"Quantile Tracking Errors (QuTE)" by Aguilar, Chengan, and Custovic

Portfolio managers are judged by a metric that's supposed to measure whether they're tracking their benchmark. That metric works fine when the two portfolios differ in average return or volatility. But what happens when they differ in ways the metric cannot see? The most common tracking error measures have been blind to skewness and kurtosis in returns for decades, flagging no problem while distributional differences pile up.

• The industry's primary tracking tool measures only two things. Traditional tracking error metrics focus on mean return and variance. When a portfolio's returns become more skewed or develop fatter tails than its benchmark, the standard measures don't register a difference at all.
• Asset returns in the real world don't cooperate with this framework. Research documents that stocks and bonds consistently exhibit excess skewness and kurtosis. Yet the compensation structures and risk management systems that depend on tracking error never evolved to account for this.
• The problem isn't theoretical. Case studies of major indices during the 2001 dot-com aftermath and the 2008 Great Recession reveal moments when tracking error stayed flat while the actual distributional differences between portfolio and benchmark were shifting dramatically.
• What gets missed matters differently depending on how risk flows through markets. A portfolio might track its benchmark perfectly in terms of average returns but develop a vulnerability to tail events that goes completely undetected by traditional measures.
• This creates a management system that's blind to categories of risk that actually determine outcomes. If the metric that justifies a manager's compensation and guides portfolio construction cannot see what's happening, what else is the industry missing about how portfolios actually behave?