Portal de Periódicos da CAPES

Revisiting the Meaning of Development: A Multidimensional Taxonomy of Developing Countries

2013; Taylor & Francis; Volume: 49; Issue: 12 Linguagem: Inglês

10.1080/00220388.2013.822071

1743-9140

Sergio Tezanos Vázquez, Andy Sumner,

Social and Economic Development in India

AbstractMany have challenged the use of income per capita as the primary proxy for measuring development since Seers's seminal works. This article continues this tradition with a more recent twist. We use cluster analysis to build a multidimensional taxonomy of developing countries using a set of indicators covering four conceptual frames on 'development'. The value-added of the article is not to suggest that our classification is the end in itself, but – more modestly – to demonstrate that more work on taxonomies is required in light of the weakness of classifications based solely on income and the changing distribution of global poverty. AcknowledgementsWe would like to thank Richard Manning, Ainoa Quiñones, José María Larrú, Duncan Green and Rafael Domínguez for their comments and suggestions.Notes1. See http://blogs.worldbank.org/developmenttalk/a-review-of-the-analytical-income-classification2. The Atlas method is a moving average of official exchange rates adjusted for inflation relative to the Euro Zone, Japan, UK and the US.3. The World Bank's thresholds are discussed in-depth in Sumner (Citation2012a). According to the short history of the Bank's classifications available on their website (World Bank, 2012a), the actual basis for the original thresholds was established by: 'finding a stable relationship between a summary measure of well-being such as poverty incidence and infant mortality on the one hand and economic variables including per capita GNI based on the Bank's Atlas method on the other. Based on such a relationship and the annual availability of Bank's resources, the original per capita income thresholds were established.' The actual documentation containing the original formulae are not publically available.4. See Nielsen (Citation2012) for a detailed explanation on how the World Bank, the IMF and the UNDP determine the number of countries that make up each income group. Nielsen also proposes an alternative 'data-driven' methodology for identifying the 'development threshold', which overcomes the arbitrary definition of the income intervals. The 2012 World Bank thresholds are as follows: the 'low income countries' (LIC) are those countries that have less than $1,025 income per capita GNI in 2011; the 'lower middle income countries' (LMICs) are those with per capita incomes between $1,026 and $4,035; the 'upper middle income countries' (UMICs) are those with incomes between $4,036 and $12,475; and the 'high income countries' (HICs) with more than $12,476 per capita income.5. We initially considered (un)employment but we finally ruled it out due to comparability problems of the unemployment rates across countries.6. One group of databases includes basic data such as the nature of democratic regimes (for example, the Alvarez, Cheibub, Limongi and Przeworski Regimes Index and the Cheibub and Gandhi extension; Freedom House; POLITY; and Vanhanen Democracy and Polyarchy Index and Data set). A second group of data sets expands this to democratic governance rather than regime type (the World Governance Indicators; the Polcon Political Constraints Index). Finally, there is a group of data sets that relate to the way governments treat their citizens (for example, the Freedom House datasets, the Heritage Foundation and Wall Street Journal Index of Economic Freedom).7. For example, the World Bank data sets on 'adjusted net national product', 'genuine savings' (also known as 'net adjusted savings'), and 'wealth estimates'; and from international NGOs such as the 'ecological footprint' and the 'Living Planet Index'. Additionally, 'Resource Flow' measures mix economic and non-economic measures.8. See the Online Appendix 1 for descriptive statistics of the data set.9. This section draws upon Tezanos (Citation2012) and Tezanos and Quiñones (Citation2012), who previously used cluster analysis for classifying the middle income countries of Latin America and the Caribbean.10. Regarding the standardisation method, we use the 'range -1 to 1' which is deemed to be preferable than other methods 'in most situations' (Mooi & Sarstedt, Citation2011, p. 247). The analysis was conducted using SPSS software.11. The countries not included in the analysis are either insular states with less than one million inhabitants (Antigua and Barbuda, Dominica, Fiji, Grenada, Kiribati, Maldives, Marshall Islands, Mauritius, Mayotte, Palau, Samoa, Sao Tome and Principe, Seychelles, Solomon Islands, St. Kitts and Nevis, St. Lucia, St. Vincent and the Grenadines, Tonga, Tuvalu and Vanuatu), or countries with limited statistical information (Afghanistan, Bosnia and Herzegovina, Cuba, Eritrea, Kosovo, Lebanon, Libya, Mongolia, Myanmar, North Korea, Somalia, Sudan, Timor-Leste, Uzbekistan, West Bank and Gaza, and Zimbabwe).12. If highly correlated variables are used for cluster analysis, specific aspects covered by these variables will be overrepresented in the outcome. Everitt et al. (Citation2011) and Mooi and Sarstedt (Citation2011) argue that absolute correlations above 0.9 are problematic.13. See correlation matrix in the Online Appendix 2.14. See the Online Appendix 3. For example, in the first stage, Malawi (country 59) and Mozambique (67) are merged at a distance of 0.149. From here onward, the resulting cluster is labelled as indicated by the first country involved in this merger (in this case, country 59).15. See the scree plot in the Online Appendix 4.16. See the dendrogram plot in the Online Appendix 5. SPSS re-scales the distances to a range of 0 to 25. Therefore, the last merging step to a 1-cluster solution takes place at a (re-scaled) distance of 25.17. See the VRC in the Online Appendix 6.18. See the ANOVA output in the Online Appendix 7.19. Online Appendix 8 shows the complete set of countries classified by clusters, GNI per capita and HDI.20. LIC, LMIC and UMIC World Bank country classifications as of 2011.21. See table in the Online Appendix 8 for a comparison of the three classifications.