Assessing Consistency and Reproducibility in the Outputs of Large Language Models: Evidence Across Diverse Finance and Accounting Tasks

TL;DR

This paper systematically evaluates the consistency and reproducibility of large language models in finance and accounting tasks, revealing task-dependent variability and the effectiveness of aggregation strategies, with implications for research reliability.

Contribution

It provides the first comprehensive analysis of LLM output consistency across multiple finance tasks and models, highlighting patterns and mitigation strategies.

Findings

Binary classification and sentiment analysis show high reproducibility.

Aggregation of multiple runs improves consistency and accuracy.

Downstream inferences remain robust despite output variability.

Abstract

This study provides the first comprehensive assessment of consistency and reproducibility in Large Language Model (LLM) outputs in finance and accounting research. We evaluate how consistently LLMs produce outputs given identical inputs through extensive experimentation with 50 independent runs across five common tasks: classification, sentiment analysis, summarization, text generation, and prediction. Using three OpenAI models (GPT-3.5-turbo, GPT-4o-mini, and GPT-4o), we generate over 3.4 million outputs from diverse financial source texts and data, covering MD&As, FOMC statements, finance news articles, earnings call transcripts, and financial statements. Our findings reveal substantial but task-dependent consistency, with binary classification and sentiment analysis achieving near-perfect reproducibility, while complex tasks show greater variability. More advanced models do not consistently demonstrate better consistency and reproducibility, with task-specific patterns emerging. LLMs significantly outperform expert human annotators in consistency and maintain high agreement even where human experts significantly disagree. We further find that simple aggregation strategies across 3-5 runs dramatically improve consistency. We also find that aggregation may come with an additional benefit of improved accuracy for sentiment analysis when using newer models. Simulation analysis reveals that despite measurable inconsistency in LLM outputs, downstream statistical inferences remain remarkably robust. These findings address concerns about what we term "G-hacking," the selective reporting of favorable outcomes from multiple generative AI runs, by demonstrating that such risks are relatively low for finance and accounting tasks.