Land rental market,off-farm employment and agricultural production in Southeast China:A plot-level case study
Shuyi FENG a,⁎,Nico HEERINK a,b ,Ruerd RUBEN c ,Futian QU a
a College of Public Administration,Nanjing Agricultural University,Nanjing 210095,PR China
b Development Economics Group,Wageningen University,The Netherlands
c
Centre for International Development Issues,Radboud University,Nijmegen,The Netherlands
a r t i c l e i n f o a
b s t r a
c t
Article history:
Received 4April 2009
Received in revised form 31May 2010Accepted 15June 2010This paper performs a plot-level analysis of the impact of land rental market participation and off-farm employment on land investment,input use,and rice yields for 215plots cultivated by 52households in three villages in Northeast Jiangxi Province.Our findings show that households that rent extra land are relatively more productive,but contradict results of earlier studies which found that tenure status of plots affects the level of land investments.We further find that off-farm employment does not signi ficantly affect rice yields.This result contradicts those of earlier studies which found that the negative lost-labor effect of off-farm employment dominates the positive income effect.Another novel finding is that people working locally off-farm tend to switch from green manure planting towards the use of organic manure on their rice plots.We conclude that policies that will further stimulate the development of land rental markets,which is still in its infancy,can contribute signi ficantly to higher rice production in Southeast China.Another implication of our results is that worries about the negative impact that the continuously growing off-farm employment may have on China's goal to remain self-suf ficient in grain production are less relevant at the moment for the region examined in our study.
©2010Elsevier Inc.All rights reserved.
JEL classi fications:J22Q12Q15
Keywords:
Land rental market Land tenure contracts Off-farm employment Rural households China
1.Introduction
Economic reforms implemented since the end of the 1970s have stimulated the development of land and labor markets in rural China.Recent studies show an increasing incidence of land rental activities (Deininger &Jin,2005;Kung,2002),while off-farm employment has become a signi ficant phenomenon since the mid-1980s.By 2000,more than 200million rural laborers worked off-farm (de Brauw,Huang,Rozelle,Zhang,&Zhang,2002;Zhang,Huang,&Rozelle,2002).As China's economy continues to grow,the development of land and labor markets is expected to continue or even accelerate.These developments may have important consequences for agricultural production.
Land rental markets can enhance allocative ef ficiency and agricultural productivity by equalizing the marginal product of land across households with different land –labor endowments and by facilitating transfers of land from less productive households to more productive ones (Carter &Yao,2002;Deininger,2003;Jin &Deininger,2009).However,in present-day China land rental arrangements are generally informal,short term,and between households living in the same village.R
ented plots are therefore subject to tenure insecurity,which may discourage land investment and hamper agricultural productivity increases.
The effect of off-farm employment on agricultural production is ambiguous.Off-farm employment reduces the labor available for agricultural production,especially if hiring agricultural labor to replace family labor incurs transaction costs and if hired labor is not as ef ficient as family labor.But off-farm employment also enables households to increase their incomes,to overcome credit
China Economic Review 21(2010)598–606
⁎Corresponding author.Tel.:+862584396009;fax:+862584396531.E-mail address:shuyifeng@njau.edu (S.
Feng).1043-951X/$–see front matter ©2010Elsevier Inc.All rights reserved.doi:
10.1016/j.chieco.2010.06.002
Contents lists available at ScienceDirect
China Economic Review
and insurance constraints and to increase their investment in agricultural production (Rozelle,Taylor,&de Brauw,1999;Taylor,Rozelle,&de Brauw,2003).In addition,the reduction in food consumption by household members working off-farm (e.g.those who migrate)may have an impact on agricultural production decisions if household production and consumption decisions are non-separable (Burger,1994).
Few studies provide empirical estimates of the effect of land rental market development on allocative ef ficiency and agricultural productivity in rural China.Carter and Yao (2002)find for a panel data set of 339households,surveyed in 1988,1993and 1997in five counties in Zhejiang and Jiangxi Provinces,that allocative ef ficiency is achieved by households that rent out land,but surprisingly not by households that rent in land.Lohmar,Zhang,and Somwaru (2001)find for a sample of 830households,surveyed in 1998in 6provinces,that allocative ef ficiency and aggregate agricultural production increase,because the households that rent in land have a signi ficantly higher marginal product of land than households that do not rent additional land.Jin and Deininger (2009)examine the impact of land rentals using a panel of almost 8,000households in about 800villages that are representative of China's nine agriculturally most important provinces for the period 2001–2004.Their study finds that,by transferring land from less able and more af fluent households who joined the non-farm sector to poorer ones with ample family labor,land markets allow more effective use of potentially idle land that can contribute to signi ficant agricultural productivity gains.
Related research on land tenure in rural China focuses on the land tenure insecurity resulting from frequent land reallocations,and the impact of this insecurity on household investment (as measured by application of green manure or organic manure)and agricultural productivity (Benjamin &Brandt,2
002;Jacoby,Li,&Rozelle,2002;Li,Rozelle,&Brandt,1998).Rural land in China belongs to the village collectives,with farm households having user rights for a fixed contract period (at the moment 30years for most cultivated land).Land is generally allocated to farm households on the basis of their household (and/or labor force)size,but may be reallocated to correct for demographic changes (Tan,Heerink,&Qu,2006).Most studies investigating the insecurity caused by reallocations find that it has a signi ficant but small negative effect on investment ( manure,organic manure),but no signi ficant effect on productivity.The main explanation is that land investment plays a minor role in agricultural production compared with other agricultural inputs such as land,labor,and chemical fertilizers (Yao,2007).
Many empirical studies have investigated the effect of off-farm employment on agricultural production in rural China.Recent studies in this field apply the “new economics of labor migration ”(NELM)framework developed by Stark and Bloom (1985),in which the migration decision is part of a set of interwoven economic choices made by households facing imperfect markets.Rozelle et al.(1999)and Taylor et al.(2003)estimate a simultaneous-equation model from data collected among 787farm households from 31villages in Hebei and Liaoning Provinces in 1995,and find the positive income effect of migrant remittances nearly offsets the negative lost-labor effect of migration on crop
production.On the other hand,using a village computable general equilibrium model that takes into account existing factor market imperfections,Shi et al.(in press)find for a remote village in Jiangxi Province that the negative lost-labor effect of off-farm employment on agricultural incomes is much stronger than the (small)positive income effect.None of these studies,however,pays attention to the reduced food consumption effect identi fied by Burger (1994).
To our knowledge no studies have analyzed until now the joint impact of land rental market development and off-farm employment on agricultural production in China.Rural China is characterized by surplus and underemployed rural labor,while land rental markets are rather thin (Brandt,Huang,Li,&Rozelle,2002;Brandt,Rozelle,&Turner,2004).Rural households facing such land and labor market imperfections are likely not to make decisions on land and labor market participation in isolation,but to decide on them simultaneously.The purpose of this paper is therefore to analyze the joint effects of land rental market participation and off-farm employment on agricultural production in rural China.We will investigate the impact on land investment,input use,and agricultural productivity separately,as the land renting and off-farm employment may affect each aspect differently.The remainder of this paper is structured as follows.Section 2presents the theoretical framework of our analysis.Section 3introduces the estimation procedure.In Section 4,estimation resul
ts are presented.The paper concludes with a summary of its main findings and with some policy implications in Section 5.2.Conceptual framework
In this paper we will use plot-level data to examine the impacts of land renting and off-farm employment on agricultural production.The model employed is based on the structural models developed by Feder,Onchan,Chalamwong,and Hongladarom (1988),Place and Hazell (1993)and Hayes,Roth,and Zepeda (1997)to examine the impact of tenure security on investments,input use and productivity in agriculture.Assuming that households decide on land renting and off-farm employment before making decisions on land management and investment 1,the reduced-form equations can be written as follows:
LI i =LI i A in ;l o ;TS i ;Z p i ;Z q ;Z h
;L ;A ;r ;w ;p X  ð1Þl a i =l a i A in ;l o ;TS i ;Z p i ;Z q ;Z h
;L ;A ;r ;w ;p X  ð2ÞX i =X i A in ;l o ;TS i ;Z p i ;Z q ;Z h
;L ;A ;r ;w ;p X  ð3ÞQ i =Q i A in ;l o ;TS i ;Z p i ;Z q ;Z h
;L ;A ;r ;w ;p X
ð4Þ
1
This assumption will (partly)be relaxed in our empirical analysis.
599
S.Feng et al./China Economic Review 21(2010)598–606
600S.Feng et al./China Economic Review21(2010)598–606
where LI,l a,X,and Q are land investment,labor use,use of non-labor inputs,and yield on each plot(subscript i),respectively.The explanatory variables A in and l o are household decisions on whether to rent additional land and participate in off-farm employment,respectively.TS is the plot-specific tenure security indicator,Z p are other plot characteristics,Z q are farm characteristics,Z h are household preferences,L is the household labor endowment,A is the land endowment,r is the market land rent,w is the market labor wage,and p X is the price for non-labor inputs.By estimating Eqs.(1)–(4),the direct plus indirect)effects of land renting(A in),the resulting tenure security of plots(TS),and off-farm employment(l o)on land investment, labor use,non-labor inputs,and land produ
ctivity can be assessed.
3.Model specification and estimation
This paper uses data from a farm household survey and a plot-level survey that were held in three villages in Northeast Jiangxi Province,Southeast China.The villages were selected using a series of criteria including economic development level,market access and geographical conditions.Local researchers and policy makers were consulted and several site visits were made as part of this process.The three villages are considered representative of the diversity of rural conditions that can be found in Northeast Jiangxi Province and in the much larger hilly area of Southeast China with rice-based production systems(Kuiper,Heerink,Tan,Ren,&Shi, 2001).The three villages selected are Banqiao in Yujiang County,Shangzhu in Guixi City and Gangyan in Yanshan County2.
The farm household survey was carried out in2000and the beginning of2001.The questions in the survey referred to the entire year2000.In each village,23%of the households were interviewed.A stratified random sample was used for selecting the households,with the hamlets within each village forming the strata(Kuiper et al.,2001).The information collected includes demographic characteristics,assets,land tenure,and participation in factor markets.Data on household migration an
d local off-farm employment decisions collected during that survey are used for the empirical analysis below.
Plot-level data,however,were not collected then.Out of the329households interviewed in2000and2001,52households were randomly selected,and plot-level agricultural production data were collected in January2003for the entire year2002.Of the52 households in the three villages,19households rented in irrigated land in2002,while33households did not(and are therefore considered ‘self-sufficient in land’).Plots cultivated with rice,the most important crop in the research area,are included in the sample.In total215rice plots were surveyed;172were contracted directly from village collectives 20%of the plots)were rented from other farm households.Contracted plots occur in both self-sufficient households and households that rent in,while rented plots are restricted only to households that rent in.Collected information includes input use and output of each plot,plot characteristics,and soil quality.
3.1.Model specification
The conceptual framework in Section2suggests the variables that potentially affect agricultural production decisions.An overview of the dependent and explanatory variables included in the analysi
s,subdivided by household land market regimes(land renting in vs.self-sufficient in land)and tenure status of the plot(d),is presented in Table1.
Land investment is represented by green manure planting(LI gm)and organic manure application(LI om)3;long-term land investments,such as wells,fences or fruit trees,are virtually absent on the rice plots examined in our study.The land investment variables are both dummy variables which equal one if the household invests on the plot.As can be seen from Table1,green manure is planted on33%of the plots,and organic manure is applied on46%of the plots in the sample.
Input use includes labor(l a)and chemical fertilizers(X cf)4.These inputs are expressed per unit area to correct for differences in plot size.Labor is measured in man days.There arefive different commonly used types of chemical fertilizers,which are aggregated and measured in value terms.Households work on average around41days and use47yuan of chemical fertilizers on each mu of land(see Table1).
Households grow either a single or a double rice crop in the surveyed area.Rice yields(Q)vary for different varieties and are aggregated and measured in values per unit area.As can be seen from Table1,the average rice yield is297.73yuan per mu5.
Explanatory variables in the analysis include household decisions in terms of land renting in(A in)and off-farm employment (l o),indicators of plot tenure security(TS),plot characteristics(Z p),farm characteristics(Z q),household characteristics(Z h), household land and labor endowments(A and L),market land rent(r),market wage rate(w),and non-labor input prices(p X).
Households that rent land(A in)are expected to have a higher marginal product of land,and therefore a higher agricultural productivity,than those that do not.The impact of off-farm employment(l o)on agricultural investment,input use and production is ambiguous,as explained in Section1.If the income effect dominates the sum of the lost-labor and reduced food consumption effects,the net impact will be positive;if the latter two effects are stronger,it will be negative.In our analysis we distinguish between local off-farm employment(l ol)and migration(l om)6.Local off-farm employment may only have an income effect,
2See Feng and Heerink(2008)for a detailed introduction of the three villages.
3The research on the effect of tenure security on land investment focuses on the use of green manure planting and organic manure application as the main forms of land investment in rural China(Jacoby et al.,2002;Li et al.,1998).
4Data are also available on seed,herbicides and pesticides and animal traction.These inputs are not included in the analysis for simplicity,because they are relatively minor inputs.
5The calculation of the rice yield is based on prices for the year2002.Since the autumn of2003,rice prices have increased rapidly.
6Migrants are family members working off-farm and not living together with other household members.Households participate in migration if at least one household member works as a migrant.They participate in local off-farm employment if they have at least one household member working off-farm but no household members working as migrants.
because household members that are involved in local off-farm employment normally live and consume at home and are often able to combine local off-farm employment with on-farm agricultural production.For migration,on the other hand,all three effects may be relevant.
Tenure security (TS )is represented by the tenure status of the plot.Plots are either contracted from the village collectives (for a 30-year period)or rented from other households.A dummy variable,which equals one if the plot is rented,is used to indicate the tenure status.The land rental arrangements are normally verbal and of short duration.Rented plots are therefore less secure than contracted plots an
d are expected to receive less land investment,and thus to produce less output.On the other hand,the use of variable inputs is expected to be higher on rented plots because households renting additional land tend to maximize short-term agricultural pro fits on these plots.Thus the net effect of tenure security on yield is ambiguous.Following Pender and Fafchamps (2001),the interaction between the household land renting variable and the rented plot dummy (A in *TS )is included in the model.In this way it is possible to examine whether rented plots are managed in a signi ficantly different way compared to non-rented plots,given that households that rent additional land may have a higher productivity than ‘self-suf ficient ’households.
The control variables in model (1)–(4)consist of plot characteristics (Z p ),farm characteristics (Z q ),household characteristics (Z h ),household time and land endowments (L ,A )and market land rent,wage rate and non-labor input prices (r ,w ,p X ).Plot characteristics comprise the following indicators:
-Soil fertility (Z pf ).Households in the survey were asked their perceptions about soil quality,which were given a value of one if households perceived soil fertility as poor,two if households perceived soil fertility as average,and three if households perceived soil fertility as good.It is expected that land productivity is higher on plots with good soil fertility.Land investment and input use may be higher on
fertile plots,if the marginal returns of land investment and input use are higher than on plots with poor soil fertility.They are expected to be lower,if the marginal returns are lower.
Table 1
Descriptive statistics of the variables used in the plot-level analysis.Source:Farm household survey.Item
Symbols
Unit
Renting in household (19)Self-suf ficient household (33)All plots
Contracted
Rented
Contracted Number of sample plots
6043
112
215Mean (standard deviation)
Dependent variables Land investment Green manure LI gm 0or 10.32(0.47)0.19(0.39)0.39(0.49)0.33(0.47)Organic manure LI om 0or 10.50(0.50)0.37(0.49)0.47(0.50)0.46(0.50)Input use Labor
l a Man day/mu b 37.6(20.0)37.7(36.0)43.4(19.6)40.6(23.9)Chemical fertilizers X cf Yuan c /mu 54.0(30.9)54.9(13.5)40.3(21.7)47.0(24.4)Rice yield
Q
Yuan /mu
327(79.8)
355(81.4)
260(79.1)
298(89.3)
Independent variables
Household factor allocation Renting land a
A in
0.72(0.26)0.76(0.24)0.14(0.20)0.43(0.37)Renting land a ×rented plot interaction A in *TS 0.00(0.00)d
0.76(0.24)d
0.00(0.00)0.15(0.32)Participating in migration
l om 0or 10.53(0.50)0.40(0.49)0.54(0.50)0.51(0.50)Participating in local off-farm employment l ol 0or 10.20(0.40)0.12(0.32)0.29(0.45)0.23(0.42)Plot characteristics Rented plot TS 0or 10.00(0.00)  1.00(0.00)0.00(0.00)0.20(0.40)Fertility
Z pf    1.93(0.84)  2.09(0.84)  1.93(0.80)  1.96(0.82)Topsoil depth Z ptd cm 16.7(4.54)17.8(3.86)16.4(3.74)16.8(4.02)Plot size
Z pps Mu    1.83(1.17)  2.46(2.59)  1.49(1.23)  1.78(1.62)Distance from home Z pd Minute
11.8(8.22)d 17.9(13.6)d
11.2(10.3)12.7(10.8)Farm characteristics
Total number of cattle Z qc    1.17(1.76)  2.16(3.22)0.67(0.47)  1.11(1.82)Age of household head
Z qa Years 51.2(13.1)46.8(13.3)45.5(9.81)47.3(11.7)Education of household head Z qe Years
3.78(2.49)land
4.37(2.61)
5.24(2.86)  4.66(2.78)Female to male adult ratio Z qfm    1.12(0.67)  1.14(0.57)  1.09(0.69)  1.11(0.66)Number of plots
Z qp    4.33
(1.74)
4.30
(1.32)
4.98
(1.96)
4.67
(1.81)
Household characteristics Household size
Z hhs Persons    5.60(1.44)  5.40(1.43)  4.79(1.91)  5.13(1.73)Number of dependents Z hd Persons
1.98(1.32)
2.26(1.42)  1.39(1.02)  1.73(1.24)Number of durable assets
Z hda 7.08(1.41)7.30(1.32)  6.46(1.67)  6.80(1.57)Household land and labor endowment Irrigated land per adult IA Mu    2.39(1.65)d
3.25(2.10)d
2.06(1.38)  2.39(1.68)Village dummies Banqiao Dummy BQ 0or 10.22(0.42)0.12(0.32)0.34(0.48)0.26(0.44)Shangzhu Dummy
SZ
0or 1
0.15(0.36)
0.07(0.26)
0.55(0.50)
0.34(0.48)
Note:a :Household renting in decisions are predicted probabilities derived from a probit model.b :1mu=1/15ha.c :1USD =6.83yuan (exchange rate on October 27,2009).d :Difference in means within renting in household group is statistically signi ficant for results highlighted in bold.
601
S.Feng et al./China Economic Review 21(2010)598–606
-Topsoil depth (Z ptd )is estimated by soil scientists,and measured in centimeters.The expected impact on yields,land investments and input use are similar to those for soil fertility.
-Plot size (Z pps ).Large plots are easier to manage and have higher input use ef ficiency (economies of scale).They are therefore expected to receive fewer inputs,have higher yields,or both.
-Distance from home (Z pd ).The distance between the homestead and each plot is measured in minutes travel time.Longer travel time raises the cost of carrying organic manure and other inputs to the plots.Planting green manure on distant plots is sometimes risky because of high supervision costs involved in keeping out wild animals.A larger distance from home is thus expected to reduce land investment,input use and rice yield.The square of this variable is added to the equation to capture possible nonlinearities in its impact.
Farm characteristics included in the model comprise:
-Number of cattle (Z qc ).Cattle are very important draft animals for small-scale households in rural China.Their use is expected to have a positive impact on land productivity.Moreover,organic manure
cannot be exchanged in the market in the research area.Therefore a larger number of cattle in a household are expected to increase land investment,and thereby also raise land productivity.
-Age of the household head (Z qa ).Older farmers have more experience and are therefore expected to be more productive in agriculture.In addition,older farmers like to stick to farming traditions,so green manure planting and organic manure application may be higher while chemical fertilizer application may be lower.
-Education of the household head (Z qe ).Household heads with more schooling (in years)are expected to have more farming skills and therefore to be more productive in agriculture.Well-educated farmers may also be more aware of the potential bene fits of land investment.
-Female to male adult ratio (Z qfm )is used to test for differences between females and males in physical strength or other factors in fluencing agricultural production.Transporting organic manure and cutting the roots of green manure for land preparation requires much physical strength.It is therefore expected that a higher female –male adult ratio leads to less land investment and more chemical fertilizer use.
-Number of plots of the household (Z qp )is an indicator of land fragmentation.It can have either nega
tive or positive effects on agricultural production (Tan,Heerink,Kruseman,&Qu,2008).On the one hand,a larger number of plots need more labor to manage (Nguyen,Chen,&Findlay,1996).On the other hand,it enables the households to diversify agricultural production and reduce risk (Bentley,1987),and to optimize their labor allocation over different crop varieties and seasons,especially when there is no market for agricultural labor (Fenoaltea,1976).
Household characteristics have a direct effect on household consumption preferences,and can have either positive or negative effects on the demand for leisure and consumption goods.If household decisions are non-separable,this will affect production decisions as well.The household characteristics included in our model are:
-Household size (Z hhs ).A larger household needs more food than a small one,so the household size is expected to have a positive impact on land investments,input use and rice yields.
-Number of dependents in a household (Z hd ).A household with more dependents needs less food than a household with many adults.So,the impact of this variable is expected to be opposite to that of household size.
-Female to male adult ratio (Z qfm )does not only have a direct impact on production decisions (see
above),but is also likely to affect consumption decisions,because female adults tend to consume less food than male adults.Hence,the impact of this ratio is expected to be similar to that of the number of dependents.
-Number of durable assets in a household (Z hda ).Wealthier households tend to consume more food,hence the number of durable assets is expected to have a positive impact on land investments,input use and rice yields.
The household time endowment equals household size (Z hhs )minus the number of dependents (Z hd ).In addition,household time endowment may depend on the ratio of female to male adults (Z qfm ),as taking care of children and doing housework is usually a female task in Chinese society.If household decisions are non-separable,land investment,labor use,and rice yield are expected to be positively related to the household labor endowment.
The household land endowment is represented by the total size of contracted irrigated land per adult (IA ).If household decisions are non-separable,land investment,labor use and rice yield are expected to be negatively related to the household land endowment.
The market land rent (r ),wage rate (w ),and non-labor input prices (p X )are assumed to be the same
for all households living in the same village.They are therefore captured by village dummy variables for Banqiao (BQ )and Shangzhu (SZ ).
The resulting model that is used for estimation can be speci fied as follows:
DV i =α0+α1A in
+α2A in
*TS i +α3l
om
+α4l
ol
+α5ln Z pf i  +α6ln Z ptd i  +α7ln Z pps i ÀÁ+α8ln Z pd i  +α9ln Z pd
i    2+α10ln Z
qc À
Á
+α11ln Z qa ÀÁ+α12ln Z qe ÀÁ+α13ln Z
qfm
+α14ln Z qp ÀÁ602S.Feng et al./China Economic Review 21(2010)598–606

版权声明：本站内容均来自互联网，仅供演示用，请勿用于商业和其他非法用途。如果侵犯了您的权益请与我们联系QQ:729038198，我们将在24小时内删除。