AI-Enabled Financial Information Systems for Credit Risk Forecasting to Support Small Business Growth

Abstract

This quantitative study examined how AI-enabled financial information systems (AI-FIS) influenced credit risk forecasting performance, credit allocation outcomes, and small business growth indicators in SME lending. Using a cross-sectional dataset of 420 SME borrower records, the analysis measured AI-FIS adoption intensity through integrated data sources, underwriting automation, model update frequency, real-time monitoring capability, and explain ability module availability. Descriptive results showed that institutions integrated an average of 4.21 data sources (SD = 1.37), automated 62.40% (SD = 18.55) of underwriting decisions, and updated forecasting models 3.10 times per year (SD = 1.25). Real-time monitoring capability averaged 3.88/5 (SD = 0.92), and explain ability modules were present in 73% of institutions. Credit allocation outcomes showed moderate approval probability (M = 3.41/5, SD = 0.86) and strong pricing consistency (M = 3.92/5, SD = 0.74). SME outcomes indicated high survival (89%) and positive average revenue growth (M = 3.62/5, SD = 0.83), with working capital improvement (M = 3.58/5, SD = 0.82). Regression results indicated that AI-FIS adoption intensity significantly predicted forecasting performance (β = 0.54, p < .001), explaining 41% of the variance (R² = 0.41). Forecasting performance significantly predicted approval probability (β = 0.38, p < .001; R² = 0.29), pricing consistency (β = 0.42, p < .001; R² = 0.33), loan size alignment (β = 0.31, p < .001; R² = 0.25), and monitoring intensity (β = 0.35, p < .001; R² = 0.27). Credit allocation outcomes significantly predicted SME growth. Approval probability and loan size alignment predicted revenue growth (β = 0.29 and β = 0.24, p < .001; R² = 0.26) and employee growth (β = 0.21 and β = 0.19, p < .001; R² = 0.19). Pricing consistency predicted working capital improvement (β = 0.27, p < .001; R² = 0.31). Logistic regression showed monitoring intensity increased survival odds (OR = 1.48, p < .001), while approval probability increased survival odds (OR = 1.36, p = .003). Mediation analysis confirmed forecasting performance mediated the adoption-to-allocation relationship (p < .01), and loan approval partially mediated forecasting-to-growth outcomes (p < .05). Overall, findings supported a measurable mechanism in which AI-FIS adoption improved forecasting quality, strengthened credit allocation efficiency, and was associated with higher SME growth and survival outcomes.