I recently sent a suggestion into the GIPS(R) (Global Investment Performance Standards) Help Desk, suggesting that they introduce a recommendation, if not an outright ban, on firms deducting the annual fee from the annual gross-of-fee return, to arrive at their annual net-of-fee return, as part of **GIPS 20/20**. To me, this is far from accurate. But, there is nothing in the Standards to forbid it.

Yesterday, I had a brief email conversation with my colleague and friend**, David Yuska**, about this. He had chaired the development of a position paper on calculating net-of-fee returns several years ago; sadly, this paper never resulted in anything re. GIPS. He seems to agree with me that this practice is inappropriate.

I thought it fitting for those working on GIPS 20/20 to, at the same time they [hopefully] do as I suggest, remove the recommendation to accrue fees. A few years back I did a brief study to determine whether accruing advisory fees provided any benefits; I discovered that they did not.

**But surely accruing advisory fees must make sense?**

I shared my thinking with another colleague, who responded “if done correctly it obviously works so clearly there is something wrong in execution, your research or your conclusion.” This motivated me to have another look.

I believe that the suggestion to accrue arose from conjecture, only. That is, *it must surely make sense, right? *But, did anyone do any research to determine whether it actually did? I can’t say, but I did (both a few years back, and today!).

**A test (actually two) to determine if accruing advisory fees provides benefits**

I’m sure I can find my earlier research on this topic, but decided to conduct a brand new study, which is rather simple to do.

I did it twice. First, I used the S&P monthly returns for 2017. My annual fee is 1.00 percent. On a monthly basis this is not 1.00%/12, but rather [(1+1.00%)^(1/12)]-1, which yields 0.0829538 percent. Simply dividing 1.00% by 12 gives us 0.0833333 percent. While the difference is slight, it’s more accurate to do the additional bit of math.

Okay, so that’s the monthly fee.

The quarterly fee is therefore [(1+0.0829538%)^3]-1 or 0.2490679 percent. Note, I display the additional precision here, but normally wouldn’t.

With the accrual approach, we will deduct the monthly fee from each monthly gross-of-fee (GOF) return to arrive at the net-of-fee (NOF) return. With the non-accrual, we deduct the quarterly fee when charged/paid, on a quarterly basis.

**Here are the first results (sorry if it’s a bit blurred)**:

As you can see, the difference between the annual returns is zero! We get the exact same results.

I tried this a second time, using the S&P monthly returns for 2016.

**And the results are**:

Here, we find a one basis point difference; what I would characterize as *de minimis* [need to toss in some Latin every once-in-awhile].

**And so, does accruing advisory fees make sense?**

As I discovered a few years back, it doesn’t. Why do this extra work?

“Recommendations” in the Standards are considered “best practice,” but in a few cases, I respectfully disagree. And this is one of those cases.

**And what about deducting the annual fee from the gross-of-fee return?**

As I stated above, this all began with my suggestion that the practice of simply deducting the annual fee from gross-of-fee should either be disallowed or at least discouraged. Consider what the results would be with what appears above.

In the first example, the GOF return is 19.41%, while the NOF (for both approaches) is 18.24 percent. By simply deducting 1.00% from the GOF we get 18.41%, which is materially different.

And, for the second example, the GOF is 9.55% while the NOF for the non-accrual method is 8.48%. By deducting 1.00% from GOF we get 8.55 percent. Perhaps not a “material” difference, but still incorrect.

We will have to wait and see if my suggestions result in anything with GIPS 20/20.

If you’d like a copy of my spreadsheet, just send me an email: **DSpaulding@SpauldingGrp.com**. Oh, and if you have a different view, or can find something incorrect in my procedures, please let me know! Your comments are always welcome!!!

**A post script**

Based on comments from Neil Riddles and Kim Cash, I thought it useful to compare the methods to derive what the quarterly or monthly fees should be. I do it slightly differently than apparently what most folks do. As Neil points out, most firms simply divide the annual by 1/4 to get quarterly, or 1/12 to get annual. Simple, but arguably both problematic and wrong. If you link these values, you will not get the annual fee amount; rather, you’ll get higher amounts (recall that simple averages fail to take compounding into consideration). Consequently you (don’t say this too loudly) may be overcharging your clients. Perhaps not by a lot, but ask them if they mind!

This table provides a comparison. As always, chime in!

Accruing fees can be helpful for shorter periods, especially when billing quarterly. For example, when clients view MTD returns it can be helpful to know your actual net of fee return, even if the fee did not post in that period of time.

Caitlin, thanks for “chiming in.”

I understand this point. For example, in what I provided, if you’re looking at the January and February returns, the GOF and NOF are the same, which might cause some issues with the client. But is THIS the basis for the recommendation by GIPS? I think not. I believe that “way back when” (under the AIMR-PPS days), someone had the thought that accruing fees made sense, and so it got adopted as a recommendation, and then carried over to GIPS. But, as I show, mathematically, across periods, we get essentially get the same results, so other than this monthly comparison (where no fees are taken, or when the fee hits on the third month), there is (in my view) no justification for it.

It would be curious to know WHY firms accrue. Is it for the reason you cite, or is it because “well, GIPS recommends it!”? Unknown to me. Perhaps a worthy mini-survey.

Agreed, it doesn’t make sense in the GIPS context, especially with the presentation of annual returns.

And yes I suspect a firm would accrue fees for these two reasons, but curious if there are others!

Thanks, Caitlin. We’ll be posting a brief survey tomorrow; hoping to get some answers!

Hi Dave,

I’m getting different results than you with these inputs – I’m guessing that you are using full precision with the calculations, and only displaying 2 digits. When using the rounded/truncated numbers, the difference between the return streams is larger (6bps in the first example, 3bps in the second).

In any case, I think that there are a few benefits to fee accrual that apply even if the return over the entire period is exactly the same:

1. If fees are charged in advance and someone runs a performance report for the first month, their net returns will be “penalized” for the fees that are supposed to apply in the future, and will therefore seem understated. The accrual will smooth this out.

2. Risk Statistics are also impacted by this. Standard Deviations will be lower when fees are accrued, for example.

3. In the case of Fee Rebates, you can have situations where Net returns are HIGHER than Gross returns for certain periods. (For example, a 10k charge in month 1, and a 5k rebate in month 2.) Using Fee Accrual, this can be adjusted so that Gross is always higher than Net.

Thanks,

-Sam

Thanks, Sam. Yes, I used “full precision,” and will be happy to share the spreadsheet (as I noted in the post).

You raise additional interesting points. As I mentioned to Caitlin Kneram, in another post, my point deals with accuracy, which I believe is what GIPS’ intention was, in making this a recommendation. One could craft other reasons to or not to accrue. For example, with performance based fees: if you accrue, and then your performance drops, you get penalized, and have to put money back (and your NOF will be greater than your GOF) (you allude to this).

These reason are, in my view, separate and apart from GIPS. Yes, firms may wish to accrue, so that the returns are “smoothed out.” I guess in one sense, we can see that by accruing, you are not coming up with a materially different result, which could cause one to “game” the returns.

While GIPS recommends accruing, it fails to provide a reason. And since GIPS deals with presenting performance, and these presentations are, at least from a requirement standpoint, limited to annual, there doesn’t seem to be a reason to accrue.

We will soon do a “mini-survey,” to (a) find out who IS accruing and (b) why. Will be interesting, I’m sure. Thanks for your comments!

Dave,

Interesting topic. I agree that accruing fees for a simple asset management fee is of little use and adds complexity (and possible mistakes). However, for sizable performance fees it may be necessary to accrue the fee or individual periods may be noticeably distorted.

Further complicating things, in my experience most institutional firms charge one quarter of the annual fee each quarter. That is, a 1.00% annual fee is charged as 0.25% each quarter not as [(1+1%)^1/4]-1 which would be 0.249068%. So, the way to accrue monthly a 1% annual fee billed quarterly would be [(1+0.25%)^(1/3)]-1 which is 0.083264% monthly.

Since you are testing a compounding effect it might be interesting to use 2007 and 2008 numbers. The larger magnitude of returns would make a worst case scenario.

Neil

Thanks, Neil. I suspect that you are correct, that most managers do it the easy way (Annual fee / 4 = quarterly fee). However, that is technically incorrect. If we link these values we will get an annual result of 1.00375625%; however, if you link the value I came up with, you get 1.00000000%. Thus, you’re technically overcharging your clients a bit doing it the simple way. But, that’s a different matter.

As for accruing performance-based fees, this is often debated. While I agree that taking a big hit once a year can cause a shock effect, you also have the impact of “surprise,” when your accrual doesn’t match reality, and you have to adjust your values, resulting in NOF > GOF.

My point here had to deal with WHY DID THE FRAMERS OF THE STANDARDS RECOMMEND ACCRUALS? I suspect for accuracy, though there is nothing to prove or disprove this. But what other reason might there have been? Anyway, I’m trying to find that out.

I’ve obviously “opened a can of worms,” which is always a good thing. And, if I can get comments from Neil Riddles, well, that makes it even more fun for me! Thanks!!!

Dave,

The firms I have worked with charged the annual fee divided by 4 each quarter. However, they did not overcharge their clients. The fees, and how they were to be calculated, were spelled out clearly in the contract. The contract was reviewed and understood by financial and legal people on both sides. You are correct to say that 25 basis points per quarter is more than 1.00% per year. However, that does not mean clients are being overcharged.

Neil

Thanks, Neil. As long as the client’s aware of the details, then definitely no problem.

I could be way off, but here is my attempt to reconcile your findings and GIPS guidance. When asset managers charge clients fees, it is at a point in time, often beginning or end of period, not based on performance throughout the period, and it is typically by taking the annual fee and dividing by 4. Does that make simple division of the annual fee the better choice? The problem with simple division is that it doesn’t pick up the impact of compounding during the shorter interim periods the way your formula more accurately does. So GIPS recommends accruing for shorter periods in order to capture that, similar to the difference between taking a one time annual fee versus shorter periods. When, as demonstrated above, you divide the fee accrual geometrically you accomplish the same thing, getting much closer to an accurate depiction. It would be interesting to see how close your quarterly method and a simple annual fee divided by 12 come to an actual fee taken out each quarter in an account with no cash flows, perhaps carried out a few more decimal places…. ? I look forward to your rebuttal…

My rebuttal? Wow!!! You’re anticipation is interesting … though somewhat accurate.

I just pointed out to our friend Neil Riddles that by compounding the simple annual / 4, you get an overstatement of the fee (i.e., you’re charging too much).

What if we do it with the monthly? Well, if you take the annual (1.00%) and divide by four, and then link, you get 1.00459609%, vs. the 1.00000000% you get if you do it with my monthly fees. As a result, you’re actually charging MORE than if you do it quarterly!!! And so, your conjecture that surely this is why the GIPS Implementation people picked up on what the AIMR-PPS people did, and recommend accruing, seems not to be valid; at least in this case.

But Kim I do appreciate you chiming in … I think the first time. Come again!

Hi David,

Most of these questions are addressed in the paper Andre and I wrote, but Ill leave a few highlights here:

1) It seems like the topic of “arithmetically deducting a fee” is being combined with the topic of “accruing,” but these are 2 entirely different questions. When you arithmetically deduct a fee, it implies that the fees are charged on beginning assets. When you geometrically deduct a fee [(1+ gross)*(1-fee) = 1 + net], it implies fees are charged on ending assets. Ideally the formula would most align with how billing is done. Whether to accrue or reflect fees as paid is another question.

2) You are correct that many times the net of fee calculation method utilized does not result in material differences, but they could, especially in cases of performance based fees (but that’s a whole different animal).

3) Most firms do bill by dividing the fee rate by 4 or 12. I believe this is industry practice in most areas like interest calculations and such. I don’t think people are thinking of how this will affect performance returns when doing that.

Thanks, Krista. I should re-read your paper, to see what’s there and how it compares.

As noted elsewhere, I suspect that you (and Neil) are correct: that most firms divide by 4 or 12, without “thinking of how this will affect performance returns,” or the actual fees. The example here is a $1 million account. I just repeated my exercise, with a 0.25% annual fee, and a $50 million account. Again, not surprisingly, you’re overcharging with the simple approach, though the amount isn’t very high ($117-$143). But, would clients mind? I can’t say. To do it the “right” way (I think it’s right, because if you link the quarterly or monthly, you tie out to the agreed upon annual), is really rather simple math. It’s just that most people don’t know this. In fact, I don’t think most people even think about this stuff, only us “performance nerds,” to quote a certain woman I know (initials: AR).

David,

If you want the quarterly fees to compound to the annual fee of 1% and to have your Yearly Net “Accrual” Return match you Yearly Net “Non-Accrual” Return, you need to compound returns geometrically throughout.

Thus, first replace the Annual fee of “1%” with “-1%, and then replace each use of subtraction in your arithmetic calculation {R – f} of the ‘Accrual” and “Non-Accrual” columns with a geometric calculation {(1+R)*(1+f) – 1}.

When you correctly geometrize in that manner, your calculations will produce the EXACT same value for the Year “Accrual” and the Year “Non-Accrual” (18.2169% in Scenario #1 and 8.4542% in Scenario #2) and all your fees values will roll up to their correct annual fee value, which is less than zero. Thus, there will be never be any “overcharging”.

This approach will also equally work for the method that takes the “Annual fee” and divides it by 4. In this case the geometric roll-ups to each quarter will be correct and the annual return will be (1 – 1%/4)^4 – 1 = -0.9963% as one would expect when one no longer is geometrically subdividing the -1% to determine the actual fee payments.

There are a lot of other points to be made about fee accruing calculations, as you will find when you re-read Krista’s and my paper.

Andre

Thanks, Andre

I think there is a conflict between theory and practice on the recognition of fees on an accrual versus a cash basis in performance calculations. Many fee structures are complex and cannot be accurately calculated, by finance, until well after the measurement period has closed (e.g. 10-15 business days). Also, we typically see billing occur on a quarterly basis for most SMAs so if you do not accrue fees for your performance calculation not only will you have two months per quarter where your gross = net but also your net-of-fees performance will not be available until sometimes 3 weeks after quarter end. This likely runs afoul of marketing and client reporting requirements. Each quarter-end marketing, consulting services and sales are anxious to update databases and standard books for meetings with clients/prospective clients and demand a quicker delivery of data. Also, explaining the nuances of how net-of-fees performance is calculated to clients, prospects and even your own firm’s portfolio management and distribution would be a full time job in and of itself. Especially when there is a wide range of practices in place across the industry and little agreement or comprehensive guidance, other than Krista’s and Andre’s paper of course, out there for the for the best practices on calculating net-of-fees performance. I believe that they would prefer to see an even accrual off fees reflected in the performance stream which would appear intuitive from their standpoint. This would seem to be the basis, at least in part, for the JPMIM request for no-action on the use of model. I think the extra effort exerted is for the benefit of the marketing/sales process which is my guess as to why GIPS would recommend accruing fees.

I do agree that arithmetically “netting down” performance figures is not appropriate and guidance on that issues should be addressed as part of GIPS 2020 along with many other aspects of the calculation of net-of-fees performance.

Thanks, James. You, as well as others, offer good reasons TO accrue, which many should consider. My issue is with the GIPS recommendation, and the reason it is there.

I have it on “very good authority” that the reason this recommendation was made was to improve accuracy. I believe I’ve clearly demonstrated that there is no “improvement in accuracy.”

You may recall that the AIMR-PPS(R) had strongly recommended asset-weighted standard deviation. And why was this? The logic was that since composite returns are asset-weighted, should dispersion? Well, whoever came up with that idea convinced the other members of the committee to go along, and it was adopted back in 1997. Apparently no one thought about what the actual results would represent. And while Carl Bacon, CIPM and I disagree about a great deal (with new items being added on a regular basis, I might add), he and I both agree that this measure is not a good idea. Apparently the GIPS folks recognized this, too, as it is no longer recommended. It can still be employed, but the “recommendation” has been dropped.

My recent suggestion to the GIPS 20/20 folks that they consider either banning or at least discouraging the practice of deducting the annual fee from the gross-of-fee return to derive the net-of-fee return prompted me to suggest that it replace the recommendation to accrue fees. Knowing (or at least strongly suspecting, with affirmation recently being made) that the basis for this recommendation was purely about accuracy, and knowing there was none, was where this post came from.

No one has disputed my argument that there is no benefit on the accuracy front; that would be difficult to do, I believe.

I’m quite pleased that so many have chimed in with OTHER reasons to accrue. These should be taken into consideration when contemplating whether or not to do this. However, if you’re doing it because GIPS recommends it, or because you believe it’s more accurate, you shouldn’t.

Thanks again for your comments.

Good points and I believe it is a healthy exercise to regularly examine these issues to determine if they make sense. With GIPS 2020 on the horizon even more so.

Hi Dave,

Here is the part of the paper that addresses why we believe the accrual approach is used and more accurate.

The following reasons support the Accrual Approach:

• Other sources of returns within the portfolio, such as income on bonds and dividends on stocks, should be reflected on an accrual basis. We believe that liabilities, such as fees, should be as well. For example, if fees are earned on each day of a 5-day period, but the sum of these fees is only accounted for on day five, then the net-of-fee performance will not accurately reflect these liabilities on each of the days one through five.

[Additional comment: This is similar to the concept that Andre refers to as the “liquidation value,” which is the value of the portfolio that the client net if they liquidated today. It is also analogous to a mutual fund that trades at its NAV.]

• If there is a large external client flow after the billing period begins but before the fee payment

date, there could be an unintended performance impact. For example, if you billed a $3,000 fee based on an opening $100,000 market value and then the client withdraws $40,000 before the fee is paid, the fee payment taken out of the $60,000 market value, if markets did not move, might decrease the gross-of-fee returns by 5% instead of an intended 3%.

• The As-Paid Approach could result in inflated volatility measures due to the fact that fees are only reflected on performance on the dates fees are paid.

• From a reporting standpoint, the Accrual Approach might allow a firm to report an appropriate net-of-fee performance before the end of the billing period. Although the above reasons show that there can be many advantages to using the Accrual Approach, the As- Paid Approach continues to be used due to its simplicity. Furthermore, when fee information is required to calculate or estimate accurate accruals but is unknowable till it is charged, the As-Paid Approach might be the only applicable method.

Krista, thanks for sharing this.

You offer a hypothetical example which, of course, could occur. Often, with accruals, especially with performance-based fees, managers determine later in the year that they were WAY OFF in their accruals, and therefore must pay back money. My point: we can find hypotheticals that support or detract both approaches.

While you’re probably correct that the reason many favor the “As-Paid Approach” (APA) is simplicity, I believe many do the accrual method because GIPS recommends it, and since recommendations are “best practice,” they follow it, not realizing that its basis was the mistaken belief that accruing of fees would result in more accurate returns.

My example demonstrates, I think pretty clearly, that net-of-fee results will be the same, regardless of whether you accrue or not. Yes, there will be more volatility with the APA method, and perhaps this is a good reason to use it. That said, one should NEVER calculate standard deviation against net of fee returns; gross-of-fee should be used. This would counter the argument that you’ll experience volatility, since the fees are taken away. I failed to point this out in an earlier response, so appreciate you giving me the chance to do so now.

Glad that we’re seeing so much interest in this topic; thanks for joining in!

Thank you

You are VERY welcome!