Data-Driven Decision Making, Predictive Analytics, and Management Practices in U.S. Manufacturing

Category: Applied Microeconomics and Organization Seminar
When: 12 June 2019
, 17:15
 - 18:30
Where: RuW 4.201


Management in America has become significantly more data-intensive, yet the economic and strategic implications of this shift are poorly understood. Working with the U.S. Census Bureau, we developed measures of how manufacturing firms have used data to guide decision making over the past decade. In our large and representative sample, data-driven decision making (DDD) is strongly correlated with productivity. The benefits attributable to DDD are distinct from those associated with other structured management practices or investment in IT, though the latter is an important complement. Moreover, instrumental variables estimates and timing falsification tests suggest a causal relationship. Implications for firm strategy, however, are nuanced; we find evidence of significant advantages for early adopters of DDD, particularly in the 2005-2010 window, when adoption rates in the sector were lower. Yet, we also observe timing-dependent complementarities and some effective following. The frontier of data-centric practices shifts during our study period, with some uses of predictive analytics taking over as a key driver of productivity gains from 2010 to 2015. Our study reveals the importance of certain managerial practices for deriving value from data, highlighting the strategic challenge of keeping up with a constantly-evolving technological and managerial frontier in the digital age.