Putting performance and attribution into context
People form habits and after a while they tend to lose focus on why they do things. They do them because they do them. Asset managers do performance measurement so they buy performance measurement. That all makes sense surely?
Well no, not really. For example, people don't want mortgages, but they do want a home, the benefit of the mortgage is that you can buy your home now. Similarly, asset managers don't want performance measurement; they want investors to give them money so they can earn fees and grow rich. Performance measurement is all about winning clients and then keeping them sweet.
Now you might say, that is completely obvious, so what? But in fact, thinking about it that way shows two things: 1. that performance measurement is essential if you want to run an asset management business and 2. That in an ideal world it should cost you as little as possible to do. It only exists because of the need to sell to and keep clients; it is not in itself an activity that is profitable for an asset manager. However, it is often viewed as a mysterious and dark art that requires extremely complex procedures to make it all work and, as a result, enormous resources are piled into it at some asset managers with no obvious benefit or improved result.
Indeed, in StatPro's experience (and this covers not just the 230 clients we have won but also the several hundred we nearly won) there sometimes seems to be a baffling desire to over-complicate what should be straightforward. There are of course plenty of examples of asset managers that are extremely efficient and we can learn from them. At a time of economic turmoil, it makes sense to re-evaluate what is really required of performance measurement and how best to deploy it so that asset managers can minimise its cost whilst maximising the benefits.
Back in the 80's the leading performance company was WM. They offered a great service where they centralised the portfolio information of most UK pension funds and calculated the performance of each portfolio once a quarter. This centralised service on a per portfolio basis meant that the price for this was not high for the pension funds. Because they measured over 2,000 portfolios they also had a valuable database which became a peer group that they could use to compare the performance of different fund managers. The service lost ground because the data frequency was not sufficient or detailed enough and asset managers decided they needed to get better information which WM could not provide them.
This meant that in the mid 90's a number of performance and attribution systems started to appear. The first was Frank Russell's RPA which proved very successful, but again failed to keep up with the development of the markets and so has withered significantly. FMC was an early provider with Sylvan a much more sophisticated system that became very popular. StatPro appeared on the scene in 2000 with our own offering we called SPA (StatPro Performance & Attribution). Between then and now, 30 or so companies have produced similar systems with varying degrees of success.
One consequence of all this competition has been for industry specialists in and out of asset managers to focus on the level of functionality of these systems and not question why they are needed in the first place. The early systems were simpler applications which had their databases in silos. The later ones have developed sophisticated data warehouses to handle the increasingly large demands for data management. But why is it necessary to go to such lengths to gather and manage such huge volumes of similar data at each asset manager? We all need electricity but we don't build generators and nowadays if we don't like our electricity supplier we can switch to a different one. It is time to re-think the way an asset manager accesses performance measurement.
For a start, how about asking the supplier to manage the IT platform? The last thing an asset manager needs is more servers and all the hassle of managing them. Secondly, all the data for indices can be managed by the supplier as well. Why should the asset manager have to unravel the complexities of the MSCI, FTSE, Russell, Lehman, Stoxx, Ibox, Euronext, DAX, Nikkei, JP Morgan, S&P and all the other indices? Much better to tick a box to choose which one you want and let someone else sort it out.
The third thing is slightly more complex but absolutely crucial. Amongst performance professionals there is consensus that Transaction Based Performance is much superior to Holdings Based Performance as of course it takes into account the trades that are made by the manager at the transaction price rather than the end of day price. However, the improved accuracy comes at a cost in terms of extra data required and the size of team needed to manage the vast data flows and it may be sensible for asset managers to consider whether that cost is worth it.
I wonder how many people in the industry stop and ask "Is transaction based performance actually accurate?" If they did they would surely say "No, it is not accurate, but it is a better approximation." It is only an approximation because for each investor their circumstances are different. They invest with the asset manager at different times and sell at different times. They have varying tax bills depending on their total income and domicile. The tax rules themselves vary over time and domicile. The valuations used to provide the daily valuations are snapshots from a market feed, usually end of day but possibly any time and in any case, if someone tried to liquidate the entire portfolio at once, they would surely not get the same prices. On top of that, conventions like the re-investment of dividends to show total performance are artificial. The result is that even if the portfolio has a single owner the performance is still approximate.
So if it is just an approximation why do it? Well the answer to that is that asset managers need to give a true view of their money management skills. The investor has to accept that his personal performance depends on him (calculated between when he puts the money in and when he takes it out), while his money is in, he will get a report that uses conventions like Year to Date performance, 1, 3, 6, 12 month performance and so on. The idea is to make comparisons between asset managers possible, but depending on the nature of the investor, it maybe that Holdings Based Performance is perfectly adequate for their needs. If that is the case, it will cost the asset manager (and thus the client) much less to provide. As long as the asset manager makes it clear in his report to his client the basis upon which the report is calculated, there should not be a problem.
This means that another aspect of any new system should be that it can do both Transaction Based Performance and Holdings Based Performance as this will allow the fund manager to reduce running costs. Because of this, the supplier also needs to provide corporate actions and end of day valuations. That way the asset manager will just need to provide the daily holdings for the relevant portfolios and the rest can be provided by the supplier. The beauty of such a service is that implementation can be calibrated to meet the required return on investment for the project.
At the moment, most projects face what seems like a fairly daunting choice of either going for the simpler route of Holdings Based Performance but thereby making it impossible to provide Transaction Based Performance for fussier clients, or having to swallow the cost of taking on Transaction Based Performance for all their portfolios and running the risk that the project will last months if not years before any benefit (like reporting to their clients) becomes apparent.
Using this Performance as a Service should also mean that the client can get lots of other things done by the supplier. Risk measurement, attribution, compliance, GIPS composites, client reporting and other services all work off the same basic data and it is uneconomic for every asset manager to try and do it all themselves. Inevitably, it will be the smaller companies that will see the benefit of leveraging the expertise and capacity of a supplier that can do all of these things, but over time, the larger companies will wonder why they didn't do the same themselves.
So in summary, in order to provide the benefit of excellent and flexible performance measurement that will help an asset manager win and keep clients whilst minimising the cost of this service, the supplier needs to provide the IT platform, the index data, the corporate actions, the end of day valuations and the option to choose between Holdings Based Performance and Transaction Based Performance per portfolio as a first step. Once that is done, the bells and whistles can be added. Not providing these basic services is like selling a car that runs on coal to power a steam engine.
Now where can I find a supplier like that?