The Bond of trust Yielding from central bank transparency
A tired pun, in service of a continued campaign on this blog for increased transparency in central banking. Or rather at the Bank of England in particular.
Previous rants have focused on the fact that the BoE does not yet provide the code, data and forecast judgements it applies to the main model it uses for monetary policy, when it should, although it has been asked to think about it; on the fact that transcripts of MPC meetings were, up until now, erased; and on numerous transparency problems with the forecast and monetary policy-setting.
This post is to point out that the Bank could and should but doesn't yet provide the codes and adjustments and data it uses in order to estimate its yields curves.
These curves are probably and unsurprisingly, the most pored-over charts inside the Bank of England, and outside of it by BoE watchers. Government bonds are bought and sold every day at prices that are discounts on the face value. They are bonds of irregular and different maturities, trading in differing and fluctuating volumes. The yield curve estimation industry is the practice of drawing bendy lines through these irregularly appearing bond prices [leave aside whether there is, truly a single price] and inferring the interest rate on those and intermediate maturity bonds [ie where none were traded]. And, using the same bendy line drawing techniques applied to index-linked bonds, to infer the 'term structure' [the trajectory, or sequence] of expected inflation rates. If you aren't familiar with central bank business, it ought still to be clear now why this is important; it's about deriving what the market thinks is the expected path of its policy instrument and one of its goal variables.
The strategy for drawing these bendy lines through bond prices is set out in Bank of England working papers that are, therefore, published. But it's still an awful lot of work to get from that to a set of codes that would replicate the charts MPC would be looking at. Not least because there are routines for excluding misbehaving bond prices all the time, which observers would find very difficult to replicate.
There are a few arguments for releasing these tools.
Most compellingly, these codes were paid for by the public, with taxpayers money, ultimately, so - using the same logic as post I wrote about the BoE forecasting model - why should they be kept private?
Another point is that many BoE watching conversations will run: ok, what do they think we think they are going to do in the future? And what will they do about that? Validate what we think they are going to do? Or put us right? Confusingly, policymakers want to know about these conversations too! And their responses can be refined if the whole process of thought extraction can be made accurate, common to all, and cheap.
Moreover, by releasing the codes, the BoE could potentially tap into free expertise in the community of code-literate BoE watchers [yes, a non-empty set] and refine what it does over time.
I surmise that - as was the case for the forecasting model - the BoE are worried about becoming a bond-yield-estimation-code-support-facility. But I don't see why it could not successfully say, simply: 'here it is, here is the data, off you go'. Keeping the code and the adjustment routines to itself looks, to those inclined to suspect its motives, like the BoE is trying to dodge scrutiny, and claim property rights it does not rightly possess.
Many top journals require researchers to deposit working codes and data for replication purposes as a matter of course. Why should our central bank not conform to this norm?
This isn't a huge deal. But it is part of an addictive privacy habit on the part of central banks that needs to be kicked. The BoE has been travelling on a journey - like other central banks - of ever-increasing transparency. This would be a small positive step that would fit in with this mission.