It’s an inexact science because we have different ways of doing the same thing. Just think about how many formulae exist for calculating the most fundamental of performance metrics: a rate of return. You will get different values for a performance return depending on which formula you use. But even though they generate different results it isn’t the case that one is right and all the others wrong, it’s just that there are different ways of doing the same thing. Each of the formulae is correct within the context for which it is to be used.
None of this bothers me, as long as people understand the nuances of the different calculation methods, and use the appropriate method in the correct context. In fact, I think it is a good thing. It shows that we have come a long way over the past 25 years in refining performance measurement into a sophisticated and detailed discipline.
What bothers me is that we haven’t made similar advances with performance analysis. In fact, I think the evolution of performance analysis is in danger of stalling, if it hasn’t already done so.
Performance measurement gives us information not explanation
Performance measurement tells us how much we have made (or lost) by investing in a portfolio. But that’s all it tells us. It doesn’t tell us why our investment decisions were or were not successful. And unless we know why the decisions we’ve made in the past were good or bad, we can’t repeat success and avoid repeating failure in the future.
That’s where performance analysis is supposed to kick in. Performance analysis gives us insight into the sources of out- and under-performance, and that insight should enable us to identify the good investment decisions and the bad ones. The problem is that it doesn’t do this; or, more precisely, the way we are doing performance analysis doesn’t do this.
Hang on a second, I hear you say. Isn’t this what performance attribution gives us? Well, no actually, it doesn’t; it only gives us part of the story.
The problem is that the attribution methods we use take the relative return between a portfolio and an index-based benchmark, and decompose it into a set of constituent contributions. Attribution shows us where the portfolio was structured differently to its benchmark, and whether those differences resulted in positive or negative contributions to the relative return.
Now there’s nothing wrong with that - it is indeed very useful information and insight - it’s just that we’re not learning everything that we need to know.
We don’t just need to know how our portfolio is doing relative to an index. We need to know whether investment managers are doing a good job at investing our money. And we need to know whether our investment objectives are going to be achieved.
Relative returns don’t always deliver real value
Let’s say my portfolio is overweight relative to its benchmark in an asset class that performed well over a six month period. This asset class will make a positive contribution to the relative return over the past six months. So the decision to go overweight was a good one.
Now let’s now say that six months ago the portfolio weight in this asset class was greater than it is today. This means that at the end of the period we have less of the portfolio invested in an asset class that performed well over the period. That will make a negative contribution to the actual return of the portfolio over the past six months.
Here’s the key point: the attribution techniques we use assess investment decisions in the context of relative returns, but we should also be assessing them in the context of absolute returns. A positive relative return indicates that my portfolio has outperformed an index. A positive absolute return indicates that my investment has increased in value. It is possible for a portfolio to deliver a positive relative return over a period in which a negative absolute return was achieved. In other words, the fact that my portfolio is outperforming an index, does not mean that my investment is increasing in value. More importantly, it is possible for an investment manager to continue outperforming an index even if they get worse at value generation.
So, why don’t we routinely perform attribution relative to how portfolios have performed over time? We could call this ‘temporal attribution’ as opposed to ‘index attribution’. It would look at how portfolios were structured at the start of reporting periods to see what’s different, whether these differences generated or lost value, and why these changes were made – and then fine-tune investment processes to make better decisions in the future.
This is only one example of additional insight that performance analysis could provide. There are others.
Performance analysis needs to look into the future as well as at the past
Increasingly, investment objectives are expressed in terms of absolute goals, where the objective is not to beat an index. In fact, it isn’t even simply to make money; the objective is to make a defined amount of money.
Attribution relative to an index-based benchmark can’t give us insight into whether our investments are on track to achieve goal-based objectives. What we need is attribution relative to a value-based projection. This ‘value attribution’ would probably require new attribution techniques.
The classical age of performance analysis
It feels to me like innovation in performance analysis has stalled. We’ve developed useful techniques that allow us to provide valuable insight into the positive and negative contributors to relative performance. And that’s a great achievement, but it’s not enough.
I see an analogy with classical physics. Classical physics is the physics of Newton and other 17th century scientists. It works perfectly well at a macro level, explaining the way that everyday objects react when forces are applied to them. It works for small things like golf balls and big things like planets; for things that move slowly like glaciers and things that move fast like rockets.
But classical physics doesn’t provide complete explanations for the things that are outside our everyday lives, like sub-atomic particles and things that move at the speed of light. So 20th century physicists developed relativity theory and quantum theory, which give us additional insight into how the universe works.
It seems to me that what we have today is ‘classical’ performance analysis. And we need more innovative thinking to take performance analysis to the next level. The big question is where does that innovation come from: universities, system vendors, investment management firms, investment consultants? I’m not sure. But I am sure that if someone doesn’t start pushing for it, it won’t happen.
BI-SAM was named in the Journal of Performance Measurement’s November 2014 survey as the #1 provider of Equity Attribution, Fixed Income Attribution, Performance Measurement and Composites. Visit our product page to learn how we can help your firm to more effectively calculate, analyze and distribute information about investment performance.