The top-quartile structured manager had an active return of 92 basis points, while the bottom-quar-tile manager had an active return of negative 4 basis points. Consistent with the differences in risk taking, the top-quartile traditional manager had an active return of 201 basis points, while the bottom-quartile manager underperformed the benchmark by 120 basis points. Thus, the historical performance record seems to indicate that, on average, structured and traditional managers outperformed by roughly the same amount. However, manager selection is much more important for traditional managers because the spread in results is much wider. Historical returns alone provide an incomplete comparison between manager styles; to complete the picture, we should also look at risk. For this reason, Table 14.1 also includes a summary of the distribution of historical tracking errors for structured and traditional managers. Given that we intentionally classified managers using realized tracking errors, we shouldn't be surprised that the tracking errors for structured managers are lower than those for traditional managers. For example, the median tracking errors are 221 and 769 basis points, respectively, for the structured and traditional managers. At the extremes, the top-quartile structured manager had an historical tracking error of 266 basis points, while the bottom-quartile manager had a realized tracking error of 147 basis points. By contrast, the top-quartile traditional manager had a tracking error of 971 basis points, while the bottom-quartile manager had a tracking error of 619 basis points. Thus, consistent with the way we've defined our sample, investors were likely to see higher realized active risk levels from their traditional managers than from their structured managers. A useful way to assess the risk/reward trade-off is with the information ratio, defined as active return per unit of active risk (or active return divided by tracking error). Table 14.1 also shows information ratios. These figures are perhaps the most interesting, as they suggest significant differences between these active management styles. That is, the historical information ratios for structured managers are higher than those for traditional managers at all skill levels. For example, the median structured manager had an information ratio of 0.28, while the median traditional manager had a realized information ratio of 0.07. Table 14.1 also shows that the dispersion of information ratios was more pronounced for traditional managers. The top-quartile information ratio for traditional managers was almost four times greater than the median. For structured managers, the top-quartile information ratio is only 57 percent higher than the median. Taken together, these figures suggest that structured managers added more active return per unit of active risk,3 and further that manager selection would have been incredibly important in developing a portfolio of traditional managers. Table 14.1 also explores the level of pairwise correlations between active returns. For the most part, these figures show no difference by active management style. The median correlation between structured managers was 0.08, while for 3These results are consistent with the study of mutual funds by Brown and Harlow (2002), which shows a clear connection between consistency of investment style, active risk levels, and persistence of performance. Generally, a high level of consistency corresponds to lower active risk levels and more persistent benchmark outperformance.