Advertisement

Impact of Bioenergy Crop Adoption on Total Crop Incomes of Farmers in Northern Ghana: The Case of Jatropha Curcas

  • Lauretta S. Kemeze
  • Akwasi Mensah-Bonsu
  • Irene S. Egyir
  • D. P. K. Amegashie
  • Jean Hugues Nlom
Open Access
Chapter

Abstract

The present study examines the adoption of Jatropha Curcas as a bioenergy crop in West Mamprusi and Mion districts of Northern Ghana. A gender analysis is also undertaken. Using data from 400 farmers, the study employs a propensity score matching method to analyze the impact of the adoption of Jatropha Curcas on total crop incomes of farmers. The study concludes that Jatropha adoption significantly reduces the level of crop income per hectare of farmers. The reduction is worse for women-headed households compared to men. Specifically, the Average Treatment effect on the Treated (ATT) estimate is GHC −385.23 per hectare for the whole sample. The study recommends the development of appropriate strategies and a regulatory framework to harness the potential economic opportunities from Jatropha cultivation.

6.1 Introduction

Energy services have the potential to boost social and economic welfare of people . Access to energy is a crucial component of poverty alleviation, improving human welfare, and raising living standards (UNDESA 2005).

Most countries in sub-Saharan Africa rely on traditional biomass (crude oil, natural gas, and coal) as the primary energy source used and imported fossil fuels (IEA 2014). In the region, nearly 730 million people live in rural areas where they rely on traditional biomass for cooking (IEA 2014). In sub-Saharan Africa, energy demand grew by around 45% between 2000 and 2012. In Africa, over 80% of electricity generated is from fossil fuels; about 620 million people do not have access to electricity (IEA 2014). Energy demand is predicted to double from 500 million tonnes oil equivalent (Mtoe) in the year 2000 to 1000 Mtoe in 2030 (Denruyter et al. 2010).

The heavy reliance on fossil fuels raises serious environmental issues such as depletion of non-renewable resources, ozone depletion, and global warming. According to Intergovernmental Panel of Climate Change (IPCC) (2007), global Greenhouse Gas (GHG) emissions should be reduced by 50–80% by 2050 to slow down global warming. This means that the use of fossil fuels for energy generation should be restrained. Even though Africa’s GHG emissions account for less than 4% of the global GHG emissions, the continent is the most vulnerable to climate change effects such as droughts and flooding (World Bank 2009). This is because Africa is exposed to climate risks such as extreme droughts, flooding, and storms. In addition, its low adaptive capacity worsens the situation because the continent is characterized by high rates of poverty, financial and technological constraints, and heavy reliance on rain-fed agriculture.

Biomass is the dominant source of energy supply in Ghana. The country depends entirely on imports in order to meet oil requirements. The production of oil started with a capacity of 85,000 barrels of oil per day in Jubilee field (Abdulai 2013). In 2007, biomass energy consumption (wood fuel and charcoal) was about 11.7 million tonnes (Ministry of Energy 2010). It is used mainly for cooking, employing traditional inefficient technologies. Less than 10% of people use modern cooking fuels (improved stoves, kerosene, or liquefied petroleum gas) in the country (Ahiataku-Togobo and Ofosu-Ahenkorah 2009). In 2007, petroleum products and electricity consumption accounted for 1.955 million tonnes and 6269 GWh respectively (Ministry of Energy 2010). Biomass (fuelwood and charcoal) consumption in Ghana accounted for 64%. Petroleum products and electricity accounted for 27% and 9% respectively (Duku et al. 2011). However, according to IPCC (2007), combustion of fossil fuels contributes to global warming.

The energy sector faces some challenges in Ghana (Energy Commission 2006). These are related to the increase in energy demand, the potential imbalance between national energy production and indigenous sources of supply, inadequate investments in the energy sector, and overreliance on fuel imports and wood fuels. In order to address the issues of imbalance, low investments, and overreliance, Jatropha has been promoted as a panacea and promising feedstock for biofuels. However, Jatropha industry started in Ghana without any biofuel policies (Campion et al. 2012). In Ghana, prior to its introduction as a bioenergy crop, Jatropha was traditionally grown as gardens and a hedge or fence plant around homes in order to protect houses and fields against animals and sun exposure (Acheampong and Betey 2013). Jatropha was considered in Ghana for its ability to generate energy just recently, in 2005 (Boamah 2014). Its cultivation was promoted on marginal lands so as to not compromise food security (Boamah 2014). According to Brittaine and Lutaladio (2010), Ghana was predicted to be among the largest Jatropha producers in Africa by 2015. Projects related to Jatropha development started from 2005; by 2006, there were 17 biofuel projects in Ghana (Schoneveld et al. 2010). Several foreign companies (Agroils, Kimminic Estates, Jatropha Africa, Viram Plantation Limited, etc.) acquired large-scale land to produce both edible and non-edible crops for ethanol and biodiesel generation for exports (Dogbevi 2009). Large-scale Jatropha (100 hectares and more) development was highly criticized by Ghanaian NGOs for issues such as land grabbing and food insecurity as many of them were actually on fertile lands. Many of these large-scale Jatropha projects have failed. Of late, mainly participatory and small-scale Jatropha developments are ongoing in Ghana.

Jatropha seeds are not directly marketable in the open market . Farmers mostly sell their seeds to an NGO called New Energy. Previously, some farmers were selling the seeds to foreign investors under contract farming. However, these foreign investors are no more buying Jatropha seeds.

Income constitutes a key determinant of food security for poor people in rural areas since adequate income can help them afford appropriate food for their nutritional diet (FAO 2010; Faaij 2008). Jatropha can provide new income sources for farmers through Jatropha-generated activities such as seed selling. This supplementary income from Jatropha can impact the food security status of farmers helping them to afford food. According to the Government of Ghana (2010), the investment in land is predominant in the northern part of Ghana due to their high prevalence of food insecurity, poverty, and illiteracy. In addition, Northern Ghana agriculture accounts for more than 90% of household incomes and employs more than 70% of the population in the region.

The objective of this chapter is to measure the impact of Jatropha Curcas adoption on total crop incomes of farmers in Northern Ghana. Section 6.2 presents the theoretical framework and estimation technique. Section 6.3 provides data and descriptive analysis. Section 6.4 presents the empirical results. Finally, Sect. 6.5 provides the conclusions.

6.2 Theoretical Framework and Estimation Technique

6.2.1 Theoretical Framework: The Random Utility Framework

Following Hoque et al. (2015), a household’s decision to adopt a bioenergy crop can be analyzed within a random utility framework. Let UhA be the utility obtained by a household h from adopting Jatropha and UhN the utility of non-adoption. Let Zh be a vector of farm and household characteristics affecting bioenergy crop-adoption decisions and εh be the error term. According to the state of adoption, the household h utility is approximated as:
$$ \left\{\begin{array}{l}{U}_{hA}=f\left({Z}_h\right)+{\varepsilon}_{hA}\\ {}{U}_{hN}=f\left({Z}_h\right)+{\varepsilon}_{hN}\end{array}\right. $$
(6.1)
A household will choose to adopt Jatropha only if the utility derived from adopting is greater than the utility from not adopting: UhA ≻ UhN. Since these utilities are not observable, they can be expressed in the following latent structure model for adoption of bioenergy crop:
$$ {V}_h^{\ast }=\beta {Z}_h+{\varepsilon}_h $$
(6.2)
$$ {V}_h=\left\{\begin{array}{l}1,{V}_h^{\ast}\succ 0\\ {}0,{V}_h^{\ast}\prec 0\end{array}\right. $$

Where Vh is a binary indicator taking the value of 1 in the case the household adopts bioenergy crop and 0 otherwise.

The outcome variable (total crop incomes per hectare of the household) is considered as a linear function of the binary variable for bioenergy crop adoption along with a vector of some other explanatory variables (X):
$$ {Y}_h=\lambda {X}_h+\gamma {V}_h+{\mu}_h $$
(6.3)

Where Yh is the outcome variable, Vh is a binary variable for adoption, λ and γ are vectors of parameters to be estimated and μ is the error term. However, from Eq. 6.3, since γ measures the impact of bioenergy crop adoption (treatment variable) on total crop incomes per hectare (outcome variable), then, households should be randomly assigned to the group of adopters or non-adopters. However, technologies are rarely randomly assigned. Instead, new technology adoption usually occurs through self-selection. In other words, it translates the fact that in Eq. 6.3, μ is correlated with V or Z. Equation 6.2 which does not take into account the self-selection might lead to a biased estimation. The propensity score matching (PSM) is employed in this study in order to deal with selection bias.

6.2.2 Estimation Technique: Propensity Score Matching

Rosenbaum and Rubin (1983) defined the Average Treatment Effect (ATE) as follows:
$$ \mathrm{ATE}={Y}_i^A-{Y}_i^N $$
(6.4)
where \( {Y}_i^A \) is the total crop income per hectare of household i that adopted and \( {Y}_i^N \) is the total crop income per hectare of household i that did not adopt. It is difficult to estimate the impact from Eq. 6.4. The issue is that either \( {Y}_i^A \) or \( {Y}_i^N \) is normally observed but not both of them for each household. What is normally observed is expressed as follows:
$$ {Y}_i={K}_i{Y}_i^A+\left(1-{K}_i\right){Y}_i^N\kern2.5em \mathrm{K}=0,1 $$
(6.5)
where K = 1 represents the situation when the household i adopts Jatropha and K = 0 is the situation when the household has not adopted Jatropha.
The ATE can be re-specified as follows:
$$ {\displaystyle \begin{array}{l}\mathrm{ATE}=P\cdot \left[E\left({Y}_i^A/K=1\right)-E\left({Y}_i^N/K=1\right)\right]+\\ {}\kern2.16em \left(1-P\right)\cdot \left[E\left({Y}_i^N/K=0\right)-E\left({Y}_i^N/K=0\right)\right]\end{array}} $$
(6.6)
where P is the probability for a household to adopt Jatropha (K = 1).
Equation 6.6 is based on the assumption that the unobserved counterfactual of adopters if they had not adopted, \( E\left({Y}_i^N/K=1\right) \), can be approximated by the one of non-adopters \( E\left({Y}_i^N/K=0\right) \). Without that assumption, the estimation of Eq. 6.4 representing the ATE cannot be done because \( E\left({Y}_i^N/K=1\right) \) is not observed. However, that procedure might highly result in a biased estimation because of the issue of selection bias. Indeed, the treated group (adopters) might not be statistically similar to the control group (non-adopters). Fortunately, the PSM approach of Rosenbaum and Rubin (1983) first reduces the pre-treatment characteristics of each household into one variable. Secondly, PSM uses the propensity score to match households with similar characteristics. Rosenbaum and Rubin (1983) defined ‘propensity score’ as the conditional probability of receiving a treatment given pre-treatment characteristics:
$$ p(X)\equiv \Pr \left\{K=1/X\right\}=E\left\{K/X\right\};\kern0.62em p(X)=F\left\{h\left({X}_i\right)\right\} $$
(6.7)
where F{.} is a normal or logistic cumulative distribution and X a vector of pre-treatment characteristics. An estimation of the propensity of Jatropha adoption is run taking into account the restriction of the region of common support. After computing the propensity scores, the Average Treatment effect on the Treated (ATT) is estimated as follows:
$$ {\displaystyle \begin{array}{l}\mathrm{ATT}=E\left\{{Y}_i^A-{Y}_i^N/K=1\right\}\\ {}\kern2.74em =\mathrm{E}\left[\mathrm{E}\left\{{Y}_i^A-{Y}_i^N/K=1,p(X)\right\}\right]\\ {}\kern2.74em =\mathrm{E}\left[\mathrm{E}\left\{{Y}_i^A/K=1,p(X)\right\}-E\left\{{Y}_i^N/K=0,p(X)\right\}/K=1\right]\end{array}} $$
(6.8)
ATT is performed using a single matching algorithm named Nearest Neighbor Matching with replacement. When there are few comparison units, matching with replacement allows one comparison unit to be matched more than once with each nearest treatment unit. However, matching without replacement forces the matching between the treatment group and the comparison group that is quite different in propensity scores. This enhances the likelihood of bad matches (increase the bias of the estimator). The quality of the matching is undertaken using a balance test called the mean absolute standardized bias. For each variable, the mean standardized difference is computed before and after matching as follows:
$$ B(X)=100\frac{\overline{X_T}-\overline{X_C}}{\sqrt{\frac{V_T(X)+{V}_C(X)}{2}}} $$
(6.9)
where \( \overline{X_T} \) and \( \overline{X_C} \) are the sample means for the treated and control groups, VT(X) and VC(X) are the associated sample variances (Lee 2006). The bias reduction can be generated as follows:
$$ BR=100\left(1-\frac{B_{\mathrm{after}}}{B_{\mathrm{before}}}\right) $$
(6.10)

Rosenbaum and Rubin (1985) recommended that the mean standardized bias after matching greater than 20% is perceived as an indicator of failed matching. In addition, according to Sianesi (2004), the balance test can be done comparing the pseudo R2 and p-values from the propensity scores estimated before and after matching. After matching, there should not be any systematic differences in the distribution of covariates between adopters and non-adopters. As a result, the pseudo R2 should be low. The test should be rejected after matching and not before.

6.3 Data and Descriptive Statistics

Data used in this study were collected from 400 farmers in the West Mamprusi and Mion districts of Northern Ghana using a questionnaire from September to October 2015. These districts were selected because of their involvement in Jatropha production. These districts are among the poorest in Ghana; hence, issues of innovation , crop diversification, or technology adoption for wealth creation which Jatropha promises become pertinent. For the purpose of this study, adopters are classified as farmers who planted Jatropha and still have it in their plots, while non-adopters refer to farmers who did not cultivate Jatropha. Focus group discussions consisted of meeting with community leaders and some Jatropha farmers in each district to gather preliminary information on the number of Jatropha growers in the communities and the size of their Jatropha plot. The survey used a structured questionnaire to collect data from the households on socioeconomic characteristics of households and information on Jatropha. From the preliminary study, the estimated number of Jatropha farmers in these two districts was 344 (256 farmers in West Mamprusi District and 88 farmers in Mion District). This information was used to calculate the minimum sample size.

Yamane’s formula of sample size is used (Yamane 1967):

\( n=\frac{N}{1+N{(e)}^2} \) Where n is the sample size, N is the population size, and e is the error term. Assuming an error of 5% and a confidence interval of 95%, the following is obtained: \( n=\frac{344}{1+344{(0.05)}^2}=184.94 \). The number of Jatropha adopters was increased to 200.

A stratified random sampling technique consisting of dividing the population into groups called strata and proceeding with a simple randomization was used to select the list of Jatropha farmers to be surveyed. The stratification was done at the district level. The first stage involved purposive selection of Jatropha growing districts in Northern Ghana. These are West Mamprusi District and Mion District. A total of 120 Jatropha farmers were randomly selected in West Mamprusi District and 80 in Mion District in order to have a fair representation of farmers in both districts based on the preliminary study.

The study needed a counterfactual to evaluate the effect of Jatropha farming adoption on key outcomes such as income. An equal sample size of 200 non-Jatropha farmers was used. The procedure to survey the non-Jatropha farmers was as follows: In each community where the Jatropha farmers were surveyed the equal number of non-Jatropha farmers was also surveyed. To choose a non-Jatropha farmer, a sample list of some non-Jatropha farmers in the community was collected from community leaders. A random list with three back-ups was then formed for each community to survey the non-Jatropha farmers.

Table 6.1 shows the distribution of respondents per district and community.
Table 6.1

Distribution of respondents per district and community

District

Community

Adopters

Non-adopters

West Mamprusi District

Zagsilari

20

20

Nasia

19

19

Boamasa

20

20

Janga

20

20

Wungu

20

20

Loagri

21

21

Mion District

Jimle

47

47

Kpachaa

30

30

Tuya

03

03

Total

 

200

200

Source: Authors

Figure 6.1 shows the map of the study area.
Fig. 6.1

Map of study area. Source: Authors

Table 6.2 describes the variables used in the study.
Table 6.2

Variables definition for Jatropha adoption

Variable

Type

Definition and measurement

Expected signs

Treatment variable

 

Adopter

Dummy

1 = Grows Jatropha, 0 = Does not grow Jatropha

 

Outcome variable

 

Total crop incomes per hectare

Continuous

Income from all crops including Jatropha per hectare (in Ghana cedis)

 

Independent variables

 

Gender

Dummy

Gender of the head of household 0 = male, 1 = female

Age

Continuous

Age of the head of household (in years)

Education

Continuous

Level of education of the head of household in years

+

Number of adults

Continuous

Number of adult members of the household (count units)

+

Farming experience

Continuous

Farming experience of the head of household (years)

+

Farm size

Continuous

Farm size (hectares)

+

Extension services

Continuous

Number of times the household had access to extension services

+

Off-farm activities

Dummy

Engagement in off-farm activities 0 = No, 1 = yes

+

Livestock

Dummy

Livestock ownership 0 = No, 1 = yes

FBO

Dummy

Farmer based organization membership 0 = No, 1 = yes

+

District

Dummy

0 = Mion, 1 = West Mamprusi

Credit access

Dummy

Access to credit 0 = No, 1 = yes

+

Distance to market

Continuous

Distance from home to the nearest agricultural market (in km)

Size of land owned

Continuous

Size of land owned (in hectares)

+

Hired labor

Continuous

Number of man-days hired during 2014 cropping season

+

Irrigation

Dummy

Practice of irrigation 0 = No, 1 = yes

+

Risk attitude

Continuous

Degree of Risk attitude

+

Discount factor

Dummy

1 = preference for present, 0 otherwise

+

Source: Authors

Table 6.3 shows the descriptive statistics of continuous variables used in the econometric models for the entire sample and the two sub-samples of non-adopters and adopters indicating the variable means and standard deviations. Adopters are distinguishable in terms of household characteristics such as age, farming experience, number of visits by extension services officers, and number of man-days labor hired.

On average, adopters allocate half a hectare to Jatropha cultivation. The mean age of farmers interviewed was about 43 years old. There is a significant difference in the age of adopters and non-adopters. On average, adopters are 45 years old while non-adopters are 41 years old. There is no statistically significant difference in the number of adult members in the household of both adopters and non-adopters. On average, the number of adult members in the household is five. There is no statistically significant difference in the years of education for farmers both adopters and non-adopters. On average, farmers spent two years of education. There is a significant difference in the years of farming experience at the level of 5% between the two groups. On average, Jatropha adopters have been farming for about 26 years while non-adopters for about 23 years. There is no statistical difference in farm size between the two groups and the average farm size is 3.75 hectares. On average, farmers own 3.45 hectares of land. There is no statistically significant difference in distance from the nearest agricultural market for the full sample. On average, the distance from home to the nearest agricultural market is 8.3 kilometers. There is a statistically significant difference at the level of 1% in the number of times farmers had access to extension services. Adopters had more access to extension services than non-adopters. On average, adopters had access to extension services 0.49 times compared to 0.29 for non-adopters during the 2014 cropping season. Adopters hired more labor than non-adopters. The mean number of hired man-days for adopters is 125 compared to 79.16 man-days for non-adopters. The difference is significant at the level of 5%. There is no statistical difference in the degree of risk attitude of the farmers. On average, the degree of risk attitude is 5.73.

Table 6.3

Descriptive statistics for continuous variables

Variables

Adopters

Non-adopters

Total

t-value

Mean

SD

Mean

SD

Mean

SD

Age

45.13

15.25

41.06

13.76

43.09

14.65

−2.80***

Number of adults

5.81

5.54

5.21

3.86

5.51

4.78

−1.25

Farming experience

26.46

15.53

23.32

14.20

24.90

14.94

−2.11**

Farm size

3.79

2.65

3.70

0.19

3.75

2.68

−0.32

Education

2.17

4.24

1.68

3.80

1.93

4.03

−1.20

Distance to market

8.11

7.08

8.47

7.32

8.30

7.19

0.51

Extension services

0.29

0.46

0.15

0.36

0.22

0.42

−3.39***

Size of land owned

3.56

2.46

3.35

2.61

3.45

2.53

−0.86

Hired labor

125.71

240.71

79.16

142.01

102.43

198.74

−2.35**

Risk attitude

5.93

2.69

5.53

2.66

5.73

2.68

−1.49

Source: Authors

Note: **, *** show significance at 5% and 1% levels, respectively

Table 6.4 shows the descriptive statistics of categorical variables used in the econometric models for the entire sample and the two sub-samples of non-adopters and adopters indicating the frequencies and percentages.

There is no statistically significant difference in gender of farmers between the two groups. On average, 85.5% farmers are male. Adopters are less engaged in off-farm activities (30%) compared to non-adopters (40.5%). There is a significant difference in farmer based organization (FBO) membership at 1%. The percentage of FBO membership is higher for adopters. A total of 45% of adopters are members of FBO compared to 22% for non-adopters. On average, 64% of respondents own livestock. There is no statistically significant difference in access to credit between both groups. On average, only 19% of farmers had access to credit. The same for the discount factor, about 76% of farmers have a preference for the present. There is no statistically significant difference in irrigation practice ; only 2% of respondents practiced irrigation.

Table 6.4

Descriptive statistics for categorical variables

Variable

Category

Adopters

Non-adopters

Total

X2 value

No.

%

No.

%

No.

%

Gender

Male

166

83

176

88

342

85.5

2.02

Female

34

17

24

12

58

14.5

 

Off-farm act

Yes

60

30

81

40.5

141

35.25

4.83***

No

140

70

119

59.5

259

64.75

 

FBO

Yes

90

45

44

22

134

33.5

23.75***

No

110

55

156

78

266

66.5

 

Livestock

Yes

130

65

126

63

256

64

0.17

No

70

35

74

37

144

36

 

Credit access

Yes

38

19

38

19

74

19

0.00

No

162

81

162

81

324

81

 

Disc. factor

Yes

148

74

159

79.5

307

76.75

1.69

No

52

26

41

20.5

93

23.25

 

District

Mion

80

40

80

40

160

40

0.00

 

WMa

120

60

120

60

240

60

 

Irrigation

Yes

4

2

4

2

8

2

0.00

No

196

98

196

98

192

98

 

Source: Authors

Note: **, *** show significance at 5% and 1% levels, respectively

aWest Mamprusi district

Table 6.5 shows the descriptive statistics for total crop incomes per hectare of farmers for the whole sample and for male and female-headed households. There is a statistically significant difference at the level of 1% for the level of total crop incomes per hectare between adopters and non-adopters. On average, adopters have GHC 641.92 per hectare as total crop incomes while non-adopters have GHC 1243.41 per hectare.

There is a statistically significant difference at the level of 1% for the level of total crop income for both male-headed and female-headed households. On average, male-headed household adopters have GHC 673.57 per hectare while male-headed household non-adopters have GHC 1206.98 per hectare. On average, female-headed household adopters have GHC 487.39 per hectare while female-headed household non-adopters have GHC 1510.26 per hectare.

Nonetheless, descriptive statistics cannot explain whether the observed difference in crop income per hectare between adopters and non-adopters for the whole sample and by gender is due to Jatropha cultivation. The noted differences in total crop income per hectare depending on the adoption status of the household might not be the result of Jatropha adoption but rather might be due to other factors, farm and farmers’ characteristics, for instance. The impact analysis of Jatropha adoption on total crop incomes per hectare is preceded by the determination of the propensity scores for the treatment variable (adoption status of Jatropha).

Table 6.5

Descriptive statistics for total crop incomes per hectare

Total crop incomes

(GHC/ha)

Adopters

Non-Adopters

Total

t-value

Mean

SD

Mean

SD

Mean

SD

Whole sample

641.92

528.92

1243.41

1235.74

942.67

995.90

6.32***

Male

673.57

558.28

1206.98

1269.32

948.07

1024.18

4.98***

Female

487.39

314.68

1510.26

932.09

910.78

816.27

5.96***

Source: Authors

6.4 Empirical Results

The impact analysis of Jatropha adoption on total crop incomes per hectare is preceded by the determination of the propensity scores for the treatment variable (adoption status of Jatropha).

6.4.1 Estimation of the Propensity Scores

A probit model is used to predict the probability of adopting Jatropha. The results of the propensity scores are reported in Table 6.6.

Several variables are statistically significantly associated with adoption of Jatropha. The number of times of access to extension services, the number of man-days hired, the FBO’s membership, and the risk attitude of the head of household are positively associated with adoption. Access to extension services could play an important role in Jatropha adoption in Northern Ghana, for instance, in educating farmers in land use decisions concerning Jatropha. Jatropha is known as labor intensive; the ability of farmers to hire labor could increase its adoption. The coefficient of the risk attitude variable shows that the greater the degree of risk loving, the higher the probability of adopting Jatropha. Membership of FBO significantly increases the likelihood of Jatropha adoption. Indeed, FBO might assist farmers to manage the crop, find a market , and get access to loans.

The variables district and off-farm activities membership are negatively associated with Jatropha adoption. Being located in West Mamprusi District significantly reduces the likelihood of adopting Jatropha compared to Mion District. This is likely due to greater access to a potential market in Mion District. The NGO New Energy is buying the seeds from the farmers in Mion district. Engagement in off-farm activities has a negative influence on the probability of Jatropha adoption. The current result might be due to the fact that farmers engaged in off-farm activities have less time and resources to engage in Jatropha cultivation.

Table 6.6

Probit estimates of the propensity to adopt Jatropha

Variables

Probit

Coefficient

Standard error

Gender

0.286

0.205

District

−0.345*

0.181

Education

0.028

0.018

Age

0.014

0.007

Number of adults

0.026

0.016

Farming experience

−0.001

0.007

Farm size

−0.112

0.076

Extension services

0.200***

0.072

Off-farm activities

−0.299**

0.146

Livestock

0.097

0.146

Credit access

0.042

0.187

Distance to market

−0.001

0.011

Hired labor

0.002***

0.000

Size of land owned

0.063

0.081

FBO

0.583***

0.155

Risk attitude

0.086***

0.029

Discount factor

−0.178

0.179

Irrigation

−0.043

0.461

Constant

−1.176

0.335

Pseudo R2

0.1269

Log-likelihood

−242.08

Observations

400

Source: Authors

Note: *, **, *** show significance at 10%, 5%, and 1% levels, respectively

Table 6.7 provides the distribution of the propensity scores.

For adopters, the estimated propensity scores vary between 0.06356 and 0.98953 with a mean of 0.50168. For non-adopters, it varies between 0.06356 and 0.89317 with a mean of 0.42073. The results suggest that the region of common support is satisfied in the interval [0.06356, 0.89317]. The consequence of this restriction is that observations falling outside this range of the region will be discarded from the analysis. As a result, nine observations have been removed from the analysis. The common support condition is imposed in the regression models by matching in the region of common support only.

Table 6.7

Estimated propensity scores

Sample

Observations

Mean

SD

Minimum

Maximum

Whole sample

400

0.50168

0.20187

0.06356

0.98953

Adopters

200

0.58263

0.19964

0.11644

0.98953

Non-adopters

200

0.42073

0.16935

0.06356

0.89317

Source: Authors

The distribution of the propensity scores and the region of common support before and after matching are represented in Fig. 6.2.
Fig. 6.2

Propensity score distribution and common support for propensity score estimation. Source: Authors

6.4.2 Estimation of Average Adoption Effect: Matching Algorithms

Table 6.8 reports the estimates of the average adoption effects estimated by Nearest Neighbor Matching with replacement for the whole sample and for male and female. The analysis is based on the restriction of the region of common support.
Table 6.8

ATT of Jatropha adoption on crop income

Matching algorithm:

Nearest neighbor

Matching with replacement

Sample

ATT

Number of treated

Number of control

Whole sample

400

−385.23***

(−3.05)

170

200

Male

342

−293.22**

(−2.48)

143

176

Female

58

−624.23***

(−3.76)

15

24

Source: Authors

Note: **, *** show significance at 5% and 1% levels, respectively

Table 6.8 shows that adoption of Jatropha significantly reduces total crop income per hectare of farmers. For the whole sample, the ATT estimate per hectare is negative (GHC −385.230). This is probably due to the fact that most farmers grow Jatropha on their fertile lands. Jatropha production is likely to compete for land with food production resulting in smaller areas cultivated and less food grown at the local level. The absence of an open market for Jatropha especially in the West Mamprusi district makes income generation from Jatropha very difficult. This results in reducing total crop incomes of farmers. The adoption of Jatropha significantly reduces the total crop income per hectare of both male-headed and female-headed households. The ATT estimate is GHC −293.22 for male and GHC −624.23 for female. Comparing the impact on male-headed and female-headed households, the results reveal that the participation in Jatropha cultivation affects more negatively female-headed than male-headed households. The reduction in total crop income per hectare is higher for female-headed than male-headed households. This can be explained by the fact that bioenergy crops such as Jatropha are inputs intensive (land, labor, water, fertilizers, and pesticides) and female farmers have traditionally limited access to inputs (Tauli-Corpuz and Tamang 2007). In Ghana, for instance, female farmers have very little access and control over resources due to patriarchy. They face challenges such as unequal access to land, finance and credit, and so on.

6.4.3 Indicators of Matching Quality Before Matching and After Matching

Table 6.9 provides the indicators of matching quality. It reveals the results of covariate balancing tests before and after matching. It can be seen that all indicators of matching quality before matching significantly exceed those after matching. After matching, the results show an insignificant likelihood ratio test supporting a rejection of the joint significance of covariates. In addition, after matching the results reveal a lower pseudo R2. Indeed, the pseudo R2 dropped from 0.127 to 0.051 after matching. After matching, there is also a reduction in absolute bias for overall covariates used to estimate the propensity score . Table 9 also reveals a mean standardized bias lower than 20% after matching as recommended by Rosenbaum and Rubin (1985). The standardized mean difference for overall covariates used in the propensity score around 15% before matching is reduced to about 13.3% after matching. This leads to a substantial reduction of the total bias of 11.33% through matching. All these statistics suggest that the specification of the propensity score is fairly successful in balancing the distribution of covariates between adopters and non-adopters. These results can then be used to assess the impact of Jatropha adoption among groups of farmers having the same observed characteristics.
Table 6.9

Matching quality indicators before and after matching for the whole population

Matching quality indicators

Before matching

After matching

Pseudo R2

0.127

0.051

LR χ2

70.35

23.89

p ≻ χ2

0.000

0.159

Mean standardized bias%

15.0

13.3

Total % |bias| reduction

11.33

Source: Authors

6.5 Conclusions

This study examined the adoption of Jatropha Curcas and its impact on total crop incomes of farmers in two districts of Northern Ghana. The PSM method was used to account for selectivity bias. The results suggested the presence of bias in the distribution of covariates between groups of adopters and non-adopters, indicating that accounting for selection bias is a significant issue. The results showed that Jatropha adoption reduces the total crop incomes per hectare of farmers. The study also highlighted the potential gender-differentiated impacts of Jatropha adoption on total crop incomes per hectare of farmers. The impact is worse for female-headed households compared to male-headed households, though the impact is negative for both. The ATE on the Treated estimates are GHC −385.230 per hectare for the whole sample, GHC −624.23 per hectare for female-headed households, and GHC −293.22 for male-headed household. Therefore, Jatropha cultivation might constitute a threat to farmers’ crop incomes. There are a couple of reasons that can justify this finding, but the primary reason is the fact that Jatropha is cultivated in most cases on fertile lands and therefore conflicting with household staple and cash crops. The lack of market for Jatropha seeds is another reason which is worthy to highlight in this context.

The study recommends that the promotion of Jatropha cultivation should be properly regulated to avoid the massive conversion of fertile land used for crop production for Jatropha cultivation. There is a need to develop appropriate strategies and a regulatory framework to harness the potential economic opportunities from Jatropha cultivation while protecting rural people from converting part of their fertile lands to Jatropha cultivation at the expense of food crops. In this view, the Energy Commission (Ministry of Energy) should present the final bioenergy policy for Ghana in order for the country to move a step forward in the biofuel sector . The policy support is needed for improving income generation from Jatropha. Pro-women Jatropha development such as the promotion of Jatropha by-products (traditional soap and fertilizers) should be encouraged.

References

  1. Abdulai, N. (2013). Ways to Achieve Sustainable Development in the Oil and Gas Industry in Ghana. International Journal of ICT and Management, 1(2), 107.Google Scholar
  2. Acheampong, E., & Betey, B. (2013). Socio-Economic Impact of Biofuel Feedstock Production on Local Livelihoods in Ghana. Ghana Journal of Geography, 5, 1–16.Google Scholar
  3. Ahiataku-Togobo, W., & Ofosu-Ahenkorah, A. (2009). Bioenergy Policy Implementation in Ghana. In Comptete International Conference 26–28 May 2009. Lusaka, Zambia. Retrieved from http://www.compete-bioafrica.net/events/events2/zambia/Session-2/2-2-COMPETE-Conference-Lusaka-Togobo-Ghana.pdf.
  4. Boamah, F. (2014). Imageries of the Contested Concepts “Land Grabbing” and “Land Transactions”: Implications for Biofuels Investments in Ghana. Geoforum, 54, 324–334. Retrieved from  https://doi.org/10.1016/j.geoforum.2013.10.009.CrossRefGoogle Scholar
  5. Brittaine, R., & Lutaladio, N. (2010). Jatropha: A Smallholder Bioenergy Crop the Potential for Pro-Poor Development. Rome.Google Scholar
  6. Campion, B., Essel, G., & Acheampong, E. (2012). Natural Resources Conflicts and the Biofuel Industry: Implications and Proposals for Ghana. Ghana Journal of Geography, 4(1), 42–64. Retrieved from http://www.ajol.info/index.php/gjg/article/view/110786.Google Scholar
  7. Denruyter, J. P., et al. (2010). Bioenergy in Africa—Time for a Shift? Sud Sciences & Technologies, 145–158.Google Scholar
  8. Dogbevi, E. K. (2009). Update: Any Lessons for Ghana in India’s Jatropha Failure? Ghana Business News. Retrieved from https://www.ghanabusinessnews.com/2009/05/23/update-any-lessons-for-ghana-in-indias-jatropha-failure/.
  9. Duku, M. H., Gu, S., & Hagan, E. B. (2011). A Comprehensive Review of Biomass Resources and Biofuels Potential in Ghana. Renewable and Sustainable Energy Reviews, 15(1), 404–415.  https://doi.org/10.1016/j.rser.2010.09.033.CrossRefGoogle Scholar
  10. Energy Commission. (2006). Strategic National Energy Plan (2006–2020). Ghana.Google Scholar
  11. Faaij, A. (2008). Bioenergy and Global Food Security, Utrecht, Berlin: Wissenschaftlicher Beirat der Bundesregierung Globale Umweltverssnderungen (WBGU). Retrieved from http://www.wbgu.de/wbgu_jg2008_ex03.pdf.
  12. FAO. (2010). Bioenergy and Global Food Security: The BEFS Analytical Framework. Rome.Google Scholar
  13. Government of Ghana. (2010). Medium-Term National Development Policy Framework: Ghana Shared Growth and Development Agenda (GSGDA), 2010–2013 Volume I Policy Framework Final Draft. Accra. Retrieved September 9, 2017, from http://www.mofep.gov.gh/sites/default/files/docs/mdbs/2010/final_draft_mtdpf.pdf.
  14. Hoque, M. M., et al. (2015). Producer Participation in Biomass Markets: Farm Factors, Market Factors, and Correlated Choices. Journal of Agricultural and Applied Economics, 47(3), 317–344.CrossRefGoogle Scholar
  15. IEA. (2014). Africa Energy Outlook: A Focus on Energy Prospects in Sub Saharan Africa. World Energy Outlook Special Report.Google Scholar
  16. IPCC. (2007). Mitigation of Climate Change: Contribution of Working Group III to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change.Google Scholar
  17. Lee, W. (2006). Propensity Score Matching and Variations on the Balancing Test. In 3rd Conference on Policy Evaluation 27–28 October. Mannheim (Germany).Google Scholar
  18. Ministry of Energy. (2010). National Energy Policy, Republic of Ghana. Retrieved from http://old.sheltercentre.org/shelterlibrary/items/pdf/Ghana.pdf.
  19. Rosenbaum, P. R., & Rubin, D. B. (1983). The Central Role of the Propensity Score in Observational Studies for Causal Effects. Biometrika, 70(1), 41–55.CrossRefGoogle Scholar
  20. Rosenbaum, P. R., & Rubin, D. B. (1985). Constructing a Control Group Using Multivariate Matched Sampling Methods that Incorporate the Propensity Score. The American Statistician, 39(1), 33–38.Google Scholar
  21. Schoneveld, G. C., German, L. A., & Nutakor, E. (2010). Towards Sustainable Biofuel Development: Assessing the Local Impacts of Large-Scale Foreign Land Acquisitions in Ghana. In World Bank Land Governance Conference. Retrieved from http://siteresources.worldbank.org/EXTARD/Resources/336681-1236436879081/5893311-1271205116054/schoneveld.pdf.
  22. Sianesi, B. (2004). An Evaluation of the Swedish System of Active Labor Market Programs in the 1990s. Review of Economics and Statistics, 86(1), 133–155.CrossRefGoogle Scholar
  23. Tauli-Corpuz, V., & Tamang, P. (2007). Oil Palm and Other Commercial Tree Plantations, Monocropping: Impacts on Indigenous Peoples’ Land Tenure and Resource Management Systems and Livelihoods. In United Nations Permanent Forum on Indigenous Issues (UNPFII), 6th session, New York, 14–25 May 2007.Google Scholar
  24. UNDESA. (2005). The Millennium Development Goals Report 2005. Retrieved from http://unstats.un.org/unsd/mi/pdf/mdg book.pdf.
  25. World Bank. (2009). Making Development Climate Resilient: A World Bank Strategy for Sub-Saharan Africa. Washington, DC.Google Scholar
  26. Yamane, T. (1967). Statistics: An Introductory Analysis (2nd ed.). New York: Harper & Row.Google Scholar

Copyright information

© The Author(s) 2018

Open AccessThis book is licensed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

The images or other third party material in this chapter are included in the chapter's Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the chapter's Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder.

Authors and Affiliations

  • Lauretta S. Kemeze
    • 1
  • Akwasi Mensah-Bonsu
    • 1
  • Irene S. Egyir
    • 1
  • D. P. K. Amegashie
    • 1
  • Jean Hugues Nlom
    • 2
  1. 1.University of GhanaAccraGhana
  2. 2.University of MarouaMarouaCameroon

Personalised recommendations