Natural Experiments and Causation in Strategy Research




Knowledge management, Financial performance, Tobin's q, Merger & acquisition, Natural experiments


Research to inform organizational strategy decisions studies the impact of specific managerial decisions on performance.  While links from decisions to outcomes can be convincingly identified in individual, specific circumstances, trying to do the same with a larger set of decisions and outcomes can be problematic.  In a situation with a substantive sample, relying on standard metrics such as financial statements, the link to particular outcomes can be difficult to establish.  Especially when trying to choose metrics not already correlated in some way.


Our research is generally in the field of knowledge management (KM).  The discipline has an extensive history of attempts to link better management of knowledge and other intangibles to performance outcomes.  While the performance outcomes are fairly straightforward (financial success such as profitability or return on investment, innovation success, etc.), the metric for knowledge management is harder to pin down.  The literature shows dozens of approaches.  Some are for single firms or a small sample, but even if focusing only on metrics for a larger sample of firms, disagreement exists. 


But even when a KM metric is chosen for a larger sample, there is often a direct tie between input and output measures, making correlation difficult to establish, let alone causation.  If profitability is part of both the input and output measures, of course they are correlated.  The most advanced statistical techniques don’t establish anything more than what the researcher should already know from a cursory look at the logic of the study and the specifics of the variables.  As a result, whether KM actually results in better organizational performance is still an open and much debated question in the field.


One research approach that can get around this issue is the natural experiment.  Pioneered by Nobel winner David Card and others, the natural experiment uses two measurements with a clear, consequential event separating the two—minimizing the impact of samples that might otherwise be connected contemporaneously.  The event, such as a change in law, provides a change in circumstances delinking the measures.  Whether structured as a pre/post experiment (measuring before or after the intervening event) or control/experimental group experiment (one group exposed to the event, the other not), the outcome can provide convincing results and can even suggest causation.


This paper will cover our work with natural experiments in relation to KM and performance outcomes for merger and acquisition (M&A) activity.  Can KM metrics before an M&A event predict the success of the event?  Success can be defined by the change in financial performance after the event takes place. 

Author Biographies

Scott Erickson, Ithaca College

Scott is Charles A. Dana Professor and Chair of Marketing in the School of Business at Ithaca College, Ithaca, NY where he has also served as Department Chair and Interim Associate Dean.    He has published widely on big data, intellectual capital, and business analytics.  His latest book is Marketing Essentials for a Digital World from Flip Publishing.

Helen Rothberg, Marist College

Helen is Professor of Strategy in the School of Management at Marist College, Poughkeepsie, NY.  She has published extensively on topics including competitive intelligence and knowledge management.  Her latest book, The Perfect Mix: Everything I Know About Leadership I Learned as a Bartender was published by Simon & Schuster.