How and why the allocation concept killed individualization

De Lillois Fractale Wiki
Révision datée du 30 mars 2010 à 17:36 par Pge (discussion | contributions)
(diff) ← Version précédente | Voir la version actuelle (diff) | Version suivante → (diff)
Aller à la navigation Aller à la recherche

The concept of allocation is widely used today. It is now like a basic concept, impossible to circumvent in the design of any portfolio management system. But this concept actually is an usurper! It happened to be used in an inappropriate place, and this caused dramatic damages to the Asset Management industry. One of the basic rules in modeling - in any kind of application – is the orthogonality (or independence) of the concepts assembled to describe a system. Regarding Asset Management, this rule was completely broken when the first Asset Management software solutions were written. And more precisely the rule was broken by the allocation concept. This merits some explanation.

What is an allocation? Everyone knows that an allocation is a set of weights given to asset classes within a portfolio, or within some subset of a portfolio. One may add that this set should be mutually exclusive (no intersection), that it has to be exhaustive (to cover the whole portfolio, or portfolio subset), and similar remarks. But our key point is simple and obvious: an allocation always relates to a portfolio, more precisely to a portfolio profile. Now, how is used this key concept? The answer is at least two-fold. The most common usage of the allocation concept is an expert expression. For instance, an investment committee, or an investment team, translates its analysis into a ‘recommended allocation’, and probably in several recommended allocations, typically one per risk profile. Besides this usage, the second frequent usage of the allocation concept is profile definition. For funds at least, and often for high net worth individual profiles, the allocations (global, sectorial, geographic, etc.,) are introduced as specifications by the fund promoter or by the client.

Is it good to use the same allocation concept on these two layers? Actually it is a big mistake, and here is the simple argument for this statement. As a portfolio profile definition component, it is wise and reasonable to use allocation as a key concept, since weights are intrinsically related to portfolio assets. But as an expert expression, it is too specialized and inappropriate, because the expert analysis and expression may be completely described without any link to a portfolio! The fact that an expert has positive or negative feelings (be it a forecast, an opinion or an estimation) on sector entities, on geographic entities, on explicit instruments has intrinsically nothing to do with the presence or absence of one or many portfolios! However it has occurred that experts had to translate their know-how into a simple convenient structure, and the allocation structure was present as a common concept in most of the early financial software systems.

The result of this modeling confusion was the difficulty, and even the impossibility, to properly manage multiple and various individual profiles. If the expert expression is a set of allocations, then obviously the number of profiles that will be potentially followed efficiently will not exceed the number of standard allocations suggested by the experts. And this is why today, even if most private bankers use individualization as a key marketing element, most of them are de facto forced to consider their multiple individualities as sets of clones of three to six standard profiles. Individualization has become unaffordable and, even worse, conceptually unreachable.

Of course, the origin of this confusion was an industrialization need. How would a bank efficiently manage thousands of mostly similar, but slightly different individual profiles? The trap was wide open, and no alternative was considered: instead of rethinking the existing with new concepts, several small adaptation steps were added, resulting in today’s landscape, with handicapped individualization.

After considering this deficiency, the following question arises: is there an alternate and appropriate concept, answering both individualization needs and industrialized processing? Our key conclusion regarding profile definition is “the allocation is a good concept”. But regarding expert expression, it should be replaced by something else. The new, simple and efficient concept introduced here is opinions.

Let us briefly introduce this newcomer. An opinion is a general scalar expression of a subjective perception by an expert, completely portfolio independent. Using opinions instead of allocations as expert expression requires some redesigns for Asset Management software. Not only does it reconcile the individualization promises and industrialization needs; it also opens several promising new business paths: the opinion material becomes an exchangeable commodity; a kind of Asset Management fuel… but this is another story!

Philippe Gonze