the difference between managed and benchmark portfolio weights. And we know that the larger (smaller) the active weights are in absolute value, the larger (smaller) a portfolio's tracking error is. Since-holding all other things equal-an increase in cash reduces a portfolio's managed weights in equities, the impact on tracking error depends on the original values of the active weights and whether the corresponding stocks are positive or negative contributors to tracking error. Take a particular stock, for example, whose active weight, prior to any change in cash, is close to zero and is a positive contributor to tracking error. Holding all other things equal, if we were to add enough cash to the portfolio, then that stock's active weight would become negative, and thus add to tracking error. If, on the other hand, the stock starts off with a small overweight, then it would be possible to add enough cash to the portfolio to make this stock's active weight zero. In this case (holding all things equal) the stock's contribution to the portfolio's tracking error would decrease. In summary, the impact that changes in cash have on a portfolio's tracking error depends on the original (i.e., prior to adding cash) values of the active weights and the stocks' contributions to tracking error. Broadly, if most of a portfolio's tracking error is coming from overweights, then an increase in cash will decrease (increase) the portfolio's tracking error. Similarly, if most of a portfolio's tracking error is due to underweights, then an increase (a decrease) in cash will reduce the managed portfolio weights and increase the absolute value of the active weights even more, thereby leading to higher tracking error. An Example We conclude the section with an example of a risk decomposition as provided by Goldman Sachs' PACE system. Figure 20.4 shows the first page of a PACE Risk Monitor Summary report for a hypothetical portfolio as of December 31, 2001. The report shows that the portfolio's value on this day is $488,481,650. Its benchmark portfolio (i.e., the portfolio that is used in calculating tracking error) is the S&P 500. This portfolio currently contains 106 assets and 2.63 percent of its total value is in cash. Looking at the left-hand side of the top box in this report, we see that its annualized predicted tracking error (US Predicted TE) is 4.81 percent. The target TE for this portfolio is 4.50 percent, so this portfolio is running slightly above target. For comparison, this report shows a predicted tracking error from an alternative model (US RMG Daily TE). According to this model, the predicted tracking error is 4.84 percent. Reading down, we come to a box called "Risk Decomposition." This information shows that about 49.94 percent of the risk is specific and the remaining (50.06 percent) is coming from factors. In particular, 23.04 percent and 27.02 percent of the total tracking error is attributed to industries and investment styles, respectively. Continuing down the page, we come to a table that shows contribution to risk by asset (stock). These contributions are based on the RCTE calculations. For example, Intel is the highest contributing stock to tracking error. It consumes 7.14 percent of the overall risk budget and has an underweight (versus the benchmark) of 114 basis points. Similarly, Oracle is also an underweight (its ac-