Smallholder Participation in Land Rental Market in a Mountainous Region of
Southern China: Impact of Population Aging, Land Tenure Security and Ethnicity
Shi Min1, Hermann Waibel1, Jikun Huang2
1. Institute of Development and Agricultural Economics, Leibniz University Hannover, Germany;
2. Center for Chinese Agricultural Policy, Chinese Academy of Sciences, China
March 15, 2016
2016 WORLD BANK CONFERENCE ON LAND AND POVERTY
The World Bank - Washington DC, March 14-18, 2016
OUTLINE
• BACKGROUND
• MOTIVATION
• OBJECTIVES
• DATA
• DESCRIPTIVE STATISTICS
• EMPIRICAL MODELS
• RESULTS
• SUMMARY AND CONCLUSIONS
[2]
BACKGROUND
Xishuangbanna Dai Autonomous Prefecture (XSBN)
•
Cultural diversity including several indigenous ethnic minorities
– 95% of smallholder rubber farmers are minorities, only 5% are the Han majority
Dai
•
Hani
Lahu
Bulang
Yi
Jinuo
Yao
Land use change: the transition from tropical rainforest to rubber farming (Zhang et al., 2015)
– Unclear land use right in the past  More complex than in other regions
– Land tenure certificate is lagging behind other regions
• XSBN 52.6%, 2012; Other region 70%, 2008 (Huang and Ji, 2012; Deininger et al., 2014)
BACKGROUND MOTIVATION OBJECTIVES DATA DESCRIPTIVE STATISTICS EMPIRICAL MODELS RESULTS SUMMARY AND CONCLUSIONS
[3]
MOTIVATION
Three significant changes motivate to develop land rental markets in Rural China
•
1) Population aging: 13.3%, 2010  17%, 2020  24%, 2030 (NBSC, 2010; Du et al., 2005)
Nation
XSBN
8075-80
70-75
65-70
60-65
55-60
50-55
45-50
40-45
35-40
30-35
25-30
20-25
15-20
10-15
5-10
0-5
0.10
0.08
Proportion
0.06
0.04
0.02
0.00
0.02
0.04
0.06
0.08
0.10
Proportion
Data sources: National Bureau of Statistics of China (2011); Authors’ survey
Fig.1. Demographic structure respectively at national level and in XSBN
BACKGROUND MOTIVATION OBJECTIVES DATA DESCRIPTIVE STATISTICS EMPIRICAL MODELS RESULTS SUMMARY AND CONCLUSIONS
[4]
MOTIVATION
•
2) Urbanization: 53.7%, 2013 (17.7% migrant farm workers)  70%, 2030 (NBSC, 2014;OECD, 2015)
– Mostly younger people migrate
(Wang et al., 2011)
 Farmers are getting older
 Land rental markets are proposed to facilitate redistributing land to younger farmers
•
3) Rural land market reform:
– Promulgation of “Rural Land Contract Law” in 2002
•
Long-term land tenure certificate
– Impacts of land tenure certificate on the development of rural land rental market (Jin and Deininger, 2009)
•
Administratively land reallocation has become more complicated
(Wang et al., 2011)
– Government encourages advancement of rural land rental markets (Wang et al., 2011)
•
Increase farm size, raise efficiency and labor productivity (Huang et al., 2012)
BACKGROUND MOTIVATION OBJECTIVES DATA DESCRIPTIVE STATISTICS EMPIRICAL MODELS RESULTS SUMMARY AND CONCLUSIONS
[5]
RESEARCH OBJECTIVES
1.
To quantify the effect of land tenure security on farmers’ participation in the local
land rental market
2.
To examine the role of population aging for farmers’ participation in the local
land rental market
3.
To test the difference in land rental market participation between ethnic minorities
and the Han majority
BACKGROUND MOTIVATION OBJECTIVES DATA DESCRIPTIVE STATISTICS EMPIRICAL MODELS RESULTS SUMMARY AND CONCLUSIONS
[6]
STUDY AREA AND DATA
Household survey (2013)
• Stratified random sampling
• 612 households, 42 villages,
• 8 townships, 3 counties
• Comprehensive
questionnaire
−
Socioeconomic characteristics of
household members
−
Land use history
−
Natural land conditions
−
Current land tenure status
−
Land productivity
−
Farm and off-farm activities etc.
Fig. 1: The map of study area and sample distribution
BACKGROUND MOTIVATION OBJECTIVES DATA DESCRIPTIVE STATISTICS EMPIRICAL MODELS RESULTS SUMMARY AND CONCLUSIONS
[7]
DESCRIPTIVE STATISTICS
Participation in land rental markets and its association with land tenure certificate,
population aging and ethnicity
Categories
Total sample
Land tenure certificate
Yes
No
Elder in household
Yes
No
Ethnicity
Han
Minority
Data sources: Authors’ survey
Rent out
Households (%)
32
Rent in
Households (%)
4
53
8
4
3
32
31
2
5
36
32
7
4
BACKGROUND MOTIVATION OBJECTIVES DATA DESCRIPTIVE STATISTICS EMPIRICAL MODELS RESULTS SUMMARY AND CONCLUSIONS
[8]
EMPIRICAL MODELS
General model of farmer participation in land rental market (Deininger and Jin, 2005; Huang et al., 2012)
𝑦𝑖1 = 𝛼1 + 𝛽1𝐷𝑖 + 𝛾1𝐶𝑖 + 𝛿1𝐸𝑖 + 𝜃1𝑍𝑖 + 𝜀𝑖
(1)
𝑦𝑖2 = 𝛼2 + 𝛽2𝐷𝑖 + 𝛾2𝐶𝑖 + 𝛿2𝐸𝑖 + 𝜃2𝑍𝑖 + 𝜇𝑖
(2)
where the subscript i represents the ith household.
𝑦𝑖1 is a dummy variable; 𝑦𝑖1 = 1 represents the ith household rented out land
𝑦𝑖2 is a dummy variable; 𝑦𝑖2 = 1 represents the ith household rented in land
𝐷𝑖 Demographic structure, the proportions of family members belonging to different age groups
Population aging, the proportion of family members aged 60 years and above
𝐶𝑖 is a dummy variable; it is equal to 1 if the ith household owned land tenure certificate
𝐸𝑖 ethnicity; 𝐸𝑖 = 1 ethnic minorities i.e. Dai, Hani, Bulang and so on, 𝐸𝑖 = 0 represents the Han majority
𝑍𝑖 is a vector of control variables that might influence renting out land and renting in land
BACKGROUND MOTIVATION OBJECTIVES DATA DESCRIPTIVE STATISTICS EMPIRICAL MODELS RESULTS SUMMARY AND CONCLUSIONS
[9]
EMPIRICAL MODELS
Definitions and statistical descriptions of all variables
Variable
Definition and description
Dependent variables
y1
Rent out land (1=Yes; 0= No)
y2
Rent in land (1=Yes; 0= No)
Independent variables
Hhsize
Household size
Demographic structure
Age16
% of family members (age<16)
Age16-40
% of family members (16≤age<40)
Age40-60
% of family members (40≤age<60)
Age60
% of family members (age≥60)
Certificate
Land tenure certificate (1=Possess; 0= No)
Ethnic
Ethnicity (1=Minority; 0=Han)
Land
Household owned land size (mu/person)
Rubber
Percent of rubber planting area
Altitude
Altitude of household location (MASL)
Remoteness Distance to the center of county(km)
Data sources: Authors’ survey
Mean
Std. Dev.
Min
Max
0.32
0.04
0.47
0.19
0
0
1
1
5.12
1.46
2
11
0.18
0.41
0.30
0.11
0.53
0.95
12.89
0.87
756.11
79.31
0.15
0.15
0.18
0.16
0.50
0.21
12.33
0.16
160.27
46.54
0
0
0
0
0
0
0
0.06
541
25
0.6
1
1
1
1
1
145.8
1
1468
190
BACKGROUND MOTIVATION OBJECTIVES DATA DESCRIPTIVE STATISTICS EMPIRICAL MODELS RESULTS SUMMARY AND CONCLUSIONS
[10]
EMPIRICAL MODELS
Estimation approach
1) Simultaneity  Bivariate probit regression
−
Assume unobserved error terms 𝜀𝑖 and 𝜇𝑖 : standard bivariate normal distributions with unit variance
𝑣𝑎𝑟 𝜀𝑖 = 𝑣𝑎𝑟 𝜇𝑖 = 1and zero mean 𝐸 𝜀𝑖 = 𝐸 𝜇𝑖 = 0
−
The correlation coefficient between 𝜀𝑖 and 𝜇𝑖 : 𝜌 = 𝑐𝑜𝑣(𝜀𝑖 , 𝜇𝑖 )
−
To identify whether unobserved heterogeneities of renting out land and renting in land are correlated
2) Endogeneity of land tenure certificate  Instrumental variable approach
𝐶𝑖 = 𝑎 + 𝑏𝐷𝑖 + 𝑐𝐸𝑖 + 𝑑𝑍𝑖 + ℎ𝐶𝑒𝑟𝑡_𝑣𝑖𝑙𝑙𝑎𝑔𝑒𝑖 + 𝜑𝑖
−
(3)
where 𝐶𝑒𝑟𝑡_𝑣𝑖𝑙𝑙𝑎𝑔𝑒𝑖 is an instrumental variable defined as the proportion of households owning
land tenure certificate in the village
−
Validity test: IV is significantly correlated with 𝐶𝑖 , but insignificantly correlated with 𝑦𝑖1 and 𝑦𝑖2
when 𝐶𝑖 is equal to 0 (Di Falco et al., 2011; Ayuya et al., 2015; Huang et al., 2015; Parvathi and Waibel, 2016)
BACKGROUND MOTIVATION OBJECTIVES DATA DESCRIPTIVE STATISTICS EMPIRICAL MODELS RESULTS SUMMARY AND CONCLUSIONS
[11]
EMPIRICAL MODELS
3) Selection bias of land tenure certificate  Endogenous switching probit model
(Lokshin and Glinskaya, 2009; Gregory and Coleman-Jensen, 2013; Ayuya et al., 2015)
𝐶𝑖 = 1 if 𝑎 + 𝑏𝐷𝑖 + 𝑐𝐸𝑖 + 𝑑𝑍𝑖 + ℎ𝐶𝑒𝑟𝑡_𝑣𝑖𝑙𝑙𝑎𝑔𝑒𝑖 + 𝜑𝑖 > 0
(4a)
𝐶𝑖 = 0 if 𝑎 + 𝑏𝐷𝑖 + 𝑐𝐸𝑖 + 𝑑𝑍𝑖 + ℎ𝐶𝑒𝑟𝑡_𝑣𝑖𝑙𝑙𝑎𝑔𝑒𝑖 + 𝜑𝑖 ≤ 0
(4b)
∗
𝑦1𝑖𝑗
= 𝛼1𝑗 + 𝛽1𝑗 𝐷1𝑖 + 𝛿1𝑗 𝐸1𝑖 + 𝜃1𝑗 𝑍1𝑖 + 𝜀1𝑖
∗
𝑦1𝑖𝑗 = 𝐼 (𝑦1𝑖𝑗
> 0)
(5a)
∗
𝑦0𝑖𝑗
= 𝛼0𝑗 + 𝛽0𝑗 𝐷0𝑖 + 𝛿0𝑗 𝐸0𝑖 + 𝜃0𝑗 𝑍0𝑖 + 𝜀0𝑖
∗
𝑦0𝑖𝑗 = 𝐼 (𝑦0𝑖𝑗
> 0)
(5b)
where j is equal to 1 or 2 , renting out land (j=1) and renting in land (j=2)
∗
∗
𝑦1𝑖𝑗
and 𝑦0𝑖𝑗
are latent variables that determine the observed behaviors 𝑦1𝑗 and 𝑦0𝑗
Observed 𝑦𝑖𝑗 is defined as 𝑦𝑖𝑗 = 𝑦1𝑗 if 𝐶𝑖 = 1 and 𝑦𝑖𝑗 = 𝑦0𝑗 if 𝐶𝑖 = 0
Significance of the correlations (𝝆) between error terms can reflect whether there is selection bias
Counterfactual analysis: Treatment effect on the treated (TT), Treatment effect on the untreated (TU)
𝑇𝑇𝑗 = 𝑃𝑟(𝑦1𝑗 = 1|𝐶 = 1) − 𝑃𝑟(𝑦0𝑗 = 1|𝐶 = 1)
(6)
𝑇𝑈𝑗 = 𝑃𝑟(𝑦1𝑗 = 1|𝐶 = 0) − 𝑃𝑟(𝑦0𝑗 = 1|𝐶 = 0)
(7)
BACKGROUND MOTIVATION OBJECTIVES DATA DESCRIPTIVE STATISTICS EMPIRICAL MODELS RESULTS SUMMARY AND CONCLUSIONS
[12]
RESULTS
Results of bivariate probit regression (Simultaneity)
•
The correlation coefficient ρ= 0.107 is insignificantly different from zero
Validity test of instrumental variable (Endogeneity)
Variables
Cert_village
Constant
Number of
observations
Wald chi2
Log pseudo
likelihood
Pseudo R2
Land tenure certificate
(Probit)
Coef.
R. Std. Err.
3.418 ***
0.216
-1.716 ***
0.126
Rent out (Certificate=0)
(Probit)
Coef.
R. Std. Err.
-0.850
0.563
-1.165 ***
0.180
Rent in (Certificate=0)
(Probit)
Coef.
R. Std. Err.
0.409
0.627
1.743 ***
0.231
612
290
290
250.73***
2.280
0.42
-284.688
-78.336
-39.874
0.328
0.025
0.006
Notes: *,**, and *** indicate significance at the 1%,5%, and 10% level, respectively
BACKGROUND MOTIVATION OBJECTIVES DATA DESCRIPTIVE STATISTICS EMPIRICAL MODELS RESULTS SUMMARY AND CONCLUSIONS
[13]
RESULTS
Probit regression with a discrete endogenous regressor (Two-step)
Variables
First step (land tenure certificate)
R. Std.
Coef.
Err.
0.006
0.046
0.027
0.563
-0.456
0.487
-0.326
0.499
Second step (Rent out)
R. Std.
Coef.
Err.
-0.057
0.041
0.039
0.521
0.684
0.429
0.892 **
0.450
1.800 ***
0.202
-0.590 **
0.271
-0.002
0.005
0.756 *
0.441
-0.001 *
0.001
-0.007 ***
0.002
Hhsize
Age16
Age40-60
Age60
𝐶𝑖
Ethnic
0.056
0.254
Land
0.011 **
0.005
Rubber
-1.587 ***
0.476
Altitude
-0.001 **
0.0005
Remoteness
-0.0004
0.002
Cert_village
3.401 ***
0.220
Constant
0.401
0.754
-0.396
0.774
Number of observations
612
612
Wald chi2
249.60***
99.31***
Pseudo R2
0.3496
0.1676
Notes: *.**.and *** indicate significance at the 1%.5%.and 10% level. respectively
Second step (Rent in)
R. Std.
Coef.
Err.
-0.070
0.070
-0.293
0.954
-1.067
0.700
-1.970 **
0.890
-0.655 **
0.316
-0.133
0.378
0.001
0.009
-1.326 **
0.601
0.000
0.001
-0.002
0.002
1.037
1.042
612
20.11**
0.0836
BACKGROUND MOTIVATION OBJECTIVES DATA DESCRIPTIVE STATISTICS EMPIRICAL MODELS RESULTS SUMMARY AND CONCLUSIONS
[14]
RESULTS
Endogenous switching probit regression (Rent out land, Selection bias)
Variables
Land tenure certificate
Coef.
R. Std. Err.
0.004
0.046
0.002
0.557
-0.500
0.482
-0.387
0.503
0.053
0.254
0.011 **
0.005
-1.601 ***
0.488
-0.001 **
0.0005
-0.0005
0.002
3.359 ***
0.215
0.473
0.786
Rent out (Certificate=1)
Coef.
R. Std. Err.
-0.034
0.056
0.820
0.654
1.206 **
0.566
1.313 **
0.649
-0.818 **
0.389
-0.014 *
0.008
1.110 *
0.588
-0.002 **
0.001
-0.011 ***
0.002
Rent out (Certificate=0)
Coef.
R. Std. Err.
-0.086
0.074
-1.244
1.210
-0.038
0.695
0.268
0.786
-0.185
0.528
0.011
0.007
-0.672
0.812
-0.002 ***
0.001
0.005 **
0.002
Hhsize
Age16
Age40-60
Age60
Ethnic
Land
Rubber
Altitude
Remoteness
Cert_village
Constant
1.865 *
1.119
1.243
𝜌11 / 𝜌01
-0.348 **
0.161
0.326
Number of observations
612
Wald chi2 (Joint significance)
257.04***
Wald chi2 (Wald test of independent eqns.)
5.34*
Notes: *,**, and *** indicate significance at the 1%,5%, and 10% level, respectively
1.411
0.249
BACKGROUND MOTIVATION OBJECTIVES DATA DESCRIPTIVE STATISTICS EMPIRICAL MODELS RESULTS SUMMARY AND CONCLUSIONS
[15]
RESULTS
Endogenous switching probit regression (Rent in land, Selection bias)
Variables
Land tenure certificate
Coef.
R. Std. Err.
0.011
0.046
-0.029
0.554
-0.382
0.474
-0.296
0.490
Hhsize
Age16
Age40-60
Age60
Ethnic#
Land
0.012 **
0.005
Rubber
-1.631 ***
0.483
Altitude
-0.001 **
0.000
Remoteness
0.000
0.002
Cert_village
3.427 ***
0.221
Constant
0.466
0.746
𝜌12 / 𝜌02
Number of observations
Wald chi2 (Joint significance)
Wald chi2 (Wald test of independent eqns.)
Rent in (Certificate=1)
Coef.
R. Std. Err.
-0.139 *
0.077
0.456
1.116
-0.877
0.764
-1.199
0.936
0.010
-1.653 **
-0.001
-0.003
1.300
0.908 ***
Rent in (Certificate=0)
Coef.
R. Std. Err.
0.020
0.120
-1.623
1.423
-1.398
1.183
-3.107 **
1.548
0.014
0.697
0.001
0.003
-0.042
-0.719
0.002 **
-0.006
0.026
0.932
0.001
0.005
1.281
0.194
-0.736
0.372
1.533
0.314
612
251.48***
3.12
Notes: *,**, and *** indicate significance at the 1%,5%, and 10% level, respectively;
BACKGROUND MOTIVATION OBJECTIVES DATA DESCRIPTIVE STATISTICS EMPIRICAL MODELS RESULTS SUMMARY AND CONCLUSIONS
[16]
RESULTS
Treatment effects of land tenure certificate
Categories
Mean
Observations
Rent out
Rent in
ATT
322
0.393 ***
-0.026 ***
ATU
290
0.637 ***
-0.029 ***
Data sources: Authors’ calculations
•
Households possessing a land tenure certificate have a 39.3% higher probability of renting out land,
and a 2.6% lower probability of renting in land
•
If farmers would possess a land tenure certificate this would increase the likelihood of renting out
land by 63.7% , and decrease the likelihood of renting in land by 2.9%
 Improving land tenure security can facilitate rural land transactions in land rental markets
BACKGROUND MOTIVATION OBJECTIVES DATA DESCRIPTIVE STATISTICS EMPIRICAL MODELS RESULTS SUMMARY AND CONCLUSIONS
[17]
SUMMARY
•
Much more lands are rented out rather than rented in by smallholder rubber farmers in
XSBN
•
Population aging can foster the advancements of rural land rental market by transferring
land from older to younger farmers
•
The availability of a land tenure certificate increases farmers’ participation in land rental
market by improving the land tenure security
•
Participation in land rental market is sensitive to ethnicity, i.e. ethnic minority groups are
significantly less likely to rent out land
•
Altitude and remoteness negatively impact on renting out land
BACKGROUND MOTIVATION OBJECTIVES DATA DESCRIPTIVE STATISTICS EMPIRICAL MODELS RESULTS SUMMARY AND CONCLUSIONS
[18]
CONCLUSIONS AND RECOMMENDATION
•
Implementation of effective land rental markets in a remote mountainous and ethnic
minority region of Southern China is more difficult and take more time than in other
agricultural areas of China
•
To facilitate the advancements of rural land rental markets in XSBN, we recommend that
government agencies:
– 1) more effectively implement the issuance of land tenure certificates, and
– 2) give higher priority to ethnic minority groups and farmers located in remote
mountainous area
BACKGROUND MOTIVATION OBJECTIVES DATA DESCRIPTIVE STATISTICS EMPIRICAL MODELS RESULTS SUMMARY AND CONCLUSIONS
[19]
Thank you for your attention !
谢谢！
[20]