Bt Cotton and Employment Effects for
Female Agricultural Laborers in
Pakistan: An Application of DoubleHurdle Model
599
Shahzad Kouser1, Matin Qaim2 and Abedullah3
1COMSATS
Institute of Information Technology, Islamabad, Pakistan
2Georg-August-University of Goettingen, Germany
3International Livestock Research Institute, Islamabad, Pakistan
Contents
 Introduction
• Motivation
– Existing literature
– Research gaps
• Research objectives
 Data collection and descriptive statistics
 Econometric model
 Econometric analysis
 Conclusions and recommendations
Shahzad Kouser
2
Introduction
• Bt cotton – Bacillus thuringiensis
• Bollworms infest 88% of global cotton area (James, 2002)
• 15% of world cotton loses due to pests = $3b + $1.7b (EJF, 2007)
Cry crystal
• Monsanto commercialized Bt cotton in USA in 1996
• Bt successfully planted on 247 million acres (James, 2014)
• Pakistan officially approved Bt in 2010
• 4th largest Bt producer in 2011 (81% adoption)
• 1.3 million cotton farmers and millions of landless rural households
• On-farm earning is vital source of income for non-farm households
• Employment links with innovation in agricultural technology
Shahzad Kouser
3
Motivation
Existing literature
• Bt technology has reduced crop damage, pesticide use, and increased yield and
farm income (Ali and Abdulai, 2010; Kathage and Qaim, 2012; Nazli et al.,
2012; Pray et al., 2002; Qaim et al., 2006, Klumper and Qaim, 2014)
• Few studies estimated through descriptive statistics that Bt has increased labor
cost and returns on labor (Pray et al., 2001; Kouser and Qaim, 2013)
Research gaps
• No study quantified on-farm employment effects of Bt adoption
• Bt may have gender implications as females mainly do picking
• Subramanian and Qaim (2010) - simulation of a self selected Indian village
Shahzad Kouser
4
Research Objectives
Estimate the spillover effects of Bt cotton adoption
on demand for total hired labor particularly for
female and male hired labor in Pakistan
Shahzad Kouser
5
Data Collection
• Farm survey in Punjab province in 2010
• Punjab comprises 80% of cotton area
• 4-stage sampling procedure
• 4 major cotton grwing districts
–
–
–
–
Bahawalnagar
Bahawlapur
Rahim Yar Khan
Vehari
• Total sample of 352 farmers
• 248 Bt and 104 non-Bt adopters
Shahzad Kouser
6
Table 1: Descriptive statistics of sample farmers and farms by production technology
Bt adopters
(N = 248)
Non-adopters
(N = 104)
Age (years)
40.557
42.442
Education (years of schooling)
8.044**
6.769
7.295
6.636
14.604***
8.837
9.116
8.067
Credit constrained (%)
27.016***
90.385
Off-farm employment (%)
41.532***
58.654
Bt awareness exposure (years)
4.206***
1.837
Variables
Household size (adult equivalent)
Total area owned (acres)
Cotton area (acres)
***, **,*
Mean values are significantly different at the 1%, 5%, and 10% level, respectively.
Shahzad Kouser
7
40
35
30
25
20
15
10
5
0
Hired total labor days
Female hired labor days
Bt
Male hired labor days
Non-Bt
Figure 1: Hired labor days per acre by gender and production technology
Shahzad Kouser
8
Double-Hurdle Model
• Bt impact on demand for hired labor days may be a two-step decision
• Farmers may initially decide whether to hire or not (first hurdle)
𝑑ℎ𝑖∗
= 𝛾𝑥𝑖 + 𝜇𝑖 :
𝜇𝑖 ∼ 𝑁 0, 1
and
1 𝑖𝑓 𝑑ℎ𝑖∗ > 0
𝑑ℎ𝑖 =
0 𝑜𝑡ℎ𝑒𝑟𝑤𝑖𝑠𝑒
• If hiring then deciding how many labor days to hire (second hurdle)
𝑄ℎ𝑖∗
= 𝛽𝑧𝑖 + 𝜈𝑖 :
𝜈𝑖 ∼ 𝑁 0, 𝜎
2
and
𝑄ℎ𝑖∗ 𝑖𝑓 𝑄ℎ𝑖∗ > 0 𝑎𝑛𝑑 𝑑ℎ𝑖 = 1
𝑄ℎ𝑖 =
0 𝑜𝑡ℎ𝑒𝑟𝑤𝑖𝑠𝑒
Shahzad Kouser
9
Double-Hurdle Model
• Cragg (1971) proposed double-hurdle (DH) model
• Likelihood specification of DH model designed by Jones (1989)
𝐿 𝑄ℎ𝑖 |𝑥𝑖 , 0 =
1 − Φ 𝛾 𝑥𝑖 𝜎𝜇
𝑄ℎ𝑖 =0
×
Φ 𝛽𝑧𝑖 𝜎𝑣
𝜙 𝑄ℎ𝑖 − 𝛽𝑧𝑖 𝜎𝑣
𝜎𝑣 Φ 𝛽𝑧𝑖 𝜎𝑣
∗
Φ 𝛾 𝑥𝑖 𝜎𝜇 Φ 𝛽𝑧𝑖 𝜎𝑣
𝑄ℎ𝑖 >0
• Control function approach is employed to test possibility of endogeneity
in Bt variable
Shahzad Kouser
10
Table 2: Results of reduced form equation
Bt adoption
Coefficient
Standard error
Bt awareness exposure (years)
0.153***
0.036
Credit constraint (dummy)
-0.634***
0.138
Market distance (km)
-0.051***
0.009
Total area owned (acres)
-0.005
0.005
Cotton area (acres)
-0.007
0.004
Wage rate (Rs/day)
-0.001
0.001
-0.022***
0.003
Price of insecticide (Rs/liter)
0.000*
0.000
Total irrigation (No.)
0.024
0.016
Off-farm employment (dummy)
-0.224*
0.132
Household size (adult equivalent)
0.031**
0.015
Price of fertilizer (Rs/kg)
***, **,*
Significant at the 1%, 5%, and 10% level, respectively.
Note: Farmer’s age, education and district variations are controlled.
Shahzad Kouser
11
Table 3: Model specification tests (Tobit vs DH model)
Likelihood ratio tests
Total hired
labor days
Hired female Hired male
labor days
labor days
Log-likelihood of Tobit regression
-1957.175
-1937.250
-1311.795
Log-likelihood of Probit regression
-141.548
-142.769
-255.343
Log-likelihood of Truncated regression
-1698.002
-1742.610
-1021.150
χ2 (16)
235.249
103.744
70.605
p-value
0.000
0.000
0.000
Shahzad Kouser
12
Table 4: Determinants of labor demand (Double-hurdle model)
Hired total labor
days
Variables
Hurdle 1
Hired female
labor days
Hurdle 2 Hurdle 1 Hurdle 2
Hired male labor
days
Hurdle 1 Hurdle 2
Predicted Bt adoption
1.308*** 14.261*** 1.234*** 13.328*** 0.990***
2.506
Total area owned
0.019**
0.103*
0.019**
0.119*
0.002
0.012
Cotton area
-0.006
0.210***
-0.005
0.144**
-0.005
0.095***
Wage rate
-0.007*** -0.139*** -0.007*** -0.159*** -0.003***
-0.002
Price of fertilizer
-0.003
-0.036
-0.004
-0.030
-0.002
-0.006
Price of insecticide
0.000
0.003
0.000
0.004
-0.000
-0.001
Cotton price
0.000
-0.003
0.000
-0.003
0.000
-0.001
Total irrigation
-0.003
-
-0.004
-
-0.030*
-
-
0.082***
-
0.065**
-
0.027***
Off-farm employment
0.393**
5.032**
0.408**
5.806***
0.005
0.591
Household size
-0.010
-0.369
-0.008
-0.414*
0.001
-0.071
Crop length
***, **,*
Significant at the 1%, 5%, and 10% level, respectively.
Note: Farmer’s age, education and district variations are controlled.
Shahzad Kouser
13
Table 5: Marginal effects for double-hurdle model
Hired total labor
days
Variables
Hired female
labor days
Hired male labor
days
Hurdle 1 Hurdle 2 Hurdle 1 Hurdle 2 Hurdle 1 Hurdle 2
Predicted Bt adoption
0.193*** 10.505*** 0.184***
Total area owned
0.003**
0.076
Cotton area
-0.001
0.155**
Wage rate
8.658**
0.270***
1.650*
0.003*
0.077
0.001
0.008
-0.001
0.094**
-0.001
0.063***
-0.001*** -0.103*** -0.001*** -0.103*** -0.001***
-0.002
Price of fertilizer
-0.001
-0.027
-0.001
-0.019
-0.001
-0.004
Price of insecticide
0.000
0.002
0.000
0.003*
-0.000
-0.001
Cotton price
0.000
-0.002
0.000
-0.002
0.000
-0.000
Total irrigation
-0.000
-
-0.001
-
-0.008*
-
-
0.060**
-
0.042**
-
0.018**
Off-farm employment
0.058*
3.707**
0.061**
3.771***
-0.001
0.389
Household size
-0.002
-0.272*
-0.001
-0.269*
0.000
-0.047
Crop length
***, **,*
Significant at the 1%, 5%, and 10% level, respectively.
Note: Farmer’s age, education and district variations are controlled.
Shahzad Kouser
14
Table 6: Unconditional marginal effects of Bt cotton on labor demand
Unconditional
expected demand for
labor days in non-Bt
cotton
Unconditional
average marginal
effects
25
13.713***
210.98
Hired female labor days
21
11.056***
(53%)
166.61
Hired male labor days
5
2.890***
45.21
Hired total labor days
Total income
benefits
(US$ million)
***, **,*
Significant at the 1%, 5%, and 10% level, respectively.
Note: Values in parentheses represent percent increase
Shahzad Kouser
15
Conclusions and Recommendations
• Bt adoption has increased income benefits for rural
laborers and for resource poor women in particular
• Potential for benefit exploitation through seed purity
• Address farmers financial constraints
• Better extension and information system
• Future research using Bt toxin concentration
• Need to explore resource allocation
Shahzad Kouser
16
References
Ali, A. and A. Abdulai 2010), ‘The adoption of genetically
modified cotton and poverty reduction in Pakistan’,
Journal of Agricultural Economics 61(1): 175-192.
Klümper W, Qaim M (2014) A Meta-Analysis of the
Impacts of Genetically Modified Crops. PLoS ONE
9(11): e111629. doi:10.1371/journal.pone.0111629
Cragg, J.G., 1971. Some statistical models for limited
dependent variables with application to the demand for
durable goods. Econometrica. Journal of the Econometric
Society 1, 829–844.
Kouser, S., Qaim, M.. 2013. Valuing financial, health, and
environmental benefits of Bt cotton in Pakistan.
Agricultural Economics, 44(3): 323–335.
EJF, 2007. The deadly chemicals in cotton. Environmental
Justice Foundation (EJF) in collaboration with Pesticide
Action Network UK. London, UK.
James, C., 2002. Global Review of Commercialized
Transgenic Crops: 2001 Feature: Bt Cotton. ISAAA Brief
No. 26, Ithaca, NY: ISAAA. International Service for the
Acquisition of Agri-biotech Applications, Ithaca, New
York..
Jones, A.M., 1989. A double hurdle model of cigarette
consumption. Journal of Applied Econometrics, 4(1): 23–
39.
Kathage, J., and Qaim, M., 2012. Economic impacts and
impact dynamics of Bt (Bacillus thuringiensis) cotton in
India. Proceedings of the National Academy of Sciences
USA, 109(29): 11652-11656.
Nazli, H., D. Orden, R. Sarker, and K. Meilke (2012), ‘Bt
cotton adoption and wellbeing of farmers in Pakistan’,
contributed paper at the 28th Conference of the
International Association of Agricultural Economists
(IAAE), 18-24 August, Foz do Iguacu, Brazil.
Pray, C., J. Huang, R. Hu, and S. Rozelle (2002), ‘Five
years of Bt cotton in China–the benefits continue’, The
Plant Journal 31(4): 423-430.
Pray, C., Ma, D., Huang, J., Qiao, F., 2001. Impact of Bt
cotton in China. World Development, 29(5): 813–825.
Qaim, M., A. Subramanian, G. Naik, and D. Zilberman
(2006), ‘Adoption of Bt cotton and impact variability:
Insights from India’, Review of Agricultural Economics
28(1): 48-58.
Subramanian, A., Qaim, M., 2010. The impact of Bt
cotton on poor households in rural India. The Journal of
Development Studies, 46(2): 295–311.
Shahzad Kouser
17
Thank You
Shahzad Kouser
18