The treasury function has been established as a distinct financial discipline in all top companies for many years. More recently, however, FDs have increasingly been questioning the size, and hence cost, of the function.

A lot of the top 100 companies have simplified the function and a hard-nosed approach to measuring treasury's contribution has developed. There are five reasons for this:

•   we are operating in an extreme cost-reduction business environment;

•   interest rates are so low that top management does not focus on interest rate, debt or cash management to quite the same extent as when rates were much higher;

•   many top corporate treasury departments grew too big in the 80s and 90s;

•   technology is simplifying the function;

•   many of the benefits are achieved on initial set up and do not need an ongoing organisation of the same size to continue to realise the savings.

Against this background it is no surprise that FDs are also becoming tougher on performance measurement.

What to measure?

For a treasury that acts like an in-house bank, meeting a treasury income target is a simple and in most cases accurate measure of performance.

The net earnings in the treasury management accounts will often simply reflect the difference between what the treasurer achieved and what the position would have been if treasury did not exist. This follows if the treasury gives arm's-length prices to the business units. These prices should be at least as good as those the business would have got if they had accessed the financial markets independently. Most of the benefits from having the central function will then be apparent from the treasury management accounts as income against which the costs can be set. Although treasury income is being measured, this is quite different from setting up the function as a pure profit centre, which might imply allowing view-taking on fx or interest rates.

For a pure cost centre treasury, which may not even be doing transactions in its own name but as agent for the businesses, a simple process of not exceeding a budgeted cost and outperformance against benchmarks is the normally preferred route.

Benchmarking rules

Setting benchmarks presents problems. Benchmarking regularly produces contentious and unproductive debate. The determination of instrument type, credit rating, maturity or liquidity characteristics of the benchmark should be as close as possible to the transactions that treasury is actually expecting to undertake in the markets.

The benchmarks should also reflect the company's treasury policy. For example, if an investment policy is to allow investment in counterparties rated single A, then this should be (or be a component of) the benchmark.

If the benchmark is the AAA rate then it will be easily beaten. A similar point arises with interest rate exposure such that if the risk management policy is to limit exposure to six months, then the benchmark should not be, say, the one-month rate. It will often be a good idea to set the benchmark at the mid-point of an allowed policy range. In the case of fx transaction risk, suppose the policy is that the next 12 months' cash flows should be 30% to 70% hedged, then there is some logic in setting the benchmark at 50%. This is on the basis that if the treasurer had a neutral view of the risk, then the policy mid-point would be the logical level of cover.

It is clear that benchmarks must be carefully crafted, but treasury yardsticks are similar to other management measures; these should also be simple to calculate and understand, and applied consistently.