Signal or Cheap Talk? AI Activity Disclosure and Firm Valuation on Nasdaq Stockholm, 2014-2024

Sammanfattning

This thesis examines whether disclosed AI activity in annual reports is associated with firm valuation and fundamental performance among Swedish non-technology companies listed on Nasdaq Stockholm Main Market between 2014 and 2024. Using textual analysis of 1,098 annual reports across 102 firms, we construct a firm-year measure of disclosed AI activity and relate it to i) the market-to-book ratio, ii) four measures of fundamental performance over horizons up to three years, and iii) R&D and SG&A intensity. We find that disclosed AI activity is positively associated with the market-to-book ratio only in our strictest specification, with significance at the 10 percent level, implying market-to-book ratios 8% to 23% higher for AI-disclosing firms than non-disclosing peers. It is also associated with subsequent improvements in three of four fundamental performance measures at the three-year horizon (including ROA of 12.5% to 37.5% higher than industry peers), with effects largely absent at shorter horizons. The findings indicate no significant association with R&D or SG&A intensity, suggesting that firms are disclosing AI activity without corresponding investment, so called AI-washing. Taken together, however, the deferred fundamental improvements are the clearest evidence against this AI-washing reading and suggest that markets price real underlying activity, though the small Swedish sample leaves room for the alternative reading.