TITLE:
The “One-Size-Fits-All” Illusion in SME Forecasting: An Analytical Examination of Financial Account-Level Forecasting Models in SMEs
AUTHORS:
Aygul Farzaliyeva
KEYWORDS:
SME Financial Forecasting, Firm-Level Heterogeneity, Data-Constrained Forecasting, Model Transferability, Time-Series Forecasting in SMEs, Forecast Evaluation Framework, Financial Account-Level Prediction, Small Sample Econometrics
JOURNAL NAME:
Journal of Data Analysis and Information Processing,
Vol.14 No.3,
June
30,
2026
ABSTRACT: This study investigates the transferability of standardized financial account-level forecasting models across 50 small and medium-sized enterprises (SMEs) operating under conditions of short and heterogeneous time series data. Despite the widespread use of uniform forecasting practices in applied settings, the extent to which such models remain stable across firms with differing volatility patterns and data histories remains insufficiently understood. Five commonly used forecasting approaches including inter-month percentage change, year-over-year percentage change, shrinkage-based, median-based, and moving average models are applied uniformly across all firms and financial accounts. Forecast performance is evaluated using a categorical framework that distinguishes between Failed, Unstable, and Fair outcomes based on statistical stability and economic plausibility. Standard accuracy measures, including Mean Absolute Error (MAE) and Mean Absolute Scaled Error (MASE), are also computed to ensure comparability with the existing forecasting literature. In addition, a na?ve persistence benchmark is included to assess baseline predictive performance. Empirical results indicate that percentage-change-based models exhibit high instability, with failure rates ranging from 70% to 85% across firm-account evaluations. Shrinkage- and median-based approaches partially reduce extreme forecasting errors but remain unreliable for more than half of the sample. Moving average models demonstrate relatively improved stability; however, they still fail to produce consistently fair forecasts across SMEs. The na?ve benchmark further confirms that simple persistence is insufficient under the observed data conditions. Overall, the findings provide strong evidence that forecasting performance in SME environments is constrained primarily by limited model transferability rather than model selection. These results challenge the validity of one-size-fits-all forecasting frameworks and highlight the importance of firm-level heterogeneity in financial forecasting under data scarcity.