The Divergence of High- and Low-Frequency Estimation: Causes and Consequences

摘要

Most of us use monthly or higher-frequency returns to estimate the inputs for portfolio construction and risk models, even when our investment horizons are much longer. We assume, either implicitly or explicitly, that standard deviations scale with the square root of time and correlations estimated from higher-frequency returns are similar to correlations estimated from lower-frequency returns. And we're not alone. The "square root of time" rule for annualizing standard deviation is everywhere. It's hard-coded into risk management and optimization software packages, mandated by the Global Investment Performance Standards (GIPS), and embedded in the Blaek-Scholes equation. But surprising new evidence suggests that it is often wrong.In this report, Will Kinlaw, Mark Kritzman and David Turkington analyze the causes and consequences of this discrepancy. They advise caution when extrapolating high-frequency risk estimates to long-term scenarios, drawing on divergence between US and emerging-market stocks from 1990 to 2013 to illustrate the dangers of this approach.