Measuring Firm-Level Inflation Exposure: A Deep Learning Approach

#artificialintelligence #machinelearning #finance #research Abstract: We develop a novel measure of firm-level inflation exposure by applying a deep learningapproach to firms’ earnings conference call transcripts. Our methodology not only identi-fies sentences that discuss price changes, but also differentiates price increases from pricedecreases, input prices from output prices, and future changes from past changes. To val-idate our methodology, we show that the correlation between our aggregate measure andPPI is 0.75. In the cross section, firms that have higher inflation exposure experience astrong negative stock reaction to earnings calls. This negative reaction is mostly driven bydiscussions on input price increases. Firms’ ex ante market power attenuates the negativemarket reaction. Consistent with the market reaction, firms with higher inflation exposurehave higher future costs of goods sold. We also observe a negative drift in the firm’s stockreturn after the earnings call, suggesting that it takes time for investors to fully incorporatefirm-level inflation exposure into stock prices.