The nature of information and overconfidence on venture capitalists' decision making

Venture capitalists (VCs) are considered experts in identifying high potent ial new ventures-gazelles. Thus, the VC decision process has received treme ndous attention within the entrepreneurship literature. Yet, most studies o n VC decision-making focus on which decision criteria are central to select ing gazelles. Although informative, the majority of these studies has negle cted cognitive differences in how VCs make decisions. This is surprising co nsidering the influence cognitive differences are likely to have on the exp loitation of an opportunity as well as its influence on likely success. The current study investigates whether VCs are overconfident, as well as the f actors surrounding the decision that lend to overconfidence. Overconfidence describes the tendency to overestimate the likely occurence of a set of events. Overconfident people make probability judgments that ar e more extreme than they should, given the evidence and their knowledge. In the case of the new venture investment decision, overconfident VCs may ove restimate the likelihood that a funded company will succeed. The results of the current study indicate that VCs are indeed overconfident (96% of the 51 participating VCs exhibited significant overconfidence) and that overconfidence negatively affects VC decision accuracy (the correlati on between overconfidence and accuracy was 0.70). The level of overconfiden ce overconfidence depended upon the amount of information, the type of info rmation, and whether the VC strongly believes the venture will succeed ol f ail. As more information becomes available, people tend to believe they will mak e better decisions; they are making a "more informed decision." More inform ation ideally should enable the VCs to assess any potential pitfalls. Howev er, additional information makes the decision more complex. Information fac tors may contradict and relate to other information in unexpected ways. Eve n if more information is available, people usually don't analyze all of it (even though they believe they do). Thus, more information creates greater confidence but it also leads to lower decision accuracy. The type of information that is available also impacts overconfidence and d ecision accuracy. VCs are intuitive decision? makers. When people are famil iar with a decision and the structure of the information surrounding that d ecision, they resort to automatic information processing. On the other hand , if information surrounding the decision is structured in an unfamiliar wa y, people need to decipher what each piece of information means and how tha t impacts their overall accuracy. In the case of expert VCs, that means the y must deviate from their intuitive style. It seems that forcing them outsi de their "comfort zone" has a negative effect on their confidence and has a n even greater effect (negative) on their accuracy. There is evidence of an "availability bias" in VC decision-making; VCs rely on how well the current decision matches past successful of failed investm ents. VCs are overconfident in, their prediction of venture success when th ey predict a very high level of success. VCs are also overconfident in thei r prediction of venture failure when they predict a very low likelihood of success. This high level of overconfidence in success predictions (or failu re predictions) may encourage the VC to limit information search and fund a lower potential investment (or prematurely reject a stronger potential inv estment). Although overconfidence in itself does not necessarily lend to a wrong deci sion, the bias is likely to inhibit learning and improving the decision pro cess. Overconfident VCs may not fully consider all relevant information, no r search for additional information to improve their decision. Moreover, th e natural tendency for people to recall past successes rather than failures may mean that VCs will make the same mistakes again. VCs can take simple s teps to reduce the effect of overconfidence, including counterfactual think ing (i.e., imaging scenarios where current assumptions might not hold), for mally recording how past decisions were made at the time of the decision (v ersus trying to recall how that decision was made from memory), and using a ctuarial decision aids that decompose decisions into core components. Reduc ing overconfidence may lead to stronger decisions. It is hoped that this st udy illustrates the power of cognitive theories for understanding VC decisi on-making. (C) 2000 Elsevier Science Inc.