Essays on predictive and non-predictive strategies: real and simulated experiments

The dissertation studies how adopting a more systematic approach to decision-making impacts decision's outcomes and compares it with the adoption of non-predictive strategies. Coherently with previous literature (Camuffo et al. 2020), I call the systematic approach "scientific" since managers and entrepreneurs are asked to act as scientist would do in a business context. Indeed, this approach consists in developing theories and logic connections about the mechanism underlying future outcomes and test them with tailored experiments. "Scientific" managers and entrepreneurs are then called to analyse test results and make decisions accordingly. As mentioned, this approach helps decision-makers to improve their predictive power by probing the future with theory-based experiments but remains explorative. Managers and entrepreneurs explore other alternative ideas on which they can theorize on. On the non-predictive side, research on effectuation (Sarasvathy 2001; Dew et al. 2009; Chandler et Al. 2011) has shown how managers and entrepreneurs can deal with uncertainty by adopting a decision-making approach aimed to control the future instead of predicting it. Effectual decision-makers select alternative ideas based on loss affordability experimentation, and flexibility. But, in this case, experiments are not guided by well-framed theories and are not part of a systematic process. Sarasvathy (2008) uses the metaphor of a patchwork quilt: managers and entrepreneurs see the business context as a table where all the pieces are there but must be assembled or even created as the future is unpredictable. With the aim to unfold the mechanism driving diferent termination rates of ideas from the adoption of these two approaches, I propose a model with the aim to predict empirical results. The model proposes a Bayesian framework, where decision-makers acquire costly information to improve the precision of signals. Based on these informative signals, they act accordingly. I expect scientic decision-makers to react promptly to very informative bad signals. While I expect effectual decision-makers to react less to bad signals since they weight less predictive information. This translates into higher rates of termination for scientic decision-makers than effectual decision-makers. Moreover, scientic decision-makers terminate earlier than effectual decision-makers. In the second chapter, I focus on the scientific approach solely. I provide evidence of the implications of a scientific approach to decision-making through four Randomized Control Trials, involving start-ups and small-medium firms (SMEs) across two countries, Italy and UK. The three main findings are that scientic decision-makers are more likely to terminate their idea in early stages, corfiming findings of the previous chapter. They pivot fewer times before committing to one or terminate the idea. They also perform better in terms of revenues. A model has been developed to explain empirical results. In the third chapter, I study a way to scale research findings by using a simulation game to replicate, to some extent, results of the previous two chapters about the scientic approach.