In a highly uncertain world, it is imperative to efficiently govern innovative buyer-supplier exchanges, which are often characterized by high asset specificity, the key indicator of governance choice in transaction economics (TCE). While TCE traditionally assumes asset specificity is easily observed and measured, recent research suggests managers’ perceptions of specificity typically drive these decisions instead. Additionally, since managers are boundedly rational, these perceptions are likely to be biased. We contribute by exploring the direction and potential of asset specificity perception biases, and how they may lead to inefficient governance. We first argue that false uniqueness and confirmation biases influence buyer and supplier asset specificity perceptions in opposite directions. We then propose that perceptions of human asset specificity are more likely to be biased than other types and that bias is more likely when the exchange features radical as opposed to incremental innovation. Finally, we examine the impact of these unconscious biases, arguing that buyer bias is more likely to be dominate in discrete exchanges, resulting in costly over-governance compared to that predicted from standard TCE theory. This paper expands our understanding of biased specificity perceptions and their impact on efficient governance and value capture for innovation partnerships.