African Island Development Experiences:
A Cluster of Models*
Jerome L. McElroy
Department of Business Administration and Economics
Saint Mary's College
Notre Dame, Indiana 46556
*Paper presented at the Indiana Association of Social Sciences Meeting, Indiana University East, Richmond, Indiana (October 18, 2002)
Since 1960 per capita GDP growth in all African countries combined has barely averaged one percent per annum (Rodrik, 1999: 107). As a result, the continent is falling behind the pace of development in Asia and Latin America (Radelet and others, 1997). Various causes have been suggested for this poor postwar performance. Easterly and Levine (1997) emphasize the role of ethnic cleavages and the region’s inability to develop democratic institutions for safeguarding the rule of law and resolving domestic and cross-border conflicts. Sachs and Warner (1997) blame climatic conditions, geography and restrictive trade regimes (overvalued exchange rates, import restrictions, prohibitive export taxes, etc.). A recent UN study (Cornia and Court, 2001) faults growth strategies that ignore inequality and that over time create conditions conducive for political instability. Finally, Rodrik (1999) believes a generation of relative stagnation is due to neglect of the fundamentals: investment in physical infrastructure and human capital, poor macroeconomic management, and demographic pressure.
Four African countries, however, have achieved increases in their standard of living since 1960 above three percent per year. They include two land-locked countries in Southern Africa—Botswana and Lesotho—and two Indian Ocean islands, Mauritius and Seychelles. The former have succeeded primarily because of accidents of nature: fortuitous resource endowments (diamonds in Botswana) and favorable geographic location (export of labor for mining in South Africa in Lesotho). The latter have succeeded primarily because of a deliberate postwar policy of economic restructuring: from sugar to export manufacturing in Mauritius and from fishing to tourism in Seychelles. Such evidence provides the basis for further exploring the socio-economic and demographic differences between African islands and their mainland counterparts.
This study seeks to determine whether African islands represent a distinct development profile. To answer this question, three analyses are provided. The first compares the socio-economic and demographic performance of 11 Atlantic and Indian Ocean islands that surround the African coastline with 48 African continental countries. The second compares the African island profile with some 40 non-African islands across the world similar in size and level of development. The third breaks down the African island profile into clusters and argues descriptively in the context of the island development literature that the two traditional models are represented (MIRAB and tourist-dependent) along with a third.
In order to compare coastal islands with their mainland counterparts, two extensive development profiles were constructed with 26 indicators. These included eight variables to measure economic resources and performance (size in area and population, GDP growth and per capita income, inflation and unemployment, and per capita electricity consumption and aid); ten measures of demographic behavior (population growth, density and age distribution, natality, fertility and migration rates, the total fertility rate and the sex ratio); six social/health indicators (infant mortality, life expectancy, adult literacy, and phones, radios and TVs per 1,000 population); and one dichotomous political status variable (with “one” representing independence and “zero” representing dependence). In addition, a so-called “misery” index was constructed for each country by summing together the inflation and unemployment rates.
In the first islands-versus-Africa comparative analysis, to minimize error, most data were taken from two standard sources for the most recent year available: The World Factbook (CIA, 2001) for the 11 islands, and African Development Indicators (World Bank, 2002) for the 48 continental countries. Other data sources particularly for the dependent islands are identified in the Appendix. The 11 selected African islands included five in the Atlantic (Azores, Canaries, Cape Verde, Madeira, Sao Tome/Principe) and six in the Indian Ocean (Comoros, Mayotte, Mauritius, Maldives, Reunion, Seychelles). In the second comparative analysis between the African and rest-of-world islands, the latter sample included 40 small countries (less than two million population): 17 from the Caribbean, nine from the Pacific, four from the Atlantic, three from the Mediterranean, and Bahrain from the Persian Gulf. For the rest-of-world sample, all data were taken from the World Factbook (CIA, 2001).
To test for statistical differences, average values were calculated for the 11 African islands, 48 mainland countries, and 40 small non-African islands. Their averages were then tested according to t-values. In the first analysis of islands versus Africa, because of the growing literature on the distinctive characteristics of islands (Caldwell and others, 1980; Dommen and Hein, 1985; Beller and others, 1990), it was hypothesized that, in contrast to their mainland neighbors, the African islands would be smaller in size, exhibit higher levels of economic growth, per capita GDP, population density, employment and aid, and demonstrate—because of their propensity to migrate and return—older population distributions, lower population growth, natality, fertility and infant mortality. Because of their assumed more advanced level of development, it was hypothesized that the islands would record higher levels of literacy, life expectancy and socio-economic progress (phone, radio and TV usage). Because of previous research on Caribbean and Pacific countries (McElroy and Mahoney, 2000), it was also assumed islands would demonstrate a propensity for political dependence absent in larger continental states. Finally, in the second analysis between African versus rest-of-world islands, it was assumed no significant differences in socio-economic and demographic structure and/or behavior would appear.
According to the average values in Table 1, the 11 islands surrounding Africa differ significantly in structure and behavior from their mainland counterparts. Not unexpectedly, size as an indicator of resource availability presents the most obvious contrast. For example, the 48 continental countries average over 17 million in population while the average island size is less than a million. Likewise in area, mainland countries average over 300 times larger than islands. However, in terms of economic performance, recent GDP growth in islands is over 50 percent faster than in continental countries. More importantly, insular per capita GDP is over three times higher, roughly $6,300 to $2,000. Similarly, per capita electricity consumption—a common standard of living indicator—is twice the level on islands than on the continent, a difference that is not statistically significant however.
Although inflation rates are equal across both profiles, unemployment rates discriminate sharply. The estimated African rate of 35 percent is over 40 percent higher than the 24 percent insular rate. The lack of statistical significance could be due to two factors: (1) a conservatively low estimate of African unemployment (see African Development Bank, 1998), and (2) the problem of small numbers in combination with the unusual circumstances of Mayotte. In recent years, this French dependency has experienced massive immigration of refugees fleeing political upheaval in Comoros. This has raised Mayotte’s net migration (11.5/1,000 in 2000) and unemployment (45%) rates considerably above the other islands and unduly influenced average values. To illustrate, when Mayotte is excluded from the analysis, the ten-island unemployment rate falls to 21.5 percent and is statistically significant (t= -1.91).
(Table 1 about here)
There are also some indications that the profiles differ in economic structure. For example, the average island contribution of agriculture to GDP (not shown) is 14 percent versus an estimated 35+ percent for Africa. On the other hand, the average island contributions of industry and services to GDP are 20 and 66 percent respectively versus an estimated 15 and 50 percent respectively for Africa. In summary, this evidence indicates that the islands have progressed further than Africa in restructuring their former colonial economies away from traditional income-inelastic staple exports (sugar, cocoa, copra, etc.) toward more income-elastic tourism and related construction, associated financial services, and export manufacturing.
These results parallel the outcomes of Armstrong and others (1998) who found the factors that discriminate high- from low-growth countries in a very large sample (204) of both large and small countries were primarily the declining significance of low-value agriculture and rising emphasis on services (tourism and financial) and, to a lesser extent, manufacturing. Finally, islands clearly have benefited more from aid, garnering over 20 times more on a per capita basis than their African counterparts ($397 versus $18). This partly reflects the so-called “small country effect” (Hein, 1990) of aid distribution, but also may indicate significant external funding for constructing the expensive transport and communications infrastructure necessary for a viable tourism industry.
Table 1 also displays distinctly different demographic profiles. On the average, islands experience significantly higher population densities but significantly lower population growth rates. This latter result is also due mainly to their lower fertility and older age structure. For example, insular rates of natural increase are nearly a full percentage point lower than the continental average, and their total fertility rate averages a third fewer children (3.4 vs. 5.0) per women of childbearing age. Likewise, the island population distribution is significantly smaller in the young cohorts (0-14 years) and significantly larger in the older cohorts (65+ years).
In addition, islands experience significantly higher emigration rates than their continental neighbors. When Mayotte is excluded, the ten-island net migration average is –2.76 versus 0.9 for Africa (t=-2.95 and above the .01 level). Previous research on Caribbean islands has shown that emigration primarily of the working age population reduces both the crude birth rate and the population growth (McElroy and de Albuquerque, 1990). On the other hand, the sex ratios (males per 100 females) are similar across the two profiles contrary to the lower ratio expected for islands because of the usual male sex selectivity of external livelihood mobility. This outcome may suggest African islands may exhibit more gender-neutral emigration. Overall, however, the demographic comparison indicates the islands have passed further along the demographic transition from high to low fertility and are experiencing reduced population pressure as a result.
African islands are also blessed with relatively better socio-economic conditions. In terms of general health measures, the islands average significantly lower infant mortality (36 deaths per 1,000 births) than the continental countries (86 deaths per 1,000 births). This is partly a function of their greater relative affluence and their higher population densities, a surrogate indicator of greater urbanization and medical access. In addition, island life expectancy is significantly longer, 69 years versus 54 years. Among other factors, this partly reflects the significantly higher incidence of HIV/AIDS in Central and Southern Africa. Similarly, the islands exhibit sharply higher levels of adult literacy, 79 percent to 56 percent. On the other hand, the measures of social progress do not discriminate the profiles. In fact, continental countries average more radios per 1,000 population than the islands. Finally, the islands do demonstrate a certain propensity for dependence. In fact, five of the 11 are political affiliates: Mayotte and Reunion with France, Azores and Madeira with Portugal, and Canaries with Spain. This may indirectly indicate that non-sovereign political status is associated not only with relatively greater access to metropolitan country aid and investment but also with certain favorable industry-specific tax and trade concessions.
Table 2 compares the African island profile with average values from a 40-island population of all small (less than two million population) non-African rest-of-world (ROW) islands for which relatively complete published data were available. Although it was assumed that the African islands would appear similar in structure and behavior to their ROW counterparts, some differences do surface. For example, African islands exhibit significantly lower levels of per capita GDP ($6,287 versus $10,281) and service contribution to GDP (66% to 72%), and correspondingly higher levels of unemployment (24% to 11%) and “misery.” The former also demonstrate significantly higher rates of natural increase and infant mortality (35 versus 17) and significantly lower life expectancy (69 to 73 years), literacy (79% to 93%), and radio and TV usage.
(Table 2 about here)
Taken together, these results indicate the African islands are somewhat less further advanced than the ROW sample. In combination with the first analysis, these results suggest the African islands lie somewhere between their continental neighbors and the ROW islands along the postwar development continuum. Because of the contrast in statistical differences between the two analyses, the data further hint that the African islands approach closer in structure and behavior to the ROW islands than to the continental countries. This conclusion is reinforced by excluding low income, high-immigration Mayotte from the African sample. When this adjustment is made (see Table 2), several of the ROW statistical differences wash out.
The two previous analyses indicate that African island performance is distinct from the continental pattern but somewhat more similar to the postwar experience of small islands across the world. A more detailed examination of the island-specific data, however, reveals that the African island sample is not homogeneous and contains more than one distinct development experience. Table 3 breaks down the 11 country sample into low- and high-income subgroups. The former comprises five islands with per capita GDP levels of $2,000 or less: Cape Verde, Sao Tome/Principe, Comoros, Mayotte and Maldives. The latter contains six islands with per capita GDP roughly from $4,800 to $16,700: Azores, Madeira, Canaries, Mauritius, Reunion, and Seychelles.
The two profiles are quite distinct. Along demographic dimensions, although the low-income islands average roughly half the size in population and area of the more affluent group, their average density is substantially higher: 399 versus 243 persons per square kilometer. Likewise their average rate of natural increase of three percent is nearly triple that for the affluent group while their fertility rate is more than double. In addition, the average proportion of the population 0-14 years is considerably higher (45% to 24%) indicating greater long-term population momentum and growth for the future.
(Table 3 about here)
The picture is similar across the socio-economic variables. Average per capita GDP and electricity consumption in the poorer subgroups is less than 15 and 10 percent respectively of the affluent subgroup. On the other hand, the percent of GDP contributed by agriculture is significantly higher (27% versus 8%) and the contribution of services (tourism) significantly lower (59% versus 69%). Not surprisingly, with an economic structure oriented towards agriculture and extensive subsistence production, the poor island average unemployment rate is significantly higher (34% to 15%) and the labor force participation rate significantly lower (26% to 47%). With the exception of aid per capita, the poor-island subgroup compares unfavorably with the more affluent islands on all the other social and lifestyle indicators: demonstrating considerably lower life expectancy, literacy and phone etc. usage and considerably higher infant mortality (64 versus 10 deaths per 1,000 births).
In fact, the poor-island profile approximates the African continental profile in many respects. The similarities include relatively low income, literacy, and life expectancy levels and high unemployment, infant mortality, fertility and rates of natural increase. But there are three major differences. First, average population densities in the poor islands are nearly 15 times higher than the mainland average. Second, rates of emigration are markedly higher. Third, per capita aid ($663) is over 35 times higher than on the continent ($18). These significant differences sketch out in broad brush the contours of the MIRAB economy model established in the island literature. According to Bertram and Watters (1986), the postwar development of many small Pacific islands can be described as a transition from subsistence and colonial monoculture to external financial dependence, i.e. from off-island labor Migration, Remittances, Aid, and the expansion of wage employment in government Bureaucracy.
In at least four of the five African islands (Maldives excepted), the contours of the MIRAB process appear evident. For example, according to Aldrich and Connell (1998), government provides most of the salaried employment in Mayotte where aid from France approaches $400 per capita. In Comoros emigration has been the traditional safety valve where presently one in five Comorans live abroad. These off-island remittances plus substantial aid from France partly cover the island’s significant annual budget and trade deficits. In Cape Verde, remittances account for 20 percent of GDP and aid another 15 percent (CIA, 2001). Likewise Sao Tome/Principe is characterized by high emigration and the highest aid per capita ($1,212) among all the countries, mainland and insular. What distinguishes Maldives from these four MIRAB islands is its reduced emigration and aid and successful diversification into high value-added tourism. The visitor industry now accounts for 25 percent of GDP, and Maldives has become “one of the best values for money underwater diving destinations in the world” (Cockerell, 1995).
In contrast to the external dependence of the poorer islands, the more affluent group with an average per capita income over eight times higher demonstrate significant postwar structural change and transformation. They have achieved a much higher level of income-elastic diversification primarily toward labor-intensive tourism (69% versus 59% services to GDP) and secondarily toward higher value-added industrial development (23% versus 13% industry to GDP). As a consequence, they maintain higher labor force participation rates and lower unemployment than their poor-island counterparts. Their lower rates of natality, fertility and infant mortality and high life expectancy and literacy that approach developed country levels suggest they have progressed far along the demographic transition. Their per capita phone, etc. usage indicate a considerably more advanced degree of modernization in comparison with both the MIRAB islands and the African continental countries.
Some cursory but illustrative examples of their different development patterns are instructive. First, the Canaries—Azores to a lessor extent—represents a classic case of the “Paradise” or tourism-dependent small island that linked its modernization to the fastest engine in the postwar global economy. For example, the Canaries receives over ten million visitors annually and is characterized as one of the most tourist-penetrated destinations in the world (McElroy, 2002). Second, Reunion may best be described as a TOURAB economy, combining features of both MIRAB dependence and tourism development (Guthunz and von Krosigk, 1996). During the 1990’s, the number of visitors roughly doubled to over 400,000 (WTO, 2002). Since 1946, when Reunion was politically assimilated into France—the so-called process of “departmentalization”—the economy has been heavily supported by remittances from labor emigration to France, and the local application of French civil service salaries, minimum family income standards, and other welfare programs. As a result, government bureaucracy now accounts for a third of all employment (Poirine, 1994).
Third, the remaining three affluent islands form a somewhat unique development model distinguished by Balanced Diversification (BD). The balance involves blending traditional agriculture with tourism expansion and export manufacturing. Mauritius is the most celebrated example. With the establishment of an export processing zone in the 1970’s, manufacturing became the lead sector and Mauritius has become the world’s second largest exporter of woolen knits (Cutter, 2001). Close behind, earnings from Mauritius tourism industry, the largest in the Indian Ocean, now rivals the contribution of traditional sugar exports (Benhamou, 1993). Madeira and Seychelles represent other BD economies. In the first case, Madeira’s development rests on three legs: export sugar, the manufacture of handicrafts and export of its fabled wine, and more recently expanding tourism. In the second case, Seychelles’ diversification has involved traditional artisanal fishing and boat-building, tuna processing for export, and a rapidly expanding high-end tourism sector. The most distinctive feature all three BD islands is the relatively high contribution of industry to GDP in the 26 to 30 percent range.
In summary, the six affluent African islands represent in broad strokes a cluster of postwar island development models different from their MIRAB neighbors: the tourist-dependent Canaries and Azores, the mixed TOURAB Reunion, and the three BD cases of Mauritius, Madeira and Seychelles. They stand out in one further respect from the low-income island subgroup. The majority are political dependencies, i.e. four of the six in contrast to only one (Mayotte) in five in the latter group. Although it is difficult at the macro level to measure how this propensity for dependence might foster superior performance, it seems plausible that non-sovereign affiliation may encourage the political stability long associated with favorable growth (Sola-i-Martin, 1997). It may also facilitate easier access to metropolitan infrastructure funding, private investment and export markets as well as provide special advantages for tourism development (ease of entry, duty-free concessions) not available to the independent islands.
In the only other systematic treatment of African islands, Cohen (1983) expressed pessimism about their future development prospects because he felt small states would be increasingly marginalized by (1) the forces of political and economic globalization, and (2) the close-off of traditional emigration outlets. This provisional study presents a more optimistic picture. By and large, the African islands have performed better than continental Africa because they have more successfully restructured their colonial economies toward rapidly growing tourism and export manufacturing. As a result, as a group they have proceeded further along the demographic transition and enjoy a higher socio-economic standard of living than their mainland counterparts. In fact, their structure and performance more closely resemble the development profile observed for other small islands across the world.
On the other hand, the African islands themselves present a heterogeneous cluster of development experiences. At one end are four low-income countries—Cape Verde, Comoros, Mayotte and Sao Tome/Principe—that resemble MIRAB states heavily dependent on emigration, remittances, foreign aid and public employment. At the other end are the affluent islands that illustrate a cluster of development experiences. These include: (1) the traditional ‘paradise’ or tourist-dependent model of Canaries, Azores and possibly Maldives; (2) a combination TOURAB case of Reunion, and (3) the Balanced Diversification model of Mauritius, Madeira, and Seychelles.
These preliminary results suggest two broad directions for future research. The first would be to further explore the factors responsible for the observed differences between African island and continental development experiences. Among other things, this list might include more favorable advantages for islands in terms of transport access, trade history, political stability, macroeconomic management and, through their greater propensity for dependence, special links to metropolitan aid, investment, overseas labor and export markets. The second would be to examine factors responsible for the differences between the African and ROW islands. This might proceed in two ways: (1) exploring the impact of regional influences with other Caribbean, Mediterranean and Pacific islands, and (2) examining, especially in the case of the island dependencies, the impact of differential economic performance of metropolitan powers on their former or present patron states, a particularly promising approach suggested by the recent work of Bertram and Karagedikli (2002).
African Development Bank. 2002. Basic Indicators on African Countries. African
Development Bank Group.
African Development Bank. 1998. Africa Economic Analysis-African Economic
Report-1998. <http://www.afbis.com/analysis/eca%20report.htm 9/11/02>.
Aldrich, R. and J. Connell. 1998. The Last Colonies. Cambridge, UK: Cambridge
Armstrong, H., R.J. de Kervenoael, X. Li and R. Read. 1998. “A Comparison of the
Economic Performance of Different Micro-states, and Between Micro-states and
Larger Countries.” World Development 26(4): 639-656.
Beller, W., P. d’Ayala and P. Hein (eds.). 1990. Sustainable Development and Environmental Management in Small Islands. Paris: UNESCO-Parthenon.
Benhamou, F. 1993. “L’ile Maurice: Un Tourisme Ltarmonieuz Mais Fragile.” Espace (Paris) 121:42-46.
Bertram, G. and O. Karagediklli. 2002. “Are Pacific Islands Converging or Diverging?”
In J. Poat (ed.), On the Edge of the Global Economy. London: Edward Edgar
Bertram, I.G. and R.F. Watters. 1986. “The MIRAB Process: Earlier Analysis in
Context.” Pacific Viewpoint 27(1): 47-59.
Caldwell, J.C., G.E. Harrison and P. Quiggin. 1980. “The Demography of Micro-
states.” World Development 8(12):953-968.
Central Intelligence Agency. 2001. The World Factbook. Washington, D.C.:CIA
Cockerell, N. 1995. “Maldives.” International Tourism Reports No. 2. London:
Economist Intelligence Unit, pp. 25-43.
Cohen, R. 1983. African Islands and Enclaves. Beverly Hills, CA: Sage Publications.
Cornia, G. and J. Court. 2001. Inequality, Growth and Poverty in the Age of
Liberalization and Globalization. Policy Brief No. 4. United Nations
University/World Institute for Development Economics Research.
Cutter, C.H. 2001. Africa 2001 36th Ed. Harpers Ferry, WV: Stryker-Post Publications.
Dommen, E. and P. Hein (eds). 1985. States, Microstates and Islands. London: Croom
Easterly, W. and R. Levine. 1997. “Africa’s Growth Tragedy: Policies and Ethnic
Divisions.” Quarterly Journal of Economics CXII (November): 1203-1250.
Europa Publications Ltd. 2000. The Europa World Yearbook 2000, Vols. I and II.
European Union. 2002. Commission Report on the Measures to Implement Article
299(2): The Outermost Regions of European Union.
European Union. 2002. Most Remote Regions-European Union Statistics. <http://europa.eu.irt/comm/secretariat_general/sgk/rup/er/statistics.htm>.
Guthunz, U. and E. von Krosigk. 1996. “Tourism Development in Small Island States:
From MIRAB to TOURAB” Pp. 18-35 in L. Brigriglio, B. Archer, J. Jafari, and
G. Wall (eds.), Sustainable Tourism in Islands and Small States: Issues and
Policies. New York: Pinter.
Hein, P. 1990. “Economic Problems and Prospects of Small Island.” Pp. 35-42 in W.
Beller, P. d’Ayala, and P. Hein (eds.), Sustainable Development and
Environmental Management in Small Islands. Paris: UNESCO-Parthenon.
Madeira Island Directory. 2002. <http://www.unep.ch/islands/CDM.htm, 6/13/02>.
McElroy, Jerome L. 2002. “The Impact of Tourism in Small Islands” A Global
Comparison.” Pp. 151-167 in F. di Castri and V. Balaji (eds.), Tourism,
Biodiversity and Information. Leiden, Neth: Backhuys Publishers.
McElroy, Jerome L. and M. Mahoney. 2000. “The Propensity for Political Dependence
in Island Microstates.” Insula 9(1)(July): 32-35.
McElroy, Jerome L. and K. de Albuquerque. 1990. “Migration, Natality and Fertility: Some Caribbean Evidence.” International Migration Review 24(4):783-802.
Model and the French Overseas Departments and Territories.” World
Development 22(12): 1997-2009.
Portugal Statistical Service. 2002.
Radelet, S., J. Sachs, and Jong-Wha Lee. 1997. Economic Growth in Asia. Cambridge,
MA: Harvard University, Institute for International Development (July).
Regional Statistical Service of the Azores. <http://www.srea.ine.pt/7/02/02>.
Regional Statistical Service of Madeira. <http://www.dre.srpc.pt./,7/02/02>.
Rodrik, D. 1999. The New Global Economy and Developing Countries: Making
Openness Work. Washington, DC: Overseas Development Council, Policy Essay
Sachs, J. and A. Warner. 1997. Sources of Slow-Growth in African Economies.
Cambridge, MA: Harvard University, Institute for International Development
Sola-i-Martin, X. 1997. “I Just Ran Two Million Regressions.” American Economic
Review 87(2): 178-183.
Statistical Institute for the Canaries. 2002. <http://www.istac.rcanaris.es/7/10/02>.
World Bank. 2001. African Development Indicators. Washington, DC: IBRD (online).
World Tourism Organization. 2002. Compendium of Tourism Statistics 2002. Madrid: