The Global FDI Country Attractiveness Index (GFICA ©) is designed to continuously evaluate the attractiveness of seven world regions as destinations for FDI using quantitative indicators. It compares the performance of each region against a broader set of countries and geographical areas, providing insights into their relative appeal for multinational enterprises (MNEs). Updated annually, the index allows for the tracking of each country's progress across key factors that influence investment decisions, offering a comprehensive overview of the critical axes that MNEs consider when choosing where to invest.

While the index compiles valuable data for FDI allocation decisions, it should not replace investors' independent research and country-specific expertise. Instead, it serves as a support tool, particularly during the initial due diligence phase, helping investors identify factors that may accelerate or hinder foreign investment in a particular region.

The analysis benchmarks the FDI attractiveness of 106 countries, which account for over 98% of global inward FDI stocks. The index incorporates 47 indicators across 10 key FDI drivers, grouped into three pillars that reflect the main competitive advantages sought by MNEs: Prerequisites, Underlying Indicators, and Externalities. The ranking highlights the specific areas where countries or regions lag behind, helping them identify opportunities for improvement to enhance their attractiveness to foreign investors. 

Ten Latent Key Drivers

The core principle of our index is to evaluate the ten underlying drivers of FDI attractiveness, which include:

• • • • • Macroeconomic Uncertainty

        • Financial Development

        • Public Governance

        • Business Environment

        • Market Size and Potential

        • Human Resources

        • Logistic Performance

        • ICT Infrastructure

        • Agglomeration Effects

        • Innovation and Differentiation

These drivers are considered "latent," meaning they are not directly measurable but are influenced by observable factors. In essence, we start by assuming that these ten drivers shape a country's FDI attractiveness. Since the drivers themselves can't be directly measured, we estimate them using observable data.

This approach is applied consistently throughout the index construction process. For each latent driver, multiple proxy indicators are used to represent it. This ensures we are not reliant on a single data source, which may be biased or incomplete due to varying data collection methods across countries. By breaking down each driver into sub-categories and using a wide range of proxies, we maintain a comprehensive and balanced assessment.

The index incorporates 47 individual indicators to measure these ten key drivers. These indicators are sourced from multiple databases with annual data, covering the period from 1980 to 2022. Key sources include the GeoDist database (CEPII), World Development Indicators (World Bank), Institutional Profiles Database (DGTPE-France), Worldwide Governance Indicators (World Bank), The Conference Board Total Economy Database, and several others such as UNCTAD, WTO, ILO, and IMF databases. To ensure a more stable assessment and reduce the influence of extreme variations in any particular year, each individual indicator is evaluated based on the average of the last three observed years. This methodology has two main advantages.  First, it prevents any single data series from having disproportionate influence, minimizing the impact of outliers. Second, it allows for transparency and granularity, enabling users to trace the overall results back to more detailed levels for better interpretation and insight. 

GFICA Index Structure

The high values of Cronbach’s alpha and MSA, along with the extraction of a single factor that explains a substantial portion of data variance, indicate that the key axes effectively serve as joint proxies for a single latent factor. This unidimensionality reflects that the key axes capture a single characteristic, confirming the appropriate selection of key drivers to assess FDI attractiveness for the studied countries. FDI attractiveness is wellrepresented through the three criteria—prerequisites, underlying factors, and agglomeration-differentiation factors—as proxies.

Additionally, the PCA (Principal Component Analysis) produces communalities, which represent the total influence of all factors on a single observed item (in this case, only one factor is identified). These communalities are equivalent to the squared factor loading for each observed indicator and similar to R² in multiple regression,. They are used to calculate the weights for the three pillars, with the square of the factor loading representing the proportion of variance explained by the factors for each indicator.

Aggregation

There are three main types of aggregation methods: additive methods, geometric aggregation, and noncompensatory multi-criteria analysis. For FDI attractiveness analysis, the linear and geometric aggregation methods are the most suitable. Linear aggregation assigns base indicators in proportion to their respective weights, making it useful when all sub-indicators share the same measurement unit, as is the case here. However, this method tends to favor countries or sub-indicators with higher scores, as a shortfall in one area can be compensated by a surplus in another. In contrast, compensability is reduced in geometric aggregation, particularly for sub-indicators with lower values. This means that countries with weaker scores in some areas would benefit more from linear aggregation.

Normalization and Consistency Analysis

To ensure the comparability of cross-sectional data and facilitate index aggregation, the raw data must be standardized into a common scale. This is achieved through the min-max rescaling method, which normalizes the sub-indicators to a specified range. In this system, a score of 100 represents the highest performance, while a score of 1 reflects the lowest. High-quality testing is crucial for assessing the reliability of data in a research study, serving as the initial step in evaluating the consistency of indices before calculating composite variables and building explanatory models. One widely used measure for this purpose is Cronbach's alpha, which gauges internal consistency. The statistic rises as the inter-correlations among a set of sub-indicators increase. A high Cronbach's alpha (≥ 0.7, the acceptable threshold) indicates that the selected sub-indicators accurately reflect the key variable. Two additional measures commonly used for consistency evaluation are related to factor analysis and data summarization: the Kaiser-Meyer-Olkin (KMO) measure of sampling adequacy and Bartlett's test of sphericity. The KMO measure, based on partial correlations among input variables, should be ≥ 0.5 to justify factor analysis. Bartlett's test of sphericity, which assesses whether the correlation matrix is an identity matrix (i.e., the sub-indicators are correlated in the population), should return a p-value below 0.05. For the ten key drivers, the reliability statistics for the sub-indicators mostly exceed Nunnally's cut-off value of 0.7. Strong KMO values and significant results from Bartlett's test (p-values less than 0.05) are obtained for all key drivers. Based on these results, it is possible to conduct a valid factor analysis.

The Weighting Scheme 

Once the performance scores for each sub-item at the lowest level are calculated, the next step before aggregation is to determine the weightings of the index items. Two weighting schemes are applied. At the lowest level, the sub-items are aggregated using equal weights, meaning that each item's weight is based on the number of components being aggregated. For the key drivers level (comprising 10 key drivers), weights are assigned according to the number of items in each driver, and the same approach is applied to the three pillars.

Specifically, equal weights are applied at the lowest level, key drivers are aggregated using weights based on the number of items they contain, and at the pillar level, weights derived from factor analysis are used. In factor analysis, each component is assigned a weight according to its contribution to the total variance in the data. This ensures that the summary indicators capture a significant portion of the cross-country variance in the sub-indicators being considered.

The Cronbach's alpha across the three axes is 0.93, indicating a high quality of data selection for all countries analyzed. The MSA (Measure of Sampling Adequacy) score is 0.72, and Bartlett's Test of Sphericity is highly significant, with a p-value of 0.000. These results confirm the reliability and appropriateness of the data for further analysis. 

Countries

Riadh Ben Jelili - IAE Bretagne Sud - University of South Brittany