Predictive Analytics in Economics: Data-Driven Approaches for Market Forecasting and Policy Planning

Abstract

The economics world is changing from a crisis that explains a paradigm to a paradigm for shocks to come. The amount of high-frequency and alternative data, coupled with low-cost computing, has transformed predictive analytics into a tool that can be used for market forecasting and policy planning. This conceptual paper brings together recent literature on data-driven analytics and provides an overarching framework for data-driven analytics that links data ingestion, feature engineering, hybrid econometric–machine learning modeling, calibration, and decision support. It will explore the ways in which new indicators help to broaden the economic toolkit, the ways in which hybrid models bridge the gap between interpretability and accuracy, and the ways in which forecasting relates to market and policy applications. It also includes the theme of "structural breaks", algorithmic feedback loops, and opacity as limitations on foresight. The paper ends with some recommendations both for policymakers and market players and an agenda for further research.