Division of the Commons and Access to Land on The Frontier :

Lessons from The Colonial Legacy in The Democratic Republic

of Congo.
Margaux Vinez∗

Paris School of Economics

February 15, 2017

Abstract

What is the importance of colonial policies in shaping today’s land tenure institu-
tions and inequalities in access to land ? This paper sheds light on this question by
analyzing ”paysannat”, a colonial intervention in the Belgian Congo attempting to push
the evolution of the tenure system from communal toward private property rights. In
the context of forced cultivation of cash crops, the Colony imposed the privatization of
collectively owned land (forests or fallows) to individual farmers in some villages. Using
spatial discontinuities of the implementation of paysannat and a unique combination of
contemporary household survey data, geographic data, as well as historic data from both
colonial records and contemporary oral history surveys, this paper shows that paysannat
had a persistent impact on local land institutions through its impact on the privatization
of collective land. We find that paysannat was successful in pushing toward the indivual-
ization of the commons, and that it had important distribution consequences between the
clanic groups.

Key words: colonization, agricultural policies, economic history, land rights, commons,
Democratic Republic of Congo
JEL classification: N37, O12, P14, P16

∗
Corresponding Author, Paris School of Economics, 48 boulevard Jourdan 75014 Paris, France. email:
[email protected]

1
 ”One of the objectives of African history, once it freed itself from an over
dependence on sources in colonial archives, was to look at African history from an
African viewpoint with sources rooted in Africa.”
Merrick Psnansky, Inauguraul adress, University of Ghana 1969

1 Introduction

There is now clear evidence that History matters for economic development, and a growing
interest in understanding why and through which channels historical events can have such a
long term influence (Nunn, 2009; Banerjee and Duflo, 2014). In particular, a growing body of
work investigate empirically the impact of colonialism on former colonies’ development. The
objective of this literature is to unveil historical dependencies and understand the importance
of institutions set up during colonization in explaining today’s development outcomes. Tack-
ling this issue raises several difficulties: it is often difficult to uncover the full development
trajectory since colonization, to identify causal impacts, and understanding the mechanisms
that link colonial interventions to today’s outcomes can be challenging.

In seminal papers, Acemoglu et al. (2002, 2000); La Porta et al. (1997); Engerman and Sokoloff
(1994) provided the first empirical evidence that the colonial rule can have a lasting impact
on colonized countries. The main channel they supported was the persistence of initial insti-
tutions: changes in institutions that were introduced by the colonizers persisted even after
those countries became independent. While La Porta et al. (1997); Engerman and Sokoloff
(1994) focused on the importance of the identity of the colonizer, while Acemoglu et al. (2000,
2002) focused on the characteristics of the colonized countries. He was the first to provide a
strong identification strategy: he instruments current domestic institutions by settlers mor-
tality, and indentifies the impact of institutions on income, relying on the assumption that
settlers mortality affects todays income only through its impact on institutions. These first
papers paved the way for other studies at the macro and country levels.

More recently, a few empirical studies have focused on limited geographic areas and a par-
ticular colonial rule to investigate the influence of colonization on todays outcomes (Banerjee
and Iyer, 2005; Iyer, 2010; Dell, 2008; Huillery, 2009, 2011). The advantage of this seemingly
narrow focus have been underlined by Cogneau and Dupraz (2015): when comparing more
homogenous areas it is easier to understand precisely what differed between them and iden-
tify the exact channels through which colonial rules shaped long term development. Those
studies provide detailed information about the trajectory of the region of interest and strong
empirical evidence. For example, Banerjee and Iyer (2005) examine differences in revenue
collection accross districts in colonial India: in some district, revenues were collected direcly
from the farmers while in others the collection was performed by local landlords. They find

2
that landlords districts have lower levels of health, education, and agricultural technology
than their non-landlord counterpart and use the date of conquest as an instrument for the
type of revenue collection that was experienced to argue that the impact is causal. Dell (2008)
uses a regression discountinuity strategy to identify the long term impacts of the forced labour
mining system ”mita” in Peru. She finds that mita districts have lower levels of consumption,
and argues that the transmission channel is that wealth was more concentrated in mita dis-
tricts which led to lower levels of provision of public goods (education and roads). Huillery
(2009) shows that colonial public investments are an important determinent of current re-
gional inequalities in West-Africa, because of the persitence of early colonial investments. To
identify the effect of colonial public investments on current outcomes, she takes advantage of
spatial discontinuities of colonial investments.

The motivation of this paper is to examine the importance of colonial history in explaining
today’s land institutions and inequalities in access to land between households of different
clanic groups, and shed some light on the consequences of the privatization of customary land.
Access to land is critical in Africa where a large share of the population depends on agriculture
for its subsistence, and is regulated by a complicated combination of customary institutions
and legal arrangments. The evolution of customary land institutions and the effects of the
privatization of the commons have been the subject of a vast litterature (Boserup, 1983;
Baland and Platteau, 1996, 1998; Ostrom, 2014; Goldstein et al., 2015; Goldstein and Udry,
2008; Weitzman, 1974). However, few studies focus on the possible distributional effects
of such privatization (Baland et al., 2009). Following (Banerjee and Iyer, 2005; Dell, 2008;
Huillery, 2009, 2011), we focus on a specific colonial rule in a specific region. The policy
of interest was implemented by the Belgian Colony during the second quarter of the 20th
century, and led to the privatization of parts of the common land. We examine the effects of
this policy in the Equateur Province in the North West of the Democratic Republic of Congo.
Paysannat aimed to target most villages in the region, but was only implemented in a subset
of them. These geographical variations in colonial interventions help us understand differences
in access to land and the agrarian structure today, and we explore how these differences also
lead to different development outcomes. To study this question, this paper uses a first-hand
household-level data set and a detailed village survey of oral history and current and past
land institutions that we collected between 2013 and 2015. We complemented and confirmed
the oral history data by collecting colonial history record data from the Ministry of Foreign
Affairs in Brussels. The advantage of this approach is that it allows us to identify precisely
the effect of this rule and examine the channels in detail.

In 1885 King Leopold was granted control of most of the Congo Basin and created the Congo
Free State. Its large territory was then still largely unexplored, so his priorities were to take
control of the vast territory and to generate financial resources to fund this expensive under-
taking (Piet, 2013). The land legislation was designed to serve these two objectives: the first

3
decree published in 1885 declared most of the Congolese land to be ”vacant” and hence to
belong to the King, and a heavy taxation in kind was imposed to collect ivory, wild rubber,
and copal. The ill-treatment of indigenous people soon triggered international protests, and
together with the state’s financial difficulties led Belgium to annex the country as a colony in
1908. The Parliament attempted to reform the land legislation but its grounding principles
remained unchanged: throughout the period, the primary objective of the land legislation
remained to allow the appropriation of the ”vacant” land and the exploitation of its resources
for the benefit of the colonial administration, missions, and European companies. By granting
huge land concessions and helping with the recruitment of local labor, the state supported the
development of mining industries as well as large scale agricultural production by commercial
companies. The agricultural policy focused on the forced cultivation of cash crops (such as
cotton, cocoa, rice or palm oil) by smallholder farmers in their villages. Commercial compa-
nies were granted large monopolies over entire regions, and farmers were forced to cultivate
cash crops and to sell their production to the monopolist at state-fixed prices under the ”com-
pulsory cultivation scheme”. This coercive system led to an increase in exportation but failed
to improve living standards and the declining soil fertility soon became a pressing issue in ru-
ral areas. In the 1930’s, a new agricultural policy called ”Paysannat” was proposed and tested
to increase indigenous production and improve living standards. The idea was to modernize
production and support the emergence of a class of small entrepreneurs by distributing com-
munal land to individual farmers and promoting the ”rationalization” of cultivation practices.
Crops, rotations, and the duration of fallows were defined by a state agronomist to maximize
production. Cotton companies, who were at the origin of its inception, pushed for the devel-
opment of the scheme, in which they saw a way to increase production and motivate farmers.
In 1949, the colony decided to expand the scheme to 500.000 farmers and their families in
several regions as part of the Ten Year Plan. However, colonial administrators faced severe
operational and logistical difficulties during prospection and implementation: transportation
was difficult, customary land rights were poorly understood, village chiefs were uncoopera-
tive, and the administrators lacked the human and financial resources to expand the scheme
at this pace. The land surveys, which were theoretically mandatory to conduct, were often
conducted in a hurry. After 5 years, the colony realized that many issues had to be solved
and decided to slow down the expansion of the scheme. By 1960 they had involved less than
half of the number of villages they had initially planed to reach. Some villages therefore con-
tinued to cultivate cash crops under the traditional ”compulsory cultivation scheme” (that
was now called the ”education scheme”), while others participated in Paysannat where they
also cultivated cash crops.

My strategy is to use the spatial distribution of Paysannat to study its influence on subse-
quent development trajectories. We compare villages who grew cash crops along the roads
and were imposed land distributions related to paysannat to village who grew cash crops

4
to study how it influenced the evolution and distribution of property rights in the villages.
The main empirical challenge is therefore that of selection: did colonial administrators chose
to implement Paysannat in some villages and not in others because they had systematically
different characteristics? In particular, since we are are intersted in the effects of privatiza-
tion, were paysannat villages chosen for being already more privatized, closer to commercial
areas, and experienced more pressure on the land ? There is no evidence that this was the
case. In fact, the paysannat villages should have been chosen for opposite reasons: the in-
structions given to territorial administrators were to chose villages where collective land was
still available in sufficient quantity to be divided among all the households in the village, and
preference was to set it up in villages were the common forest was large enough so that one
block could be identified to carry out the division. If anything, that meant chosing villages
were less privatization had already occured and where clanic claims on the land were not too
strong. In practice, since territorial administrators were constrained to set up paysannat in
some specific administrative Secteurs, they were often unable to find villages that matched
these ideal criteria, but nevertheless chose villages were land pressure was reasonably low.
Using geographic and survey data, we also show the location of paysannat villages is not cor-
related with geographic characteristics such as accessibility or distance to commercial urban
centers during colonization. Appart from selection, another important question is whether
the fact that a village had been involved in Paysannat led to different development policies
after Independence and different patterns of investments. We will show that there is not
evidence that this was the case.

First, we run a simple OLS comparing village level outcomes in villages which grew cash
crops (in this region, mainly cotton and rice) and experienced a land distribution during col-
onization to villages who, because of historical reasons and the slow down in the expansion
of the policy didn’t take part in land distribution. Those regressions include a fixed effect
at the administrative Secteur level to account for differences in geographic and colonial ad-
ministrator’s and indigenous chiefs characteristics, as well as geographic controls. We show
that paysannat villages have less ”commons”, or collective forest, so in that sense paysannat
succeeded in pushing toward more privatization. At the same time, it also seems to be more
difficult for migrants from the village to come back and get access to land. Older generations
of strangers in the village seem less integrated in the local institutions. Second, we examine
the distributional consequences of this privatization of the commons by the Belgian Colony.
We establish that there are strong inequalities in access to land between the different clanic
groups within a village, and we give evidence that part of these differences can be explained
by the presence of Paysannat. To test the robustness of our results and to control for pos-
sible endogeneity in the choice of the villages for the agricultural policy, we then follow a
nearest neighbor’s approach: we construct pairs of villages which are geographical neighbors
across administrative borders of a territory, but which happened to have different agricultural

5
policies (with or without a policy to push toward the individualization of land rights).

The remainder of the paper proceeds as follows. Section 2 presents background on the history
of the land tenure system and agricultural policies in Congo since the beginning of coloniza-
tion. Section 3 describes our experimental strategy and data. Sections 5.1 presents the results
and explores some explanations and section 6 concludes.

2 Historical Background

The process by which the Congo became a Belgian colony was long. King Leopold was eager
to participate in the colonial scramble for Africa, and spent decades sponsoring expeditions
in the central African region through his International African Association. He convinced
the European community that he was involved in philanthropic and humanitarian work, and
promised them that he would not tax trade. The colonization of the country began just after
the Berlin Conference in 1885, when he was granted control of most of the Congo Basin and
created the Congo Free State. Its large territory was then still largely unexplored. His pri-
ority was to take control of the vast territory, and to generate financial resources. The land
legislation was designed to serve these two objectives. Soon, most of the Congolese land was
declared to belong to the king, and a heavy taxation in kind was imposed (Boelaert, 1956).
When the Belgian Parliament took over the country in 1908, they reformed the land legisla-
tion but its grounding principles remained unchanged (Piet, 2013; WorldBank/UN-Habitat,
2016; Mugangu Mataboro, 2016). Throughout the period, the primary objective of the land
legislation remained to allow the appropriation of the ”vacant” land and the exploitation of
its resources for the benefit of the colonial administration, missions, and European companies.
Much of the present land law is still either a legacy from or a reaction to the land law system
established by the Belgians (Piet, 2013; WorldBank/UN-Habitat, 2016; Mugangu Mataboro,
2016). The agricultural policy, on the other hand, focused on the forced cultivation of cash
crops with limited success. In the 1940’s, a new agricultural policy was tested to encourage
the evolution of customary land right to private property rights and encourage the emergence
of a class of small entrepreneurs, with limited success but lasting consequences.

2.1 Pre-colonial institutions, land tenure and agriculture

Before exploration and colonization began in the region, several ethnic groups speaking Bantu
languages had been living in the Congo Basin forests for four millenia. They formed complex
societies that shared common traditions and food systems, as well as social and political
institutions. They had already experienced the rise and fall of several chiefdoms and forms
of power, as well as the arrival of the Atlantic trade near the Congo river, as described by

6
Vansina (1990). Customary land rights were complex, variable, and poorly understood by
the colonizer (Piet, 2013).

Political institutions in the region were very decentralized and best characterized by a tension
between the desire for local autonomy and the need for security. Three groups are critical to
understand their social and political organization: the clan, the village, and the district. The
clan 1 was a large household establishment organized around an actual or perceived common
ancestor. Several clans lived together in a settlement called a village in which the founding
clan had a prominent role. Villages often bore the name of the supposedly ancestral founder
of the elder clan in the village. A district was a network of allied neighboring villages, which
retained autonomy but occupied of precise position in a hierarchy, from the elder village to
the newest. They came together mainly when they were attacked from outside, and conflicts
within districts were settled through mediation by the leader of the elder village. Sometimes,
fractions of a clan could split up to form a new settlement in new lands in response to
impoverishment of soils and games or to social events such as attempts to enforce autocracy.
The movement of men between clans and villages was also relatively flexible. Population
densities were low and allowed creation of a relatively stable equilibrium, punctuated by wars
and the expansion of some ethnic groups.

On the eve of the colonial conquest, districts, villages and clans formed a complex and struc-
tured network of communication and exchange in the Congo Basin. Farming had been well
established, and surplus was traded along the many rivers. The agricultural process was sim-
ilar among the different ethnic groups 2 : farmers practiced shifting cultivation and used slash
and burn techniques with a strong gender division of tasks. Typically, men took charge of the
clearing of forest or savanna and of the burning of the remains, while women took over for
planting, cutting weeds, and harvesting. This gender division of labor allowed men to devote
time to fishing and hunting that brought diversity to the diet. When soil fertility started to
decline, farmers moved to a new plot of land. When fertility started to decline in the village or
fields became too far from the village dwellings, the clan or village could move further. Many
villages were semi-nomadic and moved at irregular intervals to occupy new land or reoccupy
land they had left fallow for decades. The indigenous land tenure institutions were therefore
very flexible. Land rights were not static and incorporated the fact that agriculture was
complemented in large parts by fishing, hunting, and gathering (Salacuse, 1985; Piet, 2015).
The land belonged to the lineage groups (or clans) through their ancestors. Its allocation was
under the authority of a ”land chief”, while the members of the clan who cultivated it had
usufruct rights to the land. Population density was still low and allowed cultivators to leave
the land fallow for very long periods of time. Colonization violently disrupted this pattern.
1
”Etuka” in local languages, also referred to as ”House” in Vansina’s terminology.
2
Except for some ethnic groups which specialized rather in fishing or hunting and gathering such as the
pygmies

7
2.2 Predatory System under King Leopold

In 1885 the Berlin Conference created the Congo free state, a property of Leopold II King
of the Belgian. The first decree he passed regulated land ownership 3 : all ”vacant” land
was considered to be the property of the state in either its public or private domain, while
land occupied by indigenous populations could continue to be governed by the customary
system4 . State land was withdrawn from customary law and governed by written legislation
inspired from European legal concepts. A land registration system was created, and European
enterprises, missions, or colonists could obtain grants or concessions in the private domain of
the state. The land to be obtained was surveyed, delimited, and registered before a title was
issued. This decree effectively meant the expropriation of the majority of the Congolese land
(Piet, 2013). Even if the legislation formally recognized customary land rights, in practice
the administration denied them and considered vacant all the land that was not currently
formally occupied by settlements. It was based on an illusion: many scholars consider that
no land was completely vacant at that point and that all of it was claimed by one indigenous
group or another (Vinck, 2011; Salacuse, 1985). The land legislation provided both a way
to attract investments, and a source of revenue to fund colonization, but land occupied by
indigenous populations was not subject to registation and titling, putting them in a very weak
position.

Between 1891 and 1893, the colonial administration promulgated two further decrees stating
that the state had exclusive rights to the produce of the domanial lands that provided the
legal basis for the exploitation of the resources. A heavy taxation in kind was imposed
on the Congolese farmers to generate revenue: the population was forced to collect ever
increasing quantities of ivory, and later wild rubber and copal. This forced farmers to spend
a large share of their time away from their fields and often abandon them. It eventually
led to the exhaustion of the wild rubber reserves and to international indignation. In 1905,
an international commission led by British diplomat Roger Casement published a report
that underlined the abuses and atrocities committed by the administration. The report
pointed to the land legislation as one of the main causes of the abuses and as a threat to
future development of the indigenous communities (EIC, 1905). Between 1885 and 1908, it
is estimated that the area over which Leopold II made grants and concessions covered more
than 27 million hectares, out of a total land area of 234 million hectares (Salacuse, 1985).
3
Decree of July 1, 1885, Article 2: ”No one has a right to occupy vacant land without legal title, nor to
dispossess indigenous people from the land they occupy. All vacant land must be considered to belong to the
state.”
4
Decree of September 7, 1886, Article 2: ”lands occupied by the native population, under the authority of
their chiefs, shall continue to be governed by local customs and usages”

8
2.3 Colonization, continuities, and forced cultivation

In 1908, following intense diplomatic pressure and financial difficulties, the Belgian Parliament
took control of the Congo Free State and created the Belgian Congo. The new colony put an
end to the rubber system, and attempted to reform the land legislation without changing its
fundamental principles, and introduced forced cultivation.

After the annexation, many of the concessions were renegotiated, but new ones were also
granted. New legislation aimed to define more precisely what land could be considered ”va-
cant” and how local communities should be compensated. 5 In 1935, a new decree defined
the procedures to be followed by territorial administrators before ceding domanial land6 : a
survey investigating the vacancy of the land and the customary land claims had to be con-
ducted, compensation were calculated based on that, and local chiefs had to be involved in
the determination of the boundaries. It was long, but provided the concession holder with
legal certainty. However, this proved very challenging in practice and local administrators
lacked the time and resources to implement the decree. The demarcation of the land was not
only difficult because of the lack of financial and human resources, but also made difficult
by the transitory nature of many of the villages who were reported to disappear after having
been demarcated. The presumption also remained that all the land was vacant unless local
populations could provide convincing evidence that it was not. In practice, villagers had no
bargaining power and all they could do was to hope to receive a small compensation. The
administration relied on village chiefs (also called ”capitas”) to negotiate the terms of the
arrangement. Those chiefs were often appointed by the state and replaced when not obe-
dient, and they received the compensation directly when there was one. This policy had a
broad impact in the region: thousands of hectares of land were given in concession, creating
much resentment among local populations. While population was sparse and land seemed
abundant, good arable land was not, and labor even less. After commercial concessions took
possession of large swathes of land, they set up mining activities or plantations of cocoa,
rubber, or palm tree. Those required labor resources that they tried to satisfy locally, and
indigenous people were hired, often by force, for meager salaries. As of 1944, a total of twelve
million hectares are though to have be controlled by European companies or individuals.

The new colonial administration also expressed worries about the low agricultural production
by the indigenous population. In 1917, they introduced the Compulsory Cultivation Scheme
7 to promote indigenous production of cash crops and increase exports. The administration
5
Several decrees passed in 1906 and 1909-1910 aimed to define more precisely the term ”vacant” and to
require investigations and compensations about the rights of the indigenous communities. A new definition of
vacant was introduced ”Land which the indigenous people cultivate or exploit in whichever way, in conformity
with customs and local practices”.
6
Decree of May 31, 1935
7
Decree of the 20 February 1917

9
decided which quantities of which cash crops should be produced in each village, and farmers
were obliged to grow crops and sell their production to a monopolist. Although the Colonial
Charter prohibited the use of forced labor, it was a coercive scheme, closely monitored by
state agents. Although exports of agricultural products improved, the system proved very
unpopular and the benefits for commercial companies were made at the expense of the local
populations. Low production or disobedience were sanctioned by fines and imprisonment. The
scheme increased the leverage chiefs had over their villages, and systematically undervalued
the role of women in agricultural production. While it was theoretically limited to 60 days
per year, in practice crops like cotton required many more days to be cultivated and it
limited the time farmers could devote to hunting or fishing. It further tied farmers to their
land and worsened the soil fertility issues associated with more sedentary villages and mono-
cultures, and the rural exodus accelerated (Likaka, 1997). The scheme was widely criticized in
particular by the International Labor Office, but even if was renamed to be called ”educational
scheme”, it continued with very little change until the independence of the country in 1960.

District administrators were responsible for the supervision of the scheme, but the operational
tasks were led by the territorial agents. Since 1933, the Districts were the intermediate
level between the colonial and provincial governments on the one hand and the territorial
agents and indigineous chiefs on the other hand(De Clerck, 2006). The role of the District
Commissioner was wide-ranging: he had important judiciary responsabilities (both as the
juge of the district and because he was overseeing the indigenous tribunals), he played a key
role in the attribution of Land Concessions to foreign companies and determining levels of
taxation, and he had to oversee the work of territorial agents and in particular the creation
of ”chefferies” and ”secteur” and the appointment of their chiefs. Since 1912, the territoire
was the basic unit of the administrative division of the country, and the lowest administrative
level led by a Belgian expatriate. He was in charge of maintaining public order, collecting
the taxes, lead infrastructure investments etc. He was also entrusted with conducting surveys
about indigenous organization in order to create the afformentioned ”chefferies” or ”secteur”,
a lower administrative level led by indigenous chiefs. Relationships between district and
territorial administrators were not always easy, and the implementation of paysannat was no
exception(De Clerck, 2006).

2.4 The introduction of Paysannat: a new approach to agricultural policies

and land issues

Between 1940 and 1960 a new ambitious agricultural policy was developed and implemented,
the ”paysannat indigène” (hereafter paysannat). The starting point for this scheme was the
recognition that the compulsory cultivation scheme as it was failed to improve the situation
in the countryside and to stop the rural exodus. Agricultural productivity remained very low

10
and declining soil fertility had become a pressing concern (Likaka, 1997).

The core idea of Paysannat was to reorganize and modernize the indigenous agriculture based
on a scientific approach. The objectives were twofold: first, the colony aimed to create
a class of small farmers owning their land and selling their produce on the regional and
national market. Second, they aimed to push the evolution of collective land tenure towards
a system of private property rights. Families were grouped around a homogeneous block and
were assigned plots on land that traditionally belonged to their clan. Rotation of crops and
duration of fallows were determined by the state agronomists to ensure the preservation of
the soil fertility. On their individual plots, they were asked to produce cash crops (cotton
most of the time), sometimes perennial crops (hevea or palm trees), and food crops for their
own consumption (cassava, maize, and groundnuts)(Piet, 2015). In practice, in meant the
distribution of collective land to individuals.
Figure 1: Land allotments for Paysannat, from Piet (2015)

The hope was that customary land would become privatized over time, even if official doc-
uments avoided to mention it through fear of fueling the contentious land issue. In the
propaganda, they instead insisted that the scheme would provide villagers with a way to
protect their village land from being declared vacant and given into concession to a foreign
company (Piet, 2015). This argument was meant to help convince them to voluntarily par-
ticipate in the scheme. However, local populations lacked enthusiasm and perceived it as the
continuation or at best an extension of the compulsory cultivation scheme. The territorial
agents had to resort once more to coercion to expand it. 8

The scheme started at a very small scale as an experiment conducted by the National Institute
8
This perception is very visible in the local names that were given to the scheme ”kasangisa”, meaning
”being forced, coerced” in Katanga.

11
for Agronomic Research with one family in 1936. At the beginning of the 1940’s, European
Cotton Companies became very interested in the scheme as a way to improve soil fertility
and proposed to help extend it in a combined effort with the colonial administration. Cotton
companies saw it as a way to safeguard and increase production of cotton which farmers
were forced to sell to them at fixed prices. In 1948, the systematic extension of the scheme
became a key objective of the Ten Year Plan for social and economic development. It was
decided that 500.000 farmers with their families would be allotted their own plots of land
under the program between 1949 and 1959. 9 The paysannat expanded quickly in Kasai,
Orientale Province, and in Katanga and Equateur. By 1955, 150.000 families had already
been assigned a plot under the scheme.

However, implementation turned out to be an extremely difficult task. District administrators

were respondible for the implementation of the scheme, but the operational tasks were con-
ducted by the territorial agents who led a small team of agronomists and assistants. They had
to cover large areas in very difficult transport conditions to prospect suitable land, study the
local customary rights in each village, and distribute the land within a short period of time.
10 . The prospection was meant to identify land suitable for agriculture (with a preference for
forests), and the land survey was meant to establish to which villages and which clans the land
belonged to. In practice, there were unsufficiant numbers of ”political specialists” to conduct
this delicate work, and their lack of diplomatic skills led to misunderstandings and sometimes
conflicts with local chiefs. In many instances they tried to convince villages to give parts of
their land to neigboring villages, to move the villages to areas with more untouched forest,
and failed to reckognize which clans owned the land. They were often very autoritative with
local chiefs, despite instructions from district commissioners. We will provide more detailed
information about the selection of the villages in section 3. In 1955, acknowledging there were
many implementation issues, the administration decided to slow down the expansion in order
to ”learn from past experience and to adjust when necessary”. In 1959, they had reached
210.000 farmers and their families: less than half of the objective, but a significant share of
the population.

2.5 Post-independence trajectories

The Paysannat program was swept away immediately after Congo gained independence in
1960, after achieving only limited success despite the large resources devoted to it (Salacuse,
1985). Incentives to invest remained limited for small holder farmers since their access to mar-
9
”Etude paysannats indigenes” , In African Archives Brussels, A21 (749) - Agriculture Indigene, 1934-55,
dossier 7.0.c : Paysannats indigenes, 1954-56.
10
”Compte-rendu de la reunion tenue le 5 mai 1952 Ã Itipo relative au Paysannat dans les territoires du
District de l’Equateur”, In AAB, GG 17051 - Equateur AGRI-AIMO, Paysannat Indigène, Programme Plan
Décennal.

12
kets was constrained by state-fixed prices and monopolies. Despite localized success-stories,
there was no evidence of increased living standards or significant increases in indigenous
productivity. Aggregate production of cotton however had flourished and in 1960, the Demo-
cratic Republic of Congo was one of the largest cotton producers in Africa. But production
fell dramatically after the Independence of the country, and the crisis of this sector worsened
after the ”Zairianisation” of the country launched by Mobutu Sese Seko in 1972 and the
nationalization of foreign companies.11

The land tenure system remained the same after Independence. Despite attempts to make
a full inventory of the land under grants or concessions and to redetermine their status,12 in
practice few changes were made to the existing concessions and to the law, and the view that
the land was plundered by foreign interests spread. Four types of land holdings were still
recognized: land belonging to the states, concessions (for limited periods of time), individual
registered land rights (for companies or individuals) and land governed by customary law
(WorldBank/UN-Habitat, 2016). Under Mobutu, in a movement to end the privileges of
foreign interests obtained during the colonial period, an amendment to the Constitution was
voted in 197113 and a new General Property Law was enacted in 1973 14 that still prevails
today. The general principle is that all the land belongs to the state, and that the land
can only be given in concessions to companies or individuals who can not claim ”ownership”
of it, but only ”rights of enjoyment”. Concessions can be ”perpetual” for nationals, or for
limited periods of time for foreigners, and those rights have to be registered as previously. In
practice, most of the existing contracts were transferred into the new system. While since the
beginning of colonization it had not been clear who the customary land belonged to, the new
law declared the land occupied by indigenous population to be part of the domanial land,
over which communities hold ”rights of enjoyment”. Article 387 of the 1973 Law stipulated
that such rights of enjoyment were to be regulated by an presidential ordinance, which still
has not been enacted as of today (WorldBank/UN-Habitat, 2016). While the law did not
change anything in practice, it cast doubt about the future status of this type of land which
constitutes most of the country’s land. Will it be turned into concessions ? Into collective or
individual property rights? In the mean time, land chiefs remain crucial actors locally even
11
”Zairianisation” was a nationalist political process launched in 1973 by president Mobutu, by which,
among other changes, agricultural businesses that belonged to foreigners were transferred to citizens of the
DRC. In many cases, the new owners did not continue to farm the land they received and many plantations
were abandoned.
12
Ganjika Law, 1960: the state ”retakes the full and free disposition of all land, forest and mining rights
conceded or granted prior to June 30, 1960.”
13
Law No. 71-008 of December 31, 1971 amending the Constitution” The Zairian land (sol) and subsoil, as
well as their natural products, shall belong to the State. The law shall fix the conditions for their assignment
and concession, for their retaking and retrocession. However, the retaking or the retrocession in case of
non-development (non mise en valeur) shall not give rise to the payment of any indemnity.”
14
Law no. 73-021 of July 20 ”Law Providing for a General System of Property, Land and Immovable System,
and Securities System” (Loi Portant Regime General des Biens, Regime Foncier et Immoblier, et Regime des
Suretes.)

13
if they are not mentioned anywhere in the law.

After Independence, rural population densities increased and the perception of growing
land pressure emerged, to which farmers reacted gradually by seeking security on their
land holdings. Three types of tenure can now be distinguished in the villages: traditional
customary rights, semi-formal land rights, and legal concessions (WorldBank/UN-Habitat,
2016).
First, as is many other countries in Sub Saharan Africa, traditional customary rights evolved
dynamically toward more individualization and transferability of rights in response to
increased land pressure and years of commercialization in agriculture introduced in colonial
times (Platteau, 1996). Qualitative and quantitative data collected in the region evidenced
that families or households now tend to have exclusive rights of use over parts of the land
of their clan while part of it remains collectively owned by families, clans, or villages. Sales
and lending of land have started to emerge although they are limited, and direct inheritance
between fathers and sons is now widespread. Forest that has not been cultivated for a very
long time (also called primary forest or ”ngunda” in local languages) are partly divided
between clans or families, but are still frequently common land regulated by the village or
groupement chief. Families from the village usually have a right to open fields in this new
land, and the clearing of the field often gives them the right to appropriate the land and
sometimes pass it to their descendants. The collective belief is that all the landholdings of
a family are fields that have once been opened and cultivated in the forest by their direct
ancestors, which justifies their claim on it. The only other way to obtain new land in a
village is to buy it from another farmer.
Second, an increasing number of farmers attempted to obtain some documentation to claim
their rights on the fields they had obtained through customary practice or by buying it from
local land chiefs in or outside their village: those documents do not have any legal basis,
but they provide some security locally since they are usually signed by land chiefs and local
administrators. However, the farmers who hold them would not have any legal protection
and would lose their land if someone secured a concession on them.
Last, some small farmers managed to obtain legally registered titles for land outside their
village where they set up small farms, but they still constitute a minority. Many commercial
companies still own concessions over large areas in what used to be village land.

Coming back to land distributed for Paysannat during the last 10 years of colonization, it
seems to have been often passed within the families who received them from one generation
to the next. Today, villages where Paysannat was implemented in the Equateur Province
still remember where the blocks of land were on which they cultivated cash crops. They
remember which clan received parcels of land in this period, and the owners are often still the
descendants of those who first received them. Some families even consider it as a ”collective

14
family plots” and are reluctant to split it among themselves. Because those land distributions
took place in a context of coercion and violence, they were often interpreted as mandatory as
opposed to voluntary. Families therefore perceive that they have very strong rights on them,
and they are now part of the collective oral memory.

15
3 Data and summary statistics

This paper uses a first-hand dataset collected between 2012 and 2015 in the Equateur Province
as part of a randomized experiment conducted on a World Bank funded agricultural project
called ”PARRSA” (see appendix for more details). Using this data, we construct treatment
and outcome variables at the village and household level for our analysis.

3.1 Data

3.1.1 Household level data

The data used in this paper is first hand data collected in 192 villages in the Equateur Province
in DRC. The 192 villages in the sample were selected across the region by the PARRSA team
for their accessibility by road in each of the 9 territories involved in the project, and were
surveyed three times between 2012 and 2015.

In 2012, we visited the 192 villages and in each of them a group of 4 to 5 people was selected
to answer a community level questionnaire as well as an indirect household survey about
20 randomly chosen households. The group of informants was composed of knowledgeable
people in the village, including the chief of the village, the director of the school, the director
of the dispensary, and other notables, depending on availability. At the village level, we
collected information about external interventions, agricultural practices, institutions and
conflict resolution, infrastructure, access to markets, common land, and the land market.
We also conducted a full population census, from which we randomly drew 20 households.
We then collected basic household characteristics related to agriculture, demographics, and
poverty through an indirect survey from the same group of informants (by asking information
about easily observable characteristics).

In 2013, a subsampe of 92 villages were visited in relation to the experiment and we col-
lected detailed household and plot level data about agricultural inputs and production. We
surveyed 22 households per village : 12 with a detailed questionnaire, and 10 with a shorter
questionnaire focused on program take-up and information.

In 2014 and 2015, we visited the same 12 households in 92 villages for which that we surveyd
with a long questionnaire in 2013, as well as 6 households drawn from the households surveyed
with a short questionnaires in 2013. We also added households drawn from a list of the fratry
of the household heads (a maximum of 10 such ”brothers” per village) to study inequalities in
access to land, inheritence rules, and diffusion of technology within families. We exclude them
from the sample for this paper since the selection may be different from the rest of the sample.
We also surveyed 12 households per village in an additional 100 villages. The survey covered

16
a wide range of topics ranging from agricultural inputs and production, nutrition and health,
intrahousehold decision making, external interventions, conflicts, credit and labor sharing
arrangments, networks, and access to land. We also collected detailed information about the
composition of households and the clanic affiliations of each member, as well as a history of
the arrival of the household in the village. For all these households, men and women were
interviewed separately, and two women were interviewed in polygamous households (In case
of polygamous households with more than 2 women, the first wife and a randomly selected
other wife were selected). Because of the length and complexity of the survey, we visited
households twice : a first wave was conducted between June and July 2014, and a second
wave was conducted between November 2014 and March 2015.

This quantitative data was complemented by a series of qualitative interviews conducted

between 2011 and 2016 in several villages, to deepen our understanding of several key issues
such as production choices as well as land institutions and access to the collective forests.

In this paper, we restrict the sample in two ways : first, we excclude the villages who have
not yet been created before Independence (8 were created during Mobutu’s rule and 4 were
created shortly after Independence), since they have not experienced paysannat and can not
be compared to villages that experienced it. Second, we exclude from the sample households
who are village chiefs, households who are part of the ”brothers” sample described above, as
well as households who received a gift for helping use to collect rainfall data.

3.1.2 Oral history data and colonial history record data

An original feature of this paper is the use oral history data from a community survey we
conducted systematically in the same 192 villages between 2014 and 2015. In each village,
a group of 4 to 5 people was selected to answer a community level questionnaire and detail
the history of the village. The group was composed of knowledgeable people in the village,
including the chief of the village, the director of the school, the director of the health center,
and other notables (clan leaders or elderly people). This in depth village survey provides
information about the history of village since colonization, the migration waves to and from
the village, the forced cultivation of cash crops, and historical land distributions. It also
details the current situation by providing detailed data about the institutions that govern
access to land and other natural resources available in the village, recent governmental and
non governmental projects, as well as recent conflicts and how they were settled. This data
was complemented and confirmed using histocal record data collected from the ”Archives
Africaines” at the Ministry of Foreign Affairs in Belgium, as well as contemporary research
on paysannat by historians (Piet, 2015). Those records contain detailed information about
the implementation of the scheme and the difficulties encountered.

17
There are several justifications for the use of oral history data as a primary source of in-
formation, complemented with colonial administrative data. First, Belgian colonial records
are often incomplete and biased (Likaka, 1997, 2009). Colonial administrative data was built
by colonial agents and in particular territorial administrators. The incentive for them was
to provide the best possible picture of the work they were doing in order to be promoted,
in particular when they were posted in such hostile and remote regions as Equateur. Their
work was very hard to monitor, and they often underestimated the reluctance of villagers
to participate in cotton cultivation and in Paysannat, and overestimated the success of their
undertakings (Likaka, 1997). The names of the villages were often wrongly recorded, and
the colonial maps of the region are very unreliable (Piet, 2015). Moreover, African societies
have long favored oral tradition. Likaka’s work showed that oral based history can be used to
reconstruct the history of cotton production in the region and how the villagers resisted it. In
fact, I found out about Paysannat in the first place while conducting qualitative work while
villagers were telling me about the different types of land in the village, and they pinpointed
Paysannat fields as being very different from the others because they had been distributed by
the Belgians. If dates collected this way don’t seem very reliable, villagers remember distinctly
the key events in the history of their country (colonization, independence, the rise of Mobutu,
Zairianisation, and the more recent civil wars) and they are very confident when trying to
locate village events within this framework. The oral history I reconstructed this way allowed
the creation of historical variables that I use in the analysis such as cultivation of cotton and
distributions of land related to Paysannat. Of course, oral history data may also have its own
biases. Many villages have been involved in Paysannat for a very short time, and the limited
resources of territorial administrators implies that some of them may only have been visited
a couple of times. In those cases, it is unlikely that it had any impact on the village life and
that people still remember it today. The colonial intervention variables constructed this way
should therefore be interpreted as reflecting only interventions they were sufficiently intense.
Colonial records were used to confirms that territorial administrators indeed reported the
implementation of paysannat in the same areas, even if it did not allow to check for each of
the village what exactly had been done. It gives us confidence that our oral history variables
are sufficiently reliable.

”It is no consolation to be told by others that, because there are no written

sources, no past can be recovered, as if living traces of that past were not part and
parcel of daily life.”

Vansina, Paths in the rainforest, p.xi

18
3.1.3 Geographic data

The geographic data we use in this paper comes from various sources. Geographic data for
the village geographic locations and borders comes from the village surveys described above.
Port and markets names were collected in our village surveys and localized using geonames.
Data on roads and administrative borders comes from the Ministry of Infrastructure. Finally,
data on forest cover comes from Hansen et al. (2013). Distances from villages to key locations
were computed using ARC-GIS network analysis tools, in order to identify the nearest location
(among the list of ports or commercial urban areas, administrative capitals in the region) and
calcute the distance to these locations by road.

3.1.4 Information about clanic affiliations

One piece of information that is critical in this paper is the clanic affiliations of households.
In each househod, we asked the household head and his wife to tell us the name of the clan
that each member of hi household belongs to. In most cases, his children were affiliated to
his own clan but his wife was still considered a member of the clan of her own father. In
parallel, we conducted the community survey with village leaders and drew a list of all the
clans living in the villages. After doing this, we would go through the history of the village
and ask for each period of time whether any immigrants settled in the village, and to which
clan they corresponded in the list. Among the clans who arrived after the creation of the
village (thereafter ”foundation clan”), we often realized that the village leaders had not listed
some of them in the list of clans. When this happened, we asked why: if they had simply
forgoten them, we would add them to the list. If they had not listed them for a good reason,
we added a comment but did not add them to the list of village clans. The most common
story was that they simply did not know the name of the clan of this family, because they did
not have any voice and representation at the village level. It was common for those clan to
have been ”adopted” by one of the local clans of the village, although it seemed to indicate
a very different condition from being an autochtone. We then matched the clans declared by
households with those listed by the villages using fuzzy matching techniques based on names,
and divide them into five categories : the foundation clans (the clan of the household’s head
matches one the clans listed by the village leaders as one who took part in the creation of
the village), the local clans (the clan of the household’s head matches one of the clans listed
by the village leaders, but not there at the creation of the village), women’s foundation clans
(the clan of the household’s head is not identified in the village, but the clan of one of the
wives of the household’s head was matched with one of the foundation clans in the village),
women’s local clans (the clan of the household’s head is not identified in the village, but the
clan of one of the wives of the household’s head was matched with one of the local clans in

19
the village), and marginal clans. We call marginal clans all the clans who were not listed by
the village leaders : they represent 23% on average of all the households living in the villages
we surveyed, and encompass the clans who were identified in a different village, those who
gave us the name of one of the villages in the region, and those whose clan name could not
be found either in the village clan list, in the list of villages in the region, or in the list of
clans from all the villages we surveyed. It is very reasonable to assume that we would be able
to match almost all of them with clans in other villages if we had the full list of all clans in
the region. We call them ”marginal” because they are not represented at the village level,
even if they may have been adopted by a local clan. We could also call them ”migrants” or
”newcomers” but we are reluctant to do so since some of them have been in the village a long
time, and other households recently moved to the village but were originally autochtones so
they migrated recently but have full access to their clan.

20
3.2 Background: the Equateur Province

The provinces of the Democratic Republic of the Congo are divided into 26 districts (see
Map 2, that closely resemble the districts that existed during colonization. These in turn are
divided into 192 territories. During colonial rule, districts were led by a district commissioner
who supervised territorial agents which was the lowest administrative level held by Belgian
agents (De Clerck, 2006)15 . In this paper we study three districts 16 in the North of the
Equateur Province, that are divided into 9 territories. As far as we can tell from approximate
and low resolution colonial maps, the delimitations of the districts changed very little after
1940 and only two changes had been made until Independence in 1960: the name of ”Bom-
boma” district had been changed for ”Kungu”, and ”Banziville” district had been split into
two districts named ”Businga” and ”Banziville”. By 2015, the borders of the districts were
still very close to the colonial borders of the colonial districts in 1960 even if ”Banziville”
district changed name during Zairianisation to become ”Yakoma”.

The Equateur Province is a remote region in the Congo Basin Forests, with extensive slash
and burn agriculture, high levels of food insecurity and extreme poverty, and arguably severe
constraints to economic development in other sectors (Herdeschee et al., 2012). The main
cash crop that was produced there as part of the compuslory cultivation scheme during
colonization was cotton, but rice, cocoa and palm oil were also important in parts of the
region. Cotton production started falling after Independence, and the agricultural sector was
further dislocated during the Zairianisation. The region includes many historical plantations
of hevea, cocoa or palm trees that are still abandoned or function at the smallest capacities.
Some of them resumed functionning in recent years, but there are still many cases of conflicts
with local villages and previous belgian and congolese owners of plantations who all claim
property of the land (WWF, 2015). Infrastructure was poorly maintained after Independence
and its condition degraded until they finally collapsed during the Congo Wars (1996-2003).
Inadequate infrastructure for the transport of agricultural products is still a key constraint
(Herdeschee et al., 2012). In this context, households rely mostly on subsistence smallholder
agriculture through shifting cultivation of staples 17 , as well as gathering of forest products,
fishing and hunting. While agricultural potential is believed to be large, road density is very
low, commercialization is hampered by long distances from farm to market, and most farmers
don’t have access to improved varieties or technologies. Traditional slash and burn clearing
methods are still used, and generally the only capital inputs used are hand held tools.

15
In 2015, a reform of the territorial organization turned the districts into 26 provinces, but the borders
remained very close to the borders of the districts in 2015 and remained very close to the colonial district
borders.
16
Nord-Ubangi, Sud-Ubangi, and Mongala
17
maize, rice, groundnuts, cassava, soybeans, cowpeas, yam, plantain

21
3.3 Village characteristics

Villages are characterized by a large heterogeneity in terms of demographic and geographic

characteristics, but also access to land and natural resources. Tables 1 to 12 as well as figures
4 and 5 present summary statistics of those characteristics.

Consistent with the previously described historical settlement patterns in the region, most
of the villages in our sample were formed before or during colonization (table ??). Only
a few were created later, usually the result of a preexisting village growing too large to be
administered properly and being split into two parts by the administration. The population
of the villages varies widely. There are on average 287 households per villages, but it ranges
between 13 and 2500. Population density varies widely between 4.46 inhabitants per square
kilometer to 378. The ngwaka, ngbandi, budja, and mbanja are the 5 dominant ethnic groups
in the region. As evidenced by figure ??, most villages are dominated by one principal ethnic
group, while only a few are composed of several ethnic groups who cohabit under the same
chief. There are on average 5.8 clans per village, even if a minority is composed by only one
clan. Typically, the clans are relatively balanced and only one third of the households belong
to the most numerous clan on average.

As evidenced by table 2, villages differ in terms of geographic characteristics. In terms of

access to markets, the distance to the nearest port in 55 kilometers on average, from 0 to a
maximum of 300. Distance to the nearest local market is very small, 6 kilometers on average,
because there are several local markets along the roads. In terms of access to resources,
access to forest at the village level varies widely as evidenced by figure ??: first, the area of
the village varies widely with some village land spanning over very large areas; and vegetation
is much denser in the South and East of the region, while the North-West is characterized by
forest-savanna mosaic.

Finally, land formalization varies both between and within villages. As described in section 2,
farmers in the region barely own any legally recognized titles on their land, nor do the villages.
In practice, land is managed by the chief of the groupement and the chief of the village. To
protect their rights, some farmers obtain semi-formal documents to secure their land. In our
sample, those semi-formal titles have been issued or signed either by the Groupement chief
or the Secteur chiefs, or they are documents they issued when they bought the land. We also
found evidence of 81 cases were farmers claim to own a ”Paysannat document” (see table
6. The share of people owning documents varies widely from on village to another: Table 5
shows that on average 13% of the households surveyed declare owning at least a document
on one of their fields. Additionally, 19% report owning land in paysannat (a very small share
of which is associated with a document). While traditionnally land chiefs were in charge of
allocating and managing the clanic or family land, today 26% of the households claim that

22
there is no such land chief regulating their access to land (tabe 7). 20% say that they are the
land chief for their families, and almost 50% name either their elder brother, uncle, father or
grand-father.

Commons:

Households cultivated 2.79 hectares on average in 2014, with an average of 4.3 culivated plots
and they owned 6 additional fallow plots. While half of their fields cultivated in 2014 came
from land that was already cultivated the year before, a large share came from land that was
left fallow and 4% came from a conversion from primary forest to agriculture. The main crops
cultivated in the region are maize, groundnut, cassava and rice.

Finally, table 4 presents statistics about the clanic affiliations of households in our sample.
We asked the household head and his wife to tell us the name of the clan that each member
of the family belongs to, and we then matched this name with the list of clans mentionned by
the village leaders in the community questionnaire. We find that 52% of the households in our
sample belong to one of the foundation clans, 20% belong to a men’s local clan (ie the clan of
the household’s head has been identified in the village), 2.7% belong to a women’s foundation
clans (one of the wives of the household’s head is from a foundation clan in the village), 1.8%
are from a women’s local clan (one of the wives of the household’s head is from a local clan
in the village), 9.7% of the households clans were identified in another village, and 13.6%
could not be identified. In total, we therefore have 22% of households from a marginal clan.
Compared to the households from one of the foundation clan, those households report much
less often that they own land in paysannat, they seem to have more fields with a document,
and cultivate a smaller area in 2014. They also report much more often being worried to lose
at least one of their fields. Finally, they report less often that they have access to collective
lands, whether at the village or family level.

23
4 Empirical Strategy

4.1 Exogeneity : what we learn from historical records

In this paper, we aim to compare villages of a given Secteur in which land was distributed to in-
dividual households for paysannat to villages in the same Secteur in which these distributions
did not take place although they were cultivating cash crops as well. A key methodological
question is whether they differ in systematic ways: were the villages chosen based on specific
characteristics that can also explain why they differ today? We discuss this hypothesis and
argue that it is unlikely. Many villages were therefore not part of Paysannat not because they
were not deemed suitable or had different characteristics, but simply because of the accident
of history. A second important question is wether being part of paysannat may have triggered
a differentiated treatment by subsequent administrations in the provision of public policies.
Again, we find no evidence that this is the case.

As described earlier, villages were identified by territorial agents and their team. However,
they were constrained to intervene in a list of Territoires and Secteurs that had been chosen
for paysannat (they chose areas where agricultural potential was high, and deforestation rates
were worrying or where the quality of the soil had become a concern for the colonial admin-
istrators). The historical records evidence that paysannats were set up in the territoires of
Kungu, Gemena, Budjala, Bumba and Bosobolo. In those administrative areas, the territorial
agents had to prospect the land, conduct land tenure surveys to understand the land tenure in
the area, and attribute the land to individual households within the customary boundaries of
their own clan or family. To the extent possible, they had to ensure that the land attributed
to each household was fertile, and that customary chiefs were willing to participate 18 .

Prospection and land surveys were costly and labor intensive. In practice, they often iden-
tified several blocks of land in the forests or in fallows, and reallocated land between clans
or villages. Understanding customary right was an immensly complex task : there were sev-
eral levels (clan, families) and types of rights 19 . Moreover, Territorial Administrators and
Agronomists were not popular, so the population was reluctant to participate 20 . The diffi-
culty in obtaining reliable information is evidenced by the many cases in which colonial agents
were often mistaken even in the identification of the village chief (Likaka, 1997). A report
about the implementation issues faced in the implementation of Paysannat in the Equateur
18
”Le paysannat indigǹe, Annexe lettre 99 21/77 99/2357/11.D/18 bis du 3 avril 1953. Rapport destiné
à la Conférence CCTA sur le bien être rural. Laurenco Marques”, In African Archives Brussels, AGRI 80
Dossier 1
19
20
”L’agriculture dans les perspectives d’avenir, Lebughe Pierre, Ingénieur agronome congolais”, In African
Archives Brussels, AGRI 748, Dossier 5

24
region 21 mentions their frustration and their difficulty to even identify village names because
they had to rely on local interlocutors who sometimes disagreed or tried to mislead them
(Likaka, 1997). Based on this, it is easy to imagine how difficult it was for them to conduct
the so-called survey of existing customary rights, and to identify a block of land that would
be suitable for Paysannat. They did their best to find suitable blocks of land with the limited
information they had, near the roads. When they thought that pressure on land was too
high in one area and found land that looked ”vacant” further away, they did not hesitate to
relocate the villages. The construction of roads was also extended in this period to reach the
villages growing cotton. After independence, the road network did not improve but rather
slowly fell apart, and all the villages were are studying are along these colonial roads so they
were comparable at the time of independence in terms of accessability by road.

When the expansion of the scheme was abruptly slowed in 1954 to focus on improving moni-
toring of existing Paysannats, administrators were in the middle of their prospection and had
to slow down the expansion. One of the concerns of the upper levels of administration was
that villages were chosen before a full prospection of the region was conducted, and without
a full investigation of customary rights. They were therefore chosen without full information
about which villages were more suitable in a given Secteur, and they could not complete
work in all the villages they had first identified. Even if they had had full information on
the villages, the instructions were to identify villages with fertile collective land available, in
order to increase production of cash crops. Whenever possible, they tried to find collective
forests to divide it between households. This means that these villages should have had less
land pressure, lower levels of privatization, and less population pressure than those they did
not chose in the first place (Platteau, 1996; Boserup, 1965, 1983).

One question is therefore why how they chose the villages where they started ? One plausible
hypothesis is that administrators were inclined to start in villages where they had a better
relationship with chiefs. The relationships between colonial administrators and the local
population were based on the use of threat and violence, and in this context the ”voluntary”
participation of the villages became in practice a forced participation (Likaka, 1997; Piet,
2015). Administrators relied on local chiefs for the collection of taxes and the supervision
of cotton production, and they did not hesitate to punish or replace them if they were not
cooperative. However, the villages were not given complete information about what the policy
involved: the administration avoided on purpose mentioning that they would try to modify the
land tenure system when they presented the scheme, and instead insisted on the improvement
of agricultural techniques and standard of living. In that sense, it is hard to imagine how
the local elite would have been able to distinguish the scheme from the existing compulsory
21
”Compte-rendu de la réunion tenue le 5 mai 1952 à Itipo relative au Paysannat dans les territoires du
District de l’Equateur”, In African Archives Brussels, GG 17051 - Equateur AGRI-AIMO, Paysannat Indigène,
Programme Plan Décennal.

25
cultivation scheme. The propaganda also insisted that they would be less subject to arbitrary
expropriation by commercial companies, so villages relatively closer to a concession may have
been more to this argument.

Finally, I found no evidence so far that chiefs complacent with paysannat were given other
kinds of advantages such as better access to school or other investments that could have had
a long term impact, or that Post-Independence policies took the location of Paysannat into
account when making investment decisions after the Independence of the country in 1960.
The paysannat scheme was associated with social interventions (health centers and schools)
at the level of the Secteur, but this part of the program was very late to start and there is no
evidence that villages without paysannat in the same secteur were excluded from it.

4.2 Paysannat in our sample, village level

In our sample, 104 villages report having experienced forced definitive distributions of land
related to cash crops implemented by the Belgian Colony. By definitive we mean that they
were understood by villagers as being a permanent attribution of a block of land, and not
simply a temporary allocation for the duration of the cultivation period. Among those 104
villages, the distribution was done in a single year in 41 villages while in 63 villages the
administrative the division was done more progressively.

The data from community level interviews also shows that many villages where land distri-
butions were not imposed for paysannat experienced several other types of land allocation for
cash crops. In cotton producing villages for example, it was common that the village chief
(helped by agents from cotton companies) would reserve one block of land for cotton culti-
vation for the whole village. He would then allocate parcels of land to individual households
for forced cultivation for a few years. In most cases, the land was used for a few years of
cash crop cultivation followed by subsitence crops before being returned to the community. In
some of these villages, the families decided to divide these blocks of land between those who
used it for cotton - a privatization decided by the villagers. In other villages, these blocks of
land are still considered as common village land, and managed in various ways. Finally, in
some villages cotton was cultivated on the plots that households chose freely.

When constructing village level treatment variables, we only considered as ”paysannat” vil-
lages where land was distributed with the intention of being definitive. It means that we will
compare them to villages where the privatization of the land used for cotton or cash crop
cultivation sometimes also occured spontanously. In paysanant villages, the fact that the
distribution was intended to be definitive does not necessarily mean that it was. We found
several examples of villages who decided to reallocate the land between households, where
previous owners of the land took back control over (another indication that the land was

26
not always given to the right clans), or where it was reestablished as collective village land
managed by the village chief.

Interviews with village elders also revealed the violent way in which cash crops were cultivated
in all villages, and the violent way in which common land was divided, or village were moved.
When we asked who in the village benefited from the distributions, they virtually always
replied that every household head received the same area of land and that it was mandatory
for every person in the village to participate (since they then had to sell their production to the
monopoly at state-fixed prices). The interviews also reveal small differences within villages
in the area attributed to single or married men, and don’t mention the fate of divorced
women or widows. Finally, it reveals heterogeneity between villages in the total area that was
distributed 22

22
Example of comments from village leaders about the forced cultivation of cotton and the distribution of
land : ”Selon l’enquêté tous les ménages pour chaque clan ont participé à la culture du coton car ils étaient
forcés et personne ne pouvait résister” ; ”Pour la culture du coton à cette époque, tout le monde devrait le
faire c’était obligatoire et la répartition des terres était obligatoire.” ;”Pendant l’époque des belges, ils étaient
forcés par les autorités belges de quitter et de s’installer ici pour cultiver le coton et d’être à la proximité de
la route. A cette époque tous les clans ont participé à la culture du coton, la distribution des terres était
définitive, tous les clans ont bénéficié de cette distribution et après l’arrêt de la culture de coton ces terres sont
restées propriété des familles qui les avaient recues.”

27
4.3 Baseline characteristics and balance tests

To further investigate how villages differed in terms of observable characteristics before Inde-
pendence, we use all the Pre-Independence characteristics we have about the villages in our
sample and conduct simple balance tests. Table 16 confirms that paysannat villages don’t
seem to have had different characteristics. We also put the villages on a map of the regrion
to examine their geographic distribution.

First, they had the same number of clans at foundation, which could indicate there is not
difference in population pressure at the time of colonization. Second, there is no statistically
significant difference in the distance to the nearest commercial town nearest river, the distance
to the nearest port along the roads (the location of ports has not changed since Independance).
All of these variables give an idea of access to markets, whose proximity could mean more
pressure on the land. Finally, we compare distances to the administrative capitals to which
the village belongs. Sometimes, a village can be very close to a commercial town but because
of the administrative borders he depends on an administrative capital that is far away. We
find that distance to Secteur and Territoire capitals are not statistically different (again, the
local of administrative capitals does not seem to have changed since Independence). All the
villages in our sample are located on a dirt road that connects these administrative towns.
However, they seem to differ in terms of distance to the district capital. This could indicate
that territorial agents tried to chose villages that were closer to the District Capital so they
could show their progress to their superiors more easily when they would visit. We don’t
think this is evidence of any difference in terms of access to markets. In the rest of the paper,
we will control for this variable to account for this difference.

Maps 8 9 and 10 show the geographic distribution of paysannat villages. From the maps, it
is clear that paysannat villages are present in all vegetation types in the region, and most
secteurs have both paysannat as well as non paysannat in close distance. The (dirt) road
network in the region is very limited and in a bad state but all villages in our sample are
along a road and paysannat villages don’t look obviously advantaged nor disadvantaged.

4.4 Empirical Strategy

First, we run a simple OLS comparing village level outcomes in villages which grew cash
crops (in this region, mainly cotton and rice) and experienced a land distribution during col-
onization to villages who, because of historical reasons and the slow down in the expansion
of the policy didn’t take part in land distribution. Those regressions include a fixed effect
at the administrative Secteur level to account for differences in geographic and colonial ad-
ministrator’s and indigenous chiefs characteristics, as well as a set of controls for geographic

28
characteristics that have not changed over time. For this analysis, the critical hypothesis is
that paysannat villages were not chosen for having higher levels of privatization.

Second, we investigate the heterogeneity of the impact between households from different
clanic groups in the villages. In this part, a new exogeneity issue will needs to be considered:
since we are going to compare households from different clanic groups within a secteur, a key
assumption is that the different types of clanic groups are comparable in paysannat and non
paysannat villages for a given period of arrival of the family in the village. We argue this
assumption is not problematic.

Finally, to test the robustness of our results and to control for possible endogeneity in the
choice of the villages for the agricultural policy, we follow a nearest neighbor’s approach:
we construct pairs of villages which are geographical neighbors across administrative borders
of a territory, but which happened to have different agricultural policies (with or without a
policy to push toward the individualization of land rights). On all our observed geographical
variables, these villages look very similar, and we assume that such neighboring villages would
be similar in other unobservable variables as well.

5 Empirical Results

The aim of this section is to study the differences in contemporary land institutions and
inequalities between villages with and without Paysannat. First, we show that paysannat
villages have less ”commons”, or collective forest, so in that sense paysannat succeeded in
pushing toward more privatization. At the same time, it also seems to be more difficult
for migrants from the village to come back and get access to land . Older generations of
strangers in the village also seem less integrated in the local institutions. Second, we examine
the distributional consequences of this privatization of the commons by the Belgian Colony.
We establish that there are strong inequalities in access to land between the different clanic
groups within a village, and we give evidence that part of these differences can be explained
by the presence of Paysannat.

5.1 Results: Impact at the village level

Table 17 to 21 report the results of the OLS regressions at the village level. In all regressions,
Secteur fixed effects and geographic control variables are included.

29
5.1.1 Impact on collective forests

Table 17 shows the impact of paysannat on village level outcomes related to natural resources.
The coefficient on a dummy indicating whether there exists land in primary forest that belongs
to the village and is collective (column 1) is negative but not significant. However, the nearest
collective primary forest is almost 2 km further away in paysannat villages compared to non
paysannat villages (column 2), and villages leaders are 15% more likely to consider that the
shortage of forest is an issue in the present, a strong and significant difference (column 3).
They are also more likely to be concerned that there will be shortages of forest in the near
future, but the difference is not significant (column 4). These results seem to indicate (as will
be further confirmed in the next section) that paysannat succeeded in pushing toward more
privatization of the forests near the village centers. Even if a large share of villages still report
there is some forest in their villages, there is also a high level of awareness that the frontier is
close, and the forests are relatively far from the village dwellings (4.9 kilometers on average).
One key question is therefore whether this is the mechanical consequence of the division of
the land that occured during colonization, or whether this is the result of individualization
processes or increased pressure from migration that occured later.

5.1.2 Impact on the clanic composition of the village and population

In tables 17 to 21 we show the results of regressions on the clanic composition of the villages to
investigate whether paysannat led to differentiated patterns of migration and settlement. We
do not find any differences in the share of marginal clans 23 , local clans, or foundation clans
between paysannat villages and others. One category of clan is less represented in paysannat
villages : households heads whose wife is from a founding clan in the village. This could
suggest that it is harder for those women to come back to their village with their husband
(the village of their father) and get access to land for their family. In tables 20 and 21, we
differenciate the households by the period of arrival in the village of their own family. We
call ”newcomers” households whose family arrived in a village this generation or the previous
one (for example their brother, their father, or their uncle moved to the village). We call
households whose families have been in the village for more that two generations ”oldcomers”.
We find that there are on average 4% less newcomers from local clans and 5% more marginal
newcomers in paysannat villages. The newcomers for local clans are households who have
family in the village but had migrated elsewhere or come to join a member of their family.
One interpretation of this result could be that it is more difficult for family members of local
clans to get access to land in the village because there is less collective land. Moreover, the
fact that there are more marginal oldcomers (households whose family has been in the village
23
Marginal clans : remember that is how we decided to call clans who are not recognized by village leaders
but are present in the village

30
for 2 or more generations but are not recognized by village leaders) could mean either that
there was more migration in the past, or that the migrants following independence were less
likely to have been formerly recognized by their host village. Since there are not more local
oldcomers, the second interpretation seems more likely. Overall, these results suggest that
paysannat succeeded in privatizing the commons but also seems to have led to tighter land
ownership pattern, in the sense that it secured access for local clans but made it more difficult
for others.

5.2 Results at the household level

Table 22 to ?? report the results of OLS regressions at the household level. In all regressions,
Secteur fixed effects and geographic control variables are included, and the reference category
are the households from Founding Clans.

5.2.1 Characteristics of the fields at the household level

Tables 22 and 23 shows the difference in ownership of Paysannat land and documented land
by the households depending on the clan they belong to, as well as insecurity about land
rights. We find strong differences in ownership of these types of land between marginal
clans and founding clans, as well as between households where the wife is from a founding
clan. The difference with other local clans is negative but not significant. Unsuprisingly,
village founders have benefited more from paysannat, and this advantage has been passed
on to the next generations. As evidenced by the first column, households from marginal
clans are much less likely to belong to a family who owns land in Paysannat. They are
also more likely to own land in fallow that has a document, have a higher share of plots
with a document cultivated in 2014, and are more likely to cultivate at least one plot with
a document in 2014. Finally, both local clans, whether founders or simply locals, have lower
levels of insecurity over their fallows and fields in cultivation, as evidenced by columns 5 and
6. By differentiating by period of arrival in the village, table 23 reveals that these results
are driven primarily by newcomers for marginal clans, even if insecurity on the fallows seems
to be slightly higher among oldcomer marginals. It also reveals that local clans who came
back recently experience a similar situation : they are more likely to have documents on their
fields or to own paysannat, but they express much more confidence over their land rights
than marginal clans. The fact that newcomers own more documents is likely to reflect the
fact that they have to negociate with individual farmers or with village chiefs to gain access
to arable land. It probably becomes more and more common to require a written document
as pressure on the land increases, and its more common to seek a document when you don’t
have customary rights on the land you obtain. It can also reflect the fact that newcomers

31
seek more protection from village chiefs or other administrative officials to prevent disputes
as they can not rely on their clan to protect their rights.

Tables 24 and 25 show that there is a strong persistence of ownership of Paysannat rights over
time and that paysannat explains a large part of the differences in documentation. Local clans
are the clanic groups who still benefit from Paysannat land the most: in villages where oral
history reports occurence of Paysannat related land distributions, 23% more households from
local clans report owning land in Paysannat compared to non-founding local clans in other
villages. Marginal clans are only 10% more likely to have land in Paysannat in treatment
villages compared to non treatment villages. Founders are almost 20% more likely to report
owning paysannat in treated villages, but they are also 8% more likely than marginal clans
to do so even when their villages was not treated. This result reflects the fact that land
distributions for plantations of spontanous land distributions after cotton cultivation also took
place in non paysannat villages as described in Section 2. When a block of land was reserved
for cotton, even when it was not attributed to specific households during colonization, there
are many cases where the farmers decided to share the land between themselves after the end
of cotton cultivation. This table also suggests a very different pattern of land documentation
in villages with and without Paysannat : in non-treated villages, local clans (whether founding
or not) are less likely than marginal clans to have documents on their fallows or on the fields
they cultivate. Founding clans, on the other hand, are more likely to do so. This suggests
that it is more difficult for marginal clans to access and secure land in treated villages. This is
further evidenced by column 5: marginal clans are 7.6% more likely of feeling insecure about
some of their fallows in treated villages compared to marginal clans in non-treated villages,
while local clans are not. Table 25 shows that this pattern is true for both marginal clans
who settled in the village a long time ago and those who arrived recently. These result further
point to more privatization and tigher land claims in paysannat villages.

5.2.2 Characteristics of the land portfolio, household level

Tables 26 to 29 show the results of a simple regression on the land portfolio of the households
in our simple (land cultivated or owned by households). Table 26 shows that marginal clans
have on average 2.4 less fallow plots than founding and local clans, and that own on average
2.4 less fallow plots than founding clans and 2 plots less than local clans. The total area
they cultivated in 2014 is also 0.3 hectares smaller for marginal clans compared to founders.
Table 26 also reveals that households who live with the clan of the wife of the households
head own less fallow plots. Table 29 shows that this pattern is the same for marginal clans
irrespective of the period of arrival of their family in the village. There is also no statistically
significant difference between newcomers from local clans and oldcomers from local clans. It
confirms that access to land seems to be much more difficult for clans that are not recognized

32
by village institutions, and for women, while local founders who have stayed in the village are
those who have the largest land portfolio. Interestingly, households who settle in the village
of the wife seem to have a large share of their plots in the primary forest, suggesting that
they have access not their family land but rather to collective forest.

Tables 28 and 29 show that paysannat led to smaller number of plots cultivated per housheold,
a smaller area cultivated, and a smaller number of fallows for most groups. The two exceptions
are the oldcomers from marginal or local clans, whose situation is comparable in paysannat
and non paysannat villages. All other groups have less fallows, cultivate less plots and smaller
areas in paysannat than non paysannat villages.

5.2.3 Distributional impact

We expect that the distribution of collective land, if it leads to a reallocation of the land
between the local clans, can lead to a lesser level of inequalities in the village which may end
up benefit everyone by promotting better management of the resources (Baland et al., 2009).
However, our results suggest that, while the differences between local clans may be lower in
paysannat villages, land rights also seem to be tighter and less flexible, and it seems more
difficult for strangers to access land. Overall, our results point toward more pressure on the
resources.

33
6 Conclusion

This paper studies one specific colonial intervention, Paysannat, which impacts the customary
land tenure system. It shows that this led translated to sustained land privatization, and had
strong distributional implications?. This papers uses a first hand dataset with household and
village oral history data to investigate this question, and takes advantage of the arbitrary
geographical distribution of those interventions. It suggests that land rights are very persis-
tent over time and that Paysannat had a lasting impact on land access depending on clanic
affiliation. In future work, I will explore explore whether there is evidence that they led to
differences in the sustainability of the management of natural resources and the regulation of
conflicts.

34
References

Acemoglu, Daron, Simon Johnson, and James A. Robinson, “The Colonial Origins
of Comparative Development: An Empirical Investigation,” June 2000.

, , and , “Reversal of Fortune: Geography and Institutions in the Making of the

Modern World Income Distribution,” The Quarterly Journal of Economics, November 2002,
117 (4), 1231–1294.

Baland, J.-M. and J.-P. Platteau, “Division of the Commons: A Partial Assessment
of the New Institutional Economics of Land Rights,” American Journal of Agricultural
Economics, August 1998, 80 (3), 644–650.

Baland, Jean-Marie and Jean-Philippe Platteau, Halting Degradation of Natural Re-

sources: Is There a Role for Rural Communities?, Food & Agriculture Org., 1996. Google-
Books-ID: JSMOllcGHxIC.

, Kjetil Bjorvatn, and others, “On the distributive impact of privatizing the commons:
the case of renewable resources,” Technical Report, Citeseer 2009.

Banerjee, Abhijit and Esther Duflo, “Under the Thumb of History? Political Institutions
and the Scope for Action,” Working Paper 19848, National Bureau of Economic Research
January 2014.

and Lakshmi Iyer, “History, Institutions, and Economic Performance: The Legacy of
Colonial Land Tenure Systems in India,” American Economic Review, September 2005, 95
(4), 1190–1213.

Boelaert, E, “L’Etat indépendant et les terres indigènes,” 1956.

Boserup, Ester, The conditions of agricultural growth: The economics of agrarian change
under population pressure, Transaction Publishers, 1965.

, “The impact of scarcity and plenty on development,” The Journal of Interdisciplinary

History, 1983, 14 (2), 383–407. 00044.

Clerck, Louis De, “L’administration coloniale belge sur le terrain au Congo (1908-1960)
et au Ruanda-Urundi (1925-1962),” in “Annuaire d’Histoire administrative européenne,”
Vol. 18 2006.

Cogneau, Denis and Yannick Dupraz, “Institutions historiques et développement

économique en Afrique,” Histoire & mesure, June 2015, XXX (1), 103–134.

Dell, Melissa, “The mining mita, explaining institutional persistence,” Mimeogr. MIT, 2008.

35
EIC, “Bulletin Officiel,” Technical Report 152, Etat Indépendant du Congo 1905.

Engerman, Stanley L. and Kenneth L. Sokoloff, “Factor Endowments: Institutions,

and Differential Paths of Growth Among New World Economies: A View from Economic
Historians of the United States,” Working Paper 66, National Bureau of Economic Research
December 1994.

Goldstein, Markus and Christopher Udry, “The Profits of Power: Land Rights and
Agricultural Investment in Ghana,” Journal of Political Economy, December 2008, 116
(6), 981–1022.

, Kenneth Houngbedji, Florence Kondylis, Michael O’Sullivan, and Harris

Selod, “Formalizing Rural Land Rights in West Africa,” Development Research, 2015.

Hansen, M. C., P. V. Potapov, R. Moore, M. Hancher, S. A. Turubanova,

A. Tyukavina, D. Thau, S. V. Stehman, S. J. Goetz, T. R. Loveland, A. Kom-
mareddy, A. Egorov, L. Chini, C. O. Justice, and J. R. G. Townshend, “High-
Resolution Global Maps of 21st-Century Forest Cover Change,” Science, November 2013,
342 (6160), 850–853.

Herdeschee, Johannes, Kai-Alexander Kaiser, and Daniel Mukoko Samba, Re-

silience d’un Geant Africain: Accelerer la Croissance et Promouvoir l’emploi en Republique
Democratique du Congo, Volume I Synthese, contexte historique et macroeconomique, Kin-
shasa: MEDIASPAUL, 2012.

Huillery, Elise, “History Matters: The Long-Term Impact of Colonial Public Investments
in French West Africa,” American Economic Journal: Applied Economics, March 2009, 1
(2), 176–215.

, “The Impact of European Settlement within French West Africa: Did Pre-colonial Pros-
perous Areas Fall Behind?,” Journal of African Economies, March 2011, 20 (2), 263–311.

Iyer, Lakshmi, “Direct versus Indirect Colonial Rule in India: Long-Term Consequences,”
Review of Economics and Statistics, June 2010, 92 (4), 693–713.

Likaka, Osumaka, Rural society and cotton in colonial Zaire, Madison, Wis: University of
Wisconsin Press, 1997.

, Naming colonialism: history and collective memory in the Congo, 1870-1960 Africa and
the diaspora: history, politics, culture, Madison, Wis: University of Wisconsin Press, 2009.

Mataboro, Severin Mugangu, “Revue historique du secteur foncier en République

Démocratique du Congo,” Technical Report 2016.

36
Nunn, Nathan, “The Importance of History for Economic Development,” Annual Review
of Economics, September 2009, 1 (1), 65–92.

Ostrom, Elinor, “Collective action and the evolution of social norms,” Journal of Natural
Resources Policy Research, October 2014, 6 (4), 235–252.

Piet, Clément, “The land tenure system in the Congo, 1885-1960. Actors, motivations and
consequences,” in “Colonial Exploitation and Economic Development, The Belgian Congo
and the Netherlands Indies Compared,” frankema, ewout and frans buelens ed., London:
Routledge, 2013, pp. 88–108.

, “Agricultural development in the Belgian Congo: the origins, implementation and impact
of the indigenous peasantry scheme, 1920-1960,” August 2015.

Platteau, Jean-Philippe, “The Evolutionary Theory of Land Rights as Applied to Sub-

Saharan Africa: A Critical Assessment,” Development and Change, January 1996, 27 (1),
29–86.

Porta, Rafael La, Florencio Lopez-De-Silanes, Andrei Shleifer, and Robert W.

Vishny, “Legal Determinants of External Finance,” The Journal of Finance, 1997, 52 (3),
1131–1150.

Salacuse, Jeswald W., The National Land Law System of Zaire, Land Tenure Center,
University of Wisconsin-Madison, 1985.

Vansina, Jan, Paths in the rainforests: toward a history of political tradition in equatorial
Africa, Madison, Wis: University of Wisconsin Press, 1990.

Vinck, Honoré, Conflits fonciers au Congo belge opinions congolaises: texte intégral d’une
enquête de 1954, Bruxelles: Académie royale des sciences d’outre-mer, 2011. OCLC:
949794234.

Weitzman, Martin L, “Free access vs private ownership as alternative systems for managing
common property,” Journal of Economic Theory, June 1974, 8 (2), 225–234.

WorldBank/UN-Habitat, “Revue du Secteur Foncier en République Démocratique du

Congo,” Technical Report 2016.

WWF, “Etude sur l’identification et la localisation des anciennes plantations industrielles

de palmier à huile inactives ou abandonnées dans les Provinces de Bas-Congo, Bandundu,
Equateur et Orientale,” Technical Report, République Démocratique du Congo June 2015.

37
Tables, maps and Summary Statistics

Maps

38
 Figure 2: Map of DRC Provinces in 2015

Figure 3: Map : Administrative borders of the Secteurs and Territoires

in the sample
Legend
"
) Ports
Navigable rivers
Administrative borders of the Secteurs
DRC_Collectivite_wgs Bandas
Republique centrafricaine
Country borders

Bili

Soa

Deni
ZONGO
"
)
Bosobolo Gaya
Libenge-Nord

Kasa
Otto-Banza Mobayi-Mbongo
Duaru

Yakoma

Libenge-Centre Bodangabo Wapinda

Nguya Mobenge-Mondila

Abumombazi

Avuru-Gatanga
Mbari
BUSINGA
"
) Mongwandi
Libenge-Sud Businga
MOGALO Karawa Avuru-Duma
République du Congo "
) Banga-Kungu POPOLO
"
)
Bowase

Lua Gbandi
Yandongi
LIKIMI
Banza "
) Mongala-Motima
DONGO
"
) Songo
Ngombe-Doko de Lekimi Bondongola
Bolingo
Banda-Yowa

Bomboma Mongala-Motima
Loeka
AKULA Monzamboli
Dongo "
)
Itimbiri
Ngombe-Doko Molua
Ndolo-Liboko
BUMBA
LISALA "
)
Yamandundu
Ngombe-Mombangi "
)
Mwanda

Mobango Itimbiri
Boso-Simba

Ngiri Mweko
Mombesa
Djamba Boso-Melo Boso-Djanoa
Yaliwasa
Bangala
Gombalo Bongandanga Bongandanga Yala

39
 Figure 4: Map : Ethnic groups in the sample

0 15 30 60 90 120 Legend
Kilometers

ngwaka
ngbandi
ngombe
budja
mbandja
lobala
Bosobolo monzombo
mono
fulu
Bondo
Gbadolite swahili
Zongo
mundunga
Mobayi-Mbongo autre2
bomboma
Libenge yakpa
lingonda
tembo
Yakoma
langbase
ngugbu
Gemena Businga togbo
gobu
bamwe
Aketi
autre
pagibete
kunda

Budjala Bumba
Kungu
Lisala

Bongandanga Yahuma Basoko

Bomongo
Bomongo Makanza

Figure 5: Map : Forest Cover in the Equateur Province

40
 Figure 6: Map : Cotton cultivation during Belgian colonization

Figure 7: Map : Cotton cultivation during Belgian colonization and for-

est cover

41
Figure 8: Map : Definitive land distributions during Belgian colonization
and administrative borders

Legend
"
) Capital of the Secteur
No definitive land distribution
Definitive land distribution "
)
"
) Ports Republique centrafricaine

Navigable rivers
Administrative borders of the Secteurs
"
)
Country borders
ZONGO
"
)
"
) "
)
"
)
"
)
"
)

"
)
"
)
"
)

"
)
BUSINGA
"
) "
) "
)
"
)

MOGALO
République du Congo "
) "
) POPOLO
"
)

"
) "
) "
)
LIKIMI
"
)
"
) "
)
DONGO "
) "
)
"
)
"
)
"
)
"
)
"
) "
)
"
) AKULA
"
)
"
) "
)

"
) " "
) "
)
) BUMBA
"
) LISALA "
)
"
)

Figure 9: Map : Definitive land distributions during Belgian colonization

and roads
Legend
No definitive land distribution
Definitive land distribution
"
) Ports
National
Republique centrafricaine
Provincial, prioritary
Provincial, secondary
Local
Navigable rivers
ZONGO
"
)
Equateur
Country borders

BUSINGA
"
)

MOGALO
République du Congo "
) POPOLO
"
)
Equateur

LIKIMI
"
)
DONGO
"
)

AKULA
"
)

BUMBA
LISALA "
)
"
)

42
 Figure 10: Map : Definitive land distributions during Belgian coloniza-
tion and forest cover
Legend
No definitive land distribution
Definitive land distribution
"
) Ports
National Republique centrafricaine
Provincial, prioritary
Provincial, secondary
Local
Navigable rivers
ZONGO
"
)
Hansen Forest Cover Equateur
Value
0 - 14
14 - 41
41 - 69
69- 91
91 - 100
Equateur
BUSINGA
Country borders "
)

MOGALO
République du Congo "
) POPOLO
"
)
Equateur

LIKIMI
"
)
DONGO
"
)

AKULA
"
)

BUMBA
LISALA "
)
"
)

Summary Statistics

Table 1: Creation of the village

Freq Pct
Before colonization 138 71.88
During colonization 42 21.88
Mobutu 8 4.17
Between independence and Mobutu 4 2.08
Total 192 100.00

43
Table 2: Summary Statistics : demographic and geographic characteristics of the villages in 2014

Mean SD 25th Median 75th Min Max N

Population
Number of households 307 357 120 120 350 13 2500 174
Number of individuals 2785 3067 919 919 3432 160 18000 179
Number of men 755 991 200 200 855 41 6800 175
Number of women 999 1219 270 270 1250 40 7000 175
Number of children 1048 1233 318 318 1304 45 7700 175

Ethnic composition
Number of clans 5.62 3.07 4.00 4.00 7.00 1.00 15.00 177
Share of the largest clan 0.35 0.18 0.25 0.25 0.40 0.10 1.00 175
Number of ethnic groups 1.25 0.69 1.00 1.00 1.00 1.00 5.00 174
Share main ethnic group 0.96 0.14 1.00 1.00 1.00 0.00 1.00 174

Geographic characteristics
Forest cover (km sq) 54 58 16 16 66 0.33 306 173
Nearest river (km) 65 47 27 27 94 0.04 194 178
Nearest port (km) 96 63 47 47 128 2.96 256 178
Capital of the district (km) 117 63 69 69 164 0.26 287 178
Capital of the Territoire (km) 58 40 27 27 77 1.54 191 178
Capital of the Secteur (km) 25 21 7 7 38 0.00 92 178

Note: Distances are based on computations made by the authors using the geographic lo-
cation of the villages and capital in ArcGIS to compute the nearest path along the road
network. Forest cover is calculated based on Landsat images from Hansen Global Forest
Watch dataset and village borders drawn by the author based on village surveys.

44
 Table 3: Summary Statistics : clanic groups

Other clan Foundation clan Total

Own clan/brother identified in the village 18.3 53.5 71.9
Clan of the wife in the village 1.6 2.7 4.3
Clan identified in a different village 9.3 0.0 9.3
Clan could not be identified/unknown 14.6 0.0 14.6
Total 43.8 56.2 100.0

Table 4: Summary Statistics : clanic groups

Oldcomer Newcomer Total

Own clan/brother identified in the village 65.4 6.6 72.0
Clan of the wife in the village 2.4 1.8 4.2
Clan identified in a different village 5.7 3.6 9.3
Clan could not be identified/unknown 8.1 6.4 14.5
Total 81.6 18.4 100.0

45
 Table 5: Summary Statistics : characteristics of the land

All By Clanic Group

Mean Founding Men’s Women’s Marginal

Land Formalization
Family owns Paysannat 3413 0.19 0.23 0.19 0.08 0.12
Owns at least one documented field 2309 0.12 0.11 0.10 0.10 0.14
Documented field in 2014 3466 0.13 0.12 0.14 0.13 0.15
% fields in 2014 with a document 3466 0.07 0.06 0.08 0.07 0.10
Worried about losing a field 2306 0.07 0.06 0.05 0.10 0.10

Fields cultivated in 2014

Number of plots cultivated in 2014 3479 4.31 4.43 4.36 3.97 4.09
Total area cultivated in 2014 3459 2.75 2.80 2.81 2.60 2.65
% fields from long fallow in 2014 3340 0.41 0.41 0.43 0.40 0.39
% fields from short fallow 3340 0.40 0.40 0.38 0.40 0.41
% Conversion primary forest in 2014 3476 0.04 0.05 0.04 0.04 0.04
Dummy conversion primary forest in 2014 3476 0.15 0.16 0.15 0.16 0.14

Fields cultivated in 2014

Number of plots cultivated in 2014 2676 4.34 4.50 4.41 4.04 4.01
Total area cultivated in 2014 2657 2.78 2.84 2.95 2.75 2.57
sh conversion lastyear a14 2675 0.17 0.17 0.18 0.15 0.17
sh forest lastyear a14 2675 0.17 0.17 0.18 0.15 0.17
sh cultivated a14 2674 0.53 0.53 0.54 0.52 0.52
Dummy conversion primary forest in 2014 2675 0.15 0.15 0.15 0.14 0.14

Fields cultivated in 2014

Number of plots cultivated in 2014 2676 4.34 4.50 4.41 4.04 4.01
Total area cultivated in 2014 2657 2.78 2.84 2.95 2.75 2.57
sh conversion lastyear a14 2675 0.17 0.17 0.18 0.15 0.17
% Conversion from forest in 2014 2675 0.17 0.17 0.18 0.15 0.17
% Already cultivated in 2014 2674 0.53 0.53 0.54 0.52 0.52
Dummy conversion primary forest in 2014 2675 0.15 0.15 0.15 0.14 0.14

Fields cultivated in 2014

Number of plots cultivated in 2014 2676 4.34 4.50 4.41 4.04 4.01
Total area cultivated in 2014 2657 2.78 2.84 2.95 2.75 2.57
sh conversion lastyear a14 462675 0.17 0.17 0.18 0.15 0.17
% Conversion from forest in 2014 2675 0.17 0.17 0.18 0.15 0.17
% Already cultivated in 2014 2674 0.53 0.53 0.54 0.52 0.52
Dummy conversion primary forest in 2014 2675 0.15 0.15 0.15 0.14 0.14
Table 6: Summary Statistics : Place of issuance of the land documents in the village for fields culti-
vated in 2014

Freq Pct
Groupement Chief 536 43.47
Secteur Chief 147 11.92
Paysannat title 91 7.38
Certificate of sale 459 37.23
Total 1233 100.00

Table 7: Summary Statistics : Who is the land chief for your household ?

Freq Pct
No land chief 49 25.65
Household head 38 19.90
The father of the Household Head 31 16.23
The uncle of the Household Head 29 15.18
The elder brother of the Household Head 18 9.42
The grandfather of the Household Head 12 6.28
Other 6 3.14
A relative of the wife 4 2.09
Nephew of the Household Head 3 1.57
A Cousin or brother of the Household Head 1 0.52
Total 191 100.00

47
 Table 8: Summary Statistics : knowledge and access to collective land

All By Clanic Group

Mean Founding Men’s Women’s Marginal

Collective primary forest or savanna

Declares this type of land exists 2645 0.62 0.62 0.62 0.55 0.61
Authorization to open a field (free) 2645 0.55 0.58 0.54 0.44 0.50
Has already opened a field there 2566 0.26 0.27 0.27 0.22 0.23

Collective secondary forest or old plantation

Declares this type of land exists 2645 0.12 0.12 0.12 0.11 0.13
Authorization to open a field without paying 2641 0.07 0.08 0.08 0.07 0.07
Has already opened a field there 2581 0.03 0.03 0.03 0.04 0.04

primary forest or savanna, other village

Declares this type of land exists 2638 0.07 0.07 0.06 0.09 0.07
Authorization to open a field (free) 2637 0.03 0.03 0.03 0.06 0.03
Has already opened a field there 2619 0.02 0.01 0.02 0.02 0.02

Collective primary forest or savanna, clan

Declares this type of land exists 2645 0.20 0.22 0.20 0.16 0.16
Authorization to open a field (free) 2645 0.19 0.20 0.19 0.15 0.15
Has already opened a field there 2618 0.10 0.11 0.11 0.06 0.09

Collective family land, uncultivated

Declares this type of land exists 2645 0.23 0.25 0.23 0.19 0.19
Authorization to open a field 2643 0.21 0.23 0.22 0.19 0.18
Has already opened a field there 2610 0.14 0.15 0.17 0.11 0.12

Land that belongs to the family

Declares this type of land exists 2642 0.55 0.62 0.53 0.40 0.44
This land can not be individualized 1446 0.96 0.96 0.96 0.98 0.96

48
 Table 9: Summary Statistics : Access to collective village primary forest or savana

All By Clanic Group

Mean Founding Men’s Women’s Marginal

Reason for not opening a field

Physically too demanding 2646 0.00 0.00 0.00 0.00 0.01
Too far 2646 0.01 0.00 0.01 0.02 0.01
Enough Family land 2646 0.01 0.01 0.01 0.03 0.02
Protect family land 2646 0.00 0.00 0.00 0.00 0.00
No money to pay for labour 2646 0.00 0.00 0.00 0.00 0.00
No money to pay authorization 2646 0.01 0.00 0.00 0.00 0.02
Other 2646 0.00 0.00 0.00 0.00 0.01

Reason for opening a field

Land is more fertile 2645 0.13 0.14 0.13 0.12 0.13
Increase arable family land 2645 0.07 0.08 0.06 0.05 0.06
Increase duration of fallows 2645 0.02 0.02 0.02 0.03 0.02
Ensure enough land for children 2645 0.05 0.06 0.07 0.03 0.02
Mimic ancestors 2645 0.01 0.01 0.01 0.00 0.01
Reduce risks of divagation 2645 0.00 0.00 0.00 0.02 0.01
Be closer to campement for fishing/hunting 2645 0.01 0.00 0.01 0.00 0.00
Some crops need this type of land 2645 0.08 0.08 0.09 0.08 0.06
Other 2645 0.01 0.01 0.01 0.02 0.01

Transmission of fields opened there

Transfer to children 664 0.86 0.86 0.90 0.83 0.86
Lend 664 0.79 0.79 0.86 0.74 0.76
Rent 664 0.72 0.74 0.78 0.70 0.64
Sell 662 0.35 0.35 0.34 0.52 0.34
Only autochtones can do it 2645 0.01 0.00 0.00 0.01 0.02
Only some clans can do it 2645 0.00 0.00 0.00 0.01 0.00
Only possible when chief decides 2645 0.03 0.03 0.03 0.02 0.03
Other 2645 0.01 0.01 0.01 0.00 0.01

49
 Table 10: Summary Statistics : Who took the decision to give this land to your clan

Freq Pct
Appropriation by an ancestor 26 72.22
Families 5 13.89
Kapita 2 5.56
Notables 2 5.56
Other 1 2.78
Total 36 100.00

Table 11: Summary Statistics : When was the decision to give this land to your clan

Freq Pct
During Colonization 16 44.44
Before colonization 8 22.22
Independence-Mobutu 6 16.67
Mobutu 5 13.89
After Mobutu 1 2.78
Total 36 100.00

50
Table 12: Summary Statistics : Summary Statistics : Transmission of land from one generation to the other

All By Clanic Group

Mean Founding Men’s Women’s Marginal

Transmission of land to children

Already given some for cultivation 2643 0.20 0.21 0.20 0.12 0.19
Already lent some for cultivation 2643 0.16 0.15 0.18 0.13 0.16
Already given some for futur cultivation 2643 0.09 0.10 0.10 0.05 0.06
Decided after-death sharing of land 2621 0.14 0.15 0.15 0.07 0.13

Children who already got some land

All boys and girls 2643 0.02 0.02 0.02 0.01 0.02
Only the eldest boy 2643 0.10 0.11 0.09 0.06 0.10
All boys 2643 0.06 0.06 0.07 0.04 0.07
Part of the boys 2643 0.12 0.11 0.14 0.08 0.11
Some of the boys and girls 2643 0.03 0.03 0.03 0.04 0.02
All the girls 2643 0.00 0.00 0.00 0.00 0.00
Part of the girls 2643 0.01 0.01 0.01 0.01 0.01

51
 Table 13: Summary Statistics : Land distribution during Colonization

Freq Pct
Definitive Distribution, progressive 63 35.00
Definitive distribution, one 41 22.78
Temporary allocation, progressive 37 20.56
Temporary allocation, one 23 12.78
No Distribution 14 7.78
No cash crops 2 1.11
Total 180 100.00

52
Table 14: Summary Statistics : Mean of obtention of the fields in paysannat

Freq Pct
Inherited 483 35.83
Clearing of a family field 294 21.81
Confiage, family 173 12.83
Clearing 113 8.38
Confiage, other 96 7.12
Bought 55 4.08
Other 54 4.01
Rented 45 3.34
Confiage, wife’s family 19 1.41
Given by the wife’s family 16 1.19
Total 1348 100.00

53
 Table 15: Summary Statistics : Forest cover and loss from satellite data

Mean SD 25th Median 75th Min Max N

Area in 2000 (km)

Primary Forest 6.75 9.71 0.50 0.50 8.93 0.00 57.08 173.00
Secondary Forest 2.88 2.85 0.84 0.84 3.98 0.00 20.74 173.00

Area Deforested 2000 14 (km)

Primary Forest -8.63 10.07 -12.52 -12.52 -1.70 -67.25 0.00 171.00
Secondary Forest -20.05 11.48 -28.07 -28.07 -11.36 -55.68 0.00 172.00

54
Balance Test of pre-Independance Characteristics

Table 16: Balance tests : Characteristics of the villages before independence

N Paysannat No Paysannat Diff No P - P S.E.

Preindependence characteristics
Number of clans at foundation 179 3.7 3.8 0.11 0.38
Distance to nearest commercial town (km) 178 75.9 84.8 8.89 8.63
Nearest river (km) 178 68.0 61.8 -6.25 7.14
Nearest port (km) 178 94.0 98.7 4.73 9.50
Capital of the Secteur (km) 178 24.6 25.6 1.01 3.22
Capital of the Territoire (km) 178 55.0 62.6 7.64 6.03
Capital of the district (km) 178 109.1 126.9 17.71* 9.45

Source: The distances are computed with Arc GIS based on the geograhic information collected in the
village surveys as well as road data provided my the Ministry of Infrastructure. The distances to the
ports and administrative capitals are computed using the shortest path alon the roads. The distance
to the river is computed using the direct distance to the rivers irrespective of roads.
* p < 0.1, ** p < 0.05, *** p < 0.01.

55
Regressions at the Village Level

Tables : Collective land in the village

Table 17: Collective Primary forest, OLS with secteur fixed effect

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

Shortage Forest Shortage
PF Dist
now futur
Paysannat –0.018 1.961** 0.157** 0.110
(0.080) (0.978) (0.076) (0.082)
Geo Control Vars Yes Yes Yes Yes

Observations 176 120 153 151

Mean 0.70 4.90 0.14 0.23

* p < 0.1, ** p < 0.05, *** p < 0.01.

Tables : Forest Cover in the Village

Table 18: Forest Cover in the village in 2000, OLS with secteur fixed effect

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

Deforestation Deforestation
PF SF
PF SF
Paysannat –1.305 0.446 –1.115 0.061
(1.506) (0.456) (1.403) (1.104)
Geo Control Vars Yes Yes Yes Yes

Observations 171 171 169 170

Mean 6.79 2.56 -7.87 -20.31

* p < 0.1, ** p < 0.05, *** p < 0.01.

Tables : Characteristics of the clans in the village

56
 Table 19: Composition of the clans in the village, OLS with secteur fixed effect

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

Local Founding Local Founding
Marginal
Men Men Women Women
Paysannat 0.036 –0.039 0.010 0.010 –0.018*
(0.038) (0.043) (0.052) (0.007) (0.010)
Geo Control Vars Yes Yes Yes Yes Yes

Observations 176 176 176 176 176

Mean 0.24 0.17 0.55 0.01 0.03

* p < 0.1, ** p < 0.05, *** p < 0.01.

Table 20: Composition of the clans in the village, OLS with secteur fixed effect

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

Newcomer Newcomer Oldcomer Oldcomer
Local Marginal Local Marginal
Paysannat –0.041* –0.015 0.006 0.050*
(0.021) (0.024) (0.042) (0.026)
Geo Control Vars Yes Yes Yes Yes

Observations 176 176 176 176

Mean 0.08 0.10 0.68 0.14

* p < 0.1, ** p < 0.05, *** p < 0.01.

57
 Table 21: Composition of the clans in the village, OLS with secteur fixed effect

(1) (2) (3) (4) (5) (6) (7) (8)

Newcomer Newcomer Newcomer Oldcomer Oldcomer Oldcomer Women Women
M Not founder M Founder Marginal M Not founder M Founder Marginal Founder Not founder
Paysannat –0.020* –0.013 –0.015 –0.020 0.026 0.050* 0.010 –0.018*
(0.011) (0.012) (0.024) (0.042) (0.051) (0.026) (0.007) (0.010)
Geo Control Vars Yes Yes Yes Yes Yes Yes Yes Yes

Observations 176 176 176 176 176 176 176 176

Mean 0.02 0.04 0.10 0.15 0.50 0.14 0.01 0.03

* p < 0.1, ** p < 0.05, *** p < 0.01.

58
Regressions at the Household Level

Tables : Land documents and insecurity over land rights, Household Level

Table 22: Land formalization and documents, OLS with secteur fixed effect

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

Document Document %Document Insecurity Insecurity
Paysannat
Fallow Field Field Fallows Fields
Marginal clan –0.073*** 0.032* 0.047*** 0.041*** 0.033** 0.033**
(0.018) (0.019) (0.016) (0.010) (0.016) (0.015)
Men local clan –0.010 –0.020 0.004 0.006 0.002 0.017
(0.021) (0.021) (0.018) (0.012) (0.018) (0.016)
Women from local clan –0.053 –0.005 0.026 0.009 0.011 0.058
(0.058) (0.056) (0.051) (0.033) (0.048) (0.043)
Women F clan –0.101** –0.025 0.024 0.010 0.046 0.037
(0.045) (0.046) (0.039) (0.026) (0.039) (0.036)
Geo Control Vars Yes Yes Yes Yes Yes Yes

Observations 2534 1554 2574 2574 1552 1564

Mean 0.09 0.11 0.13 0.08 0.06 0.08

* p < 0.1, ** p < 0.05, *** p < 0.01.

Note: The variable about which titles households own on their fallows is only available in a subsample
of villages

59
 Table 23: Land formalization and documents, OLS with secteur fixed effect

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

Document Document %Document Insecurity Insecurity
Paysannat
Fallow Field Field Fallows Fields
Marginal clan –0.024 0.008 0.011 0.018 0.013 0.000
(0.022) (0.024) (0.020) (0.013) (0.021) (0.019)
Marginal newcomer –0.143*** 0.067** 0.094*** 0.061*** 0.038 0.074***
(0.030) (0.032) (0.026) (0.017) (0.027) (0.024)
Men local clan –0.001 –0.028 –0.006 –0.001 0.002 0.015
(0.022) (0.023) (0.019) (0.013) (0.019) (0.018)
Local newcomer not F –0.156*** 0.089* 0.071 0.054* –0.007 0.015
(0.050) (0.051) (0.044) (0.029) (0.043) (0.039)
Women from local clan –0.064 –0.010 0.008 0.006 0.038 0.017
(0.075) (0.070) (0.066) (0.043) (0.059) (0.054)
Local newcomer F –0.134*** 0.064 0.012 0.030 –0.037 –0.032
(0.037) (0.040) (0.033) (0.021) (0.034) (0.031)
Women F clan –0.067 –0.016 0.035 0.018 –0.032 0.037
(0.059) (0.060) (0.053) (0.034) (0.050) (0.047)
Local newcomer not F W 0.005 0.018 0.044 0.012 –0.087 0.102
(0.114) (0.114) (0.101) (0.065) (0.096) (0.088)
Local newcomer F W –0.109 –0.001 –0.005 –0.003 0.191** –0.003
(0.088) (0.092) (0.078) (0.050) (0.078) (0.071)
Geo Control Vars Yes Yes Yes Yes Yes Yes

Observations 2533 1521 2538 2538 1519 1533

Mean 0.09 0.11 0.13 0.08 0.06 0.08

* p < 0.1, ** p < 0.05, *** p < 0.01.

Note: The variable about which titles households own on their fallows is only available in a subsample of
villages

60
 Table 24: Land formalization and documents, OLS with secteur fixed effect

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

Document Document %Document Insecurity Insecurity
Paysannat
Fallow Field Field Fallows Fields
Paysannat 0.136*** 0.045* –0.015 –0.011 –0.002 –0.007
(0.021) (0.024) (0.019) (0.013) (0.021) (0.019)
Marginal clan –0.085*** 0.110*** 0.097*** 0.074*** –0.014 0.009
(0.028) (0.031) (0.025) (0.016) (0.026) (0.024)
Marginal clan x P 0.010 –0.125*** –0.083** –0.056*** 0.076** 0.038
(0.036) (0.039) (0.032) (0.021) (0.033) (0.030)
Men local clan –0.091*** 0.019 –0.014 –0.005 0.000 0.022
(0.030) (0.030) (0.027) (0.017) (0.026) (0.023)
Men local clan x P 0.170*** –0.072* 0.033 0.018 0.002 –0.013
(0.040) (0.043) (0.036) (0.023) (0.036) (0.033)
Women from local clan –0.101 0.100 0.084 0.073 –0.053 –0.027
(0.104) (0.116) (0.094) (0.061) (0.098) (0.090)
Women from local clan x P 0.044 –0.150 –0.080 –0.090 0.086 0.112
(0.124) (0.133) (0.112) (0.073) (0.113) (0.103)
Women F clan –0.050 –0.033 0.021 –0.011 0.089* 0.059
(0.058) (0.061) (0.052) (0.034) (0.051) (0.046)
Women F clan x P –0.093 0.038 0.005 0.048 –0.106 –0.061
(0.090) (0.093) (0.080) (0.052) (0.079) (0.073)
Geo Control Vars Yes Yes Yes Yes Yes Yes

Observations 2534 1554 2574 2574 1552 1564

Mean 0.09 0.11 0.13 0.08 0.06 0.08

* p < 0.1, ** p < 0.05, *** p < 0.01.

Note: The variable about which titles households own on their fallows is only available in a subsample of
villages

61
 Table 25: Land formalization and documents, OLS with secteur fixed effect

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

Document Document %Document Insecurity Insecurity
Paysannat
Fallow Field Field Fallows Fields
Paysannat 0.138*** 0.059** –0.008 –0.004 0.008 –0.005
(0.022) (0.026) (0.020) (0.013) (0.022) (0.020)
Marginal newcomer –0.132*** 0.167*** 0.171*** 0.124*** 0.009 0.056*
(0.036) (0.039) (0.033) (0.021) (0.033) (0.030)
Marginal newcomer x P –0.064 –0.160*** –0.126*** –0.085*** 0.078* 0.030
(0.049) (0.052) (0.045) (0.029) (0.044) (0.040)
Marginal oldcomer –0.056 0.087** 0.048 0.047** –0.034 –0.047
(0.036) (0.042) (0.032) (0.021) (0.036) (0.032)
Marginal oldcomer x P 0.035 –0.121** –0.055 –0.043* 0.067 0.068*
(0.045) (0.051) (0.040) (0.026) (0.043) (0.039)
Local newcomer not F –0.135** 0.130** 0.116** 0.069** 0.018 0.006
(0.057) (0.060) (0.051) (0.033) (0.051) (0.045)
Local newcomer Not F x P 0.020 –0.157 –0.157* –0.044 –0.074 0.067
(0.103) (0.104) (0.092) (0.059) (0.088) (0.080)
Local newcomer F –0.069 0.130** 0.039 0.055* 0.007 –0.044
(0.054) (0.059) (0.049) (0.032) (0.050) (0.045)
Local newcomer F x P –0.097 –0.115 –0.046 –0.043 –0.078 0.020
(0.072) (0.080) (0.066) (0.042) (0.067) (0.061)
Local oldcomer Not F –0.089*** 0.013 –0.039 –0.014 0.002 0.027
(0.033) (0.033) (0.030) (0.019) (0.028) (0.026)
Local oldcomer Not F x P 0.171*** –0.068 0.059 0.023 0.002 –0.027
(0.043) (0.046) (0.039) (0.025) (0.039) (0.035)
Local newcomer not F W –0.164 0.370* 0.108 0.151 –0.034 0.052
(0.158) (0.215) (0.144) (0.093) (0.182) (0.166)
Local newcomer Not F W x P 0.124 –0.461* –0.078 –0.188* –0.020 0.084
(0.189) (0.238) (0.172) (0.110) (0.202) (0.184)
Local newcomer F W –0.085 0.003 0.028 –0.008 0.247*** –0.002
(0.083) (0.098) (0.075) (0.048) (0.083) (0.072)
Local newcomer F W x P –0.178 –0.023 0.004 0.059 –0.197 0.081
(0.135) (0.146) (0.123) (0.079) (0.123) (0.110)
Local oldcomer Not F W –0.073 0.016 0.084 0.034 –0.052 –0.058
(0.134) (0.137) (0.122) (0.078) (0.116) (0.106)
Local oldcomer Not F W x P –0.006 –0.049 –0.107 –0.039 0.120 0.103
(0.160) (0.159) (0.145) (0.093) (0.135) (0.123)
Local oldcomer F W –0.036 –0.027 0.040 0.006 0.001 0.106*
(0.079) (0.078) (0.072) (0.046) (0.066) (0.060)
Local oldcomer F W x P –0.045 0.049 –0.010 0.028 –0.083 –0.183*
(0.117) (0.121) (0.106) (0.068) (0.103) (0.096)
Geo Control Vars Yes Yes Yes Yes Yes Yes

Observations 2533 1521 2538 2538 1519 1533

Mean 0.09 0.11 0.13 0.08 0.06 0.08

* p < 0.1, ** p < 0.05, *** p < 0.01.

Note: The variable about which titles households own on their fallows is only available in a subsample of villages

62
Tables : Characteristics of the land portfolio, household level

63
 Table 26: Characteristics of the land portfolio, OLS with
secteur fixed effect

(1) (2) (3)

Nb plots Nb Area
2014 Fallows 2014
Marginal clan –0.363*** –2.426***–0.303**
(0.107) (0.697) (0.138)
Men local clan –0.005 –0.470 0.018
(0.121) (0.770) (0.155)
Women from local clan 0.041 –4.002* –0.184
(0.341) (2.048) (0.438)
Women F clan –0.390 –1.295 –0.065
(0.263) (1.660) (0.341)
Geo Control Vars Yes Yes Yes
64

Observations 2584 1561 2567

Mean 4.30 11.49 2.81

* p < 0.1, ** p < 0.05, *** p < 0.01.

Note: The variable about which titles households own
on their fallows is only available in a subsample of
villages
 Table 27: Characteristics of the land portfolio, OLS with
secteur fixed effect

(1) (2) (3)

Nb plots Nb Area
2014 Fallows 2014
Marginal clan –0.278** –2.188** –0.277
(0.131) (0.887) (0.169)
Marginal newcomer –0.175 –0.948 –0.011
(0.175) (1.147) (0.227)
Men local clan –0.024 –0.323 0.045
(0.129) (0.836) (0.167)
Local newcomer not F 0.109 –2.455 –0.299
(0.295) (1.849) (0.381)
Women from local clan –0.402 –2.428 –0.215
(0.442) (2.537) (0.570)
65

Local newcomer F –0.169 –2.337 –0.216

(0.217) (1.473) (0.280)
Women F clan –0.382 0.962 –0.199
(0.352) (2.179) (0.454)
Local newcomer not F W 1.038 –5.092 0.023
(0.675) (4.157) (0.870)
Local newcomer F W –0.156 –5.594* 0.225
(0.521) (3.313) (0.671)
Geo Control Vars Yes Yes Yes

Observations 2548 1529 2532

Mean 4.30 11.49 2.81

* p < 0.1, ** p < 0.05, *** p < 0.01.

Note: The variable about which titles households own on
their fallows is only available in a subsample of villages
 Table 28: Characteristics of the land portfolio, OLS with
secteur fixed effect

(1) (2) (3)

Nb plots Nb Area
2014 Fallows 2014
Paysannat –0.281** –1.826** –0.511***
(0.128) (0.884) (0.165)
Marginal clan –0.463*** –3.515***–0.693***
(0.168) (1.120) (0.216)
Marginal clan x P 0.171 1.804 0.650**
(0.215) (1.410) (0.277)
Men local clan –0.235 –2.299** –0.520**
(0.179) (1.096) (0.231)
Men local clan x P 0.393 3.455** 0.933***
(0.240) (1.551) (0.309)
Women from local clan 0.358 –5.399 –0.191
66

(0.624) (4.232) (0.801)

Women from local clan x P –0.406 2.268 0.090
(0.746) (4.849) (0.958)
Women F clan –0.375 –1.789 –0.000
(0.348) (2.171) (0.447)
Women F clan x P –0.135 0.504 –0.371
(0.534) (3.374) (0.692)
Geo Control Vars Yes Yes Yes

Observations 2584 1561 2567

Mean 4.30 11.49 2.81

* p < 0.1, ** p < 0.05, *** p < 0.01.

Note: The variable about which titles households own on
their fallows is only available in a subsample of villages
 Table 29: Characteristics of the land portfolio, OLS with
secteur fixed effect

(1) (2) (3)

Nb plots Nb Area
2014 Fallows 2014
Paysannat –0.329** –2.137** –0.527***
(0.134) (0.937) (0.172)
Marginal newcomer –0.405* –3.945***–0.646**
(0.219) (1.418) (0.283)
Marginal newcomer x P –0.147 1.257 0.594
(0.297) (1.891) (0.384)
Marginal oldcomer –0.575*** –3.652** –0.724***
(0.215) (1.546) (0.278)
Marginal oldcomer x P 0.489* 2.391 0.733**
(0.269) (1.871) (0.347)
Local newcomer not F –0.055 –1.748 –0.203
67

(0.342) (2.170) (0.439)

Local newcomer Not F x P 0.210 –4.617 –0.645
(0.612) (3.798) (0.786)
Local newcomer F –0.357 –3.230 –0.243
(0.325) (2.206) (0.418)
Local newcomer F x P 0.303 1.479 –0.014
(0.436) (2.958) (0.560)
Local oldcomer Not F –0.341* –2.848** –0.635**
(0.199) (1.207) (0.257)
Local oldcomer Not F x P 0.528** 4.798*** 1.144***
(0.260) (1.683) (0.335)
Women from local clan –0.322 –2.621 –0.076
(0.314) (2.013) (0.403)
Women from local clan x P 0.062 0.189 –0.184
(0.431) (2.700) (0.554)
Geo Control Vars Yes Yes Yes

Observations 2548 1529 2532

Mean 4.30 11.49 2.81

* p < 0.1, ** p < 0.05, *** p < 0.01.

Note: The variable about which titles households own on
their fallows is only available in a subsample of villages
Tables : Characteristics of the land portfolio, household level

68
 Table 30: Characteristics of the land portfolio, OLS with secteur fixed effect

(1) (2) (3) (4) (5) (6) (7)

Sh short Sh med Sh med Sh conversions Sh Forest Sh cultivated Sh fallows
fallows 2014 fallows 2014 fallows 2014 2014 2014 2014 2014
Marginal clan 0.030* –0.022 –0.026 –0.013 –0.014 0.006 –0.013
(0.016) (0.014) (0.018) (0.011) (0.011) (0.011) (0.012)
Men local clan –0.018 0.002 –0.006 –0.009 –0.014 0.012 –0.023*
(0.018) (0.016) (0.020) (0.012) (0.012) (0.013) (0.014)
Women from local clan 0.061 0.005 –0.060 –0.038 –0.037 –0.033 0.059
(0.052) (0.046) (0.056) (0.034) (0.034) (0.036) (0.038)
Women F clan 0.020 0.035 –0.055 –0.023 –0.022 0.033 –0.008
(0.039) (0.036) (0.042) (0.026) (0.026) (0.028) (0.030)
Geo Control Vars Yes Yes Yes Yes Yes Yes Yes

Observations 2483 2584 2483 2583 2583 2582 2583

69

Mean 0.31 0.24 0.41 0.17 0.17 0.54 0.26

* p < 0.1, ** p < 0.05, *** p < 0.01.

Note: The variable about which titles households own on their fallows is only available in a subsample of villages
 Table 31: Characteristics of the land portfolio, OLS with secteur fixed effect

(1) (2) (3) (4) (5) (6) (7)

Sh short Sh med Sh med Sh conversions Sh Forest Sh cultivated Sh fallows
fallows 2014 fallows 2014 fallows 2014 2014 2014 2014 2014
Marginal clan 0.020 –0.016 –0.019 –0.011 –0.012 0.007 –0.012
(0.020) (0.018) (0.021) (0.013) (0.013) (0.014) (0.015)
Marginal newcomer 0.007 –0.020 0.009 0.007 0.008 –0.006 –0.014
(0.027) (0.024) (0.029) (0.017) (0.017) (0.018) (0.020)
Men local clan –0.022 0.006 0.001 –0.005 –0.010 0.013 –0.024
(0.020) (0.018) (0.021) (0.013) (0.013) (0.014) (0.015)
Local newcomer not F –0.011 –0.035 0.023 0.003 0.006 0.004 –0.024
(0.046) (0.040) (0.049) (0.029) (0.029) (0.031) (0.033)
Women from local clan 0.001 0.014 –0.012 –0.027 –0.026 –0.044 0.031
(0.069) (0.060) (0.074) (0.044) (0.044) (0.046) (0.050)
Local newcomer F –0.070** –0.035 0.088** 0.048** 0.049** –0.005 –0.040
70

(0.033) (0.029) (0.035) (0.022) (0.021) (0.023) (0.025)

Women F clan 0.003 0.014 –0.031 –0.015 –0.014 0.054 –0.037
(0.053) (0.048) (0.057) (0.035) (0.035) (0.037) (0.040)
Local newcomer not F W 0.122 –0.030 –0.094 –0.018 –0.016 0.022 0.060
(0.103) (0.092) (0.111) (0.067) (0.067) (0.071) (0.076)
Local newcomer F W 0.011 0.030 –0.018 0.002 0.001 –0.054 0.050
(0.078) (0.071) (0.084) (0.052) (0.051) (0.055) (0.059)
Geo Control Vars Yes Yes Yes Yes Yes Yes Yes

Observations 2450 2548 2450 2547 2547 2546 2547

Mean 0.31 0.24 0.41 0.17 0.17 0.54 0.26

* p < 0.1, ** p < 0.05, *** p < 0.01.

Note: The variable about which titles households own on their fallows is only available in a subsample of villages
 Table 32: Characteristics of the land portfolio, OLS with secteur fixed effect

(1) (2) (3) (4) (5) (6) (7)

Sh short Sh med Sh med Sh conversions Sh Forest Sh cultivated Sh fallows
fallows 2014 fallows 2014 fallows 2014 2014 2014 2014 2014
Paysannat –0.022 0.027 –0.017 0.017 0.018 –0.030** 0.011
(0.019) (0.017) (0.021) (0.013) (0.013) (0.013) (0.014)
Marginal clan 0.005 –0.021 –0.012 –0.001 –0.002 0.018 –0.048**
(0.025) (0.023) (0.027) (0.017) (0.017) (0.018) (0.019)
Marginal clan x P 0.042 –0.004 –0.021 –0.020 –0.019 –0.017 0.055**
(0.033) (0.029) (0.035) (0.021) (0.021) (0.023) (0.024)
Men local clan –0.036 0.006 0.003 0.001 –0.002 0.029 –0.047**
(0.027) (0.024) (0.029) (0.018) (0.018) (0.019) (0.020)
Men local clan x P 0.030 –0.003 –0.018 –0.017 –0.019 –0.035 0.047*
(0.037) (0.033) (0.040) (0.024) (0.024) (0.025) (0.027)
Women from local clan 0.000 0.074 –0.099 –0.020 –0.018 –0.065 0.013
(0.097) (0.085) (0.105) (0.062) (0.062) (0.066) (0.070)
71

Women from local clan x P 0.088 –0.104 0.059 –0.028 –0.030 0.051 0.064
(0.115) (0.101) (0.124) (0.074) (0.074) (0.078) (0.084)
Women F clan –0.025 0.017 0.005 0.002 0.004 0.070* –0.059
(0.052) (0.047) (0.056) (0.035) (0.034) (0.037) (0.039)
Women F clan x P 0.097 0.050 –0.146* –0.054 –0.055 –0.094* 0.123**
(0.080) (0.072) (0.086) (0.053) (0.053) (0.056) (0.060)
Geo Control Vars Yes Yes Yes Yes Yes Yes Yes

Observations 2483 2584 2483 2583 2583 2582 2583

Mean 0.31 0.24 0.41 0.17 0.17 0.54 0.26

* p < 0.1, ** p < 0.05, *** p < 0.01.

Note: The variable about which titles households own on their fallows is only available in a subsample of villages
 Table 33: Characteristics of the land portfolio, OLS with secteur fixed effect

(1) (2) (3) (4) (5) (6) (7)

Sh short Sh med Sh med Sh conversions Sh Forest Sh cultivated Sh fallows
fallows 2014 fallows 2014 fallows 2014 2014 2014 2014 2014
Paysannat –0.027 0.027 –0.016 0.017 0.018 –0.031** 0.011
(0.020) (0.018) (0.022) (0.013) (0.013) (0.014) (0.015)
Marginal newcomer –0.013 –0.034 0.014 0.015 0.016 0.018 –0.061**
(0.033) (0.030) (0.036) (0.022) (0.022) (0.023) (0.025)
Marginal newcomer x P 0.072 –0.002 –0.048 –0.034 –0.035 –0.033 0.063*
(0.045) (0.040) (0.049) (0.030) (0.029) (0.031) (0.034)
Marginal oldcomer 0.004 –0.018 –0.015 –0.005 –0.007 0.013 –0.046*
(0.032) (0.029) (0.035) (0.021) (0.021) (0.023) (0.024)
Marginal oldcomer x P 0.027 0.000 –0.004 –0.011 –0.010 –0.006 0.049
(0.041) (0.037) (0.044) (0.027) (0.027) (0.028) (0.030)
Local newcomer not F –0.059 –0.020 0.046 0.032 0.030 0.027 –0.073*
(0.052) (0.046) (0.056) (0.034) (0.034) (0.036) (0.039)
72

Local newcomer Not F x P 0.055 –0.014 –0.086 –0.100* –0.096 –0.045 0.075
(0.100) (0.083) (0.108) (0.061) (0.060) (0.064) (0.069)
Local newcomer F –0.073 –0.026 0.084 0.029 0.033 0.010 –0.024
(0.049) (0.044) (0.053) (0.032) (0.032) (0.034) (0.037)
Local newcomer F x P 0.002 –0.013 0.006 0.038 0.034 –0.032 –0.028
(0.065) (0.059) (0.071) (0.043) (0.043) (0.046) (0.049)
Local oldcomer Not F –0.041 0.009 0.005 –0.001 –0.004 0.029 –0.047**
(0.030) (0.027) (0.033) (0.020) (0.020) (0.021) (0.022)
Local oldcomer Not F x P 0.030 –0.003 –0.009 –0.005 –0.008 –0.032 0.043
(0.040) (0.035) (0.043) (0.026) (0.026) (0.027) (0.029)
Women from local clan –0.037 0.025 0.001 0.005 0.007 0.035 –0.049
(0.047) (0.043) (0.051) (0.031) (0.031) (0.033) (0.035)
Women from local clan x P 0.114* –0.012 –0.085 –0.048 –0.050 –0.057 0.114**
(0.065) (0.059) (0.070) (0.043) (0.043) (0.045) (0.049)
Geo Control Vars Yes Yes Yes Yes Yes Yes Yes

Observations 2450 2548 2450 2547 2547 2546 2547

Mean 0.31 0.24 0.41 0.17 0.17 0.54 0.26

* p < 0.1, ** p < 0.05, *** p < 0.01.

Note: The variable about which titles households own on their fallows is only available in a subsample of villages