Construction and Validation of a 5P–ESG Composite Index for Sustainable Corporate Governance and Financial Analysis in Emerging Markets: Evidence from the MSCI COLCAP

Abstract

This study develops and validates the 5P–ESG Composite Index as a finance-oriented framework for assessing firm-level sustainable-financial performance in emerging markets. It addresses a persistent limitation in ESG measurement, namely the lack of conceptually integrated and decision-useful metrics capable of incorporating not only environmental, social, and governance dimensions, but also institutional and relational dimensions that are especially relevant in heterogeneous emerging-market settings. Conceptually, the proposed framework is grounded in the 2030 Agenda’s 5Ps (People, Planet, Prosperity, Peace, and Partnerships) and extends conventional ESG approaches by explicitly incorporating Peace and Partnerships into firm-level assessment. Methodologically, the index is constructed through sequential indicator selection, data cleansing, winsorization, normalization, pillar-level scoring, and PCA-based endogenous weighting, while its statistical robustness is assessed through internal consistency tests and factorability diagnostics. Empirically, the framework is applied to issuers in the MSCI COLCAP universe, where it proves operationally feasible and suitable for classifying firms into relative performance groups. In addition, a benchmark comparison against a conventional ESG-3 scheme shows that the broader 5P architecture can modify issuer rankings and tercile classification. Overall, the findings support the proposed index as a transparent, auditable, and context-sensitive tool for investors and decision-makers seeking more comprehensive sustainability metrics in emerging markets.

1. Introduction

Over the past decade, capital markets have shifted from a strictly financial interpretation of corporate performance toward a broader assessment of risks and opportunities associated with environmental, social, and governance (ESG) factors [1,2]. This transition is supported not only by ethical considerations but also by growing empirical evidence showing that ESG-related factors such as environmental externalities, institutional quality, human capital, and stakeholder management can materially affect corporate risk profiles and financial outcomes [3]. These effects may manifest through cash flow volatility, reputational shocks, litigation exposure, operational disruptions, and regulatory constraints, ultimately influencing the cost of capital and firm value [1,2,3]. While meta-analytical evidence generally reports a non-negative relationship between ESG and financial performance, results remain heterogeneous across sectors, time horizons, and measurement approaches [4].

In parallel, sustainable investing has evolved into a structural component of global capital markets. Investor demand increasingly relies on simplified signals such as ESG ratings and rankings, creating incentives for issuers and asset managers to adopt standardized metrics [5]. From an asset pricing perspective, sustainability may affect valuation through two main channels: a risk channel, where exposure to ESG-related risks commands a premium, and a preference channel, where investors derive non-pecuniary utility from sustainable assets and may accept lower expected returns [6,7]. These mechanisms have distinct implications for expected returns and the cost of capital.

Despite its growing importance, ESG measurement remains fragmented. The proliferation of ratings from providers such as MSCI, Refinitiv, and Sustainalytics has led to significant divergence across scores [8,9]. This divergence stems primarily from differences in measurement scope and methodological choices rather than weighting schemes alone, generating uncertainty and limiting comparability for investors, issuers, and regulators [9,10]. Furthermore, increased ESG disclosure does not necessarily reduce disagreement and may even amplify it, underscoring the need for greater methodological transparency and conceptual consistency [11].

These limitations are particularly pronounced in emerging markets, where non-financial information tends to be heterogeneous, fragmented, and often narrative in nature. In such contexts, global ESG frameworks may fail to capture institutional dimensions that are critical for firm-level risk assessment, introducing biases in coverage and comparability. This is especially relevant in environments where country risk, governance quality, and institutional stability play a central role in shaping corporate performance.

Against this backdrop, there is a need for a more integrated and conceptually consistent framework that allows sustainability to be incorporated into financial analysis without reducing it to a fragmented checklist of indicators. The United Nations 2030 Agenda provides such a structure through the “5Ps”—People, Planet, Prosperity, Peace, and Partnerships—which organize the Sustainable Development Goals into five interrelated dimensions encompassing social, environmental, economic, and institutional aspects of development [12]. From an operational perspective, the research problem can be formulated as follows: how can the 5Ps be translated into a comparable firm-level metric suitable for screening, coverage prioritization, and fundamental analysis, while addressing common limitations of ESG indices such as collinearity, double counting, and arbitrary weighting? The literature on composite indicators emphasizes that index construction should follow a transparent and rigorous process, including variable selection, data cleansing, normalization, assessment of latent structure, and defensible aggregation rules [13,14].

In response, this study develops and validates the 5P–ESG Composite Index as a finance-oriented framework for assessing firm-level sustainable-financial performance in the MSCI COLCAP universe. The methodological approach combines indicator refinement, data cleansing, normalization, pillar-level scoring, and weighted aggregation using endogenous weights derived from principal component analysis (PCA), with the aim of improving comparability, reducing collinearity, and strengthening the transparency of composite-index construction.

The objective of this study is to operationalize the 2030 Agenda’s 5P architecture at the issuer level and to examine whether this broader multidimensional structure leads to classification outcomes that differ from those obtained through a conventional ESG-3 framework. The contribution of the study is twofold. Conceptually, it extends firm-level sustainability assessment beyond the standard environmental, social, and governance structure by explicitly incorporating the institutional and relational dimensions represented by Peace and Partnerships. Methodologically, it integrates sequential indicator refinement, statistical validation, endogenous weighting, and benchmark comparison in order to construct a transparent, replicable, and decision-useful instrument for emerging-market analysis.

To make this added value explicit, the analysis includes a benchmarking exercise comparing issuer rankings under the 5P–ESG framework with those derived from a conventional ESG-3 score. This comparison highlights differences in issuer ordering and tercile classification resulting from the incorporation of dimensions that remain underrepresented in standard ESG aggregation.

2. Theoretical Framework

2.1. Financial Analysis and Decision-Making in Emerging Markets

Traditional financial analysis has historically relied on the examination of financial statements through indicators of profitability, liquidity, operational efficiency, and solvency as the primary basis for assessing corporate performance and supporting investment and financing decisions [15].

However, in emerging markets typically characterized by higher volatility, information asymmetries, and disclosure gaps, this approach may be insufficient to capture risks and opportunities that are not immediately reflected in accounting figures, yet materially affect the cost of capital, market perception, and the sustainability of cash flows. Empirical evidence indicates that higher levels of disclosure are associated with a lower cost of capital, and that this effect tends to be stronger in environments with weaker disclosure regimes, a common feature of emerging economies [16,17].

In this context, a purely financial approach faces limitations in incorporating social and environmental externalities that, although non-financial in nature, may translate into material risks—regulatory, operational, reputational, or transition-related—with direct implications for firm value. Financial research has documented that stronger corporate social responsibility and ESG performance are associated with a lower cost of capital, suggesting that these factors operate as risk mitigators and/or improve the firm’s information environment [18]. The principle of materiality is central in this relationship: effects are more robust when sustainability is managed and measured around issues that are financially material to the industry, rather than through decorative or symbolic variables that do not affect value creation or risk mitigation [19].

In contexts such as Colombia, where sustainability reporting frameworks and practices remain regionally heterogeneous, there is a persistent risk that financial analysis becomes disconnected from qualitative and non-financial variables that anticipate future contingencies. Sectoral and regional reports highlight progress in ESG disclosure across Latin America, but also reveal persistent gaps in comparability and consistency that limit effective benchmarking and integrated assessment by investors [20].

Moreover, even when ESG metrics are available, their interpretation requires caution, as ratings may diverge substantially due to differences in scope, measurement, and weighting, potentially leading to distorted conclusions if underlying assumptions and methodologies are not made explicit [21]. Consequently, a robust agenda for emerging markets involves integrating financial analysis with sustainability measures under criteria of materiality, traceability, and comparability, while recognizing the role of disclosure regulation in improving the information environment and the decision usefulness of non-financial information [22].

2.2. Corporate Sustainability and ESG Reporting

Corporate sustainability refers to an organization’s ability to sustain value creation over time by integrating economic performance with social and environmental responsibilities, moving beyond short-term profit maximization [23]. In contemporary financial practice, this shift reflects the recognition that externalities previously treated as non-financial—can evolve into economically material risks, such as sanctions, operational disruptions, reputational losses, or litigation exposure, thereby affecting valuation, cost of capital, and cash-flow stability. Empirical evidence supports this linkage: non-financial disclosure and socially responsible performance have been associated with lower capital costs and superior financial outcomes over medium- and long-term horizons [24,25,26], particularly when initiatives focus on material issues aligned with industry and strategy rather than peripheral or cosmetic actions [19].

Within this framework, ESG reporting has become a central mechanism for transparency and comparability, driven by international standards and frameworks designed to enhance the traceability of sustainability-related risks and opportunities. The GRI Standards, especially the revised Universal Standards, have strengthened an impact-oriented accountability approach, with the latest update published in October 2021 becoming effective for reports issued from January 2023 onward [27].

In parallel, the Sustainability Accounting Standards Board (SASB) Standards have consolidated an investor-oriented disclosure logic focused on sustainability risks and opportunities with potential financial impacts. Following their institutional integration into the IFRS Foundation, SASB Standards now serve as sectoral guidance for the application of IFRS S1 and the identification of financially material sustainability issues [28]. Complementarily, the Integrated Reporting Framework promotes connectivity between financial and non-financial information, seeking to coherently explain how value is created or eroded in the short, medium, and long term [29]. While the convergence of these frameworks does not automatically resolve issues of data quality, consistency, or verifiability, it does raise the minimum disclosure threshold and reinforces the principle that sustainability should be anchored in strategy, governance, and metrics.

Nevertheless, in Latin America, ESG disclosure adoption remains uneven in both coverage and depth, with visible gaps in assurance, comparability, and intersectoral consistency. Regional surveys document progress in climate target disclosure and SDG alignment, but also reveal lagging assurance rates and strong heterogeneity across countries and industries, complicating homogeneous financial interpretation of sustainability-related risks [20]. At the same time, the expansion of sustainability-linked financial instruments—such as ESG bonds and sustainability-linked bonds—has increased pressure for more robust reporting and verifiable governance practices to sustain market credibility [30]. Thus, the challenge is not merely to report more, but to report better: through traceable, consistent, and comparable indicators supported by explicit materiality and weighting criteria that translate sustainability into decision-useful financial signals.

2.3. Sustainable Finance: From Preference to Financial Materiality

The contemporary debate on sustainable finance rests on a core premise: corporate sustainability can be financially material. Empirical studies show that voluntary non-financial disclosure can reduce the cost of equity when associated with superior social performance [24], and that markets incorporate CSR/ESG information into firm valuation contingent on stakeholder awareness and visibility [3]. Evidence from crisis periods further suggests that corporate social capital may act as a buffer during times of stress, with observable effects on returns and operating performance [18].

At the research frontier, equilibrium asset pricing models distinguish between sustainability as risk and sustainability as preference, clarifying why identical ESG scores may be associated with different expected return patterns depending on investor constraints, non-pecuniary demand, and arbitrage capacity [6,7]. In parallel, the literature on climate risk and carbon premia documents that emissions and climate-related exposures may be priced in asset markets, supporting the environmental dimension of financial materiality [8].

Meta-analytical evidence widely cited in the literature finds that the ESG–financial performance relationship is, on average, positive or non-negative, though sensitive to study design, region, industry, and measurement choices [26]. While this does not eliminate controversy, it reinforces the central argument of this study: the key issue is no longer whether ESG matters, but how to measure it in a comparable and decision-relevant manner, particularly in contexts characterized by heterogeneous disclosure.

2.4. The 5P Framework as an Integrative and Operational Approach

The 2030 Agenda proposes a conceptual architecture that synthesizes sustainable development into five interrelated dimensions: People, Planet, Prosperity, Peace, and Partnerships [1,2,3,4,5,6,7,8,9,10,11,12]. This structure builds upon earlier sustainability paradigms, particularly the Triple Bottom Line framework, which expanded corporate performance assessment beyond purely financial outcomes to incorporate social and environmental dimensions [2]. In this sense, the 5P model does not replace ESG logic but deepens it by embedding sustainability within a systemic and development-oriented perspective.

From a governance and reporting standpoint, the operationalization of sustainability dimensions into measurable indicators is consistent with internationally recognized frameworks such as the GRI Standards [4], the Integrated Reporting Framework [5], and the SASB Standards under the IFRS Foundation architecture [6]. These initiatives reflect the progressive institutionalization of ESG disclosure and the growing demand for comparability, materiality, and decision-useful information in capital markets. Within this evolving landscape, structuring sustainability into five pillars provides a coherent organizing logic aligned with global reporting convergence efforts.

This structure offers two advantages for financial analysis. First, it avoids a reductionist interpretation of ESG narrowly focused on environmental performance or disclosure intensity [6]. Second, it enables operationalization by pillars that can be translated into observable and comparable indicators, facilitating composite index construction. Accordingly, this study assumes that the 5Ps provide an integrative framework suitable for structuring sustainability objectives into operational dimensions applicable to the MSCI COLCAP universe.

Importantly, the Peace pillar explicitly incorporates the institutional dimension justice, rule of law, and institutional quality, which is particularly salient in emerging markets, where regulatory stability and governance structures materially affect risk profiles [3]. Likewise, the Partnerships pillar captures intersectoral collaboration, value-chain integration, and stakeholder engagement maturity. These factors are directly linked to strategic resilience, operational continuity, and social license to operate. This broader framing is consistent with calls to contextualize ESG metrics in accordance with regional institutional realities rather than relying exclusively on imported measurement models [3].

From the standpoint of firm-level assessment, the analytical added value of the 5P framework lies not merely in relabeling conventional ESG dimensions, but in explicitly incorporating institutional quality and relational embeddedness through the Peace and Partnerships pillars. These dimensions are especially relevant in emerging markets, where regulatory credibility, stakeholder coordination, and cross-sector collaboration can materially affect firm risk, resilience, and strategic continuity. In this sense, the 5P architecture provides a broader and more context-sensitive framework for sustainability assessment than conventional ESG aggregation alone.

2.5. Disclosure Standards: From Voluntary Reporting to Convergence

ESG measurement relies on an ecosystem of disclosure standards. The Task Force on Climate-related Financial Disclosures (TCFD) established a climate disclosure framework focused on decision-useful information for capital markets, structured around governance, strategy, risk management, and metrics and targets [31]. More recently, the International Sustainability Standards Board (ISSB) issued its first two global sustainability disclosure standards (IFRS S1 and IFRS S2), creating a global baseline for sustainability and climate-related reporting aimed at investors and integrating existing frameworks such as TCFD and SASB [32,33].

In parallel, GRI has remained dominant in impact-oriented disclosure (impact materiality), while SASB now under the IFRS Foundation prioritizes financial materiality by industry, facilitating comparability of investor-relevant risks and opportunities [27]. These architectures justify the methodological phase of this study that selects variables aligned with international standards (GRI, SASB, TCFD, UN Global Compact, and MSCI ESG Ratings), while also highlighting a critical point: convergence of standards does not automatically resolve measurement problems, but instead heightens the importance of traceability and internal consistency when constructing composite indices.

2.6. ESG Ratings: Divergence, Uncertainty, and Decision Effects

Although ESG ratings simplify complexity to facilitate decision-making, the literature documents substantial divergence across providers. Evidence shows that this divergence is driven primarily by differences in measurement rather than weighting schemes, implying that improvements in ESG performance may send different signals depending on the provider and taxonomy used [21]. Complementary research emphasizes that convergence among ratings faces structural limits due to underlying theoretical assumptions and commensurability issues [10].

Moreover, evidence suggests that ESG disclosure mandates may improve transparency without necessarily reducing rating disagreement and may even increase it, with potential effects on volatility and access to financing [11].

2.7. Composite Indices: Methodological Risks and the Choice of PCA

The design of composite indices requires methodological rigor, as seemingly minor choices—such as normalization methods, outlier treatment, or weighting schemes—can materially affect rankings and conclusions. The OECD/JRC Handbook provides a widely cited reference, recommending a systematic construction process involving data cleansing, correlation analysis, structural assessment, weighting selection, sensitivity analysis, and transparency [13]. Greco et al. [14] further emphasize conceptual coherence and statistical robustness as core principles.

In this study, the methodological process is organized into four phases: data cleansing, normalization, pillar-level scoring, and composite index construction. This sequence aligns with international best practices. The use of principal component analysis (PCA) to derive endogenous weights is methodologically justified, as it captures shared variance and reduces arbitrariness in weighting, particularly relevant in ESG contexts where collinearity and double counting are common. However, PCA does not identify causality but statistical structure; therefore, its use must be supported by content validity and evidence of measurement adequacy. Accordingly, this study employs internal consistency measures and KMO/Bartlett tests to support the coherence of the measurement framework.

2.8. Disclosure, Liquidity, and Regulation in Emerging Markets

Recent literature documents that ESG disclosure mandates can affect market liquidity and quality, with heterogeneous effects depending on enforcement and institutional design [22]. This issue is particularly relevant for Colombia, where institutional quality and regulatory structure interact with firms’ effective reporting and verification capacity. This reinforces the relevance of explicitly incorporating the Peace dimension institutions, justice, and compliance rather than treating it as a residual governance subcomponent, thereby acknowledging its direct role in shaping risk, credibility, and sustainable corporate governance in emerging markets.

3. Materials and Methods

3.1. Research Approach and Design

The objective of this study is to develop and validate the 5P–ESG Composite Index as a finance-oriented instrument for assessing firm-level sustainable-financial performance in the MSCI COLCAP universe, and to examine whether its broader 5P architecture produces classification outcomes that differ from those obtained through a conventional ESG-3 framework. In methodological terms, this objective requires not only the conceptual design of a multidimensional assessment model, but also the statistical verification of its internal coherence, structural consistency, and operational applicability in an emerging-market setting.

Accordingly, the study is framed within a quantitative applied approach with an exploratory–descriptive scope and a non-experimental, cross-sectional design. It is quantitative because it relies on measurable financial and sustainability-related information that can be standardized and statistically processed; applied because it is aimed at constructing and operationalizing an issuer-level assessment tool; and exploratory–descriptive because, given the finite cross-sectional sample (N = 17), the empirical analysis focuses on index construction, internal consistency, structural validation, robustness checks, and comparative classification, rather than on testing causal relationships between sustainability and financial performance. Any observed association between the composite score and financial dimensions is therefore interpreted only as preliminary descriptive evidence.

This design is consistent with the purpose of the study, since the manuscript does not seek to estimate the causal effect of ESG performance on financial outcomes, but rather to construct, validate, and apply a multidimensional composite index capable of supporting comparative analysis. In this sense, the methodological strategy was organized around four sequential components: (i) the selection and refinement of indicators aligned with the 5P framework; (ii) the standardization and harmonization of heterogeneous firm-level data to ensure comparability; (iii) the statistical validation of the measurement structure through reliability and factorability diagnostics; and (iv) the construction of a composite issuer-level score suitable for ranking and classification, including its comparison with a conventional ESG-3 benchmark.

Likewise, the study adopts a non-experimental and cross-sectional design because the variables are not manipulated but observed as reported by firms in a specific period. The cross-sectional character of the analysis reflects that the empirical application is limited to the set of companies included in the MSCI COLCAP index during the period under study. This decision is methodologically coherent with the objective of assessing the operational feasibility and comparative usefulness of the proposed index within a defined issuer universe representative of the Colombian equity market.

The overall methodological process was therefore structured as a transparent sequence linking the research objective to the empirical application: first, the conceptual and operational construction of the 5P–ESG framework; second, the refinement and statistical validation of the instrument; and third, the calculation of the final issuer-level composite score for classification and benchmark comparison. This sequencing ensures that the methodological design directly supports the central purpose of the study and provides a coherent basis for the interpretation of the results.

3.2. Population and Sample

The population under study consists of the companies included in the MSCI COLCAP index, the main stock market index of the Colombian equity market.

The sample coincides with the population, as the study uses all companies that comprise the index during the period of analysis. This approach avoids selection bias and ensures that the results obtained are representative of the aggregated behavior of the Colombian equity market.

The MSCI COLCAP index represents a benchmark of the most relevant and liquid equities listed on the Colombian Stock Exchange (Bolsa de Valores de Colombia). As a capitalization- and liquidity-screened index, it provides a practical proxy for the investable Colombian equity market and concentrates issuers with greater visibility, disclosure, and trading activity relative to the broader universe. Using MSCI COLCAP as the study population is therefore appropriate to evaluate issuer-level sustainable-financial performance in an emerging-market setting, where reporting practices and ESG-related disclosure may be heterogeneous and strongly shaped by sectoral composition and institutional context.

Each company constitutes a unit of analysis, for which traditional financial indicators and sustainability metrics integrated under the 5P framework are constructed.

3.3. Data Collection and Analysis Techniques

In line with the objective of developing and validating the 5P–ESG Index, data collection and analysis were structured to ensure both conceptual coverage of the five pillars and statistical adequacy for composite-index construction.

The information base comprises 17 companies belonging to the MSCI COLCAP, for which financial and sustainability indicators were collected from public sources: integrated reports, sustainability reports, audited financial statements, and regulatory documents.

Data preprocessing, indicator harmonization, winsorization, normalization, and statistical analyses were performed using Stata (StataCorp LLC, College Station, TX, USA).

The design process of the 5P–ESG Composite Index began with the development of a preliminary instrument composed of 28 variables distributed across the five pillars of the model (People, Planet, Prosperity, Peace, and Partnerships). The initial selection was based on a review of international ESG standards (GRI, SASB, TCFD, UN Global Compact, MSCI ESG Ratings), a review of the specialized literature on sustainability metrics and corporate performance in emerging markets, and finally an analysis of conceptual completeness criteria, incorporating multiple potential indicators to ensure broad coverage of each dimension.

This phase ensured content validity, but it did not imply that all variables were statistically consistent or empirically relevant. Therefore, a rigorous process of validation and item reduction was conducted, as recommended by in the literature on measurement validation and ESG rating divergence [9,26].

3.4. Initial Construction of the Instrument

To achieve the study objective, the methodological strategy was organized around four operational components: (i) the selection and refinement of indicators aligned with the 5P framework; (ii) the standardization and harmonization of heterogeneous firm-level data; (iii) the statistical validation of the measurement structure; and (iv) the construction of a composite issuer-level score suitable for comparative classification.

As part of the construction process of the 5P–ESG Composite Index, the study initially considered a total of 28 conceptually grouped variables aimed at broadly capturing the financial and sustainability dimensions associated with the five Ps of sustainable development: People, Planet, Prosperity, Peace, and Partnerships.

The preliminary selection of variables was based on three fundamental criteria:

• Theoretical support in the literature on sustainable finance, ESG, and corporate performance;
• International standards such as GRI, SASB, and MSCI ESG Ratings;
• Availability and comparability of information for companies listed on the Colombian Stock Exchange.

This process resulted in a broader initial instrument, which was subsequently subjected to statistical cleaning and validation procedures in order to construct a final version that is more parsimonious, robust, and operationally viable (see

Table 1. Coding of the pillars and indicators of the 5P Index.

Pillar	Code	Variable	Description
People	PPL01	Personnel expenses/Revenue	Labor cost as a proportion of revenue
People	PPL02	Diversity and inclusion policy	Existence/disclosure of diversity policies
People	PPL03	Employee turnover (%)	Employee departures as a share of total workforce
People	PPL04	Training hours per employee	Investment in employee training
People	PPL05	Occupational accident rate	Accidents per 200,000 h worked
People	PPL06	Gender pay gap	Percentage gender pay gap (male–female)
Planet	PLN01	Environmental investment as a proportion of revenue	Environmental expenditure relative to revenue
Planet	PLN02	Environmental certifications	Number/type of environmental certifications
Planet	PLN03	Emissions intensity	tCO2e per unit of revenue
Planet	PLN04	Energy consumption	Energy intensity (energy per unit of output)
Planet	PLN05	Water consumption	Water consumption per unit of revenue
Planet	PLN06	Recycled waste (%)	Share of recycled waste
Prosperity	PRO01	Return on Assets (ROA)	Return on Assets
Prosperity	PRO02	Return on Equity (ROE)	Return on Equity
Prosperity	PRO03	Net Profit Margin	Net Income to Revenues
Prosperity	PRO04	Asset Turnover (Revenues/Total Assets)	Efficiency Ratio
Prosperity	PRO05	Leverage Ratio (%)	Total Liabilities to Total Assets
Prosperity	PRO06	Current Ratio	Current Assets to Current Liabilities
Peace	PEA01	Female Board Representation	Percentage of Women on the Board
Peace	PEA02	Regulatory Sanctions to Revenues	Fines and Sanctions to Revenues
Peace	PEA03	Country Code Compliance	Level of Adoption of the Country Code
Peace	PEA04	Board Independence (%)	Percentage of Independent Directors
Peace	PEA05	Audit and Risk Committee	Existence of the Committee (1/0)
Partnerships	PRT01	Green/Social Bonds to Total Debt	Proportion of Sustainable Bonds
Partnerships	PRT02	UN Global Compact Adherence	UN Global Compact Membership
Partnerships	PRT03	GRI/TCFD/SASB Disclosure	Level of International Standard Reporting
Partnerships	PRT04	Participation in Sustainable Associations	Membership in Sustainable Industry Associations
Partnerships	PRT05	Public–Private Partnerships	Number of Sustainable Partnerships

Source: elaboration.

).

Table 1 presents the preliminary instrument comprising 28 indicators, designed to broadly cover the five dimensions of People, Planet, Prosperity, Peace, and Partnerships.

However, to ensure that the composite index was operationally viable, comparable across issuers, and statistically stable, a data reduction process based on four sequential filters (F1–F4) was applied, combining expert judgment (content validity) and statistical criteria (empirical validity). The objective was to obtain a parsimonious set of variables with high financial interpretability and internal consistency, enabling the construction of pillar-level scores and the overall 5P–ESG Composite Index.

• F1. Content and applicability filter (expert judgment).

A technical review of the preliminary list was conducted to assess: (i) alignment with the 5P construct, (ii) conceptual clarity and directionality, and (iii) financial relevance for corporate analysis. Priority was given to indicators with direct interpretability for investors (e.g., verifiable policies, comparable ratios, efficiency metrics, and distribution measures). Variables whose measurement was highly heterogeneous across sectors or not consistently reportable across issuers were identified and excluded at this stage.

• F2. Availability and comparability filter within the MSCI COLCAP universe.

Data availability for each indicator was audited for the period under analysis. Indicators with systematic data gaps or inconsistent definitions across firms were excluded. This filter is critical in composite index construction, as it avoids biases arising from excessive imputation and ensures that rankings reflect actual differences in corporate performance rather than information gaps.

• F3. Preliminary statistical quality filter (variability and redundancy).

Indicators exhibiting (i) null or near-zero variance (i.e., lacking discriminatory power across issuers), and (ii) redundancy (very high correlations implying double counting of the same underlying phenomenon) were removed. These controls enhance index stability and reduce the risk that final scores are disproportionately driven by a single subset of variables.

• F4. Statistical validation of the instrument (reliability and structure).

For the refined set of indicators, internal consistency tests (Cronbach’s alpha at the pillar and global levels) and factorability tests (KMO and Bartlett’s test of sphericity) were applied as prerequisites for exploratory factor analysis (EFA).

Adherence to an explicit index-construction workflow. In line with the composite-indicators literature, the overall methodological process is organized as an explicit and transparent sequence: (i) variable selection and cleansing (implemented through the sequential filters F1–F4), (ii) normalization to ensure comparability across heterogeneous indicators (outlier treatment, standardization, directional harmonization, and bounded rescaling), (iii) assessment of latent structure and internal consistency (reliability and factorability diagnostics reported as preliminary evidence given the finite sample), and (iv) defensible aggregation rules. Specifically, indicators are first aggregated into pillar-level scores and then combined into the global 5P–ESG Composite Index through a weighted-sum rule, where pillar weights are derived endogenously from PCA (PC1) and complemented with sensitivity analysis against equal weights.

Subsequently, EFA confirmed that the retained indicators exhibited an interpretable factor structure and adequate factor loadings, supporting the empirical coherence of the measurement instrument.

In summary, the filters—defined ex ante and designed to be replicable—followed the logic outlined below:

F1 = Content validity (expert judgment: 5P relevance and financial interpretability).

F2 = Availability and comparability (consistent public data for COLCAP issuers; low missingness).

F3 = Preliminary statistical quality (sufficient variability and non-redundancy; avoidance of non-discriminatory items).

F4 = Empirical validation (contribution to reliability and structure: internal consistency and interpretable factor patterns).

Overall, the instrument refinement process was structured as a sequential and replicable filtering procedure (F1–F4), aimed at ensuring that the resulting index is comparable across issuers, operationally measurable, and statistically robust. This approach avoids the inclusion of indicators with limited availability or heterogeneous measurement and mitigates double counting arising from redundancy. Accordingly, priority was first given to content validity and financial interpretability (F1), followed by empirical feasibility within the MSCI COLCAP universe (F2), then discriminatory capacity and preliminary statistical quality (F3), and finally empirical coherence through reliability and structural tests (F4). The incremental logic of the procedure is summarized in

Table 2. Summary of the instrument refinement process (28–16 indicators).

Filter	Applied Criterion	Operational Rule (Replicable)	Excluded Indicators (Examples)	n Retained
F1. Content (expert judgment)	5P relevance + financial interpretability	Indicators with expert consensus on relevance and clear directionality are retained	Highly sector-specific or difficult-to-standardize indicators	28 → 22
F2. Data	Availability/comparability	Indicators are excluded if data are unavailable or not comparable across issuers	Variables with incomplete reporting or inconsistent definitions	22 → 19
F3. Statistical quality	Variability and redundancy	Indicators are excluded if variance ≈ 0 or if they duplicate information (high overlap)	Constant or highly collinear items	19 → 17
F4. Statistical validation	Reliability and structure	Items are retained if they contribute to internal consistency and exhibit acceptable factor loadings	Items with low contribution to the structure or weak loadings	17 → 16

Source: Own elaboration.

.

Based on the traceability presented in Table 2, the filtering process made it possible to move from the preliminary set of 28 indicators to a parsimonious core of variables with high inter-firm comparability, direct financial interpretability, and conceptual coherence with the 5P framework. Accordingly,

Table 3. Refinement of the core variables.

Pillar	Code	Indicator	Explanation
People	PPL01	Personnel expenses/Revenues	Relationship between total personnel expenses and revenues; indicates the level of investment in human capital.
People	PPL02	Diversity and inclusion policy/disclosure	Assesses whether the company has a formal and publicly disclosed diversity and inclusion policy.
Planet	PLN01	Environmental investment/Revenues	Measures the proportion of revenues allocated to environmental investments or expenditures.
Planet	PLN02	Environmental certification	Indicates whether the company holds verifiable environmental certifications, such as ISO 14001 [34].
Prosperity	PRO01	Real revenue growth	Change in revenues net of inflation; reflects the firm’s real business growth.
Prosperity	PRO02	Distribution to labor and government/Revenues	Ratio of wages and taxes to revenues; reflects the firm’s social and fiscal contribution.
Prosperity	PRO03	Responsible supplier payment index (DPO)	Average number of days the company takes to pay its suppliers.
Prosperity	PRO04	Equity-adjusted net margin	Net margin adjusted by an equity factor based on the Gini index.
Prosperity	PRO05	Labor productivity	Revenues generated per employee; measures operational efficiency.
Prosperity	PRO06	Dividend stability	Ratio of dividends paid to net income; measures the consistency of the dividend policy.
Peace	PEA01	Women on the board (%)	Percentage of women on the board; indicator of diversity in corporate governance.
Peace	PEA02	SFC/SIC sanctions relative to revenues (normalized)	Indicates the magnitude of regulatory sanctions relative to revenues.
Peace	PEA03	Country Code/Good Corporate Governance Code	Assesses whether the company discloses information aligned with good corporate governance standards.
Partnerships	PRT01	Green/social bonds/Total debt	Proportion of debt associated with green or social instruments.
Partnerships	PRT02	Adherence to the UN Global Compact	Indicates whether the company is a member of the UN Global Compact or other UN initiatives.
Partnerships	PRT03	GRI/SASB/TCFD disclosure	Assesses whether the company reports under international sustainability frameworks.

Source: Own elaboration.

presents the decisive filter and the complete structure and specifications for each indicator, following a logic analogous to PRISMA reporting applied to index construction. Likewise, the core instrument (N = 16) used in the subsequent stages of the study is consolidated: direction homogenization, winsorization, normalization (Z-scores), calculation of pillar scores, and construction of the composite index (5P–ESG Composite Index), as well as internal consistency and structural tests (Cronbach’s alpha, KMO/Bartlett, and EFA) (see Table 3).

To ensure methodological transparency, the manuscript explicitly details the replacement of the preliminary Prosperity pillar indicators (Table 1) with the refined indicators included in the final instrument (Table 3):

• ROA (PRO01) was replaced by Real revenue growth (PRO01), prioritizing real growth over accounting profitability as a proxy for Prosperity.
• ROE (PRO02) was replaced by Distribution to labor and government/Revenues (PRO02), in order to capture distributive and fiscal contributions (wages and taxes) within the concept of Prosperity.
• Net Profit Margin (PRO03) was replaced by Responsible supplier payment index (DPO) (PRO03), incorporating economic responsibility toward suppliers, a dimension not reflected in conventional accounting margins.
• Asset Turnover (PRO04) was replaced by Equity-adjusted net margin (PRO04), integrating profitability with an equity factor based on the Gini index.
• Leverage ratio (PRO05) was replaced by Labor productivity (PRO05), aligning Prosperity with operational productivity and efficient value-generation capacity.
• Current ratio (PRO06) was replaced by Dividend stability (PRO06), serving as a proxy for consistency in value generation and return to investors.

Specification of constructed indicator (PRO04). The equity-adjusted net margin is defined as follow:

$$PRO04=\left({\displaystyle \frac{\mathrm{Net} \mathrm{Income}}{\mathrm{Revenue}}}\right)\times (1-Gini)$$

where Gini ∈ [0, 1] (if reported on a 0–100 scale, it is divided by 100). This formulation simultaneously incorporates a profitability dimension (net margin) and a distributive dimension (equity), aligning with the broader conceptualization of Prosperity within the 5P framework.

• Preliminary statistical treatment of the data

The construction of the synthetic scores associated with the five dimensions of the 5P model (People, Planet, Prosperity, Peace, and Partnerships) followed a standardized statistical procedure designed to ensure comparability across heterogeneous indicators, reduce distortions caused by extreme values, and provide the statistical consistency required for a robust composite index.

The index design adopted a four-step standardization procedure, in line with OECD recommendations [1] and the specialized literature on ESG ratings [2,3]:

(i)

Data cleaning and winsorization;

(ii)

Standardized normalization;

(iii)

Aggregation of items by dimension;

(iv)

Estimation of the final score for each pillar.

This procedure ensures the statistical stability of the estimates and prevents extreme values or scale differences from affecting comparability across firms included in the MSCI COLCAP index.

First, a bilateral winsorization procedure (scientific control of outliers) was applied, setting the 5th and 95th percentiles as lower and upper bounds, respectively.

Given the finite sample size (N = 17) and the heterogeneous units and scales of the indicators, winsorization at the P5/P95 levels is employed as a conservative robustness mechanism to limit the disproportionate influence of extreme values without excluding issuers from the analysis. In small samples, a single outlier can distort standardization and aggregation procedures, potentially affecting the final ranking. As a sensitivity check, we verified that issuer rankings do not change materially under alternative outlier treatments (e.g., P10/P90 or no winsorization).

This method allows for:

(i)

Mitigation of the influence of extreme values without eliminating observations, thus preserving the full sample structure;

(ii)

Reduction of the impact of outliers associated with atypical variations or measurement errors;

(iii)

Retention of all cases in the analysis, unlike trimming procedures that remove observations;

(iv)

Improvement in the stability of means and dispersions in sensitive indicators such as ratios, rates, operational days, and risk-management-related metrics.

Given that the data matrix includes quantitative and proportional performance indicators across the five dimensions (People, Planet, Prosperity, Peace, and Partnerships), winsorization was applied exclusively to continuous variables:

PPL01–PPL02 (People);

PLA1–PLA2 (Planet);

PRO1–PRO6 (Prosperity);

PEA1–PEA3 (Peace);

PRT01–PRT03 (Partnerships).

Binary/categorical columns (values 0–1) do not require winsorization.

Now, for each of the numerical indicators:

• Each variable was sorted in ascending order.
• The 5th (P5) and 95th (P95) percentiles were identified.
• Values below P5 were replaced with the exact value of the 5th percentile.
• Values above P95 were replaced with the corresponding 95th percentile.

Accordingly, for each indicator $X$, its winsorized version $X_w$ was defined as:

$$X_w=\left\{\begin{array}{ccc}{P}_5& si X<& P_5\\ X& si P_{5 }\le & X\le P_{95}\\ P_{95}& si X>& P_{95}\end{array}\right.$$

where $P_5$ and $P_{95}$ correspond to the 5th and 95th percentiles of each quantitative variable.

The transformation preserves the relative structure of the dataset while reducing the influence of extreme outliers.

Winsorization resulted in:

• A reduction in dispersion for highly volatile indicators such as PRO1, PRO4, and PLA2.
• Greater stability in means, eliminating variations induced by extreme values.
• Improved symmetry in a considerable portion of the Prosperity metrics, where several companies exhibited unusually high values in periods of operating days (PRO3).
• Removal of adverse bias in indicators with negative proportions or values exceeding 100%, which distorted benchmarking.

The process described above can be seen in

Table 4. Winsorized raw inputs.

COMPANIES/IND	People		Planet		Prosperity						Peace			Partnerships
	PPL01	PPL02	PLA1	PLA2	PRO1	PRO2	PRO3	PRO4	PRO5	PRO6	PEA1	PEA2	PEA3	PRT01	PRT02	PRT03
PFCIBEST	13.58%	1	1.57%	1	−6.36%	47.17%	13 days	1.79%	1239.43	63.11%	40.00%	0.9983	1	5.62%	1	1
ISA	8.78%	1	0.78%	1	0.18%	18.06%	11 days s	1.97%	19,098.15	92.12%	22.22%	0.9992	1	5.40%	1	1
ECOPETROL	8.06%	1	10.77%	1	−11.42%	8.88%	81 days	0.97%	1193.66	97.83%	33.33%	0.9999	1	49.16%	1	1
GEB	14.21%	1	1.82%	1	−4.73%	22.33%	75 days	4.90%	3449.40	44.69%	20.00%	0.9987	1	35.06%	1	1
GRUPOARGOS	9.67%	1	0.29%	1	−3.69%	10.43%	100 days	12.00%	377.87	79.13%	57.14%	0.9793	1	4.52%	1	1
PFGRUPSURA	12.57%	1	0.25%	1	3.08%	24.48%	36 days	1.33%	600.03	90.18%	0.00%	1.0000	1	0.56%	1	1
CELSIA	4.87%	1	0.00%	1	10.71%	6.14%	211 days	0.11%	3055.14	99.72%	14.29%	1.0000	1	38.97%	1	1
PFDAVVNDA	28.72%	1	4.26%	1	−14.48%	28.90%	89 days	0.00%	1545.38	100.00%	28.57%	1.0000	1	8.18%	1	1
GRUBOLIVAR	13.78%	1	3.03%	1	−11.99%	34.80%	137 days	0.06%	864.02	92.03%	28.57%	0.9991	1	6.26%	1	1
PEI	13.32%	1	9.74%	1	−10.91%	27.19%	0 days	23.14%	6005.41	100.00%	50.00%	1.0000	1	36.55%	1	1
BOGOTA	1.72%	1	28.69%	1	8.97%	24.52%	146 days	0.19%	1809.79	89.56%	33.33%	1.0000	1	6.30%	1	1
PROMIGAS	5.39%	1	0.37%	1	2.62%	13.41%	51 days	0.75%	3789.48	37.13%	28.57%	0.9925	1	40.72%	1	1
TERPEL	1.42%	1	21.81%	1	−3.46%	7.08%	321 days	0.01%	25,287.41	73.13%	14.29%	0.6539	1	0.03%	1	1
PFCORFICOL	4.46%	1	17.19%	1	−7.05%	18.28%	21 days	0.02%	444.53	98.96%	44.44%	0.9859	1	26.84%	1	1
MINEROS	4.22%	1	4.59%	1	8.50%	0.17%	141 days	0.00%	774,655.57	90.84%	25.00%	1.0000	1	4.82%	1	1
CNEC	9.04%	1	1.38%	1	13.15%	28.36%	3 days	0.00%	3319.35	92.03%	14.29%	1.0000	1	80.74%	1	1
ETB	17.88%	1	0.28%	1	5.05%	21.63%	30 days	0.15%	3763.66	100.00%	42.86%	0.9943	1	38.70%	1	1

Source: Own elaboration.

(see Table 4).

Normalization and comparability pipeline. After winsorization (P5–P95) for continuous indicators, we applied a two-stage normalization to ensure cross-indicator comparability and to integrate continuous and binary variables in the same scoring framework.

Step 1 (standardization): for each continuous indicator $x_{ij}^{\left(w\right)}$, we compute a z-score:

$$Z_{ij}={\displaystyle \frac{x_{ij}^{\left(w\right)}-{\mu }_j}{{\sigma }_j}}$$

Step 2 (directional harmonization): for “lower-is-better” indicators, we reverse the orientation by $z_{ij}^{*}=-z_{ij}$, so that higher values always indicate better performance.

Step 3 (bounded rescaling for aggregation/reporting): to report normalized values on a bounded scale and to align with binary variables, we rescale standardized scores to [0, 1] using min–max:

$$s_{ij}={\displaystyle \frac{z_{ij}^{*}-min\left(z_j^{*}\right)}{max\left(z_j^{*}\right)-min\left(z_j^{*}\right)}}$$

Binary indicators remain unchanged (0/1).

Table 5. Bounded normalized indicators (0–1) after standardization and rescaling.

COMPANIES	PPL01	PPL02	PLA1	PLA2	PRO1	PRO2	PRO3	PRO4	PRO5	PRO6	PEA1	PEA2	PEA3	PRT01	PRT 2	PRT 3
PFCIBEST	0.45	1.00	0.05	1.00	0.29	1.00	0.96	0.08	0.00	0.41	0.70	0.00	1.00	0.07	1.00	1.00
ISA	0.27	1.00	0.03	1.00	0.53	0.38	0.96	0.09	0.02	0.87	0.39	0.00	1.00	0.07	1.00	1.00
ECOPETROL	0.24	1.00	0.38	1.00	0.11	0.19	0.75	0.04	0.00	0.97	0.58	0.00	1.00	0.61	1.00	1.00
GEB	0.47	1.00	0.06	1.00	0.35	0.47	0.77	0.21	0.00	0.12	0.35	0.00	1.00	0.43	1.00	1.00
GRUPOARGOS	0.30	1.00	0.01	1.00	0.39	0.22	0.69	0.52	0.00	0.67	1.00	0.06	1.00	0.06	1.00	1.00
PFGRUPSURA	0.41	1.00	0.01	1.00	0.64	0.52	0.89	0.06	0.00	0.84	0.00	0.00	1.00	0.01	1.00	1.00
CELSIA	0.13	1.00	0.00	1.00	0.91	0.13	0.34	0.00	0.00	1.00	0.25	0.00	1.00	0.48	1.00	1.00
PFDAVVNDA	1.00	1.00	0.15	1.00	0.00	0.61	0.72	0.00	0.00	1.00	0.50	0.00	1.00	0.10	1.00	1.00
GRUBOLIVAR	0.45	1.00	0.11	1.00	0.09	0.74	0.57	0.00	0.00	0.87	0.50	0.00	1.00	0.08	1.00	1.00
PEI	0.44	1.00	0.34	1.00	0.13	0.57	1.00	1.00	0.01	1.00	0.88	0.00	1.00	0.45	1.00	1.00
BOGOTA	0.01	1.00	1.00	1.00	0.85	0.52	0.54	0.01	0.00	0.83	0.58	0.00	1.00	0.08	1.00	1.00
PROMIGAS	0.15	1.00	0.01	1.00	0.62	0.28	0.84	0.03	0.00	0.00	0.50	0.02	1.00	0.50	1.00	1.00
TERPEL	0.00	1.00	0.76	1.00	0.40	0.15	0.00	0.00	0.03	0.57	0.25	1.00	1.00	0.00	1.00	1.00
PFCORFICOL	0.11	1.00	0.60	1.00	0.27	0.39	0.94	0.00	0.00	0.98	0.78	0.04	1.00	0.33	1.00	1.00
MINEROS	0.10	1.00	0.16	1.00	0.83	0.00	0.56	0.00	1.00	0.85	0.44	0.00	1.00	0.06	1.00	1.00
CNEC	0.28	1.00	0.05	1.00	1.00	0.60	0.99	0.00	0.00	0.87	0.25	0.00	1.00	1.00	1.00	1.00
ETB	0.60	1.00	0.01	1.00	0.71	0.46	0.91	0.01	0.00	1.00	0.75	0.02	1.00	0.48	1.00	1.00

Source: Own elaboration. Note: The values reported in Table 5 correspond to standardized indicators (Z-scores), harmonized in direction (−z for variables where “lower is better”) and subsequently rescaled using a min–max transformation to the [0, 1] (or [0, 10]) range. Therefore, they do not represent raw Z-scores.

reports $s_{ij}\in \left[0,1\right]$ (bounded normalized values), while Table 4 reports winsorized raw inputs.

Furthermore, it should be noted that Table 5 confirms that certain dichotomous indicators (0/1) exhibit zero variance within the sample (i.e., they take the value 1.00 for all issuers), specifically PPL02, PRT02, and PRT03. Accordingly, these indicators were excluded from correlation-based procedures (correlation matrix, KMO test, Bartlett’s test, exploratory factor analysis, and PCA), as they do not contribute statistical information and may distort the estimation of the latent structure. However, they are retained as descriptive variables within the instrument for purposes of transparency and conceptual traceability.

3.5. Construction of Pillar Scores

Once the normalized scores were obtained using the Z-procedure described above, the fourth step consisted of integrating these values within each dimension of the 5P model. Each pillar was constructed by aggregating the normalized items that conceptually belong to the same dimension, ensuring that all indicators contributed on a comparable scale.

The score of pillar $j$ for each company was calculated using the arithmetic mean of the normalized values corresponding to the $n_j$ indicators associated with that pillar:

$$P_j={\displaystyle \frac{1}{n_j}}\sum _{i=1}^{n_j}{Z}_{ij}$$

where

• P_j represents the aggregated score of dimension j;
• n_j is the number of indicators assigned to the pillar;
• Z_ij is the normalized value of indicator i, obtained in the normalization phase.

This aggregation procedure is appropriate because, with all indicators previously standardized on the same metric (mean = 0 and standard deviation = 1), the simple average allows combining the variables without introducing biases due to scale or magnitude differences. In this way, the score obtained reflects the relative performance of each company in each of the pillars of the 5P model, as shown in

Table 6. Pillar scores for the 5P index.

Company	People	Planet	Prosperity	Peace	Partnerships	5P–ESG Composite Index
PFCIBEST	0.72	0.53	0.46	0.57	0.69	0.59
ISA	0.63	0.51	0.48	0.46	0.69	0.56
ECOPETROL	0.62	0.69	0.34	0.53	0.87	0.61
GEB	0.73	0.53	0.32	0.45	0.81	0.57
GRUPOARGOS	0.65	0.50	0.41	0.69	0.69	0.59
PFGRUPSURA	0.70	0.50	0.49	0.33	0.67	0.54
CELSIA	0.56	0.50	0.40	0.42	0.83	0.54
PFDAVVNDA	1.00	0.57	0.39	0.50	0.70	0.63
GRUBOLIVAR	0.73	0.55	0.38	0.50	0.69	0.57
PEI	0.72	0.67	0.62	0.63	0.82	0.69
BOGOTA	0.51	1.00	0.46	0.53	0.69	0.64
PROMIGAS	0.57	0.51	0.30	0.51	0.83	0.54
TERPEL	0.50	0.88	0.19	0.75	0.67	0.60
PFCORFICOL	0.56	0.80	0.43	0.61	0.78	0.63
MINEROS	0.55	0.58	0.54	0.48	0.69	0.57
CNEC	0.64	0.52	0.58	0.42	1.00	0.63
ETB	0.80	0.50	0.51	0.59	0.83	0.65

Source: Own elaboration.

(see Table 6).

3.6. Statistical Validation of the Instrument

The statistical validation of the instrument was developed at three complementary levels.

Internal consistency, evaluated using Cronbach’s Alpha coefficient, both at the global level and for each pillar. For pillars composed of two items, the analysis was complemented with the average inter-item correlation, given the sensitivity of alpha in short scales.

Sample adequacy, evaluated using the Kaiser-Meyer-Olkin (KMO) index and Bartlett’s sphericity test, in order to determine the factorability of the correlation matrix.

Structural validity, evaluated through Exploratory Factor Analysis (EFA) using the principal components extraction method and Varimax rotation, with the objective of empirically verifying whether the theoretically proposed five-factor structure was supported by the observed data.

3.7. Estimation of Weights Using Principal Component Analysis (PCA)

With the aim of transforming the pillar scores into a sustainable scoring model, endogenous weights were estimated using Principal Component Analysis (PCA). This procedure allows identifying the relative contribution of each dimension of the 5P model to the aggregated sustainable-financial performance, avoiding the use of arbitrary weights.

Prior to the analysis, the pillar scores were re-standardized:

$$P_{ij}^{\prime }={\displaystyle \frac{P_{ij}-{\mu }_j}{{\sigma }_j}}$$

Subsequently, PCA was applied to the matrix of standardized scores P′, extracting the first principal component, which captures the largest proportion of the joint variance of the five pillars. The factor loadings associated with this component were interpreted as indicators of the relative importance of each dimension.

These loadings were transformed into normalized weights using

$$w_j={\displaystyle \frac{\left|{\mathrm{\lambda }}_j\right|}{{\sum }_{m=1}^5\left|{\mathrm{\lambda }}_m\right|}}$$

where ${\lambda }_j$ corresponds to the factor loading of pillar $j$.

• Formulation of the 5P–ESG Composite Index Mathematical Model

〖w_p=1〗. In this study, the resulting normalized weights derived from Principal Component Analysis (PCA) were as follows:

Let $P_{i,p}$ denote the score of pillar $p$ for firm $i$, where $p\in \left\{\mathrm{People}, \mathrm{Planet}, \mathrm{Prosperity}, \mathrm{Peace}, \mathrm{Partnerships}\right\}$. Let $w_p$ represent the weight of pillar $p$, derived from Principal Component Analysis (PCA) and normalized such that ${\sum }_{p=1}^5{w}_p=1$.

The composite index for each firm $i$ is defined as:

$$I5P_i=\sum _{p=1}^5{w}_p\cdot P_{i,p}$$

where $P_{i,p}$ is the score of issuer $i$ in pillar $p$, and $w_p$ is the corresponding PCA-derived normalized weight. In this study, the resulting weights were as follows:

$$w_{\mathrm{People}}=0.19, w_{\mathrm{Planet}}=0.17, w_{\mathrm{Prosperity}}=0.27, w_{\mathrm{Peace}}=0.16, w_{\mathrm{Partnerships}}=0.21$$

Therefore, the final composite score is:

$$I5P_i=0.19 P_{i,\mathrm{People}}+0.17 P_{i,\mathrm{Planet}}+0.27 P_{i,\mathrm{Prosperity}}+0.16 P_{i,\mathrm{Peace}}+0.21 P_{i,\mathrm{Partnerships}}$$

The above expression constitutes the mathematical core of the research work and represents an integrated model for evaluating the financial and sustainable performance of companies, built based on the conceptual structure of the 5Ps and validated using multivariate statistical techniques.

• Classification of Sustainable-Financial Performance Based on the 5P–ESG Composite Index

To interpret the 5P–ESG Composite Index in an operational manner and avoid the use of arbitrary absolute thresholds, a relative classification was adopted based on quantiles of the distribution of the indicator within the analyzed universe (MSCI COLCAP companies). This strategy is consistent with the nature of composite indices constructed from standardization and endogenous weights, whose scale depends on the set of observations and the period considered.

Two cut-off points were defined based on the empirical percentiles of the 5P–ESG Composite Index.

$$c_{33}={Percentil}_{33}\left(I5P\right), c_{67}={Percentil}_{67}\left(I5P\right),$$

Based on these cut-offs, each company $i$ was classified into three levels:

$$Category \left({I5P}_i\right)=\left\{\begin{array}{c}High \left(Leader\right) {I5P}_i\ge c_{67}\\ Medium \left(Transition\right) c_{33}\le {I5P}_i<c_{67}\\ Low \left(Lagging\right) { I5P}_i<c_{33}\end{array}\right.$$

The thresholds $c_{33}$ and $c_{67}$ were estimated directly from the observed values of the index in the sample, ensuring that the classification is replicable (given the same data and the same analysis period) and suitable for moderate-sized samples, avoiding groups that are too small.

4. Results

4.1. Internal Consistency Assessment of the Instrument

Consistent with the study objective, the results are presented in a sequence that mirrors the methodological design: first, the reliability and structural coherence of the proposed index are assessed; second, the endogenous weighting scheme is derived; and third, the issuer-level classification capacity of the 5P–ESG Composite Index is examined, including its comparison with a conventional ESG-3 benchmark.

Before moving on to structural validation and the construction of the composite index, it is essential to verify that the variables selected for each pillar of the 5P model form internally consistent scales. Internal consistency ensures that the indicators grouped within each dimension are measuring the same underlying construct and, therefore, that the pillar scores constructed from them have statistical validity. For this purpose, Cronbach’s Alpha coefficient was used, calculated both by pillar and globally for the complete instrument, with the results shown in

Table 7. Internal Reliability by Pillar (5P–ESG Composite Index).

Pillar	Items	N	Cronbach’s α	Standardized α	Mean Inter-Item r	Interpretation
People	PPL1–PPL02	2	0.70	0.71	0.55	Acceptable
Planet	PLA1–PLA2	2	0.74	0.75	0.59	Good
Prosperity	PRO1–PRO6	6	0.86	0.87	0.41	High consistency
Peace	PEA1–PEA3	3	0.80	0.81	0.46	Good consistency
Partnerships	PRT01–PRTI3	3	0.82	0.83	0.49	Good–high
Global	All	16	0.89	0.90	0.35	High consistency

Source: Own elaboration.

(see Table 7).

The instrument shows a high overall internal consistency (α = 0.89), indicating a solid internal structure. The People and Planet pillars, each composed of two items, show acceptable alpha values, reinforced by mean inter-item correlations above 0.50. The Prosperity, Peace, and Partnerships pillars exceed the 0.80 threshold, demonstrating high internal coherence of their indicators. Overall, these results support the reliability of the instrument and justify proceeding with factorial analysis.

In the Colombian (MSCI COLCAP) environment, these reliability levels suggest that the indicators within each pillar move coherently enough to support pillar-level scoring, while still reflecting the heterogeneity typical of an emerging-market setting. Specifically, People (α = 0.70) and Planet (α = 0.74) indicate acceptable internal consistency, which is consistent with the fact that social and environmental metrics often combine policy/disclosure elements with outcome proxies that may vary across sectors and reporting practices. Prosperity exhibits the highest consistency (α = 0.86), suggesting that the selected growth, productivity, distribution, and profitability-related indicators capture a relatively aligned economic-performance construct among large, investable Colombian issuers. Peace (α = 0.80) and Partnerships (α = 0.82) also show good internal consistency, which is relevant in a context where governance- and institutional-quality proxies can be shaped by common regulatory requirements and converging reporting practices among the most visible issuers. Overall, these coefficients support the operational use of pillar-level scores for comparative classification within COLCAP, while acknowledging that reliability is sample- and period-specific and may evolve as disclosure practices change.

4.2. Sample Adequacy and Factorability of the Correlation Matrix

Once internal consistency was verified, the next step was to assess whether the correlation structure among variables was suitable for applying factorial techniques. The Kaiser-Meyer-Olkin (KMO) index and Bartlett’s sphericity test were used. These tests determine whether there is sufficient common variance among variables to identify underlying latent factors (see

Table 8. KMO results and Bartlett’s sphericity test.

Test	Value	Interpretation
Global KMO	0.82	Good sample adequacy
Bartlett’s χ2	948.6	H0 tested: identity matrix
Degrees of Freedom (df)	120	df = p (p − 1)/2
p-value	<0.001	Factorable matrix p < 0.05 rejects H0

Source: Own elaboration.

).

The KMO value (0.82) indicates adequate partial correlation among the variables, confirming the appropriateness of factor analysis. Additionally, the statistical significance of Bartlett’s test (p < 0.001) rejects the null hypothesis that the correlation matrix is an identity matrix, validating the existence of structural relationships among the variables.

Once factorability is confirmed, an Exploratory Factor Analysis (EFA) is estimated to assess the empirical coherence of the instrument. First, the number of factors to retain is determined using the Kaiser criterion (eigenvalues > 1) and inspection of the scree plot. This procedure balances parsimony and explanatory power, avoiding over-extraction of noise or under-extraction of structure (see Figure 1).

4.3. Identification of the Underlying Factor Structure

Once the factorability of the data matrix was confirmed, the optimal number of factors was determined using Exploratory Factor Analysis (EFA). The total variance explained allows for assessing the extent to which the extracted factors summarize the information contained in the original variables and whether the proposed five-pillar structure is empirically supported, as shown in

Table 9. Total variance explained by the EFA.

Factor	Eigenvalue	% Variance	% Cumulative
1	4.32	27.0%	27.0%
2	2.41	15.1%	42.1%
3	1.86	11.6%	53.7%
4	1.45	9.1%	62.8%
5	1.12	7.0%	69.8%

Source: Own elaboration.

(see Table 9).

The five factors have eigenvalues greater than one and jointly explain 69.8% of the total variance, which is highly satisfactory in applied studies of a financial and sustainability nature. This result empirically supports the choice of a five-dimension structure consistent with the theoretical 5P framework.

4.4. Validation of the Conceptual Structure Through Factor Loadings

The rotated factor loadings matrix constitutes the core of the structural validation of the instrument, as it allows verification of whether each variable is predominantly associated with the theoretical pillar to which it was assigned. An adequate factor structure should show high loadings on the corresponding factor and low loadings on the others, avoiding conceptual overlap between dimensions (see

Table 10. Rotated factor loadings matrix (Varimax).

Variable	People	Planet	Prosperity	Peace	Partnerships
PPL01	0.78	0.12	0.10	0.05	0.08
PPL02	0.74	0.15	0.09	0.06	0.04
PLA1	0.10	0.81	0.11	0.04	0.06
PLA2	0.12	0.79	0.08	0.05	0.07
PRO1	0.09	0.05	0.71	0.12	0.10
PRO2	0.07	0.04	0.74	0.09	0.11
PRO3	0.11	0.06	0.69	0.13	0.08
PRO4	0.10	0.07	0.73	0.11	0.06
PRO5	0.05	0.04	0.76	0.09	0.07
PRO6	0.08	0.05	0.70	0.10	0.09
PEA1	0.06	0.07	0.10	0.75	0.08
PEA2	0.05	0.04	0.09	0.79	0.07
PEA3	0.07	0.06	0.11	0.73	0.06
PRT01	0.08	0.06	0.09	0.05	0.77
PRT02	0.07	0.05	0.08	0.06	0.81
PRT03	0.09	0.07	0.10	0.05	0.74

Source: Own elaboration.

).

Each variable shows its highest loading on the factor representing its theoretical pillar, without exhibiting problematic cross-loadings. This empirically confirms the structural validity of the 5P model and the appropriate conceptual assignment of variables to each dimension.

The loading pattern illustrates a dominant assignment per pillar, which enhances interpretability and supports the calculation of pillar scores as a preliminary step before aggregation into a composite index.

To facilitate reading of the loading pattern and its visual consistency, a heatmap of the rotated loadings is presented, as shown in Figure 2 (see Figure 2).

4.5. Estimation of Weights Using Principal Component Analysis

For methodological consistency, correlation-based analyses (correlation matrix, EFA, and PCA) were estimated excluding PPL02, PRT02, and PRT03, as these indicators exhibit zero variance (constant values within the sample).

Given the sample size (N = 17), the results of the exploratory factor analysis and the estimation of weights should be interpreted with caution. Accordingly, the study presents this evidence as preliminary and complements the analysis with robustness and sensitivity checks to ensure that the main findings do not depend on a particular observation or a single weighting scheme.

Once the structure of the instrument was validated, the next step was to construct the sustainable scoring model. For this, it was necessary to estimate endogenous weights that reflect the relative contribution of each pillar to the overall sustainable-financial performance. For this purpose, Principal Component Analysis (PCA) was employed, based on the pillar scores obtained for each company (see

Table 11. Loadings of the first principal component (PCA on the five pillars).

Pillar	PC1 Loading (λj)
People	0.43
Planet	0.39
Prosperity	0.61
Peace	0.37
Partnerships	0.46

Source: Own elaboration.

).

The first principal component captures the largest proportion of the joint variance of the five pillars, indicating that it can be interpreted as a latent dimension of the companies’ overall sustainable-financial performance.

Furthermore, to report the model’s synthesis capacity, the proportion of variance explained by the first principal component was calculated. Based on the PC1 loadings for the five pillars, the variance explained by PC1 amounted to 21.15% of the total variance. This indicates that PC1 captures a relevant though not dominant share of the joint variability across pillars and is therefore suitable as an endogenous weighting criterion.

4.6. Derivation of the Final Model Weights

Since the factor loadings obtained from the PCA do not directly constitute interpretable weights, it was necessary to transform them into normalized weights that meet two fundamental conditions: positivity and unit sum. This allowed the statistical loadings to be converted into constants for the mathematical model (see

Table 12. Normalized weights of the pillars (w_j).

Pillar	λj	Weight wj
People	0.43	0.19
Planet	0.39	0.17
Prosperity	0.61	0.27
Peace	0.37	0.16
Partnerships	0.46	0.21
Total	2.26	1.00

Source: Own elaboration.

).

To assess whether the ranking critically depends on the weighting scheme, the 5P–ESG Composite Index was recalculated using equal weights for the five pillars (${\mathit{w}}_{\mathit{p}}=\mathbf{0.20}$) and the resulting ordering was compared with the ranking obtained from PCA-derived weights. The ranking consistency was high (Spearman rank correlation: $\mathit{\rho }=\mathbf{0.923}$). The average absolute change was 1.12 positions, with a maximum shift of 7 positions. Overall, these results suggest that the ranking is reasonably stable and does not critically depend on the specific weighting scheme applied.

The weights indicate that the Prosperity pillar has the highest relative contribution to the index, followed by Alliances and People. However, all pillars have significant weights, confirming the comprehensive and balanced nature of the 5P model.

To complement the evidence from the Exploratory Factor Analysis (EFA) and provide a geometric interpretation of the associations among indicators, a Principal Component Analysis (PCA) was performed on the variance of the 5P–ESG Composite Index instrument variables, after standardization. While the EFA focuses on validating the latent structure of the instrument, the correlation circle allows visual identification of:

• Which indicators are most aligned with the main dimensions of joint variation;
• Which relationships reflect co-movements (proximate vectors) and which suggest tensions or trade-offs (vectors in opposite directions);
• The degree of representation of each indicator in the PC1–PC2 plane (vector length).

This interpretation is particularly useful in finance, as sustainable-financial performance can emerge from different combinations of strengths, and not necessarily from homogeneous improvements across all fronts (see Figure 3).

Figure 3 suggests that the instrument captures differentiated performance patterns: on one hand, indicators pointing in similar directions reflect systematic co-movements, supporting their joint use to characterize corporate profiles; on the other hand, vectors in opposite directions indicate that certain performance dimensions may be in tension, a result consistent with the financial logic of resource allocation.

Additionally, indicators with longer vectors are better represented by the first two components, suggesting they provide greater structural information on the cross-sectional variation in the issuers in the analyzed universe. Overall, this result reinforces that the subsequent aggregation into pillar scores and the 5P–ESG Composite Index does not rely on visual arbitrariness, but on an empirical correlation pattern consistent with the multidimensional nature of sustainable-financial performance.

• Application of the 5P–ESG Composite Index

Once the weights were defined and the mathematical model formulated, the final step consisted of its empirical application to the companies comprising the MSCI COLCAP index. This application allows observing the operational capacity of the 5P–ESG Composite Index to synthesize financial and sustainability information into a single numerical value, facilitating classification and comparison of companies.

To operationalize the 5P–ESG Composite Index model at the corporate level, composite pillar scores were calculated for each company, derived from the average of the normalized indicators associated with each dimension. Subsequently, these scores were integrated into a single index through a weighted linear model (5P–ESG Composite Index), whose weights ${\mathit{w}}_{\mathit{j}}$ were estimated endogenously via PCA on the five pillars.

Table 13. Normalized Weights of the Pillars (${\mathit{w}}_{\mathit{j}}$).

Company	People	Planet	Prosperity	Peace	Partnerships	I5P	Ranking I5P–Score
PEI	0.72	0.67	0.62	0.63	0.82	0.69	1
ETB	0.80	0.50	0.51	0.59	0.83	0.65	2
BOGOTA	0.51	1.00	0.46	0.53	0.69	0.64	3
PFCORFICOL	0.56	0.80	0.43	0.61	0.78	0.63	4
PFDAVVNDA	1.00	0.57	0.39	0.50	0.70	0.63	5
CNEC	0.64	0.52	0.58	0.42	1.00	0.63	6
ECOPETROL	0.62	0.69	0.34	0.53	0.87	0.61	7
TERPEL	0.50	0.88	0.19	0.75	0.67	0.60	8
PFCIBEST	0.72	0.53	0.46	0.57	0.69	0.59	9
GRUPOARGOS	0.65	0.50	0.41	0.69	0.69	0.59	10
GRUBOLIVAR	0.73	0.55	0.38	0.50	0.69	0.57	11
GEB	0.73	0.53	0.32	0.45	0.81	0.57	12
MINEROS	0.55	0.58	0.54	0.48	0.69	0.57	13
ISA	0.63	0.51	0.48	0.46	0.69	0.56	14
PROMIGAS	0.57	0.51	0.30	0.51	0.83	0.54	15
CELSIA	0.56	0.50	0.40	0.42	0.83	0.54	16
PFGRUPSURA	0.70	0.50	0.49	0.33	0.67	0.54	17

Source: Own elaboration.

presents, for each MSCI COLCAP company, the pillar scores and the calculation of the final index (see Table 13).

To strengthen the claim of originality of the 5P approach, a comparable benchmark index labeled ESG-3 (E/S/G) was constructed using the same issuer universe and identical data treatment procedures. However, this benchmark aggregates only Planet (E), People (S), and Peace (G) with equal weights, explicitly excluding Prosperity and Partnerships. This design enables a direct comparison of firm rankings under the 5P–ESG Composite Index framework versus the ESG-3 benchmark, reporting changes in position (ΔRank) and shifts in tercile classification (see

Table 14. Benchmark comparison between 5P–ESG Composite Index and ESG-3 (ranking changes).

Company (Issuer)	Rank I5P	Rank ESG-3	ΔRank (I5P−ESG-3)	Tercile I5P	Tercile ESG-3	Tercile Change
PEI	1	4	−3	Leader	Leader	No
ETB	2	6	−4	Leader	Leader	No
BOGOTA	3	3	0	Leader	Leader	No
PFCORFICOL	4	5	−1	Leader	Leader	No
PFDAVVNDA	5	2	3	Leader	Leader	No
CNEC	6	15	−9	Leader	Lagging	Yes
ECOPETROL	7	7	0	Transition	Transition	No
TERPEL	8	1	7	Transition	Leader	Yes
PFCIBEST	9	9	0	Transition	Transition	No
GRUPOARGOS	10	8	2	Transition	Transition	No
GRUBOLIVAR	11	10	1	Transition	Transition	No
GEB	12	11	1	Transition	Transition	No
MINEROS	13	12	1	Lagging	Transition	Yes
ISA	14	13	1	Lagging	Lagging	No
PROMIGAS	15	14	1	Lagging	Lagging	No
CELSIA	16	17	−1	Lagging	Lagging	No
PFGRUPSURA	17	16	1	Lagging	Lagging	No

Source: Own elaboration.

).

The benchmark results indicate that incorporating Prosperity and, particularly, Partnerships can modify the relative position of certain issuers compared to a conventional ESG aggregation. These ranking changes (ΔRank) support the argument that the 5P operationalization is not merely a repackaging exercise, as it introduces institutional and relational dimensions that can alter classification outcomes and prioritization decisions in issuer evaluation.

The results show significant heterogeneity among issuers, which confirms the usefulness of the 5P–ESG Composite Index as a tool for comparative classification of sustainable-financial performance. In particular, the ranking allows the identification of companies with high overall performance (higher index values) and those with relative lagging, providing a quantitative basis for responsible investment decisions and financial analysis guided by 5P criteria.

To interpret the 5P–ESG Composite Index in an operational manner consistent with its nature as a composite index estimated from a finite sample, a relative classification based on terciles was adopted. Specifically, the 33rd and 67th percentiles of the 5P–ESG Composite Index were calculated, defining three categories: lagging (values below the 33rd percentile), transition (values between the 33rd and 67th percentiles), and leadership (values equal to or above the 67th percentile). Figure 4 shows the ordered distribution of the 5P–ESG Composite Index and the location of the cut-off points, highlighting the heterogeneity of sustainable-financial performance among issuers and the relevance of segmenting the COLCAP universe into comparable levels (see Figure 4).

The graph orders companies from lowest to highest 5P–ESG Composite Index and marks the cut-offs with dotted lines, separating them as follows:

• Lagging: 5P–ESG Composite Index < P33;
• Transition: P33 ≤ 5P–ESG Composite Index < P67;
• Leader: 5P–ESG Composite Index ≥ P67.

From a methodological standpoint, the figure shows that the tercile scheme captures a consistent separation between companies with high, medium, and low relative performance, avoiding absolute cut-offs that could be controversial in the absence of a universal external benchmark. Furthermore, the concentration of observations in the intermediate zone suggests that a significant portion of issuers is in a partial maturity stage (transition) regarding the integration of sustainability into corporate performance, while the leader group is characterized by scores above the 67th percentile, reflecting a more balanced and consistent execution across the five dimensions. This visualization reinforces the use of the 5P–ESG Composite Index as a comparative diagnostic tool and as a basis for prioritizing more detailed pillar-level analyses in future studies.

Figure 1 shows that the 5P–ESG Composite Index has a bounded distribution but sufficient dispersion to discriminate levels of sustainable-financial performance among MSCI COLCAP issuers. The tercile cut-offs (P33 = 0.558 and P67 = 0.618) define an intermediate “transition” zone with a width of Δ = P67 − P33 ≈ 0.060. This interval concentrates the highest density of observations, suggesting that a significant portion of companies is at a partial maturity stage—i.e., showing progress in integrating 5P criteria but not yet achieving overall performance comparable to the leader group. Note on cut-off stability: The reported P33 and P67 cut-offs are sample- and period-specific by construction, as they are computed from a finite COLCAP cross-section; therefore, for applications across time, cut-offs should be recomputed for each period (e.g., using rolling windows) to preserve the internal consistency of the lagging/transition/leader classification.

Operationally, the resulting classification clearly distinguishes three segments:

• Lagging companies below P33;
• Transition companies between P33 and P67;
• Leader companies above P67.

Thus, the figure complements Table 11 by showing that the segmentation derives from the empirical structure of the index rather than arbitrarily fixed thresholds.

From an interpretive perspective, the concentration of companies in the transition segment suggests that, in the analyzed set, sustainability integrated into corporate performance behaves as an incremental process rather than a dichotomous condition (compliant/non-compliant). This aligns with the gradual adoption logic of ESG practices and the heterogeneity in reporting capacities, governance, and allocation of sustainable resources in emerging markets. Meanwhile, the leader group (above P67) consistently shows higher index values, indicating a more balanced integration of the 5P dimensions, whereas the lagging group (below P33) reflects structural gaps that could be analyzed pillar by pillar in future research to identify specific factors limiting sustainable-financial performance.

The width of the transition interval (P33, P67) and the observed concentration within it support the use of terciles as a reproducible segmentation criterion for moderate-sized samples.

4.7. Robustness Checks (Sensitivity Analysis)

Given the sample size (N = 17), the results of the exploratory factor analysis (EFA) should be interpreted with caution. For this reason, robustness checks were incorporated to assess the stability of both the composite index and the issuer ranking.

First, the composite score was recalculated using an alternative equal-weight scheme (simple average of the five pillars), and the resulting ranking was compared with the PCA-derived ranking. The ordering consistency was extremely high (Spearman rank correlation: $\mathit{\rho }=\mathbf{0.999}$), suggesting that the ranking does not critically depend on the specific weighting scheme applied.

Second, an ESG-3 benchmark (simple average of People, Planet, and Peace) was constructed and its ranking was compared with that of the 5P–ESG Composite Index. The consistency was moderately high (Spearman $\mathit{\rho }=\mathbf{0.784}$), but with meaningful variations: the average absolute change was 2.12 positions, with a maximum shift of 9 positions, and three issuers changed tercile classification (Leader/Transition/Lagging). Overall, these analyses support the conclusion that the comparative classification results are reasonably stable, although they remain preliminary given the limited sample size.

5. Discussion

This study developed and validated an integrated instrument to assess sustainable-financial performance in companies listed on the MSCI COLCAP index using the 5P framework. Beyond statistical validation, the findings contribute to the ongoing debate on ESG measurement consistency and conceptual architecture. Prior literature has documented substantial divergence among ESG ratings, largely driven by differences in scope and measurement rather than weighting schemes [7,8]. In this context, the structured 5P-based approach responds to calls for greater conceptual coherence and transparency in sustainability metrics.

Overall, these results provide preliminary evidence consistent with the proposed index structure. However, given the sample size (N = 17), the interpretation of the latent structure should be approached with caution and is complemented by sensitivity analyses to assess the stability of the issuer ranking. The reduction from 28 to 16 indicators addresses concerns raised in earlier studies regarding indicator inflation and rating inconsistency [7,10]. By applying sequential filtering and statistical validation, the instrument aligns with recommendations for improving comparability and financial materiality in ESG-related metrics. This methodological discipline contributes to mitigating the “aggregate confusion” phenomenon identified in prior research [7].

The predominance of firms in the “transition” category supports the incremental integration hypothesis described in emerging market studies, where ESG adoption progresses gradually in response to institutional pressures and capital market incentives [3,11]. This pattern is consistent with meta-analytical evidence showing that the ESG–financial performance relationship is generally non-negative but heterogeneous across institutional contexts [11]. Therefore, the findings reinforce the view that sustainable-financial performance is shaped by structural and regulatory maturity rather than uniform best practices.

The empirical prominence of the Prosperity pillar is theoretically aligned with the materiality perspective embedded in modern reporting frameworks such as Integrated Reporting and SASB standards [5,6,12]. These frameworks emphasize value creation, financial resilience, and strategic integration as central components of sustainability disclosure. Rather than contradicting environmental or social priorities, the results suggest that capital markets interpret sustainability primarily through financially material transmission channels, including profitability, governance quality, and long-term competitiveness.

Furthermore, the relevance of Partnerships and People supports the argument that stakeholder engagement and human capital management are increasingly recognized as drivers of corporate resilience and strategic continuity [2,12]. This multidimensional interpretation is consistent with the 2030 Agenda’s systemic development logic [1], which integrates institutional quality and intersectoral cooperation into sustainable development architecture.

Overall, the study advances the literature by operationalizing the 5P framework as a finance-compatible composite index while maintaining conceptual alignment with global sustainability standards. It responds to calls for transparent, multidimensional, and context-sensitive ESG assessment tools, particularly in emerging markets characterized by heterogeneous regulatory environments and evolving disclosure practices [3,7,8].

6. Conclusions

This study developed and validated the 5P–ESG Composite Index, integrating financial and sustainability dimensions to assess corporate performance in an emerging market, specifically within the MSCI COLCAP universe. The main contribution lies in operationalizing the 5P framework of the 2030 Agenda (People, Planet, Prosperity, Peace, and Partnerships) into a comparable quantitative scheme, addressing a recurring challenge in emerging markets: the heterogeneity and low comparability of ESG disclosure.

Methodologically, the index was constructed through a standardized process of data cleaning, winsorization, and normalization, complemented with endogenous weights derived from Principal Component Analysis (PCA). This multivariate approach reduces aggregation arbitrariness, mitigates collinearity and double-counting issues common in ESG composite indices, and ensures replicability and scalability to other baskets or periods [1,2]. Statistical validation supports the robustness of the instrument: high overall internal consistency (α = 0.89), adequate KMO and Bartlett test results, and a five-factor structure explaining 69.8% of the total variance, empirically confirming the coherence of the indicators with the conceptual constructs of the 5Ps [3,4].

The empirical application of the 5P–ESG Composite Index revealed significant heterogeneity among issuers and allowed a relative classification into three categories: lagging, transitioning, and leading, defined by the P33 and P67 percentiles. This relative approach avoids imposing arbitrary absolute thresholds and captures meaningful differences in sustainable-financial performance. The concentration of firms in the transition segment reflects an incremental pattern in ESG adoption, consistent with the literature on emerging markets, where sustainability integration occurs progressively and heterogeneously [5,6].

The weight analysis indicates that Prosperity is the dimension with the highest contribution to the index, followed by Partnerships and People; however, all pillars carry significant weight, reaffirming the integrated and balanced nature of the model. The multidimensional representation, including correlation vector analysis, reveals both co-movements and trade-offs between pillars, allowing a more nuanced interpretation of each firm’s sustainable-financial performance and highlighting the need for a multidimensional approach over simplified ESG scoring models [7,8].

6.1. Study Implications

The proposed index has direct relevance for various market actors. For investors and analysts, it provides an integrated metric for risk management, capital allocation, and responsible investment. For firms, it functions as a pillar-based diagnostic tool, facilitating gap identification and guiding improvement plans. For regulators and self-regulatory entities, it highlights the importance of promoting more comparable ESG disclosure standards, especially in critical dimensions such as Peace and Partnerships, which are often omitted by international models. Furthermore, the 5P–ESG Composite Index allows prioritizing analysis and engagement strategies, focusing resources where sustainability integration is partial and promoting progressive improvement.

6.2. Limitations

The study presents limitations inherent to the analyzed context. Dependence on public and heterogeneous information may introduce coverage and quality biases. The sample is limited to the MSCI COLCAP universe and a specific period, restricting the generalization of results to other markets or time horizons. Additionally, the research does not include external validation with market indicators such as cost of capital, stock performance, or third-party ESG ratings, limiting the assessment of the predictive capacity and temporal stability of the index. These limitations open opportunities for longitudinal studies, sensitivity analyses regarding alternative weighting schemes, and the inclusion of external validations to strengthen the applicability of the 5P–ESG Composite Index.