Multiplicative growth processes that are subject to random shocks often have a skewed distribution of outcomes. A simple laboratory experiment shows that participants either strongly underestimate skewness or ignore it completely. The participants' choices reveal bounds on their subjective medians of a financial asset's price that is subject to stochastic growth. The observed bias in expectations is irrespective to risk preferences and fairly robust to feedback. It is consistent with a behavioral model in which geometric growth is confused with linear growth. The bias is a possible explanation of investors' misunderstandings of real-world financial products like leveraged ETFs.