Policy discussion makes up between 5% and 70% of central bank minutes, study finds

As the European Central Bank (ECB) grapples with the question of what to include in the "accounts" of its governing council meetings, a Central Bank of Ireland (CBI) study shows there are a great many differences in the scope and characteristics of central bank minutes.

Market participants, unlike the general public, often pour over minutes of central banks' monetary policy meetings in search of clues about future policy moves.

Some central banks make that routine easier than others. While the practice of releasing minutes has caught on in recent years, there is no common standard, and a majority of central banks still do not publish them.

With regards to length, minutes range from 600 to 13,000 words, according to the Irish study, which draws on a sample of 24 banks. Of course, adding length, the authors point out, "may not enhance the information content", but could instead "act to hide the most important issues raised in a meeting".

The proportion of minutes dedicated to policy discussion, as opposed to rehashing economic forecasts, varies greatly too. One central bank used 70% of its minutes to depict the back-and-forth between committee members, another just 5%.

There appears to be greater consensus on the importance of getting minutes out in a timely fashion – perhaps because, as the study notes, the "length of the publication lag can affect the amount of market-relevant information in the minutes". All 24 central banks in the study published minutes within eight weeks of meetings, while 17 even released them within two.

A bigger bone of contention concerns whether or not to attribute comments and how best to do it – something that has "shown to provide information that can help market participants anticipate future policy decisions", according to the study.

The ECB plans to introduce "accounts" of its governing council discussions in the new year, but is yet to decide how to deal with this issue of attribution. The Irish study reveals there is not much in the way of consensus on the topic.

"We see three broad categories: two central banks assign names to comments within the discussion, 13 give different positions, but make them anonymous and nine simply relate a generic commentary of the discussion, with no specific ‘views'," the authors note.

On the other hand, the attribution of names to voting is much more common, with 11 central banks reporting voting records of individual members. Four central banks report voting records "quantitatively" while seven give no information at all about voting. Fourteen provide an explicit explanation of the reasons for dissent.