Machine Learning-Based Cash-Flow Forecasting for Improving Financial Planning in U.S. Small and Mid-Sized Businesses

Abstract

This study addresses the difficulty faced by U.S. small and mid-sized businesses in producing accurate cash-flow forecasts to support liquidity control, budgeting, working-capital allocation, financing, and resource planning, particularly where approaches do not capture financial patterns. The purpose of the study was to determine whether machine learning-based cash-flow forecasting improves financial planning effectiveness by examining relationships among financial data quality and integration, machine learning forecasting capability, cash-flow forecast accuracy, and financial planning effectiveness. A quantitative, explanatory, cross-sectional, embedded multiple-case design was applied across U.S. small and mid-sized businesses, with owners, finance managers, accountants, controllers, FP&A professionals, analysts, treasury personnel, and senior managers serving as embedded units of analysis. From 350 distributed questionnaires, 318 were returned and 312 usable responses were retained, yielding an 89.1% usable response rate. Data were collected using a structured five-point Likert questionnaire and analyzed through descriptive statistics, reliability and validity testing, Pearson correlation, and hierarchical multiple regression in SPSS. Mean scores indicated favorable levels of financial data quality and integration (M = 3.84, SD = 0.63), machine learning forecasting capability (M = 3.76, SD = 0.68), cash-flow forecast accuracy (M = 3.81, SD = 0.64), and financial planning effectiveness (M = 3.89, SD = 0.61). Significant positive correlations were found between machine learning forecasting capability and forecast accuracy (r = .67, p < .001) and between forecast accuracy and financial planning effectiveness (r = .71, p < .001). The final regression model explained 59% of the variance in financial planning effectiveness, F(8, 303) = 54.49, p < .001, with forecast accuracy emerging as the strongest predictor (β = .41), followed by machine learning forecasting capability (β = .29) and financial data quality and integration (β = .14). The findings indicate that integrated financial data, machine learning capability, and accurate forecasts strengthen financial planning in smaller firms.