IVLldU T < w i\j *_i*kj* i^aigv *^jaij Active   Historical IR Constant IR Risk Budget U.S. Large Cap Equity 1.00 1.00 21.7 U.S. Small Cap Equity 0.47 1.05 24.0 International Equity 0.55 1.19 30.6 Emerging Markets Equity 0.14 0.17 0.7 Core+ Fixed Income 46.59 0.27 1.7 High Yield 0.04 0.12 0.3 Overlay 0.48 0.98 21.0 our view is that investors gain more insight into the investment process by focusing on factors that would set their relative confidence levels. Here is a partial list of factors that could guide setting relative confidence levels for active risk at the asset class level. II Is the source of the historical alpha a one-time event that all market participants shared? If the historical alpha represents a one-time event that is not likely to repeat itself, then the confidence level should be lowered relative to other sources of active risk. Consequently, less of the active risk budget would be allocated to these strategies. An example of such a phenomenon is the historical performance of international managers relative to EAFE, where most managers were underweight Japan. II Is the source of the historical alpha a function of benchmark anomalies? Poor benchmark construction (e.g., benchmarks where index arbitrage is difficult) give rise to an embedded ability for active managers to add value. To the extent that the investor thought that benchmark construction was unlikely to change, a higher relative confidence could be assigned to the active strategies, and more of the active risk budget allocated to them. Two examples of such sources of alpha include the EAFE benchmark and the Russell 2000 benchmark. II Does the source of the historical alpha represent a structural inefficiency? Structural inefficiencies can occur when one for more) market participant has an objective function that is other than mean-variance optimization. In these cases, mean-variance optimizers have the ability to generate alpha. Consequently, relatively more confidence could be placed in these strategies, and more of the risk budget allocated to them. An example of a structural inefficiency is the currency market, where central banks have macroeco-nomic policy objectives that cannot be easily represented in a mean-variance framework. So far, we've focused on the allocation of portfolio risk between active and asset class risk, and on allocating active risk across portfolios of active strategies (e.g., a portfolio of active U.S. Large Cap managers versus a portfolio of active U.S. Small Cap managers. The same analysis can be easily extended to the manager-specific level. In that case, we would be calculating the allocation of ac-