Wednesday, November 5, 2008

Nine Measures of Inflation

I have been thinking/reading about core inflation a lot lately, but I wanted to wait for the November inflation before posting this.

To sum up: I am growing more and more disturbed by the CBT’s focus on core inflation. Not because the CBT is the only central bank that looks at core inflation. In fact, many central banks have their own version(s) of core inflation, as there are at least two plausible rationales for policymakers’ focus on core inflation measures: First, food & energy prices and the like are subject to international supply shocks, about which a central bank can not do much, and are therefore more volatile. Second, core inflation is thought to be a better predictor of total inflation over longer time periods, which are of more importance to policymakers. So, it could make sense to concentrate on core inflation.

My concern is that the CBT has not provided us enough evidence on either of these two claims.- or other justifications for using core indices, for that matter. I am not worried much about the first claim; a casual look reveals that food and energy are among the most volatile components of Turkish CPI. However, I have not seen any work from the Central Bank that shows core inflation as a better predictor of overall inflation.

Even more alarming is the Bank’s somewhat discretionary way of choosing which core measure to follow closely. For a long time, the Bank’s favorite core measure was H, -excluding energy, unprocessed food products, liquor & tobacco and gold-, and then a new measure – I , excluding energy, food & beverages, liquor & tobacco and gold, - popped out of the blue a few months ago. I was hoping for a well-written research piece from the bank’s able researchers or at least a box in one of the inflation reports explaining the rationale for the new measure, but I guess I was too greedy.

To give the CBT credit where it is due, its introduction of the core indices has not been totally out of the blue. The indices were introduced in early 2005 along with a Report on Core Inflation (prepared jointly by SPO, TURKSTAT and CBT) that explains and justifies these indices. The choice of different indices has been, in retrospect, quite useful, as these have given the Bank some flexibility to switch to another index as the index it most closely follows becomes less relevant because of temporal or structural changes. I would argue that the Bank could have done a much better job in communicating the rationale behind the switches from one index to another, but its avoidance of the term core for these indices (the CBT calls them special price indices) hints that the Bank has not been all that sloppy…

To illustrate what I have been looking for from the Bank, I recently stumbled upon a FRB Philadelphia working paper, where the authors go over the two rationales I have outlined above and find little support for either. A similar paper from the CBT explaining the rationale behind their choice of the specific core indicator would have made me very happy.

Incidentally, I have recently found out that my wish is close to being realized: Two economists from the CBT’s research department are putting the finishing touches on a paper, which, essentially, creates a core index every month by taking out the most volatile components of the CPI index. Unfortunately, while the methodology is in fact less arbitrary than switching from one index to another, is unlikely to find much practical use because adopting it would make the CBT prone to attacks that it is changing the core index arbitrarily. I could not access the draft because of confidentiality agreements between the CBT and TURKSTAT, but the authors have promised me to email it to me as soon as it is published as a BIS working paper in a couple of months, so you’ll have to hold out for my complete review of the paper, but it is exactly this kind of work I’ve been expecting from the CBT to justify their core (or special to use CBT-speak) indices.

No comments: