BrightonRock Group head of research and Transparency Task Force costs & charges team member Con Keating questions the robustness of the Investment Association’s recent research on investment costs and performance
The strapline to the press release of the latest Investment Association (IA) report “Hidden Fees: The Loch Ness Monster of Investments” was eye-catching and amusing. Having read the report several times, a painful process I would not recommend even to a devout masochist, I have to conclude that if the IA had been swallowed and lay, decomposing, in the belly of the beast, we would receive another press release from them announcing its capture, containment and domestication.
I first studied undergraduate level statistics and probability theory in 1963, and formal training in econometrics followed in 1971. I have edited several text-books on these subjects written by academic friends and I have served, in august, honoured company, on the steering committee of the Financial Econometrics Research Centre for many years. In the intervening years, I have read many hundreds of empirical financial studies and reports, perhaps even thousands. This report is by far the worst; so bad that it is offensive, insulting both our common-sense and intelligence.
The report has a veneer of transparency and rigour, but left me with over 30 questions. In a thirteen-page document, including four pages of cover, contents and blank space, that is a lot of questions. As many others have published similar concerns, I will raise only a few here.
The sample dataset is odd, covering two full years and then just 11 months. What was the hurry – could it just have been to feed the report into and corrupt the FCA’s ongoing work on asset management, on which we expect a preliminary paper in September?
With any piece of empirical work there are always choices and assumptions to be made, which introduce bias. Of the assumptions disclosed, all but one serves to present fund management performance in a positive light. The Investment Association has long, and rightly, criticised the use of the SEC algorithm for calculating portfolio turnover, and the much maligned Brussels bureaucracy concurred, writing their own, now-abandoned, standard; yet here it is, central to their report. As the lower of sales or purchases in a period divided by the average fund value, it is clearly the metric which will return the minimum turnover rate, and in this report, the lowest transaction costs estimate. In fact, transaction costs are incurred on both purchases and sales.
It is irksome that we are repeatedly told that 1350 “equity fund accounts” are contained in the dataset, when there are in fact 387 funds in the first year, 504 in the second and 457 in the third period. This is, in other words, an unbalanced panel with a common maximum possible of 387 funds, and likely many fewer. We have no idea how many funds are actually included in the “UK All Companies” active and passive universes in any period.
In any rigorous academic study, the distributions, or at least their descriptive statistics would be disclosed. Such disclosure allows us to make our own judgement of appropriate measures of central tendency, but we are repeatedly treated to “bundled” measures, and even an average bundled ongoing charges figure. It would also allow us to consider the riskiness of these investments, but risk is a word entirely absent from this report.
The aggregation taking place is between funds with a wide range of investment mandates, including FTSE 250 and AIM stocks, an inference drawn from their index presence in the final graphic of the report- but what purpose is there to showing those indexes for a different period from the data under investigation?
There is a real problem with the benchmarks used as comparators in that these are presented without any description of their compilation. In 2012-13, the (simple average) active benchmark reported was 13.61% and the tracker 17.27%, a huge difference not present in any other period which simply demands explanation. The weighted average was slightly less extreme at 14.71% and 16.84%. Both are implausible.
Bundling and averaging these different mandates simply serves to confuse and possibly mislead; it would have been so much easier to interpret had the various FTSE and AIM mandates been left separate and compared with their usual (market cap weighted) benchmarks.
The radical departure in this report from all other studies that I have seen is that it reports that active fund management adds materially to performance. It seems that active fund managers are very consistently able to add value across mandates – a further finding not reproduced anywhere else, in my experience. We are asked to believe that, in 2013/14, the average active UK fund manager added 649 basis points relative to the simple average benchmark (555 basis points weighted). Quite apart from the significance of this one figure to the overall average quoted repeatedly, this level of gain is more commonly associated with highly leveraged hedge funds, even if rarely achieved by them. I was sufficiently taken with this that I spoke with friends in seven of the major fund management houses and asked how many of their funds had equalled or exceeded this, and these fund managers probably account for a plurality of the funds in this marketplace. The question was greeted with incredulity; as one said: “If those figures had been achieved, I would have blown the advertising budget sky-high!”
Of course, the returns cited cannot be chain-linked to produce a full period outcome, though we are treated to the three period average of the Fitz Partners dataset and the wider sector sample (whatever that is) – in one, active returns are 14.83% and in the other 17.17%, while tracker are 11.83% and 13.45%. This raises another question: what statistical significance should we assign to any of the figures in this report?
Moving specifically to hidden costs, at best, it is disingenuous to search public documents looking for them as this report perhaps did. Though surprisingly, stock lending revenue (simple average) is reported as zero in all periods. It is also perfectly possible to strip value from a fund by nefarious trading practices and other means. Contrary to their assertion that the published return of a fund will include all the explicit and implicit costs, and implicitly that we should not trouble our little heads, there are both revenues that rightly belong to the fund which never make it there and payments that may be made through poor execution which simply reduce the headline return. Let us not forget that all costs and fees are a direct charge on alpha, a very serious business.
As for the claims of significant benchmark relative added value, time will doubtless tell. In the meantime, I see that monkeys have again beaten hedge fund managers, as they have in every year since 2012. Regrettably, their retirement must be postponed yet again.