Second-order uncertainty

The financial crisis has drummed home the dangers of basing analysis on unreliable data. Despite its amorphous character, risk managers must begin to increase their focus on second-order uncertainty, argues David Rowe

Those not trained in statistics often find the expression 'stable random process' quite puzzling. "How", they ask, "can a process be both stable and random?" The answer, of course, is that any one draw is random and, hence, unknowable in advance. If the random process is stable, however, then sizeable samples will exhibit broadly similar characteristics, such as the mean, the dispersion (standard