[论文解读] The Innovation Tax: Generative AI Adoption, Productivity Paradox, and Systemic Risk in the U.S. Banking Sector

Tatsuru Kikuchi|arXiv (Cornell University)|Feb 2, 2026

FinTech, Crowdfunding, Digital Finance被引用 0

一句话总结

该论文将动态空间Durbin模型与合成差分中的差分相结合，研究美国银行业的 GenAI 采用，揭示在 frontier 表现更高的同时存在短期生产率税以及显著的正向网络溢出效应。

ABSTRACT

This paper evaluates the causal impact of Generative Artificial Intelligence (GenAI) adoption on productivity and systemic risk in the U.S. banking sector. Using a novel dataset linking SEC 10-Q filings to Federal Reserve regulatory data for 809 financial institutions over 2018--2025, we employ two complementary identification strategies: Dynamic Spatial Durbin Models (DSDM) to capture network spillovers and Synthetic Difference-in-Differences (SDID) for causal inference using the November 2022 ChatGPT release as an exogenous shock. Our findings reveal a striking ``Productivity Paradox'': while DSDM estimates show that AI-adopting banks are high performers ($β> 0$), the causal SDID analysis documents a significant ``Implementation Tax'' -- adopting banks experience a 428-basis-point decline in ROE as they absorb GenAI integration costs. This tax falls disproportionately on smaller institutions, with bottom-quartile banks suffering a 517-basis-point ROE decline compared to 129 basis points for larger banks, suggesting that economies of scale provide significant advantages in AI implementation. Most critically, our DSDM analysis reveals significant positive spillovers ($θ= 0.161$ for ROA, $p < 0.01$; $θ= 0.679$ for ROE, $p < 0.05$), with spillovers among large banks reaching $θ= 3.13$ for ROE, indicating that the U.S. banking system is becoming ``algorithmically coupled.'' This synchronization of AI-driven decision-making creates a new channel for systemic contagion: a technical failure in widely-adopted AI models could trigger correlated shocks across the entire financial network.

研究动机与目标

• Motivate understanding of GenAI adoption in banking and its impact on productivity and systemic risk.
• Identify whether AI adoption signals frontier firm status or incurs an implementation tax.
• Examine heterogeneity by bank size and the role of network spillovers in productivity.
• Assess potential systemic risk from algorithmic coupling due to shared AI architectures.

提出的方法

• Dynamic Spatial Durbin Model (DSDM) to estimate direct and spillover productivity effects and to quantify network spillovers (beta for own adoption, theta for spillovers).
• Synthetic Difference-in-Differences (SDID) to identify causal effects using the 2023 ChatGPT shock as an exogenous treatment and construct ATT estimates.
• Two spatial weight matrices: network (asset-based) and geographic (distance-based) to capture different spillover channels.
• Outcome variables: ROA and ROE; treatment defined by GenAI mentions in SEC filings; controls include assets, Tier 1 ratio, digitalization index, and CEO age.
• Event-study extension to examine dynamic treatment effects and pre-trends.
• Estimation via Maximum Likelihood, Quasi-Maximum Likelihood, and Bayesian MCMC to ensure robustness.

(a) ATT on ROA (%)

实验结果

研究问题

• RQ1 Does GenAI adoption improve productivity for adopting banks relative to non-adopters?
• RQ2 What is the causal effect of GenAI adoption on productivity in the short run?
• RQ3 How does bank size affect the cost (implementation tax) of GenAI adoption and its short-run productivity impact?
• RQ4 Do AI adoption efforts generate positive spillovers across banks, indicating network effects or algorithmic coupling?

主要发现

• AI-adopting banks show higher productivity in cross-sectional comparisons, consistent with frontier firm selection (beta > 0).
• Causal SDID analysis finds a significant negative short-run effect of AI adoption on productivity: ROA down by 46 basis points and ROE down by 428 basis points.
• Smaller banks (bottom quartile by assets) suffer a larger ROE decline (517 bps) than larger banks (129 bps).
• DSDM reveals positive spillovers: theta = 0.161 for ROA and theta = 0.679 for ROE (p < 0.01 and p < 0.05 respectively).
• Spillovers among large banks are particularly strong for ROE (theta = 3.13).
• The results point to an “Innovation Tax” during implementation and to increasing systemic coupling as AI adoption spreads across the banking network.

(b) ATT on ROE (%)