AI-Based Forecasting Capability and Perceived Inflation Forecasting Effectiveness in Pakistan: The Roles of High-Frequency Data Integration, Nonlinear Pattern Recognition, and Macroeconomic Volatility
Abstract
This study examines whether AI-based forecasting capability improves Perceived Inflation Forecasting Effectiveness and how this effect operates through high-frequency data integration and nonlinear pattern recognition under macroeconomic volatility. Integrating dynamic capabilities theory, information-processing theory, and contingency theory, the study conceptualizes AI forecasting capability as an adaptive analytical capacity that converts heterogeneous, fast-moving signals into policy-relevant predictive knowledge. Survey data were collected from 215 forecasting-relevant professionals in Pakistan, including respondents from universities/research institutes, central banking, commercial banking, macroeconomic analysis units, research departments, and data/statistics functions. The model was tested in R using confirmatory factor analysis, reliability and validity diagnostics, HC3 structural regressions, and 5,000-sample bootstrapping for serial mediation and moderated mediation. AI-based forecasting capability showed a significant total effect on Perceived Inflation Forecasting Effectiveness (β=.516, p<.001). The direct effect became nonsignificant after mediators were included, while high-frequency data integration (β=.370, p<.001) and nonlinear pattern recognition (β=.277, p<.001) remained significant. Conditional indirect effects increased under high macroeconomic volatility, supporting first-stage moderated serial mediation. The study shifts debate from whether AI forecasts better to why, when, and through which forecasting mechanisms AI improves inflation prediction. Institutions should invest in algorithms, real-time data pipelines, nonlinear modeling expertise, volatility-responsive governance, transparency, and model discipline for credible policy decision-making under uncertainty.
Keywords: Artificial Intelligence Capability; Inflation Forecasting; Pakistan, An Emerging-Market Economy; High-Frequency Data; Nonlinear Pattern Recognition; Moderated Mediation; Macroeconomic Volatility
