ECONOMETRIC ACCELERATOR
Paolo Savona, Pierfrancesco Savona, Antonio Simeone
The Econometric Accelerator is a pioneering research initiative from the Italian International Economic Center (IIEC) and the LUISS Quantum & AI Lab at LUISS Guido Carli University. Its core mission is to redefine how economic forecasting is conceived, modeled, and applied by fusing the rigor of econometrics with the advanced analytical power of artificial intelligence and complex systems science.
Instead of relying on a single model or a traditional statistical paradigm, the project introduces a framework of multiple, interconnected models that cooperate to interpret and predict economic phenomena. Each model is designed to capture different dimensions of the economy—from classical macroeconomic relationships to high-frequency market signals and alternative data sources—enabling a richer, more adaptive understanding of economic dynamics.
The Accelerator extends the frontiers of econometrics by integrating advanced methods like deep learning, econophysics, and network theory. Furthermore, the initiative is pioneering the use of multi-agent systems powered by Large Language Models (LLMs).
In this innovative approach, AI agents are not used to simulate economic actors but instead function as specialized analytical units.
Each agent is tasked with continuously monitoring and interpreting vast streams of unstructured data— from real-time social media sentiment and news reports to complex geopolitical analyses. They work to distill this qualitative information into structured, actionable signals that enrich the core econometric models. This allows the framework to organically embed emerging narratives and real-time human sentiment directly into its forecasting process, capturing crucial context that traditional quantitative data alone cannot provide. The framework processes a wide spectrum of continuously updated datasets, including financial transactions, energy flows, satellite observations, and sentiment indicators.
The immense computational demands of this work are met through a strategic partnership with CINECA (https://www.cineca.it/en), Italy's leading supercomputing consortium, which provides the essential high-performance computing infrastructure to power the Accelerator's analytical engine.
This approach allows the generation of both real-time analyses and medium-term forecasts, such as quarterly projections and scenario assessments.
Ultimately, the project seeks to bridge the gap between academic research and real-world economic decision-making. It aims to provide policymakers, financial institutions, and research centers with an analytical framework that evolves simultaneously with the economy itself.
The Econometric Accelerator was presented at LUISS Guido Carli University in Rome on May 6th, 2025, and was featured by Il Sole 24 Ore (https://en.ilsole24ore.com/art/the-accelerator-revolution-econometrics-AH0ic7XB), which highlighted how the initiative is accelerating the evolution of econometrics through AI-driven experimentation and open collaboration among institutions, researchers, and data partners.
