FARM PESTICIDE, RICE PRODUCTION, AND HUMAN HEALTH
by
Jikun Huang, Fangbin Qiao,
Linxiu Zhang and Scott Rozelle
Center for Chinese Agricultural Policy
Chinese Academy of Agricultural Sciences
FARM PESTICIDE, RICE PRODUCTION, AND HUMAN HEALTH
Jikun Huang, Fangbin Qiao, Linxiu Zhang and Scott Rozelle1
I. Introduction
Pesticides of various kinds have been used on a large scale to protect crops from
damage inflicted by insects and diseases in China since 1950s. Annual pesticide
production has reached more than 500 thousand metric tons after the mid-1990s
(Huang et al., 2000). China was the second largest pesticide using country in the world
in 1980s and has been likely to be the largest pesticide using country in the world since
the middle 1990s.
Although grain production increases have been realized, intensive pesticide use
can have several drawbacks. In addition to the direct costs of the pesticides, long term,
highly concentrated application of pesticides may not only contaminate the products of
field crops, but also pose a serious danger to the agro-ecosystem (e.g., the surrounding
soil and water quality) and human health (Rola and Pingali, 1993).
As employment in rural enterprises (industries) and urban sector expanded, the
opportunity cost of agricultural labor has risen. The farming sector has been becoming
a part-time job in many areas of the eastern and coastal regions of China. This change is
likely to have negative impacts on the environment because of the substitution of
chemicals for labor (Ye, 1991; Huang and Rozelle, 1996). Labor shortages may also
lead to untimely or improper application of pesticides. Deterioration of agricultural
extension system due to inadequate financial support from government since the
mid-1980s further add to the concern of proper use of chemical in crop production
(Huang, et al.,1999a).
Given the prospectus of China's food situation in the coming century and the
central goal of China's food security policy (Fan and Agcaoili, 1997; Huang, et al.,
1999b), however, intensity cropping system will likely continue to be the dominant
farming practice in China. Most observers of Chinese agriculture believe there will be
an increasing use of farm pesticides because of the large impacts that the pesticides are
perceived by farmers to have on crop yields. Under current farming practices, plant
protection reliance on pesticides is expected to lead to more dependence of crop
production, and to rising use of pesticides due to rapid development of resistance within
China's current varieties (Widawsky et al, 1998).
This is not to say that the government has ignored the environmental and health
risk connected with farm pesticide use. Effort has been made by Chinese government in
the areas of regulating the production of highly toxic pesticide, the safety use, and
1
Jikun Haung, Fangbin Qiao, Linxiu Zhang are Professor, Senior Research Assistant, Associate
Professor, respectively, Center for Chinese Agricultural Policy, Chinese Academy of Agricultural
Sciences; Scott Rozelle is associate professor of University of California at Davis. The authors
would like to acknowledge the help comments and insights received from Fenglin Lu, David Glover,
Agness Rola, Prabhu Pingali, Ruifa Hu, Herminia Franciscoin, and Shengxiang Tang, and thank to
Jun Song, Yanmei Lu, Lei Minguo, Mingming Fan, Caiping Zhang, Ms. Yuxian Lin, and Li Wang
for their assistance in data collection and field survey. The financial supports of Economy and
Environment Program for Southeast Asia (EEPSEA) and China National Natural Sciences
Foundation (79800026 and 79725001) are greatly appreciated.
1
management of pesticides during the reform era. Before the 1980s, policies focused on
the safe handling of agricultural chemicals. Regulations on safety use of pesticide, the
bans on highly hazardous organochlorine (BHC and DDT), organomercurial and
organoarsenical pesticides, creation of standards covering acceptable daily intake level,
and maximum allowable levels of residue on crops were developed in 1980s. However
as experience has shown, the simple promulgation of rules and regulation are not
enough to guarantee improvements in the proper and safe use of pesticides. There has
been increasing concern on the environmental pollution and harmful human health
consequences of agricultural chemical pesticide use. The amount of farm pesticide use
has increased rapidly in recent years and farmers in many areas show little knowledge
of efficient and safety use of pesticides. Few researchers have seriously considered the
negative impacts of chemical pesticide use, on grain, environment and human health.
In seeking to have a clear understanding of consequences of increasing farm
pesticide use on agricultural production, environment and human health, and identify
the policies that will help farmers to reduce the pesticide use and maintaining the
profitability of crop production, several critical questions arise. What are the trends of
production, utilization and trade of chemical pesticides in China? How much pesticide
is currently applied to crop production? How the government policies and regulation
affect the production, utilization and trade of chemical pesticides? What are the
determinants of pesticide adoptions by farmers? What are the impacts of pesticide uses
on crop production, environment, and farmer's health? What are the benefits and costs
of farm pesticide uses in grain production? And what are alternatives to the pesticide
use in crop production?
Answers to the above questions are by no means clear in China. Some claim that
grain yield and production are the first important in an economy where per capita
income of farmers is still low. The others believe that it is important for China to
consider them as an integrated part of sustainable development of agriculture in future.
This study is the first attempt to quantify the impacts of pesticide use on agricultural
production and farmer health in major grain production region in China. To narrow the
scope of analysis, however, rice in Zhejiang province is selected as the focus of the
research. Rice accounts for about 40% of China's grain production and major grain
crop in Zhejiang province.
The overall goal of this paper is to have a better understanding the impacts (both
positive and negative) of farm pesticide use on rice production and farmer’s health. In
order to achieve this general goal, the paper is organized as follows. After an overview
of the pesticide economy in China in the next section, the trends of pest and weed
related problems are examined in the third section. In the fourth section, the profile of
pesticide use in rice production and farmer's perception, knowledge, and pesticide use
practices are provided. Section fifth examines the determinants of pesticide adoption by
farmers and impacts of pesticide use in rice production. The results of the analyses will
provide useful information for policy makers in designing the more conductive policies
and regulations on the farmers' pesticide uses. The impacts of pesticide use on the rice
farmer's health are examined in the sixth section. The results provide empirical
evidence of the pesticide hazardous to rice farmers. The final section summarizes the
findings of the study and their policy implications.
2
II. Overview of Pesticide Sector in China
China's chemical pesticide industry has developed rapidly since the early 1950s.
The pesticide production was about 1,000 metric tons (tons thereafter) only in 1950,
dramatically raised to 321 thousand tons in 1970 (SSB), and reached a historical high in
1980 (Table 1). Insecticide production dominated the pesticide industry. More than
three fourth of all pesticides produced in China were insecticides in 1980s. Fungicides,
at nearly 10% of all pesticides produced, and herbicide, at 6-7%, were the next
important.
Table 1. Production, net import and total supply of pesticides (in active ingredients, 000
metric tons) in China, 1985-96
1980
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
Production
Net import
Total supply
537.0
204.3
203.2
152.3
169.3
196.8
226.6
253.4
262.0
230.7
263.7
349.0
381.2
11.0
6.6
-11.1
-11.8
3.2
7.2
3.7
1.0
8.1
-16.1
-38.6
-58.1
-41.6
548.0
210.9
192.1
140.5
172.5
204.0
230.3
254.4
270.1
214.6
225.1
290.9
339.6
Sources: Trade data are from the State Statistical Bureau, Statistical Yearbook of China.
Production data are from the Ministry of Chemical Industry, Statistical Yearbook of
Chemical Industry.
The high toxicity and high residue of pesticides used in crops were major concerns
of the country as the pesticides were dominated by organochlorines (OC) and
organophosphates (OP) in the 1970s and the early 1980s. Among 537 thousand tons of
pesticide produced in 1980, the pesticides classified as "high toxicity and high residue"
category accounted for 64% (MOA). A dramatically decline in the total supply of
pesticides in the early 1980s reflected the considerable progress made in the pesticide
industry with the introduction of less persistent and high efficient compounds as
substitutes for organochlorinu in insecticides (i.e., BHC and DDT). Several policies
and regulations on pesticide production and utilization were formulated in the early
1980s. These included the bans on the extremely hazardous OC and OP production in
1983, phasing out BHC (666) production in 1984, and reducing DDT production and
only allowing DDT to be used in non-crop production. Since the mid-1980s,
methamidophos, dimthypo and parathionmethyl have been gradually replacing BHC,
dichlorvos, dimethoate and DDT as the dominant insecticides.
Although the changes in the varieties of fungicide herbicides produced in China
were less significant than that of insecticides in the past, but they have been
diversifying in favor of improving the efficient and reducing the hazardous of both
3
fungicide and herbicide. Among all pesticides, the most significant increase in the
production is herbicides (Table 2). The herbicide production raised by more than 4
times from 1985 (13.5 thousand tons) to 1996 (60.3 thousand tons) as the opportunity
cost of labor increased. Other pesticides produced in China include plant growth
regulators, acaricide, molluscicides, roderticides, and etc. By the 1996, there were
more than 1100 pesticide factories in China.
Table 2. Pesticide production (000 tons) for selected major varieties in China, 1980-96.
1980
Insecticide
Top 4 varieties in 1980
Top 3 varieties in 1995
386.8
28.7
Fungicide
Top 3 varieties in 1980
Top 3 varieties in 1995
6.1
4.0
Herbicide
Top 3 varieties in 1980
Top 3 varieties in 1995
17.3
1.1
1985
1990
1995
1996
155.7
178.5
245.9
271.7
52.5
54.4
48.4
61.4
32.7
109.7
46.6
110.4
19.0
24.9
37.5
37.2
11.2
6.8
9.8
8.4
10.2
14.2
10.2
16.0
13.5
21.1
53.3
60.3
10.1
1.3
12.0
4.8
8.0
17.7
8.8
14.4
Note: Top 4 varieties of insecticides in 1980 are 666 (BHC), Didiwei (Dichorvos), Leguo (Dimethoate),
and DDT, and Top 3 varieties of insecticides in 1995 are Jiaanlin (Methamidophos), Shachongshuang
(Dimthypo), and Jiaji1605(Parathionmethyl). Top 3 varieties of fungicide varieties in 1980 are
Duojunling (Carbendazim), Jinggangmeisu (Validamycin A), and Yidaowenjing (MAFA), and in 1995
they are Duojunling (Carbendazim), Daishenmengxin (Mancozeb), and Yidaowenjing (MAFA). Top 3
varieties of herbicide varieties in 1980 are Wulufenna (PCP-Na), Chucaomi (Nitrofen), and 2,4-D, and
they are Dingcaoan (Butachlor), Yicaoan (Acetochlor), and 2,4-D in 1995.
Source: Data are from the Ministry of Chemical Industry, Chemical Industry Yearbook of China, various
issues.
China has been both the importer and exporter of chemical pesticides since the late
1970s. The growth rate of pesticides export has been higher than that of pesticide
imports, leading China to be a net pesticide exporter since the early 1990s. By 1997,
China exported 87.6 thousand tons of pesticides and earned 309.4 million US dollars.
Compared to 48.6 thousand tons of pesticide imports, China was a net exporter of
pesticides with a trade surplus of 143.6 million US dollars in 1997, and indeed China
has been a net exporter since the middle 1990s (Table 1).
However, increasing the frequency of the pest diseases since the mid-1980s has
raised the pesticide consumption despite the progress made in the pesticide industry
since the mid-1980s. After a decline of annual pesticide total supply from 548 thousand
tons in 1980 to 140 thousand tons in 1987, the average pesticide available for
agriculture uses doubled again between 1987 and 1995. By 1996, the total pesticide
supply reached 339.8 thousand tons, likely becoming the largest pesticide consumption
countries in the world.
Pesticide use also differs significantly across regions. Top five provinces
accounted for more than half of the national pesticide applications (Huang, et al., 2000).
Zhejiang is one of the most pesticide intensive use provinces in China. The rate of
pesticide use in Zhejiang has been more doubled than that at national level.
4
III. Extent of Crop Diseases and Pesticide Use in Crop Production in China
Extent of Diseases and Efforts to Control the Diseases
We use the ratio of pest related epidemic (or occurred) area to the total crop sown
areas to indicate the extent of pest problems, and the proportion of weed epidemic areas
to the total crop sown areas to show the extent of weed problem. Table 3 shows that,
on the average, pest (insect and disease) occurred 1.3 times during one crop season in
1988 in China2. This figure rose to 1.7 times in 1996, increased by 31 percent over 8
years. Increases in the intensity of crop production and pesticide uses (leading to a
higher resistant of pests to pesticides) might explain most of the rise in the frequency of
the pest occurrences.
Table 3. The extent of pest (insect and disease) problems in Zhejiang and China, 1988-96.
National
Zhejiang
Year
Ratio of epidemic Ratio of treated
Ratio of epidemic Ratio of treated
area to total crop area to total crop
area to total crop area to total crop
sown area
sown area
sown area
sown area
1988
1989
1990
1991
1992
1993
1994
1995
1996
1.3
1.5
1.5
1.6
1.5
1.6
1.6
1.7
1.7
1.1
1.2
1.4
1.5
1.5
1.7
1.6
1.8
1.8
2.6
2.8
2.7
2.8
2.5
3.0
2.5
2.8
2.9
2.4
3.0
3.4
3.4
2.9
3.2
3.1
4.5
3.9
Source: Computed by authors based on the data from MOA, Agricultural Yearbook of China.
Zhejiang province, with higher multiple-cropping-index and higher pesticide use
rate by farmers, has a much higher ratio of epidemic to total crop sown areas for both
pest and weed, particular for insect and disease. In 1988, the extent of insect and
disease problem in Zhejiang (2.6) was the twice as much as that for the country as a
whole (Table 3). Although the rising trend of pest attacks in Zhejiang was less
significant, the ratio of pest epidemic areas to total crop sown areas in Zhejiang (3.9)
was still more than 70 percent higher than that in the national level (1.7) in 1996.
The frequency of weed problems is much less than that of pest. For grain
production in the national level, the ratio of weed epidemic areas to total grain sown
areas was 0.35 (Table 4). But it has increased significantly over last 8 years. By 1996,
the ratio rose to 0.57, 60 percent higher than that in 1988. On the average, the extent of
2
The data reported in this section are based on a reporting system from the village to township and
then to county, province and finally aggregating to a national level each year. This information used
internally by government as guidance for making their decisions on the policies related to pest
control, and assessment of local organization's pest control effort and performance. The caution
should be taken by reader in the extent of pest related problem as there is an incentive for the local
leaders to under-report the crop area actually affected by pest diseases, and over-report the crop
prevention areas and crop loss abatement.
5
weed problem in Zhejiang is about 50 percent higher than the national figures. The
evidence of increasing importance of weed problem in China also confirms to the
findings in the other countries (Rola and Widawsky, 1998).
Table 4 Extent of grain areas affected by weed problem in Zhejiang and China,
1989-96.
Ratio of epidemic area to
total grain sown area (%)
Ratio of the areas with weed control
treatment to total grain sown area (%)
Year
National
Zhejiang
National
Zhejiang
1989
1990
1991
1992
1993
1994
1995
1996
0.35
0.38
0.42
0.40
0.49
0.51
0.55
0.57
0.53
0.56
0.72
0.66
0.67
0.87
0.88
0.74
0.16
0.19
0.21
0.24
0.27
0.36
0.38
0.41
0.27
0.34
0.41
0.46
0.50
0.67
0.75
0.65
Source: Computed by authors based on the data from MOA, Agricultural Yearbook of China.
Tables 3 and 4 also show the effort taken by farmers to prevent crop loss due to pest
and weed problems. The higher ratios of prevented areas to total crop sown areas than
those of the epidemic areas for pest attacks in recent years may reflect the fact of
increasing availability of pesticides in the market and the ability of farmers to purchase
pesticides. At the national level, the areas with control treatment have been larger than
the estimated epidemic areas since the early 1990s (Table 4). But if we account for the
over-report of the prevention areas (a rate of 20 percent over-reporting was found in our
case study) and under report of affected areas, the area affected by pest may still higher
than the prevention areas in recent years. A higher rate of prevention in Zhejiang than
the national average is consistent with the extent of pest attacks and the level of market
development as well as income (credit constraint) in the province. A similar trend is
found in weed control effort by the farmers (Table 4).
Cost of Pesticide Use in Major Crop Production
Per hectare pesticide expenditures in all crop productions increased considerably in
the past decades. After deflating current value of pesticide use by the retail price index
of pesticide, the results indicate that per hectare real cost of pesticide rose 2.5 times for
rice, 3 times for wheat and 4.8 times for maize in 1980-97. The real costs of pesticides
per hectare of cropped areas increased by 1.5-2.6 times for fruits, vegetables and cotton
in the same period (Table 5).
The rates of pesticide application vary significantly among crops, reflecting the
extents of pest related problems differ across crops. On per hectare basis, fruit and
vegetable production uses much more pesticide than the other crops. Cotton farmers
apply 3 times more pesticide than rice farmers. Within grain productions, rice is the
most intensive pesticide use crop. Per hectare rice production costs 231 yuan of
pesticide in 1997, nearly 3 times of wheat and 4 times of maize (Table 5).
6
Table 5 Pesticide uses in major crop productions in China, 1980-97.
Year
Rice
Wheat
Maize Cotton Tomato Cucumber
1980
1985
1990
1995
1996
1997
29
40
103
207
224
231
1980
1985
1990
1995
1996
1997
87
118
129
207
204
214
1980
1985
1990
1995
1996
1997
5.8
6.0
7.5
7.0
6.9
7.3
Apple
Orange
8
8
31
64
99
83
Per hectare pesticide cost (yuan at current prices)
4
85
na
na
349
4
98
na
na
460
16
305
297
373
1342
59
834
868
803
2058
56
721
759
899
1702
59
728
1041
1035
1888
347
687
1695
1709
2083
1741
25
23
38
64
90
77
Per hectare pesticide cost (yuan at 1995 prices)
11
257
na
na
1048
12
292
na
na
1365
20
381
371
466
1677
59
834
868
803
2058
51
658
693
821
1554
55
674
964
958
1749
1044
2041
2118
1709
1902
1613
Share (%) of pesticide cost in total material costs of crop production
1.9
1.0
13.1
na
na
36.1
17.8
1.4
0.8
11.5
na
na
29.1
17.7
2.7
1.6
18.1
4.8
6.3
29.3
25.6
2.8
2.7
21.7
7.9
9.2
27.0
20.8
3.5
2.3
18.5
6.7
9.4
20.4
21.5
2.9
2.4
18.2
9.0
10.6
23.2
22.1
Note: Rural retail price index of pesticides is used to deflate the current value.
Source: State Economic Planning Commission.
It is worth to note that the increase in pesticide use has been higher than the
increase in other input uses in all crop productions over time. The cost shares of
pesticide in total material inputs raised from 5.8% in 1980 to 7.3% in 1997 in rice
production, 1.9% to 2.9% in wheat and 1.0% to 2.4% in maize during the same period.
The most significant increase in the pesticide cost share is cotton production, raised
from 13.1% in 1980 to 21.7% in 1995.
On the aggregate basis, rice is the largest pesticide "consumer". For the country as
a whole, rice farmers spent more than 7.3 billion yuan in chemical pesticides to control
pest problems in 1997 (Table 6). This was followed by apple (5.4 billion yuan), cotton
(3.3 billion yuan), wheat (2.4 billion yuan), orange (2.3 billion yuan), and maize (1.4
billion yuan) in 1997. For the 6 crops as a whole, farmers spent about 22.2 billion yuan
(or US$ 2.67 billion) on the pesticides to control their pest problems. These 6 crops
accounted for about 61.5% of the total crop areas in the past 2 decades. Applied a
similar rate of pesticide use to the rest of crops that are not included in Table 6, we
estimate that currently Chinese farmers may spend as much as 36.1 billion yuan
(US$ 4.34 billion) annually in the chemical pesticides. In 1992, Japan was the world
largest pesticide consumption country with a total amount of US$ 3.5 billion (Wood
Mackenzie Co., Ltd., 1993. cited in USAID 1994 and Yudelman et al 1998). Although
there is no comparable data available for Japan in 1997, our estimate for China's
7
pesticide use may indicate that China is likely to be the largest pesticide consumption
country in the world since the mid-1990s.
Table 6. Total pesticide expenditure in major crop productions in China, 1980-97.
Area
Selected 6 major plants
Year
share of 6 crops
Rice
Wheat
Maize
Cotton
Apple
6 crops
(%)
Orange
2070
2824
10397
22223
22472
22154
Million yuan at current prices
981
246
76
420
1275
228
69
506
3412
946
343
1705
6359
1840
1346
4524
7024
2923
1374
3403
7347
2498
1402
3268
257
398
2192
6079
5083
5359
90
348
1799
2075
2665
2280
1980
1985
1990
1995
1996
1997
6227
8384
12990
22223
20522
20519
Million yuan at 1995 prices
2950
738
230
1263
3785
677
205
1502
4263
1182
429
2130
6359
1840
1346
4524
6415
2669
1255
3108
6805
2313
1298
3027
774
1182
2738
6079
4642
4964
272
1035
2248
2075
2434
2112
1980
1985
1990
1995
1996
1997
Million US$ converted at official exchange rate
1382
655
164
51
280
172
962
434
78
23
172
136
2173
713
198
72
356
458
2661
762
220
161
542
728
2703
845
352
165
409
611
2672
886
301
169
394
647
60
119
376
248
321
275
1980
1985
1990
1995
1996
1997
61
60
63
61
62
61
Note: Rural retail price index of pesticides is used to deflate the current value.
Source: State Economic Planning Commission for cost of production data, crop areas that are used
in computing the total pesticide cost are from SSB, and exchange rates are also from SSB.
IV. Farmer Knowledge, Attitudes and Practices of Pesticide Use
It is often observed that farmers, the major pesticide users, are not fully aware of
the risks related to pesticide uses (Rola and Pingali, 1993). Misuse and overuse of
pesticides are often observed in the developing countries (e.g, Warburton et al., 1995;
Heong et at., 1995; Yudelman, 1998). A clear understanding of farmer's knowledge,
attitudes and practices of pesticide use is the first step in understanding the reasons
underlined the overuse of pesticide by farmers in crop production.
Farmers' Perception of Crop Loss due to Pest
Perceptions of pest related yield loss by farmers are important as their perceptions
will have a directly effect on amount of pesticide use by the farmers. A large literature
on the pest related yield loss show that the farmers' perceptions are far away from the
reality, which often stimulate overuse of pesticides (Tiat, 1997; Mumford 1981, 1982;
8
Norton and Mumford 1983; Pingali and Carlson 1985; Carlson and Mueller 1987; Rola
and Pingali 1993).
Table 7 summarized the studies that show rice yield losses due to insects, plant
diseases, weed or all pest-related yield losses in Asia. Cramer (1967) reported that all
pest-related yield losses were about 55%, among these, 34% due to insects, 10% due to
diseases, and 11% due to weed, in Asia. Ahrens et al (1982) reports a slight lower rate
of insect yield losses (24%) for East and Southeast Asia. While Pathak and Dhaliwal's
study (1981) in the Philippines recorded a higher rate of insect related losses
(35%-44%). Although the yield losses due to insect related problem vary from one
study to the others, most studies reported insect related yield losses of a similar
magnitude, ranging from 20% to 40% in most cases.
Table 7. Cross loss due to aggregate damage of pests in selected countries.
Country
Sources of
loss
Crop loss
estimate (%)
Authors
Year
Asia
Insects
Diseases
Weed
34
10
11
Cramer
Cramer
Cramer
1967
1967
1967
East & Southeast Asia
Insects
24
Ahrens et al
1982
Philippines
Philippines
Philippines
Philippines
Philippines
Philippines
Insects
Insects
Insects
Insects
Insects
All pests
20-25
16-30
35-44
29
18
35
Pathak and Dyck
Way
Pathak and Dhaliwal
Kalode
Litsinger et al
Waibel
1973
1976
1981
1987
1987
1990
Thailand
All pests
50
Waibel
1990
India
Sri Lanka
Insects
Insects
35
20
Way
Fernando
1976
1966
China
All pests
42
This report
1999
Sources: all figures are from a summarized table in Rola and Pingali (1993) and Waibel (1990) except
for China that are from the yield trials in this study.
In order to justify the pest related yield losses in our sample studied areas, with
help from the local plant protection bureau, a yield constraint trial was conducted in one
of our studied counties, Anji, to evaluate the impacts of pest related yield losses (insects,
diseases and weed).3 The results show that rice yield loss is about 41-43 percent of
production if "no pesticide is applied at all" (the last row, Table 7). The figure is lower
than those founded by Cramer (55%, 1967), and close to the average results for the
Philippines and Thailand in a more recent study by Waibel (1990).
3
The experiment was conducted in the same year using the most common variety (Xianyou 63)
adopted by the local farmers in Yushanwu village, the same village we conducted the survey and in
the same rice production season. The total experiment area is 4 mu (or 0.267 hectare, 15 mu = 1
hectare), which was divided into 4 plot with 2 cross classification: with and without pesticide uses,
and with chemical fertilizer (no organic fertilizer) and with organic fertilizer (no chemical
fertilizer).
9
But even assumed the true pest related yield loss is about 42% in our studied areas,
the figure is still much lower than the local farmers' estimation or perceptions on the
yield loss.4 On the average, the farmers in our samples expect that rice yield losses due
to pest problems are equivalent to 75.6% of the production (Table 8). It is also worth to
note that the variation of crop loss due to pest disease perceived by farmers is small
(ranging from 70.2% in Zhengbei to 80.8% in Yushanwu and relative small standard
errors within village, Table 8), indicating that the over perception of yield losses is very
common for the farmers in our sample.
Table 8. Farmers' perception and field experiment of crop loss due to pest problems in rice
production in the sampled areas of Zhejiang province, 1998.
Total samples
Zhengbei village
Yangzhuang village
Yushanwu village
Shuikou village
Experiment in Yushanwu
Crop loss estimates (%)
Standard errors
75.6
19.6
70.2
74.3
80.8
77.0
18.5
18.9
17.6
22.6
41.9
Sources: all data are based on authors' survey except for experiment data in Yushanwu that are
provided by Anji Agricultural Bureau, Zhejiang province.
While there is no significant difference in the yield loss perceptions for the farmers
who have education less than 9 years (lower than high school), the farmers with higher
education (more than 9 years, mainly high school) do have a lower yield loss perception
(Table 9). On the other hand, no significant differences of farmers' perceptions on rice
yield losses by ages are found in our samples.
Table 10 shows the farmers' perceptions on yield losses due to pest related
problems by land endowment and income. No formal conclusive results could be made
from the figures presented in this table. The differences of yield losses perceptions by
farmers are very marginal among various land endowments (per capita farm land) and
income groups.
The overestimation of pest related yield losses by farmer has been often mentioned
in the literature, but few of them have carefully investigated the reasons underlined the
estimates. Based on our intensive survey of 100 rice farmers in Xiaxing and Anji
counties of Zhejiang province, the econometric analysis of farmer’s perception on yield
losses due to pest related diseases concluded that education, farm size, occupation or
source of income, and village level extension system are major determinants of
farmer’s yield loss perception (Huang, et al., 2000).
4
The total samples are 100, with 25 farmers from each village, 2 villages from each county. The
surveys are conducted in Jiaxing and Anji counties, Zhejiang province in 1998.
10
Table 9. Perceived pest related yield losses (%) by the farmer's (respondent's)
characteristics in Jianxing and Anji counties of Zhejiang province, China.
Jianxing
Anji
Total
72.2
78.9
75.6
All samples
(50)
(50)
(100)
By age: < 40 years
73.9
(19)
76.0
(14)
74.8
(33)
40-50 years
64.8
(14)
83.9
(22)
76.4
(36)
> 50 years
76.5
(17)
73.9
(14)
75.3
(31)
Male
71.1
(42)
79.5
(48)
75.5
(90)
Female
78.4
(8)
65.0
(2)
75.7
(10)
72.3
(27)
81.0
(35)
77.2
(62)
Middle school
74.4
(17)
78.6
(11)
76.1
(28)
High school
65.8
(6)
61.0
(4)
63.9
(10)
By sex:
By education: Primary or less
Note: figures in the parenthesis are the number of sample. The total samples are 100, with 25
farmers from each village, 2 villages from each county. The surveys are conducted in Jiaxing
and Anji counties, Zhejiang province.
Table 10. Farmers' perceived pest related yield losses (%) by the farm's characteristics in
Zhejinag province, China.
Jianxing
Anji
Total
50
50
100
Per capital land
< 1.25 mu
75.2 (21)
78.3 (12)
76.4 (33)
1.25-1.75 mu
70.1 (20)
81.5 (13)
74.6 (33)
>1.75
70.0 (9)
77.8 (25)
75.7 (34)
Per capital income
<3000 yuan
71.8 (14)
74.2 (19)
73.2 (33)
3000-5000 yuan
67.7 (15)
79.7 (18)
74.2 (33)
>5000
75.8 (21)
84.6 (13)
79.2 (34)
71.0 (20)
80.4 (9)
73.9 (29)
15-30%
74.3 (15)
78.2 (17)
76.4 (32)
>30%
71.8 (15)
78.8 (24)
76.1 (39)
Total samples
Agricultural share in total
income
<15%
Note: figures in the parenthesis are the number of sample. Source: see Table 9.
11
Table 11. Farmers' knowledge on pest management.
By education
Average
<=6 years >6 years
All samples
Knowing pesticide hazard
-- Yes
66
36
30
-- No
34
26
8
Knowing pest enemies
-- Yes
35
27
8
-- No
65
35
30
Heard about IPM
-- Yes
49
28
21
-- No
51
34
17
Jiaxing county
Knowing pesticide hazard
-- Yes
30
14
16
-- No
20
13
7
Knowing pest enemies
-- Yes
13
8
5
-- No
37
19
18
Heard about IPM
-- Yes
20
9
11
-- No
30
18
12
Anji county
Knowing pesticide hazard
-- Yes
36
22
14
-- No
14
13
1
Knowing pest enemies
-- Yes
22
19
3
-- No
28
16
12
Heard about IPM
-- Yes
29
19
10
-- No
21
16
5
By sex
Male
Female
63
27
3
7
30
60
5
5
48
42
1
9
28
14
2
6
9
33
4
4
20
22
0
8
35
13
1
1
21
27
1
1
28
20
1
1
Note: The figures in the table are the numbers of farmers responded. The total samples are 100,
with 25 farmers from each village, 2 villages from each county. The surveys are conducted in
Jiaxing and Anji counties, Zhejiang province.
Farmers' Knowledge on Pest Management
Farmers' knowledge on pest management is examined based on their awareness of
the types of pesticides, pest enemies, alternative pest management measures, and the
changes in the extent of pest problems over time. Although the kinds and amount of
pesticide applied by farmers have increasing significantly over last 2 decade, the
farmers' knowledge on pest management is far away from we expected. Among 100
respondents, 34 farmers can not tell any differences in pesticide hazard for the
pesticides they used (Table 11). Even for those who replied they know the pesticide
hazard (66 farmers), their knowledge is very minimal.
About two third of farmers (65) do not know that pests have their "enemies", and
about half of farmers they even never heard about the integrated pest management
(IPM). This is surprise to us as both these 2 counties have been the experiment areas for
ecology-agricultural production bases for a long time.
12
No significant difference in the farmers' knowledge on the pest management is
found between Jiaxing and Anji counties. But education is found to have a positive
linkage with farmers' knowledge on pesticide hazard to the environment and human
health, and on integrated pest management (Table 11).
Most of farmers perceive that the kinds of and frequency of crop insects and
diseases did not change over last 15 years. Among 100 farmers interviewed, only 9 and
12 farmers perceive the increasing trend of insect and disease problems, respectively, in
the past 15 years (Table 12).
Table 12. Farmers' perceptions on the kind of and frequency of pest diseases over last 15
years.
By schooling
By sex
Average
<=6 years >6 years
Male
Female
Total Samples
Insect
No changes
88
55
33
78
10
Declining
0
0
0
0
0
Increasing
9
6
3
9
0
Do not know
3
1
2
3
0
Disease
No changes
83
55
28
73
10
Declining
2
1
1
2
0
Increasing
12
5
7
12
0
Do not know
3
1
2
3
0
Jiaxing county
Insect
No changes
46
27
19
38
8
Declining
0
0
0
0
0
Increasing
2
0
2
2
0
Do not know
2
0
2
2
0
Disease
No changes
44
26
18
36
8
Declining
0
0
0
0
0
Increasing
4
1
3
4
0
Do not know
2
0
2
2
0
Anji county
Insect
No changes
42
28
14
40
2
Declining
0
0
0
0
0
Increasing
7
6
1
7
0
Do not know
1
1
0
1
0
Disease
No changes
39
29
10
37
2
Declining
2
1
1
2
0
Increasing
8
4
4
8
0
Do not know
1
1
0
1
0
Note: The figures in the table are the numbers of farmers responded. The total samples are
100, with 25 farmers from each village, 2 villages from each county. The surveys are
conducted in Jiaxing and Anji counties, Zhejiang province.
Analyzing the data by schooling, sex and location do not find a significant
difference among their perceptions on changing trends of the pest diseases over last 15
years. The small sample on female farmer limits us to have a better understanding of
farmer's knowledge on pest management by sex. But 10 females interviewed all
13
responded that there has been no any change in the crop insect and diseases in their
fields (Table 12).
Farmers' Attitudes on Recommended Pesticide Usage
The township plant protection stations normally provide recommendation on the
pesticide uses for pest control in our sampled villages. In one of our surveyed villages,
Zhengbei, farmers received a notice in written (1-2 pages) from the technician at the
township plant protection station for every application of pesticide in controlling pest
diseases in rice production. The notice includes the timing of the pesticide application
and the kinds of pesticide uses for the insect or diseases. For the other three villages,
while there is no this kind of formal notice, similar information is passed to farmers
through radio broadcast system.
However, most farmers do not believe the recommendation and the prescription
labeled in pesticide products. They consider them as under-use of pesticide if they
apply the pesticide based on the recommendation package. In our samples, only 30
(among 100) farmers consider that the recommendation is "adequate" (Table 13). But
when farmers apply the pesticides in their fields, only 14 farmers use pesticide
following the introduction provided in the label to some extents. All other farmers
apply more than the recommendation dosage. The proportion (8 out of 10, or 80 percent)
of female who consider the recommended amount of pesticide use as "too less" is
higher than that for male (51 out of 90).
Table 13. Farmers' attitudes on recommended pesticide usage.
Questions:
Average
By schooling
<=6 years >6 years
By sex
Male
Female
How do you evaluate the prescription labeled in pesticide products?
-- Adequate
-- Too little
-- Too much
-- Do not know
30
59
4
7
23
32
2
5
7
27
2
2
28
51
4
7
2
8
0
0
Do you apply more/less pesticide than prescription labeled in pesticide products?
-- About same levels
-- Apply more
-- Apply less
-- Do not know
14
84
0
2
11
50
0
1
3
34
0
1
13
75
0
2
4
9
0
0
Note: The figures in the table are the numbers of farmers responded. The total samples are 100,
with 25 farmers from each village, 2 villages from each county. The surveys are conducted in
Jiaxing and Anji counties, Zhejiang province.
Farmers' over perception of crop yield loss due to pest and their doubts on the
technician's recommendation and pesticide prescription provided by pesticide industry
(label) directly related to farmers' over use of the pesticide in crop production. Risk
consideration is often cited as major reason in the literature for farmer pest management
behavior. There are less than half (8+38=46) of the farmers (100 interviewed) who
primarily use the information provided by technician and pesticide label (Table 14).
The traditional information channels (own experience, friend, relatives), new channel
from pesticide salesman and others accounted for 54 percent of the sources of
14
information.
Table 14. Sources of information on pest management
Total samples
-- Jiaxing
-- Anji
Pesticide
label
Techni-c
ian
Own
experience
Friend &
relatives
Pesticide
salesmen
Others
8
38
25
4
18
7
5
10
54
23
19
30
5
2
13
24
3
11
Note: The figures in the table are the numbers of farmers responded. The total samples are 100,
with 25 farmers from each village, 2 villages from each county. The surveys are conducted in
Jiaxing and Anji counties, Zhejiang province.
Farmers' Pesticide Purchasing Behaviors
Farmers' response to the pesticide prices: One of the most interesting findings
from these experiments is the farmers' response to pesticide price changes. Within a
total number of 100 farmers interviewed, only 5 farmers responded that they would
reduced their pesticide use by 22% (last column) if the pesticide price were raised by
50% (Table 15). Even if the pesticide prices are assumed to be doubled, the marginal
increases in the number of farmers who response to reduce pesticide uses are only 2
farmers. Further increasing the pesticide price by triple (rise by 200%) still does not
have any effect on 87 farmers’ decisions on their pesticide uses (Table 15). For average
of the samples, the reduction due to a rise (50 percent) in pesticide price is only 1.1%. A
similar result is found when we asked farmers to respond to the declines in the pesticide
prices. This price response is much lower than that found in a similar study in the
Philippines (Rola et al, 1988). Rola et al found that about 60% of rice farmers their
decisions on pesticide use did not affected by any changes in pesticide prices. Our field
interview with farmers on the pesticide use reveals that the inelasticity of pesticide use
with response to price changes is mainly due to the exceptionally high level of farmers'
expected pest yield loss, ranging from 30% to 100% with an average of 76% for the
whole samples.
Table 15. Farmers responded to the hypothesized pesticide price changes.
Pesticide
price changes
Price raised by:
50%
100%
200%
Price declined by:
50%
75%
All
farmers
Farmers
responded
without
changes in
pesticide use
Farmers
responded
with change
in pesticide
use
Changes in pesticide
use (%)
For those
All
responded
farmers w/ changes
100
100
100
95
93
87
5
7
13
-1.10
-1.61
-9.10
-22
-23
-70
100
100
97
95
3
5
+0.69
+1.90
+23
+38
Note: The total samples are 100, with 25 farmers from each village, 2 villages from each
county. The surveys are conducted in Jiaxing and Anji counties, Zhejiang province.
15
Farmers' choices on pesticide brand and frequency of false pesticide. With various
kinds/brands of pesticides available in the markets, and frequent reports of false
pesticide (most are low quality or less effective than it should be) sold in the markets,
about half of farmers (51%) do not have any idea on the brand which they are going to
purchase. There are 16 farmers who actually applied these false pesticides at least once
in the past 15 years. Twenty-one farmers replied that they think the false pesticide has
becoming more severe over time.
Sources of pesticide purchased. On the average, each household purchases more
than 15 kilograms of pesticides within one year. Among them, about 54 percent of
pesticides were purchased from the nearby Agricultural Input (Materials) Cooperative
and Agricultural Technology Extension Station at both county and township levels.
With the liberalization of pesticide market at retail level recently, private trade is
becoming a dominant force in pesticide market. Private traders provided 46 percent of
the total pesticides used by the farmers.
Safety and Storage Practices
Farmers' perception on pesticide residue. Farm pesticides are components
artificially introduced into and generally incompatible with agricultural ecosystems.
When applying pesticides, about 20-30% of them absorbed by the crops, the rest of the
pesticides mainly enter agro-environment. However, many farmers do not have any
knowledge on whether pesticide could have impact on the environment. Nearly one
third (31%) of farmers do not realize that pesticide residue may remain in paddy grain.
More educated farmers are more awareness of pesticide residue problem. Most of
farmers who completed mid-school or higher education believe pesticide residue will
remain in rice paddy (74%), the figure is 8% higher than those farmers who only
attended 6 years or less in the primary school.
Pesticide storage, disposal and application practices.
The survey found that
90 percent (1st row, Table 16) of farmers place the pesticide bottle in a special location
inside their houses and difficult for children to reach. However, there are still 10
percent of farmers who place the bottle in unsafe places such as on the ground and
mixed with other kinds of bottles without any prevention.
Disposal of empty pesticide bottles is the other concern of the safety. Many farmers
(43 percent, 4th row of Table 16) dispose the empty bottle in river, field and outside
house. About one out of five farmers kept the empty bottle for other uses, and one third
of farmers sold the bottles to the bottle collectors. Disposal of remaining pesticide in
sprayer presents a less problem than the disposal of the empty bottle in our samples.
Only about 5 percent of farmers dump the remaining pesticide into rive and other place
that may hurt human and animal health (9th row, Table 16).
Sprayer maintenance. All farmers interviewed have their own sprayers. But about
half (48%) of farmers report their current sprayers have linkage problem, among them
15% are severe in linkage. Only those with severe linkage sprayers are planned to
purchase new one in the near future. As the prices of the sprayer is cheap (about 70-80
yuan, or US$ 9), farmers normally purchase new sprayers instead of repairing it when they
decide to change one.
16
Table 16. Farmers' practices on pesticide storage and disposal practices.
Total
Male
Female
a
Pesticide storage after purchased :
Safe storage practices (%)
unsafe storage practices (%)
90
10
100
0
89
11
Disposal of empty pesticide bottles:
Sold (%)
Disposed of in river, field, outside house (%)
Keep for other uses (%)
Continued for pesticide storage or buried (%)
31
43
19
7
28
45
20
7
60
20
20
0
Disposal of remaining pesticide in sprayer:
Continually used in the same field (%)
Dumped in the field (%)
Dumped in river and other place (%)
83
12
5
84
10
6
70
30
0
Sprayer use practices:
Wash every time after use (%)
Wash most of time (%)
Wash rarely or never wash (%)
60
16
24
60
17
23
60
10
30
Places of wash sprayer
River (%)
Paddy field (%)
Other places (%)
94
3
3
94
2
4
89
11
0
a: Safe storage practice = the bottle is placed in a special place inside the house and difficult for
children to reach; unsafety = all other practices.
Source: Author's survey.
Our survey also found that not all farmers wash their sprayers after use. Only 60
percent of farmers wash sprayer every time after use (Table 16). The farmers who
rarely or never wash their sprayers account for 24 percent. They most often wash
sprayers in river (94 percent).
Protection measures. Most farmers spray pesticides away from the wind.
Although almost all applicators are partially covered with protective clothing (long
sleeves or/and long pants), none of them use masks. Due to the lack of awareness of
pesticide hazards, nineteen percent of farmers normally do not take a bath after
pesticide application. Thirteen percent farmers commonly wash face only after the
application and 6 percent farmers even wash their hands only.
Although farmers' knowledge on pesticide safety issues is better than pest
management, but it is also very limited and varies among different levels of education.
Because of their limit knowledge on the pest management and high risk aversion of
farmers in pest control, farmers' over estimation of crop yield losses and very
inelasticity of the farmers' pesticide use response to the pesticide price are inevasible.
The limit knowledge on pest management, much high perception of crop yield
losses, and very inelasticity of the pesticide price response indicate that the effort to
convincing farmers to reduce the extent of pesticide use will mainly rely on the factors
17
that could significantly reduce farmers' risk aversion. These might include pest
management related information, education, training and extension services.
V. Chemical Pesticide Adoption and Its Impacts on Rice Production
There is a wide debate on the contribution of pesticides to crop production
(reducing loss) and the negative impacts of their uses on the environment and human
health5. For example, Oerke et al. (1995) reviewed a large number of literatures where
chemical pesticides have significantly reduced pest related crop yield losses. On the
other hand, there are also a large number of literatures that show the negative impacts of
chemical pesticides are substantial. Several economic studies have questioned whether
current patterns of pesticide use are economically and socially efficient (e.g., Pimentel,
D. and H. Lehman. 1992; Pingali and Roger, 1995; Yudelman et al.,1998). Some
studies show that the costs (both economic and social costs) related to pesticide use in
crop production even higher than the gains from the reduction of crop yield losses (e.g.,
Pingali and Roger, 1995).
While studies of pesticide productivity are relatively common, few researchers
have assessed the farmer's pesticide adoption behaviors, and no study on the impacts of
chemical pesticide was found in China except for a recent study by Widawsky et al.
(1998). The latter focuses on pesticide productivity and host-plant resistance in China.
That study show that under intensive rice production systems in Jiangsu and Zhejiang
provinces of eastern China, pesticide productivity is low compared to the productivity
of host-plant resistance. Indeed, they found that returns to pesticide use are negative at
the margin.
To achieve optimal pesticide use in economic sense (in the next section we will
deal with human health aspects), answers to the following questions are essential to
formulate effective policies and regulations on pesticide uses. What is the extent of over
use of pesticides by farmers? What are the major factors that have significant effects on
farmer's pesticide applications? And what is the productivity of pesticide use in rice
production?
The objectives of this section are three folds: to examine the extent of pesticide
uses at farm level in rice production, to investigate the determinants of pesticide
adoption by rice farmers, and to estimate the productivity of pesticide use in rice
production. To achieve these objectives, this section is organized as follows. After rice
production and inputs in our sampled households are discussed, adoption of pesticide in
rice farming is provided. An empirical model then is developed to investigate the
factors that determine the pesticide adoption behaviors and impacts of pesticide on rice
production. Based on our 100 primary rice farms survey, rice production function with
endogenous of pesticide adoption model is estimated.
Rice Production in the Sample Households
Farm size in our sample (0.50 hectare) is slightly smaller than the country average
(0.57 hectare). Farmers allocate more than 80 percent of their land for grain production.
Rice is the most important grain and accounts for 58 percent of total crop sown areas
5
A comprehensive review of the literature could be found, for example, in Oerke et al.,1995;
Pingali and Roger, 1995, Pimentel, D. and H. Lehman. 1992,and Yudelman et al. (1998), etc.
18
(Table 17).
Table 17. The importance of rice production in the sampled households by village.
4 villages
Zhengbei Yangzhuang Yushanwu Shuikou
Farm size (ha)
0.50
(0.33)
0.51
(0.29)
0.47
(0.47)
0.50
(0.19)
0.54
(0.32)
Rice sown area (ha)
0.50
(0.39)
0.63
(0.29)
0.66
(0.58)
0.23
(0.10)
0.46
(0.28)
Rice share in total crop sown
area (%)
58
(17)
60
(7)
64
(8)
40
(19)
68
(17)
Share of doubled-season rice
in total rice sown areas (%)
37
(41)
69
(32)
80
(15)
0
(0)
0
(0)
Share of single-season rice in
total rice sown areas (%)
63
(41)
31
(32)
20
(15)
100
(0)
100
(0)
Proportion of paddy land (%)
86
(15)
91
(6)
92
(7)
68
(17)
93
(7)
Note: Standard errors in the parentheses. The statistics in the table are from 100 households in 4
villages of Jiaxing and Anji counties in Zhejiang.
Most of rice production in the two villages of Jiaxing county is planted under
double-season rice cropping system, accounted for 69 percent of the total rice sown
areas in Zhengbei village, and it is 80 percent in Yuangzhuang village. In these 2
villages, the common cropping patterns are rice-rice and cash crop-wheat. While in
Yushanwu and Shuikou villages of Anji county, all rice is planted under single-season
rice cropping system. The common cropping patterns in Anji county are rice-wheat,
rice-rapeseed, and cash-cash crops.
Rice yield in our sampled households is slightly higher than the national average,
ranging from about 6 ton/ha to 7.3 ton/ha. The yields of single-season rice (6.34-7.30
ton/ha) are general higher than double-season rice (5.96-6.57 ton/ha) because of longer
growing period of single-season rice (Table 18).
Table 18 Yield (ton/ha) of rice production in the sampled households, Zhejiang, China, 1998.
Double-season
Double-season
Single-season
Single-season
early rice
late rice
middle rice
late rice
Village
Mean
Std.
Mean
Std.
Mean
Std.
Mean
Std.
Zhengbei
6.08
0.84
6.57
0.93
7.30
1.05
Yangzhuang
5.96
0.51
6.25
0.80
6.66
0.94
Yushanwu
6.34
1.33
6.24
1.07
Shuikuo
6.63
0.76
6.21
0.70
Source: Authors' survey.
On the average, farmers use 179 labor days and 282 kilogram of fertilizer for every
hectare of rice area. Imbalance of chemical fertilizer elements (N:P:K) found in the
early studies (Huang et. al., 1994 and 1995) is also evidenced in our study areas. The
ratio of N:P:K is about 14:1:4, obviously the phosphate and potash fertilizers are
19
relatively underused compared to the nitrogen fertilizer.
Table 19. Per hectare input in rice production in the sampled households, Zhejiang, 1998.
Mean
Standard error
Labor (days/ha)
Fertilizer (kg/ha)
Pure N (kg/ha)
Pure P (kg/ha)
Pure K (kg/ha)
Pesticide use:
Volume (kg/ha)
Cost (yuan/ha)
Average mixed price (yuan/kg)
179.23
282.37
205.48
15.02
61.87
66.05
106.36
71.66
28.63
50.26
27.65
287.14
10.89
11.82
142.60
5.07
Source: Authors' survey.
Average pesticide use per hectare reaches 27.67 kilograms for each rice growing
season, which costs farmer 287 yuan (or US$ 34.7). As the surveys were conducted in
the same time in the 4 villages in the same provinces, we expected that, for the cross
section data, the prices paid by farmers for individual pesticides should be the same or
nearly same for all farmers in the surveyed areas.6 However, there is a larger difference
in the prices among different pesticides, reflecting differences in the quality of
pesticides. The mixed average price (a better term might be “quality index”) of all
pesticides is 10.89 yuan/kg, ranging from 3.70 yuan/kg to about 50 yuan/kg with a
standard error as high as 5.07 (Table 19).
Adoption of Pesticide in Rice Farming
Rice farmers in our sample area apply pesticides 8.8 times per year for rice
production (Table 20). Desegregating farmers into different groups based on farm and
farmers' characteristics indicate that the pesticide application levels are associated more
with sex, age and sources of pest management information (Table 20). Because the
variable presented in Tables 20 are strong linkage with farmer's perception on rice yield
loss due to pest related diseases (their impacts on pesticide use might be through
farmer's perception variable), testing the impacts of these variables requires a more
sophisticated model that is developed in next section.
Compared to the other Asian countries, farmers in our sample areas apply
exceptional amount of pesticide in rice production. Average application of pesticide per
hectare rice (per season) amounts to 27.7 kilograms in dosage or about 12-14 kilograms
in active ingredients (Table 8.5). This is about a similar level in Japan (14.3 kilograms
in active ingredients) and The Republic of Korea (10.70 kilograms in active
ingredients), and much higher than the rest of Asian countries (Barker and Herdt,
1985).
On the average for each season (not each year as some farmers plant two seasons
rice), farmer applies 8 times pesticides with about 3 times during seedling period and 5
times in the field (after transplanting, Tables 21 and 22). Table 8.5 also shows the
6
Indeed the prices of individual pesticides which we obtained from both the local pesticide retail
survey and farmers’ survey show no significant price variation for each pesticide (only 1-3%).
20
variations of pesticide use among villages. The differences between villages could be
attributed to the differences in rice cropping patterns, quality of agricultural extension
service, location specific fixed factors, etc. Lower values of pesticide use in this table
compared to the figures in Table 20 are due to higher weight given to the farmers
planting single-season rice, the later has a higher pesticide application rate.
Table 20. Pesticide use in rice production, farmers' characteristics and application
information sources in the sampled households in Zhejiang, China, 1998.
Sample
size
(n)
Farmer’s group
All farmers
By sex
Male
Female
By age
< 40
= 40-50
>= 50
By Schooling years
=< 6 years
> 6 years
By per capita land
< 1.25 mu
= 1.25-1.75 mu
> 1.75 mu
By information sources of pest
management
ATES
Others
Pesticide use per hectare per year
Times
Dosage
Cost
(n)
(kg)
(yuan)
100
8.8
29.5
318.7
95
5
8.9
6.3
29.9
22.8
322.8
242.2
33
36
31
8.8
8.8
8.7
30.2
28.2
30.4
304.4
327.2
324.2
62
38
9.0
8.4
30.8
27.5
330.4
299.7
33
33
34
7.9
8.5
9.8
29.1
25.3
34.1
312.0
275.2
367.6
39
61
9.0
8.4
30.7
27.8
340.1
285.4
Note: See Table 17. ATES – Agricultural Technology Extension Station. Others include own
experience, other farmers, pesticide retails etc.
Table 21. Pesticide use in rice production in the sampled households in 1998.
Village
All 4 villages
Sample
size
(n)
Pesticide use per hectare rice sown area
Times
Dosage
Cost
(n)
(kg)
(yuan)
200
8.0
27.7
287.1
Zhengbei
68
7.3
23.7
206.4
Yangzhuang
72
7.0
26.6
284.3
Yushanwu
28
9.2
32.7
436.7
Shuikou
32
10.5
33.9
334.1
Note: The data are summarized by double-season early rice, double-season late rice,
single-season middle rice and single-season late rice. The effective samples are 200, that is, on
the average each farm plan 2 kinds of rice. The conversion rates of "dose" to active ingredients
vary among the pesticides applied by farmers. But the average of conversion rate is about 45%.
The figures in this table differ slightly from Table 8.4 since the later is based on farm as base
observation (or sample base unit).
21
During the field survey, we told by the local farmers that the frequency of pesticide
application was only about 4-5 times for each rice growing season in 1970s. Significant
increase in the number of pesticide application occurred in the early 1980s when the
collectively owned land was distributed to individual household. The reform increased
the production incentives of farmers (Lin, 1992; Fan, 1991; Huang and Rozelle, 1996),
but there is also evidence that China's technology generation and extension systems
may have been weakened after the reform which could increase the cost of technology
adoption (Lin, 1991; Huang and Rozelle, 1996).
The frequency of pesticide application in Zhejiang province is about 2-3 times
higher than the average for the rest of Asian countries (Table 22). Although the
protection rate of pesticide price is relatively low in China, pesticide expenditure per
hectare is also one of the highest in Asian countries.
Table 22. Frequency and expenditures of pesticide application.
Region/Country
Number of applications
(n)
Expenditure
(US$/ha)
India
Philippines
Indonesia
Northern Vietnam
Southern Vietnam
2.4
2.0
2.2
1.0
5.3
24.9
26.1
7.7
22.3
39.3
China, Villages (this study, rice)
8.0a
38.5
a: 3 in seedling period and 5 after transplanting
Sources: China data are from the authors' rice farming survey in Zhejiang province in 1998.
Others are data for 1990-91 reported in Dung and Dung, 1999.
Pesticide use also differs by rice cropping system. While double-season rice uses
much higher dosage of pesticides measured at annual basis (add-up double-season early
and late rice), computed at per crop season base, pesticide used in singe-season rice is
higher than that in double-season rice (Table 23). The low level of pesticide
applications in double-season early rice is mainly due to less application of pesticides
during the rice seedling period and less pest related problems during April to June. The
longer rice growing period, the larger pesticide use such as the single-season middle
and late rice (Table 23).
Table 23. Pesticide use by rice season in the sampled households, 1998.
Rice season
All rice
Sample
Size
(n)
Pesticide use per hectare
Times
Dose
Cost
(n)
(kg)
(yuan)
200
8.0
27.7
287.1
Double-season early rice
47
5.7
20.6
174.5
Double-season late rice
47
8.7
28.4
288.9
Single-season middle rice
35
10.0
31.0
325.0
Single-season late rice
71
7.9
30.2
341.8
Note: see Table 21.
22
The levels of pesticide application is also found to have a positive relationship with
farmer's perceived yield losses due to pest attacks (Table 24). For example, the farmers
with less than 70 percent of rice yield loss perception, their pesticide application dosage
and costs are about 15 percent and 24 percent less than other farmers with more than 80
percent of rice yield loss perception (26.2 kg vs 31 kg, 3rd column, and the last column
of Table 24, 275.1 yuan vs 342.1 yuan).
Table 24. Pesticide adoption by farmers' perceptions on rice yield loss due to pest in the
sampled households in Zhejiang, 1998.
Sample
size
(n)
Times
(n)
All farmers
100
8.8
29.5
318.7
Perceived yield losses
< 70%
28
7.8
26.2
275.1
=70-80%
37
9.2
30.1
329.7
> 80%
35
9.0
31.0
342.1
Farmer’ group
Pesticide use per hectare
Dosage
Cost
(kg)
(yuan)
Note: the figures presented in this table are analyzed based on farm households (100 ).
Pesticide Adoption and Rice Production Model
Damage Control Production Function
To estimate the impact of chemical use on rice productivity, a production function
approach will be used. However, the roles of pesticide and other inputs on rice
production differ by nature. Inputs such as fertilizer and labor are treated as "normal"
inputs that contribute to yield increase, while pesticide is damage abatement input. In
the production function, the latter is a combination of yield loss and the output that is
recoverable through limiting this loss with damage control inputs. Following the works
by Headley (1968) and Lichtenberg and Zilberman (1986), a damage abatement
function will be incorporated into the traditional models of agricultural production.
The nature of damage control suggests that the observed crop yield, Y, can be
specified as a function of both standard inputs and damage control measure as:
(1) Y = f (X) G(Z),
where X is a vector of standard inputs such as labor, fertilizer, and other inputs. G(Z) is
a damage abatement function that is a function of the level of control agent, Z (e.g.,
pesticide use level). The abatement function possesses the properties of a cumulative
probability distribution. It is defined on the interval of [0, 1]. G(.) = 1 indicates a
complete abatement of crop yield losses due to pest attacks with certain high level of
control agent, while G(.) = 0 represents complete destroy of crop production by the pest
damage at a certain low level of control measure. G(.) is a function of non-decreasing
in Z and approaches one as damage control agent use increases. Assuming a
Cobb-Douglas production function for f(X) and the damage abatement function, G(Z),
follow either exponential or logistic specification, then equation (1) can be written as:
23
(2) Y = a0 ∏in Xiai [ 1 - exp(- c Z)],
or
(3) Y = b0 ∏in Xibi { 1/[1 + exp (d - h Z)]}
where a0, ai, c in (2) and b0, bi, d and h in (3) are parameters to be estimated. i index
inputs, including labor, various fertilizers (N, P, and K). Z is the pesticide use by farmer.
The model (2) and (3) are estimated by nonlinear methods. In order to compare the
results from the traditional production approach, a C-D production function is also
estimated using OLS, where pesticide use is specified as the same as other inputs such
as labor and fertilizer.
Farmers’ Pesticide Adoption Model
The models specified above, however, do not account for endogenous problem of
pesticide uses. Since pesticides are applied in response to pest pressure, and high levels
of infestations are correlated to lower yields, then a relationship between pesticides and
negative residuals in production function might bias parameter estimates for pesticide
and other variables. In the other word, pesticide adopted by farmers may be endogenous
to production and there is a systematic relationship among plant diseases, pesticide use,
and rice yield, a system model of pesticide adoption and production response should be
estimated using a maximum likelihood estimation technique. To empirically account
for this endogenous problem, the farmer’s pesticide adoption model is estimated first.
The predicted values of the pesticide use then are used in the estimation of model (2)
and (3).
Farmers' pesticide adoption behavior is hypothesized to depend on the incentive of
pesticide application (private profitability), farmer' characteristics and environments
where the crop production activities are based. Although no input and output price
variations observed for individual pesticide at given time, there is a large variation on
the mixed average prices of pesticides by household (reflecting the quality and
composition difference among farmers). The private profitability of pesticide use is
proxied by farmer's expected yield losses due to pest problems and the mixed average
price of pesticide.
Farm and farmer's characteristics could include farmer's education, age, gender,
occupation, farm size, etc. Impacts of these variables on farmer's pesticide application
might have two forms. One is their direct impact on farmer's pesticide use and the other
is their indirect impact through farmer's perception on yield loss. For example,
education and farming experience both are expected to relate to the quality of farmer's
judgement and expectation, and therefore the farmer's perception on yield loss. So in
specifying the pesticide adoption model, we will include both farmer's perception and
farm and farmer's characteristics variables.
Environmental factors that might have impacts on farmer's pesticide adoption
could include the sources of pest management information (quality of plant protection
extension services), rice cropping seasons, and a set of variables for village dummies
that accounted for any other unexplained regional fixed factors.
Based on the above discussion, farmer's pesticide adoption (Pesticide) model can
be explained by the following equations:
24
(4) Pesticide = f (Profitability; Farm Characteristics; Extension Service; Others)
= f ( Yloss; Pmix, Education, Age, Sex, RiceArea, LaborCost;
ExtService, Season Dummies, Village Dummies).
where Yloss is farmer’s perception on rice yield loss due to pest problems. Pmix is
average mixed unit value of pesticide (a proxy for pesticide quality). Education is a
dummy variable with 1 for the farmers with education levels higher than middle school,
0 otherwise. Sex equal 1 for male and 0 for female. Age is measured in years.
LaborCost is the opportunity cost of labor and is proxied by per capita income in the
regression. ExtService is a variable representing the quality of service provided by the
local extension system. Since village dummies will take part of ExtService effects on
farmer's perception, we use the sources of pesticide application information as a proxy
for plant protection extension service. RiceArea is average rice area per season. Based
on rice cropping system, rice is divided into double-season early rice, double-season
late rice, single-season middle rice and single-season late rice. Dummy variables are
used to differentiate the variations of pesticide uses among various seasons of rice.
Among above variables, Pmix, Age, Sex, RiceArea, LaborCost and ExtService are
considered as instrumental variables in estimating pesticide use, the later is used as a
right hand side variable in rice production function. The dependent variable, Pesticide,
is defined in three alternatives: frequency (times), quantity (kg/ha) and cost (yuan/ha)
of pesticide application for each season of rice. The model is estimated by ordinary
least square method.
Estimation Results and Discussion
While the damage control rice production model and pesticide adoption model are
estimated simultaneously with the predicted values of pesticide use in damage control
model, we will discuss the results of adoption and production function estimations
separately.
Pesticide Adoption
Alternative specifications of models show that the results are robust. All models
give reasonable high R2 values, ranging from about 0.4 to 0.6, for cross-sectional
household data. As we expected, farmer’s perception on yield loss has the largest
positive and statistically significant effects on the pesticide applications.
In terms of volume and cost of pesticide use, a 10 percent decline in farmer’s
perception on yield loss due to pest (i.e., from current average level of 75.6% to 65.6%)
will reduce farmer’s pesticide use by 2.2 percent (the 2nd row in Model II of both Tables
25 and 26). The field experiment on yield loss in our study area indicates that the over
estimation of yield loss from pest problems is about 35 percent. Therefore a significant
reduction on farmer’s pesticide application could be achieved through providing
farmers better information on the real impact of pest problems and reducing farmer’s
attitude to the risk aversion in the pest management. Similarly, a 10 percent decline in
farmer’s yield loss perception will reduce the number of pesticide application by 2-3
times (from current 8.0 to 5-6) for each season rice (Appendix Table 1).
25
Table 25. Estimated parameters for amount of farmers' pesticide application per rice growing
season in Zhejiang province.
Pesticide use (kg/ha)
Ln (kg/ha)
Variables
Model I
Model II
Intercept
43.17
(6.35)***
5.58
(0.71)***
Farmer's perception on yield loss
7.86
(3.71)**
0.22
(0.13)*
Average mixed price of pesticides
(AMPP)
Ln (AMPP)
-1.04
(0.15)***
Information from ATES
3.00
(1.60)*
Ln (Rice area)
0.12
(0.21)
-0.63
(0.07)***
Rice area
Income/person
-0.05
(0.04)
0.0004
(0.0003)
Ln (Income/person)
Age
0.06
(0.05)
-0.23
(0.09)***
Ln(Age)
Dummies:
Double-season later rice
0.13
(0.06)**
-0.41
(0.13)***
9.26
(1.96)***
0.46
(0.07)***
Single-season middle rice
-0.92
(3.27)
0.17
(0.12)
Single-season late rice
8.86
(1.98)***
0.42
(0.04)***
Higher than middle school
1.59
(2.51)
0.08
(0.09)
Male
0.19
(2.23)*
-0.02
(0.08)
Village dummies:
Zhengbei
-17.13
(3.10)***
-0.65
(0.11)***
Yangzhuang
-9.92
(3.05)***
-0.30
(0.11)***
Yushanwu
0.20
(2.78)***
-0.03
(0.10)
R2
0.41
0.48
Adjusted R2
0.37
0.44
a: the figures in the parentheses are standard errors of estimates. ***, **, * denote significance
at 1%, 5% and 10%, respectively.
26
Table 26. Estimated parameters for pesticide cost of per hectare rice production in Zhejiang
province.
Pesticide cost (yuan/ha)
Ln (yuan/ha)
Variables
Model I
Model II
Intercept
279.26
(71.38)***
5.58
(0.71)***
Farmer's perception on yield loss
58.92
(41.67)
0.22
(0.13)*
Average mixed price of pesticides
(AMPP)
Ln (AMPP)
5.79
(1.65)***
Information from ATES
30.00
(17.94)*
0.37
(0.07)***
Ln (Rice area)
-0.05
(0.05)
Rice area
2.08
(2.39)
Income/person
0.0015
(0.004)
Ln (income/person)
Age
0.06
(0.05)
-1.66
(0.97)*
Ln(Age)
Dummies:
Double-season late rice
0.13
(0.06)**
-0.41
(0.13)***
103.51
(22.09)***
0.46
(0.07)***
Single-season middle rice
-4.81
(36.75)
0.17
(0.12)
Single-season late rice
97.68
(22.31)***
0.42
(0.07)***
Higher than middle school
29.37
(28.21)
0.08
(0.09)
Male
-35.06
(25.10)
-0.02
(0.08)
Village dummies:
Zhengbei
-182.12
(34.88)***
-0.65
(0.11)***
Yangzhuang
-95.76
(34.26)***
-0.30
(0.11)***
Yushanwu
31.91
(31.25)
-0.03
(0.10)
0.49
0.45
0.59
0.56
R2
Adjusted R2
Note: the figures in the parentheses are standard errors of estimates. ***, **, and * denote
significance at 1%, 5% and 10%, respectively.
Interesting is that rice area has a significant positive impact on the frequency of
27
pesticide application, while it dose not on the volume and cost of pesticide use by
farmer. Doubling farm size will increase pesticide application by 0.67 time (column 2
of Appendix Table 1), but this does not have impact on the total pesticide used (Tables
24 and 25). The significant negative coefficients for age in pesticide cost and volume
specifications, but not significant in frequency indicate that old farmer applies as many
times of pesticide in rice production, but the amount of pesticide use is less than that by
youth farmers (Tables 25 and 26).
Average mixed pesticide price (or quality of pesticides) has expected negative
impact on the volume and positive impact on the total cost of pesticides. Improvement
in pesticide quality seems to be very important in reducing the pesticide uses. For
example, doubling the average mixed pesticide price with better quality of pesticides
available in the market, the total cost of pesticide use increases only by 37 percent
(Model II of Table 26), while the total pesticide use declines by 63 percent (Model II in
Table 25).
Significant differences of pesticide uses among various rice seasons are also
expected. For example, after control for all other factors, the pesticide application in
the double-reason late rice is 9.26 kilograms (Model I, Table 25), 2.85 kilograms or 46
percent (Model II, Table 25) more than the double-reason early rice. The pesticide
application in the single-season late rice is about 42 percent more than the
double-season early rice. While the pesticide use does not differ significantly between
the single-season middle rice and the double-season late rice. This is somewhat
unexpected. The highly correlation between single-season middle rice (almost all in
Jiaxing county) with village dummies might contribute to the insignificant of the
single-season middle rice coefficient.
A very significant (both statistically and largely) lower level of pesticide use in
Zhengbei and Yangzhuang of Jiaxing county compared to the other villages confirms
the fact that a better crop plant protection service provided by the local agricultural
technology extension station in Zhengbei and Yangzhuang villages.
Impacts on Rice Production
Given the robust estimates of pesticide adoption models presented in the last
sub-section, we use the predicted value of pesticide adoption from Model II of Table 26
as explanation variable in rice yield equation because the model has the highest R 2
value among all estimated adoption models. Non-linear estimates of damage control
rice production functions with the endogenous pesticide use are summarized in Table
27.
After control for other inputs and location impacts, higher yield for both
single-season middle and late rice (compared to double-reason early and late rice) is
expected. The yield of single-season middle rice is about 12-14 percent higher than the
yields of both double-season early and late rice. And the yield of single-season late rice
is about 8-10 percent higher than the yields of double-season rice. The results are very
similar in the three models estimated in Table 27.
After control for the yield differences due to rice growing season, a significant
positive coefficient of Zhengbei reflects a better infrastructure development and a
28
better local extension system in the village compared to the other three villages.
Table 27. Estimated parameters for rice yield (ton/ha).
C-D function
Ln (yield)
Model I
Ln (yield) with damage control function
Exponential
Logistc
Model II
Model III
Intercept
1.00
(0.37)***
1.35
(0.21)***
1.38
(0.21)***
Ln(labor)
0.07
(0.03)**
0.07
(0.03)**
0.07
(0.03)**
Ln(Fertilizer)
-0.002
(0.03)
-0.002
(0.03)
-0.003
(0.03)
Ratio of phosphate fertilizer
0.10
(0.15)
0.09
(0.15)
0.10
(0.15)
Ratio of potash fertilizer
0.16
(0.07)***
0.17
(0.07)**
0.17
(0.07)**
0.01
(0.04)
0.02
(0.04)
0.001
(0.05)
Single-season middle rice
0.14
(0.05)***
0.13
(0.05)**
0.12
(0.06)**
Single-season late rice
0.09
(0.04)**
0.10
(0.04)**
0.08
(0.04)*
Higher than middle school
-0.02
(0.04)
-0.02
(0.04)
-0.02
(0.04)
Village dummies:
Zhengbei
0.14
(0.06)**
0.12
(0.05)**
0.14
(0.07)**
Yangzhuang
0.08
(0.05)*
0.07
(0.05)
0.07
(0.05)
Yushanwu
-0.05
(0.04)
-0.04
(0.04)
-0.04
(0.04)
Variables
Dummies:
Double-season late rice
Ln(Predicted pesticide)
0.06
(0.06)
0.03
(0.01)**
c (in exponential model)
d (in logical model)
-1.13
(1.56)
h (in logical model)
0.01
(0.01)
R2
Adjusted R2
0.19
0.14
0.19
0.13
0.19
0.13
Note: the figures in the parentheses are standard errors of estimates. ***, **, * denote significance
at 1%, 5% and 10%, respectively.
Confirm to the findings of early studies on very low elasticity values of both labor
and fertilizer in rice production (Huang et. al, 1994 and 1995; Widawsky et.al., 1998).
Our estimated labor elasticity is about 0.07 and marginal contribution of fertilizer in our
29
sample farms is approaching zero. Low labor elasticity in rice production is consistent
with the observation of large labor surplus in rural China. Farmers in our sampled areas
apply 282 kg/ha fertilizer in effective nutrition base, a level already considering as one
of the highest in the world. Therefore insignificant marginal contribution of fertilizer to
rice production is expected.
On the other hand, Table 27 also shows that a significant improvement in rice yield
could be achieved if fertilizer would be applied more balance in favor of potash
fertilizer. Given the same amount of fertilizer, raising the share of potash fertilizer by
10 percent, rice yield will increase by 1.7 percent. The result also confirms to the early
findings by Huang et al (1994 and 1995).
Among three alternative models, only production function with exponential
damage control specification show a statistically significant impact of pesticide use on
rice production (Table 27). Statistically insignificant impacts of pesticide variables in
C-D model and a model with logistic damage control specification might suggest that
the marginal contribution of pesticide use at current level to rice production is zero, one
possible indicator of over use of pesticide by farmer in rice production. This result
confirms to Widawsky et al. study (1998), the later found the elasticity and marginal
value of pesticide are either zero or insignificant negative.
Because only pesticide parameter in exponential damage control specification
found to be statistically significant from zero, our following discussion will base on the
parameter implied by C in Table 27 (0.03). Based on this parameters and mean values
of rice yield and all right hand side variables, elasticity, average product, and marginal
product of pesticide use in rice production at mean level are computed. On the average,
the farmers apply 287 kg/ha pesticides. Average rice production per kilogram of
pesticide use is 22 kilogram (Table 28). Pesticide does significantly contribute to rice
production through yield loss abatement at the average level. However, at that high
level of pesticide use, rice output elasticity of pesticide use is close to zero. And
marginal product of pesticide declines to 0.02 kilogram for single-season late rice, 1.46
kilogram for double-season early rice. For all rice together, the marginal product of
pesticide is only 0.07 kilogram.
Table 28. Pesticide productivity in rice production based on exponential damage control model.
Average Double-sea Double-sea Single-seaso Single-seas
son early
son late n middle rice on late rice
rice
rice
Pesticide use (yuan/ha)
287
175
289
325
342
Y (ton/ha)
6.4
6.0
6.4
6.5
6.7
Average product (kg rice for every
yuan of pesticide use)
22
34
22
20
20
0.0033
0.0430
0.0030
0.0013
0.0008
Marginal product (kg rice for
additional yuan of pesticide use)
0.07
1.46
0.07
0.03
0.02
Optimal pesticide use (kg/ha)
201
190
200
200
202
Ratio of actual/optimal use of
1.43
0.92
1.45
1.63
1.69
Elasticity
30
pesticides
Figure 1 shows the trend of rice marginal product value with respect to pesticide
cost evaluated at means of all non-pesticide variables. The marginal rice production
values decline significantly with the increases in pesticide uses. The increases in rice
output approach to zero as pesticide use level increase to a level above 200 yuan/ha.
Based on the trend, the optimal pesticide use level is 201 yuan/ha evaluated at the mean
value of rice price (Table 28). This is substantial lower than the actual expenditure of
pesticide use. The pesticide use by the farmers in our samples is 42 percent higher than
the optimal level (287/201=1.43).
Among various season rices, the over use of pesticide is the most severe in the
single-season late rice (69%), then followed by single-season middle rice (63%) and
double-season late rice (45%). An unexpected result is found for double-season early
rice, which shows the current pesticide use is still 8 percent bellow the optimal level.
This might be due to the damage control function that applies the same parameter of C
in the exponential specification to all seasons' rice. On the other hand, the parameter C
that is estimated for the average of 4 different seasons rice might be under valued for
the double-season early rice, and therefore over estimation of the marginal rice
production value of pesticide use. Effort was made to add a dummy variable for
double-season rice in the exponential damage control function. But it was not
successful as the non-linear model fail to converge.
The above analyses show that while pesticides contribute significantly to rice
production through limiting yield losses, the marginal contributions of pesticide use
decline considerably with increase in pesticide use, and approach zero at current
average pesticide use level by the rice farmers. The overuse of pesticide by farmer is
substantial. Among various factors that determine farmers’ decision on pesticide use,
farmer’s own perception on the yield loss due to pest diseases, quality of pesticides,
local agricultural and extension services, particular crop plant protection service, are
the most importance determinants of farmers’ pesticide use. Therefore a significant
reduction on farmer’s pesticide application could be achieved through providing
farmers a better information on the real impact of pest diseases, improvement in the
quality of pesticides and strengthening the local agricultural extension system.
VI. Impact of Pesticide Use on Farmer Health
People poisoned by pesticide ranged from about 53,300 to more than 123,000 each
year in China in the past decade (Huang, et al., 2000). Among poisoned people, average
about half of them relates to pesticide use in crop production 7. China had about, on
average, 10,000 pesticide poisoned death every year, though the number has been
declining significantly since the late 1980s and reduced to less than 4000 in 1996. The
deaths due to improper and over use of pesticide in crop production are about 300-500
persons in normal years (Huang, et al., 2000).
The influence of the pesticides on human health includes not only the acute
diseases but also chronic diseases in both production and non-production related
7
The other half is due to purposive use of pesticides (e.g., suicide) or careless/mistakenly using
pesticide.
31
activities. China classifies pesticides into four categories based on their relative acute
hazard levels. This classification is primarily for acute hazard, no formal classification
is available for chronic hazards. Most of the new chemical pesticides are belong to
category III and IV that are commonly regarded as less threat to human in pesticide
uses. But recently scientists have found that the pesticides under categories III and IV
can cause critical chronic diseases that are not easy to be observed visually. So, the
pesticides under categories III and IV should be paid as much attention as category I
and II pesticides.
Evidence from Rice Farmer
Effects of pesticide use on farmer’s health are examined based on the health
examinations that were conducted for each farmer by a medical team. The team
includes two physicians and two nurses. The examinations include general information
on farmers’ physical and medical situation, acute poisoning symptoms, clinic
biochemistry examination, blood examination, and pathology examination.
Farmer’s health measures derived from the above health examinations related to
pesticide uses include both visible and invisible indicators. The visible indicators are
those that can be directly obtained through interviewing farmers. The invisible
indicators are those that accumulate in the human body and reflect the functions of the
liver, kidneys, neurological and other systems. The visible health impairment measures
are often used by researchers as evidence of effects of pesticide use on farmers’ health,
but the measurement errors could lead to a wide range of conclusions and even
inconsistent results. Farmer’s response to interviewers are highly influenced by his/her
own perception on what should be considered as health impairment or illness even if
the interviewer has given the farmer a clear definition of the impairments. Consistent
results can be found only for those health incident when farmers are seriously poisoned
by pesticide or they have to take time (i.e., a day off) to recovery their health. In this
regard, the visible health impairment data analyzed in this study deal only with those
cases with “seriously poisoned” observations. Therefore, the effects of pesticide use on
farmer’s health using the visible health presented here are underestimated, but they are
consistent among farmers.
Eye effects. While many farmers reported that they feel uncomfortable in their
eyes when they applying pesticides, few of them consider it as severe. Among 100
farmers interviewed, only 3 farmers consider the effect of pesticide use on their eyes are
severe in terms of feeling in a short time. But no one considers the effect needed
medical treatment or taking time off. Further examination on farmer’s pupil shows that
the pupils range from 0.20 cm to 0.30 cm., all are in the normality range. While our
samples fail to find significant eye effect of pesticide, other studies do show the effects,
including eyesight decreasing, eye pain, et al (Li, 1998; Peng, 1998).
Headache. Among 100 farmers interviewed, seven farmers reported that they had
serious headache when they applied pesticide during the last rice growing reason. But
light headache (i.e., feeling somewhat dizzy but not rest required) is often occurred
when farmers apply pesticide for more than 2-3 hours.
Skin effects. Ten percent of the farmers who reported that they had skin pain when
they applied pesticides (Table 29). Most of the farmers felt tickle especially when the
32
pesticide splashing to their skins. Skin as the main door for pesticides entering human’s
body, can be serious harmed by the pesticides.
Table 29. Number of poisoning cases as reported by 100 farmers in 1998.
Symptom
Headache
Nausea
Skin pain
Multiple symptoms
Total
Jiaxing
Anji
7
3
10
20
5
1
2
8
2
2
8
12
Note: The figures in the table are the numbers of farmers responded with the symptoms indicated.
The total samples are 100, with 25 farmers from each village, 2 villages from each county. The
surveys are conducted in Jiaxing and Anji counties, Zhejiang province.
Respiratory tract effects. Cough are often found in the sampled farmers, but there
are only 3 percent of farmers who had nausea or felt decreasing chest expansion when
they applied the pesticides (Table 29). When farmers are applying pesticides, some
pesticides are floating in the air and can enter human body through respiratory tract.
Small samples and difficulties in measuring the farmer’s health effects of pesticide
use lead us to focus our analysis more on laboratory tests on farmers’ health 8. The
health indicators are developed to measure possible effects (particular chronic effect) of
pesticide on human health and are provided in Huang, et al. (2000). They include the
indicators for the blood system, and the functions of liver, kidneys, neurological system
and others. A summary of major indicators is presented in Table 30.
Table 30. A summary result of laboratory tests for 100 farmers in Zhejiang province.
Abnormal cases
Normal
Less than
Greater than
range
Total
lower
upper
boundary
boundary
Blood effects
Hgb
110-160
8
8
0
PLT
100-300
69
69
0
Liver effects
ALT
AST
Kidney effects
BUN
UA
Neurological effects
CHE
0-40
0-40
22
14
0
0
22
14
3-7.2
150-430
23
4
1
1
22
3
30-80
(4500-13000)
5
5
0
8
This study benefits greatly from several meetings in Beijing and Nanjing, where the medical
doctors from hospitals provided very useful comments on pesticide related health problems and
health indicators for both general and laboratory tests.
33
Note: Hgb: Hemoglobin; PLT: platelet; ALT: alanine transaminase; AST: aspartic
transaminase; UA : uric acid; BUN: Urea nitrogen; CHE: choline esterase.
Cardiovascular effects. In acute pesticide poisoning, pesticides cause the diseases
mainly through blood. As soon as the pesticides enter body, it will enter blood system
that is not easy to be observed by simple dialogue check. The blood test results in Table
30 show that a large parts of farmers their platelet and white blood cell present
abnormality. However, further analysis by stratifying results based on pesticide use
does not lead to a significant difference in these statistics among different pesticide use
groups.
Liver function. Pesticides can enter the gastrointestinal tract accidentally through
mouth. Carbamate insecticides formulated in methyl alcohol and ingested may cause
severe gastroenteritic irritation (Morgan, 1977). Organophosphates and copper salts
irritate the gastrointestinal tract, manifesting as intense nausea, vomiting, and diarrhea.
Alanine transaminase (ALT) and aspartic (AST) are two important indicators for the
liver functions. There are many factors effecting ALT and AST, pesticide is one of
them. When a person expose to the pesticide, his/her ALT and AST values will be rise
with the level of pesticide exposure. In 100 farmers examined, we find 22 farmers
whose ALT values are higher than the limited range, and 14 farmers whose AST values
exceed the normal range for common population (Table 30). Stratifying the ALT and
AST by the pesticide use indicates there is a close linkage between the abnormality of
ALT and AST and pesticide application of farmers.
Kidney effects. Kidney helps to ensure the efficient functioning of body cells
through a number of mechanisms: regulation of extracellular fluid volume, control of
electrolytes and acid-base balance; excretion of toxic and waste products, and
conservation of essential substances. Kidney highly exposes to circulation toxins and
long-term exposure to organophosphate compounds could lead to renal tubular
abnormalities. Nephrotoxic agents such as endrin and endosulfan can also cause kidney
abnormalities. Cases about pesticides effecting the kidney function have been reported
by many researchers (Lei, et al, 1998). Among 100 rice farmers in our sample, Table
30 indicates that 23 farmers’ urea nitrogen (BUN) is abnormal, with 22 cases the BUN
values exceed the upper boundary for the normal population.
Neurological effects. Most of the pesticides are neurotoxicants. The neurological
effects of pesticides are important to farmers who exposure to pesticides (Morgan,
1977). Choline esterase (CHE) is a major indicator to monitoring the neurological
effects of pesticide use. When pesticides enter human body, the pesticides will mix with
protein enzyme and affect its function. In this condition, the value of CHE will
decrease. In our samples, there are 6 percent of the farmers have a CHE value less than
the lower limit of the range for a normal population (Table 30).
Health Costs of Pesticide Use in Rice Production
Health costs related to pesticide use are computed based on the treatment required
to restore the farmer’s health. They include the expenses for medication, physical
examination fee, and the opportunity costs of farmers’ time lost in recuperation.
Because the difficult in estimating the cost related to the treatment required to restore
the normality of the health indicators presented in Table 30, the health costs discussed
in this chapter only limit to the treatments related to the visible health impairments.
34
Table 31 summarizes farmer’s health costs stratified by pesticide use levels. The
results clearly indicate a strong linkage of farmer’s health costs and pesticide uses. On
the average, each rice farm applies 14.07 kilogram of pesticide. The health cost related
to this level of pesticide application is 21.68 yuan. However, the health costs vary
largely among farmers. For the farmers with pesticide use level greater than 15
kilograms, their health costs are nearly 3 times as much as that for other farmers with
pesticide use less than 9 kilograms (33.09 yuan vs 8.59 yuan, Table 31). A similar
relationship is also observed for the health costs stratified by the per farm pesticide cost
(Table 31).
Table 31. Pesticide use and health cost in the sample households
Pesticide application
(kg or yuan)
Sample
Mean
Std error
Per farm pesticide use
Low than 9 kg
Between 9-15 kg
High than 15 kg
Per farm pesticide cost
Low than 100 yuan
Between 100-150 yuan
High than 150 yuan
Health cost (yuan)
Mean
Std error
100
14.07
13.46
21.68
55.75
34
33
33
6.07
11.76
24.62
2.27
1.80
19.06
8.59
23.76
33.09
26.60
52.31
76.34
100
144.3
137.9
21.7
55.8
36
35
29
73.8
120.9
260.0
23.6
17.6
212.7
10.9
21.1
35.7
29.4
50.9
80.0
Note: The figures in the table are from 100 households in 4 villages of Jiaxing and Anji counties in
Zhejiang.
The health costs may also link with farmer’s health or physical situations such as
age, gender, the habit of smoking and drinking, various pesticides used. A firm
conclusion on the causal relation between farmer’s health cost and pesticide use
requires the controlling the effects of all above factors on farmer’s health.
Table 32. Pesticide use and statistics on farmers who undertaken health examination.
Mean
Std error
Age (years)
Male share (%)
Smoking history (years)
Smoker share (%)
Drink in pure alcohol (kg/day)
Drinker share (%)
45.4
90
11.1
51
0.03
69
9.10
30
12.1
50
0.03
46
Pesticide use – kg/year:
Category I &II
Category III&IV
Total (I+II+III+IV)
2.47
11.59
14.07
3.41
10.87
13.46
Pesticide use – yuan/year:
Category I &II
Category III&IV
Total (I+II+III+IV)
28.18
116.09
144.27
33.10
113.05
137.94
35
No.of pesticide applications (times/year)
15.93
6.06
Note: The figures in the table are from 100 farmers in 4 villages of Jiaxing and Anji counties in
Zhejiang.
Table 32 shows the basic statistics of farmers and pesticide use by category for 100
farmers examined. Male labor is major operator of pesticide use, accounting for 90
percent. About half of farmers smoke and 69 percent of farmers drink alcohol.
Pesticides used by farmers in rice production are dominated by categories III and IV.
These are the pesticides that are expected to have more effects on chronic diseases.
Among 14.07 kilograms of pesticides used by rice farmers per year, 11.59 kilograms
belong to categories III and IV. This accounted for 82 percent of total pesticides used
by farmer in rice production (Table 32).
The Models of Health Impairments and Costs
In this study, the hypothesized impacts of farmers' pesticide use on their health and
costs related to health impairments are examined using health risk models and health
cost models, respectively. A health risk model can be specified in a general functional
form as:
(5)
Hriski = f (Fi, Pi, Zi, ei )
where, Hrisk denotes health indicator, equal 1 if the health effect occurred or the health
indicator is not in the range of normal cases, 0 otherwise. Indicators for the health
impairments are headache, nausea and skin pains for the visible health impairments and
the abnormalities of laboratory testing results for liver, kidney and neurological
functions for the invisible health impairments. F is a vector of farmers’ characteristics
that are expected to have impacts on the probability of farmers having the health
impairments investigated. These include status of sex, height and weight, the
individual habit on smoking and drinking, et al.. P is pesticide used by farmers. Z is
regional fixed factors such as local quality of drinking water and environments. The i's
denote farmers. Small number of health impairments for each individual illness
disables us to run the model separate for each impairment. Instead, we construct Hrisk
as it equal 1 if any of health impairments presented, and 0 otherwise.
In the empirical estimation, the health impairment model is specified as:
(6) Hriski = a0 + a1*Dmalei + a2 * Ln(Agei) + a3* Ln(Smokei) + a4 * Ln(Drinki)
+ a5 * Weighti/heighti + a6 * Pesticidei + a7 * Dregioni + ei
That is, variables in continuous values such as Age, Smoke (amount/day), and
Drink (kg/day) are specified in natural log form. For those with zero value for Smoke
and Drink, a small value (0.0001) was assigned. Pesticide is specified in several forms
as alternative specifications to the models: total use (in volume or cost), by categories
of pesticides, the frequency of applications (time). We use volume and cost as
alternative specifications since the cost is more reflecting "quality" of the pesticides
used by farmers. This is evidenced in our surveys that show while all farmers face the
same individual pesticide prices (for a cross section data in a given time period), the
averaged prices of all pesticides purchased by farmers differ substantially. Weight is
also specified as two alternatives as weight and weight/height. Dmale is a dummy
36
variable, is 1 for male and 0 for female. Since the dependent variable in this equation is
a discrete variable, ordinary least squares may produce biased and inconsistent
parameters estimates (Maddala, 1983). Therefore a probit model is used to estimate the
parameters.
A health cost is specified similarly as the health impairment model except for the
dependent variable. The estimated health cost model is as:
(7) Ln(Hcost)i = b0 + b1*Dmalei + b2 * Ln(Agei) + b3* Ln(Smokei) + b4* Ln(Drinki)
+ b5 * Weighti/heighti + b6 * Pesticidei + b7 * Dregioni + ei
where, Ln(Hcost) denotes health cost in natural log form. For those with zero cost of
health related to pesticide use, the observation is replaced by a small value (both 1 yuan
and 0.0001 yuan were tried, the results are robust to the alternative specified small
values). Model (7) is estimated by OLS method. Data used in the analyses are from
100-farmer surveys in 2 county of Zhejiang provinces as discussed in the previous
sections and chapters.
Estimated Results Health Impairment and Cost Models
The Results of Health Impairment Model
The estimates of health impairment models for the equation (6) are presented in
Tables 33 and 34, and Appendix Table 2. As Table 33 shown, pesticide use
significantly affects farmer health impairments. The statistically significant and
positive coefficient for category III and IX pesticides indicate that the incidence of
farmer health impairments (headache, nausea and skin illness) rises with the increase of
these pesticide use. The result holds the same for pesticide specifications in either
volume or value terms.
Insignificant coefficient for the categories I and II pesticides might be due to the
small amount of this category pesticide using by farmers. Indeed the pesticide use is
one of two factors that have significant impacts on farmer health impairments. A
significant negative coefficient for age is somewhat interesting and is consistent with
our observations during the field survey. That is, the older the farmer, the more
experience in pesticide use.
A surprising result in Table 33 is none of farmer’s characteristics except for age
having significant effect on farmer health impairments in terms of headache, nausea
and skin illness.
Using BUN, ALT and CHE as indicators for the functions of liver, kidney and
neurological system, respectively, the estimates of multiple abnormality models of
farmer health presented in Table 34 reveal that category III and IV pesticides have
significant impacts on farmer health condition. The results also confirm to previous
findings on the impact of category III and IV pesticides on chronic diseases of liver and
kidney.
The regression results presented in Table 34 also show that male’s incidence of
health abnormality is higher than the female. The significant positive sign for the ratio
37
of weight to height seems contract to our expectation. The positive sign implies that the
heavier the weight, the higher probability of health to be abnormal after control for
height.
Table 33. Estimates of health impairments (headache, nausea and skin illness) of rice farmers
in the sampled villages of Zhejiang province.
Model I
Model II
Intercept
4.97
(6.32)
1.77
(6.09)
1.27
(1.07)
1.37
(1.05)
Ln(Age)
-2.83
(1.45)**
-2.95
(1.44)**
Ln(Smoke)
0.02
(0.05)
0.02
(0.05)
Ln(Drink)
0.04
(0.07)
0.05
(0.07)
Weight/height
7.60
(5.86)
7.80
(5.77)
Respondent's characteristics:
Male dummy
Chemical use – kg/year:
Ln (Categories I&II)
Ln (Categories III&IV)
-0.02
(0.09)
1.21
(0.53)**
Chemical cost – yuan/year:
Ln (Categories I &II)
0.02
(0.07)
Ln (Categories III&IV)
0.95
(0.50)**
Ln (Number of applications)
-0.86
(1.16)
-0.15
(1.06)
County dummy:
Jiaxing
-0.63
(0.79)
-0.74
(0.80)
Chi2(9)
16.68
14.69
Note: the figures in the parentheses are standard errors of estimates. ***, **, and * denote
significance at 1%, 5% and 10%, respectively.
More observations on abnormality of BUN enable us to run a separate model on
the impact of pesticide use on kidney function. The results presented in Appendix
Table 2 confirm to our expectation that the category III and IV pesticides have
significant impact on farmer kidney function.
The Results of Health Cost Model
The estimated parameters for farmer health cost models are presented in Tables 35
and 36. While the coefficients for pesticide use and location fixed effect factors are
38
significant and consistent with our expectations, none of respondent’s characteristics
have statistically significant effects on the farmer’s health cost. This is unexpected as
significant effects of some of these variables on rice farmer’s health costs are found in
similar studies in the Philippines and Vietnam (Rola and Pingali, 1993; Dung and Dung,
1999).
Table 34. Estimates of multiple abnormalities (BUN, ALT and CHE) of rice farmers in the
sampled villages of Zhejiang province.
Multiple abnormality of rice farmers
Model I
Model II
Intercept
6.35
5.39
(4.87)
(4.82)
Respondent's characteristics:
Male dummy
-0.69
-0.68
(0.26)***
(0.26)***
Ln (Age)
-0.20
(0.27)
-0.19
(0.27)
Ln (Smoke)
0.003
(0.009)
0.003
(0.010)
Ln (Drink)
-0.02
(0.01)*
-0.02
(0.01)*
Ln (Weight)
-1.89
(1.49)
-1.76
(1.48)
Weight/height
7.94
(4.64)*
7.48
(4.61)*
Ln(Chemical use – kg/year)
Ln (Categories I&II)
-0.003
(0.02)
Ln (Categories III&IV)
0.15
(0.09)*
Ln(Chemical cost – yuan/year)
Ln (Categories I&II)
-0.002
(0.016)
Ln (Categories III&IV)
0.17
(0.10)*
Ln (Number of applications)
-0.16
(0.23)
-0.12
(0.22)
Anji county dummy
0.08
(0.15)
0.06
(0.15)
Chi2(10)
16.72
16.98
Note: the figures in the parentheses are standard errors of estimates. ***, **, and * denote
significance at 1%, 5% and 10%, respectively.
Explanation to the insignificant coefficients of the farmer’s specific characteristics
in our model might be due to different coverage of the health costs. In our cases, only
the health costs to those health impairments that are severe and require time to recover
or time to take rest are included. Although more lightly health impairments related to
headache, eye, skin and others are also recorded, there is a strong rejection from our
39
medical team to include these observations and to estimate the “explicit” health cost for
the following two reasons. One is inconsistent measurement error among the
respondents even the interviews were conducted by the medical team. The other is
difficult to estimate the health costs for the small health impairments that only lasted a
very short period.
Table 35. Estimates of health cost of rice farmers in the sampled villages of Zhejiang
province (based on pesticide use volume).
Ln (health cost)
Model I
Model II
Model III
Model IV
Intercept
4.02
(4.13)
2.92
(4.11)
4.48
(4.03)
3.19
(4.10)
0.94
(0.68)
1.01
(0.68)
0.87
(0.68)
0.98
(0.69)
Ln (Age)
-1.13
(0.90)
-0.85
(0.91)
-1.28
(0.89)
-1.12
(0.91)
Ln (Smoke)
0.02
(0.03)
0.02
(0.03)
0.02
(0.03)
0.02
(0.03)
Ln (Drink)
0.03
(0.04)
0.03
(0.04)
0.03
(0.04)
0.04
(0.04)
Weight/height
5.59
(3.64)
5.23
(3.57)
5.14
(3.71)
5.28
(3.71)
Respondent's characteristics:
Male dummy
Ln (Chemical use: kg/year):
Ln (Categories I&II)
Ln (Categories III&IV)
0.06
(0.06)
0.04
(0.06)
0.54
(0.30)*
0.87
(0.29)***
Ln (I+II+III+IV)
Ln (Number of applications)
0.03
(0.01)*
-0.72
(0.74)
-0.34
(0.66)
-1.17
(0.57)**
-1.27
(0.55)**
Yangzhuang
-0.91
(0.59)
-1.12
(0.57)*
Yushanwu
-1.34
(0.54)**
-1.57
(0.51)***
0.24
0.15
0.23
0.15
Village dummies:
Zhengbei
R2
Adjusted-R2
0.03
(0.01)**
-0.28
(0.65)
-0.46
(0.51)
0.18
0.10
0.13
0.06
Note: the figures in the parentheses are standard errors of estimates. ***, **, and * denote
significance at 1%, 5% and 10%, respectively.
On the other hand, the significant impact of pesticide use on farmer’s health cost is
consistent with previous studies in other countries. The results are invariable to the
specifications of pesticide use in volume or value. Holding the total amount of
40
pesticide use, one of model (Model III in Table 36) also show that more frequency of
pesticide use (and therefore smaller amount of pesticide applied by farmer each time),
the less the health costs.
Table 36. Estimates of health cost of rice farmers in the sampled villages of Zhejiang province
(based on pesticide cost).
Ln (health cost)
Model I
Model II
Model III
Model IV
Intercept
1.86
(4.10)
2.88
(4.12)
1.48
(4.08)
3.13
(4.12)
1.02
(0.67)
0.99
(0.68)
0.96
(0.69)
0.94
(0.70)
Ln (Age)
-1.07
(0.90)
-0.85
(0.91)
-1.34
(0.90)
-1.16
(0.92)
Ln (Smoke)
0.02
(0.03)
0.02
(0.03)
0.02
(0.03)
0.02
(0.03)
Ln (Drink)
0.04
(0.04)
0.04
(0.04)
0.05
(0.04)
0.04
(0.04)
Weight/height
5.39
(3.65)
5.05
(3.57)
5.04
(3.76)
5.09
(3.73)
Respondent's characteristics:
Male dummy
Chemical cost – yuan:
Ln (Categories I &II)
Ln (Categories III&IV)
0.07
(0.05)
0.06
(0.05)
0.65
(0.31)**
0.85
(0.31)***
Ln (I+II+III+IV)
0.002
(0.001)*
Ln(Number of applications)
-0.71
(0.69)
-0.28
(0.65)
Village dummies:
Zhengbei
-1.12
(0.56)**
-1.27
(0.56)**
Yangzhuang
-1.00
(0.57)*
-1.14
(0.57)**
Yushanwu
-1.54
(0.50)***
-1.64
(0.51)***
0.26
0.17
0.23
0.14
R2
Adjusted- R2
0.003
(0.001)*
-1.18
(0.59)**
-0.30
(0.49)
0.17
0.10
0.12
0.05
Note: the figures in the parentheses are standard errors of estimates. ***, **, and * denote
significance at 1%, 5% and 10%, respectively.
Based on the parameters presented in Tables 35 and 36, we compute farmer’s
health costs of pesticide use at the average and marginal levels. In computing these
statistics, the sample mean values for all variables are used. The results are summarized
in Table 37.
At current level of pesticide use, Table 37 shows that the average health cost per
yuan of category III&IV pesticides use is 0.19 yuan (the 1st column) or 0.17 yuan (the
41
3rd column). And marginal health cost for additional yuan of category III and IV
pesticide use will cause 0.12 yuan (the 1st column) or 0.09 yuan (the 3rd column).
Table 37. Health cost of pesticide use in rice production.
Model I in
Table 36
(pesticide use
in yuan)
Model I in Table 35
Original
Conversed
number in
kg to
model is kg
yuan
Pesticide I+II
-- Pesticide use level
28.2
2.5
26.9
-- Average health cost per pesticide use
(yuan/yuan or kg)
0.77
8.76
0.80
-- Marginal health cost (yuan / additional
unit of pesticide use)
0.05
0.53
0.05
-- Elasticity
0.07
0.06
-- Pesticide use level
116.1
11.6
126.3
-- Average health cost per pesticide use
(yuan/yuan or kg)
0.19
1.87
0.17
-- Marginal health cost (yuan / additional
unit of pesticide use)
0.12
1.01
0.09
-- Elasticity
0.65
0.54
Pesticide I+II
Note: The sample mean values for health cost per farmer is 21.7 yuan and pesticide price is 10.89
yuan/kg. The figures for pesticide I+II are based on statistically insignificant coefficient of
pesticide (I+II) use in the health model, therefore they presented here just for reference.
VII. Conclusions and Policy Implications
Intensive cultivation and broad adoption of fertilizer responsive varieties has been
widespread pest infestation. The extent of pest-related diseases grew several times in
past 2 decades in China. Rising pest problems and availability of pesticide with market
development then further induce the increasing use of pesticides in crop pest
management. China is likely to be the largest pesticide consumption country in the
world recently. Per hectare pesticide use more than tripled in grain production within
20 years. Among grain productions, rice is the most intensive pesticide use crop.
Pesticides are used as the primary method of pest control in rice farming.
Given the prospectus of China's food situation in the coming century and the
central goal of China's food security policy, intensive cropping system with increasing
use of modern input will likely continue to be the dominant farming practice in China.
An increasing use of farm pesticides is also expected to be continued if no practical
alternative pest management technologies, regulations and policies are developed to
effectively reduce the overuse of pesticides in crop production.
42
The productivity analyses of pesticide use in rice farming based on our primary
farm level data indicate the average return to pesticide use is high. Rice yield loss due to
pest-related diseases could reach as high as 40 percent if no pest control is adopted. On
the other hand, our study show that while pesticides contribute significantly to rice
production through limiting yield losses, the marginal contributions of pesticide use
decline considerably with increase use of pesticide and approach zero at current
average pesticide use level by the rice farmers. Given the current rice and pesticide
prices, the average overuse of pesticides by farmers is more than 40 percent, with the
highest value of about 70 percent for the single-season late rice.
In the battle against pests, pesticide is a double-edged sword as it also affects
human health and contaminates environment. This study shows that both visible acute
health impairments and indivisible chronic health functions (liver, kidney and
neurological system) of rice farmers closely link with the extent of their exposure to
pesticides. Although the health costs we examined in this study only limit to the
treatments related to a few visible acute health impairments (could be just a small part
of the total health costs), they still account for about 15 percent of pesticide costs. The
estimates of multiple abnormality (liver, kidney and neurological system) models of
farmer health reveal that pesticides have significant impacts on farmer health condition
and can cause chronic diseases of liver and kidney. If we include the costs related to
these chronic disease and other costs that are not computed in this study such as lightly
acute disease effect (does not require medical treatments or could be recovery shortly)
and other external costs (consumer health and environmental costs), the total health
costs even may be greater than the private cost paid by farmer for purchasing pesticide.
On the other word, while the overuse of pesticide by farm estimated in this study using
private cost reaches 40%, the optimal use of pesticide in rice production could be less
than the half of the current level of pesticide used by farmers if the external costs were
accounted.
Although overuse of pesticide in pest management has been well documented in
the literature, a little knowledge is on the determinants of pesticide overuse. This study
shows that, among various factors that determine farmers’ decision on pesticide use,
farmer’s own perception of the yield loss due to pest problems, quality of pesticides
available in the market, local agricultural and extension services (particular crop plant
protection service), and opportunity cost of farm labor are the most important
determinants of farmers’ pesticide use. Therefore, in seeking for a better solution for
pest management and externality issue, the priority issues are not just how to set up
regulations and policies to ban all pesticide use in crop production, but how to use
pesticide properly, avoid its overuse, and improve the conditions for farmers to
internalizing its externality problems (external cost) of pesticide use, and find better
alternatives of pest management, etc.
To convince farmers on the fact that their perceptions of crop yield loss due to pest
related diseases are over-estimated, improving farmers' knowledge on pest
management and pesticide safety issues is critical. Our analyses show that farmers’
knowledge on pest management as well as the consequences of pesticide uses is very
minimal. Because of their limit knowledge on the pest management and high risk
aversion of farmers (all are small farm with average farm size of 0.5 hectare) in pest
control, farmers' over estimation of crop yield losses due to pest problems and very
43
inelasticity of the farmers' pesticide use response to the pesticide price are inevitable.
Our results show that the averaged farmer's perception on crop yield losses due to
pest related diseases is nearly twice as much as actual yield loss that might occurred
without any pest control. Therefore most farmers do not believe the recommendation
and prescription labeled in pesticide products, which are commonly regarded by farmer
as too low level of pesticide uses.
The limit knowledge on pest management, much high perception of crop yield
losses, and very inelasticity of the pesticide price response indicate that the effort to
convincing farmers to reduce the extent of pesticide use will mainly rely on the factors
that could significantly reduce farmers' risk aversion. These might include pest
management related information, education, training, extension services, etc.
Increasing farmer’s awareness of pesticide hazardous to the environment and human
health should be included in the local extension activities in order to partially
internalize the external costs of pesticide uses. The results presented in this study also
questions on the services that currently provided by Agricultural Extension System, a
public extension system that has been participating heavily in pesticide marketing
activities as a results of financial stress.
The price response is surprising low and also much lower than that found in a
similar study in Southeast Asian countries. Very low price responsiveness of pesticide
use by farmer implies that the effectively incentive policy (tax and price) should focus
on the pesticide production or factory level. As availability of better quality pesticide is
the other important determinant of pesticide use, the policies to encourage the
development of new and improved pesticides have to be made both at
production/factory and marketing levels.
The efforts in regulating pesticide production, marketing and applications have
been made by the government since 1970s. Considerable progress was made in
introducing less persistent compounds as a substitute for highly hazardous pesticides
and the safety use and management of pesticides in the past. However as experience
has shown, the simple promulgation of rules and regulation are not enough to guarantee
improvements in the quality of pesticide products in the markets and the proper and
safe use of pesticides. Methods of farmer's handling and application practices of
pesticides have not been changed. Many highly hazardous pesticides are continuously
used by farmers in pest control. More conductive and concrete policies, regulations and
enforcement system are required.
Identifying alternatives to current agricultural chemical pesticide practices as a
way to reduce pollution are also of critical importance in China. It has been shown that
host-plant resistance is a productive but under utilized input (Widawsky, et al., 1998).
As a substitute for pesticides, improvements in host-plant resistance could lead to
substantial savings in pesticides without reducing crop yield. However, our field
survey reveals that few farmers have taken into account available host-plant resistance
in making pesticide use decision (mainly due to lack of knowledge on host-plant
resistance). While the concept of integrated pest management (IPM) has gained strong
support among environmental groups, extending IPM technology is facing greater
challenge as the opportunity cost of farm labor rise with non-farm sector development
in rural China.
44
45
Reference
Barker, R. and R.W. Herdt with B. Rose. 1985. The Rice Economy of Asia. Washington,
D.C., and Los Banos, Laguna, Philippines, Resources for the Future with IRRI.
Carson, G. A, Mueller R, 1987. Farmers’ Perceptions, Education and Adoption of Pest
Management Strategies for Small Farmers. Proceedings of the 11th International
Congress of Plant Protection. Vol.1. Manila. ICPP, pp. 147-154.
Cramer, H.H., 1967. Plant Protection and World Crop Production. Pflanzenschutz.
Nachrichten Bayer. 20:1-524.
Dung, N. H. and T. T. T. Dung. 1999. "Economic and Health Consequences of Pesticide
Use in Paddy Production in the Mekong Delta, Vietnam," EEPSEA Research
Report Series, IDRC, Singapore.
Fan, S. 1991. "Effects of Technological Change and Institutional Reform on Production
Growth in Chinese Agriculture." American Journal of Agricultural Economics
(73):266-275.
Fan, S. and M.C.A. Agcaoili, “Why Do Projections on China’s Food Supply and
Demand Differ?” EPTD’s Discussion Paper No. 22. International Food Policy
Research Institute, Washington, DC, 1997.
Headley, J.C. Estimating the Productivity of agricultural pesticides Amer. J. Agri.
Econ. 50(1968):13-23
Heong, K.L., M.M. Escalada, and A.A. Lazaro. 1995. Misuse of Pesticides among Rice
Farmers in Philippines. In P.L. Pingali and P.A. Roger (eds.), Impact of Pesticides
on Farmer Health and the Rice Environment. Norwell, Massachusetts and Los
Banos, Laguna, Philippines: IRRI.
Huang, J.K, F.B. Qiao, L.X. Zhang and S. Rozelle. 2000. Farm Pesticide, Rice
Production, and Human Health. CCAP's Project Report 11. Center for Chinese
Agricultural Policy (CCAP), Chinese Academy of Agricultural Sciences. And A
Project Report Submitted to EEPSEA, Singapore.
Huang, J.K., R.F. Hu, J. Song, and S. Rozelle. 1999a. “Agricultural Technology from
Innovation to Adoption: Behavior Analyses of Decision Maker, Scientist,
Extension Worker, and Farmer,” Impact of Science on Society, No.1 (1999):55-60.
Huang, J.K., Q.J. Wang and Q.G. Chen. 1995. "The Efficiency of Chemical Fertilizer
Use in Rice Production in China," Chinese Journal of Rice Science, 9(1995):39-44.
Huang, J.K., Q.G. Chen and Q.J. Wang, 1994."Rice Productivity and Fertilizer Balance
Use in China's Major Rice Regions," Journal of Agrotechnical Economics, No.5
(1994):36-41.
Huang, J.K. and S. Rozelle. 1996. "Technological Change: Rediscovery of the Engine
of Productivity Growth in China's Rural Economy," Journal of Development
Economics 49 (July):337-69.
46
Huang, J.K., S. Rozelle, and M. Rosegrant. 1999b. “China’s Food Economy to the 21st
Century: Supply, Demand, and Trade,” Economic Development and Cultural
Change. Vol.47(1999July):737-766.
Lei, H., L. Dai, C. Liu, L. Gu, W. Xu, Y. Ding, 1998, The relationship between poison
and flesh blood protein, Chinese Journal of Preventive Medicine. Volume 32, No. 3,
May, 1998.
Li, Y. 1998, The effects of OP on eyes, Journal of Occupational Health and Damage,
Vol. 13, No.4, 1998
Lichtenberg, E., and D. Zilberman. 1986. The econometrics of damage control: Why
specification matters. American Journal of Agricultural Economics 68: 261-273
Lin, J. Y. "Rural Reforms and Agricultural Growth in China." American Economic
Review, 82(1992):34-51.
Lin, J.Y. 1991 "The household Responsibility System and the Adoption of Hybrid Rice
in China, Journal of Development Economics, 36(1991):353-373.
Maddala, A., 1983. Limited and Qualitative variables in econometrics, Cambridge
University Press, Cambridge.
Ministry of Agriculture (MOA), National Plant Protection Statistical Materials,
National Agricultural Technology Extension Center.
Ministry of Agriculture (MOA) “Research about the Influences of Pesticides on the
Eco-environment,” the project coordinating group, PRC, 1983.
Ministry of Agriculture (MOA), 1998. Agricultural Development Report. China’s
Agricultural Press. Beijing.
Mumford, J. D, 1981. Pest Control Decision-Making: Sugar Beet in England. Journal
of Agricultural Economics. 32(1):31-41
Mumford, J. D, 1982. Perceptions and Losses from Pests of Arable Crops by Some
Farmers in England New Zealand. Crop Prot. 1(3):283-288
Norton, G. A, and Mumford, J D, 1983. Decision-making in pest control. In T. H.
Cooker, (ed.), Applied Biology, Vol.8. New York, N.Y.: Academic Press.
Oerke, E.C., H. Dehne, F. Schohnbeck, and A. Weber. 1995. Crop Production and Crop
Protection: Estimated Losses in Major Food and Cash Crops: Amsterdam: Elsevier.
Pathak, P. K. and G.S. Dhaliwal. 1981. Trend and Strategies for Rice Pest Problems in
Tropical Asia. IRRI Res. Pap. Ser. 64.
Peng, Y. 1998, Eight Poisoning Cases Made by Chlorothalonil, Occupational Medicine,
Vol. 25, No. 1, 1998.
Pimentel, D. and H. Lehman. 1992. The Pestcide Question: Environment, Economics,
and Ethics, New York: Chapman and Hall.
47
Pingali, P.L. and G.A. Carlson.1985. “Human Capital, Adjustment in Subjective
Probabilities, and Demand for Pest Control”. Am. J. Agric. Econ. 67(4): 853-861.
Pingali, P. L., C. B. Marquez, and F.G.Palis. 1994. Pesticides and Philippine Rice
Farmer Health: A Medical and Economic Analysis. American Journal of
Agricultural Economics 76(August):587-592.
Pingali, P. L. and P.A. Roger. 1995. Impact of Pesticides on Farmer Health and the Rice
Environment. Massachusetts: Kluwer Academic Publishers.
Rola, A. C. and P. L. Pingali. 1993. Pesticides, Rice Productivity, and Farmers' Health:
An Economic Assessment. Los Banos, Philippines, and Washington D.C.:
International Rice Research Institute and World Resource Institute.
Rola A C, Chupungco A R, Adalla C B, Hoque M M, Stuart T H, Sumayao B R (1988).
Results of the benchmark survey. Integrated Pest Management Extension and
Women Project. Los Banos, Laguna, Philippines.
Rola, A. C. and D. A. Widawsky.1998. "Pests, pesticides, and integrated pest
management in rice, In Pingali P. and M. Hossain. Impact of rice research.
Proceedings of the International Conference on the Impact of Rice Research, 3-5
June 1996, Bangkok, Thailand. TDRI/IRRI. pp.135-158.
SSB [State Statistical Bureau], Statistical Yearbook of China, various issues.
Tait, E. J, 1977. A Method for Comparing Pesticide Usage Patterns Between Farmers.
Ann. Appl. Biol. 86(3):229-240.
Waibel, H, 1990. Requirements for An Economic Interpretation of Crop Losses. In
IRRI, Crop Loss Assessment in Rice.
Warburton, H., F.G. Palis, and P.L. Pingali.1995. Farmer Perceptions, Knowledge, and
Pesticide Use Practices. In P.L. Pingali and P.A. Roger (eds.), Impact of Pesticides
on Farmer Health and the Rice Environment. Norwell, Massachusetts, and Los
Banos, Philippines: Kluwer Academic Publishers and IRRI.
Widawsky, D., S. Rozelle, S. Jin and J.K. Huang. 1998. "Pesticide Productivity,
Host-Plant Resistance and Productivity in China" Agricultural Economics,
19(1998):203-217.
Ye, Q., 1991. The Price of Fertilizer and the Behavior of Farmers’ Fertilizer Using in
Jiangsu Province, Unpublished Ph.D. Dissertation, Food Research Institute,
Stanford University.
Yudelman, M. 1998. Water and Food in Developing Countries in the Next Century.
Feeding a World Population of More Than Eight Billion People: A Challenge to
Science. Oxford University Press. Oxford, U.K.
Yudelman, M., A.Ratta and D.Nygaard.1998. Pest Management and Food Production:
Looking to the Future. Food, Agriculture and the Environment Discussion Paper 25,
International Food Policy Research Institute, Washington D.C..
ZGNYNJ, Zhongguo Nongye Nianjian [China Agricultural Yearbook]. Beijing, China:
48
China Statistical Press, 1980-98.
Figure 1. Marginal product value of pesticide use in
rice production
Margional Product Value
(yuan)
70
60
50
40
30
20
10
0
47
97
147
Pesticide Cost (yuan/ha)
49
197
Appendix Table 1. Estimated parameters for frequency of farmers' pesticide application per rice
growing season in Zhejiang province.
Frequency
Frequency
Ln (times)
(times)
(times)
Variables
Model I
Model II
Model III
Intercept
7.60
(1.31)***
8.90
(4.26)**
1.89
(0.60)***
Farmer's perception on yield loss
2.32
(0.77)***
2.32
(0.79)***
0.35
(0.11)***
Average mixed price of pesticides
(AMPP)
Ln (AMPP)
-0.02
(0.03)
0.09
(0.44)
0.01
(0.06)
Information from ATES
0.08
(0.33)
0.12
(0.34)
0.02
(0.05)
0.67
(0.27)**
0.08
(0.04)**
-0.21
(0.56)
-0.02
(0.04)
-0.13
(0.80)
0.003
(0.11)
2.85
(0.41)***
2.98
(0.42)***
0.43
(0.06)***
Single-season middle rice
0.65
(0.68)
0.86
(0.69)
0.12
(0.10)
Single-season late rice
1.19
(0.41)***
1.08
(0.43)**
0.14
(0.06)**
Higher than middle school
-0.07
(0.52)
-0.08
(0.54)
-0.02
(0.08)
-0.25
(0.46)
-0.29
(0.48)
-0.004 (0.07)
-3.26
(0.64)***
-3.18
(0.66)***
-0.40
(0.09)***
Yangzhuang
-3.66
(0.63)***
-3.52
(0.65)***
-0.49
(0.09)***
Yushanwu
-1.05
(0.57)*
-0.92
(0.62)
-0.11
(0.09)
0.49
0.45
0.46
0.42
0.45
0.40
Ln (Rice area)
Rice area
Income/person
0.18
(0.04)***
0.00005
(0.00006)
Ln (Income/person)
Age
-0.0005
(0.0179)
Ln (Age)
Dummies:
Double-season later rice
Male
Village dummies:
Zhengbei
R2
Adjusted R2
Note: the figures in the parentheses are standard errors of estimates. ***, **, and * denote
significance at 1%, 5% and 10%, respectively.
50
Appendix Table 2. Estimate of BUN abnormalities of rice farmers in the sampled villages
of Zhejiang province.
BUN abnormality of rice farmers
Model I
Model II
Model III
Model IV
Intercept
9.22
8.58
9.29
8.49
(3.95)**
(3.89)**
(4.00)**
(3.94)**
Respondent's characteristics:
Male dummy
-0.39
-0.38
-0.36
-0.36
(0.19)**
(0.18)**
(0.19)*
(0.19)*
Ln(Age)
-0.04
(0.21)
-0.03
(0.21)
-0.07
(0.22)
-0.02
(0.21)
Ln(Smoke)
-0.002
(0.007)
-0.002
(0.007)
-0.003
(0.007)
-0.003
(0.007)
Ln(Drink)
-0.02
(0.01)*
-0.01
(0.01)
-0.01
(0.10)
-0.01
(0.10)
Ln(Weight)
-2.96
(1.21)**
-2.90
(1.21)**
-3.07
(1.23)**
-2.84
(1.21)**
Weight/height
10.28
9.98
(3.69)***
10.48
(3.72)***
9.91
(3.66)***
(3.69)***
Ln(chemical use – kg)
Ln (Categories I&II)
Ln (Categories III&IV)
0.01
(0.02)
0.11
(0.07)
Ln (I+II+III+IV))
0.003
(0.27)*
0.48
(0.27)*
Volume share of I&II (%)
Ln(Chemical cost – yuan/year)
Ln (Categories I&II)
0.01
(0.01)
Ln (Categories III&IV)
0.18
(0.08)**
Ln (I+II+III+IV))
0.0004
(0.0003)
0.32
(0.29)
Cost share of I & II (%)
Ln (Number of applications)
-0.29
(0.19)
-0.33
(0.18)*
-0.11
(0.16)
-0.15
(0.16)
Anji county dummy
0.02
(0.12)
-0.01
(0.12)
0.05
(0.12)
0.04
(0.12)
Chi2(10)
18.63
21.93
19.29
17.81
Note: the figures in the parentheses are standard errors of estimates. ***, **, and * denote significance
at 1%, 5% and 10%, respectively.
51