Merginal contribution to risk

I do not understand the concept of marginal contribution to to tracking risk. Why is there emphasis on that? on page 194 of volumn 6 exhibit 7 and also BB 6 talks about this. Not clear to me.

For portfolios that are managed against benchmarks, a common measure of risk is tracking risk (TR), also often called tracking error. The objective of an attribution model for a benchmark-relative portfolio is to quantify the contribution of active decisions to TR. For bottom-up benchmark-relative investment processes, each position’s marginal contribution to TR multiplied by its active weight gives the position’s contribution to TR. For benchmark-relative top-down investment processes, the active return is explained first by the allocation decisions. Risk attribution, accordingly, will identify the total contribution of allocation and selection to TR.

For absolute mandates, the risk of the portfolio is explained by exposures to the market, size and style factors, and the specific risk due to stock selections. The attribution model quantifies the contribution of each exposure and of specific risk. Suppose that the manager follows an absolute bottom-up process where the measure of risk is the volatility (standard deviation) of returns. In this case, we want to measure the contribution of selection decisions to overall portfolio risk. To do this, we need to know the marginal contribution of each asset to the portfolio risk—the increase or decrease in the portfolio standard deviation due to a slight increase in the holding of that asset. If we know the marginal contribution of a security to absolute portfolio risk, we can then calculate the overall risk contribution of the portfolio manager’s selection decisions.

In all cases, risk attribution explains only where risk was introduced into the portfolio. It needs to be combined with return attribution to understand the full impact of those decisions. For example, if a manager has added to excess return through asset allocation (e.g., positive return attribution allocation effect), we use risk attribution to understand whether those allocation decisions introduced additional risk. As such, risk attribution complements the return attribution by evaluating the risk consequences of the investment decisions.

If I understand your question correctly, you are wondering why a manager would focus on marginal contribution to tracking error.

The reason is very simple: when a manager is tracked against a benchmark (and its risk limit is tracking error), whenever he needs to reallocate the portfolio (for any reason), the marginal contribution of a position to tracking error is what he is concerned about, in order to keep tracking error aligned to its target value.
Pratical example: if Position A has marginal contribution of 5% and position B has 1% and managers wants to lower tracking error, he will probably sell position A or buy position B.

Hope this helps.