Performance-based fees would seem to be an obvious solution. They greatly reduce the temptation to pursue unbridled growth, while also discouraging benchmark hugging. But the predominant performance-based fee structure in use today, which uses a high-water mark, simply introduces another form of incentive misalignment. Clients pay a base fee plus a share of any outperformance, and if performance subsequently suffers, the manager must recover the prior peak before again earning performance fees. While this sounds like a good deal for investors, it's really a “heads we win, tails you lose” proposition. For the manager, the payoff to investment performance is essentially a call option in which the high-water mark is akin to the strike price and the client bears the full downside if things go poorly.
With this in mind, how does the manager maximize this option value? He can't really change the time to expiration; clients will redeem when they want to. He can't change the strike price; the high-water mark is set. And he probably can't change his skill level; if he could, he already would have done so. The only thing left to change is the volatility of the portfolio, even if doing so is not in the client's interest. Indeed, research has shown that even a manager with negative skill is better off increasing portfolio risk under such a fee structure, despite the fact that this is, by definition, a worse outcome for clients.
High-water-mark performance fees are also flawed due to their path dependence. Unless an investor redeems while at their high-water mark, they will have paid a greater portion of the return in fees than the stated sharing ratio. For instance, if the manager produces strong returns in the first few years, and subsequently underperforms, the client will have paid substantial performance fees in those first years, despite a potentially negative cumulative total return. Because very few clients would redeem while at their high-water mark, it is very likely that a client will pay a substantially larger portion of their returns in fees than the headline sharing ratio implies.