8. Discussion and conclusions
8.1. Summary and discussion There are a number of competing theoretical arguments that explain why the relation between startups' innovativeness and their subsequent survival might be either positive or negative. On the one hand, innovativeness can increase the likelihood of survival if it enhances startups' market power, reduces the costs of production, or allows the creation of dynamic capabilities and absorptive capacity (Teece et al., 1997; Zahra and George, 2002). Younger firms may benefit disproportionately from the opportunities created by innovativeness due to their less rigid routines and greater flexibility (e.g., Brüderl and Schüssler, 1990; Christensen and Bower, 1996; Freeman et al., 1983; Hill and Rothaermel, 2003; Klepper and Simons, 1997; Lumpkin and Dess, 1996). On the other hand, we have argued that the stage of firm development and the stage and nature of innovativeness may moderate the association between innovativeness and firm survival adversely and can even turn it negative. An innovative startup is laden with excess liability of novelty and smallness, which reduce its chances of survival relative to its non-innovative counterparts. Startups' innovativeness may also limit their access to external finance (due to lack of collateral) and change their overall risk profile by making the distribution of revenue streams more variable and skewed (Scherer and Harhoff, 2000) and by delaying them in time. Whereas the available empirical literature mostly suggests that the relation between firms' innovativeness and their subsequent survival is positive, we find a negative association between innovativeness and survival. Our baseline estimations suggest that the survival probability of startups engaged in innovativeness is approximately 6–7 percentage points lower than that of other startups. This finding reinforces the result of Boyer and Blazy (2013), who documented that innovative French micro-enterprises established in 1998 were more likely to fail than their non-innovative counterparts. In our view, two aspects of our empirical approach are important for these findings: First, we employ an ex ante measure of innovativeness to account for the associated uncertainties and to avoid the survivorship bias of ideas (i.e., the consequences of focusing on a selected subset of firms that have previously innovated successfully). Second, we study the association between innovativeness and firm survival using a sample of startups, which allows us to account for the survivorship bias of firms (i.e., the consequences of market selection) to a reasonable extent.