“It’s difficult to apply historical down-shocks to the current low interest rate environment,” said Will Doerner, “and models have problems in the low interest rate environments of today.” Doerner is Senior Economist at the Federal Housing Finance Agency (“Agency”), and was the first presenter at a GARP webinar on how to generate historically-based interest rate shocks, which was held October 28, 2014.
An accurate estimation of market risk helps financial institutions determine the amount of capital needed to withstand adverse market events. Interest rate changes represent a key factor for institutions with large fixed income portfolios. As such, when stress testing, estimated market value of equity changes will be inextricably linked to the shocked interest rate curves (for example, Treasury, Agency, and LIBOR-Swap) included in each stress scenario.
Doerner contrasted two current methods of introducing shocks in order to calculate stressed market risk: proportional and absolute shocks. Proportional shocks apply a multiplicative factor to existing interest rates. For example, he described a case with a 31 percent change to the 1-month LIBOR rates. However, the shock was significantly dampened when applied in a low rate environment, he noted, “ultimately translating into an up-shock of 7 bps.”
“In a high interest rate environment, the effects are amplified,” Doerner explained. “In a low interest rate environment, the effects are muted.”
A drawback to proportional shocks is that they could overstate the magnitude of rate changes, leading to implausible shocks.
The second method is to apply an absolute shock, namely, shifting the interest rates across the board. “They’re transparent. They’re easy to calculate,” he said. However, the drawback are that absolute shocks can lead to negative spot rates, “multiple ‘kink’ points”, and “implausible credit spreads such as Treasury yields being greater than Libor-Swap yields”.
For this reason, Doerner and his co-author Bogin turn to yield curve parameterization. Fortunately, there is “extensive research to draw from” and their research builds on a flexible type of parametric expression for the yield curves, namely, “non-linear Laguerre functions of time to maturity” containing a “polynomial multiplied by an exponential decay factor.” ª
Click here to view the webinar presentation on modeling stressed interest rates. Doerner’s section covers slides 1 to 18.