An update on monthly consumer prices

A little more is going to be disclosed from next month.

One of my first posts here was on the absence of an official monthly consumer price index in New Zealand. While that makes us an outlier among the OECD countries, it’s partly ameliorated by the fact that a fair amount of the CPI is already available on a monthly basis.

Today Stats NZ announced that they will be going a bit further to fill the gap. From next month, they will be expanding the range of price series that they publish on a monthly basis. As well as the current food and rent price series, they will now include alcohol, cigarettes, fuel, airfares, and accommodation.

They describe this as extending the coverage from 30% to 44% of the CPI. However, as I noted in my earlier post, fuel prices are already available from MBIE, and some of the items in the CPI only change once a year. So on my cobbled-together monthly index, the coverage rises from 40% to about 50%. (It varies over time though - the weighting of air travel dropped sharply during the Covid border closure.)

Here’s what the updated “50% of the CPI” index looks like, and how well it corresponds with the full quarterly CPI. For now I’ve just used the quarterly values for the additional series; Stats NZ will be providing backdated monthly series at the next release.

As you’d expect, the fit has improved a bit, and in particular it’s helped to dilute the impact that big swings in fuel prices have had on the recent history. But there are still some important gaps. The reason that this monthly measure typically runs higher than the quarterly CPI is that it’s missing manufactured goods, which over time have tended to see muted upward pressure on prices, or even price declines (some of which is due to quality adjustment).

What’s good about these new additions is that they are quite variable, which means they account for a disproportionate share of the variation in the CPI. For instance, while airfares currently make up 2.7% of the CPI, their contribution to changes in the index is typically larger than that. (They’re also extremely seasonal, though you can get around that by looking at year-on-year changes.) In contrast, rents make up almost 10% of the CPI, but the way they’re measured means that they evolve quite gradually, so they don’t explain much of the variation in the inflation rate.

Adding these extra series is a relatively easy win, because Stats NZ was already collecting them monthly (or even weekly) and averaging them over the quarter. Any further development of a monthly CPI would require additional price collecting, which in turn would require more funding - we’ll find out in next year’s Budget whether the new government has an appetite for that.

I also noticed that Stats NZ is seeing quite a sharp drop-off in response rates, in cases where prices are collected by sending surveys to businesses (about 30% of the CPI). I don’t know at what point this becomes a major problem, but administrative burden is something that we need to take seriously in the clamour for more, and more frequent, economic data.