Land reforms, labor allocation and economic diversity: evidence from Vietnam
1. 2014 PhD Conference
Land reforms, labor allocation and economic diversity of farm
households - The case of Vietnam
Huy Nguyen, PhD Student
Arndt-Corden Department of Economics
Crawford School of Public Policy
Australian National University
2. Content
1. Motivations
2. Research question and objectives
3. Theoretical background
4. Empirical models
5. Data
6. Empirical results
7. Conclusions and policy implications
3. Motivations
— The development experience shows that the success of countries is accompanied by
agricultural growth and economic structural change (Lewis 1954, Perkin et al. 2006,
Warr 2009)
— Johnson (2000) and Haggblade et al. (2007) find that agricultural growth is
necessary for both poverty reduction and development of the non-agricultural
sector.
— Matsuyama (1992) indicates that the growth of agricultural productivity can slow
structural change in open economies
— Foster and Rosenzweig (2004, 2008) find that growth in nonfarm sector in India is
not predicated on the expansion of agricultural growth
— This raise the question whether Vietnam can release labor from agriculture in a way
that improves productivity and adopts mechanized labor saving methods of
cultivation: What policy can this objective be achieved by?
4. Motivations
— Research hypothesis: land reforms through the reduction of land
fragmentation is one of determinants to achieve the objective.
— Land fragmentation is defined as the existence of a number of spatially
separate plots of land, which are farmed as single units (McPherson, 1982).
— It can be an obstacle to agricultural development because it hinders
agricultural mechanization, and results in time loss in travel and
inconvenience and inefficiencies in production. The reduction of land
fragmentation improves agricultural productivity (McPherson, 1982;Wan and
Cheng, 2001; Jha et al., 2005; Rahman, 2008; Tan et al., 2006; Hung et al.,
2007; Kompas et al., 2012; Markussen et al., 2013) .
— In Vietnam, land was redistributed to individual household on egalitarian basis.
There were 75 million plots over nearly 14 millions of household in rural
areas (Vy 2002).
5. Research question and objectives
1. Research question:
- This study investigates whether an exogenous shock to agricultural productivity
leads to changes in labour intensity and economic diversity by looking at two
outcomes: farm and nonfarm (labour intensity and profits)
- Do land reform toward land consolidation reduce farm labour intensity and
induce economic diversity in farm households in Vietnam?
2. Objectives:
- Contributing to current debate on the relationship between agricultural growth
and rural transformation
- Examining the role of land policies in facilitating rural transformation in Vietnam
- This study provides the theoretical background of linkages between land
fragmentation and labor allocation, which is theoretically undetermined by Jia and
Petrick (2013).
6. Theoretical background
- If agricultural technical change is Hick-neutral, this process leads to more on
farm labor supply.
- If agricultural technical change is factor-biased and the elasticity of
substitution is low enough, it may reduce on farm labor supply and release
more labor to other sectors.
- Impacts of land fragmentation on labor allocation can be theoretically
determined
7. Empirical study
• Test research hypothesis: : land reforms through the reduction of land fragmentation
results in changes in labor allocation and economic diversity of farm households in
rural Vietnam.
• By solving the household’s resource allocation problem and farm household’s optimal
labor allocation to main activities:
a= farm, nonfarm
La = f (Xt ,Ak,t ,αt ,LFt ,ε a,t )
Ya,t = f (Xt ,Ak,t ,αt ,LFt ,ε a,t )
• The effect of the reduction of land fragmentation on farm outcomes: productivity,
labor supply, profits and share of individuals in farming activities
• The effect of the reduction of land fragmentation on nonfarm outcomes: nonfarm
labor supply and profits.
8. Empirical models
• First difference method: a= farm, nonfarm
ΔLit ,a = β1ΔSit +β2Xit−1 +β3Ait−1 +β4LFit−1 +β5Rk + Δε it ,a
ΔYit ,a = λ1ΔSit +λ2Xit−1 +λ3Ait−1 +λ4LFit−1 +λ5Rk + Δε it ,a
• Mundlak fixed effects:
Lit ,a = β0 +β1Sit +β2Xit +β3Ait +β4LFit +β5Rk +β6T +β7X _
h+ωit ,a
Yit ,a = λ0 +λ1Sit +λ2Xit +λ3Ait +λ4LFit +λ5Rk +λ6T +λ7X _
+ωit ,a
• Sample selection model using Wooldridge (1995): testing sample selection bias
- Exclusion restrictions: Unearned incomes followed by Gupta and Smith (2002) in the
participation equation but not in the nonfarm labor supply and profits.
- Robustness to controlling for market wages
9. Data
• VHLSSs of 2004 and 2006 provided by GSO of Vietnam
• There were 9,189 households in 2,216 communes surveyed in each VHLSS 2004 and
2006, which forms a panel dataset including 4193 households for each year.
• Follow the approach of Jolliffe (2004) by selecting farm households with at least one
member who describes the main jobs as farming and which have positive farm
profits. In addition, households with no annual crop outputs were excluded from the
analysis
• Finally, 4,028 households over the two waves of the survey after controlling attrition
bias
• An aggregate measure of wage income and self-employment profits into nonfarm
profits. The estimation of censored variables becomes less severe if merging two
types of nonfarm activities together.
10. Data
Variables Mean Std. Dev.
• Farm outcomes (Dependent variables)
Farm profits/ha/year, 1000 VND
34879.69
96583.81
Rice output/ha, tons/ha
5.6
4.3
Farming hours
2446.90
1822.19
Share of individuals in farming activities
33.8
0.34
• Nonfarm outcomes (Dependent variables)
Nonfarm profits (1000 VND)
6833.25
11266.63
Nonfarm hours
1573.37
2034.10
• Explanatory variables
Land fragmentation
Simpson index
0.54
0.25
Number of plots
5.58
3.78
Household characteristics
Land, ha
0.51
0.76
Age of the head, years
46.96
14.40
Gender of the head, 1 for male
0.59
0.49
Marital status of the head, 1 for married
0.83
0.37
Ethnic status of the head, 1 for majority
0.81
0.39
11. Data
Variables Mean Std. Dev.
Household members, from 15 to 60 years old
2.75
1.32
Dependency ratio , %
0.33
0.23
Value of assets, 1000 VND
10888.29
40606.4
Day of illness, days
19.52
43.81
Education
Mean education of working age men, years
3.85
2.40
Mean education of working age women, years
3.66
2.38
Head of household has primary education
0.25
0.43
Head of household has lower secondary education
0.38
0.49
Head of household has university education
0.01
0.09
Participation into nonfarm activities
Having member working in state economic sectors
0.098
0.297
Having member working in private economic sectors
0.052
0.224
Having member working on household’s own business
0.850
0.357
Locational factors
Regions
No of observations
4028
12. Empirical results
• The effect of land fragmentation on farm outcomes using first difference
method
Dependent variables: Farm
outcomes
Log of plots Simpson index
Coef. SE Coef. SE
Farm labor supply
No. of individuals in farming activities
Farm profits per ha
Farm outputs per ha
0.355***
0.051
-0.115***
-0.055***
0.129
0.041
0.031
0.007
0.533*
0.200*
-0.109
-0.092***
0.315
0.097
0.082
0.019
Notes: All dependent variables are expressed in the log, except number of individuals in
farming activities; ∗, ∗∗, ∗∗∗ indicates that the corresponding coefficients are significant
at the 10%, 5%, and 1% levels, respectively; In first difference method, the regression
include all of the control variables: the initial characteristics of households and
communes, a dummy for regions.
13. Empirical results
• The effect of land fragmentation on nonfarm outcomes without selection correction
using the Simpson index
Dependent variables: nonfarm
outcomes
First difference Mundlak fixed effect
Coef. SE Coef. SE
Nonfarm labor supply
Nonfarm profits
-0.646*
-0.233
0.344
0.408
-0.120*
-0.307***
0.063
0.096
• The effect of land fragmentation on nonfarm outcomes without selection correction
using the log of plots
Dependent variables: nonfarm
outcomes
First difference Mundlak fixed effect
Coef. SE Coef. SE
Nonfarm labor supply
Nonfarm profits
-0.324**
-0.225
0.143
0.168
-0.026*
-0.154***
0.027
0.038
Notes: Standard errors are robust through cluster option; ∗, ∗∗, ∗∗∗ indicates that the
corresponding coefficients are significant at the 10%, 5%, and 1% levels, respectively;
Controlling household characteristics and locational factors.
14. Empirical results
• The effect of land fragmentation on nonfarm outcomes with selection correction using
the Simpson index
Dependent variables: nonfarm outcomes Wooldridge (1995)
Coef. SE
Nonfarm labor supply
Nonfarm profits
-0.122*
-0.297***
0.063
0.080
• The effect of land fragmentation on nonfarm outcomes with selection correction using
the log of plots
Dependent variables: nonfarm outcomes Wooldridge (1995)
Coef. SE
Nonfarm labor supply
Nonfarm profits
-0.023*
-0.143***
0.027
0.037
Notes: Standard errors are robust through cluster option; ∗, ∗∗, ∗∗∗ indicates that the
corresponding coefficients are significant at the 10%, 5%, and 1% levels, respectively; The
models include Mundlak fixed effects; Controlling household characteristics and locational
factors
15. Conclusions and policy implications
— If land fragmentation reduce by 1%, farm labor supply decreases by 0.36%.
Farm profits per ha and farm output per ha increase by 0.12% and 0.055%,
respectively.
— Land consolidation may release more farm labour to nonfarm sectors and
increase the nonfarm profits.
— Factor biased technical change plays an important role in explaining the effect of
agricultural technical change on economic diversification and income in
Vietnam.
— The finding supports arguments of Johnson (2000) that the productivity
improvement in farm sector will promote the development of nonfarm economy
and economic diversification of households.
— If land polices that encourage more consolidated land holdings, they will release
more farm labour and result in economic diversification of farm households.
— Future research should capture the changes in prices of goods and factors. The
effects of uncertainties such as shocks and risk on smallholder decision-making
should be further explored.
16. Thank you
Comments are sent to Huy Nguyen, PhD student
Email: huy.nguyen@anu.edu.au