Morningstar questions Russell
Morningstar chief executive Graham Rich responds to Frank Russell's analysis of its star rating system.
Friday, October 19th 2001, 11:07AMEditor's note: This is a response to an earlier article where Frank Russell questions star ratings models. If you haven't read that story yet Click here.
Clearly neither Alan Schoenheimer nor Frank Russell have bothered to delve deeply enough to become fully-informed about the Morningstar Star Ratings model. What is the quality of Frank Russell's own research if this is an example of how deeply they research an issue before publishing their conclusions?
A simple example - the article states that four organisations in Australia offer star ratings. Van Eyk would no doubt be surprised to hear that they are one of them. Certainly, Cannex and Corporate Monitor would be surprised to hear that they're not. Since Morningstar introduced managed fund star ratings in Australia and New Zealand six years ago, at least seven other fund research houses have followed suit.
Some more material examples:
Schoenheimer states that he has a problem with the Morningstar system being a relative one. The very point of the Morningstar Star Ratings model is to show the relative quality of funds and fund managers, and for good reason: there is a finite universe from which to select a fund, and therefore of course you need to compare what's available before investing. Is Schoenheimer seriously suggesting that it would be more valid to set some absolute standard that funds must meet and then allocate ratings accordingly? What would be the point? The Olympics don't give every athlete that runs a sub-10 second 100 metres a medal - the point is to rank them from "best" to "worst".
You also quote Schoenheimer as saying that there's a risk of bias coming in as fund managers with poorly-performing funds won't seek to get them rated. Firstly, isn't Schoenheimer contradicting himself? He complains that the model is relative, then he complains that poorly-performing funds won't be included - if he's after an absolute model, it wouldn't matter which funds were included. Secondly, Morningstar doesn't permit fund managers to choose whether they'll be rated or not. We determine this, considering the wishes of our adviser subscribers, and the direct distribution power of the fund manager. We invite a fund manager to participate in the ratings process, but if they decline, we'll rate regardless. In fact, fund managers are responsible and see the value of the ratings process, and want to participate.
Schoenheimer claims that the "distribution wanders". The distribution of Morningstar Star Ratings across funds does follow a normal distribution curve. At no point has Morningstar ever claimed that the distribution among sector funds will mirror a normal distribution - it's blatantly stupid to expect that it would or should, and again shows Frank Russell's complete ignorance of the philosophies underpinning the Morningstar Star Ratings model. Once a fund's past performance is compared to that of its peers, to create a "Qt Rating" for the fund, that Qt Rating is then combined with the qualitative rating for the fund (the "Ql Rating"), to create a percentage rating for each fund. The percentage ratings of all funds are then compared to create a Morningstar Star Rating within a normal distribution. This methodology ensures that market bias is removed, and that, for example, if once market bias is removed, the universe of New Zealand equity funds is relatively stronger quality than the universe of New Zealand fixed interest funds, the former is recognised commensurately.
Schoenheimer is concerned that "the relationship between some of the items measured... and a fund's performance aren't clear". Further, he asks whether Morningstar's qualitative research components have any predictive power - again showing his ignorance of the philosophies underpinning our model. If he'd bothered to ask Morningstar, we could have enlightened him. It's a nonsense to suggest that any ratings model can predict absolutely the future performance of a fund. Morningstar's qualitative ratings reflect our opinion of the issues that a financial adviser and investor should consider before investing in a fund, and provides information to allow this. Morningstar's qualitative rating components (e.g. Corporate Strength Rating) are comprised of two parts: firstly, what's called a subcomponent rating, which benchmarks every fund manager against the same standards for that component, so a fund manager's quality in those areas can be compared to those of other fund managers; and secondly, a trend rating in which our analysts indicate whether for the component Morningstar believes the fund manager is likely to improve or worsen. Together, these summarise Morningstar's opinion of the quality of the foundation from which the fund manager is moving forward relative to other fund managers, and which is therefore going to impact on the relative quality of their funds going forward.
The article also states that Frank Russell found "there were small sample sizes, and not a long history of data". Again, Frank Russell's ignorance of how the model is constructed comes shining through. There are 2,360 data points considered when calculating the Qt Rating for a fund. There are well over 500 qualitative issues considered for a small fund manager, many more for larger managers.
The article also quotes Frank Russell's research showing that the ratings "from the various Australian agencies are well correlated". In fact, they are not. The models are very different, and outcomes are quite diverse.
Schoenheimer comments that star ratings are not yet influencing distribution to the same degree as in the US. The US has had star ratings for 15 years. Australia has had them since Morningstar first launched them four years ago - while not yet influencing funds flow at anything like the US 70-80 percent rate, the influence is growing and will continue to do so. Interestingly, during establishment in the US, similarly ignorant attempts were made to discredit Morningstar. However, 15 years later, the US market, along with the Federal Reserve Bank of Atlanta (a member of the US Federal Reserve system) accepts the influence and validity of Morningstar Star Ratings. In a recent report, the Federal Reserve Bank of Atlanta stated: "Following rating changes, we find economically and statistically significant abnormal flow in the expected direction, positive for rating upgrades and negative for rating downgrades... Overall, our results indicate that Morningstar ratings have unique power to affect asset flow." The Federal Reserve Bank of Atlanta applied considerable resources to its study, and included Morningstar's US Director of Research in the working group.
Frank Russell asks: "are star ratings merely a proxy for past performance?" It's no surprise to Morningstar that there is a correlation between past performance of a fund and its Morningstar Star Rating. The model is deliberately built to consider where a fund has come from, balancing quantitative (past performance) and qualitative (forwards-looking) issues equally, because there's no evidence anywhere that one is more relevant than the other in considering the quality of a fund.
Finally, let's take a look at the "quality" of Frank Russell's avowed $20 million research effort, one output of which is the ANZ Gateway funds. These funds are promoted by ANZ as being made possible through an alliance with Frank Russell who "leaves no stone unturned, researching and reviewing some 2,000 fund managers... to find the cream of the crop". Of the nine ANZ Gateway funds on Morningstar's infobase, none have provided better than average rolling quarterly returns compared to their peers over their 2.5 year lives, and all but one have given consistently below-average rolling quarterly value (returns) for the risk taken compared to their peers. Only two of the funds are in what Russell would consider "small" subcategories. The ANZ Gateway Australian Shares Sector Trust is operating in the largest Morningstar subcategory of 145 funds - the Australian equity diversified trusts subcategory - in which the fund has achieved an average Return Rating of 6 out of 10 (signalling its rolling quarterly returns have been consistently average compared to its peers), and a risk/return rating of just 4 out of 10 (signalling that it's achieved consistently below-average risk-adjusted returns over time compared to most of its peers). Only one of the nine funds has a risk/return rating of 6 or more out of 10 (ie achieved average or better than average risk-adjusted returns over time compared to peers) - strange when the aim of a multi-manager, multi-style investment strategy is normally to at least achieve consistently strong risk-adjusted returns, even if not superior absolute returns. Perhaps there are a few stones left unturned by Frank Russell in their review of 2,000 fund managers after all. Who'd know? Russell have a "black box" approach to research, in complete contrast to Morningstar's philosophy of complete transparency.
No ratings model is perfect - the ultimate model would indeed predict future performance with 100% accuracy. We encourage our subscribers to understand the purpose of the Morningstar Star Ratings model, and how to use it to better understand Morningstar's opinion of the relative quality of funds and fund managers.
Ignorance may be bliss, but frankly, Russell, we challenge you to do a better research job than this before commenting on issues.
Graham Rich is the CEO of Morningstar
|« Are hedge funds a seperate asset class?||King builds an empire »|
Commenting is closed
|Printable version||Email to a friend|