In this paper, we analyze the social value of accounting objectivity in maintaining financial stability. Building on an early influential accounting study by Ijiri and Jaedicke (1966), we operationalize two informational properties, accuracy (free of collective bias) and objectivity (degree of consensus), in a correlated information structure and embed them into a model of runs on financial institutions. We show that when compared with the accuracy property, the objectivity property exhibits an advantage in mitigating inefficient panic-based runs. In fact, it is possible that improving objectivity discourages such runs, whereas improving accuracy encourages them. Our model also sheds light on the design of optimal accounting systems to enhance objectivity. We find that to generate a more objective accounting report, accounting systems should be designed to be less vulnerable to intentional managerial intervention.