By Paul Sweeting
It was, unfortunately, difficult to avoid the Eurovision Song Contest that took place at the weekend. While some would argue that it has little to recommend it in terms of the music, the voting patterns can be used as useful analogies in the world of asset management.
In the contest, a large number of European countries (with Europe being re-defined to include countries such as Azerbaijan), have equally-weighted votes to decide which song is “the best”. Each country can award anything from 1 to 12 points to the countries of their choice, and the country with the most votes “wins”. So the votes from Malta count just as much as those from Russia.
I don’t know whether Russians living to the east of Urals are prohibited from voting, but in any case this approach is intended to ensure that the weight of each country has no impact on the outcome of the contest.
However, while the votes from each country are treated as being independent from those of any other, there are a number of recognised “blocs” such as the “ Viking Empire” (Scandinavia and the Baltic States) and the “Warsaw Pact” (Poland, Belarus, Ukraine, Russia, Armenia, and Moldova). Looking at the 2013 results (which are available on the Eurovision website – I didn’t have to watch the contest), it’s clear that such blocs exist. The average mark awarded to each “Viking” country by another was around twice that received from other countries; for the “Warsaw Pact” bloc, the ratio was closer to three times.
This problem is exacerbated by the fact that countries can fragment, Yugoslavia being a prime example. Any country breaking up into smaller constituent countries is likely, unless the breakup is particularly acrimonious, to produce entries that score higher marks than the original country would have – bloc voting would see to that.
Apart from suggesting that the Britain’s best hope in Eurovision is for a breakup of the United Kingdom, this put me in mind of risk parity.
Bear with me.
Like the Eurovision Song Contest voting rules, each asset class is weighted such that its contribution to risk is equal, rather than being proportional to risk (or population in Eurovision terms). However, the correlation between some groups of assets can be so high (bloc voting) that this apparent diversification (independence) is illusory. What’s more, redefining an asset class as several smaller asset classes (breaking up the UK) is arbitrary, and can just give more weight to the original asset class (more bloc voting).
An alternative is to try and take a factor risk parity-type approach, where the allocations are made only to individual risk factors, rather than to asset classes. This means that if a new asset class comes along which can be thought of as some combination of existing risk factors, it is not just automatically added to the portfolio.
In Eurovision terms, this would mean having only one set of votes per bloc. This alone would make the contest fairer. However, if the principle were extended, such that each bloc could only have one entry, it could even make the contest more bearable.
I’ll start writing the algorithm now…
Paul Sweeting is European head of strategy at JP Morgan Asset Management
Comments