Abstract
This paper proposes a new composite index to measure the multidimensional concept of food security. Although other indicators with the same objective exist, they come with several methodological shortfalls. The Proteus index makes a twofold contribution: a) we apply all the steps needed to tackle the uncertainty sources inherent to building a composite indicator (variable selection, data imputation, normalization, weighting and aggregation) and test the assumptions through a Monte Carlo procedure that applies a variance-based sensitivity analysis of model output; and b) the results are robust over time and are comparable within and between countries, thus allowing a measure that can track the country progress towards food security. We demonstrate that the main sources of output variability are weighting, normalization, data imputation, variable selection, and aggregation in descending order of importance, but interaction effects between the uncertainty sources also play a key role. The index provides a contribution to food security monitoring. While it identifies countries requiring priority attention for their chronic situation, it proves flexible enough to capture sudden onset crises. It also reflects the main drivers that can dramatically affect a country’s food security in the short run, insofar suggesting potential areas of intervention for policy makers.
Generated Summary
This research introduces the Proteus composite index, a novel metric designed to assess the multidimensional concept of food security. The study aims to address the methodological shortcomings of existing indicators by incorporating comprehensive steps to handle uncertainty in the creation of composite indicators. This includes variable selection, data imputation, normalization, weighting, and aggregation. Through a Monte Carlo simulation, the research applies a variance-based sensitivity analysis to evaluate the robustness of the model’s output. The primary objective is to develop an index capable of ranking and categorizing countries based on their food security status and tracking progress over time. The methodology involves selecting indicators from several dimensions of food security, followed by data imputation to handle missing values, data normalization, and applying different weighting and aggregation schemes. The research also examines the impact of various uncertainty sources and sensitivity analyses to determine the reliability of the index’s results and identify the most influential factors.
Key Findings & Statistics
- The average share of missing data across the selected indicators is 21 percent.
- Six indicators have missing data rates well above the average.
- The study used multiple imputation to address missing data.
- A total of 21 indicators were used to create the index.
- The study covers 185 countries between 1990 and 2017.
- The main sources of output variability were, in descending order of importance: weighting, normalization, data imputation, variable selection, and aggregation.
- The Pearson correlation of the index with the under-5 mortality rate was 0.7848.
- The Pearson correlation of the index with the prevalence of anaemia among children under 5 was 0.7825.
- The Cronbach’s alpha (0.8772) as per Table 11 in the annex
- The index dropped by almost a quarter in 27 years (from 0.416 to 0.361).
- From 2010 to 2017, China was the country with the biggest progress towards eradicating food insecurity with a reduction of 27 percent.
- The Proteus index can be used as a tool to inform how to prioritize the allocation of multilateral resources to address food insecurity, as it combines information both on the food security level and its changes over time.
- Mortality rate, under-5 (per 1,000 live births) has a correlation of 0.7848.
- The prevalence of anaemia among children (% of children under 5) has a correlation of 0.7825.
- The prevalence of overweight, weight for height (% of children under 5) has a correlation of -0.4165.
- The prevalence of severe wasting, weight for height (% of children under 5) has a correlation of 0.3653.
- The prevalence of stunting, height for age (% of children under 5) has a correlation of 0.7734.
- The prevalence of underweight, weight for age (% of children under 5) has a correlation of 0.6656.
- The prevalence of wasting, weight for height (% of children under 5) has a correlation of 0.4737.
Other Important Findings
- The index can capture sudden onset crises and reflects drivers that dramatically affect a country’s food security in the short run.
- The index identified countries requiring priority attention for their chronic situations.
- Among the input factors, aggregation has the least impact on output variance.
- Weighting, normalization, data imputation, and the exclusion of a variable all determine a higher output variance, in descending order of importance.
- The study emphasizes the value of including interaction effects among uncertainty factors for a more robust composite index.
- The study identified Somalia, South Sudan, Chad, Eritrea, and Ethiopia among the most food-insecure countries.
- The index shows a strong non-additive component, which refers to the interaction effects among the uncertainty factors.
- In 2017, the Proteus index almost always ranked Somalia, South Sudan, Chad, Eritrea, and Ethiopia among the fifteen most food-insecure countries.
- In 2017, the index registered increases in food insecurity corresponding to the war.
- Average dietary energy supply, average protein supply and people affected by disasters are key variables.
- The main drivers of Yemen’s performance in 2017 are access to basic drinking services.
- For Niger, the key variables are average protein supply and people affected by disasters.
Limitations Noted in the Document
- The study acknowledges the inherent subjectivity in the choice of methods at each step of the composite index creation.
- The research is limited to the availability and reliability of public data from 1990 to 2017, which may impact the comprehensiveness of the index.
- The reliance on data imputation to handle missing values can introduce biases, although the study uses a multiple imputation approach to minimize this.
- The choice of indicators and the potential for high correlations between them, while addressed, could still introduce redundancy.
- The sensitivity of the index to changes in weighting, normalization, and aggregation methods is acknowledged, suggesting the need to test robustness with alternative methods.
- The exclusion of the food insecurity experience scale may limit the index’s ability to capture certain aspects of food insecurity.
Conclusion
The Proteus composite index represents a significant contribution to food security monitoring by addressing limitations in existing indicators. The study’s methodology, encompassing detailed uncertainty and sensitivity analyses, highlights the critical factors influencing food security assessments. Key findings reveal that weighting, normalization, data imputation, and variable selection are the primary drivers of output variance. The index’s ability to capture sudden crises and identify vulnerable countries offers crucial insights for policymakers. The study emphasizes that the index can be used as a tool to inform how to prioritize the allocation of multilateral resources to address food insecurity. The analysis of countries like Yemen and Niger provides valuable case studies for targeted policy interventions. The index demonstrates the potential to track food security trends over time and to assess the impact of various shocks, such as natural disasters or economic crises. The findings of the index are the ability to show a strong non-additive component pertaining to the interaction effects among uncertainty factors. The study’s approach underscores the importance of considering various factors and their interactions in food security assessments. The Proteus index’s ability to reflect diverse aspects of food insecurity and sensitivity to changes over time makes it a valuable tool for monitoring country performance. The research suggests the need for a balanced approach to enhance food security, considering both macro-level economic stability and micro-level access to essential resources. In order to inform policymaking, the index identifies single elements that make a country perform poorly, in order to inform policymaking, as Maxwell’s words. The index could help shape policies that ‘recognize the diversity of food insecurity causes, situations and strategies, and [that are] contingent on particular circumstances’. (Maxwell, 1996).
DOI
10.1016/j.worlddev.2019.104709