A path to performance enlightenment

Investment performance is at the core of every investment organisation. After all, investments are made for one reason only: to generate performance. And indeed, performance measurers often find themselves at the centre of their organisations, providing performance information to nearly every functional area. With that many different consumers of performance information, how do you organise a performance measurement function to effectively and efficiently satisfy the requirements of these consumers?

This is a significant question and, of course, there is no single correct answer. The appropriate organisational structure depends on the vision for the function and the environment in which it operates. Instead of prescribing a specific organisation, I will therefore aim to identify the principles that should guide the organisation of the performance function.

Purpose

Any analysis of the optimal organisation of the performance function must start with an understanding of the purpose of performance measurement. Clearly, one of its functions is to produce a factual performance record. But that cannot be all. If it were, then there is no justification to distinguish the performance function from that of portfolio administration.

The deeper purpose is, of course, to understand why performance is what it is and what it means — in short, to create knowledge — and to transform this knowledge into information that supports the organisation in making the right decisions at the right time. If a portfolio underperforms, is that because the portfolio manager is unlucky or unskilled? Are there opportunities to improve performance by changing the investment process? When should an underperforming product be terminated? How can we manage the impact of performance on our bottom line? How can we support the claims we make to prospects about our products? All these are fundamental business decisions in an investment management environment. The purpose of the performance function is to inform these decisions.

Requirements for performance information

If the purpose of the performance function is to inform business and investment decisions, then what requirements must this information meet? Broadly speaking, three criteria can be identified:

· Relevance

Information only informs a decision if it is relevant to the decision at hand. A decision to modify the investment process within a particular portfolio requires entirely different performance information than one to mitigate the impact of all portfolios on the bottom line. Relevance includes timeliness.

· Completeness

Even if the information is relevant, it can only be used to support a decision if it is complete. This means that performance information should cover not only return, but also risk.

· Credibility

Even if the information is relevant and complete, it will only influence a decision if it is credible. Credibility results from four qualities: independence, consistency, transparency, and correctness. Independence ensures the information is unbiased. Consistency means that the same methodology is used to produce a piece of data regardless of when and for which portfolio it is produced.

Transparency means the recipient of the information is able to critically assess the methodology used to produce the information. Finally, correctness means that performance information is derived using the agreed methodology from correct input data.

So far, we have established that a) the purpose of the performance function is to inform decision-making and b) this requires that the performance information is relevant, complete and credible. The next step is to determine the implications for the organisation of the performance function.

Principles for organisation

Before we identify the organisational principles, we must add one important universal constraint: whatever organisational model is chosen, it must be efficient. Taking this into account, I believe the organisation of the performance function should be based on six key principles.

A shared service centre

The first question is whether to organise the performance function separately or to fragment it by embedding it within each function requiring performance information.

One could indeed argue an embedded performance analyst will be much better placed to provide relevant and complete information compared with an analyst in a separate performance function. The fragmented model fails, however, in terms of credibility.

Independence would be impaired by the fact that portfolio management and sales have a conflict of interest in calculating their own performance. Consistency would be lost because the same performance information would be calculated in different functions using different input data and methodology. Transparency would also become impossible as nobody would be able to explain why there is so much conflicting performance information floating through the organisation. Finally, correctness would probably be jeopardised since individual functions lack the economies of scale to establish automated, quality controlled processes to produce performance information correctly.

Perhaps the biggest weakness of the fragmented model, however, lies in its inefficiency. The costs of the fragmented model include not only substantial direct costs, such as duplicate data and system costs, but especially the prohibitive indirect costs that result from the time and effort required throughout the organisation to understand and explain the inconsistent performance information. Nevertheless, a surprising number of investment organisations underestimate these costs, resulting in structural underinvestment in a central performance function. It should be organised as a shared service centre on grounds of credibility and efficiency.

Ensure independence

As mentioned above, one of the factors determining credibility of information is whether the information has been provided independently. One the main ways to achieve independence is to ensure that the performance function does not report into a function with a conflict of interest, such as the investment manufacturing, sales or client servicing. Instead, it should report into another function such as operations, finance or risk management.

Of course, a client of an investment manager could argue that any performance figure produced by that manager, including that manager’s ‘independent’ performance function is not independent. After all, those employed in it have an interest in the success of the manager. A common solution for many institutional clients — and some investment managers — is to outsource performance measurement to an independent third party, typically their custodian or administrator, to strengthen independence. By removing the performance function even further from the ultimate consumers of this information, however, makes it even more difficult to ensure the relevance and completeness of it.

Essentially, there is trade-off between independence on the one hand and relevance and completeness on the other. I, personally, value relevance and completeness so highly, that I prefer to have performance analysts sit among portfolio managers. While that may come at the cost of a perceived of reduction of independence, I believe that independence can be adequately guaranteed by establishing an appropriate reporting line for the function and other supplementary measures.

Integrate risk and return

Historically, performance and risk measurement have been organised as separate disciplines, often reporting into separate functional lines. More recently, several organisations have combined the two disciplines into a single function and I believe that it is the right approach for several reasons.

The first reason is completeness: nearly all decisions require a combination of both risk and return information. To understand this, it is important to realise that the distribution of the future portfolio return depends on two parameters:

· the risk in today’s portfolio;

· the skill of the portfolio manager.

To understand the risk of today’s portfolio, ex ante risk will tell you all you need to know. To understand skill, however, one must measure the risk taken by the portfolio manager during each historical period as well as the return that risk generated in each period. Contrary to popular belief, ex post risk cannot tell us anything about either the risk in today’s portfolio or the risk level in the portfolio during different periods in the past. To measure skill, we must therefore combine historical ex ante risk with returns. That is, we must combine risk measurement and performance measurement.

The second reason is consistency of the information. If one must be able to measure the ex ante risk and the return of every single investment decision consistently, then one must use the same data and methods to decompose, or attribute, ex ante risk and return consistently.

The third reason is efficiency. To measure risk, one requires data on today’s positions as well as data on the current and historical characteristics of securities. Performance measurement requires exactly the same data plus historical positions data and, for transactions-based analysis, historical transactions data as well. So by combining risk and performance measurement one can realise a significant synergy in terms of data licensing and management costs. True, an important difference between the data requirements for risk and performance measurement is timeliness. In order to be of predictive value, risk data needs to be updated every day before the start of business, while performance measurement is typically allowed to take several days after month-end to collect its data. But this only strengthens the case for combining risk and performance measurement: if one has to deliver data for risk measurement at a high frequency and within tight timeframes, then why not improve the timeliness and frequency of performance information at the same time?

Establish an advisory unit

Earlier I argued that the performance function should be organised as a shared service centre and pointed out that doing so comes at a cost of reduced relevance and completeness of the information. This is because performance analysts in a shared service centre are further removed from the functions that consume performance information. Left unchecked, such a performance function will tend to lack an understanding of the decisions that need to be made in the organisation and the type of performance information that can be provided in support of these decisions. As a result, the performance function is likely to become a reactive production unit that aims to satisfy consumers of performance information by providing them with what these consumers ask for, rather than what they need. Ironically, this may leave the decision-makers frustrated with the failure of the performance function to meet their needs. It is not unusual for the end result of this to be that these decision-makers set up their own embedded performance function, thereby undermining the benefits of the shared service performance function.

To avoid this scenario, measures must be taken in the organisation of the performance function to reduce the distance between the performance function and its clients. One of the most effective ways of doing this is to establish an explicit advisory component within the performance function focused on understanding the decisions that need to be made in the organisation and the way in which performance information can help to inform these decisions. Depending on the size and complexity of the organisation, the advisory component can even be subdivided into advisory teams that each focus on the needs of different stakeholders.

When creating this advisory component, it is probably best to separate it from the production component as much as possible in order to create focus for each on its core competency: producing data versus informing decisions. Of course, there are risks to such separation. One is that separation may well lead to disconnection of the two teams to the extent that the production component no longer serves the advisory component. This can be avoided, however, by having both teams report to the same manager. Another perceived risk is that enforced specialisation will reduce job satisfaction. This consequence, while possible, should not be exaggerated. Many staff will actually prefer to specialise as it suits their skills and ambitions better. In addition, this risk can be mitigated by putting in place job rotation schemes and/or career paths.

The advisory focus can be strengthened further by thinking twice before organising the performance measurement function as part of operations. Doing so tends to create a culture in the performance function obsessed with production at the expense of analysing the performance data to extract meaning and value. Of course, both production and advisory are important, but the latter should definitely be in the lead. A reporting line into, for example, risk management is more likely to deliver the right balance.

Standardise production

Regardless of whether production is separated from advisory, it is crucial to standardise production to ensure consistency of all performance data. The performance of every portfolio should be measured only once and, unless there are good reasons, the methodology and therefore the systems used across different portfolios should be the same.

Standardisation requires that production should take place on a single, integrated data and systems platform. But that is not synonymous with centralisation. Modern technology enables staff in different locations to work as a single team on a single platform. Of course, there are management and control benefits to centralisation, but the benefits of creating local buy-in in a geographically diverse organisation should not be discounted. Even if a centralised model is the end goal, a transition from a fragmented decentralised model to a standardised decentralised model can be a useful first step towards this end goal.

Standardisation is, of course, more challenging in investment organisations that have multiple portfolio administration platforms.

Even then, however, it can and should be pursued by implementing a single central database to collect and integrate all input data. On balance, this solution will still be cheaper than having multiple data and systems platforms by reducing the number of interfaces and systems. Of course, a prerequisite to global standardisation is that the local offices are willing to cooperate and share. In some cases, governance issues and cultural issues may complicate this.

Hire the right people

Last, but certainly not least, an organisational structure will only work if it is staffed by people with the required skills. If the performance organisation is divided into production and advisory components, then this has clear implications for those skills.

The skills required in a team that manages a standardised production platform are quite different from those required in a ‘cottage industry’ approach to production. While advanced spreadsheet skills will continue to be critical, a standardised production platform requires process engineering and quality control skills, extensive knowledge of data modelling and database technology, and a profound understanding of financial data and the quantitative models used in risk and performance analytics. Conversely, the demand for manual accuracy will be drastically decline.

Within the advisory team, on the other hand, staff should essentially have the same skills as quantitative portfolio managers — strong quantitative analytical skills to understand and model the sources of risk and return of different instrument types and investment strategies. In addition, and unlike portfolio managers, they need an ability to communicate advanced quantitative concepts in layman’s terms and a talent to see how to risk and return affect the entire organisation are required.

We have identified six key principles to help organise the performance function. These principles were derived from an analysis of the purpose of the function and the requirements that this purpose implies for performance information.

Many investment organisations have already implemented some or all of the principles proposed in this paper. This is good news, because it means performance measurement will continue to professionalise and make further progress down the path of realising its value-adding potential. In the process, the operational focus will make way for a focus on extracting value from the vast amounts of performance data produced. While this is a positive change for the profession as a whole, it will, like any change, produce winners and losers.

By clarifying what the future may hold for performance functions, I hope to stimulate the potential losers to adapt and the potential winners to send me their CV.

Subscribe

Columns

Features

Profile

Workshop

Past Issues

Advertise

Editorial Board

Resources

About Us

A path to performance enlightenment

Client Login

Current Issue

Newsletter

Search Site

Events