It is well documented that people overbid in second price auctions (SPAs). Yet, this fact is conveniently ignored when eliciting willingness-to-pay (WTP) for market goods. We propose a simple design that not only tests the external validity of SPA bids, but also suggests a more accurate method of eliciting WTP in SPAs. Following the SPA, participants were offered a randomly chosen price, from the range of retail prices in actual markets, at which they can purchase any amount of the good in an onsite secondary market. The design links overbidding and underbidding behavior to violations of the weak axiom of revealed preferences (WARP). We find robust evidence that the dominance of overbidding over underbidding in SPAs leads to an upward bias in the WTP estimates. While this can compromise market good valuations by inflating the perceived value of products, our design enables utilization of Kotlarski’s identity to recover the distribution of the unobserved true valuations.