Can LLM Improve for Expert Forecast Combination? Evidence from the European Central Bank Survey

TL;DR

This paper investigates whether large language models can improve expert forecast combination by applying ensemble learning techniques to ECB survey data, considering factors like disagreement, herd behavior, and attention limits.

Contribution

It introduces a framework for evaluating LLM-based ensemble predictions in expert forecasting, addressing key challenges like disagreement and herd behavior.

Findings

LLMs can enhance forecast accuracy in expert surveys.

Ensemble methods improve robustness against expert disagreement.

The framework identifies conditions where LLMs are most effective.

Abstract

This study explores the potential of large language models (LLMs) to enhance expert forecasting through ensemble learning. Leveraging the European Central Bank's Survey of Professional Forecasters (SPF) dataset, we propose a comprehensive framework to evaluate LLM-driven ensemble predictions under varying conditions, including the intensity of expert disagreement, dynamics of herd behavior, and limitations in attention allocation.