Leadership or luck? Randomization inference for leader effects in politics, business, and sports

Anecdotal evidence suggests that some leaders are more effective than others but observed differences in outcomes between leaders could be attributable to chance variation. To solve this inferential problem, we develop a quantitative test of leader effects that provides more reliable inferences than previous strategies, and we implement the test in the settings of politics, business, and sports. We find significant effects of political leaders, particularly in nondemocracies. We find little evidence that chief executive officers influence the performance of their firms. In addition, we find clear evidence that sports coaches matter for a wide range of outcomes in football, basketball, baseball, and hockey.