will put some type of limit on duration, credit quality, sector allocations, issuer allocations, and currency exposure. Recently, some investors have given more flexibility to their managers on individual types of risks and have instead provided the manager a constraint on total volatility relative to the benchmark in the form of a tracking error target. Step 3 Determine the permissible active investment strategies. Once the constraints are known, the manager can determine what investment strategies can be utilized for the portfolio. For example, if the client does not allow any currency exposure, then obviously the currency strategy would not be implemented for that portfolio. Similarly, if a mandate does not allow for securities rated below investment grade, then the high-yield sector would not be part of the sector allocation strategy and of course there would be no security selection strategy within the high-yield sector. Step 4 Determine an appropriate maximum amount of tracking error for each available strategy. The combination of the benchmark and the constraints will determine how much the risk exposure of the portfolio could deviate from the benchmark. Also, the manager may impose his or her own constraint on risk to a strategy based on capacity issues and prudence. The maximum tracking error is determined by multiplying the average deviation of the particular risk times the estimated volatility of that risk. For example, let's say that a particular U.S. fixed income mandate has a one-year limit on the duration deviation from the benchmark. The volatility of U.S. interest rates is about 1 percent peryear. If the manager always had the maximum bet (i.e., either long one year or short one year) at all times, then the average tracking error would be about 100 basis points (1 year times 1 percent). However, for most strategies, the information ratios of the strategies will be higher if the managers take large positions when they have the most conviction and smaller positions when they have a view but with less conviction. Therefore, in this case, we would assign a constraint on the duration strategy of about 70 bps, which would correspond to an average 0.7 year exposure and would allow for the maximum one-year bet at times of maximum conviction. We go through a similar exercise for each investment strategy, examining what size of active positioning (on average) can be achieved and then estimating how much tracking error would result from that average position. Step 5 Estimate the excess return per unit of risk for each strategy as well as the correlation of excess returns between strategies. The estimation of the amount of added return for each basis point of tracking error, or information ratio, is basically a measure of how good the manager believes the strategy will be going forward. Every strategy should have a positive information ratio, or else it would not make sense to include it in an active investment process. Determining the information ratio of a strategy is not an easy task and will always have some uncertainty around it. However, a manager can make an educated estimate by combining past results of actual or back-tested performance of the strategy with a forward-looking view that incorporates the manager's assessment of the current opportunities in the market. Generally, our view is that security selection strategies can achieve the highest information ratios, followed by sector strategies and then followed by