Core Path · L5.4.08

Shadow rates at the lower bound

14 minShadow rateLower boundBlack 1995KrippnerBank Rate

By the end you will be able to

  1. 1Explain Black's shadow rate, r = max(s, floor), and why the floor turns the short rate into an option.
  2. 2Price a bond under the floor two ways, by Krippner's approximation and by a seeded Monte Carlo, and say how close they are (0.37 to 0.51bp at 5 years).
  3. 3Fit a monthly shadow rate to the Bank of England's OIS curve in the three lower-bound episodes, and read the months where it fails.
  4. 4Show that the depth of a shadow rate belongs to the model, and why two Bank of England papers put it on opposite sides of zero.

Why it matters

From March 2009 Bank Rate sat at 0.5% for seven years, then at 0.25%, and in March 2020 at 0.1%. A central bank at its floor still eases policy, by buying bonds and promising to keep rates low, but the policy rate stops moving, so it no longer says how loose policy is. The Gaussian models of this module have the opposite problem: they let rates fall as far below zero as their volatility allows (L5.2.09: Normal volatility and negative rates). Fischer Black's answer (L5.3.08: Normal against lognormal dynamics) fixes both: the paid rate is the positive part of a shadow rate, and the shadow rate keeps moving when policy does. On the Bank's own curve it was −1.93% in December 2012, with Bank Rate at 0.5%.