Land Titles and Conflicts in Guatemala

Land Titles and Conflicts in Guatemala

Karen Macours

SAIS-Johns Hopkins University

Effects of land titling ?

• Mixed evidence to date• Different reasons:

– Effect of title depends on other factors• Other market imperfections (Carter and Olinto, Boucher et al.)• Formal versus informal property rights (Braselle et al., Lanjouw and Levy)

~ titles do not necessarily increase property rights security~ other factors affecting security (rule of law, conflict resolution, … ) might matter

– Problems with identification

Contribution of this paper

• Relationship between the impact of land titles depend on conflicts and conflict resolution mechanismsPotential lessons for targeting

• Instrument exploiting correlations resulting from historical land title processes

Titles and conflicts in Guatemala

• Long-standing conflicts related to land access

• Weak rule-of-law

• Customary conflict resolution mechanisms

=> Widespread property rights insecurity

History of land titling (1)

• Late 19th century: coffee boom– government encouraged transformation from community to

private property– Titling through application by individual possessors or

community authority– Some communities/owners applied, others didn’t– Some communities lost (part of) their land to outsiders, who

obtained title or not

Geographic correlation of title status

History of land titling (2)

• Systematic titling efforts since then have not affected the regions of study

• But: Ley de titulacion Supletoria=> individuals can claim titles based on “10 year peaceful possession”=> Current title status not exogenous

Data

• 20 communities, 2 regions• Info on 1822 households and all their plots• Geographic location of all the plots (community

maps)• Detailed information about community conflicts

and conflict resolution mechanisms• Information obtained through key informants

~ sensitivity of information on land and conflicts

Outcome variables

Table 1: Descriptive statistics of outcome variablesNo title Title Significance

1440 402 differencea

Plot-level outcomesEfficient use (used at full potential) 0.74 0.86 ***Extensive use (pasture, forest, idle) 0.13 0.09 **

Household-level outcomesGood living standard 0.22 0.55 ***Access to credit 0.65 0.87 ***

a: standard errors corrected for clustering at household level

Conflicts and conflict resolution (1)

%

Community mechanismsIntervention of a community leader or council of elders 1 rights inheritanceThrough a community meeting or a conciliatory council 2 rights resources laborIntervention of the community's mayor aid. 4 misuse trespassing rentalIntervention by a community committee 12 comun. land resources trespassing

Authorities* Intervention by the mayor or municipal authority 16 comun. land invasion resourcesConciliation through judge 5 rights invasion trespassingDecision by judge 10 labor invasion trespassingJuridical decision by other authorities 0 trespassingthrough a lawyer 4 rights inheritanceINTA 1 rightsintervention of the labor inspection 1 labor

Mediation*Mediation by the "mesa de negociación de conflictos de tierras" 1 invasion comun. landContierra 2 rights invasion credit

Among themselvesAmong parties themselves 43 rental misuse border

* No conflicts were reported to be resolved by mediation of an NGO, a campesino organization, or the church

type of conflict

Conflicts and conflict resolution (2)

• No clear pattern of different institutions specializing in certain conflicts

• Conflict resolution mechanisms do not show clear patterns in duration, intensity, or the number of actors involved

• Instead: different communities rely disproportionally on different mechanisms

Conflicts and conflict resolution (3)

20 communities: 18 types of conflicts min mean maxNumber of conflicts resolved through community mechanisms 0 3 12

Number of conflicts resolved through authorities 0 5 17

Number of conflicts resolved among parties themselves 1 7 15

Number of conflicts resolved through mediation by 3rd parties 0 0 3

Identification

• Title status plot mostly determined at end of 19th century• Since then: land fragmentized among many owners, …• But title status of neighboring plots still correlated with

each other• Use average title status of upto 5 neighboring plots (from

different owners) as instrument– Should be uncorrelated to many landlord and plot characteristics– Control for plot characteristics that might be spatially correlated

(and possible related to title status)– Control for family background and other characteristics of the

owner

First stage

(1) (2) (3)

Average title status 5 closest neighboring plots 0.501*** 0.186** 0.222***(7.17) (2.52) (3.32)

Title status nearest plot 0.206*** 0.133*** 0.098**(3.66) (2.93) (2.23)

Constant 0.051** -0.019 0.170(2.29) (0.17) (0.81)

Plot-control variables no yes yesHousehold-control variables no no yes

Observations 1348 1348 1272R-squared 0.38 0.56 0.65

Efficiency of plot use: Lin. Prob.

(1) (2) (3) (4)

Registered title 0.119*** 0.091 0.110** 0.087**(3.00) (1.52) (2.22) (2.07)

Community fixed effects no yes yes yesPlot-control variables no no yes yesHousehold control variables no no no yesObservations 1348 1348 1348 1272R-squared 0.01 0.36 0.38 0.44

Efficiency of plot use: IV

(5) (6) (7)

Registered title 0.271 0.169 0.176(1.64) (1.10) (1.14)

Community fixed effects yes yes yesPlot-control variables yes yes yesHousehold control variables no yes yesObservations 1348 1272 1272First stage F-stat 14.40 16.05 30.53P-value overidentification test (Hansen J) 0.68 0.88 just id.

Extensive plot use: Lin. Prob.

(1) (2) (3) (4)

Registered title -0.031 -0.030 -0.078** -0.093***(1.32) (0.71) (1.97) (3.04)

Community fixed effects no yes yes yesPlot-control variables no no yes yesHousehold control variables no no no yesObservations 1348 1348 1348 1272R-squared 0.00 0.08 0.21 0.24

Extensive plot use: IV

(5) (6) (7)

Registered title -0.201 -0.302** -0.298**(1.55) (2.22) (2.18)

Community fixed effects yes yes yesPlot-control variables yes yes yesHousehold control variables no yes yesObservations 1348 1272 1272First stage F-stat 14.40 16.05 30.53 P-value overidentification test (Hansen J) 0.96 0.93 just id.

Credit access

Linear Probability IV

(1) (2) (3) (4) (5) (6)

Registered title 0.181*** 0.293*** 0.246*** 0.200*** 0.361** 0.300**(4.73) (7.45) (6.21) (4.53) (2.23) (1.97)

Community fixed effects no yes yes yes yes yesPlot-control variables no no yes yes yes yesHousehold control variables no no no yes no yesObservations 1344 1344 1344 1271 1344 1271R-squared 0.02 0.22 0.25 0.39First stage F-stat 14.31 16.03P-value overidentification test (Hansen J) 0.64 0.33

Conflicts, title and efficiency of plot use

(1) (2) (3) (4) (5)

Registered title -0.152 -0.029 0.363** -0.255 0.269(0.81) (0.17) (2.39) (1.32) (1.23)

Title* Average conflict duration in the community 0.009***(3.35)

Title* Community with rights conflict in last 10 years 0.685**(2.06)

Title* Rights conflicts resolved by outside authorities -1.030**(2.40)

Title* Rights conflicts resolved through community mechanisms 1.122***(4.17)

Title* Rights conflicts resolved among parties -0.195(0.74)

Observations 1269 1272 1272 1272 1272

Conflicts, title and extensive plot use

(6) (7) (8) (9) (10)

Registered title -0.049 -0.095 -0.483*** 0.081 -0.389**(0.31) (0.71) (3.47) (0.49) (2.00)

Title* Average conflict duration in the community -0.007***(2.87)

Title* Community with rights conflict in last 10 years -0.713**(2.15)

Title* Rights conflicts resolved by outside authorities 0.889**(2.53)

Title* Rights conflicts resolved through community mechanisms -0.994***(3.87)

Title* Rights conflicts resolved among parties 0.142(0.62)

Observations 1269 1272 1272 1272 1272

Preliminary Conclusions

• Effects of titles on credit does not depend on conflicts

• Effects of titles on efficiency of plot use depends on conflict-context– Titles matter (more) for efficiency when conflicts are

longer and when conflicts get resolved by community mechanisms

– Intriguing pattern: titles are negatively related to efficiency when conflicts typically resolved by outside authorities