Cluster analysis of real estate enterprises using compositional data analysis method based on DuPont model

Abstract

This study conducts a compositional data analysis (CoDa)- based clustering of listed real estate firms in Vietnam over the period 2012- 2023, using the DuPont model as the analytical framework. The dataset comprises 63 firms and 621 firm-year observations collected from the HOSE, HNX, and UPCOM exchanges. The findings reveal three distinct clusters characterized by heterogeneous financial profiles. Cluster 1, which represents the majority of the sample, maintains a long-term debt ratio of approximately 45% of total liabilities and exhibits stable profitability. Cluster 2 shows a comparable level of profitability but relies almost entirely on short-term financing, suggesting a conservative and less prevalent capital structure strategy. Cluster 3 employs the highest degree of financial leverage yet reports the lowest ROA and ROE, indicating heightened financial risk associated with excessive debt utilisation. The results highlight the critical role of capital structure-particularly long-term debt-in sustaining profitability and strengthening the competitive position of real estate firms. The study provides meaningful implications for managers, investors, and policymakers in assessing corporate financial health, formulating sustainable development strategies, and making investment decisions aligned with market conditions.