Management Accounting Section Midyear Meeting

The Informativeness of Dark Data on Firm Performance

• In Event: Determinants of Firm Performance

Abstract

This study investigates whether an organization’s “dark data” can predict future firm performance. Specifically, I examine if the aggregate sentiment from employee emails predicts future revenue and operational efficiency incremental to traditional information sources. Further, I study if differences in information environment characteristics explain cross-sectional variation in the predictive ability of aggregate employee sentiment. To test these questions, I collaborate with a large, U.S.-based medical technology company seeking to improve its planning process. The firm provides financial data as well as access to de-identified emails of key employees involved in planning across sales, operations, and accounting functions. I conduct textual analysis of product-specific emails to develop an aggregate employee sentiment and test its relation with forecast error. This study is one of the first to empirically evaluate ways that organizations can use unstructured data to harden and quantify “soft” information to improve decision-making. Additionally, the results from this study offer practical implications for firms seeking to improve their planning processes with better information aggregation through utilization of unstructured data.

Author

• Ewelina Forker, Emory University