Demand lotteries, abandonment options and the decision to start R&D and process innovation

Abstract

(ProQuest: ... denotes formulae omitted.)1. IntroductionThe decision to start a Research and Development (RD Eiser & Nadiri, 1968). The second category includes market structure which is determined by market power (Levin, Cohen, & Mowery, 1985), competition (Grenadier, 2000; Huisman, Kort, Pawlina, & Thijssen, 2005) and entry pressure (Etro, 2006; Acemoglu, 2008; Aghion, Blundell, Griffith, Howitt, & Prantl, 2009), and general industry conditions (Acs & Audretsch, 1987; Dorfman, 1987). The third category includes different types of options such as a timing option (Dixit & Pindynck, 1994; Trigeorgis, 1996) and an abandonment option (Myers & Majd, 1990; Berger, Ofek, & Swary, 1996), and different types of uncertainty such as input cost uncertainty (Pindynck, 1993), technical uncertainty (Pindyck, 1993) and market uncertainty (Tyagi, 2006).Market uncertainty is of particular interest for our analysis. Market uncertainty is related to the future value of the innovation which is strongly determined by market demand. For example, if firms have successfully developed the new product or production technology, uncertainty still exists about market acceptance and hence innovation rents. In general, market uncertainty reduces R&D investments. However, Czarnitzki and Toole (2009, 2011) show that the negative effect is mitigated when firms receive R&D subsidies or patent their innovations.In this paper, R&D leads to a cost-reducing process innovation. We build on the model of Lukach, Kort, & Plasmans (2007) that contains many aspects of real-life R&D decisions within a net present value (NPV) framework. Besides entry threat, Bertrand competition and multi-stage R&D with an abandonment option, our model differs from Lukach et al. (2007) as it includes demand (market) uncertainty rather than supply (technical) uncertainty. We deduct testable hypotheses on the basis of which we empirically analyze the non-traditional factors driving the decision to start an R&D project. The uniqueness of our data lies in the availability of proxies for demand uncertainty, the abandonment option as well as perceived entry threat.We model an R&D project as a multi-stage game where the incumbent must decide at the first stage to start and at the second stage to continue R&D. The decision to start is influenced by demand uncertainty, modelled as a lottery between a proportional increase (=good state) and decrease (=bad state) in demand. A lottery becomes more divergent when the difference between the outcomes of the lottery increases. We derive under which lottery probabilities more divergent demand lotteries positively or negatively affect the decision to start R&D. For empirical testing, we use data from the fourth Community Innovation Survey (CIS IV) in Germany for about 2600 firms to explain the decision to start R&D. Our main results, strongly confirming our model predictions, are that for firms facing lotteries where the good state is more likely to prevail a 10% increase in the degree of divergence of the demand lottery increases the likelihood of undertaking R&D by 1. …