High-Water Marks and Performance Fees, Properly Explained

The problem a high-water mark solves

Imagine a fund that charges twenty per cent of gains, calculated each year, with no memory. Year one it gains 30 per cent and takes its fee. Year two it loses 30 per cent and takes nothing. Year three it gains 30 per cent and takes a fee again. The investor is roughly back where they started — the arithmetic of a 30 per cent fall and a 30 per cent rise leaves them short — and yet they have paid a performance fee twice.

That is not a hypothetical abuse; it is what happens by default if nobody designs against it. A high-water mark is the design against it. It records the highest value the investment has ever reached, and a performance fee is only charged on gains above that mark. Losses must be recovered in full, and for free, before the manager earns again.

The essential property is that the mark never falls. It rises when new highs are made and it stays put otherwise. A mark that resets after a bad year, or on a calendar date, is not a high-water mark however it is described — and the distinction is worth checking rather than assuming, because the wording in fund documents varies more than the concept does.

A worked example

Take an investment of 100,000, a twenty per cent performance fee and a high-water mark.

It rises to 120,000. The gain above the mark is 20,000, so the fee is 4,000, and the mark is set at the post-fee value. It then falls to 95,000. No fee — and note that the mark stays where it was. It recovers to 118,000. Still no fee, because the investment is below its previous high; every euro of that recovery belongs to the investor. Only once it passes the old mark does the manager earn again, and only on the excess.

The uncomfortable implication, which is rarely spelled out, is that after a deep loss a manager may go years without earning a performance fee. That is the intended effect. It is also why some funds quietly reset marks, and why "does the mark ever reset" is a fair and important question to ask.

Crystallisation: when the fee stops being theoretical

A performance fee is usually accrued continuously — reflected in the fund's value as it goes — but only crystallises on set dates: annually, quarterly, or on redemption. Until it crystallises, an accrued fee can still evaporate if the fund falls back.

The frequency matters more than it seems. A fee that crystallises quarterly can be earned on a gain in the first quarter that is entirely given back in the second. An annual crystallisation gives losses more time to net off against gains before anyone is paid. Shorter periods favour the manager; longer periods favour the investor. Neither is improper, but they are not equivalent, and the fund's documents will tell you which one you are in.

The awkward problem: everyone joins at a different moment

Here is where it gets genuinely difficult, and where most published explanations stop.

A fund is not one investor. People subscribe at different times, at different values, and a single fund-level high-water mark cannot be fair to all of them at once. Suppose a fund is 20 per cent below its high-water mark and you invest today. The fund then rises 25 per cent. You have made a real 25 per cent gain — but the fund as a whole is only just back at its old mark, so no performance fee is due. You have had a free ride: a genuine gain on which the manager earned nothing.

Now reverse it. You invest at the fund's peak, and it immediately falls and then recovers. Investors who joined during the dip are now in profit and owe a fee. If that fee is charged at fund level, part of it comes out of a pool that includes your money — and you are still down. You would be paying a performance fee on somebody else's gain.

Neither outcome is acceptable, and both follow inevitably from treating the fund as a single unit. So funds do one of two things.

Series accounting

The fund issues a new series of units for each dealing period. Every series carries its own value and its own high-water mark, and its performance fee is worked out on its own history. An investor who joined in March is in the March series and pays on the March series' gains, entirely independently of the investor who joined in June.

When a series has paid its fee and is level with the lead series, the two can be consolidated to keep the administration from multiplying indefinitely. The method is transparent and conceptually clean — each investor can see exactly which series they are in and what it has done — at the cost of running several series at once, particularly through a drawdown.

Equalisation

The alternative keeps a single unit value for everyone and corrects at investor level instead, using an equalisation credit or a contingent redemption depending on whether an investor joined above or below the fund's mark. The result is the same fairness; the mechanism is arithmetically heavier and much harder for an investor to verify from the outside. It is more common in Europe; series accounting is more common in the United States.

There is no better method in the abstract. What matters is that a fund charging a performance fee uses one of them. A fund that charges a performance fee against a single fund-level mark with no equalisation and no series has not solved the problem — it has distributed it among its investors according to when they happened to arrive.

What to ask

Five questions, and they can all be answered from documents before you commit. Is there a high-water mark, and can it ever reset? How often does the fee crystallise? Is there a hurdle — a return the fund must beat before the fee applies — and is it absolute or measured against an index? Does the fund use series or equalisation? And is the fee calculated before or after the management fee and the fund's costs, which changes the number more than most investors expect?

None of this tells you whether a fee is good value. That depends on what the strategy does and what the alternatives charge. But it tells you what you are actually agreeing to, which is a different question and a prior one.

Related reading: Understanding Drawdown, on why recovering from a loss is harder than the loss suggests, and What Is a Fonds voor Gemene Rekening?, on the structure underneath many Dutch funds.