Investaging the Influence of Artificial Intelligence on Regional Economies and Firm Dynamics

This research dissertation investigates the intricate dynamics of artificial intelligence (AI) adoption on regional economicstructures, labor market inequality, and industry-wide spillover effects. The project's innovative edge lies in its threefoldapproach: mapping AI's geospatial diffusion, analyzing AI's economic and labor market consequences, and assessing thespillover effects from AI innovation hubs. Employing a combination of machine learning and natural language processingtechniques, the study will utilize big data sources like Istari.ai for firm-level AI adoption indicators and OpenAlex foracademic research trends. The study will be structured around three separate but interconnected Work Packages (WPs).
Work Package 1 (WP1) focuses on the spatial analysis of AI adoption patterns in various industries and regions. Twoprimary data sources will be utilized: Istari.ai, providing data on European companies, and OpenAlex, an open-accessacademic database. By linking academic research on AI with the intensity of AI application within firms, the study expectsto find a correlation that could explain regional clustering of AI-centric firms near research institutions. This symbioticrelationship between academia and industry is presumed to drive AI ecosystems and regional economic developments.The methodology combines machine learning for clustering analysis and natural language processing for topic modelingand sentiment analysis, aiming to shed light on the overarching themes and sentiments within the AI discourse.
Work Package 2 (WP2) focuses on the potential impact of firm-level AI activity on industry structure and local economies, with a focus on potential market concentration effects. Combining Istari.ai firm AI activity data with long-term local labormarket data from Germany's Institute of Employment Research (IAB), spanning over four decades, enables acomprehensive analysis. Employing AI activity measurements at regional and industry levels from WP1 analysis, this studywill probe whether regions with higher AI activity exhibit increased market concentration over time. Subsequently, the studywill delve into the implications for wage setting within firms, particularly exploring how AI activity influences wage dynamicsin labor markets with varying levels of concentration.
Work Package 3 (WP3) will evaluate AI technology adoption's spillover effects among firms, particularly analyzing thereciprocal influences of startups within AI-intense regions. This study utilizes the IAB/ZEW Start-Up Panel, acomprehensive dataset of business start-ups and new companies in Germany, renowned for its representative sample sizeand detailed information on digital technology adoption in new ventures. The study aims to estimate the causal effect ofexposure to AI intensity or market concentration on AI technology adoption, spillover effects. Treatment variables will bederived from metrics developed in WP1 and WP2 at the regional and industry levels. The study's identification strategyextends the standard unconfoundedness assumption under the potential outcomes framework to accommodateinterference among units. Estimation will be conducted using the generalized propensity score, balancing firm andneighborhood covariates across different treatment levels and exposure to neighbors' treatment.
This research is poised to offer significant contributions to policy and industry, informing strategies to foster equitable andsustainable economic growth in the age of AI.