The author’s views expressed in this publication do not necessarily reflect the views of the United States Agency for International Development 
or the United States Government.
EVALUATION OF THE LAND 
FOR PROSPERITY (LFP) ACTIVITY IN 
COLOMBIA: BASELINE REPORT
MAY 2022
ii
Evaluation of the Land for 
Prosperity (LfP) Activity in 
Colombia: Baseline Report
USAID Contract Number: GS00F061GA
Developed under the Communications, Evidence and Learning (CEL) Project
Authors:
Greg Haugan, NORC at the University of Chicago
Ali Protik, NORC at the University of Chicago
Lauren Persha, NORC at the University of Chicago
Research Support Provided by:
Ron Wendt, NORC at the University of Chicago
Miguel Albornoz, NORC at the University of Chicago
Xiran Liu, NORC at the University of Chicago
Paige Pepitone, NORC at the University of Chicago
Zach Scheffler, NORC at the University of Chicago
Special thanks to the Panagora Group MEL Activity in Bogotá, Diego Velandia, Adriana Arana, and 
Hanner Sanchez at Sistemas Especializados de Información (SEI), and Juan Carlos Muñoz at Universidad 
EAFIT.
Suggested Citation: Haugan, G., Protik, A., and Persha, L. (2022) Evaluation of the Land for Prosperity (LfP) 
Activity in Colombia: Baseline Report. Washington, DC: USAID Communications, Evidence and Learning 
(CEL) Project.
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
iii
CONTENTS
Contents iii
Acronyms v
Executive Summary vi
Introduction 1
Conceptual Framing, Evaluation Purpose and Evaluation Questions 1
Land Tenure, Productivity, Conflict and Illicit Crops Background 1
Land for Prosperity Activity Description 3
Development Hypothesis 5
Activity Implementation Status 9
Evaluation Purpose, Audience, and Intended Uses 9
Evaluation Questions 10
Evaluation Design And Methods 14
Design Overview 14
Three-step Statistical Matching 14
Baseline Data Collection 16
Household Quantitative Sample 16
Qualitative Sample 17
Household Survey 18
Challenges Encountered During Data Collection 19
Baseline Sample Characteristics 20
Findings 1: Household Well-Being, Food Security, and Poverty 24
Livelihoods 24
Access to Credit 27
Food Security 30
Non-Farm Activities 32
Findings 2: Land Use and Agricultural Production 33
Land Holdings 33
Land Use 36
Crops Produced 38
Income from Agricultural Activities 39
Productive Investments 41
Findings 3: Land Acquisition and Tenure Security 44
Land Acquisition Trend and Mode 45
Land Rights 45
Land Documentation 46
Tenure Security 49
Land Conflicts and Displacement 52
Land Governance and Land Market Activity 53
Findings 4: Public Service Delivery, Public-Private Partnerships, and Illicit Crop 
Substitution 55
Trust in Institutions 55
Public Service Delivery 58
Public-Private Partnerships 59
Participation in Alternative Development Programs 59
Illicit Crop Cultivation 60
Characteristics of Coca-growing Households 63
Findings 5: Women’s Empowerment 67
Women and Land Tenure 67
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
iv
Participation in Productive Activities and Economic Decisions 70
Control Over Income 75
Asset Ownership 76
Access to Credit 78
Balance and Power 79
Balance Tests 79
Power Calculations 80
Conclusions 83
Implications of Baseline Findings for LfP Theory of Change and Programming 83
Recommendations for Subsequent Data Collection for the IE 85
References 87
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
v
ACRONYMS
ANT Agencia Nacional de Tierras (National Land Agency)
A-WEAI Abbreviated Women’s Empowerment in Agriculture Index
CEL Communication, Evidence, and Learning
COP Colombian Pesos
DANE Departamento Administrativo Nacional de Estadística (National Administrative 
Department of Statistics)
DDI Bureau for Development, Democracy and Innovation (USAID)
DID Difference-in-Differences
EDR Evaluation Design Report
EEI Center for Energy, Environment and Infrastructure (USAID)
GD Group Discussion
GoC Government of Colombia
IE Impact Evaluation
IGA Income Generating Activity
IGAC Instituto Geográfico Agustín Codazzi (Agustin Codazzi Geographic Institute)
INCODER Instituto Colombiano de Desarrollo Rural (Colombian Institute for Rural Development)
INCORA Instituto Colombiano de Reforma Agraria (Colombian Institute for Agrarian Reform)
KII Key Informant Interview
LfP Land for Prosperity
LRDP Land and Rural Development Program
LRG Land and Resource Governance
MDES Minimum Detectable Effect Size
MDI Minimum Detectable Impact
OSPR Ordenamiento Social de la Propiedad Rural (Participatory Rural Land Use Management 
Code)
PDET Programa de Desarrollo con Enfoque Territorial (Territorially Focused Development
Plans)
PNIS Programa Nacional Integral de Sustitución (National Crop Substitution Program) 
POSPR Planes de Ordenamiento Social de la Propiedad Rural (Participatory Rural Land Use 
Management Plans)
PPP Public-Private Partnership
SNR Superintendencia de Notariado y Registro (Superintendency of Notaries and Registry)
STARR Strengthening Tenure and Resource Rights
URT Unidad Restitución de Tierras (Land Restitution Unit)
USD US Dollars
USAID United States Agency for International Development
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
vi
EXECUTIVE SUMMARY
This report provides baseline context for a mixed-methods impact evaluation (IE) of the USAID￾supported “Land for Prosperity (LfP)” activity in Colombia. The evaluation aims to provide an evidence 
base for outcomes of LfP with respect to strengthening land rights and land governance, reducing illicit 
crop cultivation, and enhancing local livelihoods. The evaluation was commissioned by USAID’s Land and 
Resource Governance (LRG) unit in the Bureau for Development, Democracy and Innovation’s Center 
for Energy, Environment and Infrastructure (USAID/DDI/EEI), along with USAID/Colombia, and is led by 
NORC at the University of Chicago under the Communication, Evidence and Learning (CEL) Project. 
This document provides findings from the baseline data collection for the evaluation, including 
background context on key demographics, household characteristics, and baseline measures on 
outcome variables. The report also examines balance across LfP (treated) and comparison communities 
for the IE components of the evaluation, and revisits power calculations from the Evaluation Design 
Report (EDR) using parameters from the baseline data.
LAND FOR PROSPERITY ACTIVITY DESCRIPTION
LfP began in August 2019 and is a five-year activity that is implemented by Tetra Tech ARD under the 
Strengthening Tenure and Resource Rights (STARR) II IDIQ.1 The activity is envisioned to improve the 
conditions of conflict-affected rural households in a sustainable manner that will support 
USAID/Colombia’s effort to promote peace and stability, and contribute to reducing illicit crops and 
help the Government of Colombia (GoC) strengthen state presence in under-attended regions.2
LfP comprises three main intervention components: (1) advancing massive land titling3 in rural areas 
along with continued support in land restitution for forcibly displaced households; (2) strengthening local 
capacity to maintain formalized land transactions; and (3) strengthening land governance and economic 
development through strategic Public Private Partnerships (PPPs). The LfP activity takes place in a total 
of seven regions, as well as the vicinity of Chiribiquete National Park.4 The IE of LfP focuses on ten of 
the eleven so-called pilot municipalities within these regions, each of which will receive all three 
components of the intervention. 
Together, the three intervention components aim to improve tenure security and access to markets and 
State services, increase private sector engagement and opportunities for generating income through licit 
activities, strengthen local land administration systems and infrastructure, reduce land conflict, and 
improve youth, ethnic, and women’s empowerment.
EVALUATION QUESTIONS
Table 1 lists the six evaluation themes addressed by the LfP evaluation. The evaluation questions, which 
the team developed in collaboration with USAID, focus on meeting USAID’s priority learning interests 
for this activity and were derived from the LfP theory of change.
1 LfP grew out of a pilot activity implemented in the municipality of Ovejas, Sucre, which was part of the Land and Rural Development Program 
(LRDP).
2 Program municipalities were selected from among a larger group of municipalities that have suffered disproportionately from the Colombian 
conflict and a lack of State presence, and were identified during the 2016 Peace Accords between the GoC and the FARC guerrilla group to be 
the focus of targeted investment and programming. This larger group of municipalities is known as the Programa de Desarrollo con Enfoque 
Territorial (Territorially Focused Development Plans - PDET) municipalities. 3
Land titles are issued by GoC, while LfP provides all necessary inputs for titling and works with relevant entities to ensure the titles are 
processed. LfP’s stated goal is to facilitate titling for 100 percent of eligible plots, excluding plots that may be located on protected public lands, 
in hazardous areas such as flood zones, encroach upon indigenous reserves, or otherwise found to be ineligible for titling.
4 Southern Tolima, Montes de Maria, Meta, Catatumbo, Tumaco, Northern Cauca, and Bajo Cauca
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
vii
5 Noting that for rural parcels, the system may be better characterized as deconcentrated rather than decentralized, as municipalities do not 
have jurisdiction for land administration in rural areas. Still, LfP aims to work with municipalities to provide related access and motivate 
registration of changes for rural areas as well.
TABLE ES1: EVALUATION QUESTIONS AND THEMATIC AREAS OF INVESTIGATION
THEME EVALUATION QUESTION
Impacts of LfP 
on individual 
households
1. What are the effects on households of strengthened land governance and property rights via 
LfP’s land titling (and other types of formal documentation, where applicable) and coupled 
land administration capacity building? 
a. Tenure security
b. Land use, investment, and illicit crop cultivation
c. Improved household wellbeing, food security, and poverty
d. Credit access and land market activity
e. Youth, ethnic, and women’s empowerment
Heterogeneity 
of Impacts on 
Household￾level 
Outcomes by 
Beneficiary and 
Context 
Characteristics
2. How do the impacts of the LfP activity on key outcomes differ for:
a. Female-headed households or joint households
b. Youth-headed households
c. Afro-Colombians, members of indigenous communities, and ethnic minority 
households
d. Households victims of the civil conflict
e. Households in coca-producing areas compared to those in areas with no coca
3. Do women recipients of land titles experience improvements on par with men for: 
a. decision-making regarding land use
b. agricultural productivity 
c. incomes
Impacts of LfP 
on Land 
Administration 
and 
Governance
4. To what extent did LfP activities improve municipal level local government self-reliance and 
public service delivery?
5. At the municipal level, what are the impacts of strengthened local government land 
administration capacity coupled with massive land titling on municipal tax revenue?
a. How do mayor’s offices use increased tax revenue gained through an increase in 
the number of title holders? 
b. Does increased tax revenue lead to improved public services for communities?
Impacts of LfP 
on Private￾Sector 
Engagement 
and Illicit Crop 
Substitution
6. Did the LfP activity increase private sector engagement in the pilot and corridor 
municipalities? 
a. If yes, how much funding and what types of activities were leveraged from the 
private sector through the LfP initiative?
b. If no, what were the major reasons and/or barriers for the private sector to engage 
effectively? 
7. What is the role of mobilized public sector funds and public-private partnership for local 
public goods in supporting sustainable reduction in coca production based on land titling?
8. Does massive land formalization contribute to more sustainable licit crop substitution, as 
indicated by increased numbers of farmers undertaking substitution and a reduced area under 
cultivation of illicit crops? 
a. Once individuals receive a land title, how easily can they connect and improve their 
income opportunities through the alliances and opportunities created through 
Component 3?
Implementation 
Effectiveness
9. How successful was the approach to adaptation of the strategies and lessons learned from 
the Ovejas pilot?
10. How effective was the strategy to engage the private sector in formalization in scaling 
progress or sustaining formality?
11. How effective was the USAID model for establishing municipal land offices within local 
government to develop and maintain a decentralized5, sustainable land administration and 
management system in municipalities (includes looking at effectiveness of land tax collection 
and tax revenue, among others)?
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
viii
EVALUATION DESIGN
This evaluation is designed as a mixed-methods study, using a quasi-experimental difference-in￾differences (DID) approach coupled with statistical matching to select the comparison group. The IE is 
complemented by pre-post qualitative data analysis based on document reviews, interviews, and group 
discussions to better understand mechanisms for observed impacts and assess broader implications of 
the LfP activity.6 
The household survey sample for the evaluation was designed to consist of 15 households 
per community across 10 communities per municipality, for a total sample of 150 households per 
municipality. This yields a total sample of 3,000 households across ten LfP and ten comparison municipalities.
To select communities for the sample, the evaluation team first divided each of the ten LfP municipalities into 
rural and urban communities (veredas and centros poblados, respectively) and then used a stratified 
probability proportional to size sampling design7 to select ten communities from each LfP municipality.
8 A 
random sample of households was then selected from each community, such that the expected number of 
completed interviews after accounting for non-response would be approximately 15.
As part of the quasi-experimental DID approach, the evaluation team constructed a comparison group
for using a three-step statistical matching technique9 — municipal-level matching in the first 
step, community-level matching in the second step, and household-level matching in the final step. The 
team used genetic matching10 at the municipality and community levels to optimize balance across the 
LfP and comparison groups on a set of key variables and select 100 matched communities across the 10 
matched comparison municipalities. Households were then sampled from these 100 communities using 
the same household sampling process used for LfP communities. 
6 See Evaluation Matrix in Annex F for additional details. 7 All communities in each municipality were stratified by urban/rural status. Communities were randomly selected within rural or urban strata 
with the probability of selection proportional to their population size. This design generates a representative sample for each municipality, while 
ensuring representation from both rural and urban communities.
8 Together, these communities comprised the totality of each municipality, except for the municipal seat (i.e., the town or small city where the 
municipal government is headquartered, and which is generally much larger than any other community in the municipality). 9 The matching strategy is described in detail in Annex B. 10 Genetic matching reduces the statistical difference between the treatment and comparison groups by reweighting the observations in the 
comparison group. It is often superior to traditional propensity score matching (PSM), in that it produces better balance between groups, and 
avoids known problems of PSM, including model dependence and biased results.
TABLE ES1: EVALUATION QUESTIONS AND THEMATIC AREAS OF INVESTIGATION
a. What strategies or technology innovations were most effective for harmonizing the 
collection and management of cadastral information and land registry records?
b. What is the effect of technology changes (e.g., document digitization) on 
institutional capacity? Are there any notable effects these institutions faced as they 
moved from a large staff / high manpower model to one more reliant on improved 
technology?
Overarching 
Lessons
12. Did LfP implementation of massive land titling lead to more sustainable illicit crop 
substitution? In what keys ways?
13. In what ways did the LfP activity contribute to land-related Peace Accord agenda objectives, 
including those specified under Chapter 1?
14. In what ways did the LfP activity contribute to broader GoC agrarian land rural reform and 
rural development objectives?
15. To what extent did GoC adopt and scale up public policy inputs and recommendations from 
LfP?
16. Were there any unintended broader consequences of the land titling component, beyond 
those related to LfP objectives?
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
ix
BASELINE DATA COLLECTION AND HOUSEHOLD SAMPLE
Baseline qualitative and quantitative data for this evaluation was collected during April – June 2021 by 
Colombian data collection firm Sistemas Especializadas de Información (SEI), with supervision by NORC 
at the University of Chicago, and support from the Mission’s MEL Activity in Bogotá that is implemented 
by the Panagora Group. The final baseline household sample was 2,965 households surveyed across 192 
communities (1,512 households from 100 LfP communities and 1,453 households from 92 comparison 
group municipalities), after accounting for non-response and communities that could not be accessed 
due to security problems. Qualitative data was collected in all municipalities, and consisted of 20 group 
discussions (GDs) across two communities in each of ten selected municipalities, 20 group key 
informant interviews (KIIs) with community leaders from communities across the 20 survey 
municipalities, and KIIs with six officials across three GoC agencies.
The household survey was administered to the household head (if available) or another adult 
knowledgeable about the household’s land use decisions. Female-headed households comprised 27.5
percent of households in the sample in treatment communities, and 26.2 percent of sampled households 
in comparison communities. Households were overwhelmingly rural: 75.5 percent of sampled 
households in treatment communities were located in rural veredas (i.e., dispersed rural communities) 
and 24.5 percent were located in urban centros poblados (i.e., small towns other than the municipal seat). 
Broad land use and ownership context was also similar across sampled LfP and comparison group 
households, including on: engagement in agriculture on plots under the household’s control (77.2 
percent of sampled households in treatment communities and 84.1 percent of those in comparison 
communities); ownership of at least one plot (71.2 percent of sampled households in treatment
communities and 70.3 percent of those in comparison communities); and ownership of at least one plot 
that already has a formal land title (22.0 percent of sampled households in treatment communities and
20.6 percent of those in comparison communities). 
Key baseline findings are summarized below for the LfP treatment sample, across the five major themes 
of interest for the evaluation. Differences with the comparison group sample are discussed in the report 
body for informational purposes, noting these results at baseline are obtained on the unmatched sample 
at the household level. Because we are able to obtain a well-balanced sample via statistical matching on 
household-level characteristics (see discussion on this in the Balance and Power section), any differences 
presented at this stage are not a concern for the validity of the IE, 
FINDINGS 1: HOUSEHOLD WELL-BEING, FOOD SECURITY, AND POVERTY
The baseline data suggests that a high proportion of LfP households in the evaluation sample are likely 
below the national poverty line, while household livelihood strategies are equally distributed across 
farming and wage labor (which could also be in the agricultural sector). Most households in the sample 
are food secure, but food insecurity does reach a level of concern in three of the regions. Household 
participation in non-farm income generating activities is low, at 17.6 percent, and the average combined 
monthly income reported across all non-farm sources is $230 USD per month11. Credit access is also 
low, at 17.2 percent of the sample, but among credit takers the average loan size is fairly high (at 
$2,408.2 USD) and the source is primarily from formal banks (56.9 percent of households who took 
loans). Qualitative findings highlighted a perception among GD respondents that formal land titles would 
increase both their access to and the amount of loans they could obtain from formal banks. 
LIVELIHOODS
• The average likelihood of living below the 2016 national poverty line ($2.73 USD per day, after 
adjusting for inflation) for sampled households in treatment communities is 42.9 percent. 
11 Here and throughout the report, we use the 2020 average exchange rate of $1 USD = $3,691.3 COP.
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
x
• 39.2 percent of sampled households in treatment communities earned their main source of 
livelihood from crop cultivation, while another 39.2 percent earned their livelihoods primarily 
from wage labor.
ACCESS TO CREDIT
• 25.8 percent of sampled households in treatment communities said they had a need for credit 
during the 12 months prior to the survey, and 17.2 percent said they had received credit from a 
source outside the household.
• Among households who took loans over the 12-month period prior to the survey, the average 
loan amount was $2,408.2 USD.
• Among those who took loans, 56.9 percent of sampled households in treatment communities
took loans from formal banks, while 10.4 percent took them from cooperatives and 23.9 
percent took loans from informal sources, such as loan sharks or family and friends.
• Qualitative findings support the program’s theory of change that formal land titles will improve 
access to credit. Access to credit was commonly cited as the most important benefit of having a 
land title in qualitative discussions, as GD participants felt a formal title would both improve the 
likelihood that a loan would be approved and increase the amount of credit they could obtain 
from formal banks. 
FOOD SECURITY
• 92.3 percent of sampled households in treatment communities suffered from little to no hunger, 
while 7.7 percent suffered from moderate to severe hunger. 
• To the extent that hunger exists, it is spatially concentrated in a few regions. Rates of moderate 
to severe hunger were highest in Coastal Nariño (25.1 percent), Montes de Maria (15.1 
percent), and Bajo Cauca (9.6 percent).
NON-FARM ACTIVITIES
• Overall, 17.6 percent of sampled households in treatment communities operate a non-farm 
income generating activity, such as a small business, household-based enterprise, or informal 
sector entrepreneurial activity.
• The average monthly revenue for entrepreneurial non-farm income generating activities, such as 
home-based businesses, running a mototaxi, or a non-farm small business, is $178.0 USD 
($657,000 COP) for sampled households in treatment communities. Overall, these activities 
tend to be oriented for sale to the final consumer (84.3 percent) and are carried out within the 
household premises (49.8 percent).
FINDINGS 2: LAND USE AND AGRICULTURAL PRODUCTION
Households in the sample manage a relatively small number of plots on average, and although the 
average total area of land managed is quite high, at 17.4 hectares, the total landholding varies widely 
across households in the sample. Nearly half of the sample has one hectare of land or less. Land is 
primarily used for agriculture, but households do not appear to be strongly diversified in their 
agricultural activities as nearly all households rely on three or fewer crops or livestock activities. There 
is a wide distribution in the amount of agricultural income reported by households, with nearly a third of 
households reporting $5 USD or less per month. Still, it is not uncommon for households in the 
treatment sample to make land investments aimed to improve productivity. The baseline results also 
provide some evidence to suggest that households with registered land titles may be more likely to 
make such land investments, or to make higher value investments (households with a title invested 
$977.9 USD in their plots, compared to $373.9 USD for those without), while highlighting that 
households perceive a lack of resources to be a key barrier to land investment. 
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
xi
LAND HOLDINGS
• On average, sampled households in treatment communities managed an average of 1.4 plots per 
household, and used an average of 0.9 plots for agricultural purposes per household. Across all 
plots managed by households under any tenancy regime, sampled households in treatment 
communities held an average of 17.4 hectares of land in total, and had an average of 8.4 hectares 
under agricultural use. The median for total land holdings among sampled households in 
treatment is 1.5 hectares, and with a median of 1.0 hectares for land under agricultural use.
• The distribution of land holdings across households in the sample is highly unequal. Overall, 31.9 
percent of all sampled households had less than 0.25 hectares, and 45.9 percent held one 
hectare of land or less, while 17.5 percent held more than ten hectares. Households with no 
registered title to any plot were also significantly more likely to have less than 0.25 hectares of 
land compared to households with at least one registered title.
LAND USE
• Among plots controlled by sampled households in treatment communities, 65.0 percent were 
used for agricultural purposes. 
• Among agricultural plots, the most important uses of land reported by households were 
permanent crops (21.1 percent of plot area), transitory crops (18.3 percent of plot 
area), unused or fallow land (13.9 percent of plot area), single species livestock (13.9 percent of 
plot area), and dwelling (12.1 percent of plot area).
• The primary reason households cited for leaving land unused was a lack of financial resources
(57.7 percent of plots with unused land), which may support the program’s theory of change for
improved land use and productivity from land titling, if land titling helps households to have
improved access to credit.
AGRICULTURAL PRODUCTION
• Agricultural diversification among households in the sample is fairly low, as households are
generally reliant on a small number of agricultural or livestock activities. Among households in 
the sample that are engaged in agricultural activities, 29.0 percent engaged in only one crop or 
livestock activity; 26.8 percent had two crop or livestock activities; and 44.2 percent listed three 
crop or livestock activities.
• Overall, the most common crops farmed or livestock raised by sampled households are plantain 
(33.8 percent), roots and tubers (32.8 percent), and poultry (27.9 percent).
• On average, sampled farming households in treatment communities earned $466.0 USD ($1.7
million COP) per month in revenue from all agricultural and livestock activities on their 
agricultural plots.12
• The distribution of income from agriculture across farming households in the sample is highly 
unequal. Nearly a third of farming households report almost no earnings (i.e., between $0 
and $5 USD per month) from agriculture, suggesting their agricultural production is almost 
completely for consumption within the household. Another 49.0 percent earn between $6 and 
$250 USD per month from agriculture, while just 20.2 percent of sampled households in 
treatment communities reported earning more than $250 USD per month from agricultural 
activities on their plots.
PRODUCTIVE INVESTMENTS
• Among all plots controlled by sampled households in treatment communities, households said 
they had made productive investments for 41.1 percent of plots in the treatment sample over
12 Note that mean revenue from agricultural activities for farming households is higher than mean income for all sources for all households
reported in Findings 1. This has two main explanations: (1) farming households have more land and higher incomes than other households, on 
average; (2) the figures for agricultural revenue and total household income come from two different sections of the survey, one asking about 
sales revenue from farm production, and the other asking about household income.
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
xii
the last 12 months. At the household level, 48.1 percent of sampled households in treatment 
communities had made an investment on at least one plot under their control in that same time 
frame. The most common types of investment that households made were housing (15.7 
percent of plots) and fruit, timber, or other commercial trees (7.9 percent).
• Households with a registered title to at least one plot had made significantly greater financial 
investment in their land ($977.9 USD) than those without a title ($373.9 USD), on average. This 
is driven, in part, by the increased share of households without a title who made no investments 
(63.5 percent) compared to households with a title (53.1 percent). Among households who had 
made any investment in their plots, the value of the investment was also higher, on average, for 
those with titles ($1,872.9 USD) compared to those without at title ($889.1 USD), though this 
difference is not statistically significant.
• The most common reason households cited for not investing in their land was a lack of 
resources (64.3 percent), supporting the program logic that improved access to credit through 
land titling may also lead to greater land investment.
FINDINGS 3: LAND ACQUISITION AND TENURE SECURITY
At the plot level, the large average size of plots also masks highly unequal distribution of plot sizes 
across households in the sample. Plots that already have formal titles are larger in size, on average, and 
also are more likely to have been acquired through purchase. Households have no documentation 
confirming the household’s occupancy rights for nearly half (46.7 percent) of plots that are owned or 
under de facto occupation by households in the sample. The baseline findings also suggest that most 
households that do currently have a title for their plots have already formally registered it. At the same 
time, findings pointed to a fair amount of confusion among respondents around the differences between 
registered and unregistered titles, reasons and resources required to pursue formalization, and services 
provided by land governance institutions. Perceived tenure security across plots was fairly high, with 
respondents reporting they felt unlikely to lose ownership or use rights over the next five years for 80 
percent of plots (respondents expressed tenure insecurity for 16 percent of plots), and the incidence of 
land disputes in the twelve months prior to survey was very low, at around two percent. 
LAND CHARACTERISTICS / LAND OVERVIEW
• The average plot size for plots belonging to sampled households in treatment communities is 
12.4 hectares.
• The distribution of land holdings across the sample is highly unequal, and plots with formal titles 
tend to be larger (19.3 hectares) than those without (7.7 hectares).
• The distribution of plot sizes even within the group of untitled plots is also highly unequal: nearly 
half of all plots without a registered title (46.4 percent) are less than 0.25 hectares, and 66.4 
percent are one hectare or less, while 9.0 percent of all plots without a registered title are 
larger than 10 hectares (including 5.3 percent that are larger than 20 hectares). This suggests the 
vast majority of titles issued through LfP’s work will go to small landowners, though most of the 
newly titled land may belong to a relatively small share of households.
MODE OF ACQUISITION AND TENURE
• Titled plots are significantly more likely to have been acquired through purchase (65.2 percent 
of titled plots, 45.4 percent of untitled plots), and significantly less likely to have been acquired 
through inheritance (27.6 percent of titled plots, 40.5 percent of untitled plots), occupying 
vacant land (0.1 percent of titled plots, 2.2 percent of untitled plots), or through exchange or as 
gifts (1.1 percent of titled plots, 3.4 percent of untitled plots).
• Qualitative interviews with communities suggest that previous GOC efforts at land formalization 
have come undone over time as plots are inherited, sold, or transferred.
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
xiii
LAND RIGHTS
• Among plots owned or under de facto occupation by households with a registered title for the 
plot, households said they had the right to sell 91.1 percent of these plots, and had the right to 
bequeath 95.5 percent; this compares to 66.3 percent and 86.0 percent, respectively, for plots 
without a formal title. 
LAND DOCUMENTATION
• Among all plots owned or under de facto occupation by respondents in the sample, 46.7 
percent had no land document to verify the household’s right to occupancy; the figure was 
significantly higher for plots under usufruct, rental, or sharecropping agreements, at 97.6 percent 
of such plots. Documents held by households for plots owned or under de facto occupation 
included registered titles (26.0 percent), unregistered formal titles (2.1 percent), other formal 
documents13 (18.9 percent), and informal or semi-formal documents14 (6.2 percent).
• Plots with any document besides a registered title are somewhat more likely to have only the 
male decisionmaker listed on the document (50.0 percent) than plots with a registered title 
(45.2 percent), but the difference is not significant. Similarly, plots with any other document are 
also somewhat more likely to have only the female decisionmaker listed on the document (29.2 
percent), compared to those with a registered title (28.3 percent), though the difference is not 
statistically significant. On the other hand, plots with a registered title are significantly more 
likely to have the male and female decisionmakers from the household listed jointly on the 
document (9.0 percent) than plots with some other document (5.9 percent).
• It appears far more common for either the male or female decisionmaker to be the sole person 
listed, than for the two to be listed together, regardless of the document type. This may be due 
to the fact that formal marriage is relatively uncommon among the sampled population.
• The most common reasons that respondents provided for not pursuing formalized land tenure 
were that it was not necessary (34.6 percent of plots) and that the household lacked the 
resources to formalize (29.8 percent). A large share (23.8 percent) also provided other 
responses, which often dealt with confusion or customs around the inheritance process for land.
TENURE SECURITY
• Among all plots, respondents said they were either “not at all likely” or “somewhat unlikely” to 
lose ownership or use rights to the plot in the next five years for 79.9 percent of plots 
controlled by sampled households in treatment communities.
• Among sampled households in treatment communities, respondents said they were either “not 
worried” or “not at all worried” they might lose the right to use the plot within the next five
years for 78.7 percent of plots.
• For 45.8 percent of plots controlled by sampled households in treatment communities,
respondents said they could not leave their land fallow for any period without losing their use 
rights to the land.
• Among sampled households with at least one formally titled plot, 4.4 percent reported having
lost their land against their will in the past 12 months for any reason (e.g., natural disasters, 
forced displacements, or other reasons), compared to 3.7 percent of households without a
formal title to any plot; this difference is not statistically significant.
• 3.0 percent of households with at least one registered title and 2.9 percent of those without had
experienced eviction threats during the 12-month period prior to the survey.
13 “Other formal documents” refers to a legal document that is not a title, such as a will, written contract, or non-title document from one of the 
national land agencies.
14 “Informal or semi-formal documents” refer to non-legal documents, such as a signed letter, extrajudicial statement, or verbal contract.
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
xiv
LAND DISPUTES
• There is a similarly low level of exposure to any type of land-related conflict or dispute in the
12-month period prior to survey for households with registered land titles and those without
(2.5 percent of households with registered titles and 2.2 percent of those without registered
titles).
LAND GOVERNANCE, LAND MARKETS, AND PUBLIC SERVICE DELIVERY
• For the statement “Land rights are clear and easy to understand”, 44.7 percent of sampled
households in treatment communities agreed with the statement, while 41.1 agreed with the
statement “Land rights are well-protected by authorities”.
• Among all households in the sample, 1.9 percent purchased any land in the municipality during 
the 12-month period before the survey. Among all households who purchased land, the mean 
amount of land purchased is 9.5 hectares. However, the distribution is highly unequal, with a 
median purchase of just 0.08 hectares.
• Among sampled households in treatment communities, 41.4 percent had usufruct, rental, or 
sharecropping agreements to occupy land belonging to someone else, while just 6.7 percent 
reported having such agreements in place for use of their own land by other households.
• Households reported low levels of satisfaction with local service delivery. Among sampled 
households in treatment communities, 29.6 percent said they were satisfied with the quality of 
the municipality’s roads, 4.6 percent said they were satisfied with the quality of irrigation 
infrastructure in their municipality, and 12.4 percent said they were satisfied with the quality of 
overall infrastructure services in their municipality.
FINDINGS 4: PRIVATE SECTOR ENGAGEMENT AND ILLICIT CROP SUBSTITUTION
Baseline findings suggest generally low trust in government institutions at municipal and other levels, 
while social capital in communities is high. Residents of coca-growing municipalities had particularly high 
levels of dissatisfaction and mistrust, but qualitative data also suggested that a lack of communication and 
widespread confusion on national land titling policies and institutions has contributed to general mistrust 
across the evaluation sample. In conjunction with this, household participation in alternative 
development programs and perceived benefits from public-private partnerships (PPP) is also low, which 
may pose challenges for LfP’s intended PPP activities, or at least suggest a need to strategize carefully on 
how to engender interest and set expectations among the target population. Findings on this theme also 
suggest potential challenges for the intended pathway to illicit crop substitution via land formalization. 
While a very low proportion of households admit to growing coca currently (2.8 percent), households 
with registered land titles have the highest prevalence of admitting to coca cultivation in the survey 
sample, but it is also possible that such households simply feel more secure admitting to coca cultivation. 
Given the complex and potentially diverse reasons why some households may be more or less willing to 
admit to growing coca, these results should be interpreted with caution.
TRUST IN INSTITUTIONS
• Levels of trust in most institutions are low. Just 28.4 percent of sampled households in
treatment communities trust judges, 43.2 percent trust their local Community Action Board 
(Junta de Acción Communal), 25.5 percent trust their town hall (Alcaldía Municipal), and 43.1
percent trust the Municipal Land Office (Oficina Municipal de Tierras).
• Just 20.0 percent of sampled households in treatment communities agreed with the statement 
“There is little corruption in the municipal government”, just 26.3 percent agreed with the 
statement “I am very satisfied with the municipal government’s work in rural development”, and 
just 34.5 percent agreed that they “trust the municipal government works on behalf of the 
interests of all its citizens”.
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
xv
• Qualitative discussion highlighted that patchwork policies and changing laws and institutions 
around land titling have not been properly communicated to communities and have created a 
situation where households do not know what is happening at the national level, eroding trust.
• Qualitative discussions also highlighted high levels of distrust in government from participants in 
coca-growing municipalities. It is clear that for some, years of militarization of their 
communities, criminalization of their main source of livelihood, and promises that never come 
to fruition have resulted in a hostile stance towards the government in Bogota. Changes in the 
qualification criteria for participating in the national government’s crop substitution program 
(PNIS) appear to be exacerbating the situation.
SOCIAL CAPITAL
• Among sampled households in treatment communities, 68.3 percent agreed with the statement, 
“I can trust most of the neighbors in my community”, 88.7 percent agreed with the statement, 
“If someone in the community needs help, we are all willing to help,” and 87.5 percent agreed 
with the statement, “I could work on a community project that doesn’t benefit me directly”.
• 49.9 percent of sampled households in treatment communities do not belong to any 
organization. The most common organizations that households do belong to include the local 
community action board, ethnic community councils, and local civic organizations.
PUBLIC-PRIVATE PARTNERSHIPS15
• Among sampled households in treatment communities, 12.1 percent agreed that their 
household had benefitted from government development projects in the past 12 months, 8.8 
percent agreed that it had become easier to find a job in the municipality in the past 12 months, 
and 7.7 percent said they had participated in or benefited from productive projects in the past 
12 months.
PARTICIPATION IN ALTERNATIVE DEVELOPMENT PROGRAMS
• Overall, 6.2 percent of sampled households in treatment communities said they had ever 
benefitted from alternative development programs. Participation is spatially 
concentrated. Among all sampled households, the areas with highest participation in alternative 
development programs are Tumaco and Coastal Nariño (16.8 percent), Bajo Cauca (12.5 
percent), Meta (3.5 percent), and Catatumbo (2.7 percent).
• Among all households who said they had benefitted from alternative development programs, the 
most common programs were Forest Guarding Families (33.1 percent), followed by PNIS (27.8 
percent), and Voluntary Substitution Agreements (11.3 percent).
• Participation in Alternative Development Programs in treatment communities increased 
dramatically between 2016 and 2018, corresponding with the first years of implementation of 
the Colombian peace accords. Participation dropped substantially after 2018.
ILLICIT CROP CULTIVATION
• Among sampled households in treatment communities, 9.0 percent admitted to ever having 
cultivated coca, while 2.8 percent admitted to growing coca currently. Among those households 
currently cultivating coca, sampled households in treatment communities said they had 1.4 
hectares under cultivation, on average.
• Due to complex issues with household self-reporting, refusals, and security in coca-growing 
municipalities, coca results from the household survey sample are best interpreted as lower-end 
estimates of the true prevalence. The results suggest at least 5.9 percent of households in coca 
growing municipalities who own at least one plot are currently cultivating coca, including at least 
8.0 percent of those with a formal title to at least one plot, and at least 4.8 percent of those 
15 At baseline, we looked at household participation in development projects in general. 
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
xvi
without any formal title. However, the issues highlighted here make it difficult to draw firm 
conclusions on the relationship between tenure and coca at baseline.
• Quantitative and qualitative findings highlight several threats to the theory of change for
reducing illicit crop cultivation through land titling: (1) the risk of expropriation must be 
sufficiently high to dissuade titled owners from illicit crops; (2) it is unclear whether LfP will be 
able to reduce illicit crop cultivation on vacant public lands or national parks; (3) pressure from 
armed groups may give households little choice but to continue cultivating coca. How these 
risks evolve depends to a great extent on factors outside the program’s control, including 
enforcement by the Colombian government, dynamics of the armed conflict, how communities 
continue to perceive and participate in the PNIS program, and whether the GoC decides to 
resume aerial spraying.
FINDINGS 5: WOMEN’S EMPOWERMENT
Findings from the wives’ module of the household survey provide several indications of lower land 
tenure security for women in the survey sample compared to men. Women in the survey sample are 
less commonly named on registered land titles or any land documents, while male and female 
respondents alike are less likely to see women as having a right to sell household land. Men are also 
more likely than women to express knowledge of their land rights, where to obtain help for land 
conflicts, and to feel they would have access to representation if a land dispute should arise. There are 
also large disparities by gender with respect to participation in income-generating activities and 
economic decisions, but less so with respect to control over income and asset ownership. Male primary 
decisionmakers in the survey sample are more likely to report obtaining loans, and their average loan 
size (among loan takers) is nearly twice that of female primary decisionmakers.
WOMEN AND LAND TENURE
• 31.3 percent of primary male decisionmakers appear on any land document, compared to 22.3 
percent of primary female decisionmakers. 15.5 and 11.5 percent of primary male and female 
decisionmakers, respectively, appear on a registered land title.
• Male and female decisionmakers appear to have a similar perception of who within the 
household has a right to sell land. 46.6 percent of all adult male decisionmakers and just 27.0 
percent of adult female decisionmakers described themselves as personally having the right to 
sell any of their plots, representing a statistically significant gap of 19.6 percentage 
points. Similarly, just 25.6 percent of male decisionmakers said that the primary adult female 
decisionmaker in the household had the right to sell any plot, while 37.0 percent of female 
decisionmakers said the same for the primary adult male decisionmaker.
• Patterns appear somewhat different for how male and female decisionmakers describe who in 
the household has a right to bequeath land. 40.7 percent of male decisionmakers said the female 
decisionmaker in the household had a right to bequeath land, while just 28.3 percent of female 
decisionmakers said the male decisionmaker had the same right. The qualitative findings suggest 
that land is often inherited without any written will from the deceased, and that it is common, 
particularly for men, to have children with multiple partners, along with unions through common 
law, as opposed to formal marriage. These dynamics contribute to tenure insecurity for women.
• Female decisionmakers were significantly more likely to say the household was not likely to lose 
its rights to any plot within the next five years (79.1 percent of females, 71.9 percent of males), 
and to say they were not worried about the household losing its rights to any plot over the 
same time period (78.8 percent of females, 73.0 percent of males). On the other hand, male 
decisionmakers were significantly more likely to say the household could leave its land fallow 
indefinitely and would never lose its land use rights (28.0 percent of males, 14.5 percent of 
females).
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
xvii
• A statistically significant greater share of male decisionmakers (43.1 percent) agreed with the 
statement “I know more about my land rights now than I did last year”, compared to female 
decisionmakers (34.4 percent). Similarly, a significantly greater share of male decisionmakers 
(54.9 percent) agreed with the statement “I know where to go if I have a conflict about my land” 
than female decisionmakers (42.4 percent), while 50.2 percent of male decisionmakers agreed 
with the statement “I have access to legal representation if I have a land dispute”, compared to 
35.9 percent of female decisionmakers.
PARTICIPATION IN PRODUCTIVE ACTIVITIES AND ECONOMIC DECISIONS
• Among all plots belonging to sampled households, 69.2 percent were under the primary 
responsibility of the primary male decisionmaker in treatment communities, while female 
primary decisionmakers were the main person responsible for 29.0 percent of all plots.
• The survey asked male and female decisionmakers about whether they participated in each of 
six income generating activities, including food crop farming, cash crop farming, livestock 
farming, non-farm economic activities, wage employment, and fishing or fishpond culture. 40.3 
percent of all female decisionmakers indicated they participated in none of the six activities, 
compared to 10.4 percent of male decisionmakers. On average, female decisionmakers 
participated in 1.0 activities, compared to 1.9 activities for male decisionmakers.
• The evaluation team created an indicator of empowerment in productive decisions, using the 
template provided by the Abbreviated Women’s Empowerment in Agriculture Index (A-WEAI).
Among male and female decisionmakers in treatment communities, 72.4 and 47.5 percent, 
respectively, met the empowerment in productive decisions criteria.
CONTROL OVER INCOME
• The evaluation team created an indicator of empowerment in control over income, using the 
definition from the A-WEAI. Among male and female decisionmakers in treatment communities, 
97.7 and 85.5 percent, respectively, met the empowerment in control over income criteria.
ASSET OWNERSHIP
• Male decisionmakers described themselves as personally owning more items (5.6 items) than 
female decisionmakers (5.3 items). The difference is statistically significant. Male and female 
decisionmakers are also different in terms of which assets they own, with males more likely to 
own agricultural land, large livestock, and farm equipment, and females more likely to own 
poultry and small livestock, non-farm business equipment, and consumer durables.
• The evaluation team created an indicator of empowerment in asset ownership, using the 
definition from the A-WEAI. Among male and female decisionmakers in treatment communities, 
99.0 percent of respondents in each group met the empowerment in asset ownership criteria.
ACCESS TO CREDIT
• Among all male and female decisionmakers, 9.6 and 7.7 percent, respectively, took out any loan 
over the 12-month period before the survey. The difference is statistically significant.
• On average, male decisionmakers obtained $265.4 USD ($1.0 million COP) in loans over the 
12-month period before the survey ($2,753.1 USD or $10.2 million COP for male 
decisionmakers who took loans), compared to $118.5 USD or $400,000 COP for female 
decisionmakers ($1,546.9 USD or $5.7 million COP for female decisionmakers who took 
loans). These differences are statistically significant.
• Few differences were observed by gender in terms of where men and women access loans from. 
Although men who took loans were more likely to obtain those loans from the Agrarian Bank 
(30.3 percent) than women (24.9 percent), women were more likely to obtain loans from other 
formal banks besides the Agrarian Bank (37.1 percent) than men (32.2 percent). These 
differences are not statistically significant, and suggest that men and women who obtain loans 
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
xviii
are approximately equally likely to obtain those loans from any formal bank (62.5 and 62.0 
percent, respectively).
BALANCE AND POWER
The evaluation team used the baseline data to revisit statistical assumptions related to the methodology 
proposed for this evaluation, including the ability to achieve balance between sampled households in 
treatment and comparison communities on key outcome variables and covariates, using the planned 
statistical matching strategy planned by the team during the evaluation design phase. The team also 
updated power calculations for the IE using the baseline data, and confirmed that the evaluation is well￾powered to detect many key outcomes. Balance tests suggest the treatment and comparison groups are 
well balanced after statistical matching, providing an appropriate counterfactual for the evaluation. 
However, the updated power calculations suggest that the evaluation will not be powered to detect
impacts of LfP on the proportion of households that engage in coca production via the intended 
statistical analysis. This is discussed in the report body and alternative strategies are suggested.
CONCLUSIONS 
Several of the baseline findings support the LfP programming logic and elements of the intended theory 
of change, although some findings call attention to issues that may present challenges for LfP’s intended 
objectives, and/or require adaptive programming to address as implementation progresses. It is also
important to recognize certain limitations of the baseline study, including the sensitive nature of some 
questions regarding illicit crop cultivation and threats from armed groups, which can affect the reliability
of responses or respondent willingness to respond. The evaluation team also notes that the apparently 
low level of understanding of the legal and institutional landscape around land titling among some 
respondents could mean that households’ self-reported land tenure status at baseline is not always 
accurate.
Key areas of support include:
• There is scope to greatly improve the proportion of households with formalized land rights, and 
LFP’s programming emphasis on strengthening the culture of formalization over time among the 
target population also appears to be highly relevant.
• Linkages between titling and credit, and potential improvements to agricultural productivity.
• Linkages between formalized titling, strengthening women’s land rights and women’s 
empowerment.
Key areas where the baseline findings underscore potential challenges for LfP’s intended programming 
include:
• Results suggest some challenges to the theory of change that land titles will lead to widespread 
improvements in tenure security, given findings from baseline related to sources of tenure 
insecurity, experience with prior formalization efforts, and eligibility for titling. For example, 
some communities have developed their own informal systems for land governance in the 
absence of State presence, and findings also suggest that holding a land title may not address 
some common sources of tenure insecurity, such as land expropriation by armed groups.
• Colombia’s highly unequal land distribution may pose a challenge for program beneficiaries to 
leverage land titles into improved livelihoods, as baseline findings suggest that many households
have insufficient land to earn a living from even if these holdings are formalized. The benefits of 
formalization could then also accrue disproportionately to a smaller share of households that 
have medium to large land holdings.
• Low levels of satisfaction with local service delivery and infrastructure present both a challenge 
and an opportunity for LFP programming. The program’s focus on local service delivery and 
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
xix
infrastructure as part of its activities appears well-placed in this respect, but the current status 
quo does present a substantial challenge given that market access will be needed to improve 
livelihoods and trust in the government infrastructure will be needed to support formal 
transactions.
• Land restitution is contentious. Given that the cadaster updates will serve as an input to the land 
restitution process, potential challenges of accelerated land restitution for LfP include the 
possibility of increased social tension.
• Trust in government is low, particularly in coca-growing areas.
• To be effective at reducing coca cultivation, LfP will likely need to be accompanied by a credible 
threat of expropriation and viable alternative livelihoods.
RECOMMENDATIONS FOR SUBSEQUENT DATA COLLECTION FOR THE IE
The baseline data collection experience and findings also highlight some key issues that will be important 
to address for the endline data collection round:
• Supplemental strategies to strengthen the evaluation’s reporting on changes to coca production, 
including an expanded focus on this in the qualitative data collection, and potential use of 
supplemental geospatial data to inform on coca expansion.
• Expanded role for qualitative data collection at endline, to strengthen the ability to interpret endline 
results. This should particularly focus on expanding the number of group discussions, holding 
gender-segregated discussions, and expanding coverage on issues related to women’s land rights, 
tenure security and empowerment. It should also aim to help inform on potential differences in 
trends across the diverse regional contexts where LfP operates.
• Modifications to the endline survey instrument, including: equipping enumerators with a picture of 
an escritura publica and adding a question on the survey to confirm this and other document 
types stated by respondents; clarifications reduce respondent confusion on whether poultry 
raising is included as part of livestock activities on the wives module; and targeted work to 
shorten the instrument (for example, efficiencies can be gained by reducing the level of detail 
collected about loans, and certain details on the crop roster).
• Ensuring at least 6 months lead time for endline data collection and evaluation team procurement, to 
ensure sufficient time to plan and engage in the data collection preparations with the endline 
firm. This is especially important given the challenging security context in the LfP implementation 
and comparison communities for this evaluation.
Independent data firm procurement via the Mission’s MEL activity and collaboration with the evaluation 
team worked well at baseline, and can be replicated in subsequent rounds.
• Maintain the same community definitions used at baseline. GoC may change maps over time, but 
subsequent data collection activities and rounds of analysis should continue to use the 
community definitions from baseline data collection. These community definitions considered 
the diverging official cartographies from DANE and IGAC, how LfP would use these 
cartographies to conduct its activities, and the realities of how fieldwork would need to be 
conducted.
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
1
INTRODUCTION
This report provides baseline information on the study population and the socio-economic context for 
the mixed-methods impact evaluation (IE) of the “Land for Prosperity (LfP)” activity in Colombia. The 
evaluation aims to provide an evidence base for outcomes of LfP with respect to strengthening land 
rights and land governance, reducing illicit crop cultivation, and enhancing local livelihoods. The 
evaluation was commissioned by USAID’s Land and Resource Governance (LRG) unit in the Bureau for 
Development, Democracy and Innovation’s Center for Energy, Environment and Infrastructure 
(USAID/DDI/EEI), and is led by NORC at the University of Chicago under the CEL Project. 
This document provides findings from the baseline data collection for the evaluation, including 
background context on key demographics, household characteristics, and baseline measures on 
outcome variables. The report also examines balance across LfP (treated) and comparison communities 
for the IE components of the evaluation, and revisits power calculations from the Evaluation Design 
Report (EDR) using parameters from the baseline data.
CONCEPTUAL FRAMING, EVALUATION PURPOSE AND 
EVALUATION QUESTIONS
LAND TENURE, PRODUCTIVITY, CONFLICT AND ILLICIT CROPS BACKGROUND
Colombia is characterized by a dualistic distribution of land ownership, which can be traced back to the 
colonial period (Ibáñez and Muñoz-Mora, 2010). A relatively small share of plots have formalized titles 
(mostly medium- and large-size plots) and operate in a formal land market, while a significant number of 
small plots have informal land rights, in large part due to the unplanned expansion of the agricultural 
frontier (C. C. LeGrand and Valencia Goelkel, 2016). For centuries, small peasants, formerly enslaved 
groups, and even entire communities found an alternative to land access by venturing beyond the 
agricultural frontier to inhabit unused land, driven by several factors: (i) conflicts between peasants and 
large landlords; (ii) land availability; and (iii) individual motivation (Machado and Vivas, 2009; Ibáñez and 
Muñoz-Mora, 2010). Forests were logged to make room for new villages, roads, and rural communities,
creating economies based on subsistence agriculture (C. LeGrand, 1988). In some regions, the State 
promoted this as a strategy to expand and to boost agricultural production. Public idle land (baldios) 
with economic potential was formally assigned to capitalists and entrepreneurs looking to establish an 
agricultural system with potential for export markets such as bananas, cocoa, and coffee, among others 
(C. LeGrand, 1988). The correlation between regions with insecure land tenure systems and low State 
presence has created an opening for land-related disputes, expropriation, presence of illegal actors, and 
other illegal activities (Centro Nacional de Memoría Histórica, 2014; Ibáñez and Muñoz-Mora, 2010; 
Machado and Vivas, 2009). 
On average, approximately 22 percent of all private rural land in Colombia has no formal title, of which 
89 percent are small plots of less than 20 hectares (Muñoz-Mora et al., 2018). In recent decades, the 
consolidation of the drug economy has also prolonged and deepened the persistence of insecure land 
tenure systems. Thanks to the country’s abundant natural resources, high prevalence of poverty, 
territorial control by illegal actors, and low public enforcement, the rural sector has become the perfect 
environment to grow coca and other illicit crops (Dávalos et al., 2011). In this context, informal tenure 
systems both push households into illicit crop cultivation by making it harder to obtain credit, make 
productive investments, and make a living from licit crops, and weaken the State’s ability to punish those 
growing illicit crops. Despite decades of effort under the War on Drugs and billions of dollars spent on 
counter-narcotics operations in the country, cocaine production and land under coca cultivation have 
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
2
been increasing since 2012, and Colombian cocaine production is currently at all-time highs, surpassing 
the 1980s and 1990s under Pablo Escobar and the Medellin and Cali Cartels.16
Yet the traditional land administration system has been slow to improve. According to 2019 data from 
Colombia’s Instituto Geográfico Agustín Codazzi (Agustin Codazzi Geographic Institute - IGAC), nearly 87
percent of Colombian municipalities (954 out 1100) had outdated cadastral information (last updated 
more than 12.2 years prior), and for 7 percent of the territory, land administration information had 
never been collected (Departamento Nacional de Planeación, 2019). 
In recent years, the Government of Colombia (GoC) has developed a new land tenure approach for 
rural property, known as the Ordenamiento Social de la Propiedad Rural (Participatory Rural Land Use 
Management Code – OSPR). OSPR was defined by Decree 902 of 2017 and grew out of the legal 
framework that was part of the peace process between the GoC and the FARC (Fuerzas Armadas 
Revolucionarias de Colombia) guerrilla group. The policy moves formalization of land tenure from a 
demand-driven model, in which landowners issued individual requests, to a supply-driven model that 
seeks to resolve all land tenure issues in a municipality at one time. OSPR consists of three stages:
• A planning stage, using government registries and community information sources to determine 
land uses and tenure forms, identify land conflicts and measure the level of informality, 
generating a formal implementation plan for approval by the National Land Agency (ANT –
Spanish Acronym); 
• An implementation stage, where a cadaster update and parcel sweep are conducted to 
multipurpose cadastre and collect requests from petitioners along with documents to prove 
their possession; and 
• A legalization stage, in which the ANT reviews requests, makes decisions on each case, and 
issues formal land titles.
OSPR’s municipal-level planning and implementation mechanism, known as POSPR (Participatory Rural 
Land Use Management Plans), was piloted in Ovejas, Sucre in 2018-2019 with the support of USAID 
under the Land and Rural Development Program (LRDP). As a forerunner to the current LfP activity, 
USAID’s LRDP accomplished several achievements, including: designing and piloting massive titling 
methodologies; strengthening local land governance by establishing municipal land offices, titling schools 
and lands with public services to allow social investment; obtaining GoC commitment for joint cadaster 
and titling efforts; mobilizing funding in remote rural areas for the provision of basic services; establishing 
PPPs and linking restituted families and beneficiaries of land titling; improving access and security of land 
information; and incorporating new methodologies to accelerate land restitution and secure sustainable 
implementation of judicial orders. Through this pilot program, approximately 3,000 rural properties are 
in process of being formalized17 (totaling 42,000 hectares), addressing 100 percent of the informal 
holdings in the area. The pilot included more than 15,000 beneficiaries and 9 government entities. 
Learning from the Ovejas pilot has contributed to an adjusted approach for the planned LfP activity that, 
among others, considers specific regional conditions such as ethnic community land rights, presence of 
illicit crops, and large private sector investments. 
The 2016 peace accords and the Victims’ and Land Restitution Law of 2011 also established a new 
framework for dealing with restitution, land titling, and illicit crops. Since the beginnings of the 
16 https://www.whitehouse.gov/ondcp/briefing-room/2021/07/16/ondcp-releases-data-on-coca-cultivation-and-potential-cocaine-production-in￾the-andean-region/
17 Per USAID, 2,046 of 3,002 titles had been issued as of October 2021, while 188 titles had been denied, and 768 cases were still pending. The 
remaining titles will be issued under LfP assistance.
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
3
Colombian conflict in the 1960s, approximately 7 million Colombians have been forcibly displaced, 
or around 15 percent of the country’s population (UNHCR, 2015). The Victims’ Law of 2011 
established a framework for returning land to victims through the Unidad Restitución de Tierras (Land 
Restitution Unit - URT). Under this law, victims who were forced to sell land, abandoned land due 
to threats, were forced to sell land under unfair selling conditions due to the needs generated by 
the violence, or whose signatures were falsified on selling documents are eligible for restitution, 
regardless of whether they have a formal title to the plot, as long as the events occurred after 
January 1, 199118.
A new framework for dealing with illicit crop cultivation was established under the 2016 peace accords. 
Prior to signing the peace deal, the Colombian government suspended aerial fumigation of coca crops, 
which had been part of Colombia’s counter-narcotics operations for over 20 years, citing the negative 
environmental and health effects.19 The peace accords created the Programa Nacional Integral de 
Sustitución de Cultivos de Uso Ilícito (Comprehensive Program for Illicit Crop Substitution - PNIS). 
Envisioned as the country’s first ever attempt at crop substitution on a massive scale, the program 
sought to change Colombia’s counter-narcotics efforts from a forced eradication to a voluntary 
substitution model. The program aimed to enlist nearly 80 percent of Colombia’s coca-growers in a 
program providing regular payments and technical assistance in exchange for voluntarily destroying their 
coca crops and replacing them with licit alternatives, initiated through community-level agreements 
covering all coca farmers in participating communities.20 However, the program has experienced 
numerous complications and changes since its original inception, including funding challenges, changes to 
definitions on who can qualify for the program, and reformulating the agreements such that they are 
made with individual farmers instead of whole communities.21
LAND FOR PROSPERITY ACTIVITY DESCRIPTION
Building on the experiences of LRDP, the LfP activity is envisioned to improve the conditions of conflict￾affected rural households in a sustainable manner that will support USAID/Colombia’s effort to promote 
peace and stability, and to reduce illicit crops and help the GoC strengthen state presence in under￾attended regions.22 LfP began in August 2019 and is a five-year activity that is implemented by Tetra 
Tech ARD under the Strengthening Tenure and Resource Rights (STARR) II IDIQ. LfP comprises three 
main intervention components: 
Component 1: Advancing massive land titling in rural areas along with continued restitution support, 
Component 2: Strengthening local capacity to maintain formalized land transactions, and 
Component 3: Strengthening land governance and economic development through strategic PPPs.
In total, these components will provide access to land titles or other types of formal rights, where relevant23, 
while supporting land restitution as part of a broader land title policy support, strengthening local 
government capacity, and integrating citizens to licit socio-economic opportunities in target areas. These 
components are advanced under the activity’s seven guiding principles: (1) integration of gender, ethnic 
minorities, and youth; (2) environmental considerations; (3) coordination with other USAID programs and 
donors; (4) coordination of high-level dialogue meetings with the GoC; (5) transition of knowledge, skills, and 
18 https://www.minjusticia.gov.co/programas/justicia-transicional/ley-victimas-restitucion-tierras
19 https://www.bbc.com/mundo/noticias/2015/05/150514_colombia_glifosato_suspension_nc
20 https://www.crisisgroup.org/latin-america-caribbean/andes/colombia/87-deeply-rooted-coca-eradication-and-violence-colombia
21 https://www.revistaciendiascinep.com/home/la-muerte-lenta-del-pnis-en-el-gobierno-duque/
22 Land for Prosperity Quarterly Report, October-December, 2019. 
23 Per LfP implementing partners, titling is anticipated to be the predominant form of formalization under LfP however in some municipalities 
other forms of formal documentation could be prevalent.
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
4
abilities; (6) citizen security; (7) evidence-based policy. 
The original geographic scope of the LfP activity covered a total of seven regions, including Southern 
Tolima, Montes de Maria, Meta, Catatumbo, Tumaco, Northern Cauca, and Bajo Cauca and Southern 
Córdoba. Ten municipalities across the seven regions will benefit from all three different components of 
the LfP activity, and these are referred to as pilot municipalities. An overarching objective of the activity 
implementation in these “proof of concept” pilots is to help inform GoC with evidence-based 
implementation to shape land policy reform. These ten municipalities in the original geographic scope of 
LfP are also the focus of the IE. Additional municipalities along the same economic corridor as the ten
pilot municipalities are planned to receive components 2 (13 additional municipalities) and 3 (56 
additional municipalities), for a total program geographic coverage of 79 municipalities.24 Below, we 
provide additional high-level details for the three components.
Early in activity implementation, LfP’s geographic scope was expanded to include Chiribiquete National 
Park and its vicinity, the municipality of Puerto Rico in the department of Meta, and two pilot
communities in the region of Southern Meta and the vicinity of Chiribiquete, while a stronger 
environmental focus and some new implementation activities were also rolled-out in these areas, 
including additional PPPs, and land use contracts in the pilot communities. LfP activities in this expanded 
geography are not included in the design or scope of this evaluation, nor are they covered in this report.
MASSIVE LAND TITLING
Component 1, the massive land titling, is at the center of the LfP activity and has four stages. The first 
stage involves the POSPR, which determines land uses and tenure forms, as well as documenting 
possible land conflicts, restrictions, and the level of informality. Several GoC agencies, such as the IGAC, 
Agencia Nacional de Tierras (National Land Agency - ANT), and the URT are also involved in this process 
for many of the intervention municipalities. In the second stage, LfP teams will go to the field to 
announce the beginning of the parcel sweep, including the methods, geographic routes, and scope. The 
third stage involves actual household visits to collect socio-economic, demographic, cartographic, and 
legal data. The fourth stage involves the document preparation by LfP for land titling. The final stage 
involves the GoC validating cadaster information and issuing titles; although this is not a direct 
responsibility of LfP, program results will depend on GoC using the LfP-provided inputs from Stages 1-4 
to carry out this phase. The timeframe from the beginning of stage 1 to households receiving land titles 
is anticipated to be potentially 12-18 months, per the LfP IP.
In terms of eligibility, Component 1 aims to include every rural plot in the municipality in the parcel
sweep. Plots in areas where the security situation prevents access may not be covered by the parcel
sweep, and the activity would use indirect or remote methods to cover these plots; it is anticipated that 
only in extreme cases, would the activity be unable to complete the cadaster update in a community and 
be unable to advance titling. Other factors may also impact which plots are eligible for receipt of titles: 
plots with conflicting claims, open land restitution cases, or located on vacant public lands (baldíos) may 
take longer to title, while those located on lands with use restrictions (such as those located on flood 
plains) may not be granted a title at all, and communities seeking to obtain or enhance collective 
community titles to land (such as ethnic minority communities, which have a separate process for titling 
and recognition by GOC) are unlikely to receive such titles through this program, although they will 
still be included in the cadaster update.
24 Figures come from ACCESO_79_MUNICIPIOS.xlsx document provided to NORC by LfP.
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
5
STRENGTHENING LOCAL GOVERNMENT CAPACITY
Component 2 promotes the sustainability of LfP efforts by strengthening local capacities for land market 
transactions and updating urban cadaster and titling, as well as public infrastructure in national lands. The aim 
is to build information management capacities and interoperability amongst GoC entities and local 
government institutions, prioritizing data security and the legal integrity of information, while increasing 
access to municipal land offices that serve as an entry point for rural citizens to access national land agencies. 
Thus, the activity works to decentralize services and access to information, while creating single access points 
for information that is currently maintained by disparate institutions, digitizing records, and expanding local 
land management capacity through municipal land offices.
The inclusion of this component in the design of the activity was motivated by a review of international 
cooperation experiences supporting land titling, in which efforts were undone over time due to insufficient 
capacity for land administration by the local actor after the end of program support. 
PROMOTING PUBLIC-PRIVATE PARTNERSHIPS (PPPs)
Component 3 seeks to expand licit economic opportunities for households in the LfP implementation 
municipalities. As part of this component, LfP will strengthen the capacity of subnational governments to 
obtain and mobilize regional and national funds, plan and execute projects, and provide services. In particular, 
the activity aims to mobilize public and private funds for local public goods and services, as well as the 
integration of smallholders into value chain alliances. The activity meets with local chambers of commerce, 
industry groups, and mayors’ offices to identify value chains and investment projects for prioritization that 
can be linked to PPPs. By identifying bottlenecks in the agricultural value chains, the activity works to mobilize 
resources in support of solving problems to make local businesses more competitive (including, but not 
limited to construction of tertiary roads, small irrigation projects, among others. 
DEVELOPMENT HYPOTHESIS
Figure 1 draws on intervention logic provided in the LfP solicitation, literature review by the evaluation team,
and discussions with USAID and LfP IPs to depict a proposed Theory of Change for the LfP activity. The 
model was first developed by the evaluation team as part of a feasibility assessment for the IE25 and was 
further refined during the evaluation design process. The theory of change presented here serves as a 
broader learning framework to examine whether and under what conditions the direct outputs resulting 
from a massive land titling program such as LfP may also achieve a range of higher-level development 
outcomes, many of which may not be explicit targets of LfP programming directly.
Figure 1 uses solid black arrows to show the direct linkages between LfP activities and shorter-term 
outcomes that are anticipated to be more directly attributable to LfP. Changes in these short-term 
outcomes may also lead to changes in the medium- and longer-term outcomes listed, particularly in the 
presence of additional actions and intervention from regional and national governments, the private sector, 
and other donor programs. These less direct linkages to LfP activities are shown by dashed black lines. At 
endline, the evaluation team will measure the changes in outcomes that are observed and will also work to 
understand the reasons for the observed outcomes and provide interpretation of the results. This 
includes, for example, understanding how LfP implementation together with a range of complex factors 
operating in the study area may have influenced the observed results (whether negatively or positively), 
and providing learning on where the evaluation results either support or may suggest a need to revisit 
elements of the overarching logic chain from land titling to some of the ultimate development objectives. 
25 NORC at the University of Chicago. 2019. “Impact Evaluation Feasibility Assessment of the Land for Prosperity (LfP) Activity”. Prepared for 
USAID under the Communications, Evidence, and Learning (CEL) Project, USAID Contract Number GS00F061GA.
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
6
It is important to keep in mind that several additional USAID-supported activities are also active within the 
broader LfP geography, across different sectors, and in many cases they also focus on achieving the same 
overarching development objectives and outcomes as LfP. It is possible that results across several of these 
programs may also be necessary to achieve widespread change on some of the longer-term outcomes and 
ultimate development goals listed in Figure 1, there are potentially a wide range of administrative, legal, and 
economic contextual factors that may also be necessary for achieving the desired changes in the outcomes
listed in the TOC. As such, the theory of change relies on a host of assumptions that are necessary 
conditions for achieving the changes as envisioned in the model. We present these assumptions below. 
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
7
Interpretation Guide:
• Dashed red arrows indicate feedback loops.
• Dashed black arrows indicate outcomes that are not directly linked to LfP performance goals. All outcomes are likely influenced by a range of contextual factors to some extent.
FIGURE 1: LFP THEORY OF CHANGE
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: DRAFT Baseline Report
8
KEY ASSUMPTIONS
• Activity implementers and government stakeholders will be able to establish effective land conflict 
resolution mechanisms and effectively coordinate with land restitution efforts in implementation zones.26
• Corruption within the land administration and related government systems will be reduced through 
improved public systems and capacity-building.
• There is effective coordination between land administration officials and other public entities (for 
example, local tax authorities and municipal service providers).
• Land titling, improved land governance, and expanded economic opportunities in licit sectors 
provide sufficient additional incentive to existing government interventions, such as the 
government’s Comprehensive Program for Illicit Crop Substitution / Programa Nacional Integral de 
Sustitución de Cultivos de Uso Ilícito (PNIS) and Strategic Areas of Integral Intervention / Zonas 
Estratégicas de Intervención Integral (ZEII), for farmers to participate in such programs and shift from 
illicit crop production to licit crop or other economic activities.
• The State will enforce prevailing laws, such that obtaining a formal land title will increase the risk of 
state sanctions by landowners who have coca production on their land. Receipt of formal land title is 
therefore assumed to be tied to increased risk of land expropriation by the State for such 
landowners27.
• There is sufficient public and private sector interest and commitment to engage in public-private 
partnerships, and sufficient capacity to engage farmer-based organizations and provide public services. 
• Similarly, farmer-based organizations will have sufficient interest and capacity to participate and 
benefit from these partnerships.
• There will be programs or efforts from other agents to provide technical assistance and build 
capacity for households to increase farm productivity.
• Rural banks, other local financial institutions or formal lenders have or will be willing to create loan 
products (because of training or incentive programs) that are accessible and of interest to under￾served borrowers in the implementation area (rural smallholder farmers, micro- and small-scale 
enterprises and entrepreneurs).
• It will be possible within the LfP lifetime to understand market systems behavior, actors and 
opportunities within the implementation zones; and, implement new opportunities at scale, including 
any technological or other innovations needed to bolster market systems and address constraints.
• GoC provides security and proper coordination for de-mining and coca eradication, in cases where 
voluntary substitution in exchange for the land title is not achieved. In addition, the ANT is able to 
properly plug into the Formalizar para Sustituir Program, which is mandated to do the crop 
substitution in exchange for the land title.
• Broader policy and/or macro-economic assumptions:
o Land values and demand for land experience gradual change rather than sharp volatility.
o There are no major changes to in/out-migration dynamics as a result of program activities or 
other factors during the activity lifetime.
o Coca prices do not increase to a point such that revenue earned to smallholders through 
alternative crops is non-competitive.
o Current levels of violence in implementation zones do not increase significantly.
o Introduction of additional counter-drug policies and programs, or pre-existing such programs in 
the implementation area, are synergistic with LfP-promoted economic activities and enabling 
environment.
26 Note that municipalities are overseen by local government and mayors, while rural areas in the implementation regions are administered by 
national agencies.
27 Under current GoC policies, the State can expropriate land from such landowners, and they may also be charged with a criminal offense. 
This is seen as a key mechanism by which strengthened land rights may alter the risk preferences and incentives landowners face with respect 
to coca production (Munoz-Mora et al 2019).
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: DRAFT Baseline Report
9
• LfP will be able to effectively coordinate with other USAID portfolios relevant to the outcomes of 
interest, such as environment, justice, governance, and economic development.
ACTIVITY IMPLEMENTATION STATUS
LfP had already begun operating in some municipalities at the time of baseline data collection for this 
evaluation, although the nature of implementation activities was not expected to affect baseline results 
or threaten the validity of the IE. Table 1 shows the status of LfP implementation by treatment 
municipality as of the close of baseline data collection on June 17, 2021. Preparation activities were 
underway in six municipalities, including conducting social mapping to identify intervention units, 
developing work plans, submitting plans for approval with ANT, and holding meetings with local and 
regional entities. Three other municipalities had already had their implementation plans approved and 
were in a pre-implementation phase. Parcel sweeps were already underway in two municipalities, though 
massive land titling hand not yet begun. No implementation had started yet in one municipality. 
TABLE 1: LFP IMPLEMENTATION STATUS AT BASELINE DATA COLLECTION
REGION MUNICIPALITY STATUS AT BASELINE
Montes de Maria San Jacinto Implementation plan approved
El Carmen de Bolivar Preparation stage
Sur de Tolima Ataco Parcel sweeps in progress
Chaparral Implementation plan approved
Catatumbo Sardinata Preparation stage
Meta Fuente de Oro Preparation stage
Puerto Lleras Implementation plan approved
Bajo Cauca Cáceres Parcel sweeps in progress
Norte de Cauca Santander de Quilichao No implementation started
Tumaco Tumaco Preparation stage
EVALUATION PURPOSE, AUDIENCE, AND INTENDED USES
The overarching purpose of the LfP IE is to (1) provide an evidence base for impacts of the LfP activities 
with respect to strengthening land rights and land governance on several outcome categories, how impacts 
vary across key sub-groups of interest, and reasons why; and (2) provide targeted learning on overall 
lessons learned and several ‘deeper dive’ learning interests around titling effects, elements of 
implementation effectiveness, women’s empowerment and household food security, participation in licit 
market opportunities, illicit crop substitution, and improvements to municipal-level public service delivery. 
The primary audiences for the evaluation results are USAID/DDI/EEI/LRG, USAID/Colombia, and the 
LfP IP. Key secondary audiences for the evaluation results include the broader donor community, 
various GoC agencies related to land formalization efforts in Colombia, and other stakeholders involved
in land and related development sectors.
The evaluation findings are expected to have accountability and learning value to USAID. The evaluation 
will inform the design of future activities that aim to integrate massive land titling with local capacity 
building to strengthen tenure security, enhance livelihoods, promote durable peace, and reduce the 
cultivation of illicit crops. It is also intended to provide targeted learning on key knowledge and theory 
of change logic gaps to inform the design of massive land titling projects that may follow.
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: DRAFT Baseline Report
10
EVALUATION QUESTIONS
The evaluation uses a mixed-methods approach combining a quantitative quasi-experimental difference￾in-difference design and a qualitative pre-post data analysis to answer a set of evaluation questions 
across six thematic areas. The evaluation questions were developed in conjunction with USAID and LfP, 
and reflect specific USAID learning interests. See Annex F for an Evaluation Design Matrix summarizing 
data sources, outcome measures and analytic approach by evaluation question.28 This IE measures the 
combined impacts of LfP across all three intervention components. This is because all three components 
of the LfP activity are being implemented in the ten pilot municipalities, and the intervention group for 
the IE consists only of these ten pilot municipalities.
Table 2 below summarizes the evaluation questions for this IE, and Table 3 below lists the household￾and municipal-level outcomes that constitute the IE focus across different outcome domains. 
TABLE 2: EVALUATION QUESTIONS AND THEMATIC AREAS OF INVESTIGATION
28 For additional evaluation design details, see Protik, A., G. Haugan, R. Wendt, L. Persha, and J.C. Muñoz Mora. (2020) Evaluation of the 
‘Evaluation of the Land for Prosperity (LfP) Activity’ in Colombia: Evaluation Design Report. Washington, DC: USAID Communications, Evidence and 
Learning Project.
THEME EVALUATION QUESTION
Impacts of LfP 
on individual 
households
1. What are the effects on households of strengthened land governance and property rights via 
LfP’s land titling (and other types of formal documentation, where applicable) and coupled land 
administration capacity building? 
a. Tenure security
b. Land use, investment, and illicit crop cultivation
c. Improved household wellbeing, food security, and poverty
d. Credit access and land market activity
e. Women’s empowerment
Heterogeneity 
of Impacts on 
Household￾level 
Outcomes by 
Beneficiary and 
Context 
Characteristics
2. How do the impacts of the LfP activity on key outcomes differ for:
a. Female-headed households or joint households
b. Youth-headed households
c. Afro-Colombians, members of indigenous communities, and ethnic minority 
households
d. Households victims of the civil conflict
e. Households in coca-producing areas compared to those in areas with no coca
3. Do women recipients of land titles experience improvements on par with men for: 
a. decision-making regarding land use
b. agricultural productivity 
c. incomes
Impacts of LfP 
on Land 
Administration 
and 
Governance
4. To what extent did LfP activities improve municipal level local government self-reliance and 
public service delivery?
5. At the municipal level, what are the impacts of strengthened local government land 
administration capacity coupled with massive land titling on municipal tax revenue?
a. How do mayor’s offices use increased tax revenue gained through an increase in the 
number of title holders? 
b. Does increased tax revenue lead to improved public services for communities?
Impacts of LfP 
on Private￾Sector 
Engagement 
and Illicit Crop 
Substitution
6. Did the LfP activity increase private sector engagement in the pilot and corridor municipalities? 
a. If yes, how much funding and what types of activities were leveraged from the private 
sector through the LfP initiative?
b. If no, what were the major reasons and/or barriers for the private sector to engage 
effectively? 
7. What is the role of mobilized public sector funds and public-private partnership for local public 
goods in supporting sustainable reduction in coca production based on land titling?
8. Does massive land formalization contribute to more sustainable licit crop substitution, as 
indicated by increased numbers of farmers undertaking substitution and a reduced area under 
cultivation of illicit crops? 
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: DRAFT Baseline Report
11
29 Noting that for rural parcels, the system may be better characterized as deconcentrated rather than decentralized, as municipalities do not 
have jurisdiction for land administration in rural areas. Still, LfP aims to work with municipalities to provide related access and motivate 
registration of changes for rural areas as well.
THEME EVALUATION QUESTION
a. Once individuals receive a land title, how easily can they connect and improve their 
income opportunities through the alliances and opportunities created through 
Component 3?
Implementation 
Effectiveness
9. How successful was the approach to adaptation of the strategies and lessons learned from the 
Ovejas pilot?
10. How effective was the strategy to engage the private sector in formalization in scaling progress 
or sustaining formality?
11. How effective was the USAID model for establishing municipal land offices within local 
government to develop and maintain a decentralized29, sustainable land administration and 
management system in municipalities (includes looking at effectiveness of land tax collection and 
tax revenue, among others)?
a. What strategies or technology innovations were most effective for harmonizing the 
collection and management of cadastral information and land registry records?
b. What is the effect of technology changes (e.g., document digitization) on institutional 
capacity? Are there any notable effects these institutions faced as they moved from a 
large staff / high manpower model to one more reliant on improved technology?
Overarching 
Lessons
12. Did LfP implementation of massive land titling lead to more sustainable illicit crop substitution? 
In what keys ways?
13. In what ways did the LfP activity contribute to land-related Peace Accord agenda objectives, 
including those specified under Chapter 1?
14. In what ways did the LfP activity contribute to broader GoC agrarian land rural reform and 
rural development objectives?
15. To what extent did GoC adopt and scale up public policy inputs and recommendations from 
LfP?
16. Were there any unintended broader consequences of the land titling component, beyond those 
related to LfP objectives?
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
12
TABLE 3: EVALUATION INDICATORS AND OUTCOMES MEASURES30
30 All household-level outcomes will be evaluated for heterogeneity of impacts by key beneficiary sub-groups and geographic variation, and reasons why. Key sub-groups of interest include: women, 
youth, conflict victims, Afro-Colombians, indigenous communities, and households in coca producing areas. Key geographic variation of interest is to be determined.
HOUSEHOLD-LEVEL OUTCOMES
ANTICIPATED TIME 
FRAME TO ACHIEVE 
CHANGE AT SCALE
FROM END OF 
INTERVENTION
TENURE SECURITY IMPROVED LAND USE AND 
INVESTMENT
IMPROVED HOUSEHOLD WELL￾BEING, FOOD SECURITY, AND 
POVERTY
IMPROVED CREDIT ACCESS 
AND LAND MARKET 
ACTIVITY
SHORT-TERM
(0-2 years)
• Formalized land documentation (+)
• Incidence of land expropriation and 
displacement (-)
• Perceived tenure security (+)
• Recent land disputes (-)
• Area under licit crop production (+)
• Area under coca production (-)
• Participation in licit cash crop 
production (+)
• Number of licit cash crops grown (+)
MEDIUM-TERM
(3-4 years)
• Farm investments (irrigation, fertilizer, 
labor, etc.) (+)
• Investment and participation in off￾farm and non-farm businesses (+)
• Women’s empowerment and 
participation in PPPs (agricultural and 
household decision making) (+)
• Household income (+)
• Income from licit crops (+)
• Income from licit off-farm and non￾farm activities (+)
• Investments on land assets 
(machinery, coffee depulpers, 
dryers, etc.) (+)
• Seeking and access to formal 
and informal sources of 
credit (+)
• Amount of credit obtained 
(+)
• Land market activity (rentals, 
sales or other transactions, 
use as collateral) (+)
LONGER-TERM
(> 5 years)
• Satisfaction with and confidence in 
land administration and governance 
(+)
• Satisfaction with PPPs (+) 
• HH involvement in PPPs (+)
•
• Agricultural productivity (+)
• Food security (+)
• Access to markets (distance and 
road quality) (+)
• Satisfaction with quality of public 
goods (e.g., roads/irrigation) (+)
• Subjective household wellbeing (+)
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
13
MUNICIPAL-LEVEL OUTPUTS OR OUTCOMES
SHORT-TERM
(0-2 years)
• Number of land titles issued (+)
• Establishment of municipal land 
offices (+)
• Digitization of land transaction 
records for preservation and 
transparency (+)
• Amount of private-sector funds 
mobilized for land formalization (+)
•
• Area under licit crop production (+)
• Area under coca production (-)
• Changes in private 
investment (+)
• Investment in public goods 
(e.g., roads/irrigation) (+)
MEDIUM-TERM
(3-4 years)
• Amount of land tax received (+)
• Incidence of displacement (-)
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
14
EVALUATION DESIGN AND METHODS
DESIGN OVERVIEW
This evaluation uses a quasi-experimental difference-in-differences (DID) approach to assess the impacts 
of LfP through household-level and municipal-level data, coupled with statistical matching to select the 
comparison group. The IE is complemented by a pre-post qualitative data analysis based on document 
reviews, interviews, and group discussions to better understand mechanisms for observed impacts or 
reasons for no impacts and to assess broader implications of the LfP activity. 
The IE is designed to estimate impacts of LfP on various outcomes of interest measured as the difference 
between outcome levels for beneficiaries and the counterfactual condition, which represents outcome 
levels of the same beneficiaries had the intervention not taken place. Because it is impossible to observe
outcome levels for beneficiaries without the intervention once the intervention has been implemented, the 
IE design approximates the counterfactual scenario by using a closely-matched group of non-beneficiaries, 
whose observed characteristics are largely similar to those of the beneficiaries at baseline—the 
comparison group. The evaluation team constructed this comparison group using a three-step statistical 
matching technique that is summarized below. See Annex B for additional technical details.
THREE-STEP STATISTICAL MATCHING
The evaluation team implemented a three-step process for identifying the comparison group—a 
municipal-level matching in the first step, a community-level matching in the second step, and a 
household-level matching in the final step. The team used genetic matching (Diamond and Sekhon, 2013)
at municipality and community levels to optimize the balance between the LfP and the comparison group 
across a set of key variables. The IE analysis that will be conducted at endline is designed to also use 
entropy balancing at the household level, to improve covariate balance between the treatment and 
comparison groups.
We conduct statistical matching at the municipality level, to take account of municipal-level effects and 
to increase the cost-effectiveness of the IE data collection by limiting the comparison group to a single 
comparison municipality for each of the 10 LfP pilot municipalities. The evaluation team used 34 
municipal-level variables, including variables that were considered by LfP in selecting the pilot 
municipalities, variables we hypothesize may be impacted by LfP, and other variables that are related to 
the treatment and may be important in shaping potential outcomes, to the extent that the data or 
reasonable proxies were available.
We implement the second step for two reasons: (1) while the parcel sweep is conducted for the entire 
municipality, the final phase involving dispute resolution and document preparation for the titling process 
will be implemented on a community-by-community basis. This implies that there is likely to be greater 
variability across communities within a given municipality in terms of household outcomes and that there 
will be variations in implementation timing across communities within municipalities; and (2) community￾level matching provides greater statistical power because there are more communities than municipalities. 
The evaluation team first divided each LfP municipality into rural and urban communities (veredas and 
centros poblados), as defined by IGAC31. Together, these communities comprised the totality of each 
municipality, except for the municipal seat (i.e., the town or small city where the municipal government is 
headquartered, and which is generally much larger than any other community in the municipality). The 
team then used a stratified probability proportional to size sampling design to select 10 communities from 
each LfP municipality, for a total of 100 selected communities. The team then employed a genetic-matching 
31 Note that IGAC’s definitions of communities are different from those from DANE, which also maintains maps of communities within 
municipalities.
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
15
process to match each LfP municipality to a comparison municipality located in the same microregion. This 
resulted in a final selection of 200 communities for household-level data collection.
Figure 2 shows a map of the LfP and the matched-comparison municipalities. The locations of the ten LfP 
municipalities are shown in blue, while the ten matched comparison municipalities are shown in red, and 
the map labels each of the seven microregions. 
FIGURE 2: MAP OF LFP AND COMPARISON MUNICIPALITIES
The final matching step is conducted on the baseline data after the survey data collection and cleaning 
have been completed. In this step, Entropy Balancing is used to re-weight household observations in 
comparison municipalities and improve the statistical balance with household observations in LfP 
municipalities across a set of key household context characteristics at baseline, such as whether the 
household has agricultural activities, number of children and adults in the household, whether the 
household head identifies as a member of a racial minority group, sex of the household head, whether 
the household head lives with a spouse or domestic partner, household head age, and household head 
education, along with pre-treatment indicator values. We undertake this final statistical matching step to 
further improve the similarity of the LfP and the comparison group sample across important 
characteristics and enhance our future ability to make inferences about the impacts of the LfP 
intervention on households. We present further details on the household-level Entropy Balancing in the 
Balance and Power chapter in the main body of the report.
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
16
BASELINE DATA COLLECTION 
HOUSEHOLD QUANTITATIVE SAMPLE
Household survey data collection at baseline took place during April 6 – June 17, 2021. The household 
survey sample for the evaluation was designed to consist of 15 households per community across 10
communities per municipality, for a total sample of 150 households per municipality or 3,000 total 
households. 
The evaluation team selected the targeted sample of 3,000 households from across the 200 matched LfP 
and comparison group communities using a modified random sampling process. Although two separate 
GoC institutions, DANE and IGAC, maintain maps of communities within municipalities, the IE used 
those from IGAC for better alignment with how LfP is implemented. The process used slightly different 
procedures for sampling households depending on whether the selected community was a rural vereda
(analogous to an unincorporated rural community in the United States) or an urban centro poblado 
(analogous to a town), considering the feasibility and cost of data collection. For centros poblados, 
communities were divided by census blocks using data from DANE32. Three blocks per centro poblado
were selected at random, enumerators then listed dwellings in selected census blocks, and then 
randomly selected seven dwellings per block. The anticipation was that approximately 30 percent of 
selected dwellings would be uninhabited, uncooperative, or unreachable, and would yield a final sample 
of 15 successfully surveyed households per community, on average. For veredas, a GIS team divided the 
community into segments using satellite imagery such that each vereda segment contained approximately 
10 building structures and that building structures were located within relatively close proximity of each
other to facilitate fieldwork. Three vereda segments were then randomly selected, and enumerators 
attempted interviews at all building structures found to be dwellings once in the field.33 The anticipation 
was that approximately 50 percent of structures would be found not to be dwellings, uninhabited 
dwellings, or dwellings with unresponsive or uncooperative households, and would yield a final sample of 
15 successfully surveyed households per community, on average.
The final sample consisted of 2,965 households across 192 communities, with 51 percent of the 
households belonging to the LfP intervention group (Figure 3). The final sample size is smaller than the 
planned sample size because data collection in eight communities, all located in comparison municipalities,
could not be completed, either because of security reasons or because of inaccessibility. Three of these 
communities were located in La Macarena, four in Convención, and one in Olaya Herrera.34
32 Although the IE uses the IGAC definition of communities, we used the DANE census blocks because the Colombian census is carried out by 
DANE using their own definition of communities. For more details on the sampling process, please see Annex C.
33 Distance between households and difficult road access made it impossible to do a listing of the entire community, or to have larger segments that 
would be listed and used to randomly sample households. Additionally, listing was not recommended due to security risks, as the process of going 
door to door listing households is similar to activities historically done by armed groups.
34 In La Macarena, the enumerator team faced an increasingly difficult security situation. Armed groups were enforcing a general strike, and a 
land restitution team in a neighboring municipality had just gone missing (and tragically, later was found murdered). The team in La Macarena 
decided to abandon activities after completing seven communities, with NORC’s full support. In Convención, the team arrived to find that 
approximately 80 percent of the municipality was a red zone, with guerrilla activity and illicit crops. While many areas were considered no-go 
areas, the team worked with local leaders to obtain access. After completing six communities, access to the remaining communities was not 
forthcoming, while protests around a proposed tax reform were getting increasingly worse, and the threat of roadblocks could have prevented 
them from leaving the municipality indefinitely unless they left immediately; again, NORC fully supported their decision. In Olaya Herrera, one 
sampled community was found to be abandoned. Olaya Herrera is an extremely isolated municipality whose main access is via boat over the 
ocean from Tumaco. By the time a replacement community could be selected and communicated to the team, they had already left the 
municipality and it was not feasible to return.
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
17
FIGURE 3: BASELINE SAMPLE
QUALITATIVE SAMPLE
Qualitative data collection took place between July 8 and August 11, 2021 after the quantitative data 
collection had already ended. A qualitative training session was held with a group of potential 
enumerators in May, but actual fieldwork had to be postponed when a participant tested positive for 
COVID-19 shortly after the training. By the time enumerators had completed their mandatory 
quarantine, the situation around the national protests had reached a level where it was not possible to 
get teams into the field until July. Three types of qualitative data collection activities were conducted: 
Group Discussions (GDs) with community members, Group Key Informant Interviews (KIIs) with 
community leaders, and KIIs with national government officials. Table 4 provides a breakdown of how 
these activities were conducted.
For GDs, communities were selected by the 
evaluation team. First, the team purposefully 
selected five treatment and five comparison 
municipalities to provide broad representation 
across the different regions in the study, while 
also trying to include as many known coca￾producing municipalities as possible. Treatment 
municipalities included Cáceres, Fuente de Oro, 
Tumaco, San Jacinto, and Sardinata. Comparison 
municipalities included Convención, La 
Macarena, Olaya Herrera, San Juan 
Nepomuceno, and Zaragoza. The team then 
randomly selected two communities per 
municipality. This resulted in a total of 13 
veredas and 7 centros poblados.35
35 During data collection, the qualitative team in La Macarena encountered security problems similar to those encountered by the quantitative 
team, and were only able to complete one GD and one Group KII. The IE team selected a replacement community from the municipality of 
Caldono by reviewing the quantitative data and selecting a municipality where respondents reported coca growing.
Total Baseline Sample N = 2,965 Households
100 Treatment Municipalities
1,512 Households (51.0%)
92 Comparison Municipalities
1,453 Households (49.0%)
Group Discussion in an LfP community PHOTO BY SEI
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
18
GDs were conducted with a total of 151 participants, of which 71 were female and 80 were male. 
Participants in each GD sought to include a range of age and ethnic minority groups. The GDs utilized 
semi-structured instruments and were conducted by a moderator, with assistance from a notetaker (see 
Annex D for instrument). Group KIIs included a total of 75 participants, of which 30 were female and 45 
were male. Again, the groups were selected to include women, and a range of age and ethnic minority 
groups. Group KIIs used the same instruments as the GDs, and were conducted by a moderator, with 
assistance from a notetaker. A total of six KIIs were conducted with government officials; potential 
participants were recommended by LfP, and NORC’s local data collection partner, SEI, conducted 
outreach to obtain their cooperation. All GDs and KIIs were conducted in Spanish. 
TABLE 4: QUALITATIVE DATA COLLECTION SUMMARY
TYPE PARTICIPANTS NUMBER OF INTERVIEWS 
CONDUCTED
DESCRIPTION
Group 
Discussions 
(GDs)
6-10 community 
members from a 
single community
• 2 communities per 
municipality in 5 treatment 
municipalities for a total of 10 
treatment GDs;
• 2 communities per 
municipality 5 comparison 
municipalities for a total of 10 
comparison GDs.
• Multiple people participating in an open, semi￾structured discussion facilitated by a moderator;
• Structured to cover specific questions and topics, 
but the conversation is allowed to flow naturally;
• Participants purposefully selected to include women, 
youth, and ethnic minorities;
• Purpose is to gain an in-depth understanding of 
selected communities.
Group KIIs 4-6 community 
leaders from different 
communities in a 
municipality 
• 20 (1 in each of the 10 
treatment and 10 comparison 
municipalities)
• Multiple community leaders interviewed at once;
• Semi-structured interviews, led by an interviewer;
• Participants purposefully selected to include leaders 
who are women, youth, and from ethnic minorities;
• Purpose is to gain an in-depth understanding of the 
municipality through the experience of different 
communities, according to their leaders.
KIIs One-on-one 
interviews with 
officials from the 
ANT, URT, and PNIS
• 1 ANT interview;
• 1 URT interview;
• 4 PNIS interviews (1 each 
with officials from Cáceres, La 
Macarena, Sardinata, and 
Cauca).
• One person interviewed at a time
• Semi-structured interviews, led by an interviewer;
• Purpose is to describe how an entity operates, its 
interactions with LfP, and potential chokepoints in 
the program TOC.
HOUSEHOLD SURVEY
The household survey used at baseline consisted of 14 modules, as listed in Table 5. The survey was 
administered to the household head or another household member knowledgeable about the 
household’s land and productive activities (see Annex E for the full household survey instrument).
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
19
TABLE 5: HOUSEHOLD SURVEY MODULES
CHALLENGES ENCOUNTERED DURING DATA COLLECTION
Data collection itself occurred under extremely challenging conditions in the field. Both the planning and 
the fieldwork for the baseline data collection coincided with the COVID-19 pandemic. The evaluation 
team followed a strict protocol to reduce the risk of COVID spread including, remote enumerator training 
with trainees gathered in smaller groups in multiple locations, specific training for enumerators on personal 
safety and social distancing, and provision of masks, sanitizer, and alcohol wipes to each enumerator. Team 
supervisors also contacted local officials before visiting each municipality and community to ensure teams 
could comply with any specific local COVID protocols. Nonetheless, the COVID-19 pandemic created 
delays in fieldwork, and logistical challenges as enumerator teams navigated travel restrictions and local 
health protocols. Qualitative data collection, originally intended to coincide with the household survey, had 
to be delayed after an interviewer tested positive for COVID shortly after the qualitative training, requiring 
a two-week quarantine for all training participants as per our protocol.
Added to this was large-scale social and political unrest throughout the country beginning in late April, 
shortly after the start of data collection, in response to a tax reform proposal put forward by the 
national government. There were roadblocks that inhibited intermunicipal travel and forced teams to 
reroute, gasoline shortages, reduced availability among the community leaders who grant fieldwork 
permissions, and general strikes and stay-at-home orders enforced by illegal armed groups. Further 
complicating matters was a general deterioration of security conditions throughout the country, which 
made some data collection areas off-limits and further complicated access to community leaders (who 
are often the targets of violence).
A final challenge came from the weather for the municipalities in Meta, in particular in Puerto Lopez. 
Recent flooding had left some communities mostly abandoned and required several to be replaced.
SECTION COVERAGE KEY NOTES
A Household identification and consent n/a
B Respondent and household information n/a
C Household assets and wellbeing n/a
D General agricultural production
Includes a roster of the household’s three most 
important crops/animal species that is only 
administered to farming households. 
E Plot roster Module administered once for each plot the 
household controls.
F General tenure security and land rights n/a
G Land conflict/restitution Only administered if the household had at least 
one conflict in the past 12 months.
H Participation in alternative development programs n/a
I Land market activity and local service delivery n/a
J Non-farm income generating activities Only administered if anyone in the household 
operates a non-farm income generating. Activity.
K Credit n/a
L Food security n/a
M Wives' module Administered separately to the principal male 
and female decisionmakers.
N Survey conclusion Collected information for future follow-up 
surveys.
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
20
BASELINE SAMPLE CHARACTERISTICS
The baseline sample was stratified by urban or rural classification of the communities. In Colombia, 
these are known as centros poblados (small towns) and veredas (unincorporated rural communities). 
Among sampled treatment community households, 24.5 percent of the sample is in centros poblados, 
while 75.5 percent is in veredas; this compares to 25 percent of the sample in centros poblados and 
75 percent in veredas among the comparison sample (Figure 4). The average household in treatment 
communities has 3.7 members, compared to 3.8 among households in the comparison sample 
(Figure 5). Throughout the report, differences between the treatment and comparison group 
sample obtained on the unmatched sample are provided for informational purposes, but are not a 
concern for the validity of the IE (see additional discussion in the Balance and Power section).
FIGURE 4: URBAN (CENTROS POBLADO) VS. RURAL (VEREDA) SPLIT OF SAMPLE
FIGURE 5: AVERAGE HOUSEHOLD SIZE AT BASELINE
Figure 6 presents the basic demographic characteristics of the sampled households at baseline. In both 
treatment and comparison community households, the majority were headed by males. Just above a 
quarter—27.5 percent of households in the treatment communities and 26.2 percent in the comparison 
communities—were female-headed households. The majority of the households were also headed by 
older members—18.6 percent of the households in the treatment communities and 21.4 percent in the 
comparison communities were headed by young adults who were aged between 18 and 35 years. In 
terms of marital status, 52.2 percent of sampled household heads in treatment communities were in a 
common law marriage and 21.6 percent were formally married. The numbers were similar for 
household heads in comparison communities. A large share of the household heads—40.5 percent in the 
treatment communities and 35.9 percent in the comparison communities—were migrants, who were 
not born in the municipality they currently live in. None of the differences in basic demographic 
characteristics between the treatment and comparison communities in Figure 6 were statistically 
significant.
 
 
Treatment Households 
in Centros Poblados
370 Households (24.5 %)
Comparison Households 
in Centros Poblados
363 Households (25 %)
Treatment Households 
in Veredas
1,142 Households (75.5 %)
Comparison Households 
in Veredas
1,090 Households (75 %)
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
21
FIGURE 6: BASELINE SAMPLE DEMOGRAPHICS
Level of education among household heads was low. About a third of the household heads—33.4 
percent of treatment community household heads and 29.5 percent of comparison community 
household heads had secondary level education or higher. The baseline sample was also represented by 
ethnic minorities: 19.9 percent and 10 percent Afro-Colombian and indigenous household heads, 
respectively, among those in treatment communities. The corresponding numbers for the comparison 
communities were 20.5 percent and 18.8 percent. 
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
22
Figure 7 provides a breakdown of 
education levels of heads among 
sampled households in treatment 
communities: 14.8 percent of 
household heads had no education, 
51.8 percent said their highest level 
of education was at the primary level 
(kindergarten through first grade), 
26.0 percent said their highest level 
of education was at the secondary 
level, and 7.4 percent said they had 
any higher education. In general, 
household heads in comparison 
communities had similar levels of 
education to those in treatment 
communities, with the percentage 
who had any higher education 
representing the only statistically 
significant difference. 
Figure 8 presents details 
regarding ethnic identity of the 
baseline sample. Among 
sampled households in 
treatment communities, 19.9 
percent identify as Afro￾Colombian, compared to 20.5 
percent of sampled comparison 
community households. As 
expected, Afro-Colombian 
households are largely 
concentrated in the Nariño 
(along the Pacific coast) and 
Montes de Maria regions, and 
to a somewhat lesser extent in 
Northern Cauca. Ten percent 
of households in the treatment 
community sample identify as 
indigenous, compared to 18.8 
percent of sampled comparison community households, a difference that is statistically significant. 
Indigenous households in the sample are largely concentrated in Southern Tolima, Montes de Maria, and 
Northern Cauca. Among the treatment community sample, 24.9 percent identify as mestizo, a fluid 
category that describes people of mixed ancestry, while 44.9 percent do not identify with any of the listed 
racial or ethnic groups (which may include white Colombians or Colombians of mixed ancestry who do 
not necessarily identify as mestizo).
FIGURE 7: EDUCATION LEVEL OF HOUSEHOLD HEADS
FIGURE 8: ETHNIC IDENTITY OF HOUSEHOLD HEADS
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
23
Finally, we look at households among the baseline sample who were engaged in agriculture, defined as 
conducting any agriculture or livestock activities on any plots controlled by the household. Across all 
sampled households in the treatment communities, 77.2 percent were engaged in agriculture. This was 
slightly less than the 84.1 percent of households engaged in agriculture among the comparison 
community sample, a difference that is statistically significant (Figure 9). We also see some regional 
variation; in particular, Meta has a lower percentage of both treatment and comparison community 
households engaged in agriculture than other regions, likely due to specific characteristics including the 
importance of the petroleum sector and the extreme inequality in the distribution of landholdings, which 
leads more households to rely on wage labor as their primary source of income.
FIGURE 9: HOUSEHOLDS WITH FARMING ACTIVITIES, BY REGION
For households engaged in agriculture in treatment 
communities, 81.1 percent live in veredas and 18.9 
percent live in centros poblados; the corresponding 
numbers for comparison communities are 81.8 
percent and 18.2 percent, respectively (Figure 10). 
Among these farming households, 23.9 percent of 
those in treatment communities were headed by a 
female, compared to 23 percent in comparison 
communities. The average age for heads of farming
households is similar to that seen previously for all 
households; average age of the head of a farming 
household is 50.7 years in treatment communities
and 51.5 in comparison communities. There was no 
statistically significant difference in these 
characteristics between agricultural households in 
treatment and comparison communities. 
FIGURE 10: DEMOGRAPHICS OF 
FARMING HOUSEHOLDS
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
24
FINDINGS 1: HOUSEHOLD WELL-BEING, FOOD SECURITY, AND 
POVERTY
LIVELIHOODS
To measure subjective well-being, the household survey asked respondents to imagine a ladder with six 
steps, where the poorest were at the bottom and the richest were at the top. It then asked respondents 
where they, their neighbors, and their friends stood on that ladder at the time of the interview. On 
average, households in treatment communities said they stood on step 1.5, their neighbors were on step 
1.7, and their friends were on step 1.8. For each of these three categories, sampled comparison 
community households averaged just slightly lower, though the differences were always statistically 
significant (Figure 11).
FIGURE 11: HOUSEHOLD SUBJECTIVE WELL BEING
We also measure poverty 
objectively using the Poverty 
Probability Index (PPI), which 
assigns each household a likelihood 
of falling below the poverty line 
based on certain household 
characteristics and asset 
ownership.36 We consider two 
different poverty lines: Colombia’s 
2016 national poverty line,37 and the 
more extreme $1.25 US Dollars 
(USD) per day 2005 Purchasing 
Power Parity international poverty 
line. Figure 12 presents the poverty 
rates in the baseline sample. When
the national poverty line is 
considered, 42.9 percent of
treatment community households 
and 47.1 percent of sampled 
comparison community households
are below the national poverty line. 
When the more extreme poverty 
line of $1.25 per day is considered, poverty rates in our baseline sample are much lower at 8.0 percent 
for treatment community households and 9.0 percent for sampled comparison community households.
The differences in poverty rates between the treatment and comparison groups for both measures were 
statistically significant.
36 The PPI is developed by the Innovations for Poverty Action at Yale University. The PPI is based on 10 questions related to household 
characteristics and asset ownership, such as number of household members, having all children aged 6-12 enrolled in school, or owning clothes 
washing machine. Responses are scored and then totaled to generate an index for each household, and each point on the index corresponds to 
a pre-determined likelihood that the household with that score would fall below the poverty line. The Colombia user guide describes that the 
poverty line is “adjusted by a spatial price index, so that it is possible to compare…between regions or rural and urban areas.” For more 
information, see: https://www.povertyindex.org/country/colombia
37 PPI documentation for Colombia states that probability of poverty at the national poverty was calculated using the 2016 poverty line of 
COP$8,556 per person per day, equal to COP$10,087 per day after accounting for inflation. This is equal to approximately $2.73 USD per 
person per day, using the 2020 average exchange rate of $1 USD = $3,691.3 COP. For more information, see: 
https://www.povertyindex.org/country/colombia
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
25
FIGURE 12: BASELINE POVERTY RATES
To explore the high poverty rates further, we examined the main income sources of households in 
Figure 13. Wage employment and crop cultivation make up the main income sources for most 
households in both the treatment and comparison groups. Among sampled households in treatment 
communities, crop cultivation was the main income source for 39.2 percent, and wage employment was 
the main income source for another 39.2 percent; there was no statistically significant differences 
between the treatment and the comparison community households on these two dimensions. The other 
common main sources of income were commercial businesses, livestock, transfers (from government 
assistance programs or from family members outside the household), and real estate.
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
26
FIGURE 13: MOST IMPORTANT SOURCE OF INCOME
The high percentage of households earning their main income from wages are not necessarily employed 
in the private sector. In response to questions related to private sector employment (which we used to 
construct the PPI), only 16.6 percent of the households in the treatment communities mentioned that 
they have at least one household member employed by a private company at the time of the interview. 
The corresponding number for households in the comparison communities was 10.9 percent.38 In 
qualitative discussions, respondents mentioned the lack of formal private sector employment, and 
frequently mentioned that many people in their communities relied on informal daily wage labor, often 
working on local farms. As has already been alluded to and will become a theme throughout this report, 
the distribution of land in many areas is highly unequal, leaving many households without any access to 
agricultural land or with only a very small amount of land that is insufficient to make a living from.
Also, it is clear from both the quantitative and qualitative data that the significant diversity across the 
seven regions in the study also correlates to the variation in main income source across regions. From 
the qualitative data, agriculture, cattle ranching, and day labor are mentioned as primary sources in the 
Montes de Maria region; ranching, illicit crops, coffee and cacao, mining, and day labor are mentioned in 
Catatumbo; coffee, cacao, and ranching are mentioned in Southern Tolima; agriculture, mining, day 
labor, and informal employment are mentioned in Bajo Cauca; cacao, palm, illicit crops, fishing, and 
logging are mentioned in Tumaco and Olaya Herrera; ranching and farming are mentioned in Meta; and 
agriculture, including both licit and illicit crops, are mentioned in Northern Cauca. In some cases, group
discussion participants mentioned environmental factors that have harmed productivity and led to shifts
in activities. For example, multiple participants in Southern Tolima mentioned that climate change has 
led to a decline in cacao productivity and a shift away from this crop, while respondents in Tumaco 
mentioned that polluted waters meant they could no longer make a living from fishing, and in Olaya 
38 The difference between the two groups is statistically significant. Also, 9.7 percent of the households in the treatment communities had at 
least one member with a written employment contract and 8.3 percent had at least one member who was a boss or employer. The 
corresponding numbers for households in the comparison communities was 6.2 percent and 6.3 percent, respectively. The difference between 
the two groups for having at least one household member with a written contract is statistically significant. 
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
27
Herrera overlogging in the past has led to a decline in that activity today.
ACCESS TO CREDIT
With crop cultivation and informal daily wage labor being the main income source, it is not surprising 
that 25.8 percent of sampled households in treatment communities said they had a need for credit 
during the 12 months prior to the survey, and 17.2 percent said they received credit from a source
outside of the household over the same period (Table 6). These figures represent statistically significant 
differences from sampled comparison community households, who reported both lower need for credit 
(19.6 percent) and actual receipt of credit (13.9 percent). On the other hand, 5.8 percent of all sampled 
households in treatment communities said they had wanted credit over the past 12 months, but were 
unable to obtain it, while a similar figure (5.3 percent) reported the same among the comparison 
group.39 The average household in treatment communities borrowed $1.5 million COP over the 12-
month period prior to the survey, or approximately $402.7 USD. Among treatment community 
households who actually obtained loans, the average amount borrowed was higher, at $8.9 million COP, 
or approximately $2,408.2 USD. The average amount borrowed was lower among comparison 
households, though the differences are not statistically significant (Table 6).
TABLE 6: LOAN TAKE-UP AND SIZE OF LOAN
Overall Treatment 
Households
Comparison Households
Outcome N Mean N Mean N Mean Diff
Had need for credit in the past 12 
months (%) 2963 22.7 1510 25.8 1453 19.6 6.1 ***
Received credit (%) 2960 15.6 1510 17.2 1450 13.9 3.4 *
Wanted credit, but could not obtain it 2962 5.5 1509 5.8 1453 5.3 0.5 
Amount borrowed (USD) - All HHs 2943 333.2 1501 402.7 1442 260.9 141.8 
Amount borrowed (USD) - Only HHs 
who took a loan 444 2208.6 251 2408.2 193 1949.1 459.1 
*** p < 0.01, ** p < 0.05, * p < 0.1.
39 Note that the survey logic on these questions was such that the total of households receiving credit and wanting credit but not being able to 
obtain it does not necessarily add up to the percentage who had need of credit. For example, a household could have received credit for one 
need, but then had another need for credit that they were unable to obtain financing for. Additionally, the household may have had a need for 
credit but did not make any attempt to obtain it, and thus may not have indicated that they were unable to obtain it.
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
28
Figure 14 presents the sources 
of loans for the sampled 
households. Of those who 
obtained loans, the most 
common sources were banks
and cooperatives not 
including the Agrarian Bank 
(32.7 percent of all loans), the 
Agrarian Bank (24.4 percent 
of loans), informal lenders or 
loan sharks known as gota a 
gota lenders (14.6 percent of 
loans), and family or friends 
(9.2 percent). Differences 
between treatment and 
comparison community 
households in the prevalence 
of these loan sources, among 
those who took loans, were 
not statistically significant.
The gota a gota loan sharks 
are generally predatory 
lenders that lend amounts smaller than those typically provided by traditional banks; they charge high 
interest rates, sometimes resort to violence to ensure the loan is repaid, and are often affiliated with 
organized criminal groups. Gota a gota loans were particularly prevalent in the Montes de Maria region, 
accounting for about 42 percent of all loans there, and to a somewhat lesser extent in Tumaco and 
Sardinata, where they accounted for about 20 percent of all loans.
The most common uses for these loans were agriculture and livestock, including labor payments, purchase of 
inputs or machinery, post-harvest processing, and other agricultural services, followed by household 
expenses, including recreation, clothes, food, and health care, business investments, and buying or improving 
housing. Purchase of land accounted for only 1.5 percent of all loans among sampled households in treatment 
communities and 2.9 percent among those in comparison communities, with the difference not being 
statistically significant (Figure 15).
4.9
3.3
9.2
11.6
14.7
14.5
10.4
12.0
36.1
28.1
20.8
29.3
Diff.: -8.5
Diff.: 8.0
Diff.: -1.6
Diff.: 0.2
Diff.: -2.4
Diff.: 1.6
0 10 20 30 40
Percentage of Loans (%)
Other
Family /
Friends /
Employer
Gota a Gota
(Informal
Lender)
Cooperative
Other
Banks
Agrarian Bank
Note: * p < 0.1, ** p < 0.05, *** p < 0.01
Statistical significance test of the difference accounts for clustered standard errors.
Comparison Treatment
FIGURE 14: LOAN SOURCE
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
29
FIGURE 15: LOAN USE
Notwithstanding the use of loan, most households did not need to provide any collateral for loans. As 
shown in Figure 16, for 57.5 percent of the loans taken by sampled households in treatment communities
and for 55 percent of loans taken by comparison community households, no collateral was needed
(difference not statistically significant). Land was provided as collateral for 8.6 percent of all loans among 
sampled treatment community households and for 14.5 percent of all loans among sampled households in 
comparison communities (difference not statistically significant). Respondents also reported not having to 
show a land document for 75.4 percent of loans obtained, including 66.6 percent of all loans obtained from 
formal sources and 87.2 percent of loans obtained from informal sources. Of those who needed to 
provide a land document, a registered title or deed was by far the most common document that was 
provided, accounting for 15.8 percent of loans from formal sources (or 42.5 percent of all loans where a 
land document was shown). However, loans where no collateral was provided were for substantially lower 
amounts; the average loan without collateral was for $3.8 million COP ($1,031 USD), versus $10.7 million 
COP ($2,888 USD) for loans with collateral. The average loan where land or a house or building was used 
as collateral was even higher at $14.6 million COP ($3,956 USD). These results support the theory of 
change assumption that providing households with a formal title to their land will likely improve their 
ability to access loans from formal banks and obtain larger loans for productive investments.
FIGURE 16: LOAN COLLATERAL
3.7
1.2
16.3
21.2
11.0
6.2
15.6
10.4
4.3
6.6
12.9
9.1
1.5
2.9
34.7
42.3 Diff.: -7.7
Diff.: -1.4
Diff.: 3.8
Diff.: -2.3
Diff.: 5.3
Diff.: 4.8*
Diff.: -4.9
Diff.: 2.4*
0 10 20 30 40 50
Percentage of Loans (%)
Other
Household
Expenses
Pay Debts
Business
Investment
Consumer
Durables
Buy/Improve Housing
Purchase
Land
Agriculture/ Livestock
Note: * p < 0.1, ** p < 0.05, *** p < 0.01
Statistical significance test of the difference accounts for clustered standard errors.
Comparison Treatment
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
30
The qualitative data from group discussions and KIIs also shows that lack of collateral in the form of land 
titles is a major barrier to obtaining formal sector loans, and participants consistently mention easier 
access to credit as a main advantage of having a formal land title. As one community leader in 
Convención put it, 
“The challenge of not having [a legal title] for your land is that, well, many times you can’t access credit 
from the bank to improve the quality of life, to be able to work the land, because well…even though 
today there’s an alternative from the Agrarian Bank…if a person is the owner of the plot, even if you 
don’t have a legal title, they’re giving out some credit, but the amount is really low so a lot of times it’s 
not what the farmer needs.” (Community Leader KII, Convención.)
This suggests that the main benefit of a title is not simply access to credit, but rather access to higher 
quality credit, in the form of larger loans, formal financial sector credit sources, and lower interest rates.
According to one female community leader for Tumaco:
“Our community is really beaten down. There’s little economic activity because they government tore up 
our [coca] crops, and it hasn’t brought the projects [that it promised as alternative development 
programs]. And the people are in a [difficult] situation…because in reality, the main bank is the Banco 
Agrario, and it doesn’t want to give people loans…it’s not working with us like it should. And the other 
banks are day-to-day lenders – in reality they are gota a gota lenders and they’re sucking people dry, 
people are afraid to get involved in that because if you don’t pay, you’ve got a big problem coming.” 
(Community Leader KII, Tumaco.)
At the same time, participants frequently described their distrust of the formal banking system, balked at 
the high interest rates charged by formal banks, and expressed fear that if they had a failed harvest they 
would still need to repay the loan and end up deeper in debt. This suggests additional barriers may need 
to be overcome to provide access to the formal credit sector. 
FOOD SECURITY
The household survey asked several questions related to food scarcity over the 30-day period prior to 
the interview, to generate a standard measure of food insecurity. In general, food scarcity is more 
prevalent among sampled comparison community households, who were 5.6 percentage points more 
likely to have no food in the homes at some point in 30 days before the interview, 5.3 percentage points 
more likely to have a member who went to bed hungry because of lack of food, and 3.7 percentage 
points more likely to have a member go a full day without eating anything at all due to lack of resources 
to get food. These differences are all statistically significant (Figure 17).
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
31
FIGURE 17: PREVALENCE OF FOOD SCARCITY
Using these responses, combined 
with additional responses about the 
frequency with which these three 
hunger situations occurred over the 
past four weeks (rarely, sometimes, 
or often), we calculated each 
household’s score on the Household 
Hunger Scale (HHS40). The HHS is a 
food deprivation scale that measures 
the percent of households 
experiencing hunger. The scale takes 
values from 0 to 6, where 0-1 
equates to little to no hunger, 2-3 
denotes moderate hunger, and 4-6 
corresponds to severe hunger.
The results presented in Table 7 shows that respondents tended to be at the lower end of the HHS, on 
average, though sampled comparison community households had a significantly higher degree of hunger 
than sampled households in treatment communities. Based on the HHS, the percentage of households 
that experienced moderate to severe hunger was 12.5 percent among sampled households in 
comparison communities and 7.7 percent among those in treatment communities, a difference that was 
statistically significant. An additional consideration with respect to hunger is that while levels of hunger 
are low overall, hunger is highly concentrated among a small group of municipalities. Among households 
in the Coastal Nariño region, 25.1 percent suffer from moderate or severe hunger, compared to 15.1 
percent in Montes de Maria, 9.6 percent in Bajo Cauca, 5.9 percent in Southern Tolima, 4.6 percent in 
Meta, 3.6 percent in Catatumbo, and 2.9 percent in Northern Cauca.
TABLE 7: HOUSEHOLDS WITH MODERATE TO SEVERE HUNGER 
Overall Treatment Households Comparison Households
Outcome N Pct. N Pct. N Pct. Diff
All regions 2958 10.0 1507 7.7 1451 12.5 -4.8 ***
Coastal Nariño 407 25.1 203 15.3 204 34.8 -19.5***
Montes de Maria 610 15.1 303 13.5 307 16.6 -3.1 
Bajo Cauca 280 9.6 105 7.6 175 10.9 -3.2 
Southern Tolima 592 5.9 298 6.0 294 5.8 0.3 
Meta 456 4.6 269 4.1 187 5.3 -1.3 
Catatumbo 302 3.6 154 4.5 148 2.7 1.8 
Northern Cauca 311 2.9 175 0.0 136 6.6 -6.6**
Note: *** p < 0.01, ** p < 0.05, * p < 0.1.
While levels of hunger are low, it is important to keep in mind that the levels seen here may be 
impacted in the near future by the COVID-19 pandemic and political unrest happening in Colombia at 
the time of data collection. In qualitative discussions, respondents mentioned that these factors had 
40 USAID, 2019. USAID Food for Peace: Indicators for Emergency Program Performance Indicator Reference Sheets. Washington, DC.
17.7
12.1 14.3
9.1 7.0
3.2
Diff.: -5.6***
Diff.: -5.3**
Diff.: -3.7***
0
10
20
30
40
Percentage of Households (%)
No food to eat Went to sleep hungry Hungry for full day
Note: * p < 0.1, ** p < 0.05, *** p < 0.01
Statistical significance test of the difference accounts for clustered standard errors.
Comparison Treatment
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
32
disrupted supply chains by interrupting transportation, and made food less accessible due to inflation. 
However, we expect any such effects on hunger to be uniform across treatment and comparison 
communities, thus this will not pose a concern for the impact evaluation’s identification strategy.
NON-FARM ACTIVITIES
Although most households rely heavily on agriculture, about 17.6 percent of sampled households in 
treatment communities and 15.6 percent of comparison community households operate a non-farm 
income generating activity (IGA), such as a small business, household-based enterprise, or informal￾sector entrepreneurial activity (Table 8). Among all IGAs, 49.8 percent are located on the household 
premises, and most of the IGAs are newer—27 percent has been operating for 1 year or less and 58.3 
percent has been operating for 5 years or less. In addition, only 55.1 percent of the IGAs operated year 
round in the 12 months preceding the survey (Table 8). These figures are likely impacted by the 
COVID-19 pandemic. As one focus group participant in Tumaco put it, 
“Of course business was affected [by the pandemic]. A lot of people were left without work because 
they had to close [their businesses] and a lot went out of business.” 
Events such as the pandemic also has implications for the households who operate IGAs, as only 4.7 
percent of these activities are affiliated with a business association and cannot rely on an association’s 
help. Also, 84.3 percent of the IGAs sell their products or services to the final consumers, likely in their 
local communities, implying that they are not part of any value chain and are not connected to larger 
consumer markets. This is one area where LfP’s work under component 3 may improve outcomes
related to household income and poverty.
TABLE 8: NON-FARM INCOME GENERATING ACTIVITIES (IGAs)
Overall Treatment 
Households
Comparison 
Households
Outcome (over the past year) N Mean N Mean N Mean Diff
Household engaged in non-agricultural IGA
(% of households) 2961 16.6 1508 17.6 1453 15.6 2.1
IGA is located on household premise
(% of IGAs) 532 49.8 243 50.9 289 48.6 2.3
IGA has been operating for 1 years or less 
(% of IGAs) 530 27.0 290 26.9 240 27.1 -0.2
IGA has been operating for 5 years or less
(% of IGAs) 530 58.3 290 57.6 240 59.2 -1.6
IGA operated year-round in the past 12 
months (% of IGAs) 532 55.1 288 54.2 244 56.2 -2.0
Sell products or services to final consumers 
(% of IGAs) 534 84.3 290 85.2 244 83.2 2.0
IGA affiliated with a business association
(% of IGAs) 530 4.7 289 4.5 241 5.0 -0.5
Average monthly revenue from IGA (USD) 505 157.1 273 178.0 232 132.5 45.6
Note: *** p < 0.01, ** p < 0.05, * p < 0.1.
The average monthly revenue for the IGAs is $157.1 USD ($580,000 COP), including $178.0 USD 
($657,000 COP) for activities carried out by households in treatment communities and $132.5 USD 
($478,000 COP) for activities carried out by sampled comparison community households; this difference 
is not statistically significant. Figure 18 shows the percentages of treatment and comparison community 
household IGAs that generate revenues in different brackets. Among all activities, 53.7 percent bring in 
$99 USD or less per month, 22.4 percent bring in $100-199 USD, 5.7 percent bring in $200-233 USD, 
and 18.2 percent bring in more than $234 USD per month. This distribution is similar across sampled 
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
33
households in treatment and comparison communities.
FIGURE 18: INCOME FROM OFF-FARM INCOME GENERATING ACTIVITIES (IGAs)
FINDINGS 2: LAND USE AND AGRICULTURAL PRODUCTION
LAND HOLDINGS
On average, sampled households in both treatment and comparison communities managed 1.4 plots 
under any tenancy regime, and used 0.9 plots for agricultural purposes (Figure 19). Across all plots 
managed by households under any tenancy regime, sampled households in treatment communities held 
an average of 17.4 hectares, versus 9.3 hectares for sampled comparison community households, though 
the difference is not statistically significant. Similarly, sampled households in treatment communities had 
an average of 8.4 hectares under agricultural use, compared to 5.8 hectares for sampled comparison 
community households, and again the difference is not statistically significant. 
26.324.5
13.413.2 12.9
16.8
12.1
6.6
10.310.6
3.0 2.6
5.6 5.9
16.4
19.8
Diff.: -1.8
Diff.: -0.2
Diff.: 3.9
Diff.: -5.5*
Diff.: 0.3
Diff.: -0.5
Diff.: 0.3
Diff.: 3.4
0
10
20
30
Percentage of Activities (%)
$0-33 $34-66 $67-99 $100-133 $134-166 $167-199 $200-233 $234+
Note: * p < 0.1, ** p < 0.05, *** p < 0.01
Statistical significance test of the difference accounts for clustered standard errors.
Comparison Treatment
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
34
FIGURE 19: NUMBER OF PLOTS AND AMOUNT OF LAND 
The average amount of land held and used for agriculture masks the highly unequal nature of land 
holdings in the survey regions, and the manner in which these inequalities favor those who already have 
formalized land titles. Despite the mean land holdings reported above, median land holdings are far 
lower; the median for total land holdings among sampled households in treatment is 1.5 hectares, and 
with a median of 1.0 hectares for land under agricultural use. Overall, 31.9 percent of all sampled 
households had less than 0.25 hectares, including 35.9 percent of those with no registered title to any 
plot, and 17.2 percent of those with at least one registered title41, representing a statistically significant 
difference (Table 9, row 1). Among all sampled households, 45.9 percent held one hectare or less (Table 
9, rows 1-3), while 17.5 percent held more than ten hectares (Table 9, rows 7-8). This is confirmed in 
the qualitative data, where participants mentioned that the titling process was expensive, so land titles 
were more commonly held by wealthier, large landowners.
The issue of unequal land distributions came up frequently in the qualitative discussions with 
communities, with nearly every community mentioning that access to land was a significant problem for 
many families in the area, and that most people did not hold enough land to make a living from, while a 
small number of large landowners held the majority of the land. Qualitative discussions also highlighted 
that decades of armed conflict have made this problem worse, as many families sold their land for low 
prices during the worst years of the conflict, while others took advantage of the low prices to 
accumulate larger land holdings. 
41 Throughout this report, we compare households with no registered title to any plot and those with at least one registered title. For our 
purposes, "household with at least one registered title” is defined as households who say they have a land document for at least one plot 
owned by someone in the household, and where the household says that document is a registered title (escritura pública). Thus, the definition is 
based on household self-reporting of their tenure status.
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
35
TABLE 9: REGISTERED VS. NOT REGISTERED PLOTS BY LAND AMOUNT
Total land held by 
household (ha)
Overall Registered title 
to 1+ plot
No registered title 
to any plot
Freq. Pct. (%) Freq. Pct. (%) Freq. Pct. (%) Diff.
<0.25 ha 946 31.9 109 17.2 837 35.9 -18.6***
0.251-0.5 129 4.4 23 3.6 106 4.5 -0.9
0.51-1.0 286 9.6 33 5.2 253 10.8 -5.6***
1.1-3.0 572 19.3 86 13.6 486 20.8 -7.2***
3.1-5.0 225 7.6 66 10.4 159 6.8 3.6***
5.1-10.0 289 9.7 98 15.5 191 8.2 7.3***
10.1-20.0 220 7.4 91 14.4 129 5.5 8.9***
20.1 ha < 298 10.1 126 19.9 172 7.4 12.6***
Total 2965 100.0 632 100.0 2333 100.0 n/a
Note: *** p < 0.01, ** p < 0.05, * p < 0.1.
Figure 20 shows the tenure status of agricultural plots in the study communities. Sampled households in 
treatment communities owned (whether titled or untitled) 63.6 percent of the plots they occupied for 
agricultural activities, compared to 66.1 percent of plots occupied by sampled comparison community 
households. However, households in treatment communities reported having formal, registered titles
for only 19.3 percent of the plots; the corresponding number for sampled comparison community 
households was 19.4 percent. This suggests a large portion of plots lack a formal land title, and that a 
large portion of the households in the treatment communities will benefit from mass land formalization.
It should be noted that due to the difficulty of the land titling and formalization process in Colombia and 
changes in laws and entities responsible for titling over the years, there is significant confusion around what, 
exactly, constitutes a formal, registered title. This was confirmed in qualitative discussions in some 
communities where participants debated whether “title” and “deed” meant the same thing, and whether 
documents issued by now-defunct land agencies were still valid. While the figures reported above for owners 
with a registered title represent respondents who correctly provided the name of an official, registered title 
(escritura publica), it is possible that in some cases respondents may in fact possess some other, quasi-formal 
document, and that the true prevalence of formal titles is lower than what is reported here.
FIGURE 20: TENURE STATUS OF AGRICULTURAL PLOTS 
66.1 63.6
19.4 19.3
46.6 44.3
23.8 23.7
5.8 8.8 2.6 2.4 1.2 1.5
Diff.: -2.5
Diff.: -0.1
Diff.: -2.3
Diff.: -0.1
Diff.: 3.0**
Diff.: -0.3 Diff.: 0.3
0
20
40
60
80
100
Percentage of Agricultural Plots (%)
Owned, titled
or untitled Owned,
registered title
Owned, no
registered title
Usufruct Rented Sharecropping De Facto
Occupation
Note: * p < 0.1, ** p < 0.05, *** p < 0.01
Statistical significance test of the difference accounts for clustered standard errors.
Comparison Treatment
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
36
After land ownership, usufruct rights42 are the most common type of tenure; 23.7 percent plots occupied 
for agricultural activities by households in the treatment communities and 23.6 percent plots for those in 
the comparison communities were under usufruct rights (Figure 20). Renting is the third most common 
tenure regime in both treatment and comparison communities, while very few households describe 
themselves as sharecropping or occupying (i.e., squatting) land. In general, qualitative discussions 
suggested that in some regions, usufruct, rental, and sharecropping arrangements fit a community need 
where some have more land than they are able to put to productive use, and many have insufficient land 
to make a living from but insufficient resources to acquire more. Within this context, usufruct or 
sharecropping arrangements may be popular, as they allow the land to be put to use, while an otherwise 
landless farmer with no means to pay can either pay later to work the land now (e.g., through a 
sharecropping arrangement), or access land in exchange for providing a service to the landowner, such 
as hired labor on another plot the landowner owns or caring for the property of an absent landowner.
LAND USE
Among plots controlled by sampled households in treatment communities, 65.0 percent were used for 
agricultural purposes, compared to 68.0 percent for those in comparison communities, a difference that 
is not statistically significant. Figure 21 shows the most important uses of agricultural plots for sampled 
households in treatment and comparison communities, as reported by the survey respondent. These 
were permanent crops (21.1 versus 26.9 percent of plot area, respectively), transitory crops (18.3 
versus 16.2 percent of plot area, respectively), unused or fallow land (13.9 versus 11.9 percent of plot 
area, respectively), single species livestock (13.9 versus 10.8 percent of plot area, respectively), and 
dwelling (12.1 versus 13.9 percent of plot area, respectively).
FIGURE 21: MOST IMPORTANT USES OF AGRICULTURAL PLOTS 
42 In the Colombian context, “usufruct” refers to having permission from the landowner to live on the plot without payment. In most cases, the 
usufruct tenant would be an employee of the landowner who receives housing and a small plot as a benefit of employment, an employee caring 
for the land of an absentee landowner (and provided some land to cultivate for themselves), or tied to the landowner through a family 
relationship.
13.9 12.1
26.9
21.1
16.2
18.3
7.2 5.6
10.8
13.9
3.3 3.8
0.8 0.7
5.8 7.1
1.3 1.4
11.9
13.9
Diff.: -1.8
Diff.: -5.9*
Diff.: 2.0
Diff.: -1.6
Diff.: 3.1
Diff.: 0.6
Diff.: -0.2
Diff.: 1.3
Diff.: 0.1
Diff.: 2.0
0
10
20
30
Percentage of Area (%)
Dwelling Permanent
Crops
Transitory
Crops
Mixed
Crops
Single
Species
Livestock
Mixed
Species
Livestock
Pasture Forest Rented
Out
Unused/Fallow
Note: * p < 0.1, ** p < 0.05, *** p < 0.01
Statistical significance test of the difference accounts for clustered standard errors.
Includes only agricultural plots.
Comparison Treatment
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
37
For plots where any part was left fallow or unused, respondents were asked the reason why the area 
was not being put to use. As presented in Table 10, primary reasons included lack of money to cultivate 
the land (57.7 percent of plots with unused land), to conserve the land (16.8 percent), lack of labor to 
work the land (11.3 percent), poor soil quality (10.4 percent), and because the land is in a period of rest 
(10.2 percent). Only 1.4 percent mentioned uncertainty of land ownership as the reason for leaving land 
unused. Thus, the provision of formal land titles may on its own resolve one barrier to productive land 
use for only a small proportion of households; however, titling may also spur increased development of 
unused land among a much greater proportion of households where lack of resources serves as the 
primary barrier to land development, if formal land titles also enable greater access to credit. 
In qualitative discussions, respondents mentioned poor quality of the land (e.g., PH level, swampy land, 
overuse) and lack of resources as reasons for leaving land unused. Lack of resources for developing the 
land was also mentioned as a reason why land might then be rented or sharecropped out, or given in 
usufruct. Qualitative participants described the lack of resources for developing the land as essentially a 
mismatch, where some in the community may have more than enough land, but no ability to pay for 
labor or inputs. 
TABLE 10: REASONS FOR LEAVING LAND FALLOW OR UNUSED
Overall
Reasons Freq. Pct. (%)
Lack of money 587 57.7
Poor soil quality 106 10.4
Lack of water 15 1.5
There is no one who works the land 115 11.3
Uncertainty due to safety conditions on the vereda 24 2.4
To conserve it or soil conservation 171 16.8
Uncertainty about land ownership 14 1.4
Because the land is in a period of rest 104 10.2
Topography 23 0.8
Other: which? 59 5.8
Total 1017 100.0
Note: This question was a multiple response question, so counts for individual items may not add up to the 
total number of respondents. *** p < 0.01, ** p < 0.05, * p < 0.1. 
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
38
CROPS PRODUCED
Sampled households in treatment and 
comparison areas are similar in terms of the 
number of crop and animals they derive 
livelihoods from. The survey asked households 
to list the three most important crops and animal 
species for the income of the household. As 
Figure 22 shows, among households engaged in 
agricultural activities, 29.0 percent of sampled 
households in treatment communities and 34.0 
percent of those in comparison communities had 
only one crop or livestock activity; 26.8 percent 
of sampled households in treatment communities
and 26.5 percent of comparison community 
households had two crop or livestock activities; 
and 44.2 percent of sampled households in 
treatment communities and 39.5 percent of those 
in comparison communities listed three crop or 
livestock activities.
In terms of the most common crop or livestock species, sampled households in treatment and 
comparison communities are similar, with the exception of a statistically significant higher share of 
households in treatment communities cultivating cacao or raising cattle (Table 11). Overall, the most 
common crops or livestock produced by sampled households are plantain (33.8 percent), roots and 
tubers (32.8 percent), and poultry (27.9 percent). The commodities most frequently listed as the main 
income generating sources in the qualitative interviews are produced by a smaller share of households: 
coffee (18.2 percent), cattle (15.3 percent), and cacao (15.3 percent). While these commodities may be 
the primary economic drivers in specific regions, they are not equally important in all of the regions in 
this study. Additionally, these commodities require larger investments and quantities of land and are thus 
likely to be cultivated by a smaller group of landholders.
TABLE 11: MOST COMMON CROPS AND LIVESTOCKS
Overall Treatment Comparison
Commodity Category Freq. Pct. (%) Freq. Pct. (%) Freq. Pct. (%) Diff.
Roots & Tubers 774 32.8 404 35.0 370 30.7 4.3
Plantain 796 33.8 407 35.3 389 32.3 3.0
Coffee 428 18.2 168 14.6 260 21.6 -7.0
Maize 401 17.0 195 16.9 206 17.1 -0.2
Fruits 367 15.6 172 14.9 195 16.2 -1.3
Cacao 275 11.7 170 14.7 105 8.7 6.0*
Cattle 360 15.3 214 18.5 146 12.1 6.4**
Poultry 658 27.9 291 25.2 367 30.5 -5.3
Fish, pigs, & other 
livestock 183 7.8 96 8.3 87 7.2 1.1
Total farming households 2358 100.0 1154 100.0 1204 100.0 n/a
Note: This question was a multiple response question, so counts for individual items may not add up to the total number of respondents. 
Figures include farming households only.
*** p < 0.01, ** p < 0.05, * p < 0.1. 
34.0
29.0 26.5 26.8
39.5
44.2
Diff.: -4.9
Diff.: 0.3
Diff.: 4.7
0
10
20
30
40
50
Percentage of Farming Households (%)
1 2 3
Note: * p < 0.1, ** p < 0.05, *** p < 0.01
Statistical significance test of the difference accounts for clustered standard errors.
Includes farming households only.
Comparison Treatment
FIGURE 22: NUMBER OF CROP AND ANIMAL 
SPECIES PRODUCED PER HOUSEHOLD
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
39
In qualitative discussions, participants described sophisticated processes for making land use and crop 
selection decisions, involving a study of market and input prices, time to harvest, and land and labor 
requirements. On this note, one community member in Fuente de Oro expressed that households were 
often too sensitive in how they responded to price fluctuations. As this participant put it, 
“There’s a difficult situation for us campesinos, where we’re all playing a blind game of chicken with 
each other. Why? Because let’s say the price of yuca goes up a little, then everyone turns around and 
plants yuca. Now there’s a problem with too much supply, the price goes through the floor. Now, since 
everyone was planting yuca, maybe everyone stopped growing plantain, let’s say, so plantain gets 
expensive and everyone plants plantain, and we come back to the same problem.” (Community 
member, Fuente de Oro)
Additionally, qualitative discussions suggested that more and more households are preferring livestock to 
crops. The reason relates to tenure security and a preference for more mobile investments. Under this 
interpretation, acquisitions of poultry, cattle, or small livestock might be seen more as investments than 
income generating activities, and households may prefer investments that are easier to take with them or 
sell if they have to leave the farm because of violence, a land restitution claim, or land tenure issue. As a 
focus group participant from San Juan Nepomuceno said: “Every year, I used to sell my harvest, and the first 
thing I’d do, I’d buy an animal. If I needed a horse or donkey, I’d buy it, and if I needed a calf, a heifer, and that, 
fine. When I left [that farm], I had almost 15 animals. We had to abandon [the farm because of the violence].” In 
the household survey, a small number of households (under 9 percent) reported starting new crops or 
livestock, but those that did, major and minor livestock was clearly the most popular choice. 
INCOME FROM AGRICULTURAL ACTIVITIES
Consistent with the fact that sampled 
households in the treatment 
communities own more land (result 
shown earlier), they earned more from 
their farm activities on all plots they 
owned. As shown in Figure 23, 
households in treatment communities
earned $466.0 USD ($1.7 million COP)
per month from all agricultural and 
livestock activities on their agricultural 
plots, compared to $278.8 USD ($1.0
million COP) for those in comparison 
communities, representing a statistically 
significant difference. However, they 
were not more productive—average 
income per hectare per month was 
$79.4 USD ($293,000 COP) among 
sampled households in treatment 
communities, compared to $123.6 USD ($456,000 COP) per hectare per month among comparison 
community households, and this difference is statistically significant. Finally, looking at income from land
under cultivation, average income was $202.0 USD ($746,000 COP) per hectare per month among 
sampled households in treatment communities, compared to $266.2 USD ($983,000 COP) per hectare 
per month among comparison community households, and this difference is not significant.
Average farm income hides the large degree of inequality in the income distribution for households in 
these regions. As shown in Figure 24, nearly a third of farming households earn essentially nothing (i.e., 
between $0 and $5 USD per month) from agriculture, suggesting their agricultural activities are devoted 
278.8
466.0
123.6
79.4
268.2
202.0
Diff.: 187.2**
Diff.: -44.2**
Diff.: -66.2
0
100
200
300
400
500
Average Monthly Farm Income (USD)
Total Income Income Per Hectare Income Per Hectare
Used for Ag.
Note: * p < 0.1, ** p < 0.05, *** p < 0.01
Statistical significance test of the difference accounts for clustered standard errors.
Includes only plots used for agricultural activities.
Comparison Treatment
FIGURE 23: AGRICULTURAL INCOME AND 
PRODUCTIVITY
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
40
almost completely to domestic consumption. On the other hand, just 20.2 percent of sampled 
households in treatment communities and 14.7 percent of those in comparison communities earns $250 
USD or more per month.
FIGURE 24: INCOME DISTRIBUTION FOR HOUSEHOLD AGRICULTURAL ACTIVITIES
The above results are also supported by 
the fact that about a third of the farming 
household grow crops for subsistence 
only and do not sell any part of it—67.7 
percent of sampled households in 
treatment communities and 62.7 percent 
in comparison communities sell any part 
of any crops they harvest, as seen in
Figure 25. A similar pattern was also 
observed for livestock. Additionally, very 
few households process any of their crop 
harvest (18.0 percent of crop farming 
households in treatment communities, 
15.7 percent of those in comparison 
communities) or products from animal 
(5.1 percent of animal raising households 
in treatment communities, 3.5 percent of 
those in comparison communities). None 
of these differences is statistically significant.
Figure 26 shows agricultural revenues over the 12-month period prior to the survey. Farming 
households among the sample in treatment communities generated a total of $1,441.5 USD ($5.3 million 
COP) from their top three crops or livestock species, compared to $1,090.9 USD ($4.0 million COP)
among those in comparison communities. Average total revenue from crops was $458.3 and $508.6 
USD ($1.7 million and $1.9 million COP) for households in treatment and comparison communities, 
respectively. Most of this revenue came from processed crops, yet less than 20 percent of crop farming 
households do any processing. This suggests that the vast majority of crop farming revenue is being 
generated by the relatively small share of households who do crop processing, which disproportionately 
includes coffee and cacao farmers. On the other hand, animal farming generated an average of $983.2 
33.4 31.4
16.7 14.3 11.2 13.0 10.0 7.0
14.1 14.7
6.2 5.6 4.2 6.1 4.1
7.9
Diff.: -2.0
Diff.: -2.4
Diff.: 1.8
Diff.: -3.0**
Diff.: 0.5
Diff.: -0.5 Diff.: 1.9
Diff.: 3.8**
0
10
20
30
40
Percentage of Farming Households (%)
$0-5 $6-33 $34-66 $67-99 $100-250 $251-500 $501-1250 $1251+
Note: * p < 0.1, ** p < 0.05, *** p < 0.01
Statistical significance test of the difference accounts for clustered standard errors.
Includes only farming households.
Comparison Treatment
FIGURE 25: SOURCES OF AGRICULTURAL
REVENUE
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
41
USD ($3.6 million COP) over the 12-month period before the survey for treatment community 
households compared to $582.3 USD ($2.1 million COP) for comparison community households. A 
smaller share of animal farming revenue comes from processing compared to crop farming. 
FIGURE 26: TOTAL REVENUE FROM AGRICULTURAL
PRODUCTIVE INVESTMENTS
Table 12 presents information on productive investments in land by study households in the 12 months 
before the survey, which is one area that LfP might have a positive impact on as the baseline results 
indicate, in line with LfP’s theory of change, that households are indeed more likely to invest in lands 
they officially own. Among sampled households in treatment communities, 48.1 said they had made 
investments in any plot under their control compared to 42.1 percent of those in comparison 
communities. On average, sampled households in treatment communities made 0.7 investments valued 
at $504.8 USD ($1.9 million COP compared to 0.5 in comparison communities valued at $500.3 USD
($1.88 million COP). Average investment value only among those who actually made land investments
was $1,067.5 USD ($3.9 million COP) for sampled households in treatment communities compared to 
$1,219.9 USD ($4.5 million COP) among those in comparison communities. Among plots with 
permanent crops, sampled households in treatment communities had last planted permanent crops 3.4 
years ago, on average, compared to 3.8 years for comparison community households.43
43 Some permanent crops have a limited productive life, and need to be replanted periodically. While replanted crops may take time to reach 
maturity and peak productivity, representing a period where they are generating little or no income, replanting old plants whose peak 
productive years have passed is an important investment.
1090.9
1441.5
649.3
518.2 508.6 458.3 408.8 329.6
582.3
983.2
240.5 188.6
Diff.: 350.7
Diff.: -131.2
Diff.: -50.3
Diff.: -79.2
Diff.: 400.9
Diff.: -52.0
0
250
500
750
1,000
1,250
1,500
Total Annual Revenue From
Top 3 Crops & Animals (USD)
Total farming revenue Total revenue
from processing
Total crop
farming revenue
Total processed
crop revenue
Total animal
farming revenue
Total processed
animal revenue
Note: * p < 0.1, ** p < 0.05, *** p < 0.01
Statistical significance test of the difference accounts for clustered standard errors.
Comparison Treatment
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
42
TABLE 12: PRODUCTIVE INVESTMENT ON LAND IN PAST 12 MONTHS
Overall Treatment 
Households
Comparison 
Households
Outcome (over the past year) N Mean N Mean N Mean Diff
Household made any investment
(% of households) 2695 45.1 1512 48.1 1453 42.1 6.0**
Number of investments per household 2695 0.6 1512 0.7 1453 0.5 0.1***
Average investment, all households
(USD) 2695 502.6 1512 504.8 1453 500.3 4.5
Average investment, only those who made 
an investment (USD) 1311 1136.7 715 1067.5 596 1219.8 -152.3
Year since last permanent crop planting, 
only those with permanent crops 1298 3.6 658 3.4 640 3.8 -0.4
Note: *** p < 0.01, ** p < 0.05, * p < 0.1.
The most common investments made, as shown in Figure 27, include investments in housing (15.7 percent 
of all plots), soil conservation, water reserves, and natural disaster repairs (6.0 percent), and fruit, timber, 
or other commercial trees (7.9 percent). It is important to note that plots with registered titles are more 
likely to have made any investment (46.9 percent) than those without a registered title (36.5 percent).
FIGURE 27: INVESTMENT TYPES IN REGISTERED AND NON-REGISTERED PLOTS
When comparing households with registered titles for at least one plot to those without a title, as 
shown in Figure 28, households with a title invested significantly more ($977.9 USD) than those without 
($373.9 USD), on average. Further, the value of investments among households who did invest in their 
plots was greater, on average, among those with titles ($1,872.9 USD) than those without ($889.1 
USD), though this difference is not statistically significant. Among plots where no investment was made, 
the most common reasons for not investing were a lack of resources (64.3 percent) and that it was not 
necessary to make investments (24.5 percent) as opposed to uncertainty about land ownership, which 
was cited by only 2.1 percent households (result not shown in a Table or Figure). The barriers to 
investment mentioned in the qualitative discussions paint a similar picture, where lack of access to 
capital is described as the main barrier. Though rarely mentioned in the household survey, security 
conditions were also mentioned as a reason for not making investments in the qualitative discussions. 
One community member in Convención described how being too ambitious could make someone a 
target for extortion and open them up to problems with others in the community.
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
43
FIGURE 28: INVESTMENT AMOUNTS IN REGISTERED AND NON-REGISTERED PLOTS
Finally, we look at plot values, including any buildings and other investments on the land, and livestock 
values as appraised by households themselves, in Figure 29. Changes in these self-reported values at 
endline would also be a proxy indicator for the impact of LfP on investments made on land (or at least 
on households’ perceived value of the land). On average, plots under the control of sampled households 
in treatment communities were valued at $18,662.4 USD ($68.9 million COP), compared to $13807.9 
USD ($51.0 million COP) among plots controlled by sampled comparison community households. The 
average value of livestock owned by sampled households in treatment communities was $1452.5 USD 
($5.4 million COP), compared to $877.0 USD ($3.2 million COP) among those in comparison 
communities with farming activities. Cattle accounted for nearly all of this value, despite being an activity 
practiced by a minority of households with agricultural activities. Cattle accounted for $1307.9 USD 
($4.8 million COP) of the livestock owned by farming households in treatment communities, and $806.7 
USD ($3.0 million COP) among farming households in comparison communities. However, qualitative 
interviews suggested that cattle ranchers may be quite different from other households, and skew 
towards larger landowners with more resources.
FIGURE 29: PLOT AND LIVESTOCK VALUE Average Value (USD)
13,807.9 877.0 806.7 33.0
18,662.4 1,452.5* 1,307.9 36.3
Plots All livestock
Comparison Treatment
Cattle Poultry
373.9
977.9 889.1
1872.9
Diff.: 604.1*
Diff.: 983.8
0
500
1,000
1,500
2,000
Average Investment Value (USD)
All Households Households Who Made Investments
Note: * p < 0.1, ** p < 0.05, *** p < 0.01
Statistical significance test of the difference accounts for clustered standard errors.
No registered public deed Has registered public deed
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
44
FINDINGS 3: LAND ACQUISITION AND TENURE SECURITY
LAND ACQUISITION TREND AND MODE
Plots controlled by sampled households in 
both treatment and comparison 
communities were similar in terms of 
when they were acquired by households. 
Figure 30 shows the percentage of all 
plots acquired each year since 1960, by 
treatment status. For example, the figure 
shows that slightly under two percent of 
all plots currently held by sampled 
households in either treatment or 
comparison municipalities were acquired 
in 2000. Acquisition rates increase over 
the period, particularly after 2000, 
suggesting many plots held by households 
in the sample were relatively recent 
acquisitions. Half of all plots in the 
baseline sample were acquired in 2011 or 
later.
For plots owned by the respondent household (whether titled or untitled), respondents were asked 
how the plot was acquired. We present the results in Table 13 broken down by whether plots have a 
registered title or not.44 Titled plots are significantly more likely to have been acquired through 
purchase (65.2 percent of titled plots vs. 46.5 percent of untitled plots), and significantly less likely to 
have been acquired through inheritance (27.6 percent of titled plot vs 40.5 percent of untitled plots). 
Over 88 percent of plots in the study areas were obtained via these two modes of land acquisition 
(purchase and inheritance). 
TABLE 13: MODE OF LAND ACQUISITION
Overall Has registered title No registered title
Mainly, how was the plot 
acquired?
Freq. Pct. (%) Freq. Pct. (%) Freq. Pct. (%) Diff.
Purchase 1364 51.5 461 65.2 903 46.5 18.7***
Inheritance 982 37.1 195 27.6 787 40.5 -13.0***
Possession 35 1.3 8 1.1 27 1.4 -0.3
Titling Vacant Land 10 0.4 3 0.4 7 0.4 0.1
Occupied vacant lot 43 1.6 1 0.1 42 2.2 -2.0***
Special land program 111 4.2 27 3.8 84 4.3 -0.5
Land restitution 5 0.2 2 0.3 3 0.2 0.1
Agrarian reform program 4 0.2 2 0.3 2 0.1 0.2
Exchange/Gift 74 2.8 8 1.1 66 3.4 -2.3***
Other 20 0.8 0 0.0 20 1.0 -1.0***
Total plots 2648 100.0 707 100.0 1941 100.0 n/a
44 This definition of "registered title” is defined at the plot level as the plot being owned by the interviewed household, and the household 
reporting they a registered title (escritura pública) for the plot. Thus, the definition is based on household self-reporting of their tenure status on 
the plot in question.
0
2
4
6
8
Percentage of Plots (%)
1960 1970 1980 1990 2000 2010 2020
Year Plot Was Acquired
Comparison Treatment
Note: Trends use polynomial smoothing function.
FIGURE 30: LAND ACQUISITION TRENDS
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
45
Note: This question was only asked for plots owned by the respondent household (whether titled or untitled). 
*** p < 0.01, ** p < 0.05, * p < 0.1. 
LAND RIGHTS
Given that purchase and inheritance are the two most common modes of land acquisition, we asked 
households who owned or occupied plots (i.e., excluding sharecropping, plots rented in or taken in 
usufruct) about their perceived land rights. As shown in Figure 31, sampled households in treatment 
communities said they had they had the right to sell the plot for 75.8 percent of plots, compared to 69.5 
percent of plots for comparison community households, representing a statistically significant difference. 
A somewhat higher percentage (89.8 and 87.1 percent of owned or occupied plots in treatment and 
comparison communities, respectively) said they had the right to bequeath the plot.
We also examined perceived land rights by plots with and without formal titles. As expected, those with 
formal titles had statistically significantly stronger perceived land rights. Also, a higher percentage of 
those without a formal title (86.0 percent) said they have the right to bequeath compared to those who 
said they have the right to sell (66.3 percent). This discrepancy may reflect local customary norms
where having a title is uncommon, together with the particular complexities of the Colombian land 
context. For example, in Sardinata one focus group participant said, “People don’t like to rent or sell. People 
here have a tradition, only [transferring land] within families and people you know…for security.” Another 
community leader in Convención said, “Convención is a red zone, a conflict zone. [You need] that permission 
document from the mayor’s office, where they certify that if you’re selling, you were selling of your own free will, 
nobody was forcing you.” In some other cases, the discrepancy between perceived rights to sell or 
bequeath land could be accounted for by ethnic communal territories where land cannot be sold. In any 
case, the findings at baseline imply that titling is likely to improve perceived land rights for those 
currently without a formal title. 
FIGURE 31: PERCEIVED LAND RIGHTS
We also examined household perceived knowledge of their land rights and access to land conflict 
resolution resources (Figure 32). Respondents were asked to rate the degree to which they agreed or 
disagreed with each of three statements on a scale from 1 (Strongly Disagree) to 5 (Strongly Agree). For 
all three, respondents rated their level of agreement as approximately a 3, corresponding to a neutral
stance. Sampled households in treatment communities rated their level of knowledge and access to legal 
resources slightly higher than those in comparison communities, and the differences are statistically 
69.5
75.8
87.1 89.8
Diff.: 6.3**
Diff.: 2.7
0
20
40
60
80
100
Percentage of Plots (%)
Right to sell owned/occupied plot Right to bequeath
owned/occupied plot
Comparison Treatment
66.3
91.1 86.0
95.5
Diff.: 24.8***
Diff.: 9.4***
Right to sell owned/occupied plot Right to bequeath
owned/occupied plot
No registered public deed Registered public deed
Note: * p < 0.1, ** p < 0.05, *** p < 0.01
Statistical significance test of the difference accounts for clustered standard errors.
Includes plots owned or occupied by respondents only. Does not include plots rented in, or taken in usufruct or sharecropping.
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
46
significant. Knowledge in treatment communities might be higher because of LfP’s field presence as they 
prepare to implement component 1. However, results indicate there is considerable room to improve 
household knowledge of land rights and conflict resolution options in the LfP implementation area. 
FIGURE 32: PERCEIVED KNOWLEDGE OF LAND RIGHTS
LAND DOCUMENTATION
For each plot controlled by households in the sample, respondents were asked whether the household 
had any document to verify their right to occupancy, and the type of document held. The evaluation team 
consolidated documents into four categories: (1) registered title (escritura publica), comprising formal land 
titles registered in the land registry with the Superintendencia de Notariado y Registro (Superintendency of 
Notaries and Registry – SNR) and in the cadaster with IGAC; (2) unregistered titles, comprising formal 
land titles that have not been registered in the cadaster or with SNR; (3) other formal documents, 
including legal documents that are not formal titles, such as a will, written contract, or other document 
from one of the national land agencies; and (4) informal or semi-formal documents, such as a signed letter, 
extrajudicial statement, or verbal contract. Figure 33 shows that among all plots owned or under de facto 
occupation by respondents in the sample, 46.7 percent had no land document to verify the household’s 
right to occupancy.45 Documents held by households for plots owned or under de facto occupation 
included registered titles (26.0 percent), unregistered formal titles (2.1 percent), other formal documents 
(18.9 percent), and informal or semi-formal documents (6.2 percent).46 For plots with legal titles, 19.3 
percent have been obtained since 2015, 16.8 percent were obtained from 2010-2014, and 24.2 percent 
were obtained between 2000 and 2009, while 30.6 percent were obtained before 2000. 
45 Table A.G.2 in Annex G presents additional statistics by municipality. 46 For plots under usufruct, rental, or sharecropping agreements, 97.6 percent of plots did not have any land document to show.
2.8 3.1 2.9 3.1 3.0 3.2
Diff.: 0.3*** Diff.: 0.1* Diff.: 0.2**
1
2
3
4
5
Level of Agreement
(1=Strongly Disagree,5=Strongly Agree)
I know more about my
land rights than I did last year
I have access to legal
representation if I have a land conflict
I know where to go if
I have a land conflict
Note: * p < 0.1, ** p < 0.05, *** p < 0.01
Statistical significance test of the difference accounts for clustered standard errors.
Comparison Treatment
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
47
FIGURE 33: LAND DOCUMENT TYPES AND TRENDS
For plots with any type of land document, households were asked whose names appeared on the 
document. Figure 34 shows that regardless of whether the plot has a formal, registered land title or some 
other document, plots are mostly similar in terms of who appears on the document. However, male and 
female decisionmakers from a household are significantly more likely to be listed jointly on registered title 
(9.0 percent) than on other document (5.9 percent). The difference is mostly accounted for by a 
statistically significant greater share of plots with a document other than a registered title where the male 
decisionmaker is the sole person listed on the document. Qualitative data sheds some light on why this is; 
any past titling done by the Instituto Colombiano de Reforma Agraria (Colombian Institute for Agrarian 
Reform - INCORA) and Instituto Colombia de Desarrollo Rural (Colombian Institute of Rural Development –
INCODER) included both male and female decisionmakers on the documents they issued. One finding in 
the figure that stands out is that it is far more common for either the male or female decisionmaker to be 
the sole person listed, than for the two to be listed together, regardless of the document type. This is 
likely because formal marriage is relatively uncommon among the sample and that people are hesitant to 
put their partner’s name on a document when they are not formally married. Nonetheless, fewer female 
decisionmakers are listed by themselves than male decisionmakers on titles of any kind.
FIGURE 34: MALE VS. FEMALE DECISIONMAKER NAME ON LAND TITLE
For plots owned by respondents without any type of land document, they were asked their reasons for 
not formalizing their tenure status. As shown in Table 14, the most common reasons provided were 
50.5 45.2
28.9 28.3
5.9 9.0
1.6 2.8 0.5 1.4
12.3 13.2
Diff.: -5.3*
Diff.: -0.6
Diff.: 3.1**
Diff.: 1.3 Diff.: 0.9*
Diff.: 1.0
0
20
40
60
Percentage of Plots (%)
Male Decisionmaker
Only
Female Decisionmaker Only Male & Female
Decisionmakers Jointly
Male Decisionmaker
Jointly w/ Other
Female Decisionmaker
Jointly w/ other
Other
Note: * p < 0.1, ** p < 0.05, *** p < 0.01
Statistical significance test of the difference accounts for clustered standard errors.
Includes plots with some form of land document only.
No registered public deed Has registered public deed
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
48
that it was not necessary (34.6 percent of plots) and that the household lacked the resources to 
formalize (29.8 percent). A large share (23.8 percent) also cited confusion or customs around the 
inheritance process for land. LfP should be particularly helpful for those who have not titled due to lack 
of resources, information, or the difficult judicial process. As LfP had already begun laying the 
groundwork for programming at the time of the baseline survey, some communities were already aware 
of this work and eagerly awaiting the benefits. As one community member from Fuente de Oro said:
“USAID has taken a really important step in updating the documents for land titles because it has given 
us an opportunity not to have to spend so much money, it’s accelerated the processes, and we’re really 
grateful USAID has been an entity that has promoted the titling process, it’s accelerated it, and that’s a 
good thing.” (Community Member, Fuente de Oro)
However, the large share of households that considers a formal land title to be unnecessary is also echoed
in the qualitative data. In general, there does not appear to be a strong sentiment that a formal title 
provides additional protection against land conflict or displacement. With the exceptions of Tumaco and 
Coastal Nariño, most communities in the qualitative discussions claimed that forced displacement was a 
thing of the past, and that nowadays everyone in the community knows where plot boundaries are, and 
that expropriation is rare. In the absence of formal titles, communities have developed their own systems
that include informal documents, authentication of informal documents at the police station, local 
knowledge and agreement amongst community members regarding land ownership, and selling or 
transferring land to only trusted community members. Additional reasons given by participants in 
qualitative discussions for not formalizing included avoiding taxes and having plots in areas that were not 
eligible for titling, such as those within forest reserves or areas at high risk for natural disasters.
TABLE 14: REASONS FOR NOT FORMALIZING
No registered title
For what reasons have you not formalized or registered 
ownership of the land ?
Freq. Pct. (%)
Not Necessary 371 34.6
Lack of Info 125 11.7
Lack of Resources 319 29.8
Difficult Judicial Process 73 6.8
In Process 120 11.2
Collective/Ethnic Territory 55 5.1
Other 255 23.8
Total Plots 1071 100.0
Note: This question was a multiple response question, so counts for individual items may not add up to the total number of respondents. 
Question only asked for plots described by households as owned and untitled. Note this does not include plots that households described 
as being titled, but where some other type of document was provided when asked what type of document the household had for the plot. 
Notwithstanding their interest in acquiring formalization documents, a large share of plots controlled by
sampled households in treatment (81.6 percent) and comparison (78.6 percent) communities were 
physically demarcated in some way (difference not statistically significant).
Finally, Figure 35 shows the amount households pay or are willing to pay for land documents. For plots 
that had some type of land document, those with registered titles paid $207.3 USD ($765,000 COP) for 
their document, on average, compared to $47.2 USD ($174,000 COP) for those with some other form of 
documentation. For plots with no land documents, households stated they would be willing to pay $54.2 
USD ($200,000 COP), on average, for plots owned or under de facto occupation, compared to $14.2 
USD ($53,000 COP), on average, for plots taken in usufruct, rental, sharecropping, or other agreements. 
However, for 35.8 and 80.9 percent of owned/de facto occupation and usufruct/rental/other plots, 
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
49
respectively, respondents said they would not be willing to pay any amount to obtain land documents. By 
providing all of the inputs necessary for the GoC to formalize each plots tenure status and issue 
documentation, LfP does not anticipate that plot owners will need to pay anything to obtain their titles.
FIGURE 35: AMOUNT PAID AND WILLINGNESS TO PAY FOR TITLE
TENURE SECURITY
For each plot, households were asked how likely it was that they would involuntarily lose ownership or 
use rights to the plot in the next five years (Figure 36).47 Among all plots, respondents said they were 
either “not at all likely” or “somewhat unlikely” to lose rights for 79.9 percent of plots controlled by 
sampled households in treatment communities and 82.7 percent of plots controlled by sampled 
comparison community households. By this measure of perceived tenure security, sampled households 
in treatment communities consider approximately 16 percent of their plots to be tenure insecure.48 The 
results also show that tenure security is notably higher for owned plots (whether titled or untitled) than 
those taken in usufruct, sharecropping, or rented in, while tenure security for titled plots is only slightly 
higher than for untitled plots that are owned. In other words, households express higher tenure 
insecurity for plots under usufruct, rental or sharecropping arrangements, although LfP’s interventions 
are not targeted to provide formal land documents for such agreements. There was no statistically 
significant difference in perceived tenure security between the treatment and the control community 
households.
FIGURE 36: PERCEIVED TENURE SECURITY
47 Table A.G.3 in Annex G presents additional statistics by municipality. 48 This excludes approximately four percent of plots where the respondent gave a neutral response to this question.
Average 
amount Paid
Has registered 
public deed $207.3
No registered 
public deed $47.2
Not willing to
pay anything
For plots owned/de 
facto occupation 35.8%
For plots taken in 
usufruct/rental/ 
sharecropping 80.9%
Amount 
willing to Pay
For plots owned/de 
facto occupation $54.2
For plots taken in 
usufruct/rental/ 
sharecropping $14.2
82.7 79.9
94.6 93.0 91.6 88.3
63.9 61.4
Diff.: -2.8
Diff.: -1.6 Diff.: -3.3
Diff.: -2.5
0
20
40
60
80
100
Percentage of Plots
Not Likely to Lose Rights (%)
All plots Titled plots, owned Untitled plots, owned
Usufruct/
Rented/
Other plots
Note: * p < 0.1, ** p < 0.05, *** p < 0.01
Statistical significance test of the difference accounts for clustered standard errors.
Comparison Treatment
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
50
This is confirmed by qualitative discussions as well, where participants mostly viewed the possibility that 
their land would be taken from them as remote. To the extent that respondents viewed expropriation as a 
possibility, the sources of expropriation that were mentioned as relevant for landowners included having 
the land taken by the bank or government, while forced displacement was generally not regarded as a 
significant risk. Other sources of expropriation risk included having the land taken by squatters or renters
who stop paying rent and refuse to leave, though these issues were only likely to impact a relatively small 
percentage of households who have large landholdings.
These findings are echoed in 
responses to another survey question 
where, for each plot, respondents were 
asked how worried they were that they 
might lose the right to use the plot 
within the next five years (Figure 37). 
Among sampled households in 
treatment communities, respondents 
said they were either “not worried” or 
“not at all worried” for 78.7 percent of 
plots, compared to 80.7 percent of 
plots for sampled comparison 
community households. Similar to the 
previous results, this measure of tenure 
security also suggests that tenure 
security is notably higher for owned 
plots (whether titled or untitled) than 
those taken in usufruct, sharecropping, 
or rented in, while tenure security for titled plots is only slightly higher than for untitled plots that are 
owned. Still, there is a notable proportion of owned plots (whether titled or not) for which respondents 
do express tenure insecurity via concern about potential loss of rights in future. There was no 
statistically significant difference in perceived tenure security between the treatment and the control 
community households
As a complementary measure 
of perceived tenure security, 
for each plot we also asked
households how many years 
they could let the plot lay fallow 
or unused without being 
worried about losing the right 
to use the plot. The results are 
presented in Figure 38. While 
households said they were not 
worried about losing their 
rights for the vast majority of 
plots via the two tenure 
security measures presented 
above, nearly half said they 
could not leave their land fallow 
for any period without losing 
their use rights, including for 
80.7 78.7
89.4 91.8 86.3 84.6
68.5 63.7
Diff.: -2.1
Diff.: 2.3
Diff.: -1.7
Diff.: -4.9
0
20
40
60
80
100
Percentage of Plots Not Worried
About Losing Rights (%)
All plots Titled plots, owned Untitled plots, owned Usufruct/
Rented/
Other plots
Note: * p < 0.1, ** p < 0.05, *** p < 0.01
Statistical significance test of the difference accounts for clustered standard errors.
Comparison Treatment
46.4 45.8
8.6
14.2
45.0
40.0
Diff.: -0.6
Diff.: 5.6***
Diff.: -5.0
0
10
20
30
40
50
Percentage of Plots (%)
Could not leave fallow Could leave fallow for limited time Would never lose land
Note: * p < 0.1, ** p < 0.05, *** p < 0.01
Statistical significance test of the difference accounts for clustered standard errors.
Comparison Treatment
FIGURE 37: WORRY ABOUT LOSING RIGHTS
FIGURE 38: ABILITY TO LEAVE LAND FALLOW
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
51
45.8 percent of plots controlled by sampled households in treatment communities and 46.4 percent of 
those by sampled comparison community households. On the other hand, 14.2 percent of plots under 
control by sampled households in treatment communities could be left fallow or unused for some 
limited amount of time, compared to 8.6 percent of plots controlled by sampled comparison community 
households, which represents a statistically significant difference. Sampled households in treatment 
communities said they would never lose land rights to 40.0 percent of plots, regardless of how long they 
left them unused, compared to 45.0 of plots controlled by sampled comparison community households.
The results here, including the finding that households with formal titles are only marginally more 
tenure secure than those who own plots without titles, likely stem from two issues. First is the 
intense demand for land by households that are either landless or have insufficient land. Many 
households are already living on land that has been previously abandoned by former owners; this 
appears somewhat common and some communities seem to accept that leaving the area would mean 
forfeiting their land. As one community member in Caldono said, “Well at least in my community, as far 
as I know right now, nobody has been forcibly displaced. But like they say, if I leave here, if I leave the 
municipality or if I go to another municipality, that means losing my land.” Similarly, in other discussions, 
renting out unused lands was described as important to make sure that “it is not just left there”, with 
the implication that this might expose the land to occupation by squatters. Second is the low incomes 
of many households and the extent to which some are indebted. Focus group participants listed debts 
to the bank as a significant source of insecurity; given improved access to loans for households with
titled plots, this may be an issue that disproportionately affects households with formal titles. The 
implication may be that if an indebted household were to leave land fallow, they would be unable to 
repay their debts and therefore lose their land.
To examine perceived tenure security further, households were asked an additional three questions 
to assess their level of agreement with four statements about whether their land could be taken by 
the government or outside groups without fair compensation, whether others in the community 
respected their property rights, and whether they were confident conflict would not arise over their 
land in the future. The results presented in Figure 39 suggest that households perceive the greatest 
risk of having land taken from them by outside groups, which many households would interpret to 
mean “armed groups”, but could also include private companies, large landowners, or groups of 
squatters: just 58.0 percent of sampled households in treatment communities and 50.3 percent of 
those in comparison communities agreed with the statement that “outside groups cannot take my 
land without fair compensation”, representing a statistically significant difference. Households 
perceived less risk of encroachment from others in their community and the government.
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
52
FIGURE 39: CONFIDENCE IN NOT LOSING LAND 
LAND CONFLICTS AND DISPLACEMENT 
Table 15 presents results on land disputes, conflicts, and displacements, which were relatively rare in the 
sampled communities. Just 2.5 percent of sampled households in treated municipalities reported having 
any land-related conflicts or disputes on any plot in the 12-month period before the survey, compared 
to 2.2 percent of sampled households in comparison communities.49 Among sampled households in 
treatment communities, 3.1 percent reported losing land in the past 12 months due to natural disasters, 
forced displacements, or other reasons, compared to 4.7 percent of those in comparison communities.50
Similarly, just 2.8 percent of sampled households in treatment communities and 3.1 percent of those in 
comparison communities said they had been threatened with eviction in the past 12 months.51
TABLE 15: LAND DISPUTES, LOSS, AND EVICTION IN PAST 12 MONTHS
Overall Treatment 
Households
Comparison 
Households
Outcome (over the past year) N Mean N Mean N Mean Diff
Has land-related disputes or conflicts
(% of households) 2963 2.3 1510 2.5 1453 2.2 0.3
Lost land in past 12 month 
(% of households) 2959 3.9 1508 3.1 1451 4.7 -1.6
Threatened with eviction in past 12 month 
(% of households) 2963 2.9 1512 2.8 1451 3.1 -0.3
Note: *** p < 0.01, ** p < 0.05, * p < 0.1.
49 Among households that experienced disputes, the most common sources of conflict were plot boundaries (30.4 percent of all households 
with disputes), inheritance (18.8 percent), and land titling (11.8 percent).
50 Land loss was mostly caused by natural disaster (81.5 percent of households with any land loss), while forced displacement accounted for a 
smaller share (11.1 percent). Households who mentioned losing land due to forced displacement in the past year were highly concentrated in 
just a few municipalities: 19 of the 32 were located in Olaya Herrera, 4 were in Tumaco, and 3 each were in the Bajo Cauca municipalities of 
Zaragoza and Cáceres.
51 The most common source of eviction threat was armed groups and it is highly spatially concentrated: of the 52 households that mentioned 
being threatened with eviction by armed groups in the past year, 22 were in Olaya Herrera, 10 were in Tumaco, 7 were in Cáceres, and 3 were 
in Zaragoza.
82.8 81.0
50.3
58.0
96.7 93.5
66.2 72.0
Diff.: -1.7
Diff.: 7.7**
Diff.: -3.2***
Diff.: 5.8
0
20
40
60
80
100
Percentage of Households Who Agree (%)
Govt cannot take
my land without
fair compensation
Outside group cannot
take my land without
fair compensation
My land boundaries
are respected by
my community
I am confident that
conflict will not arise
over my land in the future
Note: * p < 0.1, ** p < 0.05, *** p < 0.01
Statistical significance test of the difference accounts for clustered standard errors.
Includes only households that own at least one plot.
Comparison Treatment
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
53
Although the level of conflicts and displacement reported in the last 12 months were low, Colombia has 
seen massive internal migration due to forced displacement over decades of conflict, particularly 
affecting rural populations, though numbers have trended down in recent years. According to World 
Bank data from the Internal Displacement Monitoring Centre, 106 thousand Colombians were newly 
forcibly displaced due to conflict in 2020;
52 approximately 7 million Colombians, or around 15 percent 
of the country’s population, have been displaced since the beginning of the conflict in the 1960s 
(UNHCR, 2015). Within this context, families or individuals may be forced to leave home in the past 
due to threats from armed groups (e.g., due to community leadership positions, refusal to cooperate, or 
threats of forced recruitment), or have their land stolen by armed groups; they may also be victimized in 
other ways, including the targeted assassination of a family member, permanent injury from an accident 
with a landmine or other unexploded ordinance, or face sexual violence.
Among sampled households in treatment communities, 55.4 percent said no one in their household had 
ever experienced forced displacement, while 44.4 percent said they had either had to leave home or 
abandon their land at some point in the past due to threats from armed groups; 54.0 percent of 
households identified as victims of the armed conflict, and 42.8 percent said they were registered in the 
National Victims Registry, which is necessary for receiving the government benefits directed to conflict 
victims (Figure 40). These figures are significantly higher than the national averages and reflect the 
greater degree to which the conflict has impacted these municipalities. Sampled households in treatment 
communities report slightly higher levels of victimization than those in comparison communities, though 
none of the differences is statistically significant.
FIGURE 40: HOUSEHOLDS WITH ANY MEMBERS WHO ARE CONFLICT VICTIMS
LAND GOVERNANCE AND LAND MARKET ACTIVITY
To assess households’ land governance perceptions, respondents were read two statements and asked 
about their level of agreement. As shown in Figure 41, 44.7 percent of sampled households in treatment 
communities and 40.3 percent of those in comparison communities agreed with the statement “Land 
rights are clear and easy to understand”. A similar share, 41.1 and 37.4 percent, respectively, agreed 
52 https://data.worldbank.org/indicator/VC.IDP.TOCV?locations=CO
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
54
with the statement “Land rights are well-protected by authorities”. Differences between sampled 
households in treatment and control communities were not statistically significant.
FIGURE 41: HOUSEHOLD PERCEPTION OF LAND GOVERNANCE
We assess land market activities among sampled households in Table 16. Among all sampled households, 
1.9 percent purchased any land in the municipality they were surveyed in during the 12-month period 
before the survey, including 2.1 and 1.7 percent of sampled households in treatment and comparison 
communities, respectively. Among those who purchased land, the mean amount of land purchased was
9.5 hectares. However, the distribution is highly unequal, with a median purchase of just 0.08 hectares.
The unequal land distribution is also reflected in the percentage of households who had usufruct, rental, 
or sharecropping agreements. Among sampled households in treatment communities, 41.4 percent had 
usufruct, rental, or sharecropping agreements to occupy land belonging to someone else, while just 6.7 
percent gave out land their households owned to someone else under such an agreement; numbers for 
the comparison community households were similar (41.8 and 5.4 percent, respectively).
TABLE 16: LAND MARKET ACTIVITY
Overall Treatment 
Households
Comparison 
Households
Outcome (over the past year) N Mean N Mean N Mean Diff
Bought land in municipality in past 12 
months (% of households) 2962 1.9 1509 2.1 1453 1.7 0.5
Taken at least 1 plot in usufruct, rent, or 
sharecropping in past 12 months
(% of households)
2965 41.6 1512 41.4 1453 41.8 -0.4
Given at least 1 plot in usufruct, rent, or 
sharecropping in past 12 months 
(% of households)
2965 6.0 1512 6.7 1453 5.4 1.3
Note: *** p < 0.01, ** p < 0.05, * p < 0.1.
Given that the land rental market is more active than the land buying/selling market, we look further 
into the plots that sampled households accessed via rental agreements (Table 17). The average area 
rented was 2.7 hectares, with a large, statistically significant difference between plots used for 
agricultural purposes (5.5 hectares) and those that are not (0.1 hectares). Similarly, the average 
household renting a plot paid $650.1 USD ($2.4 million COP per month), including $1,338.3 USD ($4.9 
million COP) for agricultural plots and $46.8 USD ($200,000 COP) for non-agricultural plots, on 
40.3 44.7 37.4 41.1
Diff.: 4.4
Diff.: 3.7
0
20
40
60
80
100
Percentage of Households Who Agree (%)
Land rights are clear and easy to understand Land rights are well-protected by authorities
Note: * p < 0.1, ** p < 0.05, *** p < 0.01
Statistical significance test of the difference accounts for clustered standard errors.
Comparison Treatment
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
55
average. Rental price was higher per hectare for non-agricultural plots, $567.2 USD ($2.1 million COP) 
per hectare per month, compared to $248.5 USD ($900,000 COP) for agricultural plots, likely reflecting
the fact that non-agricultural plots are more likely to be located in centros poblados, where prices may be 
higher than dispersed rural areas.
TABLE 17: LAND RENTAL MARKET CHARACTERISTICS
All plots rented Rented agricultural 
plots
Rented non￾agricultural plots
Mean SD Mean SD Mean SD Diff.
Plot area (ha) 2.7 13.1 5.5 18.5 0.1 0.3 5.4***
Rental price per month 
(USD) 650.1 12,385.5 1,338.3 1,8121.7 46.8 33.5 1291.5
Rental price per month per 
hectare (USD)* 253.8 674.6 248.5 599.1 567.2 2465.6 -318.7***
Total number of plots 417 199 218 417
Note: Includes plots taken in rent only. *** p < 0.01, ** p < 0.05, * p < 0.1. 
*Rental price per month per hectare is weighted by plot area.
FINDINGS 4: PUBLIC SERVICE DELIVERY, PUBLIC-PRIVATE 
PARTNERSHIPS, AND ILLICIT CROP SUBSTITUTION
TRUST IN INSTITUTIONS
The survey asked respondents to rate the degree to which they trusted various local and national 
government institutions. In general, levels of trust in most institutions are low (Figure 42). The lowest 
levels of trust were expressed for judges, with just 28.4 percent of sampled households in treatment 
communities saying they trust judges. Similarly, low levels of trust were seen for local institutions—
trust for local Community Action Board (Junta de Acción Communal), town hall (Alcaldía Municipal), and 
the Municipal Land Office (Oficina Municipal de Tierras) was 43.2, 25.5, and 43.1 percent, respectively, 
among sampled treatment community households. Numbers were similar for comparison community 
households. Levels of trust were highest for notaries, representing the only institution where the 
percentage of households expressing trust is notably above 50 percent.
It is important to note that a substantial portion of households felt they were not familiar enough with 
many institutions to form an opinion. Specifically, respondents expressed high levels of unfamiliarity with 
the Municipal Land Office, Office of Public Instruments (Oficina de Instrumentos Públicos), Land Registry 
Office (Oficina de Catastro), National Land Agency (Agencia Nacional de Tierras), and Land Restitution Unit 
(Unidad Restitución de Tierras).
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
56
FIGURE 42: TRUST IN LOCAL AND NATIONAL GOVERNMENT INSTITUTIONS
Trust in municipal government
was also low as shown in Figure 
43. Respondents were given three 
statements about the municipal 
government and were asked about 
the extent to which they agreed 
with those statements. Just 20.0 
percent of sampled households in 
treatment communities agreed 
with the statement “There is little 
corruption in the municipal 
government”. Similarly, just 26.3 
percent and 34.5 percent of 
sampled households in treatment 
communities agreed with the 
statements “I am very satisfied with 
the municipal government’s work in 
rural development” and “I trust the 
municipal government works on 
behalf of the interests of all its 
citizens”, respectively. 
The qualitative data provides additional insight into these low levels of trust and satisfaction with both local 
and national government. Common themes from the group discussions and interviews with community 
leaders are that (1) local government is unable to resolve local conflicts, (2) the national government is 
actively against them, (3) lack of available land leads people to want the government to make land available 
and trust erodes when it fails to do so, (4) there is frustration stemming from a perceived lack of results 
from the 2016 peace accords, and (5) isolation, changing policies and institutions, and poor communication 
create a situation where many feel totally abandoned. Below, we detail specific findings from the qualitative 
data on sentiments regarding specific government entities relating to land.
48.4 43.2
28.4 25.5
39.4 43.1 46.4 50.6 51.5 52.2
67.1 65.7
51.9 52.5
26.1 28.4
51.0 51.9 49.0 49.7
Diff.: -5.2
Diff.: -2.9
Diff.: 3.7
Diff.: 4.2 Diff.: 0.7
Diff.: -1.5
Diff.: 0.6
Diff.: 2.3
Diff.: 0.9 Diff.: 0.7
0
20
40
60
80
100
Households Who Trust Institution (%)
Community
Action
Board
Town
Hall
Municipal
Land
Office
Office of
Public
Instruments
Land
Registry
Office
Notary Ombudsperson Judges National
Land
Agency
Land
Restitution
Unit
Note: * p < 0.1, ** p < 0.05, *** p < 0.01
Statistical significance test of the difference accounts for clustered standard errors.
Comparison Treatment
20.1 20.0
29.0 26.3
36.4 34.5
Diff.: -0.1
Diff.: -2.7
Diff.: -1.8
0
20
40
60
80
100
Percentage of Households Who Agree (%)
There is little
corruption in
municipal government
I am very satisfied with
municipal government's
work in rural development
I trust the municipal government
works on behalf of the
interests of all its citizens
Note: * p < 0.1, ** p < 0.05, *** p < 0.01
Statistical significance test of the difference accounts for clustered standard errors.
Comparison Treatment
FIGURE 43: TRUST IN MUNICIPAL GOVERNMENT
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
57
Trust in national land offices (INCORA, INCODER, ANT)
Patchwork policies and changing laws and institutions have not been properly communicated to 
communities and have created a situation where households do not know what is happening at the 
national level, eroding trust. The name of the national agency responsible for land titling has changed 
multiple times in recent years, from the INCORA, to the INCODER, to the ANT, creating confusion. 
The story of one focus group participant in Ortega, whose family had benefitted from a titling initiative 
under the now-defunct INCORA, is informative: “You’d go to the bank, and no, it wouldn’t work there 
either. They wouldn’t take [the land document from INCORA]…They say INCORA doesn’t exist anymore...Who 
knows if that’s true.” The importance of the problem here is echoed by a PNIS official when asked if there 
is corruption in the land titling process in Colombia:
“I think there is [corruption in the land administration system]. That changing, like of the institutions’ 
names, of changing from one entity to another, from one name to another, or one reform to another. In 
a certain sense it seeks to hide something in this process and in some way you lose the work that has
been started [under a past institution], and that becomes a form of corruption…It affects the process 
that the families use for interacting with the government, right? When you finally identify how to do the
administrative process to get a title, a change comes and that brings a new way to get that title, and so 
if that happens on repeated occasions, well, there’s never going to be a standard for how to get a title, 
because every entity is demanding something different”. (PNIS official)
These problems are compounded in remote communities where access even to local government 
offices is limited. In one group discussion in La Macarena, with a community that clearly falls within the 
borders of La Macarena according to all materials reviewed by the evaluation team, focus group 
respondents expressed frustration that they did not even know to which municipality or department 
they belonged. The group was hostile to the idea that their community belonged to La Macarena and the 
department of Meta, whose municipal and departmental capitals are substantially further away and more 
difficult to access than a municipality in the neighboring department of Caquetá. 
However, some discussion participants across multiple communities mentioned previous titling efforts 
by the INCORA or INCODER in positive terms. Importantly, these were described as efforts somewhat 
similar in nature to LfP, where the agencies brought the titling process directly to the communities. The 
main issue here, however, is that it is not always clear if these titles were officially registered in the land 
cadaster with IGAC and land registry with SNR to make them completely formal, and that changes in 
policies and land use planning can present barriers if families attempt to register those titles later on.
Additionally, even if the land was formally titled by INCORA or INCODER, the qualitative data suggests 
that some of the original title holders have died and the land has been inherited, which puts the new 
owner back under informal status if the inheritance is not properly registered.
Trust in land restitution process (URT)
Discussion participants had mixed views regarding Colombia’s land restitution process, carried out by 
the URT. Overall, the process is described in positive terms by those who have benefitted from it and in 
negative terms by those who perceive themselves to have lost from the process. In terms of specific 
challenges that URT and the land restitution process faces regarding confidence and trust, the most 
common themes are that the process lacks transparency, is being taken advantage of by people 
perceived to be undeserving of the benefits, is sometimes carried out without the input of all parties 
involved, and is insufficiently deferential to community leaders.
However, multiple respondents expressed support and gratitude for the land restitution process. For 
example, a focus group participant in San Juan Nepomuceno attributed the little that his family had today 
to the land restitution process. Similarly, a community leader in Ataco said that many in the area had 
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
58
been able to reclaim their land through the process and receive land titles. But, participants in several 
discussions mentioned former community residents who they felt had legitimately sold land as many as 
30 years ago, are now attempting to get that land back through the restitution process, after the plot in 
question had changed hands multiple times. Still others complained that when the URT comes to do plot 
studies, it takes plot measurements even if one of the interested parties is not present, and does not 
always seek the input of all neighbors when determining plot boundaries.
PUBLIC SERVICE DELIVERY
In addition to trust in local and 
national government, the survey 
also examined household access 
to and satisfaction with public 
services. Figure 44 shows average 
travel time to access different 
services. On average, sampled 
households in both treatment and 
comparison communities live 
about an hour away from the 
administrative center of the 
municipality, the nearest health 
center, and the nearest 
agricultural market, which can be 
defined as reasonably accessible in 
the Colombian context. The 
nearest school was accessible 
within a half hour or less distance. 
One the other hand sampled 
households reported low levels of 
satisfaction with local service 
delivery as shown in Figure 45. 
Households were asked to rate
their levels of satisfaction, on a 
scale from 1 (Very Dissatisfied) to 
5 (Very Satisfied), in the quality of 
their municipality’s roads, 
irrigation infrastructure, and 
overall infrastructure services. 
Among sampled households in 
treatment communities, 29.6 
percent said they were either 
satisfied or very satisfied with the 
quality of the municipality’s roads, 
compared to 24.0 percent of 
those in comparison communities. Satisfaction in irrigation infrastructure was far lower: just 4.6 percent 
of sampled households in treatment communities and 2.1 percent of those in comparison communities
said they were satisfied or very satisfied with the quality of irrigation infrastructure in their municipality. 
For satisfaction with overall infrastructure services, 12.4 percent and 14.8 percent of sampled 
households in treatment and comparison communities, respectively, were at least satisfied. 
1.0
1.3
0.9
1.0
0.3
0.5
0.8
1.1
Diff.: 0.3*
Diff.: 0.1
Diff.: 0.1
Diff.: 0.2
0
.5
1
1.5
Travel Time (Hours)
Municipal Center Nearest Health Center Nearest School Nearest Ag Market
Note: * p < 0.1, ** p < 0.05, *** p < 0.01
Statistical significance test of the difference accounts for clustered standard errors.
Comparison Treatment
24.0
29.6
2.1
4.6
14.8
12.4
Diff.: 5.6*
Diff.: 2.6***
Diff.: -2.4
0
10
20
30
Satisfied Households (%)
Quality of Roads Irrigation Infrastructure Overall Infrastructure Services
Note: * p < 0.1, ** p < 0.05, *** p < 0.01
Statistical significance test of the difference accounts for clustered standard errors.
Comparison Treatment
FIGURE 44: ACCESS TO LOCAL SERVICES
FIGURE 45: SATISFACTION WITH LOCAL SERVICES
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
59
PUBLIC-PRIVATE PARTNERSHIPS
Component 3 of the LfP activity will 
help strengthen economic development
and create enabling environment for 
licit economic activities through 
strategic public-private partnership. At 
baseline, we looked at household 
participation in development projects in 
general. As shown in Figure 46, very few 
benefitted from government 
development projects or participated in 
other productive projects (12.1 percent
and 8.8 percent of sampled households 
in treatment communities, respectively). 
Also, a very low percentage of 
households agreed that it had become 
easier to find a job in the municipality in 
the past 12 months (8.8 percent and 
11.7 percent of treatment and comparison community households, respectively). These low numbers 
implies that there are lot of room for providing households with opportunities to participate in 
productive economic development projects. 
PARTICIPATION IN ALTERNATIVE DEVELOPMENT PROGRAMS
Alternative development programs are rural development programs designed to reduce the presence of 
illicit crops by offering a variety of incentives. Overall, 6.2 percent of sampled households in treatment 
communities said they had ever benefitted from alternative development programs, compared to 2.8 
percent of those in comparison communities, representing a statistically significant difference (Figure 47). 
As expected, due to the concentration of coca in specific areas, participation in alternative development 
programs is spatially concentrated. Among all sampled households, the areas with highest participation in 
alternative development programs are Tumaco and coastal Nariño (16.8 percent), Bajo Cauca (12.5 
percent), Meta (3.5 percent), and Catatumbo (2.7 percent). Among all households who said they had 
benefitted from alternative development programs, the most common programs were Forest Guarding 
Families (33.1 percent), followed by PNIS (27.8 percent), and Voluntary Substitution Agreements (11.3 
percent).
FIGURE 47: PARTICIPATION IN ALTERNATIVE DEVELOPMENT PROGRAMS
13.2 12.1 11.7
8.8 8.1 7.7
Diff.: -1.0
Diff.: -2.9*
Diff.: -0.4
0
5
10
15
Percentage of Respondents (%)
My household has benefitted
from government development
projects, past 12 mos.
It has become easier
to find a job in my
municipality, past 12 mos.
I have participated in or
benefitted from productive
projects, past 12 mos.
Note: * p < 0.1, ** p < 0.05, *** p < 0.01
Statistical significance test of the difference accounts for clustered standard errors.
Comparison Treatment
FIGURE 46: PARTICIPATION IN PUBLIC AND 
PRIVATE PROJECTS
Participated in 
Alternative 
Development Program
Treatment 6.2%
Comparison 2.8%
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
60
The trends in terms of the percentage of 
households participating in alternative 
development programs in treatment 
and comparison communities were
roughly parallel between 2010 and 2015
(Figure 48). Starting around 2016, the 
trends diverged, with participation in 
treatment communities far surpassing 
that in comparison communities. This is 
due in part to key features of the LfP 
program design, as municipalities were 
selected in concert with the GoC’s 
development plans, and an important 
consideration was selecting 
municipalities that aligned with the 
geographies where PNIS is working.
The evaluation team will be able to 
correct for some of these differences 
later on, during the household-level 
matching process (see Balance and 
Power section of the report).
ILLICIT CROP CULTIVATION
Figure 49 shows results regarding household participation in illicit crop production. Among sampled 
households in treatment communities, 9.0 percent admitted to ever having cultivated coca, compared to 
2.3 percent of those in comparison communities and this difference is statistically significant. A 
somewhat lower share of sampled households in treatment communities, 2.8 percent, admitted to 
growing coca currently, compared to 1.2 percent of those in comparison communities. Although not 
shown here, sampled households in treatment communities said they had 1.4 hectares under cultivation
among those households currently cultivating coca, on average, compared to 1.0 hectares for sampled 
comparison community households.
FIGURE 49: PARTICIPATION IN ILLICIT CROP PRODUCTION
2.3
9.0
1.1
2.8
10.1
0.5
Diff.: 6.7***
Diff.: 1.7
Diff.: -9.6***
0
5
10
15
20
Percentage of Respondents (%)
Ever Cultivated Coca Household Currently
Growing Coca
Refused Coca Questions
Note: * p < 0.1, ** p < 0.05, *** p < 0.01
Statistical significance test of the difference accounts for clustered standard errors.
Comparison Treatment
0
.5
1
1.5
2
Percentage of Households (%)
<= 2010 2012 2014 2016 2018 2020
Year of Alt. Dev. Program Participation
Comparison Treatment
FIGURE 48: PARTICIPATION TRENDS IN 
ALTERNATIVE DEVELOPMENT PROGRAMS
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
61
The sensitive nature of these questions presented a security issue in the comparison municipality of 
Olaya Herrera, where teams were warned by community leaders not to ask questions on coca. As such, 
the data was able to identify coca growing households in Olaya Herrera among those who volunteered
coca as one of their most important crops in the crop roster section of the survey; for those who did 
not openly admit to this in the crop roster section, the coca growing status of the remaining households 
in Olaya Herrera is simply marked as “refused”. This accounts for the much larger share of refusals to 
the coca questions among sampled comparison community households (10.1 percent), compared to the 
share among sampled households in treatment communities (0.5 percent).
It is important to note that security conditions during baseline data collection prevented enumerator 
teams from accessing a small number of communities in coca-growing municipalities in the baseline 
sample. This in turn may have introduced some bias in the baseline estimates for coca cultivation in 
certain municipalities. For example, coca growing areas in Cáceres are known to be located in the south
of the municipality, but none of the eight communities in the baseline sample in this area could be 
accessed due to local security conditions. As a result, all baseline observations in Cáceres come from 
the northern and central areas of the municipality, where coca may not be present. These dynamics 
were present in other coca-growing municipalities, as well, and likely imply that our figures are lower￾end estimates of the true prevalence of coca cultivation in sampled municipalities.
There is substantial regional variation in where coca is grown, with Catatumbo and Tumaco and coastal 
Nariño accounting for the vast majority of coca growing households in the sample (Figure 50). In 
Sardinata, 17.0 percent of sampled households admitted to currently growing coca, compared to 7.1 
percent in Tumaco, 7.0 percent in Olaya Herrera, and 6.8 percent in Convención. A small number of 
other municipalities were expected to have coca-growing households but were found to have either no 
or very few households that admitted to growing coca when asked, including Cáceres (0.0 percent), 
Caldono (0.7 percent), La Macarena (0.0 percent), Zaragoza (0.5 percent), and Santander de Quilichao
(0.5 percent). This may be explained by issues related to security and access to coca cultivating areas.
Notes from enumerators during the household survey, along with information from qualitative 
discussions, suggest that the true prevalence may be even higher than reported in Tumaco, Olaya 
Herrera, Convención, and Sardinata, but that some respondents became distrustful when coca was 
mentioned in the survey.
FIGURE 50: ILLICIT CROP PRODUCTION, BY REGION
17.0
7.1
7.0
6.8
0.7
0.5
0.5
0.0 4.0 8.0 12.0 16.0 20.0
Sardinata
Tumaco
Olaya Herrera
Convención
Caldono
Zaragoza
Santander
de Quilichao
Percentage of Households (%)
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
62
The qualitative discussions suggest a number of reasons for why households grow coca: (1) a feeling that 
there is nothing else to make a living from; (2) lack of market access for other crops; (3) the relatively 
reliable prices and short harvest cycles; (4) higher productivity (yield per hectare) of coca making it 
attractive for farmers with small amount of land. Through its focus on developing value chains through 
PPPs, LfP seeks to address issues 1 and 2 directly, making participation in licit economic activities more 
attractive. However, the qualitative results suggest that issues 3 and 4 could continue to present a 
problem for crop substitution efforts. Due to the instability of prices and longer harvest periods for the 
most popular alternative crops, numerous respondents expressed that they needed to grow coca in 
order to subsidize the cultivation of crops like cacao and coffee. Furthermore, even if LfP is able to 
address other issues, the lack of available land is likely to remain a key driver of illicit crop cultivation, as 
households will face strong incentives to maximize their profits on the little land they have, or to 
encroach onto vacant public lands and national parks. As described by one quantitative enumerator in 
their field notes while working in a community in Sardinata, “[In this part of the municipality] everyone is 
cultivating coca. Each farm has its own laboratory or processor, or “cambuche” as they call it, to process the leaf 
and convert it to base [paste], whose current value is around $2,000 pesos per gram of base. Each hectare 
yields between 5,000 and 6,000 grams of base every two months.” This calculates to approximately $10 
million COP ($2,709 USD) per hectare, every two months, a sum that is similar to those reported in
qualitative interviews in Tumaco by respondents who reported earning $10-20 million COP ($2,709-
5,418 USD) every two months before substituting their coca.
Issues with programs for combatting illicit crops
Community members understood that coca brings violence and crime into their communities. Yet coca 
growing was consistently described in the qualitative discussions as the only option for a lucrative, stable 
income source; multiple respondents in different municipalities mentioned that it allowed them to send 
their children to university, when no other crop would have allowed them to. Many even said there is no 
way they could cultivate legal crops without the stable income provided by coca, so coca is what allows 
them to grow coffee or cacao. Qualitative discussion participants have expressed frustration with a lack of 
government presence and transparency, and participants in multiple discussions in coca-growing 
communities claimed that the government’s only concern was to show a military presence and eliminate 
illicit crops, which are their main form of livelihood. It was clear that for some, years of militarization of 
their communities, criminalization of their main source of livelihood, and promises that never come to 
fruition have resulted in a hostile stance towards the government in Bogota. Thus, a common sentiment 
that runs through the discussions in coca-growing communities is that the government is more concerned 
with eliminating coca than promoting their wellbeing. As described by a respondent in Cáceres:
The United States gives Colombia millions of dollars in assistance to combat illicit crops. [But] 
combatting illicit crops means looking for alternatives…It’s no secret to anybody and that’s why a lot of 
congressional representatives in the United States are asking to reduce what they keep sending to 
Colombia [for fighting illicit crops], because it’s being badly invested. And I for one hope they do, 
because it will get rid of corruption. They’re just using it to oppress the people…like fumigation, right? 
Aerial spraying, it’s been demonstrated by the global health organizations and all that, that it affects 
human beings. Here in Colombia [they say] “No, no…we’re going to fumigate, come what may.” Big 
mistake…It affects other crops…They killed a bunch of maize…and even [human] health…Here there 
have been people with rashes. Rashes - they broke out all over and everything…and respiratory 
problems. So that means there’s a lot of things that need to change. (Respondent, Cáceres)
As an alternative to aerial spraying, the Colombian government is currently targeting specific coca￾growing municipalities, many of which overlap with the LfP municipalities, through the PNIS program, 
which was created from the 2016 peace process and provides regular payments and technical assistance 
to coca-growing households in exchange for substituting their illicit crops. However, a related issue 
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
63
mentioned in the qualitative discussions is that participants felt the government was not upholding its 
promises to pay families for not cultivating coca through PNIS, and that the payments were insufficient. 
This aligns with other reporting on PNIS, which indicates the GoC originally signed community-level 
agreements in coca-growing areas, including all farmers in selected communities, and then abandoned 
these agreements in favor of farmer-level agreements. This change imposed new requirements for 
participation and excluded around half of the original 188,000 farmers included in the original 
agreements, some of whom had already pulled up their coca crops by the time the change was 
announced.53 Particularly in Tumaco, participants in the group KIIs with community leaders expressed 
that the PNIS substitution program was insufficient, the government was not consistently making 
payments, and that people were being asked to show a land title or meet other requirements in order 
to receive payments, which had not originally been part of the communities’ understanding for how 
payments would work. As one respondent in Tumaco put it,
The agreement was every two months [a payment would be made] and they took three or four 
months of arguing after each person already had [eliminated] their crops, [after] the communities put 
so much energy into the planting, and now to pay them $2 million [COP] with so much controversy! 
[The families aren’t being paid] because of problems with the SISBEN, problems with the land, 
problems with land tenancy, [they can only be recipients] as long as they are husband and wife, as 
long as they are a nuclear family. That wasn’t the agreement. It was understood that everyone who 
had illicit crops [could access the program with] no problems, [because] this was to meet the needs of 
the community, and the community upheld its end of the bargain. (Respondent, Tumaco) 
CHARACTERISTICS OF COCA-GROWING HOUSEHOLDS
It is important to recognize “coca-growing” in our context refers only to households who openly 
admitted to growing coca in the survey, and results may suffer from issues related to self-reporting, low 
overall levels of prevalence, high 
refusal rates for coca questions, 
and issues with access to coca 
growing communities in some 
municipalities described above.
Additionally, predisposal to 
admitting to coca cultivation may 
be influenced by participation in 
crop substitution programs, 
which in turn could be correlated
with household characteristics in 
ways that generate bias in the 
results. Results should therefore 
be treated with some caution, 
particularly when comparing the 
prevalence between groups. The 
IE team believes the best use of 
these results is for establishing 
lower end estimates of 
prevalence of illicit crop 
cultivation among households in coca growing municipalities with specific characteristics. With these 
53 https://www.crisisgroup.org/latin-america-caribbean/andes/colombia/87-deeply-rooted-coca-eradication-and-violence￾colombia
15.5 15.5
4.8
8.0
Diff.: 0.0
Diff.: 3.2
0
5
10
15
20
25
30
Percentage of Households (%)
Ever Cultivated Coca Household Currently Growing Coca
Note: * p < 0.1, ** p < 0.05, *** p < 0.01
Statistical significance test of the difference accounts for clustered standard errors.
Includes only agricultural households in coca-growing municipalities.
No registered title to any plot Registered title to 1+ plot
FIGURE 51: COCA CULTIVATION, BY LAND TITLE
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
64
limitations in mind, we net restrict the sample to only households engaged in agriculture in coca-growing 
municipalities (Cáceres, Zaragoza, Caldono, Santander de Quilichao, Convención, Sardinata, Tumaco, 
Olaya Herrera, Puerto Lleras, and La Macarena) to look at the characteristics of coca-growing 
households (Figure 51).
Among this group, the lower-end estimate for the percentage of households that have ever cultivated 
coca is similar regardless of whether or not the household has a registered title to any plot. For both 
households with and without registered titles, 15.5 percent admitted to ever having cultivated coca. On 
the other hand, 8.0 percent of agricultural households with at least one registered plot title in coca 
growing municipalities admitted to currently cultivating coca, compared to 4.8 percent of those without 
any registered title, though this difference is not statistically significant. Again considering the issues 
mentioned above, the best interpretation of these results is that, among households in coca-growing 
municipalities, at least 8.0 percent of agricultural households with at least one registered plot title are 
currently cultivating coca, compared to at least 4.8 percent of those without a registered title to any 
plot, though the true prevalence for either group could be higher.
Figure 52 shows the percentage of households that admit to currently cultivating coca in coca growing 
municipalities, disaggregated by household tenure type. Our lower end estimate for prevalence of 
current coca cultivation is 5.9 percent among all households who own at least one plot (including 8.0 
percent of those with at least one title, and 4.8 percent of those without), compared to 4.7 percent of 
those who do not own any plot. Somewhat surprisingly, the reported prevalence of coca growing is 
lowest among households who are occupying land without permission (4.0 percent), although the theory 
of change suggests that we should expect prevalence to be highest among this group, due to the more 
insecure nature of their property rights. However, it is possible that households are disproportionately
inclined to underreport coca cultivation on plots they do not own. An additional possibility is that 
interactions with crop substitution programs may account for the relationship between coca and formal 
titles. For example, it is the evaluation team’s understanding that PNIS may require documentation of 
land ownership in order to participate; this could provide a potential reason for why households with 
titles are more open to admitting cultivation in PNIS municipalities (since GoC is already aware of their 
relationship with illicit crops as part of their PNIS participation), or have perversely led some 
households with titles to begin cultivating coca in order to take advantage of benefits under PNIS.
It should be noted that in qualitative interviews, multiple PNIS officials mentioned their belief that it was 
uncommon to find coca growing on titled plots. On the other hand, community members and leaders 
who participated in qualitative discussions were more mixed on this point, with multiple participants
across different regions stating a belief that titling would not impact coca production. In the end, 
opinions in qualitative discussions for why titling would or would not be effective at reducing coca came 
down to beliefs about whether or not households would perceive a credible risk of expropriation or 
have alternative means of earning a livelihood.
Thus, we highlight again that the results are best interpreted as lower-end estimates, and note that the 
degree of underreporting could be correlated with tenure type. Because of this, the results do not 
necessarily mean that the true prevalence of coca cultivation is higher among households with titles. 
Nor do the results automatically imply that households with titles are growing coca on plots with titles.
This survey did not ask respondents the location of the coca, due to the sensitivity of the issue. The 
coca that households admitted to growing could be located on a titled or untitled plot managed by the 
household, or potentially even on a piece of land in a nearby park or forest reserve not formally 
managed or claimed by the household.
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
65
FIGURE 52: REPORTED COCA CULTIVATION, BY TENURE TYPE
The qualitative discussions further illuminate the type of households that might grow coca, and where 
the coca is grown. In Caldono, Sardinata, and Convención, coca was typically described in qualitative 
discussions as being cultivated on land that belonged to individual households. The land was also 
sometimes described as having at least semi-formal ownership status, for which households were paying 
taxes and had some type of ownership documents, as opposed to vacant lands or national parks. In a 
coca-growing community in Caldono, focus group participants felt that most land in their community 
had already been titled, though due to the complexity of the formalization process it is not always clear 
if their understanding of the term meets the formal legal standard for a title. Further, coca farms were 
described as small to medium-sized, and a focus group participant in Caldono said that “the biggest farms, 
at most would be 10 hectares…and they’re planted with illicit crops, but it’s not like all 10 hectares are planted 
with illicit crops.” In Sardinata, a PNIS official confirmed that there are large farms with coca and described 
their formalization status as “in limbo”, appearing to conflict somewhat with how households and 
communities themselves describe these plots. On the other hand, coca cultivation in Tumaco and La 
Macarena was mostly described as taking place on vacant public lands, national parks, or privately-owned 
land that had been abandoned and then occupied by squatters. 
Figure 53 presents basic characteristics of households who grow coca. As above, these estimates are 
best interpreted as lower-end estimates, with the understanding that households with certain 
characteristics may be more or less predisposed to underreporting. Households were approximately 
equally likely to admit to ever having cultivated coca or currently be cultivating coca regardless of the 
gender of the household head. Those with less than a secondary education were significantly more likely 
to admit to currently cultivating coca (7.1 percent) than those with a secondary education or higher (2.2 
percent). Households were also significantly more likely to admit to currently cultivating coca if the 
household head was born in the municipality (7.0 percent) than if the head was born outside the 
municipality (3.7 percent). Finally, households controlling two or more hectares were significantly more 
likely to admit to cultivating coca (8.8 percent) than those controlling less than two hectares (2.0 
percent). This is a somewhat surprising result as smallholders are expected to be engaged in coca 
cultivation more than large landholders. However, it is possible that small landholders underreport on 
coca cultivation for fear of expropriation—it is possible that they perceive that it is easier to 
expropriate land from small landholders as opposed to large, likely socially powerful landholders. It is 
also likely that small landholders do not grow coca on their own land but venture into protected areas, 
parks, and forest reserves to cultivate coca.
5.6 5.9
4.7 4.7 4.0
8.0
4.8
0
2
4
6
8
10
Percentage of Households (%)
All households in
coca municipalities
Owners of
1+ plot
Do not own
any plots Sharecrop, rent, or
usufruct on any plot
De facto
occupation
on any plot
Titled owners
of 1+ plot
Untitled owners
of 1+ plot
Note: Includes only agricultural households in coca-growing municipalities.
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
66
FIGURE 53: CHARACTERISTICS OF COCA-GROWING HOUSEHOLD
Most officials who were interviewed stated that they felt land formalization would be effective at 
reducing coca cultivation on titled lands and felt that it would be more effective than alternative 
development programs. One PNIS officer stated, 
“Believe me, people aren’t going to risk it, losing their property to expropriation [once their land is 
titled]…Beyond me saying to them ‘Come, I’ll give you a productive project for you to tear up your illicit 
crop’, I think it weighs heavier for me to say, ‘starting today, this land is yours, figure out how to make it
produce and get out of the illicit economy, because if you continue [growing illicit crops] they’ll be 
coming for you’.” (PNIS Official)
An additional risk described by this official is that even if LfP is successful at getting families to reduce
coca cultivation on their own land, illegal armed groups may compensate for this decrease in production 
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
67
on private land by increasing cultivation on protected public lands. It is unclear what households will do 
if faced with a risk of expropriation on one side, and violent threats from armed groups on the other. In 
general, focus group discussions and group KIIs with community leaders did not mention facing threats 
from armed groups as a primary reason for cultivating coca, focusing instead on the economic reasons.
However, one PNIS official described the challenges of pressures from armed groups, and the fact that 
land with illicit crops cannot be titled, as a major challenge for LfP. 
A fundamental requirement to be able to formalize is that there aren’t any illicit crops on the plot. And 
those illicit crops a lot of times are a consequence of that pressure the armed groups exercise in the 
area. So those groups don’t allow families to substitute or eradicate illicit crops and that then is a big 
difficulty that not just USAID faces, but all of us who have a presence in the area. (PNIS official)
FINDINGS 5: WOMEN’S EMPOWERMENT
The household survey was primarily directed at the household head. However, a subset of questions 
were asked separately to both the household head and the primary decisionmaker of the opposite 
gender. For example, if the household head was male, this set of questions was asked to both the male 
household head, and then again to primary adult female decisionmaker in the household. Generally, the 
relationship between the primary adult male and female decisionmakers in the sample is conjugal, though 
in some cases it was parent-child, siblings, or some other relationship.
WOMEN AND LAND TENURE
Figure 54 shows the share of 
household members who appear on 
land documents, separately for men 
and women. Among all household 
members aged 18 or older, 20.8 
percent of men and 16.9 percent 
of women appear on any land 
document; the difference is 
statistically significant. The gap 
between men and women grows 
when looking at only primary male 
and female decisionmakers; 31.3 
percent of primary male 
decisionmakers appear on any land 
document, compared to 22.3 
percent of primary female 
decisionmakers. The percentage of 
household members who appear on a formal, registered title is lower, while the gender gap for registered 
titles is somewhat narrower. Among all household members aged 18 or older, 10.4 and 8.9 percent of 
men and women, respectively, appear on a registered title. This compares to 15.5 and 11.5 percent, 
respectively, for primary male and female decisionmakers. Thus, the data shows a modest, statistically 
significant advantage for men in terms of who appears on land documents, which narrows but remains 
statistically significant when focusing only on registered titles.
Figure 55 shows the perceptions of primary male and female decisionmakers with respect to who in the 
household has the right to sell or bequeath land. Here, we narrow the scope of the analysis to include 
only households with both an adult male and female decisionmaker, where both completed the 
questions in the gender empowerment module. Male decisionmakers were significantly more likely to 
20.8
16.9
10.4 8.9
31.3
22.3
15.5
11.5
Diff.: -3.9***
Diff.: -1.5**
Diff.: -8.9***
Diff.: -3.9***
0
10
20
30
40
Percentage of Household Members (%)
Appears on
any document
(all HH members 18+)
Appears on
registered title
(all HH members 18+)
Appears on any document
(primary male & female decisionmakers)
Appears on registered title
(primary male & female decisionmakers)
Note: * p < 0.1, ** p < 0.05, *** p < 0.01
Statistical significance test of the difference accounts for clustered standard errors.
Men Women
FIGURE 54: HOUSEHOLD MEMBERS ON LAND 
DOCUMENTS, BY GENDER
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
68
describe themselves as having the right to sell any plot, while female decisionmakers were significantly
more likely to say the decisionmaker of the opposite gender had the right to sell. These results suggest 
that, in general, primary adult male and female decisionmakers have a shared understanding of who has a 
right to sell land, and that this consistently favors the male decisionmaker.
While the difference between male and female decisionmakers retains a statistically significant advantage 
for men in terms of who describes themselves as personally having the right to bequeath land, the gap 
closes substantially compared to the gender gap for right to selling land. It is also interesting to note that 
while respondents appear to have a similar understanding of who has the right to sell land, the same is 
not true of their understanding of who has the right to bequeath; 40.7 percent of male decisionmakers 
said the female decisionmaker in the household had a right to bequeath land, while just 28.3 percent of 
female decisionmakers said the male decisionmaker had the same right. It is also notable that the share 
of female decisionmakers who say they personally have the right to bequeath (51.9 percent) is much
larger than the share who say they have the right to sell (27.0 percent).
FIGURE 55: PERCEPTIONS OF RIGHT TO SELL AND BEQUEATH LAND
One explanation for these differences could be that men are more likely to have children with multiple 
women, and that this may impact how women think about the question of inheritance rights. This was 
mentioned in qualitative discussions as being problematic and impacting women in particular. For 
example, in one group discussion in Zaragoza, participants described how land titles and wills were 
uncommon, while the frequency of informal marriages and, in particular, men with multiple families, 
create problems for inheritance. As one participant in Zaragoza said, “Well, it’s normal if they have 
children, and even more if there are children outside the marriage, or whatever type of union they have, well it’s 
normal for the children to leave the women landless. They leave them with nothing.” This was also described 
as being relatively common by a community leader in Convenión, who suggested that this may even 
happen with the male partner’s consent. As this community leader described it:
That’s one of the problems that’s occurred, when the man is the only one on the title. There’s been cases 
where the man leaves the inheritance to a son, and the wife is left out. And he can do it, because he [sold 
the land to the son] while he was still living. The only way to stop it is for the wife to realize what’s 
happening and she goes and says, ‘sorry, I’m the wife and I don’t allow this transaction.’ But if he did it 
behind her back… (Community Leader, Convención).
46.6
27.0 25.6
37.0
54.1 49.2
56.6 51.9
40.7
28.3
63.3 61.5
Diff.: -19.6***
Diff.: 11.4***
Diff.: -4.9*** Diff.: -4.7***
Diff.: -12.4***
Diff.: -1.8
0
20
40
60
80
100
Percentage of Household Members (%)
Personally has right to sell any plot
Decisionmaker of
opposite gender has
right to sell
Anyone in household has right to
sell any plot
Personally has right
to bequeath
any plot
Decisionmaker of
opposite gender has
right to bequeath
Anyone in household has right to
bequeath any plot
Note: * p < 0.1, ** p < 0.05, *** p < 0.01
Statistical significance test of the difference accounts for clustered standard errors.
Includes only primary male and female decisionmakers in dual-adult households where both the male and female completed the gender empowerment module.
Male Decisionmakers Female Decisionmakers
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
69
An additional set of three questions 
probed for respondents’ 
perceptions on their tenure 
security. The primary male and 
female decisionmakers in each 
household, regardless of 
household tenure type, were 
asked about the perceived 
likelihood that the household
would involuntarily lose ownership 
or use rights to any plot within the 
next five years, how worried the 
respondent was that the 
household would involuntarily lose 
use rights to any plot within the 
next five years, and how long the 
household could leave its land 
fallow without worrying about 
losing use rights. Narrowing the 
scope of the analysis to only dual￾adult households where both the primary male and female decisionmakers responded, the results 
presented in Figure 56 paint a somewhat mixed picture of tenure security perceptions by gender. 
Female decisionmakers were significantly more likely to say the household was not likely to lose its 
rights to any plot within the next five years (79.1 percent of females, 71.9 percent of males), and to say 
they were not worried about the household losing its rights to any plot over the same time period (78.8 
percent of females, 73.0 percent of males). On the other hand, male decisionmakers were significantly 
more likely to say the household could leave its land fallow indefinitely and would never lose its land use 
rights (28.0 percent of males, 14.5 percent of females).
On the other hand, responses from 
male decisionmakers suggest a clear 
advantage over female decisionmakers 
in terms of their perceived knowledge 
of land rights and access to resources 
in the event of a land conflict, as 
shown in Figure 57. Respondents 
were asked the extent to which they 
agreed with three statements. A 
statistically significant greater share of 
male decisionmakers (43.1 percent) 
agreed with the statement “I know 
more about my land rights now than I 
did last year”, compared to female 
decisionmakers (34.4 percent). 
Similarly, a significantly greater share 
of male decisionmakers (54.9 percent) 
agreed with the statement “I know 
where to go if I have a conflict about 
my land” than female decisionmakers (42.4 percent), while 50.2 percent of male decisionmakers agreed 
with the statement “I have access to legal representation if I have a land dispute”, compared to 35.9 
percent of female decisionmakers.
71.9 79.1 73.0 78.8
28.0
14.5
Diff.: 7.2*** Diff.: 5.8***
Diff.: -13.6***
0
20
40
60
80
100
Percentage of Respondents (%)
Not likely to
lose rights to any plots
Not worried about losing
rights to any plots
Would never lose
land left fallow
Note: * p < 0.1, ** p < 0.05, *** p < 0.01
Statistical significance test of the difference accounts for clustered standard errors.
Includes only primary male and female decisionmakers in dual-adult households where both the male and female completed
the gender empowerment module.
Male Decisionmakers Female Decisionmakers
43.1 34.4
54.9
42.4 50.2
35.9
Diff.: -8.7***
Diff.: -12.5***
Diff.: -14.3***
0
20
40
60
80
100
Percentage of Respondents
Who Agree (%)
I know more about my
land rights now than I did last year
I know where to
go if I have a conflict about my land
I have access to
legal representation if
I have a land dispute
Note: * p < 0.1, ** p < 0.05, *** p < 0.01
Statistical significance test of the difference accounts for clustered standard errors.
Includes only primary male and female decisionmakers in dual-adult households where both the male and female
completed the gender empowerment module.
Male Decisionmaker Female Decisionmaker
FIGURE 56: PERCEPTIONS OF TENURE 
SECURITY
FIGURE 57: ACCESS TO LEGAL RESOURCES 
FOR LAND DISPUTES
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
70
Male and female decisionmakers were asked whether they believed there was any advantage to having a 
property title or land document. Responses are shown in Table 18. Nearly all (98.1 and 98.0 percent of 
male and female decisionmakers, respectively) said there was an advantage to having such a document. 
In general, male and female decisionmakers appear similar in how they view these advantages, with the 
main differences being a somewhat larger, statistically significant share of male decisionmakers (15.2 
percent, compared to 9.3 percent of female decisionmakers) saying the most important advantage is 
access to loans, and a somewhat larger, statistically significant share of female decisionmakers (32.1 
percent, compared to 27.9 percent of male decisionmakers) who say the most important advantage is 
tenure security.
TABLE 18: ADVANTAGES OF LAND DOCUMENTATION
What do you think is the most 
important advantage of having 
documentation? 
Overall Female 
Decisionmaker
Male 
Decisionmaker
Freq. Pct. (%) Freq. Pct. (%) Freq. Pct. (%) Diff.
No advantage 27 1.9 14 2.0 13 1.9 0.2
Increased Property Value 49 3.6 27 4.0 22 3.2 0.8
Access to loans 166 12.3 62 9.3 104 15.2 -5.9***
Tenure security 406 30.0 215 32.1 191 27.9 4.2**
Proof of ownership 549 40.6 273 40.8 276 40.4 0.5
Other 183 13.5 92 13.8 91 13.3 0.4
Total 1353 100.0 669 100.0 684 100.0 n/a
*** p < 0.01, ** p < 0.05, * p < 0.1. 
Note: Includes only respondents in dual-adult households where both the male and female decisionmaker responded to the gender 
empowerment module. 
PARTICIPATION IN PRODUCTIVE ACTIVITIES AND ECONOMIC DECISIONS
Figure 58 shows the percentage of 
plots where the primary household 
member responsible for the plot is 
the primary male or primary female 
decisionmaker, or someone else.
Among all plots belonging to 
sampled households, 69.2 percent 
were under the primary 
responsibility of the primary male 
decisionmaker in both treatment and 
comparison communities; female 
primary decisionmakers were the 
main person responsible for 29.0 and 
28.4 percent of all plots in treatment 
and comparison communities, 
respectively. This reflects a theme 
from the qualitative discussions, 
where the most common means 
through which women acquire land 
is inheritance, and in general participants felt it was rare, though perhaps not unheard of, that women 
purchase land. At the same time, even when women do acquire land through inheritance, qualitative 
discussions suggested at times they may not feel comfortable managing the land on their own, which
some participants said may lead them to sell the land. Community leaders in Convención described this 
in the following terms, which may also reflect existing gender biases around women’s land ownership: 
69.2 69.2
28.4 29.0
2.3 1.9
Diff.: -0.1
Diff.: 0.6
Diff.: -0.5
0
20
40
60
80
100
Percentage of Plots (%)
Male Primary
Decisionmaker Female Primary
Decisionmaker Other
Note: * p < 0.1, ** p < 0.05, *** p < 0.01
Statistical significance test of the difference accounts for clustered standard errors.
Comparison Treatment
FIGURE 58: PLOT RESPONSIBILITY
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
71
Community Leader 1: “If [women] are left land [through inheritance], it's easy [to inherit land]. If they 
leave it to you. But that they leave you land is complicated…Because there’s machismo…No, and it also 
happens…sometimes they have the possibility to become, well, women who can manage a farm. But 
unfortunately, when the moment comes to divide up the inheritance, and it’s like they don’t feel capable or 
what they do is sell…So like always the man is running things because of that.”
Community Leader 2: “For example…my grandfather had land. He died and my grandmother, too. They 
divided up the inheritance between five men and three women. But of the three women, two were already 
mothers and were living away, and the other well, that was my mom….[The women] practically gave away 
the inheritance.” (Group KII, Convención)
In general, these differences reflect a context where female decisionmakers are far less engaged in 
agricultural and other economic activities than their male counterparts. The survey asked male and 
female decisionmakers about whether they participated in each of six income generating activities, 
including food crop farming, cash crop farming, livestock farming, non-farm economic activities, wage 
employment, and fishing or fishpond culture. Narrowing the scope of the analysis to include only dual￾adult households where both the male and female responded, Figure 59 shows 40.3 percent of all female 
decisionmakers indicated they participated in none of the six activities, compared to 10.4 percent of 
male decisionmakers. On average, female decisionmakers participated in 1.0 activities, compared to 1.9 
activities for male decisionmakers.
FIGURE 59: PARTICIPATION IN ECONOMIC ACTIVITIES
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
72
The data allows us to examine which of the six activities present the largest gender gaps in terms of 
participation, shown in Figure 60. While participation is higher for men and the gender gap is statistically 
significant for all six activities, three activities stand out as having particularly large gender gaps: wage 
employment, food crop farming (although this is still the most common activity that women participate 
in), and cash crop farming. The narrowest gender gaps were found for fishing, where participation was 
low overall, and for non-farm economic activities. The qualitative data suggests that many of these non￾farm economic activities are done within the home, such as artisan goods, hair styling, and preparation 
of food products, and therefore may be more accessible to some women who may be more confined in 
their activities outside the home. Similarly, qualitative discussions suggest another common activity 
conducted by women throughout the different regions in the survey is raising minor livestock, 
particularly poultry.54
FIGURE 60: TYPES OF ECONOMIC ACTIVITIES
The data suggests a slight increase in women’s economic participation correlated with formal title status, 
and a further increase correlated with the female decisionmaker appearing on the title. However, 
increased men’s economic participation is also correlated with these land title characteristics, meaning 
that the gender gap between male and female decisionmakers in terms of economic participation does 
not close. This is shown in Figure 61, which suggests that land titles alone may not be sufficient to close 
the gender gap in economic participation. Among households with no titled plots, male decisionmakers 
participated in 1.8 activities, on average, compared to 0.9 activities for female decisionmakers. For 
households with at least one titled plot, male decisionmakers participated in 2.0 activities, on average, 
compared to 1.1 activities for female decisionmakers, representing a slight increase for both genders 
over households with no titled plots. For households where the female decisionmaker appears on a title 
for at least one plot, male decisionmakers participated in 2.1 activities, on average, compared to 1.2 
activities for female decisionmakers, representing a slight increase for both genders over all households 
with at least one titled plot.
54 The results suggest poultry raising was not captured in the economic activity questions. Although “livestock farming” is one of the economic 
activity categories, the question from the A-WEAI instrument does not explicitly specify the inclusion of chickens. Thus, “livestock” may simply 
suggest cattle to most respondents, particularly given the Spanish translation.
69.7
45.2 45.0
21.9 19.7
8.7
17.4 12.4
31.4
5.4 5.1 2.7
Diff.: -24.5***
Diff.: -23.1***
Diff.: -11.0***Diff.: -5.0***
Diff.: -26.0***
Diff.: -2.4***
0
20
40
60
80
100
Percentage of Respondents (%)
Food Crop
Farming
Cash Crop
Farming
Livestock
Farming
Non-Farm
Economic
Activities
Wage
Employment
Fishing/
Fishpond
Culture
Note: * p < 0.1, ** p < 0.05, *** p < 0.01
Statistical significance test of the difference accounts for clustered standard errors.
Includes only primary male and female decisionmakers in dual-adult households where both the male and female
completed the gender empowerment module.
Male Decisionmakers Female Decisionmakers
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
73
FIGURE 61: NUMBER OF ECONOMIC ACTIVITIES, BY TENURE STATUS
For respondents who participated in each activity, they were asked who in the household usually makes 
decisions regarding the activity, and the amount of input they themselves had in making decisions about 
the activity. The results in Table 19 shows that although the vast majority of respondents who 
participated in each activity felt they had some input into decisions on the activity, male decisionmakers 
consistently indicated having more input into decisions than female decisionmakers, with differences that 
were sometimes statistically significant. The largest differences are found for cash crop farming, fishing, 
and food crop farming. For example, 98.4 percent of male decisionmakers who participated in cash crop 
farming felt they had at least some input into cash crop farming decisions, compared to 85.1 percent of 
female decisionmakers who participated in cash crop farming. These results indicate that in addition to 
facing barriers in participating in economic activities, women who do participate in these activities face 
additional barriers in decisionmaking.
TABLE 19: INPUT IN DECISIONMAKING
Respondent has some input 
in decisions on activity
Overall Female 
Decisionmakers
Male 
Decisionmakers
Freq. Pct. (%) Freq. Pct. (%) Freq. Pct. (%) Diff.
Food crop farming 775 96.0 293 92.1 482 98.6 -6.4***
Cash crop farming 440 94.0 131 85.1 309 98.4 -13.3***
Livestock raising 185 94.4 55 91.7 130 95.6 -3.9
Non-farm activity 200 95.7 81 93.1 119 97.5 -4.4
Wage & salary employment 249 97.3 37 100.0 212 96.8 3.2**
Fishing 52 94.5 16 84.2 36 100.0 -15.8*
Note: Percentages are calculated over the number of respondents who participated in each activity only. Includes only respondents in dual￾adult households where both the primary male and female decisionmaker responded to the gender empowerment module.
*** p < 0.01, ** p < 0.05, * p < 0.1. 
We gain further insight on this point from an additional set of questions asking respondents who 
participated in each activity about the extent to which they could make their own decisions on the 
activity if they wanted to. Again, the results in Table 20 show consistent advantages for male 
decisionmakers, with differences that are often statistically significant. The largest gender gaps are found 
for cash crop farming (31.6 percentage-point gap, favoring males), food crop farming (18.0 percentage￾point gap, favoring males), and livestock raising (17.4 percentage-point gap, favoring males).
1.8
0.9
2.0
1.1
2.1
1.2
Diff.: -0.9*** Diff.: -0.9*** Diff.: -1.0***
0
1
2
3
4
5
Average Number of Activities Participated In
Households with
no titled plots All households with 1+ titled plot Households where female decisionmaker appears on title
Note: * p < 0.1, ** p < 0.05, *** p < 0.01
Statistical significance test of the difference accounts for clustered standard errors.
Includes only primary male and female decisionmakers in dual-adult households where both the male and female
completed the gender empowerment module.
Male Decisionmakers Female Decisionmakers
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
74
TABLE 20: ABILITY TO MAKE OWN DECISION
Respondent feels they can make 
their own decisions if they wanted 
to on activity
Overall Female 
Decisionmakers
Male 
Decisionmakers
Freq. Pct. (%) Freq. Pct. (%) Freq. Pct. (%) Diff.
Food crop farming 686 85.5 235 74.6 451 92.6 -18.0***
Cash crop farming 389 84.2 95 62.9 294 94.5 -31.6***
Livestock raising 159 82.8 41 70.7 118 88.1 -17.4***
Non-farm activity 184 88.5 70 81.4 114 93.4 -12.0**
Wage & salary employment 223 88.1 34 91.9 189 87.5 4.4
Fishing 48 87.3 15 78.9 33 91.7 -12.7
Note: Percentages are calculated over the number of respondents who participated in each activity only. Includes only respondents in dual￾adult households where both the primary male and female decisionmaker responded to the gender empowerment module.
*** p < 0.01, ** p < 0.05, * p < 0.1. 
The evaluation team next constructed an indicator of empowerment in productive decisions, using the 
template provided by the Abbreviated Womens Empowerment in Agriculture Index (A-WEAI)55. 
Specifically, “empowered” status for productive decisions is defined as having sole or joint 
decisionmaking power over at least one agricultural activity (i.e., food crop farming, cash crop farming, 
livestock, or fisheries) or being able to make their own decisions on at least one of these activities if 
they wanted to. This allows us to classify respondents into one of three categories: participating and 
having input in at least one agricultural activity, participating in at least one agricultural activity but having 
no decisionmaking power, and not participating in any agricultural activities. The results in Figure 62
show that for both treatment and comparison communities, a substantially greater share of male 
decisionmakers have empowerment status in agricultural productive decisions, while treatment and 
comparison communities are similar in terms of the level of empowerment for each gender. For both 
male and female decisionmakers, the results suggest that participation is a much greater barrier to 
empowerment than decisionmaking; in other words, focusing on increasing participation may do more 
for empowerment than changes in the decisionmaking processes amongst those who already participate 
in these activities.
FIGURE 62: EMPOWERMENT IN PRODUCTIVE DECISIONS
55 For more information, see A-WEAI documentation: https://weai.ifpri.info/versions/a-weai/
81.8
0.7
17.6
72.4
0.3
27.3
0
20
40
60
80
100
Percentage of Respondents (%)
Comparison Treatment
Male Decisionmakers
47.9
1.9
50.2
47.5
3.0
49.5
Comparison Treatment
Female Decisionmakers
Has input in 1+ Ag Activity Participated in 1+ Ag. Activity,
but had no decision-making input
Did not participate
in any Ag. Activity
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
75
These findings are echoed in the qualitative discussions. Although some discussion participants felt 
women faced no particular barriers to participating in agriculture, women were more commonly 
described as having decision-making power over small gardens around the home’s patio, raising chickens 
and small livestock, and other activities that could be done within or near the home, such as a home￾based business. Participants also suggested that decision-making power over a plot usually lays with
whoever the plot belongs to, which may more commonly be a man, but could be a woman, or a man 
and woman jointly. Such respondents generally felt that the extent to which a woman would participate 
would vary from family to family, depending mostly on the interest and experience of the woman. For 
example, one female GD participant in Cáceres said, “For our crop, I made the decision on my own and he 
helped me when we were getting started, but not anymore, because he has his own crop over there on another 
side [of the plot].” However, even in cases where the woman appeared on a land title, some qualitative 
discussions described men as typically being responsible for decisions on land use, farming, and raising 
large livestock. As a community leader from Convención said, “Let’s put it like this. The title can be in the 
woman’s name, but the man is the one who decides and he does what he wants.”
CONTROL OVER INCOME
Male and female decisionmakers were asked an additional set of questions about decisions on income from 
different economic activities, as well as decisions on major and minor household expenditures. This allowed 
the team to calculate an additional component of empowerment under the A-WEAI, which examines control 
over income. This component defines “empowered” status as having control over income from at least one 
economic activity, or feeling they can make their own decisions on non-farm activities, wage and salary 
employment, or major household expenditures. The results are shown in Table 21.
TABLE 21: CONTROL OVER INCOME FROM DIFFERENT SOURCES
Overall Female 
Decisionmakers
Male 
Decisionmakers
Income Decisionmaking Area Freq. Pct. (%) Freq. Pct. (%) Freq. Pct. (%) Diff.
Has some input in decisions about income 
from food crop farming
655 96.8 238 93.7 417 98.6 -4.9***
Has some input in decisions about income 
from cash crop farming
431 95.4 127 87.0 304 99.3 -12.4***
Has some input in decisions about income 
from livestock raising
169 96.6 51 96.2 118 96.7 -0.5
Has some input in decisions about income 
from non-farm activity
198 95.2 80 93.0 118 96.7 -3.7
Has some input in decisions about income 
from wage & salary employment
244 98.0 37 100.0 207 97.6 2.4**
Has some input in decisions about income 
from fishing
40 95.2 12 85.7 28 100.0 -14.3
Can make own decisions on non-farm 
activities
184 88.5 70 81.4 114 93.4 -12.0**
Can make own decisions on wage & salary 
employment
223 88.1 34 91.9 189 87.5 4.4
Can make own decisions on major 
household expenditures
1122 81.6 501 73.7 621 89.4 -15.7***
Note: Questions only asked to respondents who participated in these activities, with the exception of major and minor household 
expenditures which were asked to all respondents. Includes only respondents in dual-adult households where both the primary male and 
female decisionmaker responded to the gender empowerment module. 
*** p < 0.01, ** p < 0.05, * p < 0.1. 
Among male and female decisionmakers who participated in the six economic activities, decisionmaking 
on income consitently favored men, with gender gaps that were often statistically significant. The largest 
gender gaps were seen for fishing (14.3 percentage-points), cash crop farming (12.4 percentage-points), 
and food crop farming (4.9 percentage-points). Similarly, a significantly larger share of male 
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
76
decisionmakers compared to female decisionmakers described themselves as able to make their own 
decisions on non-farm activities (93.4 and 81.4 percent, respectively) and major household expenditures 
(89.4 and 73.7 percent, respectively). These results suggest that major household expenditures provide 
the greatest contribution to women’s empowerment in control over income, particularly given the low 
levels of participation for women in most other economic activities.
Constructing the control over use of income empowerment indicator, the results show that in both 
treatment and comparison communities, male decisionmakers show a substantially higher level of
empowerment on this dimension. The results, shown in Figure 63, suggest that men are near universally 
empowered on control over income use, and while most women are empowered on this dimension, 
there is still have room for improvement. Male decisionmakers show similar levels of empowerment 
across treatment and comparison communities, while female decisionmakers in treatment communities
show a slighly higher level of empowerment than those in comparison communities; these differences 
will be accounted for in the impact analysis through statistical matching and they do not pose an issue 
for the validity of the impact estimates (see the power and balance section of the report).
FIGURE 63: EMPOWERMENT IN CONTROL OVER INCOME
ASSET OWNERSHIP
Male and female decisionmakers were asked about whether anyone in their household owned a series of 
items, and whether they themselves owned any of the item either solely or jointly. The results in Table 
22 show there are significant gender differences across multiple items. For example, a significantly 
greater share of male decisionmakers personally owned agricultural land, large livestock, fishing 
equipment, mechanized and non-mechanized farm equipment, housing, and transportation, while a 
significantly greater share of female decisionmakers personally owned small livestock, poultry, non-farm 
business equipment, and large and small consumer durables. No differences were seen across genders 
for cell phone and non-agricultural land ownership.
98.7 97.7
Diff.: -1.0
0
20
40
60
80
Percentage of Respondents (%)
Male Decisionmakers
79.5
85.5
Diff.: 6.0*
Female Decisionmakers
Comparison Treatment
Note: * p < 0.1, ** p < 0.05, *** p < 0.01
Statistical significance test of the difference accounts for clustered standard errors.
Includes only primary male and female decisionmakers in dual-adult households where both the male and female decisionmaker
responded to the gender empowerment module.
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
77
TABLE 22: ASSET OWNERSHIP 
Overall Female 
Decisionmakers
Male 
Decisionmakers
Personally owns any of item Freq. Pct. (%) Freq. Pct. (%) Freq. Pct. (%) Diff
Agricultural Land 694 49.4 308 43.9 386 54.9 -11.0***
Large Livestock 225 16.0 92 13.1 133 18.9 -5.88**
Small Livestock 256 18.2 135 19.2 121 17.2 2.0**
Poultry 804 57.2 461 65.6 343 48.8 16.8***
Fish pond/equipment 51 3.6 21 3.0 30 4.3 -1.3**
Non-mechanized Farm Equipment 562 40.0 211 30.0 351 50.0 -20.0***
Mechanized Farm Equipment 62 4.4 22 3.1 40 5.7 -2.6***
Non-Farm Business Equipment 86 6.1 53 7.5 33 4.7 2.9***
House 690 49.1 336 47.8 354 50.4 -2.6*
Large Consumer Durables 1028 73.1 525 74.7 503 71.6 3.1**
Small Consumer Durables (e.g. radio, 
cookware)
1278 90.9 659 93.7 619 88.1 5.7***
Cell Phone 1255 89.3 635 90.3 620 88.2 2.1
Non-Agricultural Land 64 4.6 31 4.4 33 4.7 -0.3
Transportation 582 41.4 227 32.3 355 50.6 -18.3***
Total 1406 100.0 703 100.0 703 100.0 n/a
Note: This question was a multiple response question, so counts for individual items may not add up to the total number of respondents.
Includes only primary male and female decisionmakers in dual-adult households where both the male and female decisionmaker responded to 
the gender empowerment module.
*** p < 0.01, ** p < 0.05, * p < 0.1. 
These questions allowed 
the evaluation team to 
follow the A-WEAI template 
to construct an indicator of 
empowerment in asset 
ownership. This indicator 
classifies each of the items as 
either large or small, and 
defines empowerment in 
asset ownership as owning 
at least one large or two 
small assets either alone or 
jointly. The results in Figure 
64 show that empowerment 
in asset ownership is near 
universal for both male and 
female decisionmakers.
While empowerment in 
asset ownership is higher for 
female decisionmakers, the 
difference is small and statistically insignificant. The results also show that both male and female 
decisionmakers are similar in terms of asset ownership across both treatment and comparison 
communities (Figure 64). These results are unexpected, and suggest the definition of empowerment 
based on these particular assets may be too coarse or ill-suited for this implementation context.
97.7 99.0
Diff.: 1.3
0
20
40
60
80
100
Percentage of Respondents (%)
Male Decisionmakers
99.0 99.0
Diff.: -0.0
Female Decisionmakers
Comparison Treatment
Note: * p < 0.1, ** p < 0.05, *** p < 0.01
Statistical significance test of the difference accounts for clustered standard errors.
Includes only primary male and female decisionmakers in dual-adult households where both the male and female completed the gender
empowerment module.
FIGURE 64: EMPOWERMENT IN ASSET OWNERSHIP
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
78
ACCESS TO CREDIT
Gender-based differences in access to credit is examined by looking at all loans described by the primary 
survey respondent and identifying which household members the loan is attributed to. This is shown in 
the first panel of Table 23. Overall, 6.7 percent of all male household members aged 18 or older and 6.1 
percent of female household members over 18 took out any loan over the 12-month period before the 
survey; the difference is not statistically significant. The gender gap increases slightly and becomes 
statistically significant when the analysis is restricted to primary male and female decisionmakers; 9.6 
percent of male decisionmakers and 7.7 percent of female decisionmakers took a loan over the period. 
Although the gender gap in the share of household members who took loans is small, the gap is much 
larger when looking at average loan amounts, as shown in the second panel of Table 23. Among all 
household members aged 18 or older who took loans, males took an average of $2,676.5 USD ($9.9 
million COP), compared to an average of $1,517.4 USD ($5.6 million COP) for female adults who took 
loans. The gender gap grows modestly when restricting the analysis to primary male and female 
decisionmakers. On average, male decisionmakers who took loans obtained $2,753.1 USD ($10.2 million 
COP) in loans over the 12-month period before the survey, compared to $1,546.9USD ($5.7 million 
COP) for female decisionmakers who took loans.
TABLE 23: ACCESS TO CREDIT, BY GENDER
Overall Women Men
Outcome or Covariate N Mean N Mean N Mean Diff.
Percent who took loans (%)
Household Members 18+ 7402 6.4 3485 6.1 3909 6.7 -0.6
Primary Decisionmakers 4924 8.7 2443 7.7 2481 9.6 -2.0 ***
Average amount borrowed (USD)
Household Members 18+ Who Took 
Loans 475 2154.5 213 1517.4 261 2676.5 -1159.1**
Primary Decisionmakers Who Took 
Loans 426 2223.6 187 1546.9 239 2753.1 -1206.2**
Loan Source (% of Household 
Members Who Accessed Loans)a
Agrarian Bank 477 27.9 213 24.9 261 30.3 -5.4
Other Banks 477 34.4 213 37.1 261 32.2 4.9
Cooperative 477 12.6 213 16.0 261 9.6 6.4 **
Gota a gota (informal lender) 477 10.5 213 11.7 261 9.6 2.2
Family / Friends / Employer 477 9.6 213 6.6 261 12.3 -5.7 *
Other 477 8.6 213 8.0 261 9.2 -1.2
Loan Use (% of Household Members 
Who Accessed Loans)a
Agriculture/Livestock 475 40.0 211 31.3 261 46.7 -15.5 ***
Purchase Land 475 2.5 211 1.9 261 3.1 -1.2
Buy/Improve Housing 475 12.8 211 14.7 261 11.1 3.6
Business Investment 475 13.9 211 19.0 261 10.0 9.0 ***
Household Expenses 475 14.7 211 15.2 261 14.6 0.6
Other 475 21.1 211 23.2 261 19.5 3.7
*** p < 0.01, ** p < 0.05, * p < 0.1.
Note: Statistical significance tests account for clustered standard errors.
a This question was a multiple response question, so percentages may not add up to 100. Calculated for all respondents who took loans.
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
79
Overall, male and female household members who accessed loans appear similar in terms of loan 
sources. This is shown in the third panel of Table 23. The most important differences are a somewhat 
higher share of loans to women coming from cooperatives (16.0 percent, 9.6 percent of loans to men), 
and a higher share of loans to men coming from family, friends, or employers (12.3 percent, 6.6 percent 
of loans to women). While a greater share of loans to men came from the agrarian bank, this is balanced 
out by a greater share of loans to women coming from other banks, meaning that a similar share of 
loans to women (62.0 percent) and men (62.5 percent) came from any formal bank.
Loans to men and women were also mostly similar in terms of the loan use, shown in the final panel of 
Table 23. The main differences were that a greater share of loans to men (46.7 percent) were used for 
agricultural or livestock purposes than loans to women (31.3 percent), and a greater share of loans to 
women were used for business investments (19.0 percent) than loans to men (10.0 percent).
BALANCE AND POWER
The evaluation team used the baseline data to revisit statistical assumptions related to the methodology 
proposed for this evaluation. As has been discussed throughout the report, results presented are for a 
group of randomly sampled households from communities in municipalities where the LfP intervention is 
taking place, along with a group of randomly sampled households from comparison communities that 
were statistically matched to the intervention communities. In terms of our methodology, the report 
describes results after the first two steps of the matching process—the municipal-level and the 
community-level matching. In general, the results in this report have suggested that the treatment and 
comparison sample are indeed similar, with the exception of a few important statistically significant 
differences. To further improve the comparability of the treatment and comparison groups, we 
conducted the third step of the matching at the household level using household characteristics from the 
baseline data employing entropy balancing. After implementing this household-level matching, the 
evaluation team conducted balance tests to examine and confirm the comparability of sampled 
households in treatment and comparison communities for the matched comparison component of the 
evaluation design. This is important for strengthening the validity and precision of the impact estimates 
at endline. In addition, the evaluation team re-ran power calculations from the evaluation design report,
using the parameters obtained from the baseline data. This updated power analysis using actual baseline 
data from the LfP implementation area enabled the team to re-confirm that the evaluation has sufficient 
statistical power to detect the impact of LfP on key outcomes of interest, and across the range of effect 
sizes that the team anticipated during the evaluation design stage.
BALANCE TESTS
The evaluation team examined normalized differences in baseline means on key outcome variables and 
covariates to assess balance between sampled households in treatment and comparison communities. 
The selected variables spanned a wide range of outcome categories and covariates of interest for the 
evaluation. The normalized difference statistic falls below 0.25 for nearly all of the variables tested, 
indicating adequate balance.56 The evaluation team also used entropy-balancing as a form of matching to 
improve comparability of treatment and comparison groups using the baseline data and confirm the 
ability to mitigate observable bias through matching. The aim of pre-processing the data via entropy￾weighting is to improve covariate balance between the treatment and comparison groups, so that the 
comparison group has a more similar distribution to the sampled households in treatment communities
on observed characteristics that may influence outcomes. This helps to overcome the confounding 
effects of potential selection bias, particularly in terms of communities selected for the intervention and 
56 An absolute value of 0.25 or less indicates strong balance.
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
80
household self-selection into participation into program activities. The process uses methods developed 
by Hainmuller (2012) and was conducted to achieve balance on treatment and comparison group means.
To test this at baseline, the evaluation team matched on a set of household-level characteristics that 
relate to key demographic traits or outcome indicators. The household-level covariates included a wide 
range of variables, as listed in Annex H. The results confirmed the ability to achieve balance on several 
key covariates and baseline values of outcomes across the samples. In other words, households in the 
comparison group form a viable pool to serve as a counterfactual for sampled households in treatment 
communities, to measure the effects of the LfP intervention as planned through the evaluation.
The table in Annex H presents means for the overall sample, the treatment sample, and the comparison 
sample (before entropy balancing), along with means for the matched comparison sample (after entropy 
balancing). The difference columns show the difference in means between the comparison means and 
the mean for the treatment sample, along with the statistical significance of the difference. The table 
shows a good degree of balance before entropy balancing, and an excellent degree of balance after 
entropy balancing. Across more than 100 key household traits and variables related to different 
outcome indicators, exceptionally few show statistically significant differences between the treatment 
sample and the matched comparison sample. Even among variables where the difference between the 
treatment sample and unmatched comparison is not statistically significant, nearly all show improvement 
in balance after entropy balancing.
However, the one important area where balance is not achieved is on coca cultivation. Unfortunately, 
this was inevitable due to the security problems experienced in comparison coca municipalities, in 
particular in Olaya Herrera where enumerators were forbidden by community leaders from asking 
questions about coca in all communities in the municipality. The consequence is that a statistically 
significant greater share of sampled households in treatment communities say they have ever benefitted 
from alternative development programs and have ever grown coca, and a larger, statistically insignificant 
portion of these households admit to currently growing coca. Because of the high rate of refusals to the 
coca questions in Olaya Herrera, the true prevalence of coca cultivation among the comparison sample 
(both matched and unmatched) is almost certainly higher than what is reported and may be more similar 
to the treatment sample than what is reported here. The main implication of this for the IE is that the 
endline statistical analysis may either over or underestimate the impact of LfP on cocoa cultivation, 
depending on the trajectories of coca cultivation in the treatment and comparison municipalities 
between baseline and endline. The evaluation team should supplement the analysis at endline with 
secondary data and targeted qualitative information, where possible, to help mitigate this limitation.
POWER CALCULATIONS
The evaluation team also used the baseline sample to update the power calculations for the IE. These 
results are shown in Table 24. Compared to power expectations at the IE design stage, the updated 
calculations confirmed that the evaluation remains well-powered to detect many key outcomes of 
interest. Table 24 presents illustrative minimum detectable impact (MDI) calculations for selected
outcomes. MDIs indicate the smallest impact for a given outcome that we are able to detect given the 
impact evaluation design, sample size, and a number of other parameters such as the confidence level of 
the hypothesis test (95 percent), the level of power (80 percent)57, and the amount of variation in the 
outcome explained by the covariates included in the regression analysis (30 percent). We first calculated
the minimum detectable effect size (MDES) for the LfP intervention, which is expressed in terms of 
standard deviation units (same across all outcomes). We then calculated the MDI for each outcome by 
57 “Confidence level of the hypothesis test” refers to how confident we can be that a difference is the result of a true difference between two 
groups, and not the result of random variation in the data (i.e., how confident we can be that we do not have a false positive). “Level of power” 
refers to how confident we can be that we will detect a difference between two groups, if such a difference exists (i.e., how confident we can 
be that we do not have a false negative).
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
81
multiplying the MDES with the baseline standard deviation of the given outcome. The smaller the MDES, 
the larger the power of the design.
The reduction in the evaluation survey sample size from the planned 3,000 households across 200 
communities to the achieved 2,965 households across 192 communities resulted in a small increase in 
the MDES over what was anticipated at evaluation design, from 0.213 to 0.218. The evaluation therefore
remains well-powered to detect medium-scale changes to outcomes as a result of LfP’s land 
formalization and related interventions. This MDES value implies that if the change to a given outcome 
as a result of LfP is smaller than 0.218 standard deviations from the mean at baseline, the IE statistical 
analysis will not be able to detect that change. In other words, the statistical analysis treats very small 
changes that may have resulted from LfP the same as if there was no impact at all from LfP. This is not a 
concern for most outcomes, given the magnitude of USAID’s investment in LfP, since we make the 
assumption that no impacts will be viewed similarly as very small impacts from a policy point of view.
In terms of individual outcomes of interest, the evaluation is powered to detect changes ranging from
10-156 percent of the baseline mean for key outcomes related to LfP’s theory of change, depending on 
the outcome. This is mostly in line with the MDIs estimated at evaluation design, which ranged from 19-
80 percent of the anticipated baseline means for most outcomes assessed at that stage. For example, for 
perceived land tenure security (percent of households who reported that they are not likely to lose 
rights to any plot in the next 5 years), the IE is powered to detect a 9.15 percentage-point or greater 
change from the baseline proportion of tenure secure households, which was 77 percent. Similarly, the 
IE is powered to detect an 8.9 percentage-point or greater increase in the proportion of households 
with a registered title to at least one plot, over the baseline sample average of 21.3 percent. The 
updated MDIs for many of the outcomes assessed in Table 24 are in line with results from prior land 
sector evaluations, which suggests they are in the range of impacts that can feasibly be achieved through 
a program such as LfP. For example, a 7.9 percentage point increase in households who receive credit 
appears achievable, particularly given the heavy emphasis of qualitative interview participants on the 
importance of formal titles for receiving credit and the existing gaps between need for credit and credit 
access detailed earlier in this report.
However, the estimated MDI is now substantially higher for revenue from agricultural activities, at 91
percent of the baseline value (instead of 41 percent as estimated in the EDR), mostly due to the 
extremely unequal distribution of revenue in the baseline sample. In practical terms, this means that LfP 
will need to achieve large improvements to household income from agriculture across a substantial 
portion of implementation households (or a nearly doubling, on average) for the evaluation to detect a 
statistically significant impact on this outcome.
The MDI is also substantially higher for prevalence of coca cultivation, at 154 percent of the baseline 
value, owing to the very low prevalence of coca reported via survey respondents at baseline.58 The 
updated MDI suggests the evaluation is powered to detect a three percentage-point difference in the 
proportion of households that cultivate coca. However, because only 2.0 percent of households 
admitted to cultivating coca at baseline, in practice it will likely be nearly impossible for the evaluation to 
detect an impact on coca that is attributable to LfP, through the statistical impact analysis.59 As a result,
the IE is effectively no longer powered to detect the impacts of LfP on coca cultivation prevalence from 
the quantitative household survey data. See the recommendations section below for discussion on 
potential mitigation options for this through supplemental secondary data sources and analysis.
58 Prevalence is only slightly higher, at 3.9 percent, when taking only coca-growing municipalities into consideration, but the corresponding 
reduction in sample size completely wipes out the improvements in MDI that stem from the higher prevalence in this sample.
59 The scenarios under which this might be possible seem very unlikely practice, such as the LfP intervention leading to no change or an even 
further reduction in the already low proportion of households that cultivate coca, while communities in comparison areas experience a large 
increase in coca production in the absence of an intervention like LfP.
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
82
TABLE 24: MINIMUM DETECTABLE IMPACTS FOR SELECTED OUTCOMES
Analysis 
unit Outcome
Baseline valuesa
Source
Sample 
size
Full Sample
mean SD MDESb
MDIc (in units 
of outcome)
MDI (% change 
from mean)d
Household Perceived tenure security (% who are not likely to 
lose rights to any plot, next 5 years) 77.0 42.1 Baseline 2965 0.218 9.15 11.9%
Household Has registered title to at least one plot (%) 21.3 40.9 Baseline 2965 0.218 8.93 42%
Household Access to credit - Received credit last year (%) 15.6 36.3 Baseline 2965 0.218 7.91 51%
Household Probability of Poverty at National Poverty Line (%) 45.0 21.5 Baseline 2965 0.218 4.69 10%
Household Monthly Income from Agricultural Activities (USD) 372.3 1526.1 Baseline 2174 0.223 340.57 91%
Household Household Cultivates Coca (%)e 2.0 13.9 Baseline 2965 0.218 3.01 154%
Notes: 
a. Baseline means and standard deviations are calculated from the evaluation baseline data.
b. MDES=Minimum detectable effect size, expressed in standard deviation units; 
Calculations assumed a confidence level of 95 percent, two-tailed tests, 80 percent power, 10 percent non-response rate, 15 percent 
correlations between outcome across baseline and follow-up surveys, and that covariates explain 30 percent of the variation in 
outcome.
c. MDI=Minimum detectable impact, expressed in units of outcome. 
d. Percent change is relative to baseline mean. 
e. Restricting the power calculation for this variable to only the communities in coca-growing municipalities yields an MDES of 0.315, 
and an MDI of 6.1 and 156%, in units of outcome and percent change from mean, respectively. These statistics are calculated using a
sample size (n=1507), mean (3.9), and SD (19.4) for that sub-sample. 
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
83
CONCLUSIONS
IMPLICATIONS OF BASELINE FINDINGS FOR LFP THEORY OF CHANGE AND 
PROGRAMMING
It is important to recognize certain limitations of the baseline study, including the sensitive nature of 
some questions regarding illicit crop cultivation and threats from armed groups, which can affect the 
reliability of responses or respondent willingness to respond. An additional limitation could stem from 
the apparently low level of understanding of the legal and institutional landscape around land titling, 
which may mean households’ self-reported land tenure status at baseline is not always accurate.
However, several of the baseline findings support the LfP programming logic and elements of the 
intended theory of change, while some findings call attention to issues that may present challenges for 
LfP’s intended objectives, and/or require adaptive programming to address as implementation 
progresses. 
Key areas of support include:
Scope to greatly improve the proportion of households with formalized land rights, and LFP’s programming 
emphasis on strengthening the culture of formalization over time among the target population also appears to be 
highly relevant.. Results support the programming logic that most households are currently untitled (for 
example, among sampled households in LfP communities, 22.0 percent of households reporting having a 
registered title to any plot). In qualitative discussions, respondents also expressed widespread approval 
for previous titling efforts by now-defunct land agencies INCORA and INCODER, which speaks to the 
extent to which communities are likely to welcome the efforts of LfP. However, these discussions also 
suggested that these plots can slip back into informal tenure status when land is inherited from a titled 
owner who dies, because the inheritor does not always go through the process of formalizing the 
inheritance. This highlights the importance of accompanying land formalization with efforts to create a 
culture of formalization so that LfP’s efforts are not undone over time.
Linkages between titling and credit, and potential improvements to agricultural productivity. Results support the 
theory of change that providing land titles may improve access to credit for those unable to obtain 
loans. Among all sampled households in treatment communities, 25.8 said they had need of credit in the 
past 12 months, while just 17.2 were able to obtain it. Improved access to credit was consistently 
mentioned in qualitative discussions as the most important benefit of having a formal title, and this 
sentiment is echoed in the quantitative data. A tangential benefit of improving access to formal credit is 
that it may also reduce households’ reliance on informal gota a gota lenders connected to organized 
crime, which account for 14.6 percent of all loans. In addition, the baseline found that around 12 percent 
of land on plots with agricultural activities is currently left unused, while a lack of resources to develop 
the land was mentioned as the primary reason for this for 57.7 percent of these plots. If titles do 
increase landholder’s ability to obtain loans, or the size of loans they obtain, then this in turn could lead 
to improved agricultural productivity for those with sufficient land.
Linkages between formalized titling, strengthening women’s land rights and women’s empowerment. Results 
support the programming logic that women are currently less likely to appear on any land document or 
on a formal title. The findings also support the possibility that including women on land titles may 
improve their tenure security, particularly in the context of land inheritance. However, low levels of 
formal marriage could present a challenge to including both partners on land titles. Although Colombian 
law recognizes the property rights of individuals in common law marriages, couples need to make a 
formal declaration that they are living together. LfP is working to assist couples with this step as it works 
to advance joint titling. LfP’s gender-sensitive titling approach also aims to teach women about their 
property rights and how to navigate the legal system around land transactions and disputes. Results also 
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
84
support the programming logic with respect to a gap in agricultural empowerment between men and 
women. However, the results only weakly support the theory of change that including women on land 
titles will improve their participation in agriculture. Comparing agricultural participation and control 
over decision-making at baseline between women who are included on titles and those who are not, the 
advantages for titled women are clear but relatively small. The failure of the instrument to adequately
capture activities related to small livestock and poultry, which many women engage in, could contribute 
to this and can be rectified at endline. Still, the nature of women’s activities suggests that the benefits to 
women from inclusion on land titles could come more strongly on other outcomes, such as in improved 
access to credit for starting or growing a small business, which in turn may also serve as an important 
driver of their empowerment.
The baseline findings provide less support for the following TOC elements, while highlighting some key 
issues that underscore certain challenges for LfP’s intended programming:
Results suggest some challenges to the theory of change that land titles will lead to widespread improvements in 
tenure security, given findings from baseline related to sources of tenure insecurity, experience with prior 
formalization efforts, and eligibility for titling. In general, measures of tenure security were relatively high at 
baseline, land disputes were uncommon, and there was little difference in tenure security between titled 
and untitled households who owned land. The biggest differences were seen between those who owned 
land and those who did not. Although such households are not the focus of LfP’s intervention, LfP may 
want to consider how these less secure households may be impacted by the intervention. In general, 
tenure security was also not commonly listed as a benefit of obtaining a title, and nearly half of owners 
without titles said they had not formalized because they saw no reason to, or because they preferred 
staying informal, perhaps to avoid taxes. This may suggest a role for LfP in communicating potential 
benefits of formalization to eligible households. The results do provide some evidence that having a 
formal title reduces forced displacement, but does not completely prevent it. Some communities appear 
to have developed their own informal systems for land governance in the absence of State presence, and 
findings also suggest that holding a land title may not address some common sources of tenure 
insecurity, such as land expropriation by armed groups. Debt was another commonly cited cause of 
tenure insecurity in qualitative discussions, and tenure insecurity from debt could actually increase if the 
program leads to a rise in credit seeking among newly titled households.
Colombia’s highly unequal land distribution may pose a challenge for the program to leverage land titles into 
improved livelihoods, as baseline findings suggest that many households have insufficient land to earn a living 
from even if these holdings are formalized. The benefits of formalization could then also accrue 
disproportionately to a smaller share of households that have medium to large land holdings.. Among all 
households with no registered title to any plot, 37.0 percent do not own any land and only have land 
access through usufruct or rental agreements, while an additional 21.5 percent own 0.5 hectares or less. 
Thus, many households may be dependent on improvements in wage employment or starting or 
growing their own businesses to improve their livelihoods, as they may not have enough land to 
facilitate substantial improvements to income from agricultural activities, even with land titles and 
improved access to credit. 
Low levels of satisfaction with local service delivery and infrastructure present both a challenge and an 
opportunity for LFP programming. The program’s focus on local service delivery and infrastructure as part of its 
activities appears well-placed in this respect, but the current status quo does present a substantial challenge 
given that market access will be needed to improve livelihoods and trust in the government infrastructure will be 
needed to support formal transactions. Over half of all respondents were not familiar enough with the 
Municipal Land Office to have any opinion, while of those who were familiar with the institution, only 
around 40 percent said they were satisfied with its services. While this suggests some challenges to LfP’s 
theory of change for Component 2, it may also provide an opportunity for LfP to substantially move the 
needle on public perception if it can successfully strengthen land administration capacity at the municipal 
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
85
level.
Land restitution is contentious. Qualitative discussions highlighted a desire for increased transparency, 
assurances that all parties involved could be present for plot studies, and for the process to include the 
input of neighbors. Given that the cadaster updates will serve as an input to the land restitution process, 
potential challenges of accelerated land restitution for LfP include the possibility of increased social 
tension.
Trust in government is low, particularly in coca-growing areas. Coca substitution is at a fragile point. 
Communities perceive the government to be against them, appear increasingly skeptical towards PNIS 
efforts, and suspect the government will resume aerial spraying. Changes to PNIS agreements and 
requirements for participation appear to be a main cause of skepticism. Some of these new program 
requirements may impede on LfP’s ability to achieve impacts in coca-growing municipalities. For 
example, it is the evaluation team’s understanding that households with salaried employment are 
ineligible for PNIS benefits, which could limit the impacts LfP might expect to achieve on increasing 
private sector employment opportunities and income from off-farm activities. Endline data collection 
should look at the size and duration of PNIS subsidies, and households’ rates of success at signing up for 
PNIS.
To be effective at reducing coca cultivation, LfP will likely need to be accompanied by a credible threat of 
expropriation and viable alternative livelihoods. Results suggest that not all households currently growing 
coca are untitled. And, while LfP may be effective at reducing coca cultivation on land that is newly 
titled, it is less clear whether it will be able to reduce cultivation on lands that remain untitled, such as 
plots inside of national parks. Indeed, it is reasonable to expect that reduced coca supply from titled 
plots could drive up prices, increasing the incentive to push into the agricultural frontier to grow coca 
on vacant public land, protected reserves, and national parks. This could be particularly true if the 
government resumes aerial fumigation, given that protected lands cannot be sprayed.
LfP’s success at reducing coca cultivation will depend on additional factors outside its control. As PNIS is already 
operating substitution programs, LfP’s success at reducing coca cultivation may depend on the timing of 
alternative development projects. LfP’s Component 3 for developing value chains and PPPs assumes 
PNIS and the GoC will be able to provide sufficient interim transitional support to these households 
until Component 3 can be implemented, though in practice this interim support is outside of LfP’s 
manageable interest. Additionally, LfP’s interventions are not designed to address the demand side of the 
illicit crop problem, and it is unclear whether the program can be effective at reducing coca cultivation 
without a reduction in demand. As long as the returns on coca cultivation remain high, even road 
construction, value chain development, and the creation of PPPs may not be sufficient to encourage a 
sufficient proportion of households to move away from coca to be able to detect an impact.
RECOMMENDATIONS FOR SUBSEQUENT DATA COLLECTION FOR THE IE
The baseline data collection experience and findings also highlight some key issues that will be important 
to address for the endline data collection round:
• Supplemental strategies to strengthen the evaluation’s reporting on changes to coca production, 
including an expanded focus on this in the qualitative data collection, and potential use of 
supplemental geospatial data to inform on coca expansion. The baseline results strongly suggest 
that households underreport their participation in coca production, and the resulting low 
frequency of participation in the baseline data means that the evaluation will very likely have 
insufficient statistical power to detect changes on this via the statistical impact analysis. To gain 
additional insights at endline on LfP’s influence on this, the evaluation team recommends: (1) 
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
86
prioritizing targeted qualitative data collection on this issue as part of the endline qualitative data
collection effort; (2) considering possibilities to use supplemental geospatial data on coca 
production from satellite imagery or related sources, such as data from the UNODC; and (3) 
incorporating questions at endline asking about coca production in 2021 to try to recover the 
data that could not be obtained from specific comparison geographies. Satellite data from 
UNODC likely provides the best opportunity for continuing with a quantitative evaluation of 
LfP’s impacts on coca cultivation, as it would provide more units for analysis and improve 
statistical power. Targeted GDs and KIIs at endline may provide a route to understand patterns 
and reasons for change on this as it relates to LfP’s activities, especially if it is paired with 
satellite data that can help identify if LfP may be associated with spillover effects on coca 
production, such as expansion on to protected lands. The evaluation team notes that coca 
production was sensitive for discussion in the qualitative effort at baseline, as anticipated. At 
endline, the evaluation team might also consider adding one-on-one discussions with community 
leaders if they are willing to do so.
• Expanded role for qualitative data collection at endline, to strengthen the ability to interpret endline 
results. This should particularly focus on expanding the number of group discussions, holding 
gender-segregated discussions, and expanding coverage on issues related to women’s land rights, 
tenure security and empowerment. It should also aim to help inform on potential differences in 
trends across the diverse regional contexts where LfP operates. Baseline findings highlight the 
strong role that qualitative data is likely to play at endline to help interpret results, including for 
gender disaggregated results and on particular issues related to women’s land rights, tenure 
security and empowerment. In addition, given the diverse regional contexts where LfP is 
operating, this data collection should disaggregated by region and aim to be substantive enough 
within each region to allow for trends to be distinguished across regions. This implies a larger 
qualitative data collection effort than was undertaken at baseline, and also conducting gender￾segregated group discussions, both of which have time and budget implications for the endline 
component of the evaluation. USAID is encouraged to prioritize this at endline, in addition to 
the substantive quantitative data collection effort, to strengthen the evaluation’s ability to 
identify reasons for observed impacts at endline and drivers of potential variation across 
different regions of LfP implementation.
• Suggested modifications to endline survey instrument and qualitative data collection:
o Given confusion from respondents at baseline on what constitutes a land title, 
enumerators should include a picture of an escritura publica at endline and add a question 
to the household survey that asks respondents to confirm if they have this document. 
Respondents could also be asked to show enumerators the document that they do 
have, to ensure it is coded correctly on the survey.
o Modifications to the endline survey instrument and enumerator training to improve 
reporting on the A-WEAI section of the wives module on whether women engage in 
poultry raising as a component of livestock activities. This could include adding a note at 
the start of the module noting that poultry farming should be considered a type of 
livestock activity for the purpose of the survey, and/or adding follow-up probing 
questions in that module on poultry activities specifically. The baseline results suggest 
that women’s engagement in poultry raising may be under-reported at baseline because 
of confusion over whether this was considered to be part of livestock activities or a 
separate category.
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
87
o Targeted work to shorten the endline survey instrument where possible. Key areas 
where some efficiencies can be gained are on level of detail collected about loans, and 
some details on the crop roster.
o A more targeted focus for the qualitative discussions on women’s empowerment issues, 
including on issues related to land inheritance, farm management and tenure security, to 
enable greater interpretation of household survey data on the same. Similarly, qualitative 
discussions and household surveys at endline should add questions related to areas LfP 
programming is working to address, including formal declarations of common law 
marriage and gender-based support for land rights.
o Ensure questions are included in the qualitative and quantitative instruments to look 
beyond access to credit, and examine the quality of credit obtained. Households may 
obtain loans from informal sources, illicit groups, or predatory lenders charging 
extortionate interest rates. Questions should go beyond simply asking about whether or 
not credit was obtained.
• Ensuring at least 6 months lead time for endline data collection and evaluation team procurement, to 
ensure sufficient time to plan and engage in the data collection preparations with the endline 
firm. This is especially important given the challenging security context in the LfP implementation 
and comparison communities for this evaluation.The baseline data collection effort highlighted 
the challenging security context in the LfP implementation areas, but also demonstrated that 
high quality data collection is achievable, particularly with careful planning and outreach on the 
part of the data firm. As was the case for baseline, the endline evaluation team must have 
sufficient time to plan and engage in data collection preparation with the endline data firm to 
ensure security logistics, plan for contingencies and ultimately collect high quality data. The 
implication for USAID is to procure the endline work with sufficient lead time before the data 
collection in the field is intended to take place, so that the endline evaluation team and data firm 
have enough time to work through the substantial logistics and security challenges for this data 
collection (in addition to the typical work required for the endline data collection itself). In our 
experience, this means USAID should aim to have both the evaluation team and data firm on 
board at least 6 months before the intended start of endline data collection in the field. A time 
period greater than 6 months may be required if coordination with other USAID data collection 
efforts is envisioned at endline, as was the case for the baseline data collection.
• If only one additional round of data collection will be possible for the evaluation, consider timing that 
endline data collection to take place at least 8-10 months after distribution of land documentation to 
households. This would enable measurement of short-term outcomes and impacts of the LfP 
activity. If feasible, a second round of data collection some 2-4 years after all components of the 
activity have been completed would enable measurement of longer-term impacts.
• Data firm procurement via the Mission’s MEL activity. The independently-procured data firm for the 
data collection at baseline, through the Mission’s MEL activity, worked well for this evaluation 
and can be replicated in subsequent data collection rounds.
• Maintain the same community definitions used at baseline. GoC may change maps over time, but 
subsequent data collection activities and rounds of analysis should continue to use the 
community definitions from baseline data collection. These community definitions considered 
the diverging official cartographies from DANE and IGAC, how LfP would use these 
cartographies to conduct its activities, and the realities of how fieldwork would need to be 
conducted.
Evaluation of the Land for Prosperity (LfP) Activity in Colombia: Baseline Report
88
REFERENCES
Centro Nacional de Memoría Histórica. 2014. “¡Basta Ya! Colombia: Memorias de Guerra y Dignidad.” 
Bogota: Universidad Nacional de Colombia. https://doi.org/10.15446/achsc.v41n2.48792.
Dávalos, Liliana M, Adriana C Bejarano, Mark a Hall, H Leonardo Correa, Angelique Corthals, and Oscar 
J Espejo. 2011. “Forests and Drugs: Coca-Driven Deforestation in Global Biodiversity Hotspots, 
Supporting Information.” Environmental Science Technology 45: 1219–27. 
https://doi.org/10.1021/es102373d.
Departamento Nacional de Planeación. 2019. “Documento CONPES 3958.” Consejo Nacional de Política 
Económica y Social. Bogota. https://doi.org/10.1109/ISSPA.1999.815782.
Diamond, A. and Sekhon, J. 2013. “Genetic Matching for Estimating Causal Effects: A General 
Multivariate Matching Method for Achieving Balance in Observational Studies”. Review of Economics 
and Statistics 95(3): 932-945.
Fernandez, M. 2012. “Violencia y Derechos de Propiedad: El Caso de La Violencia en Colombia”. 
Ensayos Sobre Politica Economica 30 (69): 113-147.
Hainmuller, J. 2012. “Entropy Balancing for Causal Effects: A Multivariate Reweighting Method to 
Produce Balanced Samples in Observational Studies”. Political Analysis 20: 25-46.
Ibáñez, A. M., and J. C. Muñoz-Mora. 2010. “The Persistence of Land Concentration in Colombia: What 
Happened Between 2000 and 2009?” In Distributive Justice in Transitions, edited by Morten Bergsmo, 
César Rodríguez Garavito, Pablo Kalmanovitz, and María Paula Saffon. Torkel Opsahl Academic 
EPublisher y Peace Research Institute Oslo.
LeGrand, Catherine. 1988. Colonización y Protesta Campesina En Colombia (1850 - 1950). Universidad 
Nacional de Colombia.
LeGrand, Catherine Carlisle., and Hernando. Valencia Goelkel. 2016. Colonización y Protesta Campesina 
En Colombia (1850-1950). Universidad de los Andes.
Machado, Absalon, and Julian A. Vivas. 2009. Ensayos Para La Historia de La Política de Tierras En Colombia: 
De La Colonia a La Creación Del Frente Nacional.
Muñoz-Mora, Juan Carlos, Santiago Tobón, and Jesse Willem d’Anjou. 2018. “The Role of Land Property 
Rights in the War on Illicit Crops: Evidence from Colombia.” World Development 103 (April): 268–
83. https://doi.org/10.1016/j.worlddev.2017.10.021. 
UNHCR (2015). Colombia operations 2015 year-end report.