Patterns of Population Growth in Peri-Urban Accra, Ghana

20
Patterns of Population Growth in Peri-Urban Accra, Ghana PETRA DOAN and CHARLES YAW ODURO AbstractUrbanization in Africa (and the wider developing world for that matter) has often been misunderstood by Western observers on the lookout for Western-style cities and suburbs. In these parts of the world, rapid urbanization has led to continuing changes in the form and shape of cities, as peri-urban zones shift swiftly from rural to urban. While some descriptions of the peri-urban zone suggest an amorphous area filled with low-income residents, others argue that the zone is highly diverse. Using population census data, spatial modeling and regression analysis, we show that urban expansion at the edge of the city of Accra, Ghana, is not amorphous and does indeed show some discernible patterns. These patterns are represented by four hypotheses tested in this study — the spreading pancake, development node, village magnet and ribbon hypotheses. While the assumption that urban growth occurs in concentric rings around a central city (represented by the spreading pancake model) holds for Accra, this pattern of growth combines with other patterns to create a still-evolving urban form in the city’s peri-urban zone. These include clustering of growth around a port city, a number of old villages and along improved highways radiating from the city. Introduction African urban development has had a long and often misunderstood history. Because the first European explorers did not find Western-style cities, they assumed that Africa had no indigenous urban traditions. They completely misread the evidence of ancient and existing cities. Coquery-Vidrovitch (2005) suggests that the form of traditional African cities contradicted the formality of walled European cities centered on a church, a palace and a central market. Instead, cities in Africa demonstrated the flexibility to adapt to shifting environmental conditions and ‘expressed a specific social organization, based on the extended family’ (ibid.: 51). This adaptable and family-based organizational form is still characteristic of many African cities, despite the distortions caused by the Atlantic slave trade and rigid segregation during colonization (Myers and Owusu, 2008). In the post-colonial era in Africa, rural–urban migrants overwhelmed the capacity of public authorities to provide shelter and essential services, settling in large informal settlements at the urban fringe. Some scholars continued to misread the African urban process and described it as a huge sea of informality. For example, Davis (2006: 5–6) writes about an emerging conurbation of 300 cities between Accra and Benin City (Nigeria) that will become ‘the biggest single footprint of urban poverty on earth’. Other scholars acknowledge that although this sprawling settlement pattern threatens fragile ecosystems, it has ‘created a basis for a more organic and indigenous form of urbanism’ (Stren and Halfani, 2001: 474). Simone (2004: 2) suggests cities reflect the ‘African love affair with the hodgepodge’ in which external economic forces have exacerbated an urban form that stands ‘aside from the usual notions of growth and development’. Mabogunje (1990: 121–2) claims that: Volume 36.6 November 2012 1306–25 International Journal of Urban and Regional Research DOI:10.1111/j.1468-2427.2011.01075.x © 2011 Urban Research Publications Limited. Published by Blackwell Publishing, 9600 Garsington Road, Oxford OX4 2DQ, UK and 350 Main St, Malden, MA 02148, USA

Transcript of Patterns of Population Growth in Peri-Urban Accra, Ghana

Patterns of Population Growth inPeri-Urban Accra, Ghana

PETRA DOAN and CHARLES YAW ODURO

Abstractijur_1075 1306..1325

Urbanization in Africa (and the wider developing world for that matter) has often beenmisunderstood by Western observers on the lookout for Western-style cities and suburbs.In these parts of the world, rapid urbanization has led to continuing changes in the formand shape of cities, as peri-urban zones shift swiftly from rural to urban. While somedescriptions of the peri-urban zone suggest an amorphous area filled with low-incomeresidents, others argue that the zone is highly diverse. Using population census data,spatial modeling and regression analysis, we show that urban expansion at the edge ofthe city of Accra, Ghana, is not amorphous and does indeed show some discerniblepatterns. These patterns are represented by four hypotheses tested in this study — thespreading pancake, development node, village magnet and ribbon hypotheses. While theassumption that urban growth occurs in concentric rings around a central city(represented by the spreading pancake model) holds for Accra, this pattern of growthcombines with other patterns to create a still-evolving urban form in the city’speri-urban zone. These include clustering of growth around a port city, a number of oldvillages and along improved highways radiating from the city.

IntroductionAfrican urban development has had a long and often misunderstood history. Because thefirst European explorers did not find Western-style cities, they assumed that Africa hadno indigenous urban traditions. They completely misread the evidence of ancient andexisting cities. Coquery-Vidrovitch (2005) suggests that the form of traditional Africancities contradicted the formality of walled European cities centered on a church, a palaceand a central market. Instead, cities in Africa demonstrated the flexibility to adapt toshifting environmental conditions and ‘expressed a specific social organization, based onthe extended family’ (ibid.: 51). This adaptable and family-based organizational form isstill characteristic of many African cities, despite the distortions caused by the Atlanticslave trade and rigid segregation during colonization (Myers and Owusu, 2008).

In the post-colonial era in Africa, rural–urban migrants overwhelmed the capacity ofpublic authorities to provide shelter and essential services, settling in large informalsettlements at the urban fringe. Some scholars continued to misread the African urbanprocess and described it as a huge sea of informality. For example, Davis (2006: 5–6)writes about an emerging conurbation of 300 cities between Accra and Benin City(Nigeria) that will become ‘the biggest single footprint of urban poverty on earth’. Otherscholars acknowledge that although this sprawling settlement pattern threatens fragileecosystems, it has ‘created a basis for a more organic and indigenous form of urbanism’(Stren and Halfani, 2001: 474). Simone (2004: 2) suggests cities reflect the ‘Africanlove affair with the hodgepodge’ in which external economic forces have exacerbatedan urban form that stands ‘aside from the usual notions of growth and development’.Mabogunje (1990: 121–2) claims that:

bs_bs_banner

Volume 36.6 November 2012 1306–25 International Journal of Urban and Regional ResearchDOI:10.1111/j.1468-2427.2011.01075.x

© 2011 Urban Research Publications Limited. Published by Blackwell Publishing, 9600 Garsington Road, Oxford OX42DQ, UK and 350 Main St, Malden, MA 02148, USA

The African city remains today a human agglomeration with no clear set of criteria to help itsidentification as a socially distinct entity . . . What colonialism produced in most parts ofAfrica, and especially in its cities, was a syncretized society caught between its traditionalpre-capitalist roots and a capitalist-oriented colonial economy.

As Yeboah (2000; 2003) notes, urban expansion in Africa has not been adequatelyresearched. This article seeks to correct previous misconceptions about an amorphousand ill-defined peri-urban zone through a broad-brush analysis of the patterns of urbanchange in the peri-urban zone of the Greater Accra Metropolitan Area (GAMA), Ghana.In particular, the research analyzes the growth of population at the urban fringe from1970 to 2000, using census data for a sample of 105 localities in the GAMA. While suchcensus data do not always capture precise urban population numbers due to undercountsof migrant and temporary populations, they represent the best available picture ofthe extent of urbanization in the capital region of Ghana. The authors develop fourhypotheses about the possible evolution of the urban form — the spreading pancake,development node, village magnet and ribbon hypotheses — by combining ordinary leastsquare (OLS) regression with ArcGIS spatial modeling. The methodology tests whetherpopulation growth depends on the distance of a locality from: (1) the CBD of Accra (thespreading pancake hypothesis); (2) the port city of Tema (the development nodehypothesis); (3) five relatively large peri-urban localities (the village magnet hypothesis);or (4) highways (the ribbon hypothesis). In the discussion section, the possible influenceof other less quantifiable factors is also considered. Finally, a set of policy and planningrecommendations for peri-urban development in Ghana and elsewhere in Africa ispresented.

Peri-urban growth patternsPeri-urban development in developing countries is partly a result of rural–urbanmigration that has led to an increase in the concentration of population and humanactivities at the urban fringe of many third world cities (Mbiba and Huchzermeyer, 2002;Allan, 2003; Torres, 2007). In Latin America, for example, Torres (2007: 3) states that:

Most peri-urban growth in Latin America results from massive rural land reclamationby migrants trying to settle in poor tenements. Such areas are not only typically ill-regulatedand distant from key employment hubs, but also present appalling sanitation conditionsand significant environmental problems, including deforestation and pollution of rivers andstreams. Metropolitan fragmentation — characteristic of many metro areas, which lackplanning coordination and proper funding for their peri-urban municipalities — adds to thealready unacceptable living conditions of the newly arrived.

While many descriptions of the peri-urban zone suggest an amorphous fringe area filledwith recently arrived low-income residents, others (e.g. Browder et al., 1995) havesuggested that the urban fringe is associated with diverse communities with mixedincomes and new as well as long-term residents. In West Africa, Grant (2009) describesperi-urban development that contains both upscale residential developments and low-income self-built housing. Unfortunately, much of this literature about peri-urbandevelopment does not rigorously examine the patterns of peri-urban development.

Expansion of peri-urban growth

Chinese planners have described rapid urban growth around some cities as a kind ofspreading pancake (Laquian, 2005). Several scholars writing about Africa havesuggested a growth process that is reminiscent of von Thunen’s theoretical model ofagricultural land use, with the urban fringe spreading outwardly in a more or less

Patterns of population growth in peri-urban Accra, Ghana 1307

International Journal of Urban and Regional Research 36.6© 2011 Urban Research Publications Limited

concentric pattern (Grant and Yankson, 2003). These scholars (e.g. Gough and Yankson,2006) suggest that the waves of population growth precede infrastructure provision inperi-urban areas because the limited capacity of local governments is unable to keep pacewith the rate of growth. Consequently, peri-urban lands get settled long before basicamenities like water, electricity and schools are provided. Simon et al. (2004: 236) reportthat:

Many indigenous villages, previously located in rural areas a considerable distance from thecity [Kumasi, Ghana], have experienced in-migration, growth, changes in populationcomposition, land use and economic base. As a generalization, the closer the city comes, themore pronounced is the transition from ‘rural’ to ‘urban’ characteristics. Eventually, thesesettlements become part of the built-up urban area, which then comprises a complex mixtureof formal houses, shanties, rural huts and other dwellings.

One common thread that connects these examples is the assumption of a dominantcentral city that influences the expansion of its fringe areas. Peri-urban development thusspreads outwards from a central city in a pancake-like pattern. While such a model isuseful at a broad scale, it fails to provide much new information about the variations thatoccur within peripheral areas.

The influence of global investment on peri-urban development

A number of scholars cite globalization and trade liberalization as major forcesinfluencing peripheral urbanization in Latin America, Asia and Africa (e.g. McGee,1991; Yeboah, 2000; Grant, 2009). For example, McGee (1991) observed an emergingurban form in Pacific Asia, which he describes as an extended metropolitan region(EMR) or ‘desakota’. This zone is characterized by the fusion of cities and outlyingvillages through the functional integration of urban and rural economies and thedevelopment of transportation systems (Drakakis-Smith, 2000). Singapore exemplifiesthis development:

An example of an extended metropolitan region is Singapore and its surrounding area, whichincludes the Malaysian state of Johor and the Indonesian Archipelago of Riau. In this region,market interchanges have been growing and links with areas once considered rural are gainingstrength. It is characterized by high levels of economic diversity and interaction, a highpercentage of non-farm employment (over 50%), and a deep penetration of global marketforces into the countryside (McGee, 1997: 35).

In Latin America, Gilbert (1993) describes a kind of ‘polarization reversal’ thatrepresents two phenomena: the slowing down of population growth in large cities whileother cities and towns within the national settlement system experience fast growth; anda situation where growth slows down in the core of large cities but accelerates at otherlocations within the metropolitan regions of those cities. Both situations are attributed toeconomic recessions, structural adjustment and globalization (Yeboah, 2000).

Often the impact of globalization is most acutely evident in development nodes thatemerge in urban areas, seeking to attract investments to specific sites in the peri-urbanzone through the establishment of ‘special economic zones’ and other policy initiatives.Mabogunje (1990: 151) suggests that a preference for import substitution in the vicinityof port cities has resulted in ‘the burgeoning of the population of many port citiesas country after country engaged in this form of industrialization immediately afterindependence’. An example is South Africa’s Richards Bay, which grew from a smallfishing village to a town of 28,000 inhabitants within 10 years of the government’sdecision to develop it into a port and industrial town (Wiese, 1981). In the peri-urbanzone, such nodes often result in uneven economic growth, which attracts the ‘haves’ tospeculate in land in anticipation of future development, or build before infrastructure is

1308 Petra Doan and Charles Yaw Oduro

International Journal of Urban and Regional Research 36.6© 2011 Urban Research Publications Limited

provided (Grant and Yankson, 2003). Eventually, the peri-urban zone shifts from a zoneof survival to an investment hub (Briggs and Mwamfupe, 2000).

Peri-urban growth and existing villages

Peri-urban development also appears to be attracted to pre-existing villages thatalready have basic levels of critical services, though this may result in spatialsegregation with the poor living in the least accessible areas (Torres, 2007). Forexample, in the Dakar metropolitan area, new municipalities such as Pikine ‘consist ofsome historical towns and villages as the focal points’ (Simone, 2004: 38) that act asmagnets for fast residential development (Grant and Yankson, 2003). Another exampleis the village of Dong Mei, which was surrounded by and eventually incorporated intothe peri-urban zone of Guangzhou, China; existing social capital in the village enabledresidents to adapt to rapid socioeconomic development (Leaf, 2002). In general, itappears that pre-existing villages have a variety of advantages and often become thenuclei of fast-growing densely populated pockets surrounded by slow-growing sparselypopulated areas.

Transportation and peri-urbanization

Other scholars have suggested that peri-urban development follows major roads linkingthe central city to other urban areas and rural surroundings. The presence of such roadsencourages residents to set up small-scale ‘roadside’ businesses, especially at majorintersections (see Mabogunje, 1990; Oduro, 2010). Mabogunje (1990: 145) argues thatthe construction of highways in ‘peri-urban and suburban districts has been an importantmechanism for bringing more and more land within the ambit of the emerging capitalistmode of production’. In Tanzania, for example, road improvements in the peri-urbanzone have had a marked influence on the urban form (see Briggs and Mwamfupe, 2000).In addition, improved roads enable peri-urban residents to commute to the central cityand other locations within the metropolitan area where their livelihoods are located (seeGrant and Yankson, 2003; Bryceson, 2006). Finally, better roads in the peri-urban zoneare likely to attract middle-class residents, at the expense of the poor who must locatefarther away from such roads (Kombe, 2005).

Background on the Accra metro areaAs of 2000, the Greater Accra Metropolitan Area (GAMA) consisted of the AccraMetropolis, Tema Municipality and Ga District. It is located in the Greater AccraRegion (GAR), Ghana’s smallest in land area but the most urbanized and amongthe fastest-growing in the country (see Table 1 and Figure 1). With a population of

Table 1 Population growth rates (% per annum) of Ghana, Greater Accra Region and City ofAccra, 1950–2000

PeriodGhana Ghana Greater

GAMACity

(Total) (Urban) Accra Region of Accra

1950–60 2.5 6.6 — — —

1960–70 2.4 4.7 5.1 6.5 5.9

1970–84 2.6 3.3 3.3 3.7 2.7

1984–2000 2.7 4.7 4.4 3.5 3.2

Source: Computed by authors based on census data

Patterns of population growth in peri-urban Accra, Ghana 1309

International Journal of Urban and Regional Research 36.6© 2011 Urban Research Publications Limited

2,905,726 the GAR was the second most populous region in 2000 (Ghana StatisticalService, 2005). Most of this population was made up of migrants from other regionswho had settled in Accra. The population of the city itself increased from 377,558 in1960 to 1,657,856 in 2000 (Twum-Baah, 2000). As has been the case in some sub-Saharan African countries in recent times (see Potts, 2009), urban population growth inGhana slowed between 1970 and 1984, probably due to the economic crises andpolitical instability in the 1970s and 1980s that resulted in a decline in rural–urbanmigration and a surge in international out-migration (see Table 1). However, the growthof Ghana’s urban population, including that of GAMA and the city of Accra,

Figure 1 Base map of Greater Accra area (map created by Charles Yaw Oduro)

1310 Petra Doan and Charles Yaw Oduro

International Journal of Urban and Regional Research 36.6© 2011 Urban Research Publications Limited

accelerated again between 1984 and 2000 as the economy recovered following theadoption of structural adjustment programs (SAPs) and liberalization policies (Yeboah,2000; 2003; Yankson et al., 2004; Grant, 2009).

With an area of 241 km2 (Twum-Baah, 2000), Accra’s capacity to contain populationgrowth has been exceeded, leading to spillovers into peri-urban areas (Yeboah, 2000;2003; Yankson et al., 2004). In addition, trade liberalization has eased access forimported building materials, foreign exchange restrictions and transfer of remittances(Yankson et al., 2004). Consequently, the peri-urban zone has experienced a high rate ofhousing construction by residents, expatriate Ghanaians and recently arrived migrants inneed of cheap residential accommodation. In one study Gough and Yankson (2006: 198)observed that:

Whereas only a decade ago peri-urban Accra consisted of dispersed rural settlements wheresubsistence agriculture was widely practiced, today these settlements are surrounded by newhousing at various stages of construction and little farmland remains. Richer households aremoving into the newly completed houses, whereas poorer households seek rooms to rent, oftenwithin the indigenous settlements.

Rapid peri-urban growth in Accra is characterized by a complex land market thatcombines elements of customary land tenure and formal land-management procedures(Durand-Lasserve and Selod, 2007), as well as a shift in societal perspective on land froma social good to a tradable commodity. One of the impacts of the formalization ofcustomary land tenure is the emergence of a small but growing formal housing marketoperated by private real estate development companies in and around the city.Architecturally, there is a steady shift from compound (extended family-oriented) housestowards bungalow-type (single family-oriented) houses (see Grant, 2009). In addition,peri-urban development has overwhelmed the local planning authorities (municipalassemblies) to the extent that GAMA’s built-up area has grown faster than its populationand has done so in a haphazard manner. Between 1985 and 2002, for instance, Yanksonet al. (2004) estimate that the built-up area increased 2.6 times, although populationincreased 1.9 times. Yeboah (2000; 2003: 107, 109) has examined this expansion ofhousing, describing it as ‘quality residential sprawl with unicentric tendencies’ and a‘spontaneous and unplanned’ development that has occurred in all directions from thecentral city. It is essential to understand the pattern of Accra’s growth in order to improvepolicy and planning.

Data and methodsMost of the explanations of peri-urban expansion in the developing world have not beenformally evaluated. This article provides a more rigorous approach to peri-urbandevelopment by analyzing changes in population growth throughout the GAMA to testfour hypotheses derived from the literature on peri-urban development. For the sake ofclarity, these hypotheses have been named the spreading pancake, development node,village magnet and ribbon hypotheses. The analysis tests these hypotheses (describedbelow in greater detail) by combining statistical analysis and spatial modeling usingArcGIS. The unit of analysis for the study is the locality, and population growth in eachlocality is used as a proxy for the urbanization process in the peri-urban fringe. Localityis defined by the Ghana Statistical Service as any nucleated physically distinct humansettlement ranging from a village to a city (Census Office, 1972). This study used asample of 105 localities, for which data were available in 1970, 1984 and 2000. As of2000, they ranged in population from 47 to 150,312 and accounted for 83% of GAMA’spopulation.

Patterns of population growth in peri-urban Accra, Ghana 1311

International Journal of Urban and Regional Research 36.6© 2011 Urban Research Publications Limited

Spreading pancake hypothesis

This hypothesis suggests that a locality in that part of the peri-urban zone closest to acentral city is likely to have experienced the fastest population growth and have thehighest population density. With increasing distance from the center, the rate ofpopulation growth and population density is likely to decline in a series of concentricrings about the central city in the form of a spreading pancake. In the context of thisstudy, the dominant central city is Accra, and the spreading pancake of peri-urbanizationis measured by the growth of localities outside the city’s boundary but still within theGAMA.

Development node hypothesis

The development node hypothesis indicates that localities close to a cluster of globalinvestments (e.g. ‘special economic zones’) are likely to experience faster populationgrowth and higher population density than other localities further from that developmentnode. In GAMA, the locality that fits this description is Tema, because global economicinterests led to its selection as the site for a new seaport and industrial city. With an‘Export Processing Zone’ (EPZ) intended to ‘target companies worldwide to locate in theTema enclave’ (Grant and Yankson, 2003: 69), Tema is a gateway through which theGhanaian economy is integrated into the global economy.

Village magnet hypothesis

This hypothesis predicts that localities in close proximity to existing villages that havebeen engulfed by rapid urbanization will have higher rates of population growth andhigher population density than places that are further away from the village magnets. Inthis article, Ashiaman and Adenta in Tema Municipality, and Dome, Gbawe and Madinain Ga District were selected for study as examples of village magnets. These localitieshad populations ranging from 700 to 23,000 as of 1970, and were among localities thatregistered dramatic population growth between 1970 and 2000. Dome and Gbawe grewby close to 4,000%, Adenta and Madina grew by over 900%, while Ashiaman grew byalmost 570%. Each of them had at least a primary school, a middle school and/or a healthclinic as of 1984. In addition, Ashiaman, Dome and Madina are major centres of informalsmall-scale industrial and commercial activities.

Ribbon hypothesis

This hypothesis anticipates that peri-urban communities located along highways arelikely to experience faster rates of population growth and higher population density thanplaces that are farther from the road. This article calculates the distance of each localityin the study area from an improved highway based on analysis conducted using GISmaps of the region.

As mentioned earlier, the four hypotheses were tested through the use of both spatialmodeling and regression analyses. The spatial modeling involved the calculation andvisual depiction of the change in population density over time and space relative to thecity of Accra, the development node of Tema, the five village magnets and highwaysradiating outwards from Accra through the peri-urban zone. This procedure used ArcGISto divide up the entire peri-urban zone into 3 km or 4 km wide bands around Accra,Tema, the five village magnets and the highways. After calculating the populationdensity of each band in 1970, 1984 and 2000, the results were then presented in the formof maps and line graphs.

The regression analysis employed the ordinary least square (OLS) regression methodto assess whether the spatial pattern of population growth could be explained by the fourhypotheses. As noted above, absolute population change served as the dependent

1312 Petra Doan and Charles Yaw Oduro

International Journal of Urban and Regional Research 36.6© 2011 Urban Research Publications Limited

variable, and the main independent variables were the distances from the centroid of eachlocality to various locations including the CBD of Accra, Tema, Ashiaman, Adenta,Dome, Gbawe and Madina, as well as distance from the centroid of each locality to thenearest highway.

ResultsDescriptive statistics for the dependent and independent variables show that localities inthe peri-urban zone experienced significant growth during the time period under review(see Table 2). The population of the average locality grew by almost 1,600 peoplebetween 1970 and 1984, and by 5,600 people between 1984 and 2000.

Population density around Accra also increased dramatically during the study (seeFigures 2 and 3). Between 1970 and 2000, about 99% of the peri-urban population livedwithin 20 km of the city. By the year 2000, the 20 largest localities (with populationsranging from 7,100 to 150,300 and including Tema, Ashiaman, Madina, Adenta, Domeand Gbawe) were within 10 kms of the city. Also, while population density increasedthroughout the metropolitan area, there was a marked decline in population density withincreasing distance from the city during the period.

The results from a series of bivariate regression analyses reveal that each independentvariable has a negative and statistically significant relationship to absolute populationgrowth (see Table 3). As distance increases from the city center, from Tema, from eachvillage magnet and from an improved highway, absolute population growth of a localitydecreases. However, while these results suggest that the independent variables separatelyhelp to explain some of the growth patterns in peri-urban Accra, multivariate regressionmodels tell a more nuanced story.

Table 2 Summary statistics (105 localities)

Variables Mean Std. Dev.

Absolute population change 1970–84 (000s) 1.6 5.6

Absolute population change 1984–2000 (000s) 5.6 13.6

Distance from Accra CBD (km) 16.0 5.7

Distance from Tema (km) 25.6 9.4

Distance from Ashiaman, Adenta, Dome, Gbawe or Madina (km) 7.5 5.0

Distance from Highway (km) 3.9 3.6

Source: Authors’ construct based on GIS analysis and census data

Figure 2 Population density by distance from Accra, 1970, 1984 and 2000 (source: authors’construct based on census data)

Patterns of population growth in peri-urban Accra, Ghana 1313

International Journal of Urban and Regional Research 36.6© 2011 Urban Research Publications Limited

A multiple regression model was estimated for three time periods: 1970–84 (model 1),1984–2000 (model 2) and 1970–2000 (model 3). The coefficient for distance from Accrawas positive and statistically significant in each model (models 1–3), which was notexpected (see Table 4a). However, distance from the development node (Tema) isnegative as expected and is significant during the earlier period (model 1) and in theoverall (model 3). The coefficient for distance from a highway is negative and significantin each model as expected. The coefficients for distances from the village magnets aremixed and harder to interpret, because some are positive and others are negative withvarying levels of significance.

Figure 3 Population density by proximity to Accra, 1970, 1984 and 2000 (source: authors’construct based on census data)

Table 3 Coefficients of bivariate regression of absolute population change on distancevariables in 1970–84, 1984–2000 and 1970–2000

Independent Variables

Regression Coefficients (Beta)1970–84 1984–2000 1970–2000(Model 1) (Model 2) (Model 3)

Distance from Accra (km) -0.133* -0.227** -0.209**

Distance from Tema (km) -0.406*** -0.436*** -0.448***

Distance from nearest village magnet (km) -0.238** -0.396*** -0.366***

Distance from Highway (km) -0.265*** -0.318*** -0.3168***

N 105 105 105

Significance test: *P < 0.1, **P < 0.05, ***P < 0.01Source: Authors’ construct based on GIS analysis and census data

1314 Petra Doan and Charles Yaw Oduro

International Journal of Urban and Regional Research 36.6© 2011 Urban Research Publications Limited

These somewhat anomalous results required further analysis. The researchersre-specified the models because a peri-urban locality is likely to be most stronglyinfluenced only by the nearest village magnet. Therefore another set of statisticalanalyses were conducted, using distance of a locality only to the nearest village magnetfor the same periods: 1970–84 (model 4), 1984–2000 (model 5) and 1970–2000 (model6) (see Table 4b). In these models, the coefficients for distance from Accra remainpositive but are no longer statistically significant. At the same time, the coefficient fordistance from a village magnet is negative and significant for each model, except inthe earliest period 1970–84 (model 4). Because the analysis uses standardized betacoefficients, it is also possible to compare the relative size of the coefficients, whichindicates that Tema had a large and significant influence during the early period, but bythe later period its influence was supplanted by the influence of other rapidly growingmagnets. Distance from a highway also loses significance because the selected villagemagnets are located close to highways, indicating that more remote localities travel first

Table 4a Regression of absolute population change on distance variables in 1970–84,1984–2000 and 1970–2000 (including distance from each village magnet)

Independent Variables

Regression Coefficients (Beta)1970–84 1984–2000 1970–2000(Model 1) (Model 2) (Model 3)

Distance from Accra (km) 1.387*** .838* 1.045**

Distance from Tema (km) -2.66*** -.540 -1.212*

Distance from Ashiaman (km) -.679** -.459 -0.230

Distance from Madina (km) -1.278 -1.497* -1.502*

Distance from Adenta (km) 1.047 1.709*** 1.588**

Distance from Dome (km) .102 .056 .073

Distance from Gbawe (km) -.951* -.743 -.842*

Distance from Highway (km) -.232* -.213* -.229*

R-Square 0.3403 0.2923 0.3102

N 105 105 105

Significance test: *P < 0.1, **P < 0.05, ***P < 0.01Source: Authors’ construct based on GIS analysis and census data

Table 4b Regression of absolute population change on distance variables in 1970–84,1984–2000 and 1970–2000 (including distance from nearest village magnet)

Independent Variables

Beta1970–1984 1984–2000 1970–2000(Model 4) (Model 5) (Model 6)

Distance from Accra (km) 0.082 0.233 0.198

Distance from Tema (km) -0.370** -0.208* -0.267*

Distance from nearest village magnet (km) -0.038 -0.462** -0.355*

Distance from Highway (km) -0.044 0.031 0.010

R-Square 0.1684 0.2313 0.2244

N 105 105 105

Significance test: *P < 0.1, **P < 0.05, ***P < 0.01Source: Authors’ construct based on GIS analysis and census data

Patterns of population growth in peri-urban Accra, Ghana 1315

International Journal of Urban and Regional Research 36.6© 2011 Urban Research Publications Limited

to the village magnets and then access the road from there. These overall results suggestthat, although population density remained highest in areas close to Accra, the generaleffect of proximity to the city on population growth disappears when the effects ofproximity to Tema, the five village magnets and a highway are included in the model.

The spreading pancake hypothesis

Based on the results of ArcGIS modeling and regression analyses, the spreading pancakehypothesis explains some of the variation in population growth in the GAMA.Population is more concentrated in areas close to the city than in areas farther away.However, probably because of the exhaustion of the carrying capacity of areas close tothe city and the growing influence of Tema and other localities, the effects of proximityto the city on population growth appear to have diminished. This is consistent with theobservation by Simon et al. (2004: 243, 235) that the degree of ‘urbanness’ (in this casepopulation growth and concentration) in peri-urban areas follows a negative but ‘non-linear and non-uniform gradient’. But why does the spreading pancake hypothesis seemto hold for peri-urban Accra? Yankson et al. (2004) suggest that liberalization policiesresulted in a hike in rents in Accra, triggering the migration of both poor and wealthyresidents from the city to peri-urban areas in search of cheaper land and residentialaccommodation. Nevertheless, in terms of net effect, this was not enough to cause Accraand its immediate environs to lose population to peri-urban localities as Gilbert’s (1993)‘polarization reversal’ hypothesis would suggest.

The development node hypothesis

The statistical analysis indicated that Tema had a large influence between 1970 and 1984,but its influence declined somewhat thereafter. The ArcGIS spatial modeling indicatesthat there are distinct clusters of high density in the vicinity of Tema as well as the fivevillage magnets. However, as shown in Figures 4 and 5, population density rapidlydeclines with distance from each of these centers in each period. During the 30 yearsunder review, 92–96% of the peri-urban population resided within 8 km of these centers.The mean population density of areas within 4 km of the six localities more thanquintupled, while that of areas 4–8 km away increased six-fold. Areas beyond 16 kmremained sparsely populated. Thus, both ArcGIS modeling and regression analyses

Figure 4 Population density by distance from Tema, Ashiaman, Adenta, Dome, Gbawe orMadina, 1970, 1984 and 2000 (source: authors’ construct based on census data)

1316 Petra Doan and Charles Yaw Oduro

International Journal of Urban and Regional Research 36.6© 2011 Urban Research Publications Limited

confirm the development node hypothesis, which predicts faster population growth andgreater population concentration in and around Tema than in other parts of the peri-urbanzone.

The village magnet hypothesis

As mentioned above, ArcGIS modeling illustrates that, during the reference period,population density tended to be higher in areas around Ashiaman, Adenta, Dome, Gbaweand Madina (together with Tema) than in other parts of peri-urban Accra (see Figures 4and 5). However, results of regression analyses of the effects of proximity to these‘village magnets’ on population growth are mixed. The multivariate regression models(models 4, 5 and 6) in Table 4b, which use a single distance from the nearest villagemagnet, show the expected negative coefficients (standardized betas), which werestatistically significant in both 1984–2000 (model 5) and 1970–2000 (model 6), butinsignificant in 1970–84 (model 4). These findings confirm that peri-urban populationtends to be concentrated in and around selected pre-existing localities (see Leaf, 2002;Grant and Yankson, 2003). In the case of peri-urban Accra, this pattern of growth hasbeen particularly true with respect to Ashiaman and Madina, both of which became thenuclei of fast-growing densely populated pockets in the peri-urban zone.

The ribbon hypothesis

Throughout the 1970–2000 period, peri-urban areas close to highways were moredensely populated than areas farther away from highways (see Figures 6 and 7). In 1970,1984 and 2000, 90% of the peri-urban population lived within 4 kms of a highway. Whendistance to each of the five village magnets is included in the multivariate regressionmodels as a separate independent variable, distance from a highway shows consistently

Figure 5 Population density by proximity to Tema, Ashiaman, Adenta, Dome, Gbawe orMadina, 1970, 1984 and 2000 (source: authors’ construct based on census data)

Patterns of population growth in peri-urban Accra, Ghana 1317

International Journal of Urban and Regional Research 36.6© 2011 Urban Research Publications Limited

negative but marginally significant coefficients (see Table 4a). However, when distanceto the nearest village magnet is used instead, all the regression models yield a statisticallyinsignificant coefficient for distance from a highway (see Table 4b). As noted above,these results are probably due to the fact that all the major growth centers (Tema and thevillage magnets) are located within 4 kms of a highway. As a result, including distancesfrom these localities as independent variables in the regression models renders distancefrom highway redundant. In other words, the validity of the ribbon hypothesis asrepresented by the impact of distance from highways on population growth is not clearly

Figure 6 Population density by distance from highways, 1970, 1984 and 2000(source: authors’ construct based on census data)

Figure 7 Population density by proximity to highways, 1970, 1984 and 2000(source: authors’ construct based on census data)

1318 Petra Doan and Charles Yaw Oduro

International Journal of Urban and Regional Research 36.6© 2011 Urban Research Publications Limited

shown by the multivariate regression models, because its influence is indirectlyaccounted for through the other independent variables. In other words, highways areimportant stimulators of peri-urban development in the GAMA, but their influence ismost clustered in existing growth centers.

The above finding is consistent with other studies suggesting that peri-urbandevelopment in Africa tends to follow major highways. The presence of highways isusually an opportunity for residents to set up home-based and roadside businesses, orto commute easily to the central city where their livelihoods may be located (seeMabogunje, 1990; Briggs and Mwamfupe, 2000; Mbiba and Huchzermeyer, 2002; Grantand Yankson, 2003; Bryceson, 2006; Oduro, 2010). This applies to Accra and its peri-urban zone, where the livelihoods of many residents are based on proximity to the road.In addition, there has been a boom in car ownership since Ghana adopted structuraladjustment and liberalization policies in the 1980s (Yeboah, 2000; 2003; Yankson et al.,2004; Grant, 2009), and the number of vehicles in the country (most of which are inAccra and other urban centers) is currently growing at 10% per annum (Adarkwaa,2009). Proximity to a highway is critical for many peri-urban residents, whether theyown a car or use public transport, since most of them make frequent trips to Accra fortheir daily work, to purchase essential goods and services, and for other activities notavailable in the periphery.

Unfortunately in peri-urban Accra, many secondary arterials and local access roadshave never been paved and some of them are hardly passable, making access to the pavedhighway even more important. As a result, connectivity is limited between different partsof the peri-urban zone, making the highways the most important routes for vehiculartravel and locations to do business. Moreover, many houses in peri-urban Accra withgood accessibility are ideal locations for dual use as residences and the sites of variouskinds of income-generating activities (e.g. hairdressing, textile-making, carpentry,electronics repairs, etc.). In addition, some peri-urban residents undertake vending andother petty commercial activities either directly on roads (where there are chronic trafficjams) or at the roadside. These factors help to explain the highway-induced radial patternof population distribution depicted in Figure 7.

Figure 8 provides a good visual summary of the combined results. It illustrates thegeneral pattern of the city’s growth with a tendency to follow the spreading pancakehypothesis — beginning with high population density in the immediate vicinity of thecity and spreading outwardly over time in concentric circles. However, this general

Figure 8 Schematic representation of peri-urban expansion in Accra

Patterns of population growth in peri-urban Accra, Ghana 1319

International Journal of Urban and Regional Research 36.6© 2011 Urban Research Publications Limited

pattern is distorted by the magnetic attraction of alternative development nodes, villagemagnets and highways — the tendency for population to gravitate towards Tema, thefive village magnets and highways. For example, a careful examination of Figure 8reveals that while the presence of highways produces a radial pattern of growth, thisconfiguration is in turn influenced by the spreading pancake effect. In other words, theintensity of growth along highways decreases with distance from Accra. In addition, thespreading pattern of urbanization is intensified by the growth-inducing effects of Tema,Adenta, Ashiaman, Dome, Gbawe and Madina. That is, the general pattern of concentricdevelopment around Accra is reinforced by the fact that Tema, each of the villagemagnets and a number of the other large peri-urban localities are all within 10 km of thecity, while most of the smaller localities were more than 20 km from the city.

As noted above, there are other plausible (but difficult to test) factors including thepresence of natural barriers, the existence of undeveloped state-owned lands, andongoing disputes over the ownership and/or control of land between various parties.Notable natural barriers are the Densu lake (created by the Weija dam) and the DensuDelta Ramsar site — an expanse of wetlands comprising lagoons, estuaries and marshesthat purifies surface water and serves as a habitat for many plant and animal species,including fish and migratory and resident birds (see Amankwah, 1998; Anku, 2006;Gbogbo, 2007). Examples of undeveloped state lands include protected buffers aroundthe Densu lake/basin and large tracts of military lands at various locations, such as Teshie(located along the coast between Accra and Tema) and Mitchell Camp (located along theTema–Akosombo highway). Land disputes in peri-urban Accra may result fromdisagreement over land ownership or boundaries, or over the sharing of ‘drink money’collected by traditional authorities and clan heads for the release of land to developers(Mends and Meijere, 2006). These conflicts sometimes lead to litigation in the courtsystem and even violent clashes. These barriers have contributed to the presence ofseveral pockets of undeveloped land surrounded by vast stretches of built-up areas.

RecommendationsThe findings show that, standing alone, none of the four hypotheses tested adequatelyexplains Accra’s spatial growth and form. Rather, growth in the peri-urban zone is theoutcome of simultaneous interactions of forces represented by the four hypotheses, aswell as other factors that are not as easily testable. This seemingly random growth hasbeen exacerbated by the apparent inability of the modernist planning system (with itsfocus on the central city) to adjust to the speed of changes in the peri-urban zone. Thefindings call for a rethinking of intervention strategies for the management of peri-urbandevelopment.

The patterns revealed by this article could be used to provide more explicit guidanceto local planners. Unfortunately, planning functions are severely hampered by theinability of local governments to direct physical development in an orderly manner. Anumber of studies point out that this shortcoming has exacerbated low-densitydevelopment, destruction of farmlands, encroachment on ecologically sensitive landsand various forms of environmental degradation (e.g. see Yankson et al., 2004; Goughand Yankson, 2006; Oduro, 2010). Rather than attempting to impose a modern urbanorder, the authorities should seek to recognize the underlying growth trends and developmechanisms to guide new investments in more sustainable ways.

Improvements in the planning capacity at the district level are an essential first step inreducing the negative aspects of rapid growth. An increased emphasis on planning wouldfacilitate local government decisions to encourage compact development and preserveecologically sensitive lands. In particular, district planning staff might consider measuresto promote development around existing village centers and discourage leapfrogdevelopment. Furthermore, planning capacity must be developed at a wider regional

1320 Petra Doan and Charles Yaw Oduro

International Journal of Urban and Regional Research 36.6© 2011 Urban Research Publications Limited

scale. This is important because Accra’s growth is expected to extend beyond the GAMAin the next 20 to 30 years, as evidenced by the fact that several medium-size urbancenters close to the GAMA (notably Kasoa, Nsawam, Aburi and Dodowa) have alreadybecome functionally linked to Accra by serving as dormitory towns. This growth willundoubtedly follow similar patterns to those described in this article. Developmentnodes, old village centers and highways will attract intense urban development, andplanners could begin making preparations now. With the support of the centralgovernment, the Greater Accra Regional Coordinating Council (RCC) and the NationalDevelopment Planning Commission (NDPC), the municipal assemblies need to adopt ajoint multi-level planning system, since individual assemblies lack the capacity toeffectively contain this fast and complex growth.

In addition, the central government needs to support the municipal assemblies toupgrade selected localities to serve as intermediate service centers, so as to relieve someof the pressure on Accra and Tema, and to reduce commuting and highway-inducedsprawl. The ‘village magnets’ (and possibly other centers not included in the analysis)could be strengthened to serve as regional service centers. Specific interventions to beconsidered might include: the development or upgrading of markets, health, educationaland other infrastructure; relocating some of the many government agencies in Accra tothese centers; and incentivizing private businesses to set up there. Since village magnetsare already a locus of intense development, it makes good sense to build on that trend andexpand the level of services already available.

Finally, improved planning capacity and the development of intermediate servicecenters should be complemented by a careful consideration of transportation-relatedinvestments throughout the metropolitan region. In particular, proposals for investmentsin new highways and arterial roads radiating from the central city should be scrutinizedcarefully, since such investments are likely to create new high-intensity growth corridorsthat would stimulate further sprawl. An alternative and more useful objective would beto enhance connectivity between existing centers and outlying localities, by improvingthe conditions of existing secondary and access roads in the already built-up or newlydeveloping peri-urban areas. Improving local roads may increase the pressure fordevelopment along those secondary corridors, but increasing the accessibility of moreremote localities will encourage a more fully articulated economic system. As theperi-urban zone develops, access to the full range of urban services will be increasinglydesirable and will allow localities at a greater distance from the highway to participatemore fully in the urban economy. In addition, the construction of a ring road to link majorperi-urban localities (e.g. Gbawe, Dome and Ashiaman) that are currently disconnectedfrom one another might also have a positive impact on a more orderly urban form.

Expanding Accra’s existing mass transit system to serve those peri-urban areas wherepopulation density and travel demand are high enough should be carefully explored.Although such investments would reduce congestion on the highways, sufficient demandfor mass transit in the peri-urban zone may continue to be limited by low density.However, as density increases in the vicinity of development nodes and village magnets,there may be sufficient demand to warrant investments in alternatives to the use of privatecars and the informal public transport system (locally called tro-tros). These investmentsare only likely to be viable in the highest-density corridors and may increase populationpressure in those areas.

ConclusionsThe nature and consequences of urban expansion in the developing world, especially inAfrica, have not been adequately researched. In particular, expansion patterns at the edgeof cities are often misunderstood. The naïve assumptions that peri-urbanization occurs aseither a uniform process of concentric expansion or a chaotic wave of informal

Patterns of population growth in peri-urban Accra, Ghana 1321

International Journal of Urban and Regional Research 36.6© 2011 Urban Research Publications Limited

development have been debunked by the results of this research. Instead of a hodgepodgesea of informality, social and economic conditions at the edge of the urban area are aresponse to the adaptive failure of the modernist planning model. These patterns ofperipheral urbanization are complex, but can be understood as a nexus of predictableforces. Individuals seeking to create a home or develop a business in the metropolitanarea choose locations that make economic sense, but their choices may be unrelated tothe goals established in an outdated development plan. Specifically, this study hasdemonstrated that peri-urban growth in Accra results from a complex interaction offactors represented by the four hypotheses: the spreading pancake, development node,village magnet and ribbon hypotheses. While population is found to be mostconcentrated around Accra, other factors are also influential. The cluster of relativelyintense urban development around the port of Tema and its globally oriented industrialzone forms a key node of development activity. In addition, more intense developmentalso clusters around established localities within the peri-urban zone that began asmodestly sized localities. By 2000, these localities had become powerful magnets forincreased population and land development in the immediate vicinity. Finally, highwaysare a clear catalyst for development, though some of the evidence suggests that there isa more intense clustering of population and urban activity within existing sub-centers.All four hypotheses contribute to a more complete understanding of developmentpatterns in the peripheral zone, and are better understood as a complex interweaving offorces rather than as separate factors.

These findings about the sources and patterns of spatial growth in one Africanmetropolitan region have a number of broader implications for urban policy and planningelsewhere in the continent. First, it must be recognized that attempts to expand aEuropean-style planned city outwards from the old colonial core are misguided. Theurbanization process in African cities is in part a reaction against the imposed orderlinessof the ‘Western’ planning model, and in part a longing for the density and vibrancyassociated with more traditional African urban forms. While the results may appearhaphazard, they can in fact be explained by a combination of the principles explored inthis article. These insights regarding the patterns and drivers of development can guidemunicipal authorities to devise more appropriate ways of managing continued urbangrowth and infrastructural stresses. The findings might be used by planning officials tounderstand expansion processes in other African cities that are not based on Westernmodels of central city and suburban patterns, but are clustered around existing villages,highways and economic opportunities.

Policymakers struggling to rationalize urban investments at the urban fringe would dowell to recognize the important catalytic effects of development nodes on the periphery.In addition to generating increased foreign exchange, global investments in free tradefacilities are also likely to stimulate a swath of more intense development outside theplanned area. In the case of Tema, it is possible to discern a rippling of effects resultingfrom increased global investments that far exceed the planned impacts of the port andfree trade zone. Concentrated global investments in development nodes are often wellplanned in the immediate proximity of the node, but local planning efforts shouldconsider the broader spatial implications for peri-urban expansion at the periphery of thenodes. Often such nodes include a modest amount of modern housing for workers in theindustrial enterprises, but these workers in turn require a wide range of services that willstimulate development at a wider scale in the vicinity. Each of these services generatesa demand for housing and related commercial services, creating a cascading effectstretching out into the periphery of that node.

In addition, planners need to pay closer attention to the ways in which existingsettlements at the urban periphery are incorporated into the urban region. The resultsfrom this research indicate that such localities are likely to serve as clusters of intenseurban development, especially those in close proximity to highways. Urban plannersneed to recognize this tendency and work with these communities to develop local plansthat reflect the needs of the locality, rather than seek to impose order based on modernist

1322 Petra Doan and Charles Yaw Oduro

International Journal of Urban and Regional Research 36.6© 2011 Urban Research Publications Limited

assumptions. Such plans might involve making concentrated infrastructure and serviceinvestments in such places to improve the lives of residents and encourage moresustainable patterns of development in the periphery. Future transportation investmentsshould also take into account these pre-existing localities, since clusters of densepopulation will be required to ensure the economic viability of future transportationalternatives.

Given these broad implications, it is important to consider a key limitation of thestudy. Land development patterns in Accra are partially a result of the particular socio-cultural institutions that govern land development in Ghana. While many African citieshave broad peri-urban fringes, the Ghanaian land tenure system has a number of uniqueaspects, including the use of ‘drink money’ as an equivalent to land rent that is not foundin other countries. Oduro (2010) has documented some of the inconsistencies and landdevelopment anomalies that occur in this region with a marked effect on peri-urbanlivelihoods. While this set of circumstances is specific to Ghana, it should be noted thatcustomary land tenure arrangements exist in traditional agricultural areas of otherAfrican nations and, as cities expand, there is likely to be considerable turmoil andconfusion as customary and more formal land tenure regimes come into conflict. Thismeans that while the precise patterns may not be replicable everywhere, the broaderimplications of the rapidly developing African urban periphery are likely to generalizereasonably well.

The spreading pancake, development node, village magnet and ribbon hypotheseshelp to illustrate that Africa’s urban development is not random, but the result of aninadequate and perhaps outdated planning system, and of an expanding population’sneed for places to live and work within the region. These hypotheses suggest that thelogic of these location decisions might be usefully applied to efforts to revamp theplanning systems and outcomes in other African cities.

Petra Doan ([email protected]) and Charles Yaw Oduro ([email protected]), Department ofUrban and Regional Planning, Florida State University, 330 Bellamy Building, Tallahassee,FL 32306, USA.

ReferencesAdarkwa, K.K. (2009) KNUST vice

chancellor doubts success of BRT.Interview with Joy FM, Accra [WWWdocument]. URL http://news.myjoyonline.com/news/200912/39104.asp (accessed10 December 2009).

Allan, A. (2003) Environmental planning andmanagement of the peri-urban interface:perspectives on an emerging field.Environment and Urbanization 15.1,135–47.

Amankwah, C.C. (1998) Ramsar wetlandsinformation sheet. Compiled for theCoastal Wetlands Management Project,Wildlife Department, Accra, Ghana.

Anku, S.K. (2006) Managing wetlands inAccra, Ghana. Presentation at the AfricanRegional Workshop on Cities, Ecosystemsand Biodiversity in Nairobi, 21 September(a side event at the Africities Summit,18–24 September 2006).

Briggs, J. and D. Mwamfupe (2000)Peri-urban development in an era ofstructural adjustment in Africa: the city ofDar es Salaam, Tanzania. Urban Studies37.4, 797–809.

Browder, J., J. Bohland and J. Scarpaci(1995) Patterns of development on themetropolitan fringe: peri-urban fringeexpansion in Bangkok, Jakarta andSantiago. Journal of the AmericanPlanning Association 61.3, 310–27.

Bryceson, D.F. (2006) Fragile cities:fundamentals of urban life in east andsouthern Africa. In D.F. Bryceson and D.Potts (eds.), African urban economies:viability, vitality, or vitiation, PalgraveMacmillan, Basingstoke.

Census Office (1972) The 1970 populationcensus of Ghana. Vol. II: Statistics oflocalities and enumeration areas. CensusOffice, Accra.

Patterns of population growth in peri-urban Accra, Ghana 1323

International Journal of Urban and Regional Research 36.6© 2011 Urban Research Publications Limited

Coquery-Vidrovitch, C. (2005) The history ofAfrican cities south of the Sahara: fromthe origins to colonization. Translatedby Mary Baker, Markus Wiener,Princeton, NJ.

Davis, M. (2006) Planet of slums. Verso,New York.

Drakakis-Smith, D. (2000) Third world cities.Second edition, Routledge, New York.

Durand-Lasserve, A. and H. Selod (2007) Theformalisation of urban land tenure indeveloping countries. Paper presented atWorld Bank Urban Research Symposium,Washington, DC, 14–16 May.

Gbogbo, F. (2007) The importance ofunmanaged coastal wetlands to waterbirdsat coastal Ghana. African Journal ofEcology 45, 599–606.

Ghana Statistical Service (2005) Populationdata analysis reports. Volume 1:Socio-economic and demographictrends analysis. Ghana StatisticalService, Accra.

Gilbert, A. (1993) Third world cities: thechanging national settlement system.Urban Studies 30.4/5, 721–40.

Gough, K. and P. Yankson (2006) Conflictand cooperation in environmentalmanagement in peri-urban Accra. In D.McGregor, D. Simon and D. Thompson(eds.), The peri-urban interface:approaches to sustainable natural andhuman resource use, Earthscan, London.

Grant, R. (2009) Globalizing city: the urbanand economic transformation of Accra,Ghana. Syracuse University Press,Syracuse, NY.

Grant, R. and P. Yankson (2003) City profile:Accra. Cities 20.1, 65–74.

Kombe, W.J. (2005) Land use dynamics inperi-urban areas and their implications onthe urban growth and form: the case ofDar es Salaam, Tanzania. HabitatInternational 29.1, 113–35.

Laquian, A. (2005) Beyond metropolis: theplanning and governance of Asia’smega-urban regions. Johns HopkinsUniversity Press, Baltimore.

Leaf, M. (2002) A tale of two cities:globalization and peri-urban change inChina and Vietnam. Cities 19.1, 23–31.

Mabogunje, A. (1990) Urban planning andthe post-colonial state in Africa: a researchoverview. African Studies Review 33.2,121–203.

Mbiba, B. and M. Huchzermeyer (2002)Contentious development: peri-urban

studies in sub-Saharan Africa. Progress inDevelopment Studies 2.2, 113–31.

McGee, T. (1991) The emergence of desakotaregions in Asia: expanding a hypothesis. InN. Ginsburg, B. Koppel and T.G. McGee(eds.), The extended metropolis: settlementtransition in Asia, University of HawaiiPress, Honolulu.

McGee, T. (1997) Globalization, urbanizationand the emergence of sub-global regions.In R.F. Watters and T. McGee (eds.), AsiaPacific: new geographies of the Pacificrim, Hurst and Company, London.

Mends, T.M. and J. De Meijere (2006) A studyof the institution of the customary landtenure system in the supply of propertyrights for urban development — anexample of Accra, Ghana. Paper presentedat the Promoting Land Administration andGood Governance Fifth FIG (InternationalFederation of Surveyors) RegionalConference, Accra, 8–11 March.

Myers, G. and F. Owusu (2008) Cities ofsub-Saharan Africa. In S. Brunn, M.Hays-Mitchell and D. Ziegler (eds.), Citiesof the world: world regional urbandevelopment, Rowman and LittlefieldPublishers, Lanham, MD.

Oduro, C.Y. (2010) Effects of rapidurbanization on livelihoods in theperi-urban areas of Accra, Ghana. PhDdissertation, Department of Urban andRegional Planning, Florida StateUniversity.

Potts, D. (2009) The slowing of sub-SaharanAfrica’s urbanization: evidence andimplications for urban livelihoods.Environment and Urbanization 21.1,253–9.

Simon, D., D. McGregor and K.Nsiah-Gyabaah (2004) The changingurban-rural interface of African cities:definitional issues and an application toKumasi, Ghana. Environment andUrbanization 16.2, 235–47.

Simone, A. (2004) For the city yet to come:changing African life in four cities. DukeUniversity Press, Durham, NC.

Stren, R. and M. Halfani (2001) Cities ofsub-Saharan Africa. In R. Paddison (ed.),Handbook of urban studies, Sage,Thousand Oaks, CA.

Torres, H. da Gama (2007) Social andenvironmental aspects of peri-urbangrowth in Latin American megacities.Paper presented at United Nations ExpertGroup Meeting on Population Distribution,

1324 Petra Doan and Charles Yaw Oduro

International Journal of Urban and Regional Research 36.6© 2011 Urban Research Publications Limited

Urbanization, Internal Migration andDevelopment, New York, 21–23 January.

Twum-Baah, K.A. (2000) Population growthof Mega-Accra — emerging issues. In R.Mills-Tettey and K. Adi-Dako (eds.),Visions of the city: Accra in the 21st

century, Woeli Publishing Services, Accra.Wiese, B. (1981) New ports as nodes for

industrial and urban development: thecases of Richards Bay and Saldanha inSouth Africa. GeoJournal SupplementaryIssue 2, 51–8.

Yankson, P.W.K., R.Y. Kofie and L.Moller-Jensen (2004) Monitoring urbangrowth: urbanization of the fringeareas of Accra. Unpublishedpaper.

Yeboah, I.E.A. (2000) Structural adjustmentand emerging urban form in Accra, Ghana.Africa Today 47.2, 61–89.

Yeboah, I.E.A. (2003) Demographic andhousing aspects of structural adjustmentand emerging urban form in Accra, Ghana.Africa Today 50.1, 106–19.

RésuméEn Afrique (et dans le monde en développement en général), l’urbanisation a souvent étémal comprise par les observateurs occidentaux en quête de villes et banlieues de typeoccidental. Dans ces parties du monde, l’urbanisation accélérée a entraîné desmodifications constantes dans la conformation des grandes villes, les terrains rurauxpériphériques devenant rapidement des zones urbaines. Selon certaines descriptions,la périphérie est amorphe, envahie d’habitants à faible revenu; selon d’autres, elleest empreinte d’une grande diversité. À l’aide de données de recensement, d’unemodélisation spatiale et d’une régression statistique, on peut montrer que l’expansionurbaine aux limites de la ville d’Accra au Ghana n’est pas amorphe et qu’elle présentedes schémas tout à fait identifiables. Ceux-ci correspondent aux quatre hypothèsesévaluées dans cette étude: la crêpe étalée, le pôle de développement, les villagesaimantés, le ruban. L’idée que la croissance urbaine se diffuse en cercles concentriquesautour d’une ville centrale (modèle de la crêpe étalée) s’applique à Accra, mais cemodèle est associé à d’autres schémas pour créer une forme urbaine en évolutionpermanente dans sa périphérie. On trouve ainsi une concentration de croissance autourd’une ville portuaire, de plusieurs villages pré-existants, ainsi que le long de grandesroutes rénovées qui rayonnent à partir de la ville.

Patterns of population growth in peri-urban Accra, Ghana 1325

International Journal of Urban and Regional Research 36.6© 2011 Urban Research Publications Limited