A view from the ivory tower
Study sheds surprising light on short-terming
BOSTON-With the benefits of extension almost taken for granted among money managers, it seems worthwhile to ask how these strategies stand up to more scientific scrutiny. Among academics, broad generalizations can be punishable by death, but a considerable body of opinion supports the idea of a 'liquidity premium' on very short-term investments.
Most academics agree that investors should earn a slightly higher rate on longer-term instruments to compensate for their marginally higher risk and lower liquidity. But this doesn't make mismatch strategies a free lunch. According to Alan Marcus, professor and head of the finance department at Boston College's Carroll School of Management, any benefits from extension should be fully 'paid for' in terms of added risk. "The textbook view is that the profit from riding the curve should not be zero. Gains can be more than losses on average, but no more than is commensurate with a market risk premium."
But when Marcus and co-author Robin Grieves tested the "no free lunch" principle on historical money market data, they found some surprising results. Over a study period running from 1949-1988, they found steady profits were generated by a 90-day duration mismatch between 3-month and 6-month treasury bills. The simulated profits were too large and consistent to be explained away by transaction costs or incremental risk.
Interestingly, the most consistent benefits seemed confined to the shortest end of the curve. Mismatch strategies that went further out the yield curve-3-month versus 9-month and 3-month versus 12-month-were more risky, and performed no better than the three-month mismatch.
Testing for what statisticians call "stochastic dominance," Marcus and Grieves found the short mismatch nearly always produced higher returns with no meaningful increase in risk. This poses a still unsolved puzzle. "The higher returns to riding the yield curve can't be a risk premium if there's almost no risk," says Marcus.
Marcus and Grieves' study also suggests that it doesn't pay to mess around with a good thing. When they tried applying minimum slope hurdles to mismatch trades, the results were worse than trading whenever the slope was positive. While imposing a spread hurdle eliminated trades with smaller expected payoffs, qualifying trades did not have a higher success rate, and overall returns suffered.