Design, Implementation and Evaluation of a Tax Evasion Risk Index for Small Firms

Thesis Abstract

Small firms face substantial challenges in detecting and preventing tax evasion, which undermines revenue mobilization and distorts competitive conditions. This study addresses the gap between existing compliance tools and the dynamic risk factors unique to small enterprises by designing, implementing, and evaluating a Tax Evasion Risk Index (TERI) tailored to small firms operating in dynamic regulatory environments. The aim is to develop a robust, actionable index that integrates behavioral, operational, and contextual determinants of non-compliance, and to assess its predictive validity, reliability, and practical utility for tax authorities and firm managers. Specific objectives are to (i) identify key drivers of tax evasion risk among small firms through a synthesis of theory and empirical evidence, (ii) construct a composite index that captures material, behavioral, and governance-related risk factors, (iii) validate the index using cross-sectional data from 600 small firms across manufacturing and service sectors, (iv) assess the index’s predictive performance against actual audit outcomes and reported irregularities, and (v) evaluate the feasibility and policy implications of deploying TERI in risk-based auditing. The study adopts a positivist, multi-method design, anchored in the Theory of Planned Behavior and Institutional Theory to explain compliance decisions and the influence of regulatory context. A cross-sectional survey of 600 small firms (employees <100) will be complemented by audit outcome data from the Revenue Authority covering the previous three fiscal years. Data collection will combine structured questionnaires addressing perceived norms, control beliefs, governance practices, record-keeping quality, and transaction characteristics, with archival records of tax filings, prior audits, penalties, and sector-specific risk indicators. Instrument validity and reliability will be established through content validation by tax compliance scholars and a pilot test (n=60), followed by Cronbach’s alpha assessments (target >0.7) and confirmatory factor analysis to verify the construct structure. Analysis will proceed in three stages (i) exploratory factor analysis to identify latent dimensions of evasion risk, (ii) construction of the TERI via weight estimation using logistic regression and regularization (LASSO) to prevent overfitting, and (iii) evaluation of predictive performance through out-of-sample validation (hold-out sample of 150 firms) and ROC-AUC metrics, along with calibration plots. Additional analyses will include multivariate regression to examine determinants of TERI scores, and subgroup analyses by sector and firm size to assess stability. The expected findings include a parsimonious, empirically grounded TERI comprising key dimensions such as accounting transparency, governance quality, cash-intensive transactions, sectoral risk, and regulatory exposure, with demonstrated predictive accuracy (ROC-AUC ? 0.80) for identifying high-risk firms. The study anticipates that TERI will outperform baseline models using conventional compliance indicators alone and will reveal heterogeneity in risk drivers across sectors, informing targeted enforcement and supportive interventions. The contribution to knowledge lies in (1) operationalizing a theory-driven, context-specific tax evasion risk index for small firms; (2) providing a validated measurement tool that integrates behavioral and institutional factors into risk assessment; and (3) offering evidence on the practical viability of risk-based audit prioritization for improving tax compliance while reducing compliance costs for small businesses. Policy and managerial implications include recommendations for integrating TERI into risk-scoring systems, enhancing data collection practices, and designing targeted education and support programs for firms with elevated risk profiles. The study concludes that a rigorously validated TERI can strengthen tax administration efficiency and fairness by enabling timely, proportionate responses to risk, while encouraging compliant behaviors through transparency and improved governance. Limitations anticipated include potential measurement error in self-reported behavioral constructs and limited generalizability to informal firms, with future research suggested to extend TERI to longitudinal designs and to incorporate experimental evaluation of enforcement and compliance interventions.