Thursday, 17 July 2014

Calm down, folks

Yesterday Stats told us that inflation in the June quarter was 0.3%, a little less than the 0.4% that forecasters and the Reserve Bank had expected. And out came all sorts of OTT reactions, mostly along the lines that the RBNZ would now hold off on some interest rate hikes, or had been running too tight a monetary policy all along. The kiwi dollar dropped a cent against the US dollar.
Puh-leez. Let's get a grip here.
One. Even in a world where Stats measured the CPI with complete precision, there is no real difference from any perspective whatever between a 0.3% inflation rate and  a 0.4% rate. Especially as the numbers have the capability of setting off entirely spurious tripwires because of the rounding process. Yesterday's "0.3%" was actually 2.517%. If the CPI index had been the teentsiest bit lower, it could have been 2.499%, in which case we'd have had a headline rate of 0.2% and forex dealers leaping to their deaths from their dealing rooms. All on the basis of the third decimal point in a percentage change between two large numbers. Give me a break.
Two. Stats doesn't measure the CPI with complete precision (and this, I think, is going to be my only original input into the conversation). The CPI numbers come from a survey, and just like opinion poll surveys get reported these days with their "margin of error", the CPI number has some uncertainty around it. Somewhat oddly, after I read all the CPI release stuff, even the rather boring and technical (but important) bits way down the back, I couldn't find what the CPI's margin of error was.
So I rang The Man, who in this instance is the ever helpful CPI guru Chris Pike at Stats (and everything from here on is my view, not his, by the way). It transpires that I hadn't made a cockup of browsing the Stats website: there aren't, in fact, published estimates of the margin of CPI error.
The reason is that the CPI isn't a random survey, where you can use standard statistics to estimate sampling error: it's a purposeful survey that decides on certain outlets and certain product lines. But just because we don't know exactly what the survey error is, that doesn't mean there isn't any. There surely is, for example because the CPI relies on other surveys such as the Household Expenditure Survey which have their own sampling errors. The HES might say, for example, that households spend 23% of their income on food, and Stats will use that as an expenditure weight in the CPI,  but it might be 22%, or it might be 24%. And of course the CPI will have its own survey errors as the price collectors wander through the aisles and misread the label on the baked beans. And so on.
So not only would a real difference between 0.3% and 0.4% not matter a damn,  there may not even be a real difference in the first place. Allowing for survey error, 0.3% and 0.4% could well be statistically indistinguishable. I don't know what that margin of error is, but it's highly plausible that numbers only 0.1% apart fall within it.
Three. Even if 0.3% is absolutely beyond doubt the right number, and definitely different to 0.4% or 0.2%, the over-excited "lower than expected inflation" and "pressure off the Bank" and "overzealous inflation fighting" reactions don't make much sense to me. Brian Fallow, as is his balanced wont in his article in the Herald, made the key points: the headline figure is being flattered by the high NZ$, which lowers import prices,and in any event it's more important to look at non-tradables inflation (the inflation in purely domestic sectors, like most of the services sectors). And there you don't see inflation at a comfortable level. The CPI may be running at a 1.6% annual rate. But non-tradables inflation, which for monetary policy matters more, is running at a 2.7% rate. Even if you take out the housing and household utilities bits (which are running hot at the moment), core non-tradables inflation is either 2.2% (ex housing and household utilities) or 2.6% (ex purchases of new houses).
I've gone on about this before (here and here, for example) but our domestic non-tradables inflation keeps trundling along, as the chart below shows (the hump in 2010-11 is the effect of GST going from 12.5% to 15%, so don't pay that too much mind), and generally towards the top of the RBNZ's 1%-3% target band.

It's pretty evident that as and when the kiwi dollar goes for a burton, we're going to see some very ugly headline inflation numbers indeed, unless the non-tradables inflation rate drops, and so far it's shown a remarkable ability to hang in there. I'd also point out the sizeable and unhelpful contribution of central and local government charges to the non-tradables inflation. Okay, there's an element of sin taxes in there that maybe I can go along with (ideally if it's taxes on the other fellow's fags and not on my Côtes du Rhône). But it looks very much to me as if large swathes of central and local government still feel as if they can write any number they like on the price ticket.
Bottom line - park all that guff about unexpected victory in the battle against inflation.

No comments:

Post a Comment

Hi - sorry about the Captcha step for real people like yourself commenting, it's to baffle the bots