The tournament phenomenon suggests that fund managers with above-the-median first-half performance tend to decrease the second-half risk of their investments, and vice versa. The prevailing interpretation of agency theory argues that managers compete in an annual tournament for capital inflows. However, behavioral economics provides an alternative explanation for these risk preferences Fund managers who are below their benchmarks are risk seeking (and vice versa) because they cannot bear finishing below the median. This is independent of monetary incentives. The authors' experiment removes all capital inflow incentives to eliminate the foundation of the agency theory explanation. Yet, the tournament phenomenon persists Non-risk-neutral subjects increase risk after bad relative performance, but decrease risk after bad absolute performance. Whereas agency theory would label the authors' observations as anomalies, behavioral economics explains the persistence of the tournament phenomenon quite well - by viewing it as an internal conflict, rather than an agency conflict. Therefore, the authors' findings carry implications for restructuring incentives in tournament situations.