From Equal Weights to Empirical Weights: Reassessing Economic Freedom Rankings
Approximate Word Count (text only): 2,402
Keywords: Economic freedom; composite indices; weighting schemes; exploratory factor analysis; confirmatory factor analysis; rank sensitivity; measurement validity.
JEL Codes (suggested): C38, C43, C52, O57, P16.
1 Abstract
This paper evaluates the equal-weighting assumption in the Economic Freedom Index (EFI) using the 2024 Economic Freedom of the World data release (with a 2022 cross-sectional analysis). We argue that equal weighting should be treated as a modeling assumption rather than a neutral default. Using exploratory factor analysis (EFA), confirmatory factor analysis (CFA), and a rank-sensitivity exercise, we show that an empirical 4-factor structure yields highly unequal factor weights and materially changes country rankings. Even where aggregate rank correlations remain high, many countries move substantially. The main implication is methodological: EFI results should be accompanied by explicit weighting sensitivity analysis and clearer justification for the equal-weighting choice.
2 Motivation
The equal-weighting defense is usually transparency: it is simple, easy to explain, and avoids explicit normative claims about which dimensions matter more. That is a real advantage. But equal weighting can still be a strong assumption when:
- The index is hierarchical (areas, subareas, indicators).
- The groups are not empirically homogeneous.
- The maintained measurement structure itself is uncertain.
This is the problem addressed here. The question is not only whether the five EFI Areas should each receive 20%, but whether the imposed five-Area structure maps well onto the observed covariance structure of Components and Subcomponents.
3 Data and Scope
This analysis uses:
- The Fraser Institute Economic Freedom of the World 2024 data release (Gwartney et al. 2024).
- The researcher Excel file used to reconstruct and analyze Component/Subcomponent data (Fraser Institute 2024).
- A 2022 country cross-section for factor analysis and ranking comparisons.
The analysis is a measurement critique, not a causal claim about the effects of economic freedom on growth or welfare.
Terminology: throughout the paper, we follow the EFW hierarchy (Area, Component, Subcomponent). When referring to variables used in the factor models, we report their names and codes (e.g., X1A, X1D_i).
4 Literature Background and Terminology
4.1 EFW Hierarchy and Current Aggregation Structure
The Economic Freedom of the World (EFW) index is organized as a hierarchical composite measure using an Area-Component-Subcomponent structure (Gwartney et al. 2024). The published index groups the construct into five Areas: Size of Government, Legal System and Property Rights, Sound Money, Freedom to Trade Internationally, and Regulation of Credit, Labor, and Business (Gwartney et al. 2024). These Areas are further decomposed into Components and Subcomponents constructed from external data sources and then standardized to a common scale in the EFW methodology (Gwartney et al. 2024).
This hierarchy is a major strength of the EFW framework. It provides conceptual organization, supports transparent replication, and allows researchers to work at different levels of aggregation depending on the empirical question. In this paper, we retain that terminology throughout. We use “indicator-level variables” only as a generic measurement term when discussing the variables entering the EFA/CFA, while preserving the EFW hierarchy for interpretation.
4.2 Equal Weighting as a Methodological Choice
In the published EFW construction, aggregation proceeds by averaging within the hierarchy: Subcomponents are combined into Components, Components into Areas, and the five Areas are then combined into the overall summary index with equal Area weights (Gwartney et al. 2024). This approach has clear advantages. It is transparent, easy to communicate, and avoids embedding a highly specific theory of relative institutional importance where economics does not provide a consensus weighting rule.
At the same time, the composite-indicator literature is clear that equal weighting is not a neutral baseline; it is a substantive modeling choice (Nardo et al. 2008; Greco et al. 2019). Equal weights impose assumptions about relative importance and compensability across dimensions, and they can interact with correlation structure in ways that effectively amplify some dimensions and mute others.
4.3 Why Reassess Weights? Composite-Indicator and Rank-Robustness Perspectives
The methodological case for reexamining weighting in EFW follows three established concerns in the composite-indicator literature.
First, correlated Components/Subcomponents can create redundancy. When multiple variables move together, equal weighting can unintentionally double-count closely related institutional features. Dimension-reduction methods such as principal components and factor analysis are commonly used to diagnose latent structure and redundancy (Nardo et al. 2008; Greco et al. 2019).
Second, weighting decisions affect construct validity and statistical coherence. Even when a normative index is appropriate, the empirical covariance structure may not align with the imposed aggregation structure. Composite-indicator methodology therefore emphasizes sensitivity and uncertainty analysis as part of quality assessment, rather than as an optional robustness appendix (Saisana et al. 2005).
Third, rank robustness is central whenever cross-country rankings are used for policy comparison or downstream regression work. If modest changes in weighting generate large rank reversals, then the ranking should be interpreted more cautiously. Rank robustness and related sensitivity diagnostics are therefore directly relevant to how EFW-based rankings are used in applied research (Foster et al. 2009; Greco et al. 2019).
Structural equation modeling (SEM), including the EFA/CFA sequence used here, adds a useful bridge between composite-indicator design and econometric measurement practice: it treats economic freedom as a latent institutional construct, tests whether the observed covariance structure is consistent with the maintained Area-Component-Subcomponent organization, and makes the weighting assumptions empirically inspectable through estimated loadings and fit statistics rather than leaving them entirely implicit in equal-weight aggregation.
4.4 Implication for This Paper
Our contribution is not to reject the EFW hierarchy or its theoretical motivation. Instead, we treat equal weighting as a testable aggregation assumption within an otherwise useful and influential measurement framework. The empirical exercises below ask whether the observed covariance structure supports the current aggregation pattern and how much ranking movement arises when we replace equal weights with empirically derived factor weights.
5 Empirical Strategy
The workflow is:
- Reconstruct and clean the EFI Area/Component/Subcomponent dataset.
- Examine within-area correlations (descriptive).
- Run EFA to identify a data-driven latent structure.
- Construct empirical factor weights from the EFA solution.
- Compare country rankings under the original and empirical weighting schemes.
- Run CFA to compare an empirical 4-factor model with an EFI-style 5-factor model.
6 Main Results
6.1 Correlations Across the Five Official EFI Area Scores
This figure reports pairwise correlations among the five official EFI area scores (Area 1 through Area 5). Together with the Area 1 within-area correlation evidence reported next, it highlights an important distinction: area-level aggregates can comove positively even when individual indicators within an area exhibit substantial heterogeneity.
6.2 Descriptive Correlation Evidence for EFI Area 1 (Size of Government)
Before turning to EFA/CFA, it is useful to inspect the within-area correlation structure of the official EFI Area 1 indicators. Area 1 is especially useful as a motivating case because it includes both positive and negative within-area correlations, which highlights how an imposed area grouping can combine related but non-coextensive dimensions. We include the full set of analogous charts for all five EFI areas (including Area 2) in Appendix B.
| Area | MinCorr | MaxCorr |
|---|---|---|
| Area 1 | -0.321 | 0.706 |
| Area 2 | 0.389 | 0.924 |
| Area 3 | 0.038 | 0.603 |
| Area 4 | -0.041 | 0.763 |
| Area 5 | -0.102 | 0.843 |
Note: Pairwise correlations range from -0.32 to 0.71 (excluding the diagonal). Compact Component/Subcomponent codes are used for readability; full variable definitions appear in the appendix mapping/definition tables.
Note: A labeled Component/Subcomponent summary-statistics table (code plus label) is reported below in the main text. Appendix D1 reports the full table.
Interpretation:
- Area 1 exhibits notable heterogeneity, including negative pairwise correlations among some indicators.
- At the same time, the five official area scores can still be positively correlated with one another at the aggregate level.
- This combination supports our broader concern that equal aggregation within an official area can mask distinct underlying constructs.
- These figures are descriptive, but they help motivate the EFA/CFA results reported below.
6.3 Unequal Empirical Factor Weights
The EFA-based weighting exercise produces substantially unequal factor weights.
6.3.1 Summary Statistics (Sample and Rank Sensitivity Context)
The table below summarizes the comparison sample and the scale of rank movement under empirical weighting. This gives readers a compact baseline before the detailed EFA/CFA results.
| Metric | Value |
|---|---|
| Countries in rank-comparison sample | 108 |
| EFI score (original): mean | 6.821 |
| EFI score (original): SD | 0.999 |
| EFI score (original): min / max | 3.530 / 8.580 |
| Empirical weighted score (scaled): mean | 5 |
| Empirical weighted score (scaled): SD | 2.5 |
| Empirical weighted score (scaled): min / max | -0.075 / 9.518 |
| Absolute rank change (weighted): mean | 19.5 |
| Absolute rank change (weighted): median | 14 |
| Absolute rank change (weighted): max | 71 |
Component/Subcomponent summary statistics are reported below (labeled Components/Subcomponents only) and in Appendix D1 (full table).
6.3.2 Summary Statistics (Labeled Components and Subcomponents Only)
This table includes only Components/Subcomponents with explicit labels in the analysis code (code plus label). It is provided as a concise companion to the full appendix table.
| Code: Variable | N | Mean | SD | Min | Max |
|---|---|---|---|---|---|
| X1A: Government consumption | 165 | 5.47 | 2.46 | 0.00 | 10.00 |
| X1B: Transfers and subsidies | 156 | 7.53 | 2.05 | 1.82 | 10.00 |
| X1C: Government investment | 161 | 7.12 | 3.28 | 0.00 | 10.00 |
| X1D_i: Top marginal income tax rate | 165 | 7.47 | 2.19 | 2.00 | 10.00 |
| X1D_ii: Top marginal income and payroll tax rate | 162 | 5.48 | 2.63 | 0.00 | 10.00 |
| X1E: State ownership of assets | 162 | 6.66 | 1.58 | 2.44 | 9.52 |
| X2A_scaled: Judicial independence* | 165 | 5.20 | 1.59 | 1.97 | 8.64 |
| X2B_scaled: Impartial courts* | 165 | 4.42 | 1.96 | 0.68 | 8.84 |
| X2C_scaled: Property rights* | 165 | 5.16 | 2.23 | 0.00 | 9.67 |
| X2D_scaled: Military interference* | 138 | 5.99 | 2.75 | 0.00 | 10.00 |
| X2E_scaled: Legal integrity* | 164 | 5.52 | 1.97 | 1.55 | 9.80 |
| X2F_scaled: Contract enforcement* | 165 | 3.78 | 2.02 | 0.00 | 8.73 |
| X2G_scaled: Real property restrictions* | 163 | 7.13 | 1.66 | 2.35 | 9.98 |
| X2H_scaled: Police reliability* | 165 | 4.92 | 2.38 | 0.00 | 9.77 |
| X3A: Money growth | 163 | 8.24 | 1.61 | 0.00 | 9.98 |
| X3B: Inflation volatility | 165 | 7.65 | 2.52 | 0.00 | 9.74 |
| X3C: Recent inflation | 165 | 5.88 | 2.61 | 0.00 | 9.46 |
| X3D: Foreign currency accounts | 165 | 7.15 | 3.95 | 0.00 | 10.00 |
| X4A_i: Trade tax revenue | 156 | 8.63 | 1.47 | 3.33 | 10.00 |
| X4A_ii: Mean tariff rate | 164 | 8.26 | 1.02 | 3.50 | 10.00 |
| X4A_iii: Tariff dispersion | 165 | 6.06 | 2.13 | 0.00 | 10.00 |
| X4B_i: Non-tariff barriers | 165 | 5.75 | 1.80 | 0.00 | 9.18 |
| X4B_ii: Trade compliance costs | 165 | 6.46 | 3.01 | 0.00 | 9.98 |
| X4C: Black market exchange rates | 165 | 9.22 | 2.47 | 0.00 | 10.00 |
| X4D_i: Financial openness | 165 | 5.84 | 3.02 | 0.00 | 10.00 |
| X4D_ii: Capital controls | 164 | 3.52 | 2.76 | 0.00 | 10.00 |
| X4D_iii: Freedom of foreigners to visit | 165 | 7.10 | 3.38 | 0.00 | 10.00 |
| X4D_iv: Protection of foreign assets | 165 | 5.73 | 2.10 | 0.00 | 9.27 |
| X5A_i: Ownership of banks | 151 | 7.59 | 2.78 | 0.00 | 10.00 |
| X5A_ii: Private sector credit | 162 | 7.39 | 2.53 | 0.00 | 10.00 |
| X5A_iii: Interest rate controls | 163 | 8.35 | 2.27 | 0.00 | 10.00 |
| X5B_i: Hiring regulations and minimum wage | 165 | 5.81 | 1.80 | 1.25 | 9.45 |
| X5B_ii: Hiring and firing regulations | 165 | 5.64 | 1.97 | 0.00 | 10.00 |
| X5B_iii: Centralized collective bargaining | 141 | 6.48 | 1.22 | 1.61 | 8.67 |
| X5B_iv: Hours regulations | 165 | 7.84 | 1.94 | 2.00 | 10.00 |
| X5B_v: Mandated cost of worker dismissal | 162 | 6.49 | 2.83 | 0.00 | 10.00 |
| X5B_vi: Conscription | 165 | 6.61 | 4.20 | 0.00 | 10.00 |
| X5B_vii: Foreign labor | 165 | 5.02 | 1.49 | 0.00 | 8.14 |
| X5C_i: Regulatory burden | 158 | 4.15 | 1.22 | 1.14 | 7.44 |
| X5C_ii: Bureaucracy costs | 165 | 5.04 | 2.22 | 0.00 | 9.56 |
| X5C_iii: Impartial public administration | 162 | 5.62 | 2.48 | 0.34 | 9.91 |
| X5C_iv: Tax compliance | 165 | 5.90 | 1.92 | 0.00 | 9.87 |
| X5D_i: Market openness | 165 | 6.07 | 1.80 | 1.24 | 10.00 |
| X5D_ii: Business permits | 161 | 8.39 | 1.14 | 4.98 | 10.00 |
| X5D_iii: Distortion of business environment | 165 | 5.20 | 2.29 | 0.00 | 10.00 |
Note: EFI construction is hierarchical. Subcomponents with suffixes (e.g., X1D_i and X1D_ii) are typically combined with equal weights to form a parent Component (e.g., X1D), and Components within an Area (e.g., X1A through X1E for Area 1) are then combined to form the Area score (e.g., Area 1). The table reports the lower-level Components/Subcomponents listed in the analysis code.*
6.3.3 Reported Empirical Weights
| Factor | Weight |
|---|---|
| Factor 1 | 39.2% |
| Factor 2 | 36.3% |
| Factor 3 | 15.0% |
| Factor 4 | 9.5% |
These weights are far from an equal 25/25/25/25 distribution across the empirical factors (and even further from the EFI’s equal 20/20/20/20/20 area weighting logic).
6.4 What the Four Empirical Factors Appear to Capture (and Why We Use These Names)
We view the factor-labeling exercise as part of the paper’s contribution. A weighting critique is more persuasive when we show not only that the data imply different weights, but also what the new factors are actually measuring.
The labels below are based on the primary loading assignments from the 4-factor EFA in scripts/lavaan12_cleanV6.html (loadings threshold: |loading| > 0.30). We interpret factor names from the cluster of indicators, not any single variable.
| Empirical Factor | Proposed Name | Main Indicators from EFA | Why This Name Is Justified |
|---|---|---|---|
| Factor 1 | Legal-Institutional Quality and State Burden | Area 2 legal-quality indicators (Judicial independence (X2A_scaled), Impartial courts (X2B_scaled), Legal integrity (X2E_scaled), Police reliability (X2H_scaled), Contract enforcement (X2F_scaled)) plus public administration quality (X5C_iii) and negative loadings on Government consumption (X1A) / Transfers and subsidies (X1B) | The dominant loadings are legal/institutional quality indicators, but the factor also captures government burden (via negative loadings on X1A and X1B), so a combined institutional-state-burden label is more accurate than a pure 'legal' label. |
| Factor 2 | External Openness and Market Access | Trade and cross-border openness indicators (Financial openness (X4D_i), Protection of foreign assets (X4D_iv), Non-tariff barriers (X4B_i), Mean tariff rate (X4A_ii), Capital controls (X4D_ii)), plus Market openness (X5D_i) and Foreign labor (X5B_vii) | This factor is anchored by international openness and access conditions. It captures how easily firms, capital, and people interact across borders and markets, which goes beyond a narrow 'trade' label. |
| Factor 3 | Domestic Regulatory and Administrative Frictions | Regulatory burden (X5C_i), Tax compliance (X5C_iv), Hiring regulations and minimum wage (X5B_i), Business permits (X5D_ii), and related tax-policy/administrative items (including X1D_ii) | The strongest loadings are domestic regulatory and compliance costs. The factor appears to capture business-facing frictions inside the domestic economy rather than external openness. |
| Factor 4 | Labor Market Adjustment Costs | Hiring and firing regulations (X5B_ii) and Mandated cost of worker dismissal (X5B_v) | This factor is highly concentrated in labor dismissal and employment-adjustment rules, making the labor-adjustment-cost label both parsimonious and empirically grounded. |
6.4.1 Factor Decomposition (Primary Mapping from Original Components and Subcomponents)
To make the new structure transparent for readers, the table below shows how original EFI Components and Subcomponents map into the four empirical factors (based on primary loading assignment in the EFA output).
| Empirical Factor | Mapped Indicators (illustrative primary assignments) |
|---|---|
| Factor 1 | Integrity of the legal system (X2E_scaled); Judicial independence (X2A_scaled); Impartial courts (X2B_scaled); Military interference in rule of law and politics (X2D_scaled); Reliability of police (X2H_scaled); Protection of property rights (X2C_scaled); Legal enforcement of contracts (X2F_scaled); Impartial public administration (X5C_iii); Government consumption (X1A, negative); Transfers and subsidies (X1B, negative); plus Bureaucracy costs (X5C_ii), Trade compliance costs (X4B_ii), Top marginal income tax rate (X1D_i), and Government investment (X1C) |
| Factor 2 | Freedom to enter markets and compete / Distortion of business environment (X5D_iii); Protection of foreign assets (X4D_iv); Market openness (X5D_i); Non-tariff trade barriers (X4B_i); Financial openness (X4D_i); Foreign labor (X5B_vii); State ownership of assets (X1E); Black market exchange rates (X4C); Trade taxes / tariffs / capital controls (X4A_i, X4A_ii, X4D_ii); plus some sound-money and legal spillovers |
| Factor 3 | Regulatory burden (X5C_i); Hiring regulations and minimum wage (X5B_i); Tax compliance (X5C_iv); Business permits (X5D_ii); Top marginal income and payroll tax rate (X1D_ii); Centralized collective bargaining (X5B_iii); Money growth (X3A) |
| Factor 4 | Hiring and firing regulations (X5B_ii); Mandated cost of worker dismissal (X5B_v) |
6.4.2 EFA Loading Heatmap (Paper-Friendly Visual Summary)
Note: This heatmap is exported from the analysis script (lavaan12_cleanV7.qmd) using the same 4-factor EFA specification and low-MSA exclusions as the main analysis. Rows are sorted by primary loading. Labels are printed only for |loading| >= 0.30 to keep the figure readable.
Why this matters for the paper:
- The empirical factors do not line up one-for-one with the original five EFI areas.
- The new factors are substantively interpretable, which strengthens the case that the EFA is not just a statistical artifact.
- Naming and decomposing the factors clarifies what receives higher or lower empirical weight in the re-ranking exercise.
Important caveat:
- Factor labels are interpretive. They are justified by dominant loading patterns, but cross-loadings imply overlap across constructs.
6.5 Ranking Sensitivity Is Substantive
The empirical weighting exercise produces meaningful ranking changes.
| Metric | Value |
|---|---|
| Countries in comparison sample | 108 |
| Spearman correlation (original vs empirical weighted rank) | 0.939 |
| Pearson correlation (original vs empirical weighted rank) | 0.926 |
| Mean absolute rank change | 19.5 |
| Median absolute rank change | 14 |
| Countries with absolute rank change > 10 | 64 (59.3%) |
| Countries with absolute rank change > 20 | 41 (38.0%) |
Interpretation:
- Correlation remains high (
Spearman = 0.939), so the broad ordering is not completely overturned. - However, the average absolute rank change is
19.5, which is not trivial. 41countries (about38.0%) move by more than 20 ranks.
This is the central ranking argument: high overall correlation does not imply that weighting choices are substantively unimportant.
6.6 Illustrative Movers
| Countries | Rank | Rank_Empirical_Weighted | Rank_Change_Weighted |
|---|---|---|---|
| Argentina | 159 | 88 | 71 |
| Türkiye | 138 | 72 | 66 |
| Ethiopia | 148 | 89 | 59 |
| Algeria | 161 | 103 | 58 |
| Ukraine | 150 | 93 | 57 |
| Zimbabwe | 164 | 107 | 57 |
| Angola | 155 | 104 | 51 |
| Iran, Islamic Rep. | 158 | 108 | 50 |
| Sri Lanka | 123 | 74 | 49 |
| Malawi | 140 | 92 | 48 |
| Countries | Rank | Rank_Empirical_Weighted | Rank_Change_Weighted |
|---|---|---|---|
| El Salvador | 59 | 81 | -22 |
| Panama | 23 | 43 | -20 |
| Guatemala | 39 | 59 | -20 |
| Honduras | 67 | 86 | -19 |
| Armenia | 36 | 53 | -17 |
| Malaysia | 29 | 45 | -16 |
| Albania | 38 | 52 | -14 |
| Bahrain | 34 | 46 | -12 |
| Indonesia | 64 | 76 | -12 |
| Philippines | 59 | 70 | -11 |
These tables are not presented as evidence that the empirical ranking is normatively superior. They show that the published rank order is sensitive to plausible alternative weighting assumptions.
6.7 CFA Comparison: Empirical 4-Factor vs EFI-Style 5-Factor
The CFA comparison indicates that the empirical 4-factor model performs better on several comparative fit measures and information criteria, although both models exhibit poor absolute fit.
| Model | CFI | TLI | RMSEA | SRMR | AIC | BIC |
|---|---|---|---|---|---|---|
| 4-factor (empirical) | 0.831 | 0.806 | 0.148 | 0.102 | 7439.3 | 7624.4 |
| 5-factor (EFI-style) | 0.629 | 0.604 | 0.133 | 0.121 | 16302.1 | 16658.8 |
| Statistic | Value |
|---|---|
| Chi-square difference | 1621.8 |
| Difference in df | 586.0 |
| p-value | 0.0 |
Key interpretation:
- The empirical 4-factor model improves
CFI,TLI,SRMR,AIC, andBIC. - The EFI-style 5-factor model has a lower reported
RMSEAin this comparison. - The practical conclusion is therefore a relative-fit claim, not a claim that the empirical model is well-fitting in absolute terms.
This is an important point for the paper tone: the results support skepticism toward the imposed equal-weighted structure, but they do not yet justify strong claims of a fully validated replacement model.
7 What the Evidence Supports (and Does Not Support)
7.1 What it argues
- Equal weighting in the EFI is a contestable methodological choice.
- The data support a materially different empirical weighting pattern.
- Country rankings are sensitive to weighting assumptions.
- The imposed five-area structure does not appear uniquely privileged by the current factor-model evidence.
7.2 What it does not yet argue
- That the empirical 4-factor model is the definitive structure of economic freedom.
- That one alternative weighting scheme should replace EFI in all contexts.
- That the present analysis establishes causal validity.
8 Limitations
- The analysis currently emphasizes one cross-section (2022).
- EFA/CFA results may be sensitive to missing-data handling, variable exclusions, rotation, and estimation choices.
- The CFA comparison includes a warning that the models are based on different sets of observed variables; this needs to be discussed carefully.
- Absolute fit is weak for both compared models, which limits strong structural claims.
- The empirical weights are useful as a sensitivity benchmark, but not necessarily a final index design recommendation.
9 Proposed Revisions
- Add a short literature review on composite-index weighting (equal weights, PCA/factor-based weighting, and robustness analysis).
- Run multi-year robustness checks (not only 2022).
- Add a short appendix documenting Component/Subcomponent exclusions and missingness effects.
- Tighten the interpretation of CFA results, especially the mixed fit signal (
RMSEAvsCFI/TLI/SRMR/AIC/BIC). - Add a results appendix comparing
psych::fa()andlavaanEFA outputs.
10 Conclusion
The current evidence supports a clear methodological conclusion: equal weighting in the Economic Freedom Index should not be treated as a neutral default. A data-driven factor structure yields markedly unequal empirical weights, and those weights produce substantial rank reordering for many countries. Even if one does not adopt the empirical weights as a replacement index, the results make a strong case for reporting weighting sensitivity analyses alongside headline EFI rankings.
11 Appendix A: Reference Figure
The figure below is included for reference/context and moved out of the main results so the core empirical argument reads more cleanly.
12 Appendix B: Area-Wise Correlation Charts (All Five EFI Areas)
These charts use the same red-white-green correlation format as the main-text Area 2 figure and are included here as reference diagnostics for all official EFI areas.
12.1 Area 1 (Size of Government)
Note: Pairwise correlations range from -0.32 to 0.71 (excluding the diagonal). Compact Component/Subcomponent codes are used for readability; full variable definitions appear in the appendix mapping/definition tables.
12.2 Area 2 (Legal System and Property Rights)
Note: Pairwise correlations range from 0.39 to 0.92 (excluding the diagonal). Compact Component/Subcomponent codes are used for readability; full variable definitions appear in the appendix mapping/definition tables.
12.3 Area 3 (Sound Money)
Note: Pairwise correlations range from 0.04 to 0.60 (excluding the diagonal). Compact Component/Subcomponent codes are used for readability; full variable definitions appear in the appendix mapping/definition tables.
12.4 Area 4 (Freedom to Trade Internationally)
Note: Pairwise correlations range from -0.04 to 0.76 (excluding the diagonal). Compact Component/Subcomponent codes are used for readability; full variable definitions appear in the appendix mapping/definition tables.
12.5 Area 5 (Regulation)
Note: Pairwise correlations range from -0.10 to 0.84 (excluding the diagonal). Compact Component/Subcomponent codes are used for readability; full variable definitions appear in the appendix mapping/definition tables.
13 Appendix C: Reproducibility Notes
This paper combines:
- Results reported in the manuscript are generated from the analysis workflow and exported tables/figures.
- A full analysis notebook (including EFA factor weights and CFA fit comparisons) is available from the authors on request (
lavaan12_cleanV7).
Before finalizing the manuscript, re-run the analysis notebook and update the reported values if they change.
14 Appendix D: Summary Statistics - Non-Standardized Variables
This appendix reports descriptive statistics for the Component/Subcomponent variables used to construct the EFA input. Variables are reported in their original coding (not standardized z-scores). For Area 2, * indicates the gender-adjusted scaled variables used in the factor-analysis pipeline.
14.1 Appendix D1: Full Summary Statistics Table (All Variables)
This version preserves the broader non-standardized variable list for auditability and cross-checking.
| Code: Variable | N | Mean | SD | Min | Max |
|---|---|---|---|---|---|
| Year | 165 | 2022.00 | 0.00 | 2022.00 | 2022.00 |
| Economic Freedom Summary Index | 165 | 6.53 | 1.05 | 3.02 | 8.58 |
| Rank | 165 | 82.78 | 47.80 | 1.00 | 165.00 |
| X1A: Government consumption | 165 | 5.47 | 2.46 | 0.00 | 10.00 |
| X1B: Transfers and subsidies | 156 | 7.53 | 2.05 | 1.82 | 10.00 |
| X1C: Government investment | 161 | 7.12 | 3.28 | 0.00 | 10.00 |
| X1D_i: Top marginal income tax rate | 165 | 7.47 | 2.19 | 2.00 | 10.00 |
| X1D_ii: Top marginal income and payroll tax rate | 162 | 5.48 | 2.63 | 0.00 | 10.00 |
| X1D | 165 | 6.50 | 2.23 | 1.00 | 10.00 |
| X1E: State ownership of assets | 162 | 6.66 | 1.58 | 2.44 | 9.52 |
| Area1 | 165 | 6.64 | 1.12 | 3.62 | 9.06 |
| Area1_rank | 165 | 83.00 | 47.78 | 1.00 | 165.00 |
| X2A | 165 | 5.51 | 1.45 | 2.45 | 8.64 |
| X2B | 165 | 4.66 | 1.89 | 0.83 | 8.84 |
| X2C | 165 | 5.44 | 2.16 | 0.00 | 9.67 |
| X2D | 138 | 6.29 | 2.66 | 0.00 | 10.00 |
| X2E | 164 | 5.83 | 1.85 | 1.70 | 9.80 |
| X2F | 165 | 3.97 | 1.98 | 0.00 | 8.73 |
| X2G | 163 | 7.59 | 1.52 | 2.67 | 9.98 |
| X2H | 165 | 5.21 | 2.42 | 0.00 | 9.77 |
| X2_adj | 165 | 0.87 | 0.17 | 0.29 | 1.00 |
| X2_wo_gen | 165 | 5.53 | 1.66 | 1.98 | 9.10 |
| X2_with_gen | 165 | 5.23 | 1.79 | 1.63 | 9.10 |
| Area2 | 165 | 5.23 | 1.79 | 1.63 | 9.10 |
| Area2_rank | 165 | 83.00 | 47.78 | 1.00 | 165.00 |
| X2A_scaled: Judicial independence* | 165 | 5.20 | 1.59 | 1.97 | 8.64 |
| X2B_scaled: Impartial courts* | 165 | 4.42 | 1.96 | 0.68 | 8.84 |
| X2C_scaled: Property rights* | 165 | 5.16 | 2.23 | 0.00 | 9.67 |
| X2D_scaled: Military interference* | 138 | 5.99 | 2.75 | 0.00 | 10.00 |
| X2E_scaled: Legal integrity* | 164 | 5.52 | 1.97 | 1.55 | 9.80 |
| X2F_scaled: Contract enforcement* | 165 | 3.78 | 2.02 | 0.00 | 8.73 |
| X2G_scaled: Real property restrictions* | 163 | 7.13 | 1.66 | 2.35 | 9.98 |
| X2H_scaled: Police reliability* | 165 | 4.92 | 2.38 | 0.00 | 9.77 |
| X3A: Money growth | 163 | 8.24 | 1.61 | 0.00 | 9.98 |
| X3B: Inflation volatility | 165 | 7.65 | 2.52 | 0.00 | 9.74 |
| X3C: Recent inflation | 165 | 5.88 | 2.61 | 0.00 | 9.46 |
| X3D: Foreign currency accounts | 165 | 7.15 | 3.95 | 0.00 | 10.00 |
| Area3 | 165 | 7.23 | 1.71 | 0.74 | 9.55 |
| Area3_rank | 165 | 82.99 | 47.77 | 1.00 | 165.00 |
| X4A_i: Trade tax revenue | 156 | 8.63 | 1.47 | 3.33 | 10.00 |
| X4A_ii: Mean tariff rate | 164 | 8.26 | 1.02 | 3.50 | 10.00 |
| X4A_iii: Tariff dispersion | 165 | 6.06 | 2.13 | 0.00 | 10.00 |
| X4A | 165 | 7.62 | 1.18 | 3.92 | 10.00 |
| X4B_i: Non-tariff barriers | 165 | 5.75 | 1.80 | 0.00 | 9.18 |
| X4B_ii: Trade compliance costs | 165 | 6.46 | 3.01 | 0.00 | 9.98 |
| X4B | 165 | 6.11 | 2.11 | 1.69 | 9.41 |
| X4C: Black market exchange rates | 165 | 9.22 | 2.47 | 0.00 | 10.00 |
| X4D_i: Financial openness | 165 | 5.84 | 3.02 | 0.00 | 10.00 |
| X4D_ii: Capital controls | 164 | 3.52 | 2.76 | 0.00 | 10.00 |
| X4D_iii: Freedom of foreigners to visit | 165 | 7.10 | 3.38 | 0.00 | 10.00 |
| X4D_iv: Protection of foreign assets | 165 | 5.73 | 2.10 | 0.00 | 9.27 |
| X4D | 165 | 5.55 | 2.15 | 0.69 | 9.40 |
| Area4 | 165 | 7.12 | 1.49 | 2.48 | 9.66 |
| Area4_rank | 165 | 83.00 | 47.78 | 1.00 | 165.00 |
| X5A_i: Ownership of banks | 151 | 7.59 | 2.78 | 0.00 | 10.00 |
| X5A_ii: Private sector credit | 162 | 7.39 | 2.53 | 0.00 | 10.00 |
| X5A_iii: Interest rate controls | 163 | 8.35 | 2.27 | 0.00 | 10.00 |
| X5A | 164 | 7.74 | 1.63 | 0.00 | 10.00 |
| X5B_i: Hiring regulations and minimum wage | 165 | 5.81 | 1.80 | 1.25 | 9.45 |
| X5B_ii: Hiring and firing regulations | 165 | 5.64 | 1.97 | 0.00 | 10.00 |
| X5B_iii: Centralized collective bargaining | 141 | 6.48 | 1.22 | 1.61 | 8.67 |
| X5B_iv: Hours regulations | 165 | 7.84 | 1.94 | 2.00 | 10.00 |
| X5B_v: Mandated cost of worker dismissal | 162 | 6.49 | 2.83 | 0.00 | 10.00 |
| X5B_vi: Conscription | 165 | 6.61 | 4.20 | 0.00 | 10.00 |
| X5B_vii: Foreign labor | 165 | 5.02 | 1.49 | 0.00 | 8.14 |
| X5B | 165 | 6.25 | 1.22 | 2.67 | 9.14 |
| X5C_i: Regulatory burden | 158 | 4.15 | 1.22 | 1.14 | 7.44 |
| X5C_ii: Bureaucracy costs | 165 | 5.04 | 2.22 | 0.00 | 9.56 |
| X5C_iii: Impartial public administration | 162 | 5.62 | 2.48 | 0.34 | 9.91 |
| X5C_iv: Tax compliance | 165 | 5.90 | 1.92 | 0.00 | 9.87 |
| X5C | 165 | 5.17 | 1.55 | 0.60 | 9.05 |
| X5D_i: Market openness | 165 | 6.07 | 1.80 | 1.24 | 10.00 |
| X5D_ii: Business permits | 161 | 8.39 | 1.14 | 4.98 | 10.00 |
| X5D_iii: Distortion of business environment | 165 | 5.20 | 2.29 | 0.00 | 10.00 |
| X5D | 165 | 6.52 | 1.44 | 2.59 | 9.86 |
| Area5 | 165 | 6.41 | 1.13 | 2.54 | 8.86 |
| Area5_rank | 165 | 83.00 | 47.78 | 1.00 | 165.00 |
References
Foster, James, Mark McGillivray, and Suman Seth. 2009. Rank Robustness of Composite Indices. OPHI Working Paper 26. Oxford Poverty; Human Development Initiative, Queen Elizabeth House, University of Oxford. https://EconPapers.repec.org/RePEc:qeh:ophiwp:ophiwp026.
Fraser Institute. 2024. Economic Freedom of the World 2024 Master Index Data for Researchers (XLSX). Dataset accompanying Economic Freedom of the World 2024. https://www.fraserinstitute.org/studies/economic-freedom-of-the-world-2024-annual-report.
Greco, Salvatore, Alessio Ishizaka, Menelaos Tasiou, and Gianpiero Torrisi. 2019. “On the Methodological Framework of Composite Indices: A Review of the Issues of Weighting, Aggregation, and Robustness.” Social Indicators Research 141 (1): 61–94. https://doi.org/10.1007/s11205-017-1832-9.
Gwartney, James, Robert A. Lawson, Ryan Murphy, et al. 2024. Economic Freedom of the World: 2024 Annual Report. Fraser Institute. https://www.fraserinstitute.org/studies/economic-freedom-of-the-world-2024-annual-report.
Nardo, Michela, Michaela Saisana, Andrea Saltelli, Stefano Tarantola, Anders Hoffmann, and Enrico Giovannini. 2008. Handbook on Constructing Composite Indicators: Methodology and User Guide. OECD Publishing. https://www.oecd.org/content/dam/oecd/en/publications/reports/2008/08/handbook-on-constructing-composite-indicators-methodology-and-user-guide_g1gh9301/9789264043466-en.pdf.
Saisana, Michaela, Andrea Saltelli, and Stefano Tarantola. 2005. “Uncertainty and Sensitivity Analysis Techniques as Tools for the Quality Assessment of Composite Indicators.” Journal of the Royal Statistical Society: Series A (Statistics in Society) 168 (2): 307–23. https://doi.org/10.1111/j.1467-985X.2005.00350.x.