- Email: [email protected]
Fifteen Years of Bt Cotton in China: The Economic Impact and its Dynamics
World Development Vol. 70, pp. 177–185, 2015 0305-750X/Ó 2015 Elsevier Ltd. All rights reserved. www.elsevier.com/locate/worlddev
http://dx.doi.org/10.1016/j.worlddev.2015.01.011
Fifteen Years of Bt Cotton in China: The Economic Impact and its Dynamics FANGBIN QIAO* China Economics and Management Academy, Central University of Finance and Economics, Beijing, China Summary. — Even though the economic benefit of Bt cotton adoption in the short-run has been well documented, the dynamics of this benefit remain unclear. In particular, the possibility of pest resistance build-up and secondary pest outbreaks has caused concern regarding the sustainability of this economic benefit in the long run. Hence, this study analyzes the economic impact of Bt cotton and its dynamics in China. Using nationally representative long panel data for 1997–2012, we show that this economic benefit continues 15 years after the commercialization of Bt cotton. Ó 2015 Elsevier Ltd. All rights reserved. Key words — economic benefit, Bt cotton, sustainability, China
1. INTRODUCTION
gene would die away, some farmers had switched to varieties with stacked Bt genes (James, 2013). Whether the short-run effect of Bt cotton adoption is sustainable remains a question for two notable reasons. First, most of the previous studies analyze data collected in the first few years after the commercialization of Bt cotton in a country. However, an increase in pesticide use owing to pest resistance build-up and/or secondary pest outbreaks may not appear in the short run (Qiao, Wilen, & Rozelle, 2008). Therefore, these studies may not provide satisfactory answers on the long-term impact or the sustainability of the economic benefit of Bt cotton. 1 Second, all previous studies were based on a small sample of household survey data. Even though there are more than ten million cotton farmers, these surveys usually include only data on a few hundred of them (e.g., Huang et al., 2003; Kathage & Qaim, 2012). Moreover, although all the sampled households are from the major cotton-producing regions, the estimation results based on the data set might be biased. In this study, we address this shortcoming by using nationally representative panel data over 15 years. We choose China as the focus of our study for two reasons. First, prior to 2006, China was the largest country to plant Bt cotton, and subsequently became the second largest country (James, 2013). 2 More importantly, Bt cotton was commercialized in China early in 1997. In other words, Bt cotton has presently been commercialized for over 15 years. Furthermore, nationally representative data on the detailed information of input and output of cotton production have been surveyed, recorded, and published annually. Second, the sustainability of the effect of Bt cotton adoption is still a hot topic in China. In fact, there has been widespread opposition from the public regarding Bt cotton adoption, particularly after a secondary pest outbreak in some Bt cotton fields in the early 2000s. Scientists and researchers who supported genetically modified (GM) crops were labeled “traitors” (Economist, 2013). Faced by increasing criticism against GM technology, the Chinese government delayed the
The short-run economic benefit of Bacillus thuringiensis (Bt) cotton has been well documented (e.g., Carpenter, 2010; Qaim, 2003; Stone, 2011; Wossink & Denaux, 2006). Empirical studies in China showed that the expenditure on pesticide decreased by more than two-thirds after Bt cotton adoption (Huang, Hu, Pray, Qiao, & Rozelle, 2003; Huang, Hu, Rozelle, Qiao, & Pray, 2002; Pray, Ma, Huang, & Qiao, 2001). Furthermore, the reduction of chemical pesticide use not only increased the yield and net profit of cotton farmers, but also contributed to a cleaner environment and improved the health of farmers (Hossain, Pray, Lu, Huang, & Hu, 2004; Kouser & Qaim, 2011). Because of its high profitability, Bt cotton was almost exclusively adopted in the North China Plain only a few years after its commercialization in 1997 (James, 2013). On the other hand, concerns regarding the long-run effect of Bt technology continue. In fact, the debate on the advantages and disadvantages of this technology began even before the commercialization of Bt crops. It was expected that the widespread adoption of Bt crops would lead to pests’ developing resistance to Bt toxin and/or secondary pest outbreaks. As a result, the total pesticide use would be gradually restored to the level before the adoption of Bt cotton, and the short-run economic benefit of Bt cotton would be completely offset in the long run (e.g., Pemsl & Waibel, 2007; Wang, Just, & Pinstrup-Anderson, 2008). In recent years, this negative attitude toward Bt cotton has often seemed to dominate the public debate in the news and media (Cleveland & Soleri, 2005; Kathage & Qaim, 2012). For example, in China, Bt technology had even been described as a “weapon” that developed countries used to attack developing countries (Jiang & Li, 2010). These concerns have important impact on agricultural production practices. To mitigate the development of pest’s resistance, farmers were required to plant refuges of non-Bt cotton in almost all the countries where Bt cotton is planted, as suggested by the entomologists and ecologists (Bates, Zhao, Roush, & Shelton, 2005; Gould, 1998; Tabashnik et al., 2003). Because pests adapted to one toxin may still be susceptible for another toxin, cotton varieties with stacked Bt genes are more efficient in pest control (Gould, 1998). Due to the worry that the economic benefit of varieties with a single Bt
* The authors acknowledge the financial supports of this study from the National Natural Science Foundation of China (71273290). Final revision accepted: January 24, 2015. 177
WORLD DEVELOPMENT
Data used in this study are primarily obtained from the All China Data Compilation of the Costs and Returns of Main Agricultural Products (Quanguo Nongchanpin Chengben Shouyi Ziliao Huibian – in Chinese pinyin) and China Statistical Yearbook (Zhongguo Tongji Nianjian – in Chinese pinyin). Specifically, the data regarding expenditure on pesticides, labor use, fertilizer use, seed cost, and other material cost, as well as cotton yield are obtained from the All China Data Compilation of the Costs and Returns of Main Agricultural Products (National Development and Reform Commission, various years). 3 However, the data on national total cotton sown area, provincial cotton sown areas, and price indexes are obtained from the China Statistical Yearbook (National Bureau of Statistics of China (NBSC), various years). Due to lack of national statistics, the percentages of Bt cotton in different provinces are mainly obtained from the Center for Chinese Agricultural Policy (CCAP), Chinese Academy of Sciences. 4 China has three major cotton-producing regions: the Yellow River valley, the Yangtze River valley, and the Northwest (Hsu & Gale, 2001). The Northwest region includes primarily the Xinjiang Uyghur autonomous region, which has been the largest cotton-producing province in China since the mid1990s. Another important cotton-producing province in the Northwest region is the Gansu province, which is the twelfth largest cotton-producing province in China (NBSC, 2013). The Yellow River valley is China’s largest cotton-producing region. In this study, five provinces from the Yellow River valley—Shandong, Hebei, Henan, Shaanxi, and Shanxi—are included. In addition, five provinces from the Yangtze River valley—Hubei, Anhui, Hunan, Jiangsu, and Jiangxi—are included. These 10 provinces are also the second to thirteenth largest cotton-producing provinces in China (NBSC, 2013). 5 The total cotton sown area of these 12 provinces included in this study is 4.59 million ha, which is 97.80% of the national total sown area. The basis characteristics of the variables used in this study is shown in Table 1. As shown in Figure 1, Bt cotton adoption in the major cotton-producing provinces was rapid, but significantly different in many aspects. Because of the warm and dry climate conditions and yield damage caused by cotton bollworm, Bt cotton adoption in the Yellow River valley is significantly higher than that in the other two regions (Wu & Guo, 2005). For example, yield loss caused by cotton bollworm in the Hebei province in 1992 was nearly 40% (Ministry of Agriculture, 1990–1999). Hence, Bt cotton was first commercialized in the Yellow River valley in 1997 (Hsu & Gale, 2001). As this was very successful, Bt cotton adoption then rapidly spreads to the Yangtze River
Variable names
Mean
Standard deviation
Yield (kg/mu) Bt cotton adoption (%) Pesticide cost (yuan/mu) Seed cost (yuan/mu) Labor (day/mu) Fertilizer use (yuan/mu) Other cost (yuan/mu)
142.69 32.05 67.66 36.36 35.30 610.77 438.39
38.90 41.63 39.25 18.20 12.59 574.11 121.72
Note: 1 ha = 15 mu; 1 USD = 6.25 Chinese Yuan.
Spread of Bt cotton in China
100
2. DATA AND THE IMPACT OF BT COTTON ADOPTION IN CHINA
Table 1. Basic characteristics of variables
Percentage of Bt cotton(%) 60 20 40 80
commercialization process of other Bt crops (Jiang & Li, 2010). For example, even though GM rice technology has been mature for years, the Chinese government has no plan or schedule for its commercialization. The remainder of this paper is organized as follows. The next section discusses the data used in this study. A descriptive analysis of the economic benefit of Bt cotton and its dynamics is presented by showing the quantities of pesticide cost, seed cost, labor use, and cotton yield and their dynamics since Bt cotton adoption. To isolate the impact of Bt technology and its dynamics, econometric models are set up in the third section, and the estimation results are discussed in the fourth section. The final section concludes the paper.
0
178
1996
2000 Hebei Anhui
2004 (Year) Shandong Hubei
2008 Henan Xinjiang
2012 Jiangsu Gansu
Figure 1. Spread of Bt cotton in China.
valley. As shown in Figure 1, only Bt cotton was grown in the Yellow River and Yangtze River valleys within a few years after its commercialization. On the other hand, because of their dry and hot climate, the percentage of Bt cotton adoption in Xinjiang and Gansu provinces was relatively small until 2004. Figure 2 shows the dynamics of inputs and cotton yield with the widespread adoption of Bt cotton. As shown in Panel A, the pesticide cost decreased significantly after Bt cotton adoption. Similarly, labor use also decreased significantly from 1997 onward (Panel B). Simultaneously, however, both seed cost and yield increased (Panels C and D). In order to show the changes of inputs and yields before and after the Bt cotton adoption, we compared the inputs and yields of three time periods: 1990–96, 1997–2003, and 2004–12. The first time period comprises the years before Bt cotton adoption. In the second time period, Bt cotton cultivation spread from the Yellow River valley to the Yangtze River valley (early adoption period hereafter). In the third time period, only Bt cotton was grown in the Yellow River and Yangtze River valleys and Bt cotton adoption began to increase significantly in the Northwest region (late adoption period hereafter). In fact, 2004 is also the year when the concern regarding the sustainability of the economic benefit of GM technology became the center of public debate. Table 2 shows that with the spread of Bt cotton, pesticide use decreased significantly (column 1). Moreover, as observed in column 2, the pesticide cost is 93.71 yuan/mu during 1990– 96, which decreased to 74.45 yuan/mu during 1997–2003 and to 60.54 yuan/mu during 2004–12 (row 1 to row 3). After dividing the entire sample into three major regions, we found that the reduction in pesticide use mainly comes from the
FIFTEEN YEARS OF BT COTTON IN CHINA: THE ECONOMIC IMPACT AND ITS DYNAMICS
179
Labor use
0
0
50
20
(day/mu) 40
(yuan/mu) 100 150
60
200
80
250
Pesticide cost
1992
1996
2000 (Year)
2004
2008
1992
2012
1996
Panel A
2000 (Year)
2004
2008
2012
2004
2008
2012
Panel B
Seed cost
0
50
20
100
(kg/kg) 150
(yuan/mu) 40 60
200
80
250
100
Yield
1992
1996
2000 (Year)
2004
2008
2012
1992
1996
Panel C
2000 (Year)
Panel D
Figure 2. Dynamics of pesticide cost, seed cost, labor cost, and yield after Bt cotton adoption in China.
Table 2. Input and output of cotton production in China before and after Bt cotton adoption Bt cotton (%)
Pesticide cost (yuan/mu)
Labor use (day/mu)
Seed cost (yuan/mu)
Yield (kg/mu)
China 1990–96 1997–2003 2004–12
0.00 27.60 80.73
93.71 74.45 60.54
42.49 32.43 25.45
24.99 37.94 55.21
123.25 147.31 169.75
Yellow River valley 1990–96 1997–2003 2004–12
0.00 54.65 96.44
127.12 73.47 47.70
38.50 31.20 25.28
29.55 43.54 50.45
88.11 129.91 145.52
Yangtze River valley 1990–96 1997–2003 2004–12
0.00 13.69 85.81
79.69 96.85 81.72
49.83 37.48 28.88
16.47 28.74 60.29
147.08 145.33 173.77
Northwest 1990–96 1997–2003 2004–12
0.00 1.48 30.51
25.58 22.23 38.27
25.78 22.90 17.28
44.59 47.90 53.86
155.15 190.65 217.61
Note: 1 ha = 15 mu; 1 US Dollar = 6.2 Chinese Yuan.
180
WORLD DEVELOPMENT
Yellow River valley. The pesticide cost during 2004–12 is 47.7020yuan/mu, which is less than 40% of that during 1990–96 (row 4 to row 6). A similar trend is observed for labor use over time. As shown in column 3, the labor use during 2004–12 is 25.45 day/mu, while that during 1990–96 is 42.49 day/mu. In other words, the labor use during the late adoption period is only 60% of that before the Bt cotton adoption. It is worth noting that this decrease in labor use can be attributed to the reduction in the time spent for pesticide spraying and picking up large cotton bollworms. In practice, China’s cotton farmers not only use pesticides to control cotton bollworms, but also manually pick them up, particularly large ones, in fields. The decrease in labor use caused by Bt cotton adoption has also been shown in other studies (e.g., Elbehri & Macdonald, 2004; Subramanian & Qaim, 2009). On the other hand, Bt cotton adoption led to an increase in seed cost and cotton yield. The seed cost increased from 24.99 yuan/mu before the Bt cotton adoption to 37.97 yuan/mu during the early adoption period and 55.21 yuan/mu during the late adoption period (column 4). This is as expected, as the price of Bt cotton is much higher than that of non-Bt cotton. As shown in the last column, cotton yield also consistently increased by more than one-third (123.25 kg/mu during 1990–96 vs. 169.75 kg/mu during 2004–12). In summary, Table 2 shows that Bt cotton adoption has successfully led to a decrease in pesticide use and labor use and an increase in cotton yield. Even though seed cost more than doubled after Bt cotton adoption, it still cannot offset its positive economic benefit (in fact, the increase in seed cost is similar to the decrease in pesticide use). More importantly, Table 2 shows that the economic benefit of GM technology did not diminish; rather, it remained stable and continuous during the late adoption period. These observed differences provide interesting insights into the impact of Bt technology and its dynamics. However, they cannot be interpreted as net impacts of the Bt technology, because confounding factors have to be controlled for. In the following section, we isolate the impact of Bt technology by setting up econometric models and estimate multivariate regressions. 3. REGRESSION MODELS The above results might be biased owing to the presence of confounding factors. To isolate the impact of Bt cotton adoption, we need to develop and estimate econometric models. Specifically, the yield function can be written as follows: Yield i;t ¼ a0 þ a1 Bti;t þ a2 Bti;t D2004–2012 þ a3 Inputsi;t þ a4 Yeart þ a5 Provincei þ ei;t
ð1Þ
In Eqn. (1), Yield represents the cotton yield and Bt denotes the percentage of Bt cotton. 6 The Bt variable indicates whether or not the Bt technology has a positive net effect on yield. For example, if the estimated coefficient of Bt in the yield function is positive and statistically significant, then the impact of Bt technology on cotton yield is positive, and vice versa. In addition, we include an interaction term, Bt*D2004–2012, to consider the dynamics of the impact of Bt cotton adoption. D2004–2012 is a dummy variable, which is 1 for the year 2004– 12 and 0 otherwise. If the estimated coefficient of Bt is positive and that of Bt*D2004–2012 is significantly positive, then the
impact of Bt cotton adoption on yield increased over time. On the other hand, if the estimated coefficient of Bt*D2004–2012 is statistically insignificant, then the impact of Bt technology remained stable over time. The input variables, Inputs, in the yield function include fertilizer, pesticide, seed, labor, and other inputs. As there are no quantitative data for seed and pesticide use, expenditures on seed and pesticides are used. To consider the impact of inflation, all the expenditures were deflated by the rural consumer price index. Similarly, other material cost, which is used to consider the impact of all other material inputs, was also adjusted using the rural consumer price index. Fortunately, however, quantitative data for fertilizer use (kg/mu) and labor use (day/mu) are available. Year is a vector of year dummies, which is added to consider the impact of variation each year. Similarly, Province is a vector of provincial dummies. Adding Year and Province variables makes our model a two-way fixed-effect model. Finally, subscript i is the ith province, subscript t is the year, and e is the error term. However, OLS estimation of Eqn. (1) might be biased, as Bt technology adoption affects the pesticide cost, seed cost, and labor use. To solve this problem, we need to exclude the impact of Bt cotton adoption on these inputs. After excluding the net impact of Bt cotton adoption, we then added the predicted values of these inputs, into Eqn. (1). In other words, Eqn. (1) was changed as follows: Inputsi;t ¼ b0 þ b1 Bti;t þ b2 Bti;t D2004–2012 þ b3 Yeart þ b4 Provincei þ ei;t Yield i;t ¼ c0 þ c1 Bti;t þ c2 Bti;t D2004–2012 þ c3 Inputs pi;t;Bt þ c4 Fertilizeri;t þ c5 Otheri;t þ c6 Yeart þ c7 Provincei þ fi;t
ð2Þ
In Eqn. (2), the predicted values, Inputs_pi,t,Bt, are the values of inputs (i.e., pesticide cost, labor use, and seed cost) showing the net impact of Bt cotton adoption. Fertilizer is the fertilizer use per unit of land, while Other represents the other material cost. All the other variables are as discussed above. Finally, Bt technology and pesticide use are special inputs, which are different from traditional inputs like fertilizer and labor. Thus, instead of running an OLS regression, we run the Weibull damage abatement function (Headley, 1968; Lichtenberg & Zilberman, 1986). This method has also been widely used in previous studies (e.g., Huang et al., 2003; Qaim & de Janvry, 2003). In this study, the Weibull damage abatement function that we estimate is as follows: Yield i;t ¼ ðk0 þ k1 Yangtzei;t þ k2 Northwesti;t Þ Labor ph1 Fertilizerh2 Seed ph3 Otherh4 ð1 ePesticide
pm
Þ
ð3Þ
In the above equation, Yangtze is a dummy variable, which equals to 1 if the observations are from the Yangtze River valley region. Similarly, Northwest is the Northwest region dummy. m is the coefficient to be estimated to capture the impact of pesticide use on the yield. In this study, we have two scenarios to define m. In the first scenario, m is defined as follows: m ¼ h0 þ h1 Bti;t þ h2 Bti;t D2004–2012
ð4Þ
FIFTEEN YEARS OF BT COTTON IN CHINA: THE ECONOMIC IMPACT AND ITS DYNAMICS
In this scenario, the impact of Bt cotton and its dynamics in all the three major cotton-producing regions (i.e., the Yellow River valley, the Yangtze River valley, and Northwest) are assumed to be the same. In the second scenario, we consider the differences in the impact of Bt cotton adoption and its dynamics for each cotton-producing region. Under this scenario, m is defined as follows: m ¼ h0 þ h1 Bti;t þ h2 Bti;t D2004–2012 þ h3 Bti;t
Inputsi;t ¼ /0 þ /1 Bti;t þ /2 Bti;t D2004–2012 þ /3 Bti;t Yangtzei;t þ /4 Bti;t D2004–2012 Yangtzei;t þ /5 Bti;t Northwest þ /6 Bti;t D2004–2012 Northwest þ /7 Yeart þ /8 Provincei þ li;t
Yangtzei;t þ h4 Bti;t D2004–2012 Yangtzei;t þ h5 Northwesti;t
ð5Þ
Accordingly, under the scenario, the input function is as follows:
Table 3. Impact of Bt cotton adoption and its dynamics in China, 1997–2012
Share of Bt cotton*2004–12 year dummy Constant Provincial dummies Year dummies Observations R2
ð6Þ
4. RESULTS AND DISCUSSION
Bti;t Northwesti;t þ h6 Bti;t D2004–2012
Share of Bt cotton
181
Expenditure on pesticide (yuan/mu)
Labor (day/mu)
Seed cost (yuan/mu)
0.4436** (4.21) 0.2847*
0.0750** (4.18) 0.0072
0.4514** (9.37) 0.1296*
(2.31) 85.8181** (7.56) Yes Yes
(0.34) 17.6436** (9.12) Yes Yes
(2.30) 38.7977** (7.48) Yes Yes
185 0.722
185 0.838
185 0.674
Note: Cost is in 2012 year Yuan. The symbols * and ** denote significance at 5% and 1%, respectively.
The results of the econometric estimation of input equations are shown in Tables 3 and 4, while the estimation result of yield is shown in Table 5. In general, most of the regression results are consistent with the descriptive analysis in Section 3. Most estimated coefficients on the control variables have the expected signs and are statistically significant. In the following paragraphs, we first discuss the estimation result of input equations and that of the yield equation later. (a) Impact of Bt technology on pesticide use, labor use, and seed cost, and its dynamics Estimation results show that the adoption of Bt cotton has significantly decreased pesticide use and labor use. As shown in the first column of Table 3, compared to non-Bt cotton, the decrease in pesticide use led to the reduction of expenditure on pesticide by 44.36 yuan/mu, or 50.82% of the pesticide use in 1996, the final year before the adoption of Bt cotton. Similarly, the adoption of Bt cotton led to a reduction of 7.50 day/mu of labor use, or 17.25% of the total labor use in 1996. After considering the regional differences, the negative impact of Bt cotton adoption on pesticide use and labor use is still significant. As shown in Table 4 (column 1), even though the impact of pesticide use in the Yangtze River valley is significantly smaller than that in the Yellow River valley, the decrease in pesticide use in the Yangtze River valley is still substantial. On the other hand, the impact of Bt cotton on labor use in all three major cotton-producing regions showed no difference.
Table 4. Impact of Bt cotton adoption and its dynamics in China’s major cotton region, 1997–2012
Percentage of Bt Percentage of Bt*2004–12 year dummy Yangtze River dummy*percentage of Bt Yangtze River dummy*percentage of Bt*2004–12 year Northwest dummy*percentage of Bt Northwest dummy*percentage of Bt*2004–12 year Constant Provincial dummies Year dummies Observations R2 Note: Cost is in 2012 year Yuan. The symbols
*
and
**
Expenditure on pesticide (yuan/mu)
Labor (day/mu)
Seed cost (yuan/mu)
0.9303** (7.42) 0.1047 (0.78) 0.6637** (2.94) 0.0949 (0.46) 3.0191 (1.51) 3.5121 (1.80) 80.4646** (5.06) Yes Yes
0.0569* (2.38) 0.0065 (0.26) 0.0089 (0.21) 0.0112 (0.29) 0.0109 (0.03) 0.0583 (0.16) 19.3025** (6.37) Yes Yes
0.3370** (5.55) 0.1618* (2.50) 0.3267** (2.99) 0.1639 (1.64) 0.2938 (0.30) 0.3656 (0.39) 46.6728** (6.05) Yes Yes
185 0.779
185 0.840
185 0.710
denote significance at 5% and 1%, respectively.
182
WORLD DEVELOPMENT Table 5. Impact of Bt cotton adoption on cotton yield in China, 1997–2012 Scenario 1 Coefficient
Constant Yangtze river dummy Northwest dummy Fertilizer use (yuan/mu) Labor (day/mu) Seed cost (yuan/mu) Other cost (yuan/mu) Percentage of Bt cotton Percentage of Bt cotton*2004–12 year Yangtze river dummy*percentage of Bt cotton Yangtze*percentage of Bt*2004–12 year Northwest dummy*percentage of Bt cotton Northwest*percentage of Bt*2004–12 year Observation R2 Note: The symbols
*
17.4513 4.2766 11.7255 0.0173 0.2165** 0.0536** 0.2467** 0.0013** 0.0001
185 0.62 *
and
**
Scenario 2 T-value
Coefficient
T-value
2.09 1.85 1.92 1.45 2.93 2.76 3.29 3.55 0.20
16.4812 4.7158 11.3691 0.0143 0.2238** 0.0502* 0.2481** 0.0014* 0.0001 0.0006 0.0004 0.0001 0.0004
1.94 1.71 1.81 1.08 2.93 2.45 3.20 2.56 0.39 0.77 0.66 0.02 0.06
185 0.62
denote significance at 5% and 1%, respectively.
More interestingly, the significant impact of Bt cotton adoption on pesticide use did not diminish. The estimated coefficient of the interaction term of Bt cotton and 2004–12 year dummy, Bt*D2004–2012, is negative and significant in the pesticide use equation (Table 3, column 1), which implies that farmers spray less pesticide in the late adoption period than in the early adoption period. In other words, the estimation results show that the long-term impact of Bt cotton, with regard to pesticide use, not only did not decrease, but also increased over time. This result is clearly contradictory to the expectations of those who believe that the development of resistance by pests to Bt toxin and increase in secondary pests would completely offset the economic benefit that Bt technology had generated. In fact, the estimation results are not hard to understand. Previous studies had shown that the widespread adoption of Bt cotton had successfully suppressed the cotton bollworm population density regionally; thus, all farmers, including those who planted non-Bt cotton varieties, reduced their pesticide applications (Carrie`re et al., 2003; Wu, Lu, Feng, Jiang, & Zhao, 2008). The reduction of the total pest population density caused a further reduction of pesticide use and thereby increase in the net economic benefit of Bt cotton adoption in the long run. The estimation results also show that labor use reduction does not change over time. As shown in the second column, the interaction term of Bt and D2004–2012 is not statistically significant (Table 3). Similarly, the interaction terms of Bt*D2004– 2012 and the Yangtze River valley and Northwest are also statistically insignificant (Table 4). In other words, these results show that the reduction in labor use owing to Bt cotton adoption is stable during the late adoption period. Finally, as expected, the adoption of Bt cotton led to an increase in seed cost. As shown in Table 3, compared to non-Bt cotton production, seed cost increased to 45.14 yuan/mu, which is 161.91% of that in 1996, when Bt cotton was not commercialized. However, with the increase in the price of non-Bt cotton seeds and the competition in the seed market (Huang, Chen, Mi, Hu, & Osir, 2009), the difference between Bt and non-Bt cotton reduced in the long run (row 2 of Table 3). Similar results are shown in Table 4 (row 2, column 3).
To summarize, the estimation results of the input equations show that the economic benefit of Bt cotton adoption, particularly the pesticide use reduction, did not diminish in the long run. On one hand, pest resistance development and secondary pest outbreaks may offset the economic benefit of Bt cotton adoption. On the other hand, regional reduction of pest population density caused by the widespread usage of Bt cotton may lead to larger economic benefits. Thus, in the long run, the net impact of Bt cotton adoption depends on whether the positive side (i.e., reduction of pest population) or the negative side (i.e., pest resistance development and secondary pest outbreaks) is dominant. The estimation results in this study show that the positive side dominated after 15 years of Bt cotton commercialization. (b) Impact of Bt technology on cotton yield and its dynamics The estimation results of the yield equation are presented in Table 5. As shown in Table 5, estimation results under the two scenarios are very similar, indicating the robustness of the results. The results show that most of the coefficient estimates are as expected and statistically significant. For example, the estimated coefficients of seed cost, labor use, and other material cost are all positive and statistically significant. On the other hand, fertilizer use has no significant impact on yield, which is consistent with previous studies that show the overuse of fertilizer is a common phenomenon in China (e.g., Huang, Hu, Cao, & Rozelle, 2008; Zhang, Huang, Qiao, & Rozelle, 2006). More importantly, the results show that the estimated coefficient of the percentage of Bt cotton is positive and statistically significant under both the scenarios, which implies that Bt cotton adoption has a positive impact on yield (row 8). On the other hand, the estimated coefficient of the interaction term, percentage of Bt cotton*D2004–2012 year dummy, is insignificant, which implies that the impact of Bt cotton adoption on yield remains stable over time (row 9). 7 As shown in Table 5, none of the estimated coefficients of the interaction terms of regional dummy and percentage of Bt cotton (and D2004–2012 year dummy) is statistically significant (row 11 to row 14). In other words, our estimation results show that the impact of Bt cotton adoption on the yield is similar in all cotton-producing regions.
FIFTEEN YEARS OF BT COTTON IN CHINA: THE ECONOMIC IMPACT AND ITS DYNAMICS
183
Table 6. Economic benefit of Bt cotton adoption in China Economic benefit per unit of land (mu) per year 1997–2003
Pesticide use Labor Seed cost Yield
Total benefit since Bt adoption
2004–12
Total
Annual average
Quantity
(%)
Quantity
(%)
Billion yuan
Billion yuan
19.57 1.57 12.31 12.58
22.42 3.61 44.15 10.50
43.99 4.59 27.29 40.42
50.40 10.56 97.88 33.73
4.12 8.70 2.56 22.96
0.26 0.54 0.16 1.44
33.22
2.08
Total Note: % = Quantity saving during 1997–2012/quantity of 1996 (the year before Bt cotton adoption).
(c) Economic benefit of Bt cotton adoption in China After the discussion on the estimation results of the impact of Bt cotton adoption on inputs and yield, we calculate the total impact of Bt cotton adoption in China. To calculate the total impact, we used the estimated coefficients when regional differences are considered, as shown in Tables 4 and 5, and the real percentage of Bt cotton adoption. The results of the calculation are summarized in Table 6. As shown in the first two rows of Table 6, owing to Bt cotton adoption, farmers saved 4.12 billion yuan on pesticide use and 8.70 billion yuan on labor use. On the other hand, the benefit from yield increase is even higher (row 4). As shown in the last row, the total economic benefit that Bt cotton had generated is more than 33 billion yuan over 15 years after Bt cotton was commercialized. 5. CONCLUSION Using nationally representative panel data, I analyzed the economic impact of Bt cotton adoption and its sustainability, after 15 years of its commercialization in China. Consistent with its short-term impact, this study showed that the economic benefit of Bt cotton did not diminish, but remained stable and continuous in China. To the best of our knowledge, this is the first study that uses nationally representative data and focuses on the economic benefit of Bt cotton and its dynamics. The findings of this study have important implications. First, undoubtedly, these findings will contribute to a wider public debate in China as well as in other countries where GM crops are planted. The long-term impact and the sustainability of the economic benefit that GM crops generate have been subject to a heated debate. Those who oppose GM technology believe that such economic benefit has only a short-run
impact, which would diminish in the long run, owing to pest resistance development and secondary pest outbreaks (e.g., Wang et al., 2008). However, the estimation results of this study provide empirical evidence that the economic benefit of GM technology is not limited to the short run, which will definitely help to calm down the public debate. Second, our results are helpful for Chinese policymakers in managing GM crop adoption. Faced by criticism from those who oppose GM technology, the Chinese government has stalled the commercialization of other GM crops, even though billions of dollars have been spent on these crops and China leads the world in GM rice technology. Even though the commercialization of GM rice is affected by many factors, the economic impact and its dynamics are particularly important research topics. We believe that the results of this study will contribute to the management of GM crops by the Chinese government. Finally, findings from this study also contribute to the development and implementation of biotechnology policies in other countries. Even though planting non-Bt cotton as refuge is required in most of the country, economic analysis showed that it might be inappropriate, especially for countries where natural refuge crops are available (Qiao, Wilen, Huang, & Rozelle, 2009). This study provided empirical evidence that China’s non-zero refuge still works after 15 years of Bt cotton commercialization. Hence, policy makers might need to rethink their refuge policies in countries where farmers are required to plant non-Bt cotton as refuge. As shown in this paper, the first generation of cotton varieties with a single Bt gene still can effectively control the bollworm. Even though farmers in some countries have switched from unpatented and royalty fee cotton varieties with a single Bt gene to patented cotton varieties with stacked genes, rigorous analysis is needed to answer whether this switch is economical or a result caused by many factors (Arza & van Zwanenberg, 2014).
NOTES 1. To the best of our knowledge, the only exceptions are Kathage and Qaim (2012) and Huang et al. (2010). Using data collected during 2002–08 in India, Kathage and Qaim estimated the economic impact and impact dynamics of Bt cotton. However, as Bt cotton was first commercially released in India in 2002, it had only been planted for seven years when they conducted their final survey. In this sense, the results of their study may have captured the mid-term impact rather than the long-term impact. For the same reason, Huang et al. (2010), whose study is based on data collected during 1999–2007, cannot be said to have focused on the long-term impact.
2. Since 2006, India is the largest country where Bt cotton is planted. 3. Since 1953, the National Development and Reform Commission began to conduct household surveys to understand the costs and returns for each of the major crops. For cotton, the sample counties were selected based on sown area and production in all major cotton production provinces, while households were randomly selected in sample counties. The households were revisited in the following years unless they were not representative (for example some farmers abandoned cotton production). According to the 1999–2003 year data (no sample information is released after 2003), the average sample size is more than 1,500.
184
WORLD DEVELOPMENT
4. To be more specific, the percentage of Bt cotton for most of the provinces of the Yellow River and the Yangtze River valleys are obtained from the CCAP data. However, the percentages of Bt cotton for Xinjiang and Gansu provinces are obtained from the author’s calculation based on Li et al. (2013).
am not able to directly estimate Bt cotton adoption models to deal with this issue. However, the advantage of Bt cotton in China is obvious. Hence, the adoption of Bt cotton in practice was mainly affected by the seed availability and the permission from the government. In this sense, I expect that the bias might not be a big issue in this study.
5. Due to lack of data, the tenth largest cotton-producing province, Tianjin, is excluded from this study.
7. The sign of the marginal impact of Bt cotton on yield is the same as that of the estimated coefficient of the percentage of Bt cotton. Similarly, the sign of the interaction term(s) of Bt cotton with year dummy (and regional dummies) is (are) same as its (their) marginal impact(s) on yield.
6. The potential endogeneity problem of Bt cotton adoption might lead to biased-estimation (Fernandez-Cornejo, Klotz-Ingram, & Jans, 2002; Fernandez-Cornejo & McBride, 2002). Given the data used in this study, I
REFERENCES Arza, V., & van Zwanenberg, P. (2014). The politics of technological upgrading: International transfer to and adaptation of GM cotton in Argentina. World Development, 59, 521–534. Bates, S., Zhao, J., Roush, R., & Shelton, A. (2005). Insect resistance management in GM crops: Past, present and future. Nature Biotechnology, 23, 57–62. Carpenter, E. (2010). Peer-reviewed surveys indicate positive impact of commercialized GM crops. Nature Biotechnology, 28, 319–321. Carrie`re, Y., Ellers-Kirk, C., Sisterson, M., Antilla, L., Whitlow, M., Dennehy, T. J., et al. (2003). Long-term regional suppression of pink bollworm by Bacillus thuringiensis cotton. Proceedings of the National Academy of Sciences of the United States of America, 100, 1519–1523. Cleveland, D., & Soleri, D. (2005). Debate over a GM rice trial in China. Science, 310(5746), 231–233. Economist, The. (2013). China’s food fight: A fierce public debate over GM food exposes concerns about America. The Economist, December 14, 2013.. Elbehri, A., & Macdonald, S. (2004). Estimating the impact of transgenic Bt cotton on West and Central Africa: A general equilibrium approach. World Development, 32(12), 2049–2064. Fernandez-Cornejo, J., & McBride, D. (2002). Adoption of bioengineered crops. AER-810, Economic Research Service, U.S. Department of Agriculture, May. Fernandez-Cornejo, J., Klotz-Ingram, C., & Jans, S. (2002). Farm-level effects of adopting herbicide-tolerant soybeans in the U.S.A. Journal of Agricultural and Applied Economics, 34(1), 149–163. Gould, F. (1998). Sustainability of transgenic insecticidal cultivars: Integrating pest genetics and ecology. Annual Review of Entomology, 43, 701–722. Headley, J. C. (1968). Estimating the productivity of agricultural pesticides. American Journal of Agricultural Economics, 50, 13–23. Hossain, F., Pray, C., Lu, Y., Huang, J., & Hu, R. (2004). Genetically modified cotton and farmers’ health in China. International Journal of Occupational and Environmental Health, 10, 296–303. Hsu, H., & Gale, F. (2001). Regional shifts in China’s cotton production and use, Economic Research Service, USDA, November. Huang, J., Chen, R., Mi, J., Hu, R., & Osir, E. (2009). Farmers’ seed and pest control management for Bt cotton in China. In R. Tripp (Ed.), Biotechnology and agricultural development: Transgenic cotton, rural institutions and resource-poor farmers. London: Routledge. Huang, J., Hu, R., Cao, J., & Rozelle, S. (2008). Training programs and in the field guidance to reduce China’s overuse of fertilizer without hurting profitability. Journal of Soil Water Conservation, 63(5), 165–167. Huang, J., Hu, R., Pray, C., Qiao, R., & Rozelle, S. (2003). Biotechnology as an alternative to chemical pesticides: A case study of Bt cotton in China. Agricultural Economics, 29(1), 55–67. Huang, J., Hu, R., Rozelle, S., Qiao, F., & Pray, C. (2002). Transgenic varieties and productivity of smallholder cotton farmers in China. Australian Journal of Agricultural and Resource Economics, 46(3), 367–387. Huang, J., Mi, J., Hai, L., Wang, Z., Chen, R., Hu, R., et al. (2010). A decade of Bt cotton in Chinese fields: Assessing the direct effects and indirect externalities of Bt cotton adoption in China. Science China Life Sciences, 53, 981–991.
James, C. (2013). Global status of commercialized biotech/GM crops: 2013. ISAAA briefs no. 46. Ithaca, NY: International Service for the Acquisition of Agri-biotech Applications. Jiang, G., & Li, X. (2010). On the safety and risks of genetically modified crops industry. Journal of Engineering Studies, 2(2), 112–119. Kathage, J., & Qaim, M. (2012). Economic impacts and impact dynamics of Bt (Bacillus thuringiensis) cotton in India. Proceedings of the National Academy of Sciences of the United States of America, 109(29), 11652–11656. Kouser, S., & Qaim, M. (2011). Impact of Bt cotton on pesticide poisoning in smallholder agriculture: A panel data analysis. Ecological Economics, 70, 2105–2113. Li, X., Gong, Z., Wang, J., Zheng, J., Ai, X., Liang Y., et al. (2013). Current situation and policy suggestions for Bt cotton in Xinjiang. Paper presented in annual meeting of China Cotton Association, August 8–9, Changsha, China. Lichtenberg, E., & Zilberman, D. (1986). The econometrics of damage control: Why specification matters. American Journal Agricultural Economics, 68, 261–273. Ministry of Agriculture. (1990). National plant protection statistical materials. Beijing: National Agricultural Technology Extension Center. National Bureau of Statistics of China. (various years, 2013). China statistics yearbook. Beijing: China Statistics Press. National Development and Reform Commission. (various years). All China data compilation of the costs and returns of main agricultural products. Beijing: China Statistics Press. Pemsl, D., & Waibel, H. (2007). Assessing the profitability of different crop protection strategies in cotton: Case study results from Shandong Province. China Agricultural System, 95, 28–36. Pray, C., Ma, D., Huang, J., & Qiao, F. (2001). Impact of Bt cotton in China. World Development, 29, 813–825. Qaim, M. (2003). Bt cotton in India: Field trial results and economic projections. World Development, 31(12), 2115–2127. Qaim, M., & de Janvry, A. (2003). Genetically modified crops, corporate pricing strategies, and farmers’ adoption: The case of Bt cotton in Argentina. American Journal of Agricultural Economics, 85(4), 814–828. Qiao, F., Wilen, J., Huang, J., & Rozelle, S. (2009). Dynamically optimal strategies for managing the joint resistance of pests to Bt toxin and conventional pesticides in a developing country. European Review of Agricultural Economics, 36(2), 253–279. Qiao, F., Wilen, J., & Rozelle, S. (2008). Dynamically optimal strategy to manage resistance to Genetically Modified (GM) crops. Journal of Economic Entomology, 101, 915–926. Stone, G. D. (2011). Field versus farm in Warangal: Bt cotton, higher yields, and larger questions. World Development, 39(3), 387–398. Subramanian, A., & Qaim, M. (2009). Village-wide effects of agricultural biotechnology: The case of Bt cotton in India. World Development, 37(1), 256–267. Tabashnik, B., Carrie‘re, Y., Dennehy, T., Morin, S., Sisterson, M., Roush, R., et al. (2003). Insect resistance to transgenic Bt crops: Lessons from the laboratory and field. Journal Economic Entomology, 96, 1031–1038.
FIFTEEN YEARS OF BT COTTON IN CHINA: THE ECONOMIC IMPACT AND ITS DYNAMICS Wang, S., Just, D., & Pinstrup-Anderson, P. (2008). Bt cotton and secondary pests. International Journal of Biotechnology, 10, 113–121. Wossink, A., & Denaux, Z. (2006). Environmental and cost efficiency of pesticide use in transgenic and conventional cotton production. Agricultural System, 90, 312–328. Wu, K., & Guo, Y. (2005). The evolution of cotton pest management practices in China. Annual Review of Entomology, 50, 31–52.
185
Wu, K., Lu, Y., Feng, H., Jiang, Y., & Zhao, J. (2008). Suppression of cotton bollworm in multiple crops in China in areas with Bt toxincontaining cotton. Science, 321, 1676–1678. Zhang, L., Huang, J., Qiao, F., & Rozelle, S. (2006). Economic evaluation and analysis of fertilizer overuse by China’s farmers. In Z. Zhu, D. Norse, & B. Sun (Eds.), Policy for reducing nonpoint pollution from crop production in China. Beijing: Environment Science Press of China.
Available online at www.sciencedirect.com
ScienceDirect
http://dx.doi.org/10.1016/j.worlddev.2015.01.011
Fifteen Years of Bt Cotton in China: The Economic Impact and its Dynamics FANGBIN QIAO* China Economics and Management Academy, Central University of Finance and Economics, Beijing, China Summary. — Even though the economic benefit of Bt cotton adoption in the short-run has been well documented, the dynamics of this benefit remain unclear. In particular, the possibility of pest resistance build-up and secondary pest outbreaks has caused concern regarding the sustainability of this economic benefit in the long run. Hence, this study analyzes the economic impact of Bt cotton and its dynamics in China. Using nationally representative long panel data for 1997–2012, we show that this economic benefit continues 15 years after the commercialization of Bt cotton. Ó 2015 Elsevier Ltd. All rights reserved. Key words — economic benefit, Bt cotton, sustainability, China
1. INTRODUCTION
gene would die away, some farmers had switched to varieties with stacked Bt genes (James, 2013). Whether the short-run effect of Bt cotton adoption is sustainable remains a question for two notable reasons. First, most of the previous studies analyze data collected in the first few years after the commercialization of Bt cotton in a country. However, an increase in pesticide use owing to pest resistance build-up and/or secondary pest outbreaks may not appear in the short run (Qiao, Wilen, & Rozelle, 2008). Therefore, these studies may not provide satisfactory answers on the long-term impact or the sustainability of the economic benefit of Bt cotton. 1 Second, all previous studies were based on a small sample of household survey data. Even though there are more than ten million cotton farmers, these surveys usually include only data on a few hundred of them (e.g., Huang et al., 2003; Kathage & Qaim, 2012). Moreover, although all the sampled households are from the major cotton-producing regions, the estimation results based on the data set might be biased. In this study, we address this shortcoming by using nationally representative panel data over 15 years. We choose China as the focus of our study for two reasons. First, prior to 2006, China was the largest country to plant Bt cotton, and subsequently became the second largest country (James, 2013). 2 More importantly, Bt cotton was commercialized in China early in 1997. In other words, Bt cotton has presently been commercialized for over 15 years. Furthermore, nationally representative data on the detailed information of input and output of cotton production have been surveyed, recorded, and published annually. Second, the sustainability of the effect of Bt cotton adoption is still a hot topic in China. In fact, there has been widespread opposition from the public regarding Bt cotton adoption, particularly after a secondary pest outbreak in some Bt cotton fields in the early 2000s. Scientists and researchers who supported genetically modified (GM) crops were labeled “traitors” (Economist, 2013). Faced by increasing criticism against GM technology, the Chinese government delayed the
The short-run economic benefit of Bacillus thuringiensis (Bt) cotton has been well documented (e.g., Carpenter, 2010; Qaim, 2003; Stone, 2011; Wossink & Denaux, 2006). Empirical studies in China showed that the expenditure on pesticide decreased by more than two-thirds after Bt cotton adoption (Huang, Hu, Pray, Qiao, & Rozelle, 2003; Huang, Hu, Rozelle, Qiao, & Pray, 2002; Pray, Ma, Huang, & Qiao, 2001). Furthermore, the reduction of chemical pesticide use not only increased the yield and net profit of cotton farmers, but also contributed to a cleaner environment and improved the health of farmers (Hossain, Pray, Lu, Huang, & Hu, 2004; Kouser & Qaim, 2011). Because of its high profitability, Bt cotton was almost exclusively adopted in the North China Plain only a few years after its commercialization in 1997 (James, 2013). On the other hand, concerns regarding the long-run effect of Bt technology continue. In fact, the debate on the advantages and disadvantages of this technology began even before the commercialization of Bt crops. It was expected that the widespread adoption of Bt crops would lead to pests’ developing resistance to Bt toxin and/or secondary pest outbreaks. As a result, the total pesticide use would be gradually restored to the level before the adoption of Bt cotton, and the short-run economic benefit of Bt cotton would be completely offset in the long run (e.g., Pemsl & Waibel, 2007; Wang, Just, & Pinstrup-Anderson, 2008). In recent years, this negative attitude toward Bt cotton has often seemed to dominate the public debate in the news and media (Cleveland & Soleri, 2005; Kathage & Qaim, 2012). For example, in China, Bt technology had even been described as a “weapon” that developed countries used to attack developing countries (Jiang & Li, 2010). These concerns have important impact on agricultural production practices. To mitigate the development of pest’s resistance, farmers were required to plant refuges of non-Bt cotton in almost all the countries where Bt cotton is planted, as suggested by the entomologists and ecologists (Bates, Zhao, Roush, & Shelton, 2005; Gould, 1998; Tabashnik et al., 2003). Because pests adapted to one toxin may still be susceptible for another toxin, cotton varieties with stacked Bt genes are more efficient in pest control (Gould, 1998). Due to the worry that the economic benefit of varieties with a single Bt
* The authors acknowledge the financial supports of this study from the National Natural Science Foundation of China (71273290). Final revision accepted: January 24, 2015. 177
WORLD DEVELOPMENT
Data used in this study are primarily obtained from the All China Data Compilation of the Costs and Returns of Main Agricultural Products (Quanguo Nongchanpin Chengben Shouyi Ziliao Huibian – in Chinese pinyin) and China Statistical Yearbook (Zhongguo Tongji Nianjian – in Chinese pinyin). Specifically, the data regarding expenditure on pesticides, labor use, fertilizer use, seed cost, and other material cost, as well as cotton yield are obtained from the All China Data Compilation of the Costs and Returns of Main Agricultural Products (National Development and Reform Commission, various years). 3 However, the data on national total cotton sown area, provincial cotton sown areas, and price indexes are obtained from the China Statistical Yearbook (National Bureau of Statistics of China (NBSC), various years). Due to lack of national statistics, the percentages of Bt cotton in different provinces are mainly obtained from the Center for Chinese Agricultural Policy (CCAP), Chinese Academy of Sciences. 4 China has three major cotton-producing regions: the Yellow River valley, the Yangtze River valley, and the Northwest (Hsu & Gale, 2001). The Northwest region includes primarily the Xinjiang Uyghur autonomous region, which has been the largest cotton-producing province in China since the mid1990s. Another important cotton-producing province in the Northwest region is the Gansu province, which is the twelfth largest cotton-producing province in China (NBSC, 2013). The Yellow River valley is China’s largest cotton-producing region. In this study, five provinces from the Yellow River valley—Shandong, Hebei, Henan, Shaanxi, and Shanxi—are included. In addition, five provinces from the Yangtze River valley—Hubei, Anhui, Hunan, Jiangsu, and Jiangxi—are included. These 10 provinces are also the second to thirteenth largest cotton-producing provinces in China (NBSC, 2013). 5 The total cotton sown area of these 12 provinces included in this study is 4.59 million ha, which is 97.80% of the national total sown area. The basis characteristics of the variables used in this study is shown in Table 1. As shown in Figure 1, Bt cotton adoption in the major cotton-producing provinces was rapid, but significantly different in many aspects. Because of the warm and dry climate conditions and yield damage caused by cotton bollworm, Bt cotton adoption in the Yellow River valley is significantly higher than that in the other two regions (Wu & Guo, 2005). For example, yield loss caused by cotton bollworm in the Hebei province in 1992 was nearly 40% (Ministry of Agriculture, 1990–1999). Hence, Bt cotton was first commercialized in the Yellow River valley in 1997 (Hsu & Gale, 2001). As this was very successful, Bt cotton adoption then rapidly spreads to the Yangtze River
Variable names
Mean
Standard deviation
Yield (kg/mu) Bt cotton adoption (%) Pesticide cost (yuan/mu) Seed cost (yuan/mu) Labor (day/mu) Fertilizer use (yuan/mu) Other cost (yuan/mu)
142.69 32.05 67.66 36.36 35.30 610.77 438.39
38.90 41.63 39.25 18.20 12.59 574.11 121.72
Note: 1 ha = 15 mu; 1 USD = 6.25 Chinese Yuan.
Spread of Bt cotton in China
100
2. DATA AND THE IMPACT OF BT COTTON ADOPTION IN CHINA
Table 1. Basic characteristics of variables
Percentage of Bt cotton(%) 60 20 40 80
commercialization process of other Bt crops (Jiang & Li, 2010). For example, even though GM rice technology has been mature for years, the Chinese government has no plan or schedule for its commercialization. The remainder of this paper is organized as follows. The next section discusses the data used in this study. A descriptive analysis of the economic benefit of Bt cotton and its dynamics is presented by showing the quantities of pesticide cost, seed cost, labor use, and cotton yield and their dynamics since Bt cotton adoption. To isolate the impact of Bt technology and its dynamics, econometric models are set up in the third section, and the estimation results are discussed in the fourth section. The final section concludes the paper.
0
178
1996
2000 Hebei Anhui
2004 (Year) Shandong Hubei
2008 Henan Xinjiang
2012 Jiangsu Gansu
Figure 1. Spread of Bt cotton in China.
valley. As shown in Figure 1, only Bt cotton was grown in the Yellow River and Yangtze River valleys within a few years after its commercialization. On the other hand, because of their dry and hot climate, the percentage of Bt cotton adoption in Xinjiang and Gansu provinces was relatively small until 2004. Figure 2 shows the dynamics of inputs and cotton yield with the widespread adoption of Bt cotton. As shown in Panel A, the pesticide cost decreased significantly after Bt cotton adoption. Similarly, labor use also decreased significantly from 1997 onward (Panel B). Simultaneously, however, both seed cost and yield increased (Panels C and D). In order to show the changes of inputs and yields before and after the Bt cotton adoption, we compared the inputs and yields of three time periods: 1990–96, 1997–2003, and 2004–12. The first time period comprises the years before Bt cotton adoption. In the second time period, Bt cotton cultivation spread from the Yellow River valley to the Yangtze River valley (early adoption period hereafter). In the third time period, only Bt cotton was grown in the Yellow River and Yangtze River valleys and Bt cotton adoption began to increase significantly in the Northwest region (late adoption period hereafter). In fact, 2004 is also the year when the concern regarding the sustainability of the economic benefit of GM technology became the center of public debate. Table 2 shows that with the spread of Bt cotton, pesticide use decreased significantly (column 1). Moreover, as observed in column 2, the pesticide cost is 93.71 yuan/mu during 1990– 96, which decreased to 74.45 yuan/mu during 1997–2003 and to 60.54 yuan/mu during 2004–12 (row 1 to row 3). After dividing the entire sample into three major regions, we found that the reduction in pesticide use mainly comes from the
FIFTEEN YEARS OF BT COTTON IN CHINA: THE ECONOMIC IMPACT AND ITS DYNAMICS
179
Labor use
0
0
50
20
(day/mu) 40
(yuan/mu) 100 150
60
200
80
250
Pesticide cost
1992
1996
2000 (Year)
2004
2008
1992
2012
1996
Panel A
2000 (Year)
2004
2008
2012
2004
2008
2012
Panel B
Seed cost
0
50
20
100
(kg/kg) 150
(yuan/mu) 40 60
200
80
250
100
Yield
1992
1996
2000 (Year)
2004
2008
2012
1992
1996
Panel C
2000 (Year)
Panel D
Figure 2. Dynamics of pesticide cost, seed cost, labor cost, and yield after Bt cotton adoption in China.
Table 2. Input and output of cotton production in China before and after Bt cotton adoption Bt cotton (%)
Pesticide cost (yuan/mu)
Labor use (day/mu)
Seed cost (yuan/mu)
Yield (kg/mu)
China 1990–96 1997–2003 2004–12
0.00 27.60 80.73
93.71 74.45 60.54
42.49 32.43 25.45
24.99 37.94 55.21
123.25 147.31 169.75
Yellow River valley 1990–96 1997–2003 2004–12
0.00 54.65 96.44
127.12 73.47 47.70
38.50 31.20 25.28
29.55 43.54 50.45
88.11 129.91 145.52
Yangtze River valley 1990–96 1997–2003 2004–12
0.00 13.69 85.81
79.69 96.85 81.72
49.83 37.48 28.88
16.47 28.74 60.29
147.08 145.33 173.77
Northwest 1990–96 1997–2003 2004–12
0.00 1.48 30.51
25.58 22.23 38.27
25.78 22.90 17.28
44.59 47.90 53.86
155.15 190.65 217.61
Note: 1 ha = 15 mu; 1 US Dollar = 6.2 Chinese Yuan.
180
WORLD DEVELOPMENT
Yellow River valley. The pesticide cost during 2004–12 is 47.7020yuan/mu, which is less than 40% of that during 1990–96 (row 4 to row 6). A similar trend is observed for labor use over time. As shown in column 3, the labor use during 2004–12 is 25.45 day/mu, while that during 1990–96 is 42.49 day/mu. In other words, the labor use during the late adoption period is only 60% of that before the Bt cotton adoption. It is worth noting that this decrease in labor use can be attributed to the reduction in the time spent for pesticide spraying and picking up large cotton bollworms. In practice, China’s cotton farmers not only use pesticides to control cotton bollworms, but also manually pick them up, particularly large ones, in fields. The decrease in labor use caused by Bt cotton adoption has also been shown in other studies (e.g., Elbehri & Macdonald, 2004; Subramanian & Qaim, 2009). On the other hand, Bt cotton adoption led to an increase in seed cost and cotton yield. The seed cost increased from 24.99 yuan/mu before the Bt cotton adoption to 37.97 yuan/mu during the early adoption period and 55.21 yuan/mu during the late adoption period (column 4). This is as expected, as the price of Bt cotton is much higher than that of non-Bt cotton. As shown in the last column, cotton yield also consistently increased by more than one-third (123.25 kg/mu during 1990–96 vs. 169.75 kg/mu during 2004–12). In summary, Table 2 shows that Bt cotton adoption has successfully led to a decrease in pesticide use and labor use and an increase in cotton yield. Even though seed cost more than doubled after Bt cotton adoption, it still cannot offset its positive economic benefit (in fact, the increase in seed cost is similar to the decrease in pesticide use). More importantly, Table 2 shows that the economic benefit of GM technology did not diminish; rather, it remained stable and continuous during the late adoption period. These observed differences provide interesting insights into the impact of Bt technology and its dynamics. However, they cannot be interpreted as net impacts of the Bt technology, because confounding factors have to be controlled for. In the following section, we isolate the impact of Bt technology by setting up econometric models and estimate multivariate regressions. 3. REGRESSION MODELS The above results might be biased owing to the presence of confounding factors. To isolate the impact of Bt cotton adoption, we need to develop and estimate econometric models. Specifically, the yield function can be written as follows: Yield i;t ¼ a0 þ a1 Bti;t þ a2 Bti;t D2004–2012 þ a3 Inputsi;t þ a4 Yeart þ a5 Provincei þ ei;t
ð1Þ
In Eqn. (1), Yield represents the cotton yield and Bt denotes the percentage of Bt cotton. 6 The Bt variable indicates whether or not the Bt technology has a positive net effect on yield. For example, if the estimated coefficient of Bt in the yield function is positive and statistically significant, then the impact of Bt technology on cotton yield is positive, and vice versa. In addition, we include an interaction term, Bt*D2004–2012, to consider the dynamics of the impact of Bt cotton adoption. D2004–2012 is a dummy variable, which is 1 for the year 2004– 12 and 0 otherwise. If the estimated coefficient of Bt is positive and that of Bt*D2004–2012 is significantly positive, then the
impact of Bt cotton adoption on yield increased over time. On the other hand, if the estimated coefficient of Bt*D2004–2012 is statistically insignificant, then the impact of Bt technology remained stable over time. The input variables, Inputs, in the yield function include fertilizer, pesticide, seed, labor, and other inputs. As there are no quantitative data for seed and pesticide use, expenditures on seed and pesticides are used. To consider the impact of inflation, all the expenditures were deflated by the rural consumer price index. Similarly, other material cost, which is used to consider the impact of all other material inputs, was also adjusted using the rural consumer price index. Fortunately, however, quantitative data for fertilizer use (kg/mu) and labor use (day/mu) are available. Year is a vector of year dummies, which is added to consider the impact of variation each year. Similarly, Province is a vector of provincial dummies. Adding Year and Province variables makes our model a two-way fixed-effect model. Finally, subscript i is the ith province, subscript t is the year, and e is the error term. However, OLS estimation of Eqn. (1) might be biased, as Bt technology adoption affects the pesticide cost, seed cost, and labor use. To solve this problem, we need to exclude the impact of Bt cotton adoption on these inputs. After excluding the net impact of Bt cotton adoption, we then added the predicted values of these inputs, into Eqn. (1). In other words, Eqn. (1) was changed as follows: Inputsi;t ¼ b0 þ b1 Bti;t þ b2 Bti;t D2004–2012 þ b3 Yeart þ b4 Provincei þ ei;t Yield i;t ¼ c0 þ c1 Bti;t þ c2 Bti;t D2004–2012 þ c3 Inputs pi;t;Bt þ c4 Fertilizeri;t þ c5 Otheri;t þ c6 Yeart þ c7 Provincei þ fi;t
ð2Þ
In Eqn. (2), the predicted values, Inputs_pi,t,Bt, are the values of inputs (i.e., pesticide cost, labor use, and seed cost) showing the net impact of Bt cotton adoption. Fertilizer is the fertilizer use per unit of land, while Other represents the other material cost. All the other variables are as discussed above. Finally, Bt technology and pesticide use are special inputs, which are different from traditional inputs like fertilizer and labor. Thus, instead of running an OLS regression, we run the Weibull damage abatement function (Headley, 1968; Lichtenberg & Zilberman, 1986). This method has also been widely used in previous studies (e.g., Huang et al., 2003; Qaim & de Janvry, 2003). In this study, the Weibull damage abatement function that we estimate is as follows: Yield i;t ¼ ðk0 þ k1 Yangtzei;t þ k2 Northwesti;t Þ Labor ph1 Fertilizerh2 Seed ph3 Otherh4 ð1 ePesticide
pm
Þ
ð3Þ
In the above equation, Yangtze is a dummy variable, which equals to 1 if the observations are from the Yangtze River valley region. Similarly, Northwest is the Northwest region dummy. m is the coefficient to be estimated to capture the impact of pesticide use on the yield. In this study, we have two scenarios to define m. In the first scenario, m is defined as follows: m ¼ h0 þ h1 Bti;t þ h2 Bti;t D2004–2012
ð4Þ
FIFTEEN YEARS OF BT COTTON IN CHINA: THE ECONOMIC IMPACT AND ITS DYNAMICS
In this scenario, the impact of Bt cotton and its dynamics in all the three major cotton-producing regions (i.e., the Yellow River valley, the Yangtze River valley, and Northwest) are assumed to be the same. In the second scenario, we consider the differences in the impact of Bt cotton adoption and its dynamics for each cotton-producing region. Under this scenario, m is defined as follows: m ¼ h0 þ h1 Bti;t þ h2 Bti;t D2004–2012 þ h3 Bti;t
Inputsi;t ¼ /0 þ /1 Bti;t þ /2 Bti;t D2004–2012 þ /3 Bti;t Yangtzei;t þ /4 Bti;t D2004–2012 Yangtzei;t þ /5 Bti;t Northwest þ /6 Bti;t D2004–2012 Northwest þ /7 Yeart þ /8 Provincei þ li;t
Yangtzei;t þ h4 Bti;t D2004–2012 Yangtzei;t þ h5 Northwesti;t
ð5Þ
Accordingly, under the scenario, the input function is as follows:
Table 3. Impact of Bt cotton adoption and its dynamics in China, 1997–2012
Share of Bt cotton*2004–12 year dummy Constant Provincial dummies Year dummies Observations R2
ð6Þ
4. RESULTS AND DISCUSSION
Bti;t Northwesti;t þ h6 Bti;t D2004–2012
Share of Bt cotton
181
Expenditure on pesticide (yuan/mu)
Labor (day/mu)
Seed cost (yuan/mu)
0.4436** (4.21) 0.2847*
0.0750** (4.18) 0.0072
0.4514** (9.37) 0.1296*
(2.31) 85.8181** (7.56) Yes Yes
(0.34) 17.6436** (9.12) Yes Yes
(2.30) 38.7977** (7.48) Yes Yes
185 0.722
185 0.838
185 0.674
Note: Cost is in 2012 year Yuan. The symbols * and ** denote significance at 5% and 1%, respectively.
The results of the econometric estimation of input equations are shown in Tables 3 and 4, while the estimation result of yield is shown in Table 5. In general, most of the regression results are consistent with the descriptive analysis in Section 3. Most estimated coefficients on the control variables have the expected signs and are statistically significant. In the following paragraphs, we first discuss the estimation result of input equations and that of the yield equation later. (a) Impact of Bt technology on pesticide use, labor use, and seed cost, and its dynamics Estimation results show that the adoption of Bt cotton has significantly decreased pesticide use and labor use. As shown in the first column of Table 3, compared to non-Bt cotton, the decrease in pesticide use led to the reduction of expenditure on pesticide by 44.36 yuan/mu, or 50.82% of the pesticide use in 1996, the final year before the adoption of Bt cotton. Similarly, the adoption of Bt cotton led to a reduction of 7.50 day/mu of labor use, or 17.25% of the total labor use in 1996. After considering the regional differences, the negative impact of Bt cotton adoption on pesticide use and labor use is still significant. As shown in Table 4 (column 1), even though the impact of pesticide use in the Yangtze River valley is significantly smaller than that in the Yellow River valley, the decrease in pesticide use in the Yangtze River valley is still substantial. On the other hand, the impact of Bt cotton on labor use in all three major cotton-producing regions showed no difference.
Table 4. Impact of Bt cotton adoption and its dynamics in China’s major cotton region, 1997–2012
Percentage of Bt Percentage of Bt*2004–12 year dummy Yangtze River dummy*percentage of Bt Yangtze River dummy*percentage of Bt*2004–12 year Northwest dummy*percentage of Bt Northwest dummy*percentage of Bt*2004–12 year Constant Provincial dummies Year dummies Observations R2 Note: Cost is in 2012 year Yuan. The symbols
*
and
**
Expenditure on pesticide (yuan/mu)
Labor (day/mu)
Seed cost (yuan/mu)
0.9303** (7.42) 0.1047 (0.78) 0.6637** (2.94) 0.0949 (0.46) 3.0191 (1.51) 3.5121 (1.80) 80.4646** (5.06) Yes Yes
0.0569* (2.38) 0.0065 (0.26) 0.0089 (0.21) 0.0112 (0.29) 0.0109 (0.03) 0.0583 (0.16) 19.3025** (6.37) Yes Yes
0.3370** (5.55) 0.1618* (2.50) 0.3267** (2.99) 0.1639 (1.64) 0.2938 (0.30) 0.3656 (0.39) 46.6728** (6.05) Yes Yes
185 0.779
185 0.840
185 0.710
denote significance at 5% and 1%, respectively.
182
WORLD DEVELOPMENT Table 5. Impact of Bt cotton adoption on cotton yield in China, 1997–2012 Scenario 1 Coefficient
Constant Yangtze river dummy Northwest dummy Fertilizer use (yuan/mu) Labor (day/mu) Seed cost (yuan/mu) Other cost (yuan/mu) Percentage of Bt cotton Percentage of Bt cotton*2004–12 year Yangtze river dummy*percentage of Bt cotton Yangtze*percentage of Bt*2004–12 year Northwest dummy*percentage of Bt cotton Northwest*percentage of Bt*2004–12 year Observation R2 Note: The symbols
*
17.4513 4.2766 11.7255 0.0173 0.2165** 0.0536** 0.2467** 0.0013** 0.0001
185 0.62 *
and
**
Scenario 2 T-value
Coefficient
T-value
2.09 1.85 1.92 1.45 2.93 2.76 3.29 3.55 0.20
16.4812 4.7158 11.3691 0.0143 0.2238** 0.0502* 0.2481** 0.0014* 0.0001 0.0006 0.0004 0.0001 0.0004
1.94 1.71 1.81 1.08 2.93 2.45 3.20 2.56 0.39 0.77 0.66 0.02 0.06
185 0.62
denote significance at 5% and 1%, respectively.
More interestingly, the significant impact of Bt cotton adoption on pesticide use did not diminish. The estimated coefficient of the interaction term of Bt cotton and 2004–12 year dummy, Bt*D2004–2012, is negative and significant in the pesticide use equation (Table 3, column 1), which implies that farmers spray less pesticide in the late adoption period than in the early adoption period. In other words, the estimation results show that the long-term impact of Bt cotton, with regard to pesticide use, not only did not decrease, but also increased over time. This result is clearly contradictory to the expectations of those who believe that the development of resistance by pests to Bt toxin and increase in secondary pests would completely offset the economic benefit that Bt technology had generated. In fact, the estimation results are not hard to understand. Previous studies had shown that the widespread adoption of Bt cotton had successfully suppressed the cotton bollworm population density regionally; thus, all farmers, including those who planted non-Bt cotton varieties, reduced their pesticide applications (Carrie`re et al., 2003; Wu, Lu, Feng, Jiang, & Zhao, 2008). The reduction of the total pest population density caused a further reduction of pesticide use and thereby increase in the net economic benefit of Bt cotton adoption in the long run. The estimation results also show that labor use reduction does not change over time. As shown in the second column, the interaction term of Bt and D2004–2012 is not statistically significant (Table 3). Similarly, the interaction terms of Bt*D2004– 2012 and the Yangtze River valley and Northwest are also statistically insignificant (Table 4). In other words, these results show that the reduction in labor use owing to Bt cotton adoption is stable during the late adoption period. Finally, as expected, the adoption of Bt cotton led to an increase in seed cost. As shown in Table 3, compared to non-Bt cotton production, seed cost increased to 45.14 yuan/mu, which is 161.91% of that in 1996, when Bt cotton was not commercialized. However, with the increase in the price of non-Bt cotton seeds and the competition in the seed market (Huang, Chen, Mi, Hu, & Osir, 2009), the difference between Bt and non-Bt cotton reduced in the long run (row 2 of Table 3). Similar results are shown in Table 4 (row 2, column 3).
To summarize, the estimation results of the input equations show that the economic benefit of Bt cotton adoption, particularly the pesticide use reduction, did not diminish in the long run. On one hand, pest resistance development and secondary pest outbreaks may offset the economic benefit of Bt cotton adoption. On the other hand, regional reduction of pest population density caused by the widespread usage of Bt cotton may lead to larger economic benefits. Thus, in the long run, the net impact of Bt cotton adoption depends on whether the positive side (i.e., reduction of pest population) or the negative side (i.e., pest resistance development and secondary pest outbreaks) is dominant. The estimation results in this study show that the positive side dominated after 15 years of Bt cotton commercialization. (b) Impact of Bt technology on cotton yield and its dynamics The estimation results of the yield equation are presented in Table 5. As shown in Table 5, estimation results under the two scenarios are very similar, indicating the robustness of the results. The results show that most of the coefficient estimates are as expected and statistically significant. For example, the estimated coefficients of seed cost, labor use, and other material cost are all positive and statistically significant. On the other hand, fertilizer use has no significant impact on yield, which is consistent with previous studies that show the overuse of fertilizer is a common phenomenon in China (e.g., Huang, Hu, Cao, & Rozelle, 2008; Zhang, Huang, Qiao, & Rozelle, 2006). More importantly, the results show that the estimated coefficient of the percentage of Bt cotton is positive and statistically significant under both the scenarios, which implies that Bt cotton adoption has a positive impact on yield (row 8). On the other hand, the estimated coefficient of the interaction term, percentage of Bt cotton*D2004–2012 year dummy, is insignificant, which implies that the impact of Bt cotton adoption on yield remains stable over time (row 9). 7 As shown in Table 5, none of the estimated coefficients of the interaction terms of regional dummy and percentage of Bt cotton (and D2004–2012 year dummy) is statistically significant (row 11 to row 14). In other words, our estimation results show that the impact of Bt cotton adoption on the yield is similar in all cotton-producing regions.
FIFTEEN YEARS OF BT COTTON IN CHINA: THE ECONOMIC IMPACT AND ITS DYNAMICS
183
Table 6. Economic benefit of Bt cotton adoption in China Economic benefit per unit of land (mu) per year 1997–2003
Pesticide use Labor Seed cost Yield
Total benefit since Bt adoption
2004–12
Total
Annual average
Quantity
(%)
Quantity
(%)
Billion yuan
Billion yuan
19.57 1.57 12.31 12.58
22.42 3.61 44.15 10.50
43.99 4.59 27.29 40.42
50.40 10.56 97.88 33.73
4.12 8.70 2.56 22.96
0.26 0.54 0.16 1.44
33.22
2.08
Total Note: % = Quantity saving during 1997–2012/quantity of 1996 (the year before Bt cotton adoption).
(c) Economic benefit of Bt cotton adoption in China After the discussion on the estimation results of the impact of Bt cotton adoption on inputs and yield, we calculate the total impact of Bt cotton adoption in China. To calculate the total impact, we used the estimated coefficients when regional differences are considered, as shown in Tables 4 and 5, and the real percentage of Bt cotton adoption. The results of the calculation are summarized in Table 6. As shown in the first two rows of Table 6, owing to Bt cotton adoption, farmers saved 4.12 billion yuan on pesticide use and 8.70 billion yuan on labor use. On the other hand, the benefit from yield increase is even higher (row 4). As shown in the last row, the total economic benefit that Bt cotton had generated is more than 33 billion yuan over 15 years after Bt cotton was commercialized. 5. CONCLUSION Using nationally representative panel data, I analyzed the economic impact of Bt cotton adoption and its sustainability, after 15 years of its commercialization in China. Consistent with its short-term impact, this study showed that the economic benefit of Bt cotton did not diminish, but remained stable and continuous in China. To the best of our knowledge, this is the first study that uses nationally representative data and focuses on the economic benefit of Bt cotton and its dynamics. The findings of this study have important implications. First, undoubtedly, these findings will contribute to a wider public debate in China as well as in other countries where GM crops are planted. The long-term impact and the sustainability of the economic benefit that GM crops generate have been subject to a heated debate. Those who oppose GM technology believe that such economic benefit has only a short-run
impact, which would diminish in the long run, owing to pest resistance development and secondary pest outbreaks (e.g., Wang et al., 2008). However, the estimation results of this study provide empirical evidence that the economic benefit of GM technology is not limited to the short run, which will definitely help to calm down the public debate. Second, our results are helpful for Chinese policymakers in managing GM crop adoption. Faced by criticism from those who oppose GM technology, the Chinese government has stalled the commercialization of other GM crops, even though billions of dollars have been spent on these crops and China leads the world in GM rice technology. Even though the commercialization of GM rice is affected by many factors, the economic impact and its dynamics are particularly important research topics. We believe that the results of this study will contribute to the management of GM crops by the Chinese government. Finally, findings from this study also contribute to the development and implementation of biotechnology policies in other countries. Even though planting non-Bt cotton as refuge is required in most of the country, economic analysis showed that it might be inappropriate, especially for countries where natural refuge crops are available (Qiao, Wilen, Huang, & Rozelle, 2009). This study provided empirical evidence that China’s non-zero refuge still works after 15 years of Bt cotton commercialization. Hence, policy makers might need to rethink their refuge policies in countries where farmers are required to plant non-Bt cotton as refuge. As shown in this paper, the first generation of cotton varieties with a single Bt gene still can effectively control the bollworm. Even though farmers in some countries have switched from unpatented and royalty fee cotton varieties with a single Bt gene to patented cotton varieties with stacked genes, rigorous analysis is needed to answer whether this switch is economical or a result caused by many factors (Arza & van Zwanenberg, 2014).
NOTES 1. To the best of our knowledge, the only exceptions are Kathage and Qaim (2012) and Huang et al. (2010). Using data collected during 2002–08 in India, Kathage and Qaim estimated the economic impact and impact dynamics of Bt cotton. However, as Bt cotton was first commercially released in India in 2002, it had only been planted for seven years when they conducted their final survey. In this sense, the results of their study may have captured the mid-term impact rather than the long-term impact. For the same reason, Huang et al. (2010), whose study is based on data collected during 1999–2007, cannot be said to have focused on the long-term impact.
2. Since 2006, India is the largest country where Bt cotton is planted. 3. Since 1953, the National Development and Reform Commission began to conduct household surveys to understand the costs and returns for each of the major crops. For cotton, the sample counties were selected based on sown area and production in all major cotton production provinces, while households were randomly selected in sample counties. The households were revisited in the following years unless they were not representative (for example some farmers abandoned cotton production). According to the 1999–2003 year data (no sample information is released after 2003), the average sample size is more than 1,500.
184
WORLD DEVELOPMENT
4. To be more specific, the percentage of Bt cotton for most of the provinces of the Yellow River and the Yangtze River valleys are obtained from the CCAP data. However, the percentages of Bt cotton for Xinjiang and Gansu provinces are obtained from the author’s calculation based on Li et al. (2013).
am not able to directly estimate Bt cotton adoption models to deal with this issue. However, the advantage of Bt cotton in China is obvious. Hence, the adoption of Bt cotton in practice was mainly affected by the seed availability and the permission from the government. In this sense, I expect that the bias might not be a big issue in this study.
5. Due to lack of data, the tenth largest cotton-producing province, Tianjin, is excluded from this study.
7. The sign of the marginal impact of Bt cotton on yield is the same as that of the estimated coefficient of the percentage of Bt cotton. Similarly, the sign of the interaction term(s) of Bt cotton with year dummy (and regional dummies) is (are) same as its (their) marginal impact(s) on yield.
6. The potential endogeneity problem of Bt cotton adoption might lead to biased-estimation (Fernandez-Cornejo, Klotz-Ingram, & Jans, 2002; Fernandez-Cornejo & McBride, 2002). Given the data used in this study, I
REFERENCES Arza, V., & van Zwanenberg, P. (2014). The politics of technological upgrading: International transfer to and adaptation of GM cotton in Argentina. World Development, 59, 521–534. Bates, S., Zhao, J., Roush, R., & Shelton, A. (2005). Insect resistance management in GM crops: Past, present and future. Nature Biotechnology, 23, 57–62. Carpenter, E. (2010). Peer-reviewed surveys indicate positive impact of commercialized GM crops. Nature Biotechnology, 28, 319–321. Carrie`re, Y., Ellers-Kirk, C., Sisterson, M., Antilla, L., Whitlow, M., Dennehy, T. J., et al. (2003). Long-term regional suppression of pink bollworm by Bacillus thuringiensis cotton. Proceedings of the National Academy of Sciences of the United States of America, 100, 1519–1523. Cleveland, D., & Soleri, D. (2005). Debate over a GM rice trial in China. Science, 310(5746), 231–233. Economist, The. (2013). China’s food fight: A fierce public debate over GM food exposes concerns about America. The Economist, December 14, 2013.
James, C. (2013). Global status of commercialized biotech/GM crops: 2013. ISAAA briefs no. 46. Ithaca, NY: International Service for the Acquisition of Agri-biotech Applications. Jiang, G., & Li, X. (2010). On the safety and risks of genetically modified crops industry. Journal of Engineering Studies, 2(2), 112–119. Kathage, J., & Qaim, M. (2012). Economic impacts and impact dynamics of Bt (Bacillus thuringiensis) cotton in India. Proceedings of the National Academy of Sciences of the United States of America, 109(29), 11652–11656. Kouser, S., & Qaim, M. (2011). Impact of Bt cotton on pesticide poisoning in smallholder agriculture: A panel data analysis. Ecological Economics, 70, 2105–2113. Li, X., Gong, Z., Wang, J., Zheng, J., Ai, X., Liang Y., et al. (2013). Current situation and policy suggestions for Bt cotton in Xinjiang. Paper presented in annual meeting of China Cotton Association, August 8–9, Changsha, China. Lichtenberg, E., & Zilberman, D. (1986). The econometrics of damage control: Why specification matters. American Journal Agricultural Economics, 68, 261–273. Ministry of Agriculture. (1990). National plant protection statistical materials. Beijing: National Agricultural Technology Extension Center. National Bureau of Statistics of China. (various years, 2013). China statistics yearbook. Beijing: China Statistics Press. National Development and Reform Commission. (various years). All China data compilation of the costs and returns of main agricultural products. Beijing: China Statistics Press. Pemsl, D., & Waibel, H. (2007). Assessing the profitability of different crop protection strategies in cotton: Case study results from Shandong Province. China Agricultural System, 95, 28–36. Pray, C., Ma, D., Huang, J., & Qiao, F. (2001). Impact of Bt cotton in China. World Development, 29, 813–825. Qaim, M. (2003). Bt cotton in India: Field trial results and economic projections. World Development, 31(12), 2115–2127. Qaim, M., & de Janvry, A. (2003). Genetically modified crops, corporate pricing strategies, and farmers’ adoption: The case of Bt cotton in Argentina. American Journal of Agricultural Economics, 85(4), 814–828. Qiao, F., Wilen, J., Huang, J., & Rozelle, S. (2009). Dynamically optimal strategies for managing the joint resistance of pests to Bt toxin and conventional pesticides in a developing country. European Review of Agricultural Economics, 36(2), 253–279. Qiao, F., Wilen, J., & Rozelle, S. (2008). Dynamically optimal strategy to manage resistance to Genetically Modified (GM) crops. Journal of Economic Entomology, 101, 915–926. Stone, G. D. (2011). Field versus farm in Warangal: Bt cotton, higher yields, and larger questions. World Development, 39(3), 387–398. Subramanian, A., & Qaim, M. (2009). Village-wide effects of agricultural biotechnology: The case of Bt cotton in India. World Development, 37(1), 256–267. Tabashnik, B., Carrie‘re, Y., Dennehy, T., Morin, S., Sisterson, M., Roush, R., et al. (2003). Insect resistance to transgenic Bt crops: Lessons from the laboratory and field. Journal Economic Entomology, 96, 1031–1038.
FIFTEEN YEARS OF BT COTTON IN CHINA: THE ECONOMIC IMPACT AND ITS DYNAMICS Wang, S., Just, D., & Pinstrup-Anderson, P. (2008). Bt cotton and secondary pests. International Journal of Biotechnology, 10, 113–121. Wossink, A., & Denaux, Z. (2006). Environmental and cost efficiency of pesticide use in transgenic and conventional cotton production. Agricultural System, 90, 312–328. Wu, K., & Guo, Y. (2005). The evolution of cotton pest management practices in China. Annual Review of Entomology, 50, 31–52.
185
Wu, K., Lu, Y., Feng, H., Jiang, Y., & Zhao, J. (2008). Suppression of cotton bollworm in multiple crops in China in areas with Bt toxincontaining cotton. Science, 321, 1676–1678. Zhang, L., Huang, J., Qiao, F., & Rozelle, S. (2006). Economic evaluation and analysis of fertilizer overuse by China’s farmers. In Z. Zhu, D. Norse, & B. Sun (Eds.), Policy for reducing nonpoint pollution from crop production in China. Beijing: Environment Science Press of China.
Available online at www.sciencedirect.com
ScienceDirect