A Shift-Share and Non-Parametric Associative Framework to Assess Gen AI Complementarity or Substitutability in Marketing Occupations by USA State, 2014–2024

Abstract

This paper proposes a methodology to identify complementarity or substitutability between marketing-related occupations and generative AI tools (such as ChatGPT) across U.S. states from 2014 to 2024. The suggested methodology applies a shift-share, non-parametric, associative, and descriptive statistical framework to achieve this goal. Results support a statistically significant negative association between Writers and Authors (SOC 27-3043) and per capita’s popularity index of ChatGPT suggesting a potentialsubstitutive relationship. While the visual analysis suggested a positive association between Marketing Specialists (SOC 13-1161), Marketing Managers (SOC 11-2021), and ChatGPT potentially indicating that per capita ChatGPT’s popularity was potentially complement of these marketing occupations; results were statistically insignificant. Similarly, the relationship was found to be inconclusive for graphic designers (SOC 27-1024) and per capita ChatGPT’s popularity index as mixed results were found when comparing the visual analysis with the descriptive, and non-parametric analyses. This study reinforces, from an associative perspective, the idea that AI tools have a heterogeneous and complex impact on marketing occupations.