The influence of agriculture on the structural economic vulnerability of small island spaces: Assessment using DEA based composite indicators

ABSTRACT

Small island spaces are confronted with significant disadvantages that ultimately result in strong economic vulnerability. The conventional literature emphasizes the main role of agriculture in generating structural vulnerability. Specifically, the higher the weight of agriculture compared to other sectors is, the more structurally vulnerable an economy is. However, the recent food crises revealed that the economic dependence on agriculture is not a problem on its own, but the issue is rather the efficiency of this sector along with the orientation of domestic production towards diversification and food self-sufficiency. In this paper, we thus propose a new structural economic vulnerability indicator based on the assumption that promoting the local agriculture could boost development. We insert the share of the agriculture sector in the GDP proxied by imported food dependency into the ‘official’ economic vulnerability index. Moreover, for robustness, the proposed indicator is obtained based on an endogenous weighting system derived from data envelopment analysis. Finally, our simulations for a sample of 131 developing economies point out that considering food dependency reduces dramatically the exposure of small island economies to structural vulnerability.

KEYWORDS:

• Structural vulnerability
• small island spaces
• food self-sufficiency
• DEA
• composite indicators

JEL CLASSIFICATION:

• O13
• Q18

Acknowledgments

The authors would like to thank the French Development Agency for the financial support.

Disclosure statement

The authors report no potential conflict of interest.

Notes

¹ Following Encontre (2004), Blancard and Hoarau (2013) put forward four main lists of small island developing states (SIDS): (i) a political list of 39 SIDS (32 countries and 7 territories recognized as SIDS by the Alliance of Small Island States [AOSIS]); (ii) an economic list of 48 SIDS (34 countries and 14 territories implicitly recognized as SIDS by the United Nations by the AOSIS); (iii) an institutional list of 46 SIDS (34 genuine island states, 4 continental states, and 8 non-self-governing territories) that together make up the official list according to the United Nations Secretariat; and (iv) a pragmatic list of 29 SIDS, which are self-governing, genuine island states with a population not exceeding 5 million, except for Papua New Guinea, and similar socio-economic developing country according to the United Nations Conference on Trade and Development (UNCTAD) classification. See Blancard and Hoarau (2013) for more details.

² The recent higher and more volatile international food prices rely on many factors: poor harvests in major producing countries due to extreme weather events, declining food stocks, high oil and energy prices raising the cost of fertilizers, irrigation and transportation, production of bio-fuels substituting food production, speculative transactions, export restrictions leading to hoarding and panic buying, depreciation of the US dollar, a growing change in dietary patterns from starchy foods towards meat and dairy products, and three decades of underinvestment in agriculture (Tadesse et al. 2014; FAO 2011).

³ See OECD (2008) for a survey.

⁴ http://www.fao.org/faostat/en/#data/FBS.

⁵ www.fao.org/faostat/en/#data/FA. In the special case of French Overseas Regions (FOR), data are obtained from the French customs and the French Ministry of Agriculture (https://stats.agriculture.gouv.fr/disar-web/disaron/%21searchurl/searchUiid/search.disar).

⁶ Over the considered period (2003–2012), 95 countries in our dataset have benefit at least once from the World Food Programme (see FAOSTAT, Food Aid Shipment Sheet, data from the WFP, http://www.fao.org/faostat/en/#data/FA).

⁷ As such, UNCDP’s EVI is one of the three criteria used for identifying the LDC. The two other criteria are gross national income per capita and the human assets index. Therefore, to be considered a LDC, a country must be a low-income country with a low level of human capital and high vulnerability (Guillaumont 2010).

⁸ All relevant information on the raw data (rationale, methodology of construction, time span, and data sources) are shown in Table 3.:

⁹ For all conventional dimensions, the official min and max values have been retained. In the absence of studies that define the categories of states for food dependency, it was not easy to determine the thresholds used for the min-max method. The two bounds used here (10% and 100%) are arbitrary and can be discussed.

¹⁰ Due to space considerations, we do not report the results concerning the index introducing both FDR_E and FDR_P. However, the findings are similar.

¹¹ To consider the strong heterogeneity within each group, we test for the significance of the differences between groups by the Wilcoxon procedure. The results are shows in the Appendix (Table A.2).

¹² For instance, the following assumptions are not required: (i) specific functional form of production frontier contrary to the conjoint analysis method for the utility function or (ii) absence of correlation between sub-indicators contrary to the principal component or regression analyses.

¹³ The standard FDEVI (m-score) has been normalized to ensure its comparability with the FDEVI resulting from DEA-like models. Therefore, according to a scale factor, a value of 0 is assigned to the least vulnerable country and one between 0 and 100 to all relatively more vulnerable countries.

¹⁴ For comparison with the conventional approach, the values of the endogenous weighting schemes have been reduced to unity.

¹⁵ Animal-sourced foods, which are indirectly affected by the cost and availability of feed, are also expected to see price increases, but their range is around half those of cereals (Intergovernmental Panel on Climate Change 2019).

Additional information

Funding

This work was supported by the Agence Française de Développement.