0.7 0.91 0.74
Variability related to livestock production 3.07 1.29 10 0.29 0.82 0.18 0.78 0.9 0.89
Variability related to livestock price 3.22 1.27 9 0.31 0.79 0.23 0.77 0.9 0.9
Effect of crop yield variability on income 4.14 1.09 1 0.79 0.22 0.09 0.68 0.9 0.89
Effect of crop price variability on income 3.9 1.25 3 0.82 0.28 0.12 0.77 0.9 0.91
Effect of fertilizer price variability on income 3.94 1.22 2 0.71 0.19 0.2 0.58 0.9 0.87
Effect of crop pest attack on income 2.52 1.35 13 −0.01 0.35 0.67 0.56 0.91 0.78
Effect of wages paid to hired labour on income 2.89 1.31 11 0.26 0.04 0.76 0.64 0.91 0.77
Effect of livestock production on income 3.33 1.38 8 0.36 0.77 0.19 0.76 0.9 0.88
Effect of livestock price variability on income 3.39 1.39 7 0.35 0.78 0.21 0.77 0.89 0.9
Eigenvalues 6.53 2.46 1.45
Per cent of total variance explained 28.1 22.3 17.3
Cumulative per cent of variance explained 28.1 50.3 67.6
Overall Cronbach’s coefficient α 0.91
Overall KMO 0.86
Number of variables considered for naming 4 2 2
Notes: n¼ 500. Factor loadings with absolute values greater than 0.4 are in italic
Source: Field Survey (2015)
AJEMS In addition, market risks in the form of the effect of the high variation of crop price and the effect
9,3 of the high variation of fertiliser price on households were ranked as the 2nd and 3rd most
important source of risk, respectively. Since inorganic fertiliser is a major input in the farming,
an increase in the price of inorganic fertiliser is likely to affect smallholder farmers. The finding
on market risk is consistent with previous studies by Aditto et al. (2012), Flaten et al. (2005) and
Gebreegziabher and Tadesse (2014). These researchers have indicated that the variability and
374 effect of input and product prices (market risks) are essential to smallholder farmers hence an
increase in their prices affects the production processes. Table II, further, indicates that the
unexpected variability associated with crop yield and unexpected variability associated with
fertiliser price were ranked as the 4th and 5th most relevant sources of risk, respectively.
Additionally, the three least important perceived sources of risks are the variability of
the wages paid to hired labour, the effect and the variability of crop pest on yields. Although
ranked as the least important sources of risks, their averages were above 2.5. This is an
indication that households in the semi-arid regions viewed these perceived risk sources as
equally important. It is essential to note that most farm families hired labour for their farm
activities due to the communal labour systems in the rural areas of Ghana. In the communal
labour systems, known as the “nnoboa”, farmers take turns to work on each other’s farm. As
indicated in Table II, three factors were selected for interpretation from the factor analysis
conducted on the perception of the sources of risk since their eigenvalues and the scree plots
were greater than one (see appendix). These three factors explained 67.6 per cent of the total
variance which was deemed sufficient and satisfactory (Hair et al., 2006). Additionally, for
easy interpretation, factor loadings with absolute values greater than 0.4 were labelled.
To this end, the obtained factors were grouped named based on the identified names in the
literature. The first factor is named “Market Risk” because of the relatively high loadings of
the crop price and fertiliser price for both the effect and variability of the sources of risk. This
finding is in line with the findings of Aditto et al. (2012) and Bishu et al. (2016). Additionally,
there was a relatively high loading of livestock production variability and livestock
production effect compared to livestock price variability and livestock price effective hence the
second factor is labelled “Production Risk”. The third factor is labelled “Human Risk”, because
of the highest factor loading of the variability of the wage paid to hired labour. Bishu et al.
(2016) labelled this type of risk as labour risk. Similar Gebreegziabher and Tadesse (2014)
found human risk to be the most important in source of risk of dairy farmers Ethiopia.
5.3 Relationship between the perceived risk source and farmer and farm characteristic
Consistent with studies that have used the standardised factor scores and the direct scores
from the responses to the risk perception as dependent variables, the study reports a low
value for the coefficients of determination (R2) and pseudo R2 for both the ordinary least
square regression (OLS) and the probit model of the risk factor models. The low value of the
R2 is an indication of the diversity among the individual farmers in the semi-arid region.
This confirms the numerous literature that have used the standardized factor scores for
linear regression (Aditto et al., 2012; Flaten et al., 2005; Meuwissen et al., 2001). Dependent
variables are made up of the normalised or the standardized factor score and the direct
scores given by the respondents, respectively[2]. The results from the analysis of the linear
regression model presented in Table III shows that the age of the household head was
negative and significantly related to the perception of market risks at 10 per cent level of
significance. Meaning older household heads perceived both market risks (both crop and
livestock) as less important than younger household heads.
Furthermore, the number of livestock owned (TLU) was positive and significantly related
to production risk at 1 per cent level of significance implying households with larger livestock
perceived production risks as more important than farmers owning little livestock. This
finding was consistent with the finding of Bishu et al. (2016) for the Tigray area of Ethiopia.
Downloaded by University of Ghana At 08:06 27 June 2019 (PT)
Farmers’ risk
Market risk Prod. risk Human risk
Coef. p-value Coef. p-value Coef. p-value perceptions
and risk
Intercept −2.08E-01 0.4782 −2.35E-01 0.41344 −5.16E-01 0.0686*
Age −5.28E-02 0.0854* −2.78E-02 0.35599 −5.78E-02 0.051* management
Gender 2.89E-03 0.8984 1.69E-02 0.44694 2.40E-02 0.2726
Education 2.33E-01 0.3015 1.39E-01 0.53028 −1.90E-03 0.9931
Family size 2.96E-02 0.2003 1.01E-02 0.65543 1.98E-02 0.3751 375
Farm size −2.09E-03 0.5283 2.49E-03 0.44374 7.01E-03 0.0286**
TLU 2.07E-03 0.6632 1.22E-02 0.00925*** 7.46E-05 0.987
Off-farm 7.96E-06 0.2341 −6.04E-06 0.35745 1.40E-05 0.0306**
Farm income −8.35E-02 0.3668 −2.12E-01 0.02002** −9.13E-02 0.3071
Adjusted R2 0.01371 0.04221 0.03616
Bptest 20.483 0.008654*** 20.718 0.007936*** 19.693 0.01156** Table III.Multiple regression of
F-statistics 2.678 0.006912*** 2.279 0.02116** 1.65 0.1083 the risk factors and
Notes: *,**,***Statistically significant at 10, 5 and 1 per cent, respectively the socioeconomic
Source: Field Survey (2015) variables
A probable reason for this finding is the open range nature of livestock production
and the poor livestock management strategies of farmers in the semi-arid area of Ghana which
makes livestock production a risky venture. Furthermore, the study found that wealthy
farmers ( farm income) perceived production risk to be less important compared to less
wealthy farmers. This was expected since wealthy farmers in the Northern part of Ghana
usually rely on diverse sources of income, hence are able to face risk relating to their
production activities.
Additionally, the result of Table III shows that the age of the household head was
significant and negatively related to human risk, which implied that younger semi-arid
farmers perceived human risk as more important than older farmers. Also, human risk was
perceived as highly important by farmers who had off-farm work. This was an indication
that farmers who have off-farm work were very apprehensive about the disruptive effect of
human risk on their off-farm income and their overall household income. Households in the
Semi-arid regions with larger farms were highly concerned about the human risk, because
such households had hired more labour to help with their farming activities.
The results from the probit model (Table AIII) and the marginal effect (Table VII) of the risk
sources using the direct responses also shows that households with higher education perceived
market risk as more important than those with lower years of education. Also households with
larger farm size perceived both production and human risk as important while households
having off-farm employment perceive both production and human risk as less important.
5.4 Perceptions of risk management strategies
Smallholder farmers in Northern Ghana encounter various forms of risk and adopt
heterogeneous strategies to manage these risks. Table IV shows that the two most
important risk management strategies used to stabilise household income were the growing
of different crops (4.39) and the storing of feed or seed reserve (3.71). This finding is in
line with that of Aditto et al. (2012) who found similar results for smallholder farmers in the
Northern-east and the Central Thailand. Furthermore, the spreading of sales over a time
period (3.67) and the reduction in debts (3.53) were ranked 3rd and 4th effective income
stabilising risk management strategies adopted by the rural households in semi-arid regions
of Ghana. According to the household in the semi-arid regions of Ghana, the effort of
growing different crops to stabilise household income (mean of 2.62) was the least effective
risk management strategy used to stabilise household income.
Downloaded by University of Ghana At 08:06 27 June 2019 (PT)
AJEMS
9,3
376
Table IV.
Varimax rotated
factor loadings of
management
strategies for
Northern Ghana
Downloaded by University of Ghana At 08:06 27 June 2019 (PT)
Effective risk management strategies
Management strategies Mean SD Rank Divers. Fin. Mgt. Off-farm Communality Cronbach’s α KMO
Managing risk by putting effort into
Spread sales over time 3.190 1.45 10 0.31 0.47 0.58 0.65 0.96 0.96
Work off-farm as supplement 3.022 1.51 16 0.39 0.34 0.74 0.82 0.96 0.96
Invest in non-farm business 3.078 1.45 12 0.39 0.33 0.70 0.74 0.96 0.94
Have livestock on farm 2.656 1.43 21 0.19 0.71 0.43 0.74 0.96 0.96
Choose crop varieties with lower price variability 2.638 1.38 22 0.22 0.53 0.64 0.75 0.96 0.96
Choose crop varieties with lower yield variability 3.040 1.09 15 0.18 0.34 0.59 0.49 0.96 0.93
Reduce debts 2.798 1.58 20 0.18 0.86 0.21 0.82 0.96 0.94
Store feed or seed reserves 2.982 1.58 17 0.36 0.59 0.45 0.67 0.96 0.94
Grow different crops 2.622 1.51 24 0.23 0.79 0.22 0.73 0.96 0.93
Sign rainfall insurance 2.974 1.6 18 0.31 0.11 0.62 0.49 0.96 0.94
Save cash 3.068 1.52 13 0.3 0.71 0.36 0.73 0.96 0.93
Save asset 2.848 1.58 19 0.21 0.9 0.15 0.87 0.96 0.9
Stabilising household income through
Spread sales over time 3.668 1.39 3 0.73 0.22 0.14 0.46 0.96 0.94
Work off-farm as supplement 3.320 1.42 7 0.63 0.14 0.21 0.6 0.96 0.92
Invest in non-farm business 3.200 1.43 9 0.66 0.21 0.4 0.65 0.96 0.94
Have livestock on farm 3.206 1.43 8 0.65 0.2 0.39 0.62 0.96 0.97
Choose crop varieties with lower price variability 3.154 1.39 11 0.74 0.11 0.26 0.62 0.96 0.92
Choose crop varieties with lower yield variability 3.066 1.42 14 0.71 0.14 0.2 0.56 0.96 0.91
Reduce debts 3.532 1.44 4 0.70 0.28 0.33 0.69 0.96 0.95
Store feed or seed reserves 3.706 1.44 2 0.68 0.13 0.02 0.48 0.96 0.88
Growing different crops 4.386 1.55 1
Sign rainfall Insurance 2.626 1.47 23
Save cash 3.490 1.47 5 0.77 0.3 0.31 0.78 0.96 0.95
Save asset 3.466 1.46 6 0.72 0.32 0.38 0.77 0.96 0.96
Eigenvalues 11.92 2.53 1.31
Per cent of total variance explained 26.7 22.2 18.1
Overall Cronbach’s α coefficient 0.96
Overall KMO 0.94
Cumulative per cent of variance explained 26.7 48.9 67
Number of variables 3 3 3
Notes: n¼ 500. Factor loadings with absolute values greater than 0.4 are in italic
Source: Field Survey (2015)
After the ranking, the factor analysis was conducted using the 24 risk management Farmers’ risk
variables. The first varimax orthogonal rotation indicated that stabilising household income perceptions
by growing different crops and signing rainfall insurance to stabilise household income had and risk
to be dropped due to the lower communalities than the required minimum of 0.4. A second
rotation was therefore performed with the 22 risk management variables. management
For the first factor, the study reports high loadings on choosing crop varieties with lower
yield variability, having livestock on the farm and choosing crop varieties with lower price 377
variability, hence factor one is described as “Diversification strategy”. Furthermore, factor two
loaded highly on the effort at using savings in the form of cash, the ease of converting an asset
into cash and the debt reduction strategies hence it was referred to as “Financial strategy”.
The last factor, factor three, was construed as “off-farm strategy” because it loaded highly on
the effort at using the non-farm business andworking off-farm as riskmanagement strategies.
Additionally, the factor also loaded highly on investing in a non-farm business to stabilise
household income. The finding of this study is largely consistent with those of Kisaka-Lwayo
and Ajuruchkwu (2012), Aditto et al. (2012) and Gebreegziabher and Tadesse (2014).
5.5 Correlation between perception and management
The correlation results using the standardised factor scores are presented in Table V. The
results generally show a positive and statistically significant relationship between the risk
factors and the effective risk management. For instance, the relationship between production
risk and diversification and between human risk and diversification were positive and
significant at one per cent meaning that farmers perceive diversification as a very important
strategy for managing both production and human risk. Furthermore, financial strategy was
perceived by farmers as an important strategy to adopt in the mitigation of all the three risk
sources identified in the study. Additionally, off-farm strategy was positive and significantly
related to both production risk and human risk, which is an indication that farmer in the
semi-arid region of Ghana perceived off-farm strategy as an important strategy for managing
human and production risk. The finding showed that the relationship between market risk
and diversification, and market risk and off-farm strategies were statistically not significant.
Probably most households in the rural setting are subsistence farmers and are not market
focussed, hence the limited relationship between these variables.
5.6 Relationship between perceived risk management strategies, socioeconomic
characteristics and the perceived sources of risk
The results from Table VI show that the gender of the household head was negatively related
to “diversification strategy”, an indication that that female household heads perceived
diversification management strategy as a more relevant strategy than male household heads.
In the semi-arid regions of Ghana female farmers are more diversified in their farming activities
since most are in-charge of growing different types of vegetables (Hesselberg and Yaro, 2006).
In contrast, male farmers perceived off-farm strategy as a more important risk management
strategy than their female counterpart. Since men are usually in-charge of the upkeep of their
homes, it is expected that more male heads engaged in wage employment than female heads.
Risk management strategies
Risk sources Diversification Financial Mgt. Off-farm
Market risk 0.06088092 0.5245043*** −0.04756402 Table V.Correlation between
Production risk 0.340463*** 0.5355763*** 0.457862*** perceived risk sources
Human risk 0.2820308*** 0.07991187* 0.08343481* and risk management
Notes: *,***Indicate pairwise correlation statistically significant at 10 and 1 per cent, respectively strategies
Downloaded by University of Ghana At 08:06 27 June 2019 (PT)
AJEMS
9,3 Risk managementDiversification Financial. Mgt. Off-farm
Coef. p-value Coef. p-value Coef. p-value
Intercept 4.26E-01 0.0354** −9.67E-03 0.9714 −1.29E-01 0.60244
Age −1.71E-03 0.9356 2.56E-02 0.3659 −7.73E-02 0.00317***
Gender −2.58E-02 0.0988* 1.20E-02 0.5649 4.75E-02 0.01356**
378 Education −1.98E-01 0.2025 1.10E-03 0.9958 −1.52E-01 0.42783
Family size 1.03E-02 0.5181 −3.22E-02 0.1308 4.61E-02 0.01876**
Farm size −3.40E-03 0.137 3.26E-03 0.2861 −3.09E-03 0.27252
TLU 3.53E-03 0.284 −8.77E-03 0.0466** −1.03E-03 0.79986
Farm income 1.15E-01 0.0739* −1.39E-01 0.1041 9.02E-02 0.25214
Market risk 4.89E-01 o2e-16*** 4.45E-02 0.2871 −8.50E-02 0.02733**
Production risk 5.07E-01 o2e-16*** 3.35E-01 2.37E-14*** 4.42E-01 o2e-16***
Table VI.
Linear regression of Human risk 6.90E-02 0.0332** 2.67E-01 1.35E-09*** 4.45E-02 0.262832
the effective risk Adjusted R 0.5395 0.2033 0.2516
management Bptest 23.245 0.01632 112.57 2.20E-16 12.156 0.352
strategies, risk factors F-statistics 51.44 2.20E-16*** 11.21 2.20E-16*** 14.76 2.20E-16***
and the socioeconomic Notes: *,**,***Statistically significant at 10, 5 and 1 per cent, respectively
characteristics Source: Field Survey (2015)
As expected older farmers were found to be less involved in off-farm strategy, hence they
perceived this risk management strategy as not important. In contrast, farmers who had higher
household incomes perceived the “diversification strategy” as relevant than farmers who had
lower income. In addition to the above, households with larger family members perceived the
use of off-farm strategy as an important strategy to manage risk. This is because family
members can supply more labour both on and off-farm in order to gain more income to
supplement income from the farm. Furthermore, the study reported that household with larger
livestock holding perceived as unimportant the financial strategy. Thus, households with
larger livestock holding could easily convert their livestock to cash hence might give less
attention to debt management than households with smaller livestock holdings.
Almost all the risk factors were positive and significantly associated with the effective
risk management strategies in the regression model. Production risk was found to be
positive and significantly associated with diversification strategy, financial management,
and off-farm strategy. This was an indication that farmers focus on these risk management
strategies to curb production risk.
Furthermore, human risk had a direct and statistically significantly related to both
diversification and financial strategies, implying that farmers emphasised on the
diversification and financial strategies when confronted by human risk. The study further
reported that the perception of market risk had a larger influence on the diversification and
the off-farm management strategies of adopted by these farm households (Table VII).
It is important to note that with the exception of the usage of financial strategy to
mitigate market risk and the usage of an off-farm strategy to mitigate the effect of human
risk which was statistically not significant, almost all risk management strategies were
perceived as important in the mitigation the effect of the perceived risk. Additionally, the
study shows that production risk is the commonest risk reported by most farm households
in Northern Ghana whereas diversification strategy was the dominant management
strategy applied by households to manage the negative effects associated with production,
human and market risk. Generally, the findings of the study showed that many rural
households in Northern Ghana employ multiple risk management strategies to manage risk
hence are diversified. This finding confirms Senadza (2014) who indicated that many rural
households in Ghana to be diversified.
Downloaded by University of Ghana At 08:06 27 June 2019 (PT)
Farmers’ risk
perceptions
and risk
management
379
Table VII.
Parameter estimates
of the risk sources:
marginal effect
Downloaded by University of Ghana At 08:06 27 June 2019 (PT)
Market Production Human risk
dF/dx SE pW |z| dF/dx SE pW |z| dF/dx SE pW |z|
Age −9.66E-05 1.76E-03 0.95622 −4.33E-04 1.78E-03 0.80795 −2.63E-03 1.80E-03 0.14364
Gender −1.86E-01 1.16E-01 0.10933 −1.08E-01 1.21E-01 0.36945 1.84E-01 1.06E-01 0.0837*
Education 2.18E-02 1.22E-02 0.07314* −7.31E-03 1.22E-02 0.5484 −1.29E-02 1.21E-02 0.28625
Family size 8.58E-03 1.23E-02 0.48613 −1.47E-02 1.25E-02 0.23768 −2.55E-02 1.25E-02 0.04217**
Farm size 5.01E-03 6.62E-03 0.44918 1.58E-02 6.99E-03 0.02396** 1.17E-02 6.89E-03 0.09098*
TLU 1.25E-03 2.76E-03 0.65152 −8.41E-03 5.56E-03 0.13044 2.05E-03 2.69E-03 0.44642
Off-farm −5.26E-02 4.93E-02 0.28566 −1.24E-01 4.94E-02 0.0123** −1.02E-01 4.97E-02 0.04033**
Farm income −1.25E-06 3.59E-06 0.72748 −1.40E-07 3.78E-06 0.97043 −8.86E-06 4.02E-06 0.02741**
Notes: *,**Statistically significant at 10 and 5 per cent, respectively
AJEMS The results from the probit model (Table AIV) and its marginal effect (Table VIII) also show
9,3 similar results to the OLS result on the risk management strategies. For instance, off-farm
strategy and financial strategy was perceived as important management strategies to
manage both production risk and human risk according to the results of both OLS and the
marginal effect of the probit model. Diversification management strategies was perceived as
important strategies for managing the human and production risk as depicted by the results
380 of the OLS and the marginal effect of the probit model. The result shows that larger families
perceived the use of off-farm management strategies to manage risk as important (similar to
the OLS results) while household heads with greater number of years of education and those
who cultivated larger farm size perceived the use of diversification risk management
strategies and off-farm management strategies as less important risk management
strategies, respectively. The probable reason could be that households with alternative
income source (off-farm income) due to the higher level of education and ownership of
off-farm business might not see agricultural as a very important source of income.
6. Conclusion and implications
Unlike existing literature in Sub-Saharan African that have focussed on livestock farmers, this
study investigated the risk perception and risk management strategies of a sample of 500
rain-fed smallholder farmers from the semi-arid regions on Ghana. The study applied the
descriptive statistics to identify the most important risk sources and the most effective risk
management strategies; and the factor analysis to investigate the risk factors and effective
risk management strategies of smallholder farmers in the semi-arid region of Ghana. The
linear regression and the probit model were used to examine the relationship between farm
and farmer characteristics, the effective risk management strategies and the risk perceptions
of these farmers. The results from the descriptive statistics showed that the rain-fed
smallholder farmers viewed the effect of the variation in crop yields, fertiliser prices and crop
prices as the three most important perceived sources of risks. To manage the negative impact
of these risks on household income, income stabilising strategies such as growing different
crops, storing feed/seed reserves and spreading of sales over time are implemented.
The factor analysis on the perceived sources of risks identified three interpretable risk factors:
market risk, production risk and human risk. Households in the semi-arid regions of Ghana
managed these three risk factors by using three interpretable effective risk management
strategies: diversification strategy, financial strategies and off-farm strategy. The study finds an
uphill relationship between the risk factors and the effective risk management strategy. The
results of the study suggest a statistically significant positive relationship between the risk
factors and the effective risk management strategies. Risk is diverse and different among
individual farm households in the semi-arid region as indicated by the low value of the coefficient
of determinant in both the regression and probit model. The results from the linear regression
using the standardised factor scores and the probit model indicated that age and education were
the only major determinant of market risk whereas farm income, TLU, and off-farm employment
were important determinants of human risk. Production risk was determined by TLU, farm size
and farm income. Additionally, the linear regression and the probit model results for the risk
management strategies indicated that farmers who prioritised diversification strategy were those
concerned about market risk, human risk and production risk. In addition, farmers who were
apprehensive about production risk and human risk prioritised financial strategy while off-farm
strategy was prioritised by farmers most affected by production risk. The findings generally
confirm the PMT on risk, which show that risk perception influence the management strategies
adopted. The finding showed that socioeconomic variables were important determinants of the
riskmanagement strategy. Gender was not an important determinant of diversification, though it
was an important determinant of the off-farm strategies of the smallholder farmers. Furthermore,
TLU was not perceived as an important determinant of the use of financial strategy.
Downloaded by University of Ghana At 08:06 27 June 2019 (PT)
Farmers’ risk
perceptions
and risk
management
381
Table VIII.
Parameter estimates
of the risk
management
strategies: marginal
effect
Downloaded by University of Ghana At 08:06 27 June 2019 (PT)
Diversification Financial Off-farm
dF/dx SE pW |z| dF/dx SE Pr(W |z|) dF/dx SE Pr(W |z|)
Age −1.92E-03 2.14E-03 0.36888 8.89E-04 2.14E-03 0.6781 −9.79E-04 2.02E-03 0.6278
Gender −8.91E-02 1.52E-01 0.55727 1.00E-02 1.45E-01 0.9449 −1.23E-01 1.33E-01 0.35509
Education −3.86E-02 1.56E-02 0.01352** 1.79E-02 1.48E-02 0.2288 1.87E-02 1.41E-02 0.18409
Family size 8.04E-03 1.51E-02 0.59374 −1.59E-02 1.48E-02 0.2824 2.67E-02 1.43E-02 0.06172*
Farm size −2.40E-03 8.65E-03 0.7811 1.14E-02 8.29E-03 0.1676 −1.92E-02 8.03E-03 0.01671**
TLU 4.41E-03 5.74E-03 0.44184 1.65E-01 5.85E-02 0.0047*** −2.10E-03 2.66E-03 0.42929
Farm Income −2.15E-06 4.71E-06 0.64756 −6.30E-06 4.76E-06 0.1854 3.59E-06 4.34E-06 0.40764
Market risk 2.91E-01 4.00E-02 6.42E-11*** 3.83E-02 3.44E-02 0.265 −1.18E-03 3.00E-02 0.9687
Production risk 1.56E-02 3.54E-02 0.65893 2.64E-01 3.48E-02 3.99E-15*** 5.52E-02 3.01E-02 0.06669*
Human risk 3.35E-01 3.66E-02 2.20E-16*** 3.04E-01 3.57E-02 2.20E-16*** 3.16E-01 3.55E-02 o2e-16***
Notes: *,**,***Statistically significant at 10, 5 and 1 per cent, respectively
AJEMS Household size was perceived as an important determinant of the usage of off-farm strategy.
9,3 Wealthy ( farm income) were more diversified than less wealthy households. The study
recommends that apart from the socioeconomic characteristics, attention should be paid to
understandingmarket, production and humanwhen designing policies targeted at improving the
welfare of smallholder rain-fed farmers in the semi-arid region of Ghana.
382 Notes
1. Using the 2015 rate of 1 GHc ¼ 0.303.
2. The direct scores or the nominal variables given by the respondents were treated as dummies and
also analysed using probit model.
References
Aditto, S., Gan, C. and Nartea, G.V. (2012), “Sources of risk and risk management strategies: the case of
smallholder farmers in a developing economy”, in Banaitiene, N. (Ed.), Risk Management:
Current Issues and Challenges, INTECH, pp. 449-674.
Ahsan, D.A. (2011), “Farmer’ motivations, risk perceptions and risk management strategies in a
developing economy: Bangladesh experience”, Journal of Risk Research, Vol. 14 No. 3, pp. 325-349.
Ahsan, D.A. and Roth, E. (2010), “Farmers’ perceived risks and risk management strategies in an emerging
Mussel aquaculture industry in Denmark”,Marine Resource Economics, Vol. 25 No. 3, pp. 309-323.
Bergfjord, O.J. (2009), “Risk perception and risk management in Norwegian aquaculture”, Journal of
Risk Research, Vol. 12 No. 1, pp. 91-104.
Bishu, K.G., O’Reilly, S., Lahiff, E. and Steiner, B. (2016), “Cattle farmers’ perceptions of risk and risk
management strategies: evidence from Northern Ethiopia”, Journal of Risk Research, Vol. 19
No. 5, pp. 579-598.
Cafri, G., Thompson, J.K., Jacobsen, P.B. and Hillhouse, J. (2009), “Investigating the role of appearance-
based factors in predicting sunbathing and tanning salon use”, Journal of Behavioral Medicine,
Vol. 32 No. 6, pp. 532-544.
Cameron, C.A. and Trivedi, P.K. (2005), Microeconometrics: Methods and Applications, Cambridge
University Press, Cambridge.
Canagarajaha, S., Newman, C. and Battamishra, R. (2001), “Non-farm income, gender, and inequality:
evidence from rural Ghana and Uganda”, Food Policy, Vol. 26 No. 4, pp. 405-420.
DeJoy, D.M. (1996), “Theoretical models of health behavior and workplace self-protective behavior”,
Journal of Safety Research, Vol. 27 No. 2, pp. 61-72.
Deressa, T.T., Hassan, R.M., Ringler, C., Alemu, T. and Yesuf, M. (2009), “Determinants of farmers’
choice of adaptation methods to climate change in the Nile Basin of Ethiopia”, Global
Environmental Change, Vol. 19 No. 2, pp. 248-255.
Di Falco, S. and Veronesi, M. (2014), “Managing environmental risk in presence of climate change: the role of
adaptation in the Nile Basin of Ethiopia”, Environmental & Resource Economics, Vol. 57 No. 4,
pp. 553-577.
Field, A. (2005), Discovering Statistics Using SPSS, 2nd ed., Sage Publication, London.
Flaten, O., Lien, G., Koesling, M., Valle, P.S. and Ebbesvik, M. (2005), “Comparing risk perceptions and
risk management in organic and conventional dairy farming: empirical results from Norway”,
Livestock Production Science, Vol. 95 Nos 1-2, pp. 11-25.
Gebreegziabher, K. and Tadesse, T. (2014), “Risk perception and management in smallholder dairy
farming in Tigray, Northern Ethiopia”, Journal of Risk Research, Vol. 17 No. 3, pp. 367-381.
Ghana Statistical Service (2014), Ghana Living Standard Survey: Poverty Profile in Ghana (2005-2013),
Ghana Statistical Service, Accra.
GSS (2014), Final 2012 Gross Domestic Product & Revised 2013 Gross Domestic Product, Ghana
Statistical Service, Accra.
Downloaded by University of Ghana At 08:06 27 June 2019 (PT)
Gucer, P.W., Oliver, M. and McDiarmid, M. (2003), “Workplace threats to health and job turnover among Farmers’ risk
women workers”, Journal of Occupational and Environmental Medicine, Vol. 45, pp. 683-690. perceptions
Hair, J.F., Black, W.C., Babin, B.J., Anderson, R.E. and Tatham, R.L. (2006), Multivariate Data Analysis, and risk
6th ed., Pearson Prentice Hall, Upper Saddle River, NJ. management
Hair, J.F., Black, W.C., Babin, B.J., Anderson, R.E. and Tatham, R.L. (2010), Multivariate Data Analysis,
7th ed., Pearson Prentice Hall, Upper Saddle River, NJ.
Hardaker, J., Huirne, R. and Anderson, J. (2004), Coping with Risk in Agriculture, 2nd ed., CAB 383
International: CABI Publishing, Wallingford.
Hesselberg, J. and Yaro, J.A. (2006), “An assessment of the extent and causes of food insecurity in
Northern Ghana using a livelihood vulnerability framework”, GeoJournal, Vol. 67, pp. 41-55.
Kim, J.O. and Mueller, C.W. (1978), Factor analysis: statistical methods and practical issues, Sage, CA.
Kisaka-Lwayo, M. and Ajuruchkwu, O. (2012), “Risk perceptions and management strategies by
smallholder farmers in KwaZulu-Natal Province, South Africa”, International Journal of
Agricultural Management, Vol. 1 No. 3, pp. 28-39.
Koesling, M., Ebbesvik, M., Lien, G., Flaten, O., Valle, P.S. and H, A. (2004), “Risk and risk management
in organic and conventional cash crop farming in Norway”, Acta Agriculturae Scandinavica,
Sectection C, Vol. 1 No. 4, pp. 195-206.
Le, T.C. and Cheong, F. (2010), “Perception of risk and risk management in Vietnamese catfish farming:
an empirical study”, Acquaculture Economics and Management, Vol. 14 No. 4, pp. 282-314.
Legesse, B. and Drake, L. (2005), “Determinants of smallholder farmers’ perceptions of risk in the
Eastern Highlands of Ethiopia”, Journal of Risk Research, Vol. 8 No. 5, pp. 383-416.
Martin, S. (1996), “Risk management strategies in New Zealand agriculture and horticulture”, Review of
Agricultural, Food and Environmental Studies, Vol. 64 No. 1, pp. 31-44.
Maurer, K. (2014), “Where is the risk? Is agricultural banking really more difficult than other sectors?”,
in Köhn, D. (Ed.), Finance for Food: Towards New Agricultural and Rural Finance, Springer
Heidelberg, New York, NY; Dordrecht and London, pp. 139-165.
Menapace, L., Colson, G. and Raffaelli, R. (2016), “A comparison of hypothetical risk attitude elicitation
instruments for explaining farmer crop insurance purchases”, European Review of Agricultural
Economics, Vol. 43 No. 1, pp. 113-135.
Meraner, M. and Finger, R. (2017), “Risk perceptions, preferences and management strategies: evidence
from a case study using German livestock farmer”, Journal of Risk Research, Vol. 20, pp. 1-26.
Meuwissen, M.P., Huirne, R.B. and Hardaker, J.B. (2001), “Risk and risk management: an empirical
analysis of Dutch livestock farmers”, Livestock Production Science, Vol. 69, pp. 43-53.
MoFA (2012), Agriculture in Ghana: Facts and Figures, Statistics, Research and Information Directorate
(SRID), Accra.
MoFA (2015), 2014 Agriculture Sector Annual Progess Report, Monitoring and Evaluation Directorate,
Accra.
Senadza, B. (2014), “Income diversification strategies among rural households in developing countries”,
African Journal of Economic and Management Studies, Vol. 5 No. 1, pp. 75-92.
Sonkkila, S. (2002), Farmers’ Decision-Making on Adjustment into the EU, Department of Economics
and Management, University of Helsinki, Helsinki.
van Winsen, F., de Mey, Y., Lauwers, L.V., Vancauteren, M. and Wauters, E. (2016), “Determinants of
risk behaviour: effects of perceived risks and risk attitude on farmer’s adoption of risk
management strategies”, Journal of Risk Research, Vol. 19 No. 1, pp. 56-78.
Wilde, G.J. (1994), Target Risk: Dealing with the Danger of Death, Disease and Damage in Everyday
Decisions, PDE Publications, Toronto.
Corresponding author
Richard Asravor can be contacted at: rkAsravor@yahoo.com
Downloaded by University of Ghana At 08:06 27 June 2019 (PT)
AJEMS Appendix 1
9,3
(a) Scree plot: Risk Sources Northern Ghana
6
384 5
4
3
2
1
0
2 4 6 8 10 12 14
Component number
(b) Scree plot: Risk Strategies Northern Ghana
12
10
8
6
4
2
0
Figure A1.
Scree plots 5 10 15 20
Component number
Appendix 2. Variance inflation factors (VIF)
Northern Ghana
Variables VIF Sqrt(VIF)
Age 1.195316 1.09330508
Gender 1.031338 1.01554813
Education 1.599787 1.26482687
Family size 1.367319 1.16932416
Farm size 1.214669 1.10212023
TLU 1.160878 1.07744049
Table AI. Off-farm 1.516538 1.23147797
VIF: Risk perceptions Farm income 1.128381 1.06225279
Downloaded by University of Ghana At 08:06 27 June 2019 (PT)
Eigenvalues of components Eigenvalues of components
Northern Ghana Farmers’ risk
Variables VIF Sqrt(VIF) perceptions
and risk
Age 1.214457 1.102024
Gender 1.034975 1.017337 management
Education 1.622703 1.273854
Family size 1.375942 1.173006
Farm size 1.236414 1.111942 385
TLU 1.190337 1.091026
Farm income 1.521884 1.233647
Market risk 1.01769 1.008806 Table AII.
Production risk 1.066298 1.032617 VIF: risk management
Human risk 1.025988 1.012911 strategies
Downloaded by University of Ghana At 08:06 27 June 2019 (PT)
AJEMS Appendix 3. Probit model of risk sources
9,3
386
Table AIII.
Parameter estimate of
the probit model of
the risk sources
Downloaded by University of Ghana At 08:06 27 June 2019 (PT)
Market risk Production risk Human risk
Estimate SE Pr(W |z|) Estimate SE Pr(W |z|) Estimate SE Pr(W |z|)
Intercept −9.58E-02 3.89E-01 0.8055 4.86E-01 3.89E-01 0.2107 4.68E-01 3.98E-01 0.2387
Age −2.86E-04 4.43E-03 0.9485 −1.01E-03 4.46E-03 0.821 −6.71E-03 4.48E-03 0.1336
Gender −4.71E-01 3.06E-01 0.1235 −2.70E-01 3.03E-01 0.3724 4.91E-01 3.16E-01 0.1198
Education 5.45E-02 3.05E-02 0.0738* −1.82E-02 3.05E-02 0.551 −3.24E-02 3.04E-02 0.2869
Family size 2.14E-02 3.09E-02 0.4887 −3.71E-02 3.12E-02 0.2336 −6.33E-02 3.12E-02 0.0425**
Farm size 1.29E-02 1.66E-02 0.4393 3.86E-02 1.72E-02 0.0249** 2.87E-02 1.70E-02 0.0917*
TLU 3.30E-03 6.74E-03 0.6244 −2.04E-02 1.35E-02 0.1312 5.25E-03 6.61E-03 0.4273
Off-farm −1.32E-01 1.24E-01 0.2873 −3.09E-01 1.24E-01 0.0132** −2.55E-01 1.24E-01 0.0406**
Farm income −3.32E-06 9.00E-06 0.7118 −9.71E-07 9.29E-06 0.9167 −2.21E-05 9.72E-06 0.023**
Pseudo R2 0.015207 0.0355097 0.03857573
log pseudo likelihood −335.4979 −328.5812 −328.3936
df 9 9 9
Prob W χ2¼ 0.2405713 0.002125595 0.000913853
Notes: *,**Statistically significant at 10 and 5 per cent, respectively
Appendix 4. Probit model of risk management strategies Farmers’ risk
perceptions
and risk
management
387
Table AIV.
Parameter estimate of
the probit model of
the risk management
Downloaded by University of Ghana At 08:06 27 June 2019 (PT)
Diversification Financial Off-farm
Estimate SE Pr(W |z|) Estimate SE Pr(W |z|) Estimate SE Pr(W |z|)
Intercept 8.98E-01 4.58E-01 0.05** −3.67E-01 4.48E-01 0.41284 1.64E-01 4.30E-01 0.7028
Age −4.31E-03 5.06E-03 0.3944 2.25E-03 5.06E-03 0.65702 −2.31E-03 4.82E-03 0.6311
Gender −2.72E-01 3.54E-01 0.4423 2.65E-02 3.41E-01 0.93808 −2.72E-01 3.37E-01 0.4199
Education −9.21E-02 3.63E-02 0.0111** 3.82E-02 3.50E-02 0.27513 4.49E-02 3.34E-02 0.1786
Family size 1.52E-02 3.55E-02 0.6686 −3.99E-02 3.50E-02 0.25448 6.07E-02 3.39E-02 0.073
Farm size −1.27E-03 2.01E-02 0.9497 2.80E-02 1.97E-02 0.1537 −4.52E-02 1.90E-02 0.0172**
TLU 1.03E-02 1.31E-02 0.4322 3.78E-01 1.42E-01 0.00781*** −5.01E-03 6.52E-03 0.4421
Farm income −6.74E-06 1.10E-05 0.5387 −1.54E-05 1.12E-05 0.17066 8.35E-06 1.03E-05 0.4158
Market risk 6.84E-01 9.07E-02 4.89E-14*** 1.02E-01 7.80E-02 0.19106 1.48E-02 6.99E-02 0.8321
Production risk 5.18E-02 8.01E-02 0.5185 6.39E-01 7.87E-02 4.68E-16*** 1.28E-01 7.03E-02 0.0679*
Human risk 7.94E-01 8.06E-02 o2e-16*** 7.24E-01 7.92E-02 o 2e-16*** 7.52E-01 7.86E-02 o2e-16***
Pseudo R2 0.3024257 0.2797149 0.2060768
log pseudo likelihood −239.3256 −246.4358 −272.0475
df 12 12 12
Prob W χ2¼ 2.05E-38 4.50E-35 9.16E-25
Notes: **,***Statistically significant at 5 and 1 per cent, respectively