Allocative Efficiency or Agglomeration

Allocative Efficiency or Agglomeration:
Devolution of Household Forestland Management and Rental Markets
in China
Yuanyuan Yi
University of Gothenburg, Sweden
The 2017 World Bank Conference on Land and Poverty
March 22, 2017
Yuanyuan Yi
Allocative Efficiency & Forestland Markets
WB Land Conference 2017
1/12
Background
Land reforms: change political arrangement, reallocate factors for the
poor
China: Devolution in agricultural land, in 1980s, great success
The Devolution Reform in collective forest areas (60% of area,
32% of volume)
I
I
I
I
Village collectively decide taking the reform or not
Administrative reallocation, on a per capita basis: Equalisation aim
Formal documentation and forestland certificates: Ownership security
Encouraging forestland rental markets: Efficiency enhancing
Why devolving forestland to households?
I
Community management historically unsuccessful, lessons regarding
unclearly defined and insecure tenure rights (Hyde et al., 2003)
In and post devolution: Concerns on inefficiency, agglomeration
Yuanyuan Yi
Allocative Efficiency & Forestland Markets
WB Land Conference 2017
2/12
Map of China’s Forest, 210 million ha, 2013
An example: village forest and rural households in Jiangxi Province
Hypotheses
Forestland rental market participation
1
2
3
4
contributes to the equalisation of forestland-labor ratios.
enhances allocative efficiency by transferring forestland to producers
with higher ability.
has a positive net income effect.
has a positive impact on poverty alleviation.
F
F
Yuanyuan Yi
Rent-in ⇒ Income ↑ ∵ higher forestry productivity, more factors of
production
Rent-out ⇒ Income ↑ ∵ more time in off-farm jobs, more income from
off-farm
Allocative Efficiency & Forestland Markets
WB Land Conference 2017
3/12
Data
Panel data, survey of 1,000 households in 3 provinces
2003
2005
Reform
I
I
Survey-1
2010
Reform
Survey-2
Household characteristics, farming and non-farming activities and
income, and forestland management activities (plot level)
Village socio-economic info, reform decision and implementation
Weather data: Rainfall - yearly average and variability;
temperature - effective and harmful for trees growth
Yuanyuan Yi
Allocative Efficiency & Forestland Markets
WB Land Conference 2017
4/12
Higher participation rate in reformed households
Figure: Forestland rental participation: no reform vs with reform
Yuanyuan Yi
Allocative Efficiency & Forestland Markets
WB Land Conference 2017
5/12
Factor ratio: Pre-rantal vs After-rental
(a)
2005
(b)
2010
Figure: A comparison of factor ratio with rental participation
Yuanyuan Yi
Allocative Efficiency & Forestland Markets
WB Land Conference 2017
6/12
Empirical strategy
1
Household forest productivity: α̂i = µbi + εc
ijt from FE estimation:
logyijt = γ1 logLijt +γ2 logAijt +γ3 logKijt +γ4 Xijt +ηt +ϑj +ϑjt +µi +ijt .
(1)
2
Determinants of rental participation: by multinomial logit
A
τijt = δ1 α̂i +δ2 ( )ijt +δ3 FTRjt +δ4 Tijt +δ5 Xijt +δ6 Vjt +β υˆjt +eijt . (2)
L
τijt : 0: rent-out, 1 no-rent, 2 rent-in,
ˆ jt − zj ξˆ0 : a control function approach.
and υˆjt = FTR
3
Welfare effect of rental market participation: by PSM
Ii = p(Zi ) + θDi + σi .
(3)
Ii , income pc, Pr(income below absolute poverty), off-farm income,
forestry income
Yuanyuan Yi
Allocative Efficiency & Forestland Markets
WB Land Conference 2017
7/12
Results: Forestland rental participation
Table: Determinants of Rental Participation, multinomial logit estimation
For 2005 samples
Rent-out Rent-in
For 2010 samples
Rent-out
Rent-in
0.172***
(0.054)
-0.524***
(0.186)
-0.063
(0.082)
0.673***
(0.196)
0.214**
(0.100)
-0.579**
(0.281)
0.058
(0.070)
0.471***
(0.167)
0.195***
(0.060)
-0.573***
(0.173)
0.805**
(0.314)
-0.578**
(0.263)
1.644**
(0.678)
-1.190**
(0.492)
0.531
(0.690)
-0.533
(0.574)
-0.192
(0.363)
0.088
(0.307)
0.685**
(0.279)
-0.461*
(0.241)
VARIABLES
Rent-out
Rent-in
α̂i (Productivity)
0.017
(0.051)
0.609***
(0.132)
0.251
(0.320)
-0.223
(0.251)
( AL )ijt
ˆ jt (Length with Reform)
FTR
υˆjt (Reform selectivity)
N
2,151
2,151
1,187
1,187
964
Log Lik
-567.5
-567.5
-171.4
-171.4
-339.6
2
Pseudo-R
0.230
0.230
0.367
0.367
0.236
Note: Controlled for: risk perception, credit access, opportunity cost, transaction costs,
other household and village characteristics, and year, province fixed, province year efects.
Yuanyuan Yi
Allocative Efficiency & Forestland Markets
964
-339.6
0.236
WB Land Conference 2017
8/12
Hypothesis 1 2
Forestland rental market participation:
1
contributes to the equalisation of forestland-labor ratios.
2
enhances efficiency by transferring forestland to more productive hhs.
Implications for equalised factor ratios and efficiency ↑:
I
Predicted change in probability (%) of falling into each of the three:
Variable
Type of change
Rent-out
No-rent
Rent-in
Doubling
Doubling
2.159***
0.029
-0.262
-0.618***
-1.896***
0.588*
Forestland-labor endowment
Household foretry Productivity
No evidence for agglomeration to rich, powerful hhs:
Statistically insignificant:
Yuanyuan Yi
Total livestock value
Cropland area per capita
H.head if Communist Party member
H.head if ever village leader
Allocative Efficiency & Forestland Markets
WB Land Conference 2017
9/12
Results: Welfare effect of renting
Hypothesis 3 4
Rental market participation has a positive effect on:
Household income per capita
Poverty reduction: hh income pc below the poverty line
3 treatments separately: Participation, Rent-in, Rent-out
Outcomes: pc income, poverty status, off-farm income pc, off-farm
income share, forestry income per area unit, forestry income share
Self-selecting into treatment =⇒ propensity score matching
Yuanyuan Yi
Allocative Efficiency & Forestland Markets
WB Land Conference 2017
10/12
Results: Welfare effect of renting
Table: Average Treatment effect estimates (ATT, θ)
Participation
Rent-in
Rent-out
Income per capita
1538.36**
1318.91
1849.48*
Absolute poverty
-0.093***
-0.093**
-0.088*
Off-farm income pc
1445.30***
1423.21
1645.87**
Off-farm income share
0.035
0.027
0.070*
Forestry product value per mu
1.47
76.86
-77.85***
Income share from forestry
-0.020
0.020
-0.047**
Notes: Standard errors are bootstrapped with 200 replications.
The matching algorithm: Kernel-based matching (KBM) with bandwidth 0.01.
Robust to: KBM with bandwidth 0.06, NNM 1-1, 1-5,1-10
On average:
Participation ⇒ 1500 CNY pc income ↑, 9% ↓ in likelihood of poverty
Rent-in ⇒ forestry income ↑ insignificant: no-renters also doing well
Rent-out ⇒ off-farm income ↑: by 1600 CNY and 7% higher
Yuanyuan Yi
Allocative Efficiency & Forestland Markets
WB Land Conference 2017
11/12
Concluding remarks
The Forest Tenure Reform ⇒ more forestland to households ⇒ more
rental participation ⇒ allocative efficiency, not agglomeration ⇒
household welfare
Policy implications:
1
2
3
Improve access to land of more efficient producers, management
skills
Reduce transaction costs, keep vigilant of consolidation
Encouraging forestland rentals also good for structural transformation
⇒ rural development
Yuanyuan Yi
Allocative Efficiency & Forestland Markets
WB Land Conference 2017
12/12
Propensity scores of being treated: p 1 (Zi ), p 2 (Zi ), p 3 (Zi )
Zi : hh and village social, economic characteristics
(a)
Participate
(b)
Rentout
(c)
Rentin
Figure: Distribution of propensity scores and common support
Difference-in-means tests, by rental status
Table: Mean differences by rental status
Welfare indicator
No-rent
(A)
Participation
(B)
Difference
(B-A)
Rent-in
(C)
Difference
(C-A)
Rent-out
(D)
Difference
(D-A)
Per capita income, CNY
Below poverty line
6007.6
0.305
8455.2
0.156
2447.6 ***
-0,149 ***
8939.5
0.129
2931.9 ***
-0,176 ***
7937.4
0.184
1929.8 **
-0.117 **
Off-farm income pc, CNY
Off-farm income share
3495.5
0.514
5773.7
0.568
2278,2 ***
0,054 **
6253.3
0.572
2757.8 ***
0,058 *
5261
0.564
1765.5 ***
0,05
Forestry production, CNY/mu
Forestry production share
162.4
0.101
92.6
0.07
-69,8
-0.031 **
171.9
0.098
9,5
-0.003
7.7
0.04
-154,7
-0.061 ***
Number of observations
1979
180
93
87