Sie sind auf Seite 1von 7

WFL Publisher

Science and Technology

Meri-Rastilantie 3 B, FI-00980 Journal of Food, Agriculture & Environment Vol.8 (3&4): 772-778. 2010 www.world-food.net
Helsinki, Finland
e-mail: info@world-food.net

Efficiency and its determinant factors for smallholder farms in the Grain for Green
Program on the Loess Plateau, China
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

Received 16 July 2010, accepted 25 October 2010.

Abstract
A nonparametric frontier approach, data envelopment analysis, was used to calculate technical, pure technical, cost, allocative and scale efficiencies
of 112 farms participating in the Grain for Green Program on the Loess Plateau in China. The results indicate substantial cost inefficiencies. The main
source of cost inefficiency was allocative inefficiency although considerable technical inefficiency also exists. Technical inefficiency resulted slightly
more from scale inefficiency than from pure technical inefficiency. Scale inefficiency was mostly due to the suboptimal scale of farms. Regression
analysis of explanatory variables against technical efficiency, allocative efficiency and scale efficiency suggested that LG_Agri (indicator for farm
size) was the only variable that was significantly related to all three efficiency indices. Other significant variables include Remittance Ratio (indicator
for rural-urban migration), Tenancy Ratio (indicator for land tenancy) and Simpson Index (indicator for land fragmentation). Our findings prompt
policy suggestions that would facilitate labor mobility, provide more secure land tenure and increase land consolidation, while avoiding further
withdrawal of cultivated land unless explicit measures to alleviate the potential impact on production efficiency are followed.

Key words: Semi-arid region, technical efficiency, allocative efficiency, scale efficiency, farm size, land tenure, land fragmentation, labor mobility.

Introduction
The Loess Plateau, a vast area in northwestern China comprising sloping lands. The Program also devoted to foster long-term
Shaanxi, Shanxi, Gansu provinces, the Ningxia Hui Autonomous sustainable rural livelihoods through measures like building
Region and part of others, provides home and subsistence to terraces, spreading bio-energy technologies, promoting out-
over 69 million farmers. Due to the extremely highly erodible migration and off-farm work, expanding agricultural grants-in-aid
characteristics of the topsoil to wind and water, Loess Plateau and providing extension services to farmers, etc. In general, the
was among the regions with the most serious soil erosion in the Program is intended to help the peasants escape the poverty trap
world. As a result of million years’ erosion, it is a heavily dissected through financial support, technical assistance and institutional
landscape that is, in part, too steep for cropping and too fragmented improvement so that agricultural production improves and surplus
for communication. Last several decades, however, population labors shift to off-farm jobs (State Council of the P.R.C., 2007).
growth and increasing demand for food caused widespread Agricultural production can be improved by intensifying
overgrazing and excess reclamation of marginal and sloping land, agricultural inputs, adopting new technology or improving
leading to accelerated land degradation 1. On the other hand, with efficiency. However, options to expand agricultural production
the shift of farmland from a collectively system to individual by intensifying inputs within the context of the Grain for Green
household responsibility and the occasional reallocation of land Program are limited because the program not only directly
among farms to adapt to demographic changes, the land saw an decreases the cultivated land area, but also encourages off-farm
increasing trend toward fragmentation 2, 3. While land degradation employment. Recent work by Li et al. 4 suggests that as a result of
and fragmentation are detrimental to agricultural production several years’ efforts by the program, even though the technology
(UNEP), overpopulation brings about massive potential labor available to farmers has greatly improved, farm efficiency under
surplus, lacking employment opportunities, which also reduce the improved technology has not. Therefore, the most effective
the productivity and efficiency in agricultural production. The way for farms to increase their agricultural production is to improve
combined consequence of these factors was that the subsistence production efficiency 5. Consequently, the aims of this study were
farmers of the Loess Plateau were living in poverty. to explore (1) whether and to what extent farms are able to realize
To help mitigate soil erosion, rehabilitate the environment and the full production potential of emerging technologies, or their
improve the welfare of rural communities, Chinese government production efficiency, under the prevailing circumstances brought
initiated the Grain for Green Program on the Loess Plateau in 1999. about by the program; (2) whether and to what extent involvement
The principal measure of the Program is to compensate farmers in non-agricultural employment, out-migration (Uchida et al. 6
with grain or cash for reforesting cultivated fragile marginal or and Xu et al. 7 found evidence of increase in non-agricultural

772 Journal of Food, Agriculture & Environment, Vol.8 (3&4), July-October 2010
employment and rural-to-urban migration), land rental and land Estimation of efficiencies
tenancy (Zhang et al. 8 observed the emergence of the transfer of Technical and pure technical efficiency: Following Fare et al. 14,
land use right in farming among farmers to mitigate the land suppose that K farms (indexed by k) use N inputs (indexed by n)
constraints created by the program), the degree of land to produce M products (indexed by m), we can estimate the input-
fragmentation and access to credit affect farm efficiency; and (3) oriented technical efficiency of the ith farm, under the assumption
which other factors, if any, are responsible for variation in farm of CRS, TEi (yi, xi), by solving the following linear programme:
efficiency.
TEi (yi, xi)= θmin θi ,

Methods i

We used a two-stage approach to this study. First, we used data subject to


envelopment analysis (DEA) to estimate efficiencies, including K
cost efficiency, technical efficiency, pure technical efficiency,
allocative efficiency and scale efficiency, excluding discretionary
– yim + Σy
k=1
km
λk
≥ 0, m = 1, 2, … , M

variables. Second, we used farm-specific socio-economic variables K


related to involvement in non-agricultural employment and out-
migration, land tenancy, the degree of land fragmentation and
θi xin – Σx
k=1
kn
λk ≥ 0, n = 1, 2, … , N

access to credit in an ordinary least square (OLS) regression


framework to explain variation in measured efficiencies. λk ≥ 0, k = 1, 2, … , K
Here we refer to widely used definition 9 that the efficiency of a
decision making unit (DMU, such as a firm, a farm, etc) is the where xi and yi are the input and output vectors of the ith farm,
distance between observed and optimal values (frontiers) of its respectively; xkn and ykm are the nth input and mth output of the kth
outputs and inputs 9. If the combination of the inputs and outputs farm; λk represents the weights to be used in the inputs for the kth
lies on the production frontier, it is called efficient; otherwise, farm to achieve efficiency; and θi is the technical efficiency of the
inefficient. ith farm, bound by 0 and 1, where 1 indicates efficient and less than
The concept of technical efficiency was firstly advocated by 1 indicates inefficient.
Koopmans 10 and then popularized by Farrell 11. It refers to the The technical efficiency under VRS, pure technical efficiency,
ability to avoid waste by producing as much output as possible PTEi (yi, xi), can be derived by adding the convexity constraint
as input usage allows (output orientation), or by using as little K
input as output production allows (input orientation). Farrell 11
decomposed economic efficiency — the ability of a DMU to get
Σ λ =1 , which benchmarks the farm against farms of a similar size.
k=1
k

more output at minimum cost — into two components, given the Cost and allocative efficiency: To estimate the allocative
assumption of constant returns-to-scale (CRS). One component efficiency of the ith farm, we first need to calculate cost efficiency
is technical efficiency. The other is allocative efficiency (termed by solving the following cost-minimizing DEA model under the
as price efficiency by Farrell 11), which refers to the ability to assumption of CRS:
combine inputs and outputs in optimal proportions in light of
MCi (yi, , cin) = ,
prevailing prices. Since technical efficiency as defined above takes
no account of the scale effect, Banker et al. 12 introduced the
concept of technical efficiency under the assumption of variable subject to
returns-to-scale (VRS), which was later called pure technical
K
efficiency, to distinguish scale efficiency from technical efficiency.
Scale efficiency measures whether the DMU is operating at the
–yim + Σy
k=1
km λk ≥ 0, m = 1, 2, … , M
optimal size at which change in size will not improve output or K
revenue.
The efficiency of a DMU can be estimated by using a parametric
x’in – Σx
k=1
kn λk ≥ 0, n = 1, 2, … , N

method, such as stochastic frontier approach, or a non-parametric


method, such as DEA. We chose DEA because it simultaneously
examines technical efficiency, allocative efficiency and scale λk ≥ 0, k = 1, 2, … , K
K
efficiency, and readily handles multiple inputs and outputs. DEA
requires no assumptions about the form of the production function Σ λ = 1,
k=1
k

and the distribution of the underlying data and thus is less


sensitive to misspecification and unlike parametric methods, is where MCi (yi, x’in, cin) is the minimum total cost for the ith farm
not subject to assumptions about the distribution of the error under CRS; cin is the price of the nth input used by the ith farm for
term (see more comparison between parametric and nonparametric agricultural production; and x’in is the combination cost-minimizing
method, refer to Bauer et al. 13). nth input used by the ith farm. As the actual cost for the ith farm is
We adopted the input-orientation in our study because we given by cin× xin, cost efficiency, the ratio of minimum to actual
assumed that farmers have greater ability to control land, labor cost, is expressed by:
and capital than to adjust outputs according to market prices,
which are regulated by the local government. CEi (yi, xin, cin)= [MCi (yi, x’in, cin)]/(cin× xin)

Journal of Food, Agriculture & Environment, Vol.8 (3&4), July-October 2010 773
The ith farm is cost efficient when CEi =1. affected by severe soil erosion and the farmers there lived in
Allocative efficiency is then computed by: poverty. Both catchments were among the 11 pilot catchments
selected in 1999 for the Grain for Green Program and have been
AEi (yi, xin, cin) = [CEi (yi, xin, cin)]/TEi (yi, xi) subjected to many program-sponsored measures. Considerable
changes have taken place in the rural community as a result of the
under the assumption of CRS 11. program, the most apparent being contraction of the land area.
For example, around 53.5% of the farmland in Zhifanggou
Scale efficiency: The scale efficiency of the ith farm is obtained by catchment had been withdrawn from agricultural production as a
dividing technical efficiency by pure technical efficiency: result of the Grain for Green Program by 2003. The population
engaged in agricultural production decreased from 523 to 501,
TEi (yi , xi) , most of which due to out-migration. The income from non-
SEi (yi, xi) =
PTEi (yi , xi) agricultural activities at the catchment scale increased by 32.5%,
while agricultural production decreased by 16.5% in 2003
where SE = 1 represents scale efficient (under both CRS and VRS) (provided by Wang). It is therefore urgent for policy proposal to
and 0 ≤ SE <1 represents scale inefficient, which can be either increase the farm efficiency of agricultural production for both
increasing returns-to-scale (IRS) or decreasing returns-to-scale food security and the sustainability of the program.
(DRS). In 2008 and 2009, we collected data related to agricultural
Following Coelli and Thirtle 15, the property of returns to scale in production as of 2007, the last year of the first phase of the Grain
operation can be detected by replacing the convexity constraint for Green Program. The data were collected with the joint efforts
K K of the Arid Land Research Center of Tottori University, Japan,
Σλ
k=1 = 1 (VRS) with kΣ
k k
λ
= 1 ≤ 1 (the assumption of non-increasing and the Institute of Soil and Water Conservation of the Chinese
returns-to-scale, NIRS) when estimating technical efficiency and Academy of Sciences.
then comparing the NIRS and VRS efficiency scores. If the two The catchments have a total area of 67.3 km2 and are composed
scores are equal, the farm is operating at decreasing returns-to- of 14 villages. In total, 124 households were randomly surveyed,
scale; and if not, at increasing returns-to-scale. Finally, if technical of which 112 gave valid data: 60 from Zhifanggou and 52 from
efficiency under CRS equals that under VRS, then it is by definition Xiannangou.
operating under CRS.
Inputs and outputs to estimate efficiency scores: To estimate the
Second-stage regression analysis: In efficiency analysis, the input-oriented frontier, and thus the efficiency scores for the farms,
second-stage regression usually involves a Tobit model (because the quantities and prices of all inputs and the total volumes of
the efficiency scores are censored between 0 and 1), or, less outputs were needed.
commonly, maximum likelihood estimate (MLE) or ordinary least The input variables, for which data were collected at the farm
square (OLS) regression. Recently, McDonald 16 proved that the level, included Land, the total area under cultivation for all crops;
efficiency scores by DEA are not generated by a censoring process Labor, the total labor forces engaged in agricultural activities,
and are fractional data, thus, using Tobit in the second stage adjusted by working hours; Seed; Fertilizer; Fuel; Young animals;
would be inconsistent. Instead, he advocated the use of OLS. Fodder. Except for land and labor, the quantities of all inputs were
Hoff 17 also asserted that the Tobit model is not necessarily estimated by dividing total expenditures by the average of sampled
superior to OLS in many cases. Banker and Natarajan 18 further prices. The prices for land and labor in 2007 are taken from the
demonstrated that DEA in the first stage followed by OLS or MLE Shaanxi Statistic Bureau (http://www.sei.gov.cn/). The price of
regression in the second stage yields consistent estimates of the labor was given as the average annual net agricultural income at a
impact of contextual variables as long as the contextual variables provincial level, instead of shadow wage-rate, due to data
are independent of the inputs. We therefore used OLS regression availability. The output variables were Crop and Livestock, both
to identify the exogenous factors associated with farm efficiency, measured by market value. Crop included maize, millet, soybean,
as our data meet the requirements given by Banker and Natarajan 18. potato, buckwheat, peanut, oilseed, vegetables and fruits; and
The regression equation takes the following form: Livestock products included cattle, sheep, pigs, chickens and
their products, such as eggs, milk and wool (refer to Table 1).
J
EFFi = a0 + Σ a V +e,
j=1
ij ij
i Explanatory variables related to efficiency scores: Many studies
have investigated the impact of exogenous variables that might
where EFFi is the efficiency score of the ith farm derived from the influence efficiency at the farm level 19-29 (see Bravo-Ureta and
first stage estimation; a0 is the intercept, Vij is the jth of J explanatory Pinheiro 30 for a detailed review of the socio-economic variables
variables for the ith farm; and ei is the error term. relevant to farm efficiency). In our model, we examined age (Age)
and education (Education) of the householder, farm size
Study Area and Datasets (LG_Agri), non-agricultural employment (NonAgri Ratio),
The sample farms were chosen from the Zhifanggou and remittance from out-migrant family members (Remittance Ratio),
Xiannangou catchments in Ansai County, Shaanxi Province. land tenancy (Tenancy Ratio), land fragmentation (Simpson Index),
Although agricultural production plays a crucial role in local and access to credit (Access to Credit) (refer to Table 2).
economy, the study area has a semi-arid continental climate and We used years of schooling to reflect educational level: following
agriculture is primarily rain-fed. Both catchments have been Fan et al. 31, we assumed 0 years for persons who are illiterate or

774 Journal of Food, Agriculture & Environment, Vol.8 (3&4), July-October 2010
semi-illiterate, 5 years for primary school education, 8 years for Farm efficiency estimates: Cost efficiency, technical efficiency
junior high school education and 12 years for senior high school and allocative efficiency under the assumption of CRS and pure
education. No generally accepted measures exist for farm size, technical efficiency were calculated using DEAP, a computer
although various measures of output, sales or inputs have been program developed by Coelli 34. The statistics of the efficiency
suggested in the literature 32. Land area is the most frequently scores are presented in Table 3. The results indicate considerable
used measure but because we included both crop and livestock inefficiency in agricultural production.
production as the outputs, we adopted the suggestion that an As cost inefficiency is a result of both allocative inefficiency
agricultural output-related index would be more rational 20 and and technical inefficiency, we can conclude that cost inefficiency
take the common logarithm of agricultural income (LG_Agri) to (0.726) is due mainly to allocative inefficiency (0.611) rather than
reflect farm size 21. The NonAgri Ratio is the ratio of non- technical inefficiency (0.311). The farmers could reduce costs by
agricultural income, including earnings from both non-agricultural 61.1% choosing more cost-efficient input combinations without
on-farm and off-farm activities, to total household income. Rural- compromising output.
urban migration and the remittance by out-migrant family members Technical inefficiency (0.311) results from both pure technical
are expected to affect farm production even though the influences inefficiency (0.179) and scale inefficiency (0.181). This implies that
are complex and manifold. On one hand, out-migration helps the if all farms were an optimal size, the cost would be decreased by
shift of surplus labor force and the remittance might be critical for 18.1% to produce the same output.
enhanced agricultural investment and farm management 29, 33. On Table 4 shows the distribution of the efficiency scores. More
the other hand, rural to urban migration usually involves more than 90% of the cost efficiency and allocative efficiency scores
well-educated and robust labors, which has the potential to were less than 0.6, whereas nearly 60% of the technical efficiency
sacrifice agricultural production 22, 29. To distinguish the influence scores were above 0.6 and 25% were absolutely technical efficient.
of rural-urban migration and hence remittance from that of overall This is consistent with the result above that farmers were unable
non-agricultural income, we used Remittance Ratio, the ratio of to allocate inputs so as to minimize costs, rather than unable to
remittances from out-migrant family members to total household use resources with the greatest technical efficiency. Thirty farms
income. The Tenancy Ratio is the rented cultivated land area (26.8%) had reached optimal scale but 82 (73.2%) operated with
divided by the total cultivated land area. Land fragmentation was non-optimal scale.
S Statistics related to the returns-to-scale characteristics are given
measured by the Simpson Index, which is defined as 1– ( sΣ
2
as / 2

=1 s )
A in Table 5 to explore the details of scale inefficiency. Of the 82
where as is the area of the sth plot and A is the farm’s total land area, scale inefficient farms, most (69.6% of the total sample) operated
S at increasing returns-to-scale (sub-optimal scale), implying that
composed of S plots, sΣ
a
=1 . We define a plot as a discrete parcel most farms are too small. Farms operating at decreasing returns-
s

of land physically separated from other land owned by the same to-scale (supra-optimal scale) and constant returns-to-scale
household. This index has a value between 0 and 1. A value of 0 (optimal scale) have higher average values of agricultural output
means that the farm household has only one parcel or plot of land, than those at sub-optimal scale, and achieved, on average, higher
which indicates complete land consolidation, whereas a value efficiency scores than those operating at sub-optimal scale,
close to 1 means the household has numerous plots and the farm suggesting that expanding the size of the farms would improve
is “very fragmented”. Access to Credit is the ratio of the sum of production efficiencies.
loans obtained in the past 8 years to total household assets.
Variables related to variation in farm efficiency: Because
Results and Discussion technical efficiency, allocative efficiency and scale efficiency have
Descriptive statistics of the primary data: Descriptive statistics greater practical implications than cost efficiency and pure technical
of the primary data indicates the socio-economic structure efficiency, we used OLS regression to simulate the relationship
and demographical characteristics of the farms of the study area between the explanatory variables and the technical, allocative
(Tables 1 and 2). As shown in Table 1, the farms are dominated by and scale efficiency scores obtained in the previous section. The
crop production and livestock is basically a sideline product, with results are presented in Table 6. LG_Agri was significantly
average income from crop production of 13,441.231 yuan and that positively related to all the three indices; Remittance Ratio was
from livestock production of 2868.214 yuan. Arable land per significantly positively related to technical efficiency and scale
household is only 9.580 mu with substantial variability. efficiency; and Tenancy Ratio and Simpson Index were
The householders in our sample were mostly middle-aged to significantly negatively related to technical efficiency and scale
elderly, with low education level and had an average agricultural efficiency.
income of about 16,310 yuan per household. Non-agricultural The significant positive relationship between LG_Agri and the
income had become an important income sources, accounting for three indices means that larger farms achieved higher efficiency
37.8% of total household income. In contrast, remittance from scores, which accords with the results in Table 5 that the farms
absent family members amounted to only 1.2% of total household operating at supra-optimal or optimal scale had higher efficiency
income. Land use right transfer emerged, and rented land amounted scores than those operating at sub-optimal scale. This result is
to 9.9% of total cultivated land on average. The Simpson Index consistent with the findings of Coelli and Battese 23 and Karagiannis
indicates severe fragmentation. Access to credit was rather limited. and Sarris 24. The underlying reason was given that smaller farms
The statistics are marked by great variation in the tenancy ratio usually face capital constraints, which restrain them from utilizing
and the non-agricultural income ratio. and allocating resources more efficiently. Our understanding of
this result is that when increasing returns-to-scale prevails or

Journal of Food, Agriculture & Environment, Vol.8 (3&4), July-October 2010 775
776
Table 1. Descriptive statistics of input and output variables used in DEA models to estimate Table 4. Frequency distribution of efficiency scores of
efficiency scores of sample farms on the Loess Plateau, China (n = 112). sample farms on the Loess Plateau, China (n = 112).
Ave. SD Min Max Range TE AE CE PTE SE
Inputs Q Land (mua) 9.580 4.574 0.800 23.500 (0.0-0.1] 0 4 17 0 0
Labor (persons) 1.779 0.488 1.000 4.000 (0.1-0.2] 3 13 29 0 0
Seed (kg) 11.172 7.551 0.910 38.960 (0.2-0.3] 9 15 24 0 3
Fertilizer (kg) 465.329 428.620 24.000 1838.000 (0.3-0.4] 9 22 20 1 4
Fuel (kg) 0.871 1.874 0.000 11.400 (0.4-0.5] 6 32 11 3 6
Young animal (kg) 58.552 107.632 0.000 640.000 (0.5-0.6] 18 18 7 11 9
(0.6-0.7] 8 4 1 20 7
Fodder (kg) 638.259 1486.056 0.000 15000.000
(0.7-0.8] 16 2 1 17 10
P Land (yuanb/mu) 48.000 (0.8-0.9] 8 0 0 11 14
Labor (yuan/person*year) 2645.000 (0.9-1) 7 0 0 5 29
Seed (yuan/kg) 10.000 1 28 2 2 44 30
Fertilizer (yuan/kg) 2.500
Fuel (yuan/kg) 5.882
Young animal (yuan/kg) 5.000
Fodder (yuan/kg) 0.700
Outputs Crop (yuan) 13441.231 13474.756 718.400 69960.000
Livestock (yuan) 2868.214 6104.027 0.000 60000.000
Here Q refers to quantity and P refers to price; a: 1 mu = 0.067 ha; b: 1 yuan = 0.133$ in 2007; and note that we assume farms faced same input prices.

Table 2. Descriptive statistics of explanatory variables expected to affect Table 5. Returns to Scale of sample farms on the Loess Plateau, China (n = 112).
efficiency scores of sample farms on the Loess Plateau, China (n = 112). No. of Average value
RTS
Variables Ave. SD Min Max farms Agricultural output TE AE CE PTE SE
Age (years) 46.821 9.587 22.000 75.000 IRS 78 11687.000 0.570 0.374 0.215 0.758 0.742
Education (years) 4.446 3.683 0.000 11.000 CRS 30 26790.000 0.985 0.421 0.417 0.985 1.000
LG_Agri 4.048 0.401 2.856 4.862 DRS 4 27848.000 0.797 0.451 0.358 0.819 0.974
NonAgri Ratio (percentage) 0.378 0.305 0.000 0.954 Here RTS refers to returns-to-scale; IRS refers to increasing RTS; CRS refers to constant RTS; and DRS refers to decreasing RTS.

Remittance Ratio (percentage) 0.012 0.039 0.000 0.358


Tenancy Ratio (percentage) 0.099 0.205 0.000 1.000
Simpson Index 0.658 0.165 0.000 0.885
Access to Credit (percentage) 0.024 0.104 0.000 0.800

Table 3. Technical, allocative, cost, pure technical and


scale efficiencies of sample farms on the Loess
Plateau, China (n = 112).
TE AE CE PTE SE
Ave. 0.689 0.389 0.274 0.821 0.819
SD 0.266 0.172 0.183 0.178 0.220
Min 0.105 0.076 0.029 0.344 0.210
Max 1.000 1.000 1.000 1.000 1.000
Here TE refers to technical efficiency; AE refers to allocative efficiency; CE refers to cost
efficiency; PTE refers to pure technical efficiency; and SE refers to scale efficiency.

Journal of Food, Agriculture & Environment, Vol.8 (3&4), July-October 2010


Table 6. OLS regression analysis of factors affecting technical efficiency, allocative
efficiency and scale efficiency of sample farms on the Loess Plateau,
China (n = 112).
Variable Technical efficiency Allocative efficiency Scale efficiency
Coefficient Coefficient Coefficient
t-value t-value t-value
Constant í0.780 í0.956 í0.590
í3.130*** í5.923*** í2.928***
Age í0.003 0.001 í0.003
í1.441 0.865 í1.755
Education í0.008 0.003 í0.008
í1.424 0.798 í1.894
LG_Agri 0.497 0.310 0.428
9.178*** 8.850*** 9.801***
NonAgri Ratio 0.003 0.037 í0.056
0.038 0.866 í1.043
Remittance Ratio 1.359 í0.365 0.875
2.834*** í1.176 2.257**
Tenancy Ratio í0.336 0.071 í0.227
í2.988*** 0.972 í2.501**
Simpson Index í0.557 0.023 í0.199
í4.653*** 0.298 í2.062**
Access to Credit 0.119 í0.117 0.030
0.607 í0.920 0.188
Multiple R 0.750 0.748 0.774
R Square (R2) 0.563 0.560 0.599
Adjusted R2 0.529 0.526 0.559
**
means 95% confidence level (p < 0.05);
***
means 99% confidence level (p < 0.01).

economies of scale exists, expanding farm size would be an value indicate more fragmented), was negatively related to
effective way to increase efficiencies of agricultural production. technical efficiency and scale efficiency. The implication of this
The Remittance Ratio was significantly positively related to result is that land fragmentation is significantly detrimental to
technical efficiency and scale efficiency. In contrast, the NonAgri both efficiencies, as also found by others: Chen et al. 22 from
Ratio did not have a significant effect on the efficiencies. These randomly selected samples from a Chinese national survey; Tan
two results suggest that remittances from out-migrant family et al. 27 for farms in Southeast China and Feng 28 from Jiangxi
members help to relax the farm’s capital constraints to permit Province in China. The underlying reason might be that land
investment in efficiency enhancing equipment and technology, fragmentation discourages agricultural mechanism, causes
and in expanding farm size, but non-agricultural income does not. wastage at plot boundaries, increases time spent commuting
Chen et al. 22 gave a regression analysis of village migrant ratio between plots and increases loss of industrial inputs such as
against technical efficiency and found the same result. The fertilizers, pesticides and seeds during transport from one plot to
possible reasons, though, at village or regional level, he assumed another 22, 27.
may include greater chances of voluntary land lease and
consolidation, which might benefit production efficiency 22. Conclusions
Our result about Tenancy Ratio, however, was significantly Our results suggest the existence of substantial inefficiency in
negatively related to technical efficiency and scale efficiency, which agricultural production for the sample farms participating in the
means that renting lands (or land release at farm level) had a Grain for Green Program on the Loess Plateau. The inefficiency
negative effect on both efficiencies. The results of technical was partly explained by insufficient farm size, lack of access to
efficiency are consistent with the work by Olson and Vu 21, Llewelyn capital, insecure land tenure and land fragmentation. Remittances
and Williams 25 and Latruffe et al. 26. Llewelyn and Williams 25 received from out-migrant family members provide an effective
obtained the same result for technical efficiency of crop farms in way to outbreak farms’ capital constraints, and there is an
Indonesia and asserted that farmers utilize better resource increasing trend for surplus labor forces seeking job in the cities
management practices on their own lands, thus achieve higher and towns as a result of the program 6, 7. However, the labor mobility
technical efficiency than on rented land. The negative relationship by rural-urban migration in China is still subject to significant
between Tenancy Ratio and scale efficiency, however, is somewhat institutional constraints due to the different household
unexpected, implying that renting land prohibits the scale-efficient registration system (so called “hukou” in Chinese) between rural
use of resources. The result would seem to arise from the fact that and urban residents and thus different social welfares. The current
most farms renting land had already reached optimal or supra- farmland tenure system is characterized by insecure land right,
optimal scale, so further land renting would actually undermine resulted from both the separation of land property right and use
their scale efficiency. right, the former of which being possessed by the village collective
The Simpson Index, which measures land fragmentation (higher and the latter by the individual farm households and the occasional

Journal of Food, Agriculture & Environment, Vol.8 (3&4), July-October 2010 777
11
administrative reallocations of land by the village collective to Farrell, M. 1957. The measurement of productive efficiency. Journal of
adapt to demographic changes 28. Against such background, the the Royal Statistical Society 120:253-281.
12
land rental market on the Loess Plateau after the Program showed Banker, R. D., Charnes, A. and Cooper, W. W. 1984. Some models for
to discourage land investment and undermine farm efficiency in estimating technical and scale inefficiencies in data envelopment analysis.
Manage. Sci. 30:1078-1092.
agricultural production. The land fragmentation in Loess 13
Bauer, B. W., Berger, A. N. and Ferrier, G. D. 1998. Consistency
Plateau was also exacerbated during the process of egalitarian conditions for regulatory analysis of financial institutions: A comparison
reallocations of the land by the village collective to households of frontier efficiency methods. J. Econ. Bus. 50:85-114.
and parents to married sons and creates numerous farms with a 14
Fare, R., Grosskopf, S. and Lovell, C. A. K. 1994. Production Frontiers.
number of separated small parcels of land 27. However, little Cambridge University Press, Cambridge.
15
attention has been paid to alleviate land fragmentation on the Coelli, T. and Thirtle, C. 2002. Technical, allocative, cost and scale
Loess Plateau, which might be significant to improve farm efficiencies in Bangladesh rice cultivation: A non-parametric approach.
efficiency. J. Agric. Econ. 53(3):607-626.
16
We therefore recommend that effort should be made to expand McDonald, J. 2009. Using least squares and Tobit in second stage DEA
efficiency analyses. Euro. J. Oper. Res. 197(2):792-798.
the farm size, even though practical mechanisms by which to 17
Hoff, A. 2007. Second stage DEA: Comparison of approaches for
achieve this may be difficult to identify. At the very least, it is clear modeling the DEA score. Eur. J. Oper. Res. 181:425-435.
that any further withdrawal of cultivated land from production 18
Banker, R. D. and Natarajan, R. 2008. Evaluating contextual variables
must be performed in a manner that avoids the deterioration of affecting productivity using data envelopment analysis. Oper. Res.
productive efficiencies, and thus agricultural production, on the 56(1):48-58.
19
remaining land. Also policies should be oriented to facilitate labor Amaza, P. S. and Ogundari, K. 2008. An investigation of factors that
mobility, provide more secure land tenure and increase land influence the technical efficiency of soybean production in the Guinea
consolidation. savannas of Nigeria. J. Food Agric. Environ. 6(1):92-96.
20
Alvarez, A. and Arias, C. 2004. Technical efficiency and farm size: A
conditional analysis. Agric. Econ. 30(3):241-250.
Acknowledgements 21
Olson, K. and Vu, L. 2009. Economic efficiency in farm households:
This work was supported by the Japan Society for the Promotion Trends, explanatory factors and estimation methods. Agric. Econ.
of Science (JSPS) Core University Program and the Global Center 40(5):587-599.
of Excellence (GCOE) Program in Japan. We are grateful to Prof. 22
Chen, Z., Huffman, W. and Rozelle, S. 2009. Farm technology and
Guobin Liu of the Institute of Soil and Water Conservation of technical efficiency: Evidence from four regions in China. China Econ.
the Chinese Academy of Sciences, for assistance during the Rev. 20:153-161.
23
investigations and the villagers of the study area for their kind Coelli, T. and Battese, G. 1996. Identification of factors which influence
cooperation. the technical inefficiency of Indian farmers. Aust. J. Agric. Econ.
40(2):103-128.
24
Karagiannis, G. and Sarris, A. 2005. Measuring and explaining scale
References
1 efficiency with the parametric approach: The case of Greek tobacco
Rozelle, S., Huang, J. and Zhang, L. 1997. Poverty, population and
growers. Agric. Econ. 33(supp.):441-451.
environmental degradation in China. Food Policy 22(3):229-252. 25
2 Lewelyn, R. V. and Williams, J. R. 1996. Nonparametric analysis of
Chen, L., Wang, J., Fu, B. and Qiu, Y. 2001. Land-use change in a small
technical, pure technical and scale efficiencies for food crop production
catchment of northern Loess Plateau, China. Agric. Ecosyst. Environ.
in East Java, Indonesia. Agric. Econ. 15:113-126.
86(2):163-172. 26
3 Latruffe, L., Davidova, S. and Balcombe, K. 2008. Application of a
Tan, S., Heerink, N. and Qu, F. 2006. Land fragmentation and its driving
double bootstrap to investigation of determinants of technical
forces in China. Land Use Policy 23(3):272-285.
4 efficiency of farms in Central Europe. J. Prod. Anal. 29:183-191.
Li, L., Tsunekawa, A., Tsubo, M., Koike, A. and Wang, J. 2010. Assessing 27
Tan, S., Heerink, N., Kuyvenhoven, A. and Qu, F. Impact of land
total factor productivity and efficiency change for farms participating
fragmentation on rice producers’ technical efficiency in South-East
in “Grain for Green” Program in China: A case study from Ansai,
China, NJAS. In press.
Loess Plateau. J. Food Agric. Environ. 8:1185-1192. 28
5 Feng, S. 2008. Land rental, off-farm employment and technical efficiency
Belbase, K. and Grabowski, R. 1985. Technical efficiency in Nepalese
of farm households in Jiangxi Province, China. NJAS. 55(4):363-378.
agriculture. J. Dev. Areas. 19:515-525. 29
6 Mochebelele, M. T. and Winter-Nelson, A. 2000. Migrant labor and
Uchida, E., Rozelle, S. and Xu, J. 2009. Conservation payments, liquidity
farm technical efficiency in Lesotho. World Devel. 28:143-153.
constraints and off-farm labor: Impact of the Grain for Green Program 30
Bravo-Ureta, B. E. and Pinheiro A. E. 1993. Efficiency analysis of
on rural households in China. Amer. J. Agr. Econ. 91(1):70-86.
7 developing country agriculture: A review of the frontier function
Xu, J., Tao, R., Xu, Z. and Bennett, M. T. 2010. China’s Sloping Land
literature. Agric. Resource Econ. Rev. 22:88-101.
Conversion Program: Does expansion equal success? Land Econ. 31
Fan, S. and Zhang, X. 2004. Infrastructure and regional economic
86(2):219-244.
8 development in rural China. China Econ. Rev. 15:203-214.
Zhang, S. Q., Swanson, T. and Kontoleon, A. 2005. Impacts of 32
Lund, P. and Price, R. 1998. The measurement of average farm size. J.
Compensation Policies in Reforestation Programs. Report of the
Agric. Econ. 49(I):100-110.
Environment and Poverty Programme to China Council for 33
Rozelle, S., Taylor, J. E. and deBrauw, A. 1999. Migration, remittances
International Cooperation in Environment and Development.
9 and agricultural productivity in China. American Economic Review
Fried, H. O., Lovell, C. A. K. and Schmidt, S. S. 2008. Efficiency and
Papers and Proceedings 89:287-291.
productivity. In Fried, H., Lovell, C. A. K. and Schmidt, S. (eds). The 34
Coelli, T. 1996. A Guide to DEAP Version 2.1: A Data Envelopment
Measurement of Productive Efficiency and Productivity Change.
Analysis Computer Program. Center for Efficiency and Productivity
Oxford University Press, New York.
10 Analysis. Working Paper 96/08. University of New England, Armidale,
Koopmans, T. 1951. Analysis of production as an efficient combination
NSW, Australia.
of activities. In Koopmans, T. (eds). Activity Analysis of Production
and Allocation. John Wiley and Sons. Inc., New York.

778 Journal of Food, Agriculture & Environment, Vol.8 (3&4), July-October 2010

Das könnte Ihnen auch gefallen