Year: 2026
Type: book-chapter
Abstract: Predictive modeling, especially in financial economics, has enabled new data-driven methodologies. These methodologies provide a new scope for evaluating economic theories through prediction rather than estimation. Predictive modeling methods have been well received recently in non-economics disciplines, although these techniques—non-parametric regression, time-series analysis, economic modeling, and machine learning—have been utilized long before. In the context of financial decision-making, predictive modeling often requires market forecasting, with such market forecasts being proposed by several studies. Furthermore, prediction is the clearest output of a model; thus, it can be a robust test of economic theory. Banking and financial economics perform large-scale quantitative modeling to evaluate capital and funding decisions and to re-evaluate big bank theory. Recent big bank failure hypothetical scenarios have created the need for novel predictive assets that build simple, adaptive models using just the past history of the endogenous variable. Predictive assets aim to maximize forecast accuracy rather than model fit and include several well-known alternative forecasting models—such as ARIMA, ARY, GARCH, and ensemble techniques—generated through bespoke predictive routines specifically designed for accurate and robust time-series prediction.
Author: Tarun Vakkalagadda
Institution: Goldman Sachs (United States)
Language: English
Field: Decision Sciences
Domain: Social Sciences
Open Access status: gold
OpenAlex ID: W7204023919