The Retriever has found another juicy bone. He (or she?) writes in raising a number of points about credit spreads, questioning my suggestion (following John Vickers’ assertion) that credit spreads rose in 2018 ‘Largely, if not wholly, because corporate credit risk has gone up’.
Retriever raises a number of interesting points. For the moment (and in a subsequent post) I will focus on the challenge presented to our case by the Bank of England’s structural credit model. As Retriever points out, Chart E of this 2016 inflation report (page 13 of the PDF, and copied above) suggests that the Bank of England thinks that the illiquidity premium is something like 50% of the corporate bond spread.
Presumably you must think that either 1) the Merton model is inappropriate here; 2) that the BoE has parameterised the Merton model incorrectly; or 3) that this “residual” bit of the spread is accounted for something other than credit and liquidity.
It’s interesting that Retriever raises this question, because the Bank’s credit model, together with its implication that there is a substantial illiquidity premium, was central to the discussions about the Matching Adjustment around 2013-14. The model did not originate at the Bank. As the authors (Rohan Churm at the Bank, Nikolaos Panigirtzoglou, then at QMUL) state, 1 it is an implementation of the Leland and Toft model developed in 1996, which itself is a development of Robert Merton’s credit model published in 1974. 2 The Churm-Panigirtzoglou paper has a detailed specification at end which anyone with enough patience can code up, and anyone is welcome to our Eumaeus version via our contact form.
There is a lot more to say on this subject than I have time for here, it is sufficient to note that the validity of the model depends crucially on its calibration. As Annaert et al state here
Decompositions of bond spreads have been advanced in order to empirically disentangle the credit part from other residual risks (Webber & Churm (2007); Churm & Panigirtzoglou (2005)). The usefulness of these decomposition attempts depends crucially on the heroic assumption that the credit risk model is well specified and it gives an ‘ad hoc’ interpretation to the residual part. [my emphasis]
In my next post I will examine the question (following Retriever’s case #2 above) of whether the BoE really has parameterised the Merton model correctly.
- ‘Decomposing Credit Spreads’, Bank of England Working Paper No. 253, Rohan Churm, Nikolaos Panigirtzoglou, 2005.
- Merton, Robert C. “On the Pricing of Corporate Debt: The Risk Structure of Interest Rates”, The Journal of Finance, Vol 29, 2, 1974.