Land Conflict and Agricultural
Tenancy: The Impact on Land Use and
Farm Size
Lee J. Alston
University of Colorado
NBER
Bernardo Mueller
University of Brasilia
• In Brazil rentals account for about 11% of the
farms
• For U.S. and OECD countries it’s close to 50%
• Missing market- millions of landless peasants
and unused or under-utilized land
• Tenancy associated with career mobility
• Too few rentals implies an inefficiency
• Landless move to the frontier and deforest
Figure 1 – Evolution of Tenancy over time in Brazil
25
1970
20
1960
1985
1975
% of total
1980
15
1940
1995
10
1950
2006
5
1920
0
1910
1920
1930
1940
1950
1960
1970
1980
1990
Hectares
Source: IBGE (2007). Data for 2006 from the 2006 Agricultural Census.
2000
2010
2020
Number of Farms
 Insecure
Property Rights, i.e., the threat of
expropriation - fewer land rentals
 de janvry and Sadoulet 1989
• Land reforms of the 1960s and 1970s gave
incentives for landowners to modernize and
expel tenants.
 Conning
and Robinson (2007)
• “…the anticipation of future property rights
challenges by tenants lead landlords to
defensively suppress tenancy as a costly
mechanism to protect property rights.”
 Land
Statute, 1964
 “… expropriation will be applied to: ... areas
with high incidence of renters, sharecroppers
and squatters."
 Rural Worker
 Land
Statute, 1963
Reform- Land Conflict leads to
expropriations (Alston, Libecap, Mueller, 1999,
2000, 2009)

Rented land may attract invasion

Need instruments for land conflict
Landless peasants and rural workers
became increasingly organized in 19701980s:

•
Catholic Church –organized landless
peasants from the early 70s until the mid-80s.
•
Instrument: Priests per rural pop 1966
Land Conflict and Tenancy


Security of Property rights affect contract choice
(1996 Census data) –more conflict associated with
fewer fixed rent and sharecrop contracts.
- one std dev in conflict: decreases fixed rent from
4% to 3% and sharecrop from 2.5% to 1.25%
Rented farms “too small” and sharecropped
farms “too large”- generating inefficiencies in
agriculture .
• Net result: farm size increases
Land Conflict and Land Use


Our results: land conflict pushes land use into
marginal uses and lowers investment
Increase from 0 to 4.4 conflicts1000 farms:
- natural pasture- 20% to 3% (target for invasions)
- temporary crops- 18% to 15% (highly productive)
- planted pasture – 26% to 37% (low productivity)

Total country wide impact in hectares= to the size
of small countries, e.g. Greece, Honduras plus
others
Land conflict is the result of insecure property
rights generated by government policies, e.g.
land reform
 Land Conflict Reduces Tenancy:

• 1) hurts the landless;
• 2) creates inefficiencies; and
• 3) causes deforestation through migration to the
frontier.

Conflicts skew land uses to low productivity uses
 Coase-like
bargaining?
 Urban
constituents favor redistribution of
land, affects credible commitment of
Government
 Belief
in social inclusion (for now) prevents
fixing the “missing market.”
 Coase-like
bargaining?
 Brazilian
Belief in Social Inclusion
sustains misallocation

Urban constituents favor redistribution
of land, affects credible commitment of
Government
700
160
600
140
120
Occupations
500
100
400
80
300
60
200
40
100
20
0
0
2007
2006
2005
2004
2003
2002
2001
2000
1999
1998
1997
1996
1995
1994
1993
1992
1991
1990
1989
1988
Occupations
Families Settled
Source: Ministério do Desenvovlvimento Agrário (2004: 20), MDA/INCRA Balanço de 2007 (2008).
Comissão Pastoral da Terra (2004:13). Note: Data for number of families settled from 1988 to 1994 is
the average for each government; Sarney (1988-89), Collor (1990-91), Franco (1992-94).
Families Settled (1000)
Land Conflicts and Land Reform
 In
the 1970s the Catholic Church in Brazil made
a ‘preferential option for the poor.’
 It
became “probably the most progressive
Church in Latin America, if not the world.”
(Bruneau, 1987: 271)
 Pastoral
Land Commission (CPT).
“The CPT (Pastoral Land Commission) was
the practical application of the Theology of
Liberation, which was an important contribution
to the landless peasants’ struggle from the
ideological point of view. The priests, pastoral
agents and pastors discussed with the
peasants the need for them to organize
themselves. The Church stopped doing
messianic work and saying to the peasant:
‘Wait and you will go to heaven.’ Now they
started saying ‘You have to get organized and
fight to solve your problems here on earth’”
J.P. Stedile, main leader of the MST on the Church’s role
in the foundation of the MST in the early 1980s.


Land Conflicts: data collected by the
Catholic Land Pastoral Commission.
Land Conflicts per farm by county
1986-1995 – aggregated to MCA.

Conflicts in 645 of the 3615 MCAs

Tobit estimation
Exclusion Restriction
Brazilian Bishops
conference
Priest Medellin
data conference
1966
1970
Conflict
data
Puebla
conference
1975
1980
Direct
active
Church
participation in
agrarian
issues
1985
1990
Redemocratization
Conservative shift
of the Church.
Rise of the MST.
Evangelical movement.
Contract
data
1995
Exclusion Restrictions
“In more recent years the Church has become increasingly confused
with respect to support for societal transformation. Not only has the
upper hierarchy become more fractious, a tendency toward
conservatism has also become apparent. The Church as an institution
has returned to previous modes of political influence and appears to be
abandoning its support for grassroots movements in favor of direct
pressure on political policy makers”. (pg. 148)
Hewitt, Warren E. 1990. “Religion and Consolidation of Democracy in Brazil: The Role of the Comunidades Eclesiais de Base
(CEBs),” Sociological Analysis, 50:2, 139-152.
Determinants of Priest Allocation and Opposition Party
Strength
Dep. Variable: Priest
per 1000 rural pop in 1985
I
Priest / 1000 rural pop 1966
1.09***
(11.96)
II
III
IV
V
Dep. Var: Opposition
party†
1.09***
(11.48)
Oppo. Party (1982)
VI
Opp. 1982
VII
Opp. 1996
0.010*
(1.92)
Fixed rent % (1970)
23.763**
(2.42)
3.109
(0.40)
Fixed rent %
0.144***
(2.87)
-0.012
(-0.89)
Sharecrop % (1970)
-14.756*
(-1.67)
-10.553
(-1.15)
Sharecrop %
0.213***
(3.30)
0.053
(1.63)
Occupied % (1970)
-24.222***
(-5.14)
-2.178
(-0.78)
Occupied %
-0.045
(-1.28)
0.019
(1.62)
Population growth 1970-80
3.151
(1.04)
-0.012
(-0.01)
Pop. growth
0.040***
(3.51)
-0.001
(-0.90)
GDP growth 1970-80
0.109
(0.61)
0.063
(0.42)
GDP growth
-0.0003
(-0.31)
-0.001***
(-2.76)
-0.0001**
(-2.29)
-0.00001
(-1.41)
-2.2e-7***
(-4.06)
0.0000001*
Income (1970)
(12.26)
-0.0002***
(-6.15)
0.00004***
(3.95)
Frontier
0.002*
(1.72)
0.0004
(1.24)
0.009
(0.03)
Latitude
-0.003
(-1.11)
0.002***
(2.93)
0.311***
(2.92)
0.327
(1.55)
Longitude
0.012***
(4.52)
-0.004***
(-4.31)
Constant
-0.709***
(3.45)
0.318***
(3.75)
Total: 3656
Yes
Total: 3656
Yes
0.35
84.07
0.0000
0.10
9.94
0.0000
1.453
(1.53)
(9.66)
-0.048***
-0.008**
(-7.52)
(-2.41)
Frontier
-1.411***
(-5.72)
0.088
(0.90)
Latitude
-1.036***
(-11.41)
Longitude
Distance to state capital
Constant
Number of observations
State dummies (27 states)
R2 adjusted
F(k, n-k)
Prob>F
**
(3.47)
0.014***
(8.46)
12.017***
Schooling (1970)
Income
0.805
(0.75)
17.234***
(18.74)
-0.165
(-0.13)
-0.677
(-0.17)
-23.517*
(-1.88)
Total: 3631
No
Total: 3631
No
Total: 3631
No
Total: 3631
No
Total: 3631
Yes
0.64
142.94
0.0000
0.005
13.50
0.0000
0.05
30.44
0.0000
0.07
37.13
0.0000
0.64
74.79
0.0000
Schooling
Dist. to state capital
Number observations
State dummies (27
states)
R2 adjusted
F(k, n-k)
Prob>F
0.082***
Explanatory Variables:
1. Number of Priests /rural population
2. Frontierness – number of times a MCA subdivided
3. Priests interacted with Frontierness
4. Opposition Parties: MDB and PT % of seats
5. Distance to state capital
6. Latitude and Longitude
7. Population Density: more demand for land
1st Stage Specification – Land Conflict
8. Ag/GDP: More valuable land better defended
9. Land concentration: size of farms
10. Crop variables
11. State dummies.
Data:
Land Conflict: (Pastoral Land Commission)- rural
threats, murders, murder attempts and occupations by
county
Ag Variables: Agricultural Census;
Priests: Catholic Hierarchy;.
Determinants of Rural Conflict – First Stage Equation
Dep. Var.:
Violence 1985-1996
Priests per rural population
Frontier
Interaction: Priest x Frontier
Political opposition 1982 (% seats in state
assembly – MDB
Political opposition 1996 (% seats in state
assembly – PT
Agricultural GDP growth 1985-1995.
Distance to state capital
Latitude
Longitude
Cattle per hectare1995
Tractors per hectare1995
Rural/Urban Population (1995)
Population growth 1985-1995
Constant
Number of observations
State Dummies (27 states)
Pseudo R2
2(55)
Prob>2
Priests 1966
No Opposition Parties
-0.629***
(-10.47)
0.233***
(5.01)
0.287***
(8.21)
2.683***
(4.68)
0.004
(1.39)
0.195
(0.84)
-0.359
(-1.35)
-0.656
(-0.67)
-1240.63***
(-8.57)
-1.650***
(-4.28)
1.236***
(2.65)
19.994
(1.10)
Total:
3616
Censored at 0: 2967
Uncensored:
648
Yes
0.14
1131.91
0.0000
Priests 1966
Opposition Parties
-0.614***
(-10.26)
0.232***
(5.00)
0.285***
(8.13)
6.680***
(3.52)
15.803**
(2.30)
2.810***
(4.91)
0.006**
(2.02)
0.250
(1.08)
-0.423*
(-1.59)
-0.668
(-0.66)
-1241.46***
(-8.60)
-1.394***
(-3.59)
1.265***
(2.71)
20.930
(1.15)
Total:
3616
Censored at 0: 2967
Uncensored:
648
Yes
0.15
1150.37
0.0000
Tobit Estimation. t-stats in parentheses. Statistical significance: 1% ***. 5% **, 10% *.
Weighted by the number of county subdivision from 1970-2000. Other controls not
shown to save space.
Interaction of Priests and Frontier: Effect of Priests on Conflicts.
0
5
10
15
20
Effect of Priest per hectare per Catholic on Violence
0
10
20
30
Frontier
95% confidence interval
40
50
Determinants of Contract Choice
Fixed Rent (%) Sharecropper (%)
Conflict per 1000 farms
Cotton, % of total farm area
Rice, % of total farm area
Coffee, % of total farm area
Cane, % of total farm area
Beans, % of total farm area
Manioc, % of total farm area
Corn, % of total farm area
Soy Beans, % total farm area
Frontier
GDP growth 1985-1995
Latitude
Longitude
Distance to state capital
Transport cost to São Paulo
Number of train stations
-0.008***
(-3.60)
0.428***
(4.21)
0.275***
(5.97)
-0.146***
(-4.14)
0.187***
(17.77)
-0.050
(-1.40)
0.062
(1.21)
-0.042*
(-1.70)
0.218***
(13.21)
-0.0004**
(-2.28)
0.006**
(2.47)
-0.004***
(-4.60)
-0.001
(-0.93)
-0.000002
(-0.19)
0.000003
(0.73)
0.002***
(3.57)
-0.006***
(-4.21)
0.181***
(2.61)
0.227***
(7.19)
0.042*
(1.73)
0.068***
(9.47)
0.089***
(3.61)
0.097***
(2.78)
0.021
(1.23)
0.032***
(2.82)
0.00003
(0.21)
0.004**
(2.40)
-0.0001
(-0.15)
-0.0007
(-1.01)
0.00002***
(2.66)
-0.0000002
(-0.10)
-0.0001
(-0.21)
Owner (%)
Occupant (%)
0.010***
(3.39)
-0.689***
(-4.99)
-0.512***
(-8.18)
0.124***
(2.60)
-0.225***
(-15.73)
-0.167***
(-3.42)
-0.695***
(-9.98)
0.019
(0.57)
-0.219***
(-9.73)
0.0009***
(4.05)
-0.013***
(-3.87)
0.005***
(4.64)
-0.003**
(-2.01)
0.00001
(0.58)
-0.00001**
(-2.52)
-0.003***
(-2.85)
0.004**
(2.50)
0.080
(1.05)
0.011
(0.30)
-0.021
(-0.78)
-0.030***
(-3.83)
0.129***
(4.76)
0.536***
(13.94)
0.002
(0.11)
-0.032**
(-2.55)
-0.0005***
(-4.05)
0.003**
(1.52)
-0.001**
(-2.15)
0.005***
(5.79)
-0.00003***
(3.23)
0.00001***
(3.68)
0.0003
(0.60)
(Continues)
(Continuation)
Determinants of Contract Choice
Population density 1995
Rural/Urban Population 1995
Population growth 1985-1996
Tractor /hectare growth 19851995
Cattle per hectare1995
Constant
Number of observations
State dummies (27 states)
Pseudo R2
2(44)
Prob>2
Hausman-Wu Test
H0: Conflicts are exogenous.
Fixed Rent (%) Sharecropper
Owner (%)
(%)
0.00002**
0.000001
-0.00002**
(2.06)
(1.47)
(-1.96)
**
-0.002
-0.0008
-0.00005
(-2.32)
(-1.56)
(-0.05)
0.003
0.002
-0.002
(1.12)
(1.36)
(-0.62)
-0.243
0.512***
-0.775**
(-1.06)
(3.26)
(-2.48)
-0.002
0.0002
0.007**
(-0.65)
(0.11)
(2.06)
0.077
0.075
1.100***
(1.09)
(1.54)
(11.42)
Total: 3616
Total: 3616
Total: 3616
Yes
Yes
Yes
0.17
0.05
0.18
1757.51
720.98
1754.12
0.0000
0.0000
0.0000
2
2
2
Χ (1)=25.06
Χ (1)=30.66
Χ (1)=21.18
p-value= 0.0000 p-value=0.0000 p-value=0.0000
Occupant (%)
-0.000003
(-0.52)
0.003***
(4.59)
-0.003
(-1.61)
0.507***
(2.94)
-0.005***
(-2.96)
-0.248***
(-4.68)
Total: 3616
Yes
0.22
1649.15
0.0000
2
Χ (1)=5.39
p-value=0.0202
The impact of Violence on Tenancy and Sharecropping - Brazil
0.06
0.05
0.04
Share NV
0.03
Fixed NV
Share V
Fixed V
0.02
0.01
0
1970
1975
1980
1985
1990
1995
Determinants of Land Use
Natural Forest
%
Conflict per 1000 farms
0.006**
(3.17)
Cotton, % of total farm area
-0.731***
(-4.65)
Rice, % of total farm area
-0.356***
(-4.99)
Coffee, % of total farm area
-0.223***
(-4.10)
Cane, % of total farm area
-0.157***
(-9.95)
Beans, % of total farm area
-0.097*
(-1.83)
Manioc, % of total farm area
-0.862***
(-10.78)
Corn, % of total farm area
-0.195***
(-5.07)
Soy Beans, % total farm area
-0.075***
(-2.92)
Constant
0.532***
(5.02)
Number of observations
Total: 3616
State dummies (27 states)
Yes
R2
0.37
2(44)
4904.52
Prob>2
0.0000
Planted Forest
%
0.006***
(6.21)
-0.098
(-1.19)
-0.063*
(-1.70)
-0.116***
(-4.08)
-0.053***
(-6.36)
0.047*
(1.68)
-0.029
(-0.70)
-0.052***
(-2.61)
-0.039***
(-2.89)
1.096*
(1.74)
Total: 3616
Yes
0.07
829.24
0.0000
Perm. Crops
%
0.017***
(8.97)
-0.097
(-0.60)
-0.111
(-1.51)
1.056***
(18.79)
-0.034**
(-2.12)
-0.077
(-1.41)
0.428***
(5.23)
0.023
(0.60)
-0.092***
(-3.47)
-0.140
(-1.29)
Total: 3616
Yes
0.11
993.45
0.0000
Temp. Crops
%
-0.007***
(-5.00)
0.815***
(6.59)
0.531***
(9.48)
-0.213***
(-4.97)
0.647***
(52.18)
0.623***
(14.96)
0.908***
(14.47)
0.499***
(16.53)
0.782***
(38.69)
-0.155*
(1.87)
Total: 3616
Yes
0.74
14975.28
0.0000
Nat. Pasture
%
-0.038***
(-9.60)
-0.511
(-1.45)
-0.071
(-0.45)
-0.135
(-1.11)
-0.162***
(-4.59)
-0.123
(-1.04)
-0.573***
(-3.22)
-0.299***
(-3.48)
-0.081
(-1.41)
1.681***
(7.11)
Total: 3616
Yes
0.12
1225.01
0.0000
Plant. Pasture
Fallow %
%
0.063***
0.00003
(8.20)
(0.06)
0.736***
-0.015
(2.67)
(-0.35)
-0.099
0.155***
(-0.79)
(7.75)
-0.319***
-0.018
(-3.34)
(-1.16)
***
-0.244
0.017***
(-8.81)
(3.81)
***
-0.264
0.021
(-2.84)
(1.39)
0.050
0.024
(0.36)
(1.06)
-0.096
0.088***
(-1.42)
(8.19)
***
-0.404
-0.059***
(-8.96)
(-8.17)
-1.273***
0.00004
(-6.88)
(0.00)
Total: 3616
Total: 3616
Yes
Yes
0.29
0.38
2787.73
2421.12
0.0000
0.0000
Unused %
-0.008***
(-7.70)
-0.099
(-1.03)
0.013
(0.31)
-0.032
(-0.96)
-0.014
(-1.45)
-0.129***
(-4.00)
0.055
(1.16)
0.031
(1.32)
-0.031**
(-2.02)
0.215***
(3.39)
Total: 3616
Yes
0.23
2340.63
0.0000
Estimated by Three Stage Least Squares. t-stats in parentheses. Statistical signif.: 1% ***. 5% **, 10% *. The coefficients for all eight equations are constrained to add up to 0 for every variable.
A Hausman-Wu endogeneity test rejects exogeneity of conflicts in all of the equations at 1% (except Natural Forest at 5%) except Fallow.