Vous êtes sur la page 1sur 8

WFL Publisher

Science and Technology

Meri-Rastilantie 3 B, FI-00980 Journal of Food, Agriculture & Environment Vol.8 (2): 1185-1192. 2010 www.world-food.net
Helsinki, Finland
e-mail: info@world-food.net

Assessing total factor productivity and efficiency change for farms participating in
Grain for Green program in China: A case study from Ansai, Loess Plateau
Li Li 1*, Atsushi Tsunekawa 1, Mitsuru Tsubo 1, Atsushi Koike 2 and Jijun Wang 3
1
Arid Land Research Center, Tottori University, 1390 Hamasaka, Tottori 680-0001, Japan. 2 Faculty of Engineering, Tottori
University, 4-101 Koyama-Minami, Tottori 680-8550, Japan. 3 Institute of Soil and Water Conservation, Chinese Academy of
Science, No. 26, Xinong Road, Yangling, Shaanxi 712100, China. *e-mail: lili@alrc.tottori-u.ac.jp,
tsunekawa@alrc.tottori-u.ac.jp, tsubo@alrc.tottori-u.ac.jp, koike@sse.tottori-u.ac.jp, jjwang@ms.iswc.ac.cn

Received 18 January 2010, accepted 3 April 2010.

Abstract
In 1999, the Grain for Green program (also known as Sloping Land Conversion program), one of the world’s largest land-conservation programs, was
launched in China. It emphasizes both eco-environment rehabilitation and poverty alleviation, with the ultimate goal of sustainable development.
Total Factor Productivity (TFP) has important implications for conservation of the environment, enhancement of wellbeing, food self-sufficiency,
and the sustainability of the program. Using data from the year before and the ending year of the first phase, we did a case study in the Zhifanggou
watershed on China’s Loess Plateau. We used a Data-Envelopment-Analysis based on Malmquist TFP index approach to find out the changes of
TFP, the sources of TFP growth, and the determinant factors at a farm level. We found that 1) TFP has been greatly improved after the implementation
of the program and 2) the sole source of TFP growth is technological growth. In contrast, the technical efficiency of farms under the improved
technology has decreased; 3) farms with an unfavorable initial state benefited more after the introduction of the program; and 4) land terracing and
access to credit contributed significantly to TFP growth and technological growth. Land terracing is positively related to technical efficiency change,
too. Extension services are positively related to technological growth, while age is negatively related to technical efficiency change. We therefore
suggest a focus on terracing slopes, improving access to credit and extension services.

Key words: Desertification, dryland, sustainable agriculture, Grain for Green program, total factor productivity, Malmquist TFP index, data envelopment
analysis.

Introduction
Widespread land degradation and extreme poverty continued to 1999. A program called Grain for Green, also known as the Sloping
be the most challenging problems threatening the sustainable Land Conversion program, was launched in Shaanxi, Sichuan and
development of rural China until the 1990s, despite decades of Gansu provinces, to prevent unsustainable agricultural practices.
efforts by the Chinese government and local communities. The The program stresses both eco-environmental rehabilitation and
heavy, if not sole reliance of the huge rural population on limited poverty alleviation, with the ultimate goal of sustainable
natural resources that surpassed the carrying capacity of the agricultural development. The key measure of the program is to
ecosystem was the principal cause of resource depletion and compensate those volunteers who set aside steep cropland or
impoverishment 1. With the population explosion and fast low-yield marginal land for forest and grassland, or who retain
economic development, last half century saw an accelerating trend and extend vegetation, with grain, cash or tree seedlings. Other
toward land reclamation, overgrazing and deforestation on marginal measures include terracing for soil and water conservation,
land and sloping land, exacerbating land degradation. Lin and livelihood diversification to absorb surplus labor, rural energy
Ho 2 estimated that over 70 million hectares of barren land, pasture construction, environmental resettlement, etc. (State Council of
or forest was reclaimed up to 1996, mainly in low MCI (Multiple the P.R.C.). As the program gained acceptance in the pilot areas,
Cropping Index) provinces with fragile environments in the western the government expanded it to 17 provinces and districts in 2000,
interior and northern frontiers. On the Loess Plateau and the South- and to 25 provinces in 2002, totaling 1897 counties, more than 20
west Plateau, where erosion rate is the highest, degraded land thousand towns and townships, more than 100 thousand villages,
accounted for 71.30% and 52.53%, respectively, of cultivated 73.91% of the national land, over 15 million rural households and
areas 3. The serious land degradation not only greatly undermines more than 60 million farmers. The government declared their
land productivity and sticks farmers in poverty trap, but also has intention to convert 34 million mu (1mu = 0.067ha) of marginal
substantial off-site impact, such as increased incidence and cropland on steep hillsides and slopes, together with 39.93 million
severity of flooding and droughts, as a result of sediment deposition. mu of rotational or waste land and mountains, to grassland and
Severe drought in 1997 and a devastating flood in 1998, which forests during the next 10 years (2002-2011). The total investment
claimed thousands of lives and caused billions of yuan’s worth of in the program will be 430 billion yuan 4.
damage, finally galvanized the Chinese government into action in Given the large operating scale, huge public investment and

Journal of Food, Agriculture & Environment, Vol.8 (2), April 2010 1185
profound environment and well-being implications, broad attention technical efficiency change and scale efficiency change, to draw
had been drawn from academic circles domestically and abroad. inferences about the primary drivers for TFP growth, and then we
Unsurprisingly, many studies have focused on the effectiveness compared the relative technical efficiency before and after the
or performance of the program in environmental rehabilitation, first phase of the program. Finally, we used ordinary least square
economic development and poverty alleviation 5-9. Other strands (OLS) regression analysis to investigate the factors, mainly
of literatures refer to the prospects and sustainability of the including the measures taken in the program, affecting TFP growth,
program 10-12. One concern is that the loss of cultivated land area technological growth and technical efficiency change.
might lead to a decrease in agricultural production that will threaten
food security and social stability and thus condemn any follow- Malmquist TFP index: Using the distance functions introduced
up investment and irrevocably stymie the program. Other concerns by Malmquist, Caves et al. 16 proposed a framework for input,
involve who will pay for the ecosystem services to sustain the output and productivity measurement that does not proceed from
cultivated land conversion practices and whether farmers would a continuous time representation, which they named as Malmquist
refrain from reclamation if the subsidies were to be ceased. The Index.
results of the above analysis, however, turned out to be Let x = (x1, …, xn)∈ RN+ denote an input vector and y = (y1, …,
controversial. Added to the fact that most of the previous studies yn)∈ RN+ an output vector, we can get a production possibility
were based on data collected before 2004, the motivation of this set at time s by
study is to try to contribute to literatures with a case study using
the latest data covering the first phase of the program, from a Ps(x) = {(x, y): x can produce y} (1)
productivity or efficiency perspective. The parameters have
practical importance, as new compensation standards and Assuming constant return to scale (CRS), the output distance
operation mechanisms were to be introduced when the first phase function with technology at time s, the initial period, can be defined
ended in 2007(State Council of the P. R. C.). as
Measured as the ratio of aggregate output to aggregate input, y
Total Factor Productivity (TFP), has significant environmental d s ( x, y ) inf{T : ( x, )  P s ( x )} (2)
T
and economic implications in context of land retirement. Increases
in TFP contribute to total agricultural production that would offset This function completely characterizes the technology at s, in the
or even outweigh the production loss resulting from the loss of direction that frontier (or maximum possible) output can be
arable land. This not only benefits food security and poverty produced given a certain amount of inputs. Here, θ is a scalar, and
alleviation and encourages surplus labor to shift to other its value is the efficiency score for each production unit. It satisfies
occupations, but also demotivates farmers to reclaim land later. In 0 < θ ≤ 1, where a value of 1 indicating a point on the frontier of
addition, changes in TFP can be differentiated into technological technology and hence a fully technically efficient production
growth and changes in technical efficiency. Technological progress activity.
in agriculture, as pointed out by Bennett 13, together with changes Similarly, we can define a distance function in relation to
in relative prices, could all be moving China quickly to a point technology at time t, the final period, as
where a net transition from cropland to forestland will naturally y
begin to take place. Technical efficiency, defined by Farrell 14 as d t ( x, y ) inf{T : ( x, )  P t ( x )} (3)
T
the ratio of actual to potential production, is also of great
importance to resource utilization, land management, and Thus, the output-oriented Malmquist TFP index, which measures
household wellbeing 15. We therefore employed a Data- the TFP growth between time s and t, can be defined as
Envelopment-Analysis based Malmquist TFP index approach to
Ms= d s (xt , y t ) (4)
calculate TFP growth induced by the program and its sources,
d s (x s , y s )
and then to find out important factors for improving efficiency
and productivity within the context of the program. Specifically,
the objectives were: 1) to evaluate TFP growth at the farm level if we take technology at time s as a benchmark, and as
induced by Grain for Green program and compare farm technical d t (xt , yt )
Mt= (5)
efficiency before and after the first phase of the program; 2) to d t (x s , y s )
identify the sources of TFP growth from technological growth
and technical efficiency change, the latter of which can be further if we take technology at time t as a benchmark.
differentiated into pure technical efficiency change, and scale To avoid arbitrariness in choosing the benchmark, we follow
efficiency change; and 3) to find out the factors that contributed Färe et al. 17 to define the Malmquist TFP index as the geometric
to improved TFP, technology, and technical efficiency, and our mean of the above two indices:
emphasis is on the specified measures taken in the Grain for Green
1/ 2
program. s s t t
ª d s (xt , yt ) d t (xt , yt ) º (6)
M (x , y , x , y ) « s s s u t s s »
¬« d ( x , y ) d ( x , y ) ¼»
Methodologies
We calculated the Malmquist TFP index with distance functions which is equivalent to:
for the farms before and after the first phase of the program. We 1/ 2
then decomposed the index into technological growth and technical d t (xt , y t ) ª d s (xt , yt ) d s (x s , y s ) º (7)
M (x s , y s , xt , yt ) u « u »
efficiency change, and further decomposed the latter into pure d s ( x s , y s ) ¬« d t ( x t , y t ) d t ( x s , y s ) ¼»

1186 Journal of Food, Agriculture & Environment, Vol.8 (2), April 2010
Here, the ratio outside the brackets measures the relative K

technical efficiency change between time s and t, which can be s.t. T k 's y mk 's d ¦ z ks y mks m = 1,…,M,
k 1
interpreted as the effect of inefficient farms catching up with better
ones (also called catch-up). Farms catch up with the leading farms
K
if they get closer to the frontier, and fall behind if they get further
from the frontier. The geometric mean of the two ratios inside the ¦z
k 1
ks
x nks d x nk 's n = 1,…,N,

brackets captures the shift in technology between the two periods,


or technological growth (also called innovation). Expansion of zks ≥ 0 k = 1,…,K, (10)
frontier occurs mainly due to technological advances or
infrastructure investment. In the case of an upward expansion, we
call it technological progress, and downward expansion,
technological regress.
Allowing variable return to scale (VRS), Färe et al. 17 further
decomposed the index into:
1/ 2
d Vt ( x t , y t ) ª d Vs ( x s , y s ) d Ct ( x t , y t ) º ª d Cs ( x t , y t ) d Cs ( x s , y s ) º
M (x s , y s , xt , yt ) « Ƴ »« Ƴ » (8)
d Vs ( x s , y s ) ¬ d Vt ( x t , y t ) d Cs ( x s , y s ) ¼ ¬ d Ct ( x t , y t ) d Ct ( x s , y s ) ¼

where subscript V stands for VRS, C for CRS assumption. The first In the case that the technology and observations come from
term measures the relative technical efficiency change between different period, e.g. dCs (xk't, yk't), technical efficiency can be
time s and t if VRS is assumed; it represents changes in managerial computed by solving the following linear programming:
skills to produce more with less, and is referred to as pure technical
efficiency change. The second term measures the change in the >d s
C ( x k 't , y k 't ) @
1
max T k 't
z ,T
ratio of the distance function satisfying CRS to the distance
function satisfying VRS, so it represents the efforts of farms to K

move to the optimum scale and is called scale efficiency change. s.t. T k 't y mk 't d ¦ z ks y mks m = 1,…,M,
k 1
In another words, technical efficiency change is the product of
K
pure technical efficiency change and scale efficiency change. The
third term, as mentioned above, is technological growth. Any value
¦z
k 1
ks
x nks d x nk 't n = 1,…,N,

greater or less than 1 indicates positive or negative growth,


respectively, and a value equal to 1 indicates stagnation for the zks ≥ 0 k = 1,…,K, (11)
specified efficient or technology change indicator.
Similarly, we can calculate d Ct (xk's, yk's).
Data Envelopment Analysis: Both Data Envelopment Analysis We can calculate distance functions under VRS, d Vs (xk's, yk's) and d
K
(DEA), the non-parametric method, and Stochastic Frontier t
V
(xk't, yk't), by adding the ∑ zks = 1 constraint 17.
k=1
Analysis (SFA), the parametric method can be used to estimate
the distance functions that constitute the Malmquist TFP index. Study Area and Dataset
DEA gained more popularity over SFA in recent years. In our We selected Zhifanggou watershed (latitude 36°51´30´´N,
study, because it offers a flexible environment in which multiple longitude 109°19´23´´E; Fig. 1), one of the pilot areas for the Grain
inputs and outputs, even with different units of measurement, for Green program, as the study area. It is situated in Ansai County,
can be easily processed, we construct the output-oriented Shaanxi province, on the Loess Plateau, with a total area of about
Malmquist TFP index using the distance functions calculated from 8.27 km2, and is characterized by numerous hills and gullies with
a DEA Programming. thick loess coverage. It has three villages: Zhifanggou, Siyaoxian,
Following Färe et al. 17, suppose there are K farms (indexed by k) and Washuta. With poor vegetation and a typical semi-arid
using N inputs (indexed by n) to produce M products (indexed by continental climate, the watershed has been pervaded by soil
m). Accordingly, xnks and ymks denote the nth input and mth output for erosion, and grain yield was only about 25 kg/mu in the 1970s 18.
the kth farm at time s, the base time (or xnk t and ymk t at time t, the With population growth and economic development, by the early
current time). 1990s the ratio of land under cultivation had increased to 47.9%
The data set is given by: and goat numbers had increased by half 18. In 1999, the Grain for
Green program was initiated in the watershed at a full scale. Farmers
{(xk,i,yk,i): k = 1,…K; i = s,t} (9) were offered a total 160 yuan (or grain of equivalent value in the
first 3 years) per mu for setting aside sloping land over 25° and
Assuming CRS technology, technical efficiency relative to the planting trees, providing the seedling survival rate passed the
reference technology at time s (or t), dCs (xk's, yk's) (or dCt (xk't, yk't)), standard in annual inspection. Afforestation was divided into
can be calculated by solving the following linear programming: ecological forests, subsidized for 8 years, and economic forests,
subsidized for 5 years. Zhifanggou and Washuta villages were
>d s
C ( x k 's , y k 's ) @
1
maxT k 's
z ,T
assigned ecological forests and Siyaoxian economic forests. The

Journal of Food, Agriculture & Environment, Vol.8 (2), April 2010 1187
Table 1. Inputs and outputs used to calculate TFP growth and its
sources.
Variable Definition
Outputs
Crop production Market value of crops produced on the farm (yuana).
Livestock production Market value of animals and their products produced
on the farm (yuan).
Inputs
Land Cultivated land on the farm (mub) .
Labor Number of labors in crop and livestock production on
the farm, adjusted by actual working hours (person).
Material and capital Seed, fertilizer, fodder, fuel, pesticide, irrigation,
depreciation and maintenance cost of machinery and
building, wages , rent (yuan).
a: 1yuan=0.147$. b:1mu=0.067ha.

Figure 1. Map of Zhifanggou watershed, Ansai County, Shaanxi Province, Table 2. Descriptive statistics of the inputs and outputs.
China. Outputs Inputs
Year Crop Livestock Material and
Labor Land
production production capital
(person) (mu)
program also involved terracing gentle slopes (less than 25°), (yuan) (yuan) (yuan)
1999 Ave. 5428.296 1344.068 2.085 22.164 1250.201
which was financially supported by the local government; SD 3810.336 1573.854 0.724 13.234 1297.844
providing agricultural extension services by local extension Min 879.000 0.000 0.700 2.000 27.700
agencies especially on greenhouse crop management, fruit Max 17915.000 6500.000 4.000 64.000 5956.540
growing, and livestock breeding; generating off-farm employment; 2007 Ave. 7856.991 1384.681 1.780 7.992 1530.060
SD 9826.185 4488.744 0.507 3.600 2183.706
and offering easier access to loans from Agricultural Bank of China, Min 401.825 0.000 1.000 0.800 22.373
especially to those farmers who were engaged in greenhouse crop Max 39349.070 33733.370 4.000 15.500 12254.980
production.
A research station set up by the Institute of Soil and Water The reason for an increase in crop production with a shrunk
Conservation of the Chinese Academy of Sciences followed the cultivated land area is twofold. On the one hand, it resulted from
household level activities in the study area for over 10 years. In the increase in land productivity owing to the soil and water
addition, members of the Sino-Japan Research on Combating conservation practices, especially terracing slope lands, and the
Desertification and Developmental Utilization in Inland China introduction of new seeds and technology like using plastic
project conducted further investigations in 2006 and 2008 to fill in coversheets to keep soil moisture content and temperature, etc;
gaps. Totally 59 farms were randomly selected for investigation on the other hand, resulted from the shift from extensive,
and used for our analysis. subsistence production to intensive, cash production. The share
To calculate the change in TFP after the execution of the program of land under cultivation for cash crops including fruits, melon
and to identify its sources, we collected data on agricultural and vegetables increased from 7% to 26.4% after the execution of
outputs and inputs (Table 1) for 1999, the year before the initiation the program (Table 3).
of the program, and 2007, the last year of the first phase. As shown Previous studies have shown that the variance in the
in Table 1, we adopt the most common two-outputs (crop and demographic, socio-economic characteristics and policy
livestock) three-inputs (land, labor and material and capital) model. participation explains the differences in TFP growth, technological
Here crops include millet, soybean, maize, potato, fruits, melon growth or technical efficiency change 15, 24, 25. Table 4 defines the
and vegetables, while livestock include cattle, sheep, pigs, explanatory variables in our study: age, health status, education,
chickens, and their products including eggs, milk and wool. To subsidy, off-farm employment, access to credit, land terracing and
avoid the effects of currency inflation, we deflated the market extension.
value of the data by the GDP deflator (National Bureau of Statistics The characteristics of the householders, who are generally the
of China). decision-makers or who play a key role in agricultural production
Descriptive statistics of the input and output data are shown in and influence family activities, are important factors that affect
Table 2. Eight years after the introduction of the program,
the average area of cultivated land per farm had decreased
by 63.9% as a result of setting-aside sloping land; the Table 3. Land productivity and land share for the major crop products in
labor force decreased by 14.6%, most of which had 1999 and 2007.
shifted to non-agricultural production (refer to Table 5 1999 2007
Average area Land Yield Average area Land Yield
that ratio of non-agricultural income has increased by Crops
of cultivated share (kilogram/ of cultivated share (kilogram/
12.6%); and material and capital input had increased by land (mu) (%) mu) land (mu) (%) mu)
22.4% on average, most of which resulted from increase Millet 8.255 0.373 97.174 1.015 0.127 135.132
in the usage of fertilizer, irrigation facilities, tractor and Soybean 6.725 0.303 96.679 1.226 0.153 124.961
Maize 3.481 0.157 287.710 2.733 0.342 429.923
greenhouses. For the agricultural output, while livestock Potato 2.161 0.097 425.626 0.912 0.114 652.511
production increased by 3.0%, crop production increased Fruits 1.390 0.063 1015.010 1.092 0.137 1226.609
significantly, by 44.7% on average. Melon 0.093 0.004 2266.667 0.678 0.085 2780.000
Vegetables 0.059 0.003 6366.667 0.336 0.042 6925.000
Total 22.164 7.992

1188 Journal of Food, Agriculture & Environment, Vol.8 (2), April 2010
Table 4. Explanatory variables for TFP growth and its sources. householders were middle-aged people with limited education,
Explanatory variable Definition around primary school level. Nearly 90% had reasonable health.
Age Age of householder in 1999 (years). The average terraced area is about 1.178 mu and the annual
Health status Health status of householder; 1= good and 0 = bad. subsidy received through the program was about 2.287 thousand
Education Education level of householder; 0 = illiterate, 5 = yuan although great variation existed in the samples. Off-farm
primary school, 8 = junior high school, 12 = senior
high school. employment increased substantially, as non-agricultural income
Subsidy Average annual subsidy received for participation ratio rose by 12.6%. Financing by credit remained limited, as most
in the Program in the past 8 years (thousand yuan). farmers had not access to credit; the sum of credit for the 8 years
Off-farm employment Change in the ratio of non-agricultural income to
amounted to only 2.6% of total assets, and more than half of the
total household income in the past 8 years
(percentage). farmers received training or extension services.
Access to credit Ratio of the total credit over the past 8 years to total
household assets (percentage). Results and Discussion
Land terracing Total area of land terraced on the farm in the
Program during in the 8 years (mu). Malmquist TFP index and its sources: Table 6 shows the TFP
Extension Whether the farm has received the training or growth result for the farms participating in Grain for Green program.
extension services under the Program in the past 8 The results suggest that TFP rose by 52.5% on average. Forty-
years; 1= yes and 0 = no.
four households (74.6%) showed increased TFP as a result of the
program. Technological progress, or the shift in best practice
TFP growth and its sources. Here we include their age, health technology, was the only contributor to TFP growth of the farms.
status, and education as the explanatory variables. We should In contrast, retrogression of technical efficiency, or slackened
note that we judged health status by whether the householder catch-up effect, prevented TFP from improving further.
had suffered a chronic or major disease in the past 8 years and we The adoption of better technologies led to dramatic improvements
set 1 for those who had not and 0 for those who had. We in agricultural production, with a range of 1.148 to 4.054 and an
acknowledge that the standard of judgment used here is coarse, average of 1.760. This might be a result of the specified measures
owing to the limited data availability. We used years of schooling in the program, like terracing, support for green-house farming
to reflect education level; but as the information was obtained as and fostering of livestock farming, or through financial support,
diplomacy, following Fan et al. 19, we assign 0, 5, 8, and 12 years which allowed farms to take advantage of the modern facilities,
for householders who are illiterate or semi-illiterate, primary school like tractor, coversheet, greenhouses, irrigation facilities or even
diploma, junior high school diploma and senior high school improved seeds.
diploma, respectively. The farmers witnessed a regression in technical efficiency of
We also introduced indicators to reflect farm’s participation in 13.5% on average. This implies that the farmers had not adapted
the program, which include subsidy, off-farm employment, access well to the improved technology, and had thus failed to catch up
to credit, land terracing and extension, into the model to explain with the better-off. To trace the sources of the decrease in technical
differences in TFP growth and its sources. Lack of capital limits efficiency, we decomposed it into pure technical efficiency and
the purchase of inputs and efficiency-improving equipment and scale efficiency effect, as shown in columns 3 and 4 of Table 6.
thus discourages the efficient allocation of resources and adoption The change in scale efficiency had little effect, decreasing by
of technology. We reckon here that the subsides by the program, only 2.2% on average. This result suggests that more farmers
income from off-farm employment and loans from the banks help failed to operate at an optimum scale after execution of the program.
the farmers to break out the capital limitations and make optimal This might be a result of the decreasing cultivated land area. The
decisions on agricultural production. Terracing slopes not only dominant contributor to the decrease in technical efficiency was
reduce soil erosion but also retain the water, thus is expected to pure technical efficiency change, which decreased by 10.8%. This
increase crop yield. The effect of training and agricultural extension decline can be interpreted as a general retrogression in the ability
on agricultural production has been widely studied, and its
significance verified in many studies 20-24. Table 5. Descriptive statistics of the explanatory variables.
Climate variables are major determinants of agricultural Ave. SD Min Max
production. Drought is the most prominent meteorological disaster Age (years) 39.508 10.224 22.000 71.000
that threatens crop yields and farmers’ welfare in our study area. Health status (0, 1) 0.898 0.305 0.000 1.000
Lacking field data for the study area, we collected monthly data Education (0-12 years) 4.119 3.649 0.000 11.000
Subsidy (thousand yuan) 2.287 1.576 0.640 7.200
from Yan’an city, the local administrative center (Chinese Off-farm employment (percentage) 0.126 0.299 -0.654 0.954
Meteorological Administration), instead. No abnormal event was Access to credit (percentage) 0.026 0.112 0.000 0.800
observed in the 2 years selected for the study, and difference in Terraced
Land land area
terracing (mu)(mu) 1.178 3.517 0.000 14.000
the annual precipitation is also negligible in the two years. So we Extension (0, 1) 0.610 0.492 0.000 1.000
ignored the impact of weather.
Our priori expectation is that age, health status,
Table 6. Estimated Malmquist index and the decomposition result.
education, subsidy, off-farm employment, access to credit,
Technological Technical Pure technical Scale efficiency TFP
land terracing, and extension will generally have positive
Farm growth efficiency change efficiency change growth
influences on TFP growth, technological growth and (1) (2)=(3)*(4) change (3) (4) (5)=(1)*(2)
technical efficiency change. Ave. 1.760 0.865 0.892 0.978 1.525
Table 5 presents descriptive statistics for all the SD 0.636 0.386 0.359 0.190 0.993
Min 1.148 0.145 0.141 0.289 0.343
explanatory variables listed above. In general, the Max 4.054 2.542 2.500 1.813 6.448

Journal of Food, Agriculture & Environment, Vol.8 (2), April 2010 1189
of the farmers to manage agricultural production under the new Regression analysis: We regressed the indices of TFP growth,
technology. That is, even though the farmers used conventional technological growth, and technical efficiency changes against
technology in 1999 efficiently but they are not as successfully the aforementioned eight explanatory variables. The OLS
managing the agricultural production as their leading farms under regression can easily be extended to models involving two or
the new technology. more explanatory variables against several dependent variables,
and that justifies the reason we chose this method.
Technical efficiency: DEA generates relative efficiency The OLS estimates of the effect of the explanatory variables on
measurements that make no sense if compared between different three indices are revealed in Table 7. The result shows that land
periods directly because the relative efficiencies are based on terracing is the only variable that is significantly related to all of
different technology levels. So we plotted the frequency the three indices. The coefficients are positive, which suggests
distribution of technical efficiencies for the sample farms in 1998 that land terracing benefited TFP growth, technological growth
and 2007, under the CRS assumption, to see the changes in the as well as technical efficiency change. The positive relationship
distribution of the efficiencies (Fig. 2). The average efficiency in with TFP growth and technical efficiency change, as we supposed,
1999 was 0.750 with a standard deviation of 0.197, whereas that in might result from improved water and nutrient holding capacity
2007 was 0.624 with a standard deviation of 0.230. Fig. 2 indicates under leveled land. The positive relationship of land terracing
that the farmers had a more even efficiency distribution after the and technological growth might be related to the fact that it makes
execution of the program. Both distributions show bimodal possible of the farmers to employ facilities like tractors.
features. One mode is around 0.6-0.7 and another higher mode is
around 0.9-1 in 1999; while in 2007, one mode is around 0.5-0.6 and Table 7. Factors explaining TFP growth, technological growth
another lower mode is around 0.9-1. This suggests that there were and technical efficiency change.
more farms close to or had reached the best practice frontier in Dependent variables
1999. This result accords with our previous finding that only a Independent variables Technological Technical
TFP growth
growth efficiency change
few farms caught-up with the leading farms after the execution of 1.914 1.120 1.402
the program. Constant
2.703 (t)*** 2.493 (t) ** 4.300 (t) ***
-0.018 0.002 -0.011
18 Age
-1.606 (t) 0.240 (t) -2.041 (t)**
16 0.125 0.084 0.052
Health status
14 0.749 (t) 0.343 (t) 0.294 (t)
12 -0.012 0.013 -0.013
Education
-0.367 (t) 0.657 (t) -0.905 (t)
10
1999 0.031 0.042 -0.006
8 Subsidy
2007
0.447 (t) 0.937 (t) -0.194 (t)
6 -0.236 0.036 -0.142
Off-farm employment
4 -0.633 (t) 0.153 (t) -0.827 (t)
2 2.294 1.294 0.473
Access to credit
2.261 (t) ** 2.010 (t) ** 1.014 (t)
0
0-0.1 0.1-0.2 0.2-0.3 0.3-0.4 0.4-0.5 0.5-0.6 0.6-0.7 0.7-0.8 0.8-0.9 0.9-1
0.140 0.077 0.033
Land terracing
4.439 (t) *** 3.833 (t) *** 2.281 (t) **
CRS efficiency scores
0.003 0.358 -0.207
Figure 2. Frequency distribution of technical efficiencies of the farms in Extension
0.015 (t) 2.484 (t) ** -1.978 (t)
1999 and 2007. Multiple R 0.691 0.698 0.517
R2 0.477 0.487 0.267
2
Adjusted R 0.393 0.404 0.150
t: t statistics; ***: P < 0.01; **: P < 0.05.
We were also interested in the change in farms’ efficiency rankings
after the execution of the program to see whether the originally
inefficient farms benefited more. To test the relationship between For the three variables we had reckoned to help break the capital
the efficiency ranks before and after the first phase of program, limitations, we are surprised to find out, that as the principle
we used Spearman’s Rank Correlation Analysis. The correlation measures for the Grain for Green program, subsidy provided by
value in this method ranges from -1.0 (perfect negative correlation), the government to the farmers are not significantly related to
through 0 (no correlation), to +1.0 (perfect positive correlation). improvement in TFP, technology or technical efficiency. This result
To avoid the same ranks resulting from same technical efficiency is very interesting and meaningful. Actually, Karagiannis and
values, technical efficiency was given as super-efficiency value, Sarris (unpublished) 25 found subsidies decreased technical
which is measured relative to a frontier derived without the farm efficiency for the farms participating in CAP Land Set-aside and
under consideration. The rank correlation between 1999 and 2007 Price Decline Compensation Programs in Greece. The implication
is statistically significant (t statistics =2.170, P = 0.034) and positive is that direct subsidy to farms might not be an effective way to
but not high, with a value of only 0.276, which reflects low ranking improve farms’ efficiency and productivity. The possible reason
stability for the farms. Among the 13 efficient farms in 1999, only might be that direct subsidy weakens DMU’s motivation 26. Nor is
3 of them remained efficient in 2007; and that for the formerly the variable of off-farm employment significantly related to the
inefficient farms, eight improved greatly, four of which even leaped three indices. The result rejected our hypothesis that off-farm
to the top. This result suggests that the program preferentially employment of the surplus labors would not only raise total
emphasized farms with an unfavorable initial state. household income, improving the potential to invest in capital-
intensive efficiency-improving practices, but also broaden the
farmers’ horizon, facilitating imitation and competition (see, e.g.

1190 Journal of Food, Agriculture & Environment, Vol.8 (2), April 2010
Nehring and Fernandez-Cornejo 27 found evidence that off-farm Thus we suggest that policy-makers focus resources on terracing
income boosts scale and technical efficiency for U.S. smallholder sloping lands, improving access to credit, and offering extension
farms). The underlying reason might be that competent labor services, other than providing subsidy to farmers during next
forces are distracted to more profitable non-agricultural activities phase of Grain for Green program.
and decisions are directed to improving household efficiency
instead of farm efficiency 15, 28. In contrast, the variable of access Acknowledgements
to credit is significantly related to TFP improvement and This work was supported by the Japan Society for the Promotion
technological growth. The result is consistent to the fact that of Science (JSPS) Core University Program and the Global Center
loan from the bank generally goes to farms engaged in greenhouse of Excellence (GCOE) Program in Japan. We are grateful to Prof.
vegetable production for cover-sheeting; and pig or cattle Guobin Liu, from the Institute of Soil and Water Conservation of
breeding farms for purchasing feed additives and other the Chinese Academy of Sciences, for his assistance during the
innovations. investigations and the villagers in the study area for their kind
Our result shows that contacting the farmers with training and cooperation. We appreciate the effort by Dr Mohan Saxena from
extension services improved farmers’ technological level ICARDA with the language. Three anonymous reviewer are
significantly as we expected. The interpretation in our study is acknowladged for their invaluable comments.
that introducing new seeds and varieties, spreading of the
irrigation, temperature, pest control and other vegetables and fruits References
management practices, and breeding and feeding practices by the 1
Komatsu, Y., Tsunekawa, A. and Ju, H. 2005. Evaluation of agricultural
local public agencies through the Grain for Green program, have sustainability based on human carrying capacity in drylands - A case
significantly improved the technology level for the sample farms. study in rural villages in Inner Mongolia, China. Agr. Ecosyst. Environ.
The coefficient of the age of householders on technical efficiency 108:29-43.
2
Lin, G. C. S. and Ho, S. P. S. 2003. China’s land resources and land-use
change is significant and negative, which is consistent to Tauer
change: Insights from the 1996 land survey. Land Use Policy 20(2):87-
and Nazibrola’s work 29. The implication is that the older farmers 107.
were less successful receptive or adapting to the new technology 3
Ash, R. F. and Edmonds, R. L. 1998. China’s land resources, environment
and practices. and agricultural production. The China Quarterly 156:836-879.
In short, the empirical analysis results indicate that terracing 4
Liu, J., Li, S. X., Ouyang, Z. Y., Tam, C. and Chen, X. D. 2008. Ecological
sloping land and broadening access to credit significantly and socioeconomic effects of China’s policies for ecosystem services.
increased TFP and technology; that the former also significantly P. Natl. Acad. Sci. USA 105(28):9477-9482.
5
improved technical efficiency, extension significantly improved Cao, S. X., Li, C. and Yu, X. X. 2009. Impact of China’s Grain for Green
technology and age withdrew technical efficiency from improving project on the landscape of vulnerable arid and semi-arid agricultural
regions: A case study in northern Shaanxi Province. J. Appl. Ecol.
significantly.
46(3):536-543.
6
Feng, Z. M., Yang, Y. Z., Zhang, Y. Q., Zhang, P. T. and Li, Y. Q. 2005.
Conclusions Grain-for-green policy and its impacts on grain supply in West China.
The Grain for Green program, with a mission to contribute to Land Use Policy 22:301-312.
sustainable development in rural areas in China by compensating 7
Peng, H., Cheng, G., Xu, Z., Yin, Y. and Xu, W. 2007. Social, economic,
farmers for setting aside marginal and ecologically fragile land, and ecological impacts of the “Grain for Green” project in China: A
has aroused interest in enhancing its effectiveness and preliminary case in Zhangye, Northwest China. J. Environ. Manage.
sustainability. 85(3):774-784.
8
Our results show that the program had greatly improved TFP in Wang, C., Ouyang, H., Maclaren, V., Yin, Y., Shao, B., Boland, A. and
Tian, Y. 2007. Evaluation of the economic and environmental impact
the Zhifanggou watershed, solely through technological progress.
of converting cropland to forest: A case study in Dunhua county,
In contrast, technical efficiency decreased as a result of decreases China. J. Environ. Manage. 85(3):746-756.
in both pure technical efficiency and scale efficiency. Thus, even 9
Zhou, H., Rompaey, A. V. and Wang, J. 2009. Detecting the impact of
though the program had improved the technology available to the “Grain for Green” program on the mean annual vegetation cover in
the farmers, the farmers’ ability to produce more agricultural the Shaanxi province, China using SPOT-VGT NDVI data. Land Use
products under improved technology had decreased. Policy 26(4):954-960.
10
Only a few farms successfully managed to catch up with the Deng, X. Z., Huang, J. K., Rozelle S. and Uchida, E. 2006. Cultivated
leading farms after the execution of the program, although the land conversion and potential agricultural productivity in China. Land
distribution of technical efficiency has became more equitable. Use Policy 23:372-384.
11
Uchida, E., Xu, J., Xu, Z. and Rozelle, S. 2007. Are the poor benefiting
Farms with a poorer technical efficiency benefited more than the
from China’s Land Conservation Program?. Environ. Dev. Econ. 12:593-
previously better performers in the efficiencies. 620.
Land terracing and access to credit contributed significantly to 12
Xu, Z. G., Xu, J. T., Deng, X. Z, Huang, J. K, Uchida, E. and Rozelle,
TFP growth and technological growth. Land terracing also S. 2006. Grain for Green versus grain: Conflict between food security
contributed significantly to technical efficiency change. Extension and conservation set-aside in China. World Dev. 34(1):130-148.
services contributed significantly to technological growth, and 13
Bennett, M. T. 2008. China’s sloping land conversion program:
age of the householder withdraws technical efficiency change Institutional innovation or business as usual?. Ecol. Econ. 65(4):699-
significantly. 711.
14
Farrell, M. J. 1957. The measurement of productive efficiency. J. R.
Note: We should admit that if data permits, it would be better to make a comparison between the farms
Stat. Soc. 120: 252-90.
15
participating in the Program and those which don’t; and we should note that the same method used in this Solís, D., Bravo-Ureta, B. and Quiroga, R. 2009. Technical efficiency
study might not necessary get the same result when used in other area’s analysis as some of the measures
taken in the pilot areas haven’t been extended to other areas.
among peasant farmers participating in natural resource management

Journal of Food, Agriculture & Environment, Vol.8 (2), April 2010 1191
programs in Central America. J. Agr. Econ. 60(1):202-219.
16
Caves, D. W., Christensen, L. R. and Diewert, W. E. 1982. The economic
theory of index numbers and the measurement of input, output, and
productivity. Econometrica 5:1393-1414.
17
Färe, R., Grosskopf, S., Norris, M. and Zhang, Z. 1994. Productivity
growth, technical progress, and efficiency change in industrialized
countries. Am. Econ. Rev. 84:66-83.
18
Dai, Q., Liu, G., Xue, S., Lan, X., Zhai, S., Tian, J. and Wang, G. 2007.
Health diagnoses of ecosystems subject to a typical erosion
environment in Zhifanggou watershed, north-west China. Frontiers of
Forestry in China 2(3):241-250.
19
Fan, S. and Zhang, X. 2004. Infrastructure and regional economic
development in rural China. China Econ. Rev. 15:203-214.
20
Amaza, P. S. and Ogundari, K. 2008. An investigation of the factors
that influence the technical efficiency of soybean production in the
Guinea savannas of Nigeria. J. Food Agric. Environ. 6(1):92-96.
21
Birkhaeuser, D., Evenson, R. E. and Feder G. 1991. The economic
impact of agricultural extension: A review. Econ. Dev. Cult. Change
39(3):607-650.
22
Feder, G., Lau, L. J. and Slade, R. H. 1987. Does agricultural extension
pay? The training and visit system in Northwest India. Am. J. Agr.
Econ. 69(3):677-686.
23
Oladele, O. I. and Sakagami, J. I. 2004. SWOT analysis of extension
systems in Asian and West African countries. J. Food Agric. Environ.
2(2):232-236.
24
Rosegrant, M. W. and Evenson, R. E. 1992. Agricultural productivity
and sources of growth in South Asia. Am. J. Agr. Econ. 74(3):757-761.
25
Karagiannis, G. and Sarris, A. 2002. Direct Subsidies and Technical
Efficiency in Greek Agriculture. 2002 International Congress. August
28-31. Zaragoza, Spain 24868. European Association of Agricultural
Economists.
26
Bergström, F. 2000. Capital Subsidies and the Performance of Firms.
Small Bus. Econ. 14(3):83–93.
27
Nehring, R. and Fernandez-Cornejo, J. 2005. The impacts of off-Farm
income on farm efficiency, scale and profitability for corn farms. 2005
Annual meeting for American agricultural economics association (New
Name 2008: Agricultural and Applied Economics Association).
Providence. RI 19566.
28
Pfeiffer, L., López-Feldman, A. and Taylor, J. E. 2009. Is off-farm
income reforming the farm? Evidence from Mexico. Agr. Econ.
40(2):125-138.
29
Tauer, L. W. and Nazibrola, L. 2000. Farmer efficiency and technology
use with age. Agr. Resource Econ. Rev. 29:24-31.

1192 Journal of Food, Agriculture & Environment, Vol.8 (2), April 2010

Vous aimerez peut-être aussi