World Development Vol. 63, pp.

33–42, 2014
Ó 2014 Elsevier Ltd. All rights reserved.
0305-750X/$ - see front matter
www.elsevier.com/locate/worlddev
http://dx.doi.org/10.1016/j.worlddev.2013.10.013

The Role of Rural–Urban Migration in the Structural

Transformation of Sub-Saharan Africa
ALAN DE BRAUW,
VALERIE MUELLER

and

HAK LIM LEE *

International Food Policy Research Institute, USA
Summary. — Rural-to-urban migration is an inherent part of the economic development process, yet it is relatively understudied in sub-
Saharan Africa. In this paper, we attempt to describe the present state of rural–urban migration from several different angles. Migration
rates are quite low in several countries, despite the fact that large proportions of populations continue to reside in rural areas, and that
there are clearly several types of gains to migration. We offer a number of possible explanations for low migration rates. We make rec-
ommendations for improvements in research on rural–urban migration and migration policy in Africa.
Ó 2014 Elsevier Ltd. All rights reserved.

Key words — rural-urban migration, structural transformation, Africa

1. INTRODUCTION Pritchett (2006) notes that during the 20th century, a large part
of the Great Plains region in the United States experienced
Rural-to-urban migration is a necessary component of the roughly the same rate of economic growth as the remainder
economic development process, as the migration of labor out of the country, while the population of the Great Plains fell
of agriculture has been a feature of the growth path of every in absolute terms by 27%. Without migration, the population
country that has developed (Taylor & Martin, 2001). From a would have been nearly three times higher, and living stan-
macroeconomic perspective, every country has its own dis- dards would almost certainly not have grown as quickly.
tinct growth path (World Bank, 2008), but the common fea- Moreover, migration can enhance living standards in rural
ture is that as economic growth occurs, labor moves out of areas through remittances migrants send back (e.g., Cox-Ed-
agriculture into the manufacturing and service sectors. How- wards & Ureta, 2003; Yang, 2008). If migration is slow or
ever, as noted by Foster and Rosenzweig (2008), from a not occurring, such potential growth is not occurring.
microeconomic perspective the transition out of agriculture Although African cities have grown quickly over the past
is not well understood. The path from a largely agricultural 50 years, rural–urban migration has played a relatively small
economy to a wealthier, manufacturing, and services based role in recent growth (Kessides, 2007; Potts, 2012), and in
economy seems varied, and perhaps more importantly it is some African countries, net rural–urban migration is even
not clear how to positively influence progress along that path negative (Beauchemin, 2011).
using policy. We have two primary objectives in this paper. First, we
In sub-Saharan Africa, the same negative correlation between provide a description of the state of rural–urban migration
GDP per capita and the share of the population living in rural in Africa, updating previous research on the topic (e.g., Byer-
areas exists as elsewhere in the world (Figure 1, Panel A). It does lee, 1974). Second, we explore possible reasons why migra-
not change if we replace the rural population share with the tion rates are not higher, given the difference in returns to
share of the economically active population working in agricul- labor in agriculture and other sectors. One of the primary
ture (Panel B). Moreover, both graphs indicate that the majority challenges faced in this paper is that few data sets exist in
of the population still lives in rural areas in much of sub-Saha- sub-Saharan Africa that allow for the detailed study of
ran Africa. If those economies are to grow rapidly in a sustain- migration and its implications.
able manner, additional rural–urban migration must occur.
Rural–urban migration is a topic that often gives policy
makers pause, due to perceptions about potential negative im- 2. OVERVIEW OF RURAL–URBAN MIGRATION IN
pacts of migration. In urban areas, concerns about significant SUB-SAHARAN AFRICA
rural–urban migration include increasing unemployment, pro-
viding services to new arrivals, and potential for political un- In this section, we initially describe the knowledge gaps that
rest (UN-HABITAT, 2010). In rural areas, a historical arise from the relative lack of data on rural–urban migration
concern about rural–urban migration is that a labor shortage in sub-Saharan Africa. We then present rough measures of
occurs at times when specific agricultural tasks must occur
(e.g., Binswanger and Rosenzweig, 1986; Byerlee, 1974). Yet
without migration, the returns to labor in urban and rural * We thank Kathleen Beegle, Claudio Montenegro, Derek Headey, and
areas can become significantly imbalanced. For example,
two anonymous reviewers for their feedback.
33
34 WORLD DEVELOPMENT

Panel A Brauw, Mueller, & Woldehanna, 2011, 2012). Other surveys

are also underway to follow individuals from all of the house-
holds in the panel, such as the World Bank’s Living Standards
and Measurement Study – Integrated Surveys for Agriculture
(LSMS-ISA). As the LSMS-ISA panel data sets become avail-
able, more basic information about rural–urban migration in
Africa will be revealed.
Among the few datasets that document migration patterns,
there is substantial heterogeneity across countries, which par-
tially comes from the inconsistent definitions of urban over
space and time. Different countries define an urban area in dif-
ferent ways. Urban areas have also expanded, and are often offi-
cially reclassified from rural to urban. Therefore, demarcations
of urban zones continue to accommodate the birth and growth
of metropolitan areas since 1960. As a consequence, national
statistics on rural–urban population shares are not immediately
comparable across space or time. While there are several efforts
underway to measure urban expansion (e.g., Bloom, Canning,
Fink, Khanna, & Salyer, 2010), without holding urban areas
Panel B constant, it is nearly impossible to compute rural–urban migra-
tion rates, as one cannot attribute city growth in excess of nat-
ural population growth to migration alone.
Given our description of urban expansion, it is perhaps not
surprising that urbanization is often linked to rural–urban
migration, though migration is not the only cause of urbaniza-
tion. Nonetheless, the fact that the countries of sub-Saharan
Africa are often described as rapidly urbanizing is often as-
sumed to imply high rural–urban migration rates (e.g., Tod-
aro, 2000). For example, Byerlee (1974) reports urban
growth rates of over 7% per annum in the 1960s. However,
there is recent evidence that growth has slowed or even re-
versed in some cases (e.g., Potts, 2009). Considering all of
the areas defined as urban in the CIESEN (2004) data set,
the average urban growth rate was 3.4% in sub-Saharan Africa
between 1990 and 2000.
We want to estimate the contribution of migration to urban
growth in sub-Saharan Africa. Urban population growth is
the sum of four components: net fertility, where net fertility
Figure 1. Rural population versus GDP per capita in PPP terms, sub- is the birth rate less the death rate; urban expansion or reclas-
Saharan Africa. Source: World Development Indicators (2011). sification of areas from rural to urban; rural–urban migration;
and international immigration. Although fertility rates are lar-
gely thought to be lower in urban areas than in rural areas, we
the rate of rural–urban migration in the countries of sub-Sah- assume they are the same in this case. We control for urban
aran Africa. expansion by using a definition of urban areas that is consis-
There are several reasons that information on rural–urban tent over time, as defined by the CIESEN (2004) project. 1 A
migration in sub-Saharan Africa is relatively scarce. Nation- major problem in comparing rural–urban migration across
ally representative, multi-purpose household surveys often countries is that the concept of “urban” takes on varying
lack a distinct migration module (e.g., Lucas, 2000, chap. forms in different countries, so simply comparing national sta-
16). They do often ask about the birthplace of people, which tistics across countries may not lead to correct conclusions.
means migrants can be observed at the destination, but one We then interpret the difference between the urban and rural
cannot use destination information alone to construct esti- population growth rates between 1990 and 2000 as the rur-
mates of migration rates. Panel surveys often only follow al–urban migration rate. 2
household heads, such as the World Bank Living Standards We find that the population weighted rural–urban migration
Measurement Surveys (LSMS), which implies that informa- rate was 1.07% per annum between 1990 and 2000 in sub-Saha-
tion about migrants is lost between rounds unless explicitly in- ran Africa. 3 Not surprisingly, the average masks a great deal of
cluded in the survey. Moreover, migration modules in LSMS heterogeneity at the country level (Figure 2). Although several
style surveys often lack information to characterize migrants; countries have rural–urban migration rates right around 1%,
for example, the Ghana Living Standards Survey conducted a few countries have very slow or even negative rural–urban
in 2006 included a migration module, but did not record migration rates, and only a handful experienced migration rates
whether the individual’s birthplace was in a rural or urban over 2% during the 1990s. 4 Negative rural–urban migration im-
area, and only asked about the last location. As people appear plies that re-ruralization is occurring; these findings are corrob-
to move within Ghana, the information is inadequate for char- orated by other evidence for at least some of the countries for
acterizing the flow of rural–urban migration. Recently, a few which we find re-ruralization. Specifically, Beauchemin (2011)
efforts in Africa have attempted to follow migrants, such as finds evidence that people are returning to rural areas in Cote
the Kagera Household Survey (e.g., Beegle, de Weerdt, & Der- D’Ivoire. Potts (2009) notes the copperbelt towns in Zambia
con, 2011) and the ERHS Migrant Tracking Survey (de also decreased in population during the 1990s.
 THE ROLE OF RURAL–URBAN MIGRATION IN THE STRUCTURAL TRANSFORMATION OF SUB-SAHARAN AFRICA 35

role in the structural transformation (McMillan, 2011). How-

.5

ever, before we discuss the theoretical underpinnings of rural–

urban migration, we want to note that three other types of
population movement are important in Africa: population
.4

movements due to emergencies, rural–rural migration, and

international migration. Population movements due to emer-
gencies have been relatively frequent and well-publicized in
.3
Density

sub-Saharan Africa. According to the United Nations High

Commission for Refugees. (2011), as of 2010 there are approx-
imately 2 million refugees in Africa, and an additional 6 mil-
.2

lion people living in camps for internally displaced people

(IDP camps). These movements are clearly important to deter-
.1

mining the well-being of people in sub-Saharan Africa; how-

ever, they represent a different type of economic
transformation, so we do not cover them here. Other types
0

-2 -1 0 1 2 3
of migration (rural–rural or international) have similar moti-
Rural-Urban Migration Rate (Percent) vations from a theoretical perspective. 5

Figure 2. Number of countries with different rural–urban migration rates, (a) Theory: Why people migrate
1990–2000, sub-Saharan Africa. Notes: Migration rate below average
implies it is below the population weighted average of 1.07% for sub-Saharan To begin considering the rationale for rural–urban migra-
Africa. Above average implies a migration rate between the average and 2% tion in sub-Saharan Africa, it is useful to begin with the classic
per annum. High implies above 2% per annum. Source: CIESEN (2004). model of Harris and Todaro (1970). Consider a risk-neutral
rural laborer who makes a daily wage wr(Lr) in agriculture,
where rural wages are a function of the total rural labor force.
The laborer faces the decision of whether or not to migrate to
an urban area to receive a daily wage wu(Lu), where urban
wages are a function of the total urban labor force. However,
the laborer will only receive a job in the urban area with prob-
ability pu(Lu), which is again a function of the size of the urban
labor force. Harris and Todaro postulated that if Lr + Lu = -
L, and if initially puwu > wr, then laborers would move from
rural to urban areas until an equilibrium condition was met:
wr ðLr Þ ¼ pu ðLu Þwu ðLu Þ:
At equilibrium, urban wages remain higher than rural
wages, but the wedge between rural and urban wages is caused
by the uncertainty of employment in urban areas, whereas rur-
al employment is certain on the farm.
Although the Harris–Todaro framework has been used in a
myriad of different situations, it was originally posed as an
explanation for the rural–urban wage gap and persistent high
unemployment in Latin America in the 1960s. From the point
Figure 3. Share of population that is rural versus estimated rural–urban
migration rate, sub-Saharan Africa, 1990–2000. Source: CIESEN (2004).
of view of stylized facts, this view is not inconsistent with sub-
Saharan Africa in the 2000s. Whereas there is reasonably cer-
tain rural employment in agriculture in most countries, finding
Figure 3 uses the CIESEN data to plot the population share employment is less certain in urban areas, where the informal
living in rural areas against the computed rural–urban migra- sector is large.
tion rate. In 2000, 68.2% of the population of sub-Saharan Stark (1991) and Stark and Bloom (1985) posited the New
African lived in rural areas. Around the average, there is no Economics of Labor Migration (NELM), which hypothesizes
obvious relationship between the migration rate and the pop- that migration is a household rather than an individual deci-
ulation share in rural areas; between around 60% and 80% of sion. 6 According to NELM, migration is used as a household
population living in rural areas, rural–urban migration rates level strategy for overcoming constraints facing households,
are widely dispersed. So although we know that the urban such as missing credit or insurance markets. From the perspec-
population share must increase in some countries as growth tive of rural–urban labor migration in Africa, this explanation
occurs, this graph demonstrates that there are several coun- for migration and potentially remittance behavior is quite
tries in sub-Saharan Africa with low migration rates and high attractive. Households often lack access to credit or insurance,
population shares in rural areas. and as a result they could potentially turn to rural–urban or
international migration either as a substitute for access to
credit markets (Rozelle, Taylor, & de Brauw, 1999; Wouterse
3. MOTIVATIONS FOR MIGRATION & Taylor, 2008) or as a risk coping strategy (Dillon, Mueller,
& Salau, 2011; Paulson, 2000). Compared to the potential in-
There are several reasons why people move from one place come from local off-farm work, the covariance between in-
to another. We are interested in studying the economic ratio- come from migration and other household income is lower,
nales for rural–urban migration in the African context. We which decreases overall income variance. Alternatively, migra-
choose rural–urban migration because of our interest in its tion and any remittances migrants send back can be thought
36 WORLD DEVELOPMENT

of as an informal institution designed to foster limited risk Table 1. Urban–rural wage ratio by sector in selected sub-Saharan African
sharing (e.g., Foster & Rosenzweig, 2001). 7 countries
Therefore, the reasons that individuals might move out of Formal sector Informal sector Survey year
agriculture are complex. Expected or perceived wages at the
Cameroon 1.36 1.26 2007
destination can play a role in the migration decision, but other
Chad 1.45 1.06 2006
rationales can do so as well. Migrants may move out to pro-
Ethiopia 2.05 2.22 2000
vide additional liquidity, better risk sharing, or other services
Gabon 1.08 2.71 2005
at the household level.
Kenya 1.92 2.74 2005
Even if households do not expect to receive remittances Mozambique 1.67 1.07 2006
from migrants, internal migration can potentially benefit the Niger 0.86 0.84 2002
source household as there is more for the remainder of the Nigeria 1.36 1.49 2003
household to share. Although not validated in the empirical Tanzania 1.59 1.26 2006
literature, if household production and income primarily come Togo 2.22 2.81 2006
through agriculture and are strongly linked to a fixed factor, Uganda 2.03 2.34 2005
such as land, then the opportunity cost of the household mem- Zambia 1.64 3.12 2003
ber’s departure may be low.
Source: International Income Distribution Database (2012).
(b) Rural–urban income differentials in Africa

Based on the Harris–Todaro theory, it is important to

first understand the returns to labor in rural areas relative International Income Distribution Database (World Bank,
to urban areas. From a microeconomic perspective, it is 2012a). The database compiles information from approxi-
quite challenging to measure the differential between the re- mately 750 nationally representative surveys (e.g., Labor
turns to agricultural and non-agricultural labor in sub-Sah- Force Surveys and Household, Income, and Consumption
aran Africa. Given that most workers in rural areas perform Surveys) and standardizes demographic and labor variables
agricultural labor, and often have their own land, returns to to allow for comparisons over 160 countries. Although the
labor specifically in agriculture are difficult to measure. database was established in 2005, it is updated continuously
However, returns to labor can be equally difficult to mea- and used in World Bank publications.
sure in urban areas, due to the prevalence of informal labor For selected countries in sub-Saharan Africa, we estimate
(Fox & Gaal, 2008). Lastly, individuals frequently move the ratio of urban to rural wages in both the formal and
into and out of work in sub-Saharan Africa. Using panel informal sectors (Table 1). 8 In each case except for Niger,
data collected in Ghana, Sandefur, Serneels, and Teal urban wages always exceed rural wages in both the formal
(2007) find that nearly 15% of individuals change jobs or and informal sectors. Formal sector wages appear to be gen-
employment status within a one-year interval. Such transi- erally higher on average than informal sector wages in ur-
tions make coming up with microeconomic conclusions ban areas, though the informal sector differential is
about the relative returns to labor quite difficult. actually higher in some cases. Finally, note that in six coun-
Two recent studies of labor productivity show large produc- tries, urban informal wages more than double rural informal
tivity gaps between agriculture and non-agricultural sectors. wages. Clearly, in such countries there appears to be a high
An important insight from McMillan and Rodrik (2011) is return to moving from the rural to the urban informal sec-
that there are large differences in productivity across sectors tor. As average wages may mask heterogeneity in the re-
within economies, and that returns are lowest consistently in turns to labor, for the same list of countries, we also
the agricultural sector. Moving labor out of the agricultural compare the poverty head counts in rural and urban areas
sector into other sectors would, at least in partial equilibrium using the World Development Indicators in Table 2 (World
terms, increase the returns to labor in the economy. In a re- Bank, 2012b). Consistent with evidence from the ratio of ur-
lated paper, Gollin, Lagakos, and Waugh (2011) show that ban to rural wages in the formal and informal sectors, pov-
the gap between agricultural and non-agricultural productivity erty rates are consistently lower in urban areas.
cannot be explained by differences in hours worked, human
capital, educational quality, cost-of-living differences, or fac-
tor shares. They find the gap between agricultural and non- Table 2. Poverty head count ratio at the (rural or urban) poverty line in
agricultural productivity persists even accounting for all of selected sub-Saharan African countries
these potential confounding factors. It seems safe to conclude Rural Urban Year
that the returns to labor in the agricultural sector in sub-Sah-
aran Africa are lower than in other sectors. Cameroon 55.0 12.2 2007
Though the returns to labor in agriculture are lower, we Chad 58.6 24.6 2003
have not yet demonstrated that the returns to migrating to ur- Ethiopia 39.3 35.1 2005
ban areas are higher than the returns to all economic activities Gabon 44.6 29.8 2005
Kenya 49.1 33.7 2005
in rural areas. Using individual migrant tracking data,
Mozambique 56.9 49.6 2008
Beegle, de Weerdt, and Dercon, 2011 find that consumption
Niger 63.9 36.7 2007
growth is 36 percentage points higher among all migrants leav-
Nigeria 63.8 43.1 2004
ing villages in Kagera between 1991 and 2004 than individuals
Tanzania 37.4 21.8 2007
who stayed. Among individuals who move to more connected
Togo 74.3 36.8 2006
areas, proxying for urban areas, this figure rises to 66 percent- Uganda 27.2 9.1 2009
age points. Their study suggests the returns to labor in urban Zambia 76.8 26.7 2006
areas may be higher than the returns to labor in rural areas.
An alternative measure of the relative returns to labor in the Source: World Development Indicators (World Bank, 2012b)
formal and informal sectors can be derived from the Note: The year in the database refers to the latest information available
per country.
 THE ROLE OF RURAL–URBAN MIGRATION IN THE STRUCTURAL TRANSFORMATION OF SUB-SAHARAN AFRICA 37

(c) Remittances in Africa Consider the intergenerational transfers model where the
migrant and household have an implicit contract (Cox, Eser,
Conditional on migration, remittance rates appear low in & Jimenez, 1998). In the first stage, the parent might support
most countries for which we have data (Table 3). 9 For Ethio- the migrant to further his education outside of the village and
pia, we use remittance reports from migrants who left the later obtain work (especially in areas where obtaining land for
households in the 18 villages surveyed in the Ethiopian Rural youth is problematic). The migrant may only begin to fulfill
Household Survey 2004–05 that were tracked in 2009 (de his end of the bargain in the second stage, when he is older,
Brauw Mueller, and Woldehanna, 2013a). For the rest of and when the parent’s earnings are also lower, later in the life
the countries, we use the Migration and Remittances House- cycle. When one compares the characteristics of rural–urban
hold Surveys which were conducted in Burkina Faso, Kenya, migrants to rural residents, one in fact observes that migrants
Nigeria, Senegal, South Africa, and Uganda during 2009 and are indeed younger and have a higher education level (Table 4).
2010, where source households were asked about household The demographic shift of migrants may be such that the
members who had left the household and their remittances majority of our migrants are at too early a stage in the con-
(World Bank, 2012c). While the timing and standardized tractual arrangement to support their family. Taken together,
methodology behind these six surveys allow for comparison without complete information on behalf of the migrant and
across countries, only Senegal, Uganda, and Nigeria are household over time, we cannot refute that implicit contract
nationally representative. The surveys done in Burkina Faso arrangements exist.
and Kenya sampled households from 10 provinces and 17 dis- Those few studies that have information from both actors
tricts that had higher incidences of migration. The survey per- suggest a major motivation for remittances may be to substi-
formed in South Africa sampled two migration corridors tute formal insurance. Remittances for the insurance rationale
(Limpopo and Gauteng). may be important to household productivity, as households in
There are several potential reasons that remittance rates rural areas of developing countries are exposed to many types
might be low. First, it could be that the main contribution of agricultural and health risks. Gubert (2002) finds that the
of the household member leaving is to consume less from insurance motive is important for remittances in the Keyes
the household’s production. This hypothesis would apply pri- area of Mali. On the other hand, de Brauw et al. (2011) find
marily to densely populated areas or countries. Where land that in a tracked sample of migrants in Ethiopia, migrants ap-
pressure is sufficiently high, then overall household production pear to remit for self-insurance rather than to insure the source
may not suffer much when the migrant leaves. households. As existing studies come to different conclusions
A second explanation for low remittance rates is that meth- and represent select locations in Africa, additional data collec-
ods to transfer money to source households are scarce. Where tion efforts that match source households to migrants would
there is a lack of financial development, and formal or infor- be beneficial for our understanding of migration and remit-
mal institutions to facilitate remittances do not exist, then tance behavior in Africa. Applications in behavioral econom-
remittances may be hindered. Alternatively, migrants might ics are also being made to further understand remittance
resort to bringing money or other goods home when they re- motives (e.g., Yang, 2011).
turn. In this case, new technologies may make remittances
more feasible, such as banking through cell phones. In Kenya,
where remittance rates were already high, the frequency of 4. WHY MAY RURAL–URBAN MIGRATION RATES
remittances is higher for M-PESA members than non-mem- BE TOO LOW IN SUB-SAHARAN AFRICA?
bers (Jack & Suri, 2010). However, such products might not
be obviously profitable for providers in countries with less of In the previous two sections, we establish that internal
a remittance tradition. migration rates are quite heterogeneous in sub-Saharan
One can alternatively challenge the interpretation of these Africa. As such, a set of countries exists in which a large pro-
remittance rates to suggest continued uncertainty about the portion of laborers reside and work in rural areas and have
role of remittances in motivating migration. We compute low rural–urban migration rates, despite higher available
remittance rates based on migrant reports at one point in time, returns to internal migration. Understanding why migration
which may provide an incomplete picture of the story. This is rates are lower depends on our knowledge of policies that have
particularly true if an implicit contractual arrangement exists explicitly and implicitly encouraged or hindered rural–urban
between the household and migrant. Under those circum- migration in the past and present and familiarity with the
stances, one would really need to further describe information barriers to migration. In what follows, we contextualize the
on behalf of the household and over a period of time. economic and political background of countries with

Table 3. Internal remittance rates among recent migrants, individuals aged 16 and above
Country Internal remittance rate Source
Ethiopia 33 ERHS Migrant Tracking Survey, 2009
Uganda 24 Migration and Remittance Household Survey 2010
Nigeria 32 Migration and Remittance Household Survey 2009
Senegal 55 Migration and Remittance Household Survey 2009
Kenya 46 Migration and Remittance Household Survey 2009
South Africa 43 Migration and Remittance Household Survey 2009
Burkina Faso 47 Migration and Remittance Household Survey 2010
Note: Remittance rates are computed for internal migrants who were reported to have sent money, food, or goods back to the
source household within the past 12 months. Internal migrants are classified as any household member who left the household in
the last 5 years and are above the age of 15. No distinction is made by migration pattern (e.g., rural–urban migrants or rural–
rural migrants).
 38 WORLD DEVELOPMENT

historically high rural–urban migration rates and review expli-

Sources: Malawi: Integrated Household Survey, 2004; Kenya, Nigeria, Senegal: Migration and Remittance Household Surveys, 2009; Uganda: Migration and Remittance Household Survey, 2010.
Residents
cit and implicit barriers to migration in the present. We then
Rural

3670
85.2
35.2
42.9

0.4
9.2
5.2
contrast a discussion of known individual barriers to migra-
tion with one on the hidden barriers to migration which are
Senegal

underexplored though merit future empirical investigation.

Rural-Urban
Migrants

(a) Policies
66.1
14.5
14.5
27.1
84.3

223
5
The structural transformation, as witnessed in the recent
decades in Asia, attracted rural–urban urban migrants after
a marked growth in agricultural productivity (i.e., the Green
Rural Residents

Revolution), the concerted development of industrialized cit-

ies, or a combination of both. We can ascribe high urbaniza-
Table 4. Human capital characteristics of recent migrants in urban areas versus rural residents, individuals aged 16 and above

2790
58.4
28.2
10.6
34.8
48.2

2.9

tion rates observed in some African countries to a focus on

resource exports. The legacy of the 1970s oil boom in Nigeria
Uganda

was a large increase in government revenue, followed by

investment in the development of urban centers. Rural–urban
migrants were therefore drawn to employment opportunities
Rural-Urban
Migrants

in the construction and service sectors (Akpan, 2012). Declines

58.3

25.9
31.3
31.3
11.5
278
26

in federal agricultural expenditures and shortages of rural la-

bor further contributed to the deterioration of the agricultural
sector (Mogues et al., 2008). Jedwab (2011) has since formal-
ized a structural model to describe the development of “con-
Rural Residents

sumption cities” (not production cities) in economies

dependent on resource exports. He uses historical panel data
4521
43.2
34.9
50.7

7.7
26
23

to provide evidence of this process in Ghana and the Ivory

Coast.
Nigeria

Broader development strategies also affect rural–urban

migration. However, explicit policy restrictions on migration,
Rural-Urban

though rare in Africa (Deshingkar & Grimm, 2004), exist of-

Migrants

ten out of political necessity or to reduce the proliferation of

22.8
55.3
16.3
258
5.7
26
64

urban slums. How these policies manifest in practice is often

unclear. The decentralization of services in Ghana and else-
where, among other rationales, is motivated to reduce rural–
urban migration (Owusu, 2005). In Ethiopia, the government
Rural Residents

overtly discourages rural–urban migration (National Popula-

tion Policy of Ethiopia, 1993). Moreover, urban registration
2591
15.3
39.8
32.1
12.8
39.2
47.5

cards are technically required for employment in the formal

sector and to access services. However, a recent migration sur-
Kenya

vey conducted by the World Bank indicates that as many as

94% of migrant workers in the domestic industry and 63%
Rural-Urban

of other migrants in Addis Ababa were not registered despite

Migrants

cognizance of the registration requirements (World Bank,

27.1
29.4
61.5

317
22
50
1

2010). Empirical investigations into the links between explicit

policy barriers on population movement would make signifi-
cant policy contributions to the existing literature.
Implicit policy restrictions are equally important in affecting
Rural Residents

migration. In settings where property rights are less transpar-

14732

ent, the fear of expropriation can affect the decision to mi-

75.7
21.7
35.7
47.9

2.2
0.3

grate. Whereas households often hold use rights to their

land, they do not necessarily have the right to rent out land,
Malawi

and as such they may lose their land if they leave the village.
There is only one study, that we are aware of, however, that
Rural-Urban

explicitly estimates the relationship between land rights and

Migrants

internal migration in sub-Saharan Africa. De Brauw and

1621
38.5
16.3
38.7
27.7
51.3

6.4

Mueller (2012) find weak evidence that land rights and migra-
tion are negatively correlated in Ethiopia.
Educational policies may also have implicit, negative effects
on migration. In Section 3, we found that higher education
Less than Primary

levels appear correlated with migrant status across several dif-

Education Level

ferent countries. By necessity, schools are less densely distrib-

Number of Obs.
Secondary
Percent Male

uted in rural areas than urban areas, particularly at higher

Average Age

Primary

Higher

levels of schooling, making them more difficult to reach for

rural residents. Rural–urban literacy gaps in sub-Saharan
Africa are attributed to disparities in resources and teaching
quality (Zhang, 2006). Admissions to secondary schools often
 THE ROLE OF RURAL–URBAN MIGRATION IN THE STRUCTURAL TRANSFORMATION OF SUB-SAHARAN AFRICA 39

depend on exam scores, where urban residents outperform larly, Klonner and Nolen (2010) find cell phone coverage has a
rural immigrants due to these discrepancies. As a result, substantial positive effect on employment in rural areas, par-
chances of successful migration may be lowered through edu- ticularly for women and wage occupations, and also causes
cational policies as well. a substantial sectoral shift from rural to urban employment.
Given the recent data collection efforts in Africa, it will now As such, mobile phones have the potential to reduce informa-
become possible to provide additional empirical evidence on tion asymmetries and increase rural–urban migration by relay-
these various fronts for multiple countries at the nationally ing job-related information to rural residents.
representative level. The Living Standards Measurement Migrant networks can also act to reduce the non-monetary
Study – Integrated Survey on Agriculture team at the World costs of migration. Rural–urban migration is a process leading
Bank has committed to conducting household surveys in sev- someone to move into an unfamiliar location and can lead to
eral African countries, where modules on migration and land, anxiety. If enough “like” people are not available to assist with
among other key variables, are included to study these issues this process, the process becomes less likely to occur. There-
and more. For some countries, panel data will even be avail- fore an important component of such networks is likely to
able. With these efforts there will doubtlessly emerge a more be ethnic, religious, or cultural. Given the diverse groups pres-
solidified understanding of how institutional changes may be ent in many sub-Saharan countries, a lack of migrants from
used to increase migration rates for broader development one’s own ethnic group in a particular destination likely has
strategies. The barriers we will discuss next are more difficult a strong, negative effect on the probability of migration to that
to understand with existing data sources due to the challenges destination. While ethnic, religious, or cultural based networks
in their identification and measurement. have been studied in other contexts in Africa (Bandiera & Ra-
sul, 2006; Conley & Udry, 2010; De Weerdt & Dercon, 2006;
(b) Implicit barriers to rural–urban migration Krishnan & Sciubba, 2009), there are relatively few quantita-
tive studies on migration that study such aspects of networks.
There are several implicit barriers to rural–urban migration An exception is Luke and Munshi (2006) who demonstrate
that could also be affected by policy. These barriers could be that ethnic-networks are used by migrants and their families
thought of as affecting either the individual or household deci- to reduce employment uncertainty and solidify future remit-
sion to migrate. Largely, these barriers involve information tance receipts through the migrant’s marriage at the destina-
and risk, network channels, and anticipated changes in well- tion. To the extent that such barriers affect migration,
being. Here, we discuss these barriers in more detail. individuals may be less receptive to interventions that attempt
First, a well-known barrier to migration is that risk-averse to facilitate the job search in urban areas, such as introducing
individuals may be more disinclined to migrate, particularly job brokers.
if the perceived income risks of migration are high. Percep- A second major potential barrier is the opportunity cost of a
tions of income risk could hinder migration in two ways. First, migrant’s departure. Able bodied workers may be scarce in
recall that the Harris–Todaro model predicts that migration some households, and therefore the potential migrant’s contri-
will occur until the expected returns to labor in rural areas bution to household income at home may be too high to mi-
are equivalent to the expected returns to labor in urban areas, grate. The median age of the African population is 18
accounting for the probability of employment. If rural resi- (United Nations, 2008), so even though households are large
dents lack information about the probability of urban employ- they may not include many workers who could migrate. Zezza,
ment, they may perceive the probability of urban employment Carletto, Davis, and Winters (2011) note that even if labor
as low. This problem may be exacerbated by the prevalence of markets functioned perfectly so that household members
informal labor in sub-Saharan Africa (Fox & Gaal, 2008). As could be replaced by hired labor, credit constraints preclude
a result, a larger rural–urban income gap would be necessary a household’s ability to substitute family for hired labor. In
to cause migration. Second, note that if individuals are also Malawi, where land pressure forces a substantial portion of
risk averse, then they would require an additional premium households to depend on alternative sources of income, rela-
to migrate if they perceive urban income as riskier than rural tively cheap informal labor substitutes exist either through ga-
income to compensate for the additional risk. 10 Risks need nyu labor, collective work parties, or simply helping out
not only be to income, either. For example, potential migrants relatives (Bezner Kerr, 2005; Goldberg, 2011). However, abil-
could be concerned about health risks. In Lesotho and Swazi- ities to exploit these available resources will still likely depend
land, HIV transmission is higher among migrant populations on household’s network ties with communal members or the
(Corno and de Walque, 2012). liquidity available to hire workers. 11
The NELM theory stresses the role of migrant networks as a Finally, there is little research on how anticipated compro-
potential determinant of migration, as they can reduce some of mises in individual or household well-being may serve as a
the uncertainties feared by risk-averse individuals, such as barrier to migration. A recent review on the literature of
resolving information asymmetries in distance labor markets. migration and happiness indicates several instances, primarily
However, the costs to moving are lowered as the network gets in Western countries and Asia, where migrants appear dissat-
larger (Carrington, Detraigiache, and Vishnawath, 1996) or isfied (Simpson, in press). Internal migrants in Ethiopia do not
improves in quality (Loury, 2006; Magruder 2010). There appear to have greater levels in measures of subjective well-
are good reasons why migrant networks may be weak in many being than their predecessors (de Brauw, Mueller, and Wolde-
African countries: road networks are poor, implying that tra- hanna, 2013b). A qualitative study in Ethiopia found the fam-
vel is often costly, particularly relative to rural household in- ily members of migrants also become discontent with the
come. Communication is also often difficult, as landline absence of fathers, strained family relations, and the preva-
phone networks are also poor and mobile phone coverage lence of HIV/AIDS (Feleke, Pankhurst, Bevan, and Lavers,
may not be available in many rural areas. Under this context, 2006).
facilitating communication could help strengthen existing At this juncture, we are unable to generalize the relevance
informal networks, and further encourage migration. For of social barriers on mobility decisions. As nationally
example, Muto (2009) finds a larger ethnic network and mo- representative multi-topic household surveys are collected in
bile phone coverage increases the chances of migration. Simi- sub-Saharan African countries, we can test the validity of
40 WORLD DEVELOPMENT

these factors on the migration decision. If preferences deter- Ideally, at least a portion of these surveys would be panels that
mine location decisions, we may need to rethink the role of pol- track migrants, using the design of the LSMS-ISA surveys as a
icy to stimulate migration for broader economic development. guide. Surveys that follow individuals over time and include
well-designed labor modules can help design better interven-
tions or policies to foster migration.
5. CONCLUSIONS: DIRECTIONS FOR FURTHER Second, from an economic perspective the most important
RESEARCH urban–rural linkage related to migration is through the remit-
tances that migrants send home. Yet in several countries
In this paper, we have discussed rural–urban migration in remittance rates appear to be quite low, at least among mi-
sub-Saharan Africa from several different perspectives. Migra- grants in the countries listed. It is first necessary to build a bet-
tion from rural to urban areas, we argue, may be an inherent ter understanding of the reasons that migrants do not remit. If
component of the development process. Yet using data from barriers such as high transaction costs hinder remittances, pi-
CIESEN (2004), we find that in a number of countries in lot experiments that attempt to lower remittance costs, poten-
Africa over the 1990s, rural–urban migration has actually been tially through SMS transmission as in Kenya, could lead to
relatively slow. The lack of more rapid rural–urban migration meaningful poverty reduction in rural areas with strong link-
in combination with large population shares in rural areas, ages to urban areas. Given that M-PESA replaced other remit-
may in fact hinder economic growth in the future. We show tance transmission systems in Kenya, mobile phone coverage
there are likely substantial monetary returns to mobility. Re- may be a limiting factor or other cell phone based systems
turns to labor in the agricultural sector as well as in rural areas do not work as well in other countries.
are lower than in other sectors and in urban areas. This The analysis in this paper also suggests several potential
suggests that better functioning labor markets could reduce roles for policy to help alleviate the gap between rural
poverty in many sub-Saharan African countries. and urban incomes, through migration. As noted by Dercon
Compounding the fact that migration rates are low, there and Collier (2011) in this volume, the current policy focus
are few data sets with good information on migration in ignores the fact that migration out of agriculture and rural
sub-Saharan Africa. As Foster and Rosenzweig (2008) note, areas more generally is necessary for growth in labor pro-
most available data sets are not adequate to study migration, ductivity and for narrowing the productivity gap between
because they do not ask about linkages to individuals who are agriculture and other sectors. Other than considering ways
within the country and have left the household recently nor do to improve the transmission of remittances across space,
they follow their behavior over time. Furthermore, a conse- we envision three further policy angles that can help reduce
quence of low migration rates is that even if questions were the rural–urban wage gap. The most obvious recommenda-
asked leading to the reliable identification of migrants, few re- tion is that any explicit laws or regulations that restrict
cent migrants would be found. As a result, in countries with the movement of people should be removed. Such restric-
low migration rates it is nearly impossible to study migration tions likely hinder economic growth. Second, it is worth
without a specialized migration survey, since even large sur- considering whether specific types of implicit restrictions,
veys will not include a large number of migrants (de Brauw such as ambiguous property rights over land, hinder migra-
& Carletto, 2012). tion. Additional migration might be an added benefit to
There is therefore a need for additional research potentially strengthening property rights so that land can be used as
along two dimensions. First, in high migration countries there collateral. Furthermore, we have also suggested an examina-
would be substantial research returns to including a migration tion of the social barriers to migration is crucial. Third,
module in the national multi-topic survey, assuming one is educational investments should not discriminate against rur-
conducted. Examples of survey modules tailored for multi- al residents, as migrants are clearly being selected on their
topic surveys that can be adapted to internal migration educational background in the countries we studied in Afri-
can be found in de Brauw and Carletto (2012). In countries ca. The payoff to improved educational opportunities for
considering policies to encourage migration, it would be worth rural residents may not be in agriculture or local work,
first understanding where the important sources of migrants but rather through increased migration to urban areas.
are, so that specialized migration surveys could be designed.

NOTES

1. For its Global Rural–Urban Mapping Project, CIESEN defines urban 3. Here we make the conservative assumption that rural and urban
areas using the NOAA night-time lights data set and buffered settlement population growth are equivalent; if rural population growth is naturally a
centroids when night lights are not sufficiently bright to define urban bit faster than urban population growth, more migrants would be found.
settlements, as is often the case in Africa. The project uses one definition, The point remains that it is difficult to study rural–urban population
so the definition of urban does not change over time nor does it depend movements when that movement is not very rapid.
upon official definitions. We therefore avoid the issue of reclassification of
areas from rural to urban areas. 4. Moreover, several of those countries experienced serious disturbances
leading to population movements during the 1990s, if one includes the
2. As mentioned, international migration is also a potential con- massive resettlement as the war ended in Mozambique (Rwanda, Somalia,
founding factor. However, international migration flows tend to be and the Sudan being the others).
small in much of Africa—West Africa being something of an outlier.
Our assumption of equal net birth rates in rural and urban areas 5. In certain contexts, both may be quite important; for example, in
likely understates rural–urban migration somewhat; assuming zero Ethiopia rural–rural migration may account for as much as two-thirds of
international migration implies an overstatement of rural–urban population movement to find employment (Central Statistical Agency,
migration. 2006, de Brauw et al., 2013b).
 THE ROLE OF RURAL–URBAN MIGRATION IN THE STRUCTURAL TRANSFORMATION OF SUB-SAHARAN AFRICA 41

6. Lucas (1997) also reviews the literature on internal migration. 9. See Azam and Gubert (2006) for a review of remittances from the
perspective of international migration. Where international migration is
7. Better local public goods might also attract migrants to urban areas. prevalent, international remittance rates are much higher. For example,
Ackah and Medvedev (2010) find that the incentive to migrate internally in Gallego and Mendola (2011) find that 60% of Mozambican migrants in
Ghana is highest for those that have reduced access to rural educational South Africa remit to their source households.
and health services, implying that those with the largest service differen-
tials between rural and urban areas are induced to migrate. This potential 10. Note that if migration is a household decision rather than an
rationale for migration has not been studied in many settings. individual decision, this rationale is mitigated as urban incomes are likely
to be uncorrelated with rural incomes, therefore decreasing the risk to
8. Urban and rural wage ratios are computed using the weighted overall income.
averages of monthly urban wages divided by the weighted averages of
monthly rural wages for each country. We computed these ratios for 11. Furthermore, there is a real knowledge gap in understanding the
countries where the unit of measurement did not differ within the country function of these informal labor markets; some which are a form of
(e.g., daily, weekly, monthly, etc.), with the exception of Uganda where migration.
wages were standardized across different units of measurement.

REFERENCES

Ackah, C., & Medvedev, D. (2010). Internal migration in Ghana: de Brauw, A., Mueller, V., & Woldehanna, T. (2013a). Motives to Remit:
Determinants and welfare impacts. Policy Research Working Paper Evidence from tracked internal migrants in Ethiopia. World Develop-
Series 5273. Washington, DC: World Bank. ment, 50, 13–23.
Akpan, N. S. (2012). From agriculture to petroleum oil production: What de Brauw, A., Mueller, V., & Woldehanna, T. (2013). Does internal
has changed about Nigeria’s rural development?. International Journal migration improve overall well-being in Ethiopia? IFRPI Ethiopia
of Developing Societies, 1(3), 97–106. Strategy Support Program Working Paper No. 55.
Azam, J.-P., & Gubert, F. (2006). Migrants’ remittances and the De Brauw, A., & Mueller, V. (2012). Do limitations in land rights
household in Africa: A review of the evidence. Journal of African transferability influence low mobility rates in Ethiopia?. Journal of
Economies, 15(Suppl. 2), 426–462. African Economies, 21(4), 548–579.
Bandiera, O., & Rasul, I. (2006). Social networks and technology adoption De Weerdt, J., & Dercon, S. (2006). Risk-sharing networks and insurance
in Northern Mozambique. Economic Journal, 116(514), 869–902. against illness. Journal of Development Economics, 81(2), 337–356.
Beauchemin, C. (2011). Rural–urban migration in West Africa: Towards a Dercon, S., & Collier, P. (2011). African agriculture in 50 years:
reversal? Migration trends and economic situation in Burkina Faso Smallholders in a rapidly changing world? Working paper. Oxford
and Cote D’Ivoire. Population, Space, and Place, 17, 47–72. University.
Beegle, K., de Weerdt, J., & Dercon, S. (2011). Migration and economic Deshingkar, P., & Grimm, S. (2004). Voluntary internal migration: An
mobility in Tanzania: Evidence from a tracking survey. Review of update. Overseas Development Institute working paper.
Economics and Statistics, 93(3), 1010–1033. Dillon, A., Mueller, V., & Salau, S. (2011). Migratory responses to
Bezner Kerr, R. (2005). Informal labor and social relations in Northern agricultural risk in Northern Nigeria. American Journal of Agricultural
Malawi: The theoretical challenges and implications of Ganyu Labor Economics, 93(4), 1048–1061.
for Food Security. Rural Sociology, 70(2), 167–187. Feleke, T., Pankhurst, A., Bevan, P., & Lavers, T. (2006). Migration and
Binswanger, H., & Rosenzweig, M. (1986). Behavioral and material rural–urban linkages in Ethiopia: Case studies of five rural and two urban
determinants of production relations in agriculture. Journal of Devel- sites in Addis Ababa, Amhara, Oromia and SNNP Regions and
opment Studies, 22(3), 503–539. Implications for Policy and Development Practice. Research Group
Bloom, D., Canning, D., Fink, G., Khanna, T., & Salyer, P. (2010). Urban on Wellbeing in Developing Countries, Ethiopia Programme, Univer-
settlement: Data, measures, and trends. UNU-WIDER Working Paper sity of Bath.
No. 2010/12. Foster, A., & Rosenzweig, M. (2001). Imperfect commitment, altruism,
Byerlee, D. (1974). Rural–urban migration in Africa: Theory, policy, and and the family: Evidence from transfer behavior in low-income rural
research implications. International Migration Review, 8(4), 543–566. areas. Review of Economics and Statistics, 83(3), 389–407.
Carrington, W., Detraigiache, E., & Vishnawath, T. (1996). Migration Foster, A., & Rosenzweig, M. (2008). Economic development and the
with endogenous moving costs. American Economic Review, 86(4), decline of agricultural employment. Handbook of Development Eco-
909–930. nomics, 4, 3051–3083.
Center for International Earth Science Information Network (CIESEN). Fox, L., & Gaal, M. S. (2008). Working out of poverty: Job creation and the
(2004). Global rural–urban mapping project. Available at: <http:// quality of growth in Africa. Washington, DC: World Bank.
sedac.ciesin.columbia.edu/gpw/index.jsp>. Gallego, J., & Mendola, M. (2011). Labor migration and social networks
Central Statistical Agency (CSA) (2006). The 2005 National Labour Force participation: Evidence from Southern Mozambique. INSIDE Paper
Survey. Addis Ababa, Ethiopia: The Federal Democratic Republic of No. 28. Institute for Economic Analysis.
Ethiopia. Goldberg, J. (2011). Kwacha gonna do? Experimental evidence about labor
Conley, T., & Udry, C. (2010). Learning about a new technology: supply in Rural Malawi. Working paper. University of Maryland
Pineapple in Ghana. American Economic Review, 100(1), 35–69. Department of Economics.
Corno, L., & de Walque, D. (2012). Mines, migration and HIV/AIDS in Gollin, D., Lagakos, D., & Waugh, M. (2011). The agricultural produc-
Southern Africa. Journal of African Economies, 21(3), 465–498. tivity gap in developing countries. Working paper. Williams College.
Cox, D., Eser, Z., & Jimenez, E. (1998). Motives for private transfers over Gubert, F. (2002). Do migrants insure those who stay behind? Evidence
the life cycle: An analytical framework and evidence from Peru. from the Kayes area. Oxford Development Studies, 30(3), 267–287.
Journal of Development Economics, 55(1), 57–81. Harris, J., & Todaro, M. (1970). Migration, unemployment, and devel-
Cox-Edwards, A., & Ureta, M. (2003). International migration, remit- opment: A two-sector analysis. American Economic Review, 60(1),
tances, and schooling: Evidence from El Salvador. Journal of Devel- 126–142.
opment Economics, 72(2), 429–462. Jack, W., & Suri, T. (2010). The economics of M-PESA. Working paper.
de Brauw, A., & Carletto, G. (2012). Improving the measurement and policy Massachusetts Institute of Technology Sloan School.
relevance of migration information in multi-topic household surveys. Jedwab, R. (2011). Why is African urbanization different? Evidence from
Electronically published by the World Bank. Available at: <http:// resource exports in Ghana and Ivory Coast. Mimeo. Available online at:
go.worldbank.org/4GH8S56F70>. <http://home.gwu.edu/~jedwab/JEDWAB_JMP_14112011.pdf>.
42 WORLD DEVELOPMENT

Kessides, C. (2007). The urban transition in sub-Saharan Africa: Implica- Rozelle, S., Taylor, J. E., & de Brauw, A. (1999). Migration, remittances,
tions for economic growth and poverty reduction. Washington, DC: The and agricultural productivity in China. American Economic Review,
Cities Alliance. 89(2), 287–291.
Klonner, S., & Nolen, P. (2010). Cell phones and rural labor markets: Sandefur, J., Serneels, P., & Teal, F. (2007). African poverty through the
Evidence from South Africa. Working paper. lens of labor economics: Earnings and mobility in three countries. In P.
Krishnan, P., & Sciubba, E. (2009). Links and architecture in village Paci, & P. Serneels (Eds.), Employment and shared growth. Washing-
networks. The Economic Journal, 119, 917–949. ton, DC: World Bank.
Loury, L. D. (2006). Some contacts are more equal than others: Informal Simpson, N. (in press). Happiness and migration. In K. Zimmerman & A.
networks, job tenure, and wages. Journal of Labor Economics, 24(2), Constant (Eds.), Chapter for the International Handbook
299–318. on the Economics of Migration. Amsterdam: Edward Elgar Publishing
Lucas, R. E. B. (1997). Internal migration in developing countries. In M. Ltd.
Rosenzweig, & O. Stark (Eds.), Handbook of family and population Stark, O. (1991). The migration of labor. Boston: Basil Blackwell.
economics North Holland: Amsterdam. Stark, O., & Bloom, D. E. (1985). The new economics of labor migration.
Lucas, R. E. B. (2000). Migration. M. Grosh, &P. Glewwe (Eds.), American Economic Review, 75(2), 173–178.
Designing household survey questionnaires for developing countries (Vol. Taylor, J. E., & Martin, P. (2001). Human capital: Migration and rural
2, pp. 49–82). population change. In G. Rausser, & B. Gardner (Eds.), Handbook of
Luke, N., & Munshi, K. (2006). New roles for marriage in Urban Africa: agricultural economics (pp. 458–511). New York: Elsevier.
Kinship networks and the labor market in Kenya. Review of Economics Todaro, M. (2000). Urbanization, unemployment, and migration in
and Statistics, 88(2), 264–282. Africa: Theory and policy. In D. Ghai (Ed.), Reviewing social and
Magruder, J. (2010). Intergenerational networks, unemployment, and economic progress in Africa. New York: United Nations.
persistent inequality in South Africa. American Economic Journal: United Nations. (2008). Trends in sustainable development: Africa report.
Applied Economics, 2(1), 62–85. Department of Economic and Social Affairs, Division for Sustainable
McMillan, M. (2011). A contribution to the empirics of structural Development. United Nations, New York.
transformation in Africa. Working paper. International Food Policy United Nations High Commission for Refugees. (2011). Available at:
Research Institute. <http://www.unhcr.org/cgi-bin/texis/vtx/home> Accessed 15.04.11.
McMillan, M., & Rodrik, D. (2011). Globalization, structural change, and World Bank (2008). World development report 2008. Washington, DC:
productivity growth. Working paper. International Food Policy World Bank.
Research Institute. World Bank. (2010). The Ethiopian urban migration study 2008: The
Mogues, T., Morris, M., Freinkman, L., Adubi, A., Ehui, S., with Nwoko, characteristics, motives and outcomes of migrants to Addis Ababa.
C., Taiwo, O., Nege, C., Okonji, P., Chete, L. (2008). Agricultural Poverty Reduction and Economic Management, Africa Region.
spending in Nigeria. International Food Policy Research Institute Report No. 55731-ET.
Discussion Paper No. 789. World Bank. (2012a). International income distribution database (I2D2).
Muto, M. (2009). The impacts of mobile phone coverage expansion and World Bank. (2012b). World development indicators.
personal networks on migration: Evidence from Uganda. In Paper World Bank. (2012c). Migration and remittances surveys. Available at:
presented at the International Association of Agricultural Economists <http://econ.worldbank.org/WBSITE/EXTERNAL/EXTDEC/EXT-
Conference, Beijing, China, August 16–22, 2009. DECPROSPECTS/0,,contentMDK:22759429~pagePK:64165401pi
National Population Policy of Ethiopia (1993). National Population Policy PK:64165026theSitePK:476883,00.html>.
of Ethiopia. Addis Ababa, Ethiopia: Office of the Prime Minister (April Wouterse, F., & Taylor, J. E. (2008). Migration and income diversifica-
1993). Available at: <https://cyber.law.harvard.edu/population/poli- tion: Evidence from Burkina Faso. World Development, 36(4),
cies/ETHIOPIA.htm>. 625–640.
Owusu, G. (2005). Small Towns in Ghana: Justifications for their Yang, D. (2008). International migration, remittances, and household
promotion under Ghana’s Decentralisation Programme. African investment: Evidence from Philippine migrants’ exchange rate shocks.
Studies Quarterly, 8(2), 48–69. Economic Journal, 118(528), 591–630.
Paulson, A. (2000). Insurance motives for migration: Evidence from Yang, D. (2011). Migrant remittances. Journal of Economic Perspectives,
Thailand. Working paper. Northwestern University. 25(3), 129–152.
Potts, D. (2009). The slowing of sub-Saharan Africa’s urbanization: Zezza, A., Carletto, C., Davis, B., & Winters, P. (2011). Assessing the
Evidence and implications for urban livelihoods. Environment and impact of migration on food and nutrition security. Food Policy, 36(1),
Urbanization, 21, 253–259. 1–6.
Potts, D. (2012). Whatever happened to Africa’s rapid urbanisation? Africa Zhang, Y. (2006). Urban–rural literacy gaps in sub-Saharan Africa: The
Research Institute Working Paper. Available on: <http://www.africa- roles of socioeconomic status and school quality. Comparative Educa-
portal.org/dspace/articles/whatever-happened-africas-rapid-urbanisa- tion Review, 50(4), 581–602.
tion> Accessed 01.01.13.
Pritchett, L. (2006). Let their people come: Breaking the gridlock on global
labor mobility. Washington, DC: Center for Global Development.

Available online at www.sciencedirect.com

ScienceDirect