“Exponential growth bias” – the tendency to underestimate exponential growth – has been shown to affect how much people save and borrow. I develop a simple theoretical framework to show that this bias, through its interaction with individual risk preferences, also affects how people save and borrow. Relatively risk-tolerant consumers will choose inefficiently safe investments, appearing more risk-averse than they are. In contrast, relatively risk-averse consumers will choose inefficiently risky investments, appearing less risk-averse than they are. I present survey data to support this hypothesis. The model introduces a new perspective on some apparent anomalies in intertemporal choice and raises additional considerations for consumer protection and financial literacy programs.