Interest rate models and central bank corridors

In my blog post last month about interest rate models at the zero bound, I did not consider the effect of central bank corridor policies. I realized that this is an important omission when I looked at the decision of the European Central Bank (ECB) a couple of days ago to lower the main refinancing rate (the central rate in the corridor) by 0.25% and the marginal lending facility rate (the upper rate in the corridor) by 0.50%. Why was one rate lowered by twice as much as the other? The answer is that with the deposit rate (the lower rate in the corridor) stuck at zero since July 2012, the only way to keep the corridor symmetric is to set the upper rate to be exactly twice the central rate. So the marginal lending facility rate will always change by twice the change in the main refinancing rate!

Of course, despite the deeply (biologically) ingrained love of symmetry, central banks can decide to abandon symmetry and move the central and upper rates independently. In fact, the historical data shows that in the first three months of the ECB’s existence, the corridor was not symmetric around the central rate, but since April 1999, the symmetry has been maintained.

Modelling short term rates in a symmetric corridor floored at zero is problematic. The log normal model has problems because it does not allow rates to be zero. Yet it is the natural way to model the proportionate changes in the central and upper rates.

Posted at 1:46 pm IST on Fri, 3 May 2013         permanent link

Comments