This paper studies the implications of individuals' knowledge of future job loss for the existence of an unemployment insurance (UI) market. Learning about job loss leads to consumption decreases and spousal labor supply increases. This suggests existing willingness to pay estimates for UI understate its value. But it yields new estimation methodologies that account for and exploit responses to learning about future job loss. Although the new willingness to pay estimates exceed previous estimates, I estimate much larger frictions imposed by private information. This suggests privately traded UI policies would be too adversely selected to be profitable, at any price.