How it started, how it's going

An update on some previous posts, and a quick request.

Photo by Kate Macate on Unsplash

This week marks the one-year anniversary of Local Knowledge. So far I think it’s lived up to its fairly modest ambitions - as I’ve said before, the goal was to try my hand at writing about a slightly wider range of topics, but hopefully with enough readership to reassure me that I wasn’t just shouting into the void. Thanks again to those of you who have subscribed, and have been curious enough to at least open the emails.

One of my particular goals in the early days was to write some pieces that would have a bit of a shelf life, setting out ideas that I could refer back to over time. In this post I’m going to revisit a few of those, and discuss any further thoughts or developments since then.

Before that, though, I have a quick request: are there any topics that you’d like to see me write about in the months ahead? Maybe something that you’re not seeing written about elsewhere, or something about the economy (or economists) that’s been bugging you for a long time. Preferably nothing too chunky like “how do we fix our productivity problem” - remember that I’m doing this for free in my spare time, so you get what you pay for.

If you have an idea, just pop it in the comments below, or message me directly if you want to keep it private.

OK, let’s start by going all the way back to the first proper post:

---

I also revisited this one back in October, after Stats NZ launched an expanded suite of monthly price indices. That looks to be as far as this is going to go for now, though. As I noted in the original post, the only real barrier to a full monthly CPI is funding, and the new government doesn’t seem inclined to find new things to spend money on.

I estimated that a monthly CPI could cost around $5m a year - pretty small in the context of a $400bn a year economy - but I suggested that the benefits probably wouldn’t be particularly large either. More recently, I’ve thought about how you might actually go about quantifying those benefits. The answer is not at all obvious.

Firstly, the main effect of timelier inflation data would be that the Reserve Bank might be able to identify turning points sooner, and move interest rates earlier. How much earlier? Policy reviews are typically every 6-7 weeks, and it’s unlikely that it would be much more than that. After three months, there’s no information advantage between a monthly and a quarterly release.

The next question is what would be the benefits of starting and ending monetary policy cycles slightly earlier. I suspect that most people will look at this through the lens of the current situation: if the RBNZ had been able to identify the cooling in inflation earlier, they might have already begun reducing the OCR by now. But keep in mind that the effects of timelier inflation data would operate in both directions - the RBNZ would also start raising interest rates sooner when it sees inflation picking up. So we can’t quantify the benefits in terms of what we’d save on interest payments (netted off from the interest that savers would miss out on).

Moreover, lowering and raising interest rates sooner probably wouldn’t make a difference to the long-run level of GDP. The most it’s going to do is shave off the very tops and bottoms of the economic cycle.

So the answer that we really need is: what is the value of (slightly) greater stability in the economy over the cycle? There are models than can attempt to quantify this, though there will be a wide margin of uncertainty around any estimate. But in the context of a $400bn a year economy, it’s possible that those estimates could exceed $5m a year quite comfortably.

So I’m coming around more to the idea that a monthly CPI would be worth it. And since the RBNZ would get the most benefit, the easiest solution would be for them to put up the funding for it.

---

The GDP ‘nowcast’ model that the RBNZ unveiled in their Monetary Policy Statement last August hasn’t appeared in their last two documents. They may still be using it internally, and indeed that’s where it would have its most value. The point of a ‘nowcast’ is that you can update it at any time, to incorporate the latest available information. Producing it only once every three months would make it no different from any other forecast.

At work, I’ve been toying with some nowcasting methods of my own, which we’ll be publishing weekly once we have a version that’s ready to go. I have no real experience in coding, so whatever I come up with is going to be far from the cutting edge of the work that’s being done in this area. But it’s something I can hopefully refine over time.

I don’t expect that a nowcast will produce more accurate forecasts of GDP; after all, it will use much the same data that we’ve always used for this purpose. Where I see the benefit is that it puts some structure around the question of how much you should update your view as new information arrives - the sort of Bayesian approach that I also wrote about in my post “Becoming a better forecaster”.

---

This is probably the most widely-shared piece I’ve written, although I should point out once again that the credit for this work goes entirely to Alexandra Ferguson at Stats NZ. The upshot was that there can be a persistent gap between the Jobseeker benefit numbers and the official unemployment statistics because they’re two quite different things, with much less overlap than you might think.

For those who were inclined to see this gap as evidence that the official figures were being massaged, you may or may not be pleased to hear that the gap has been narrowing again for the last few years. Recently, that’s been because the official unemployment figures have been rising faster than the benefit numbers. The narrowing also pre-dates the change of government, which makes it less likely that it’s to do with MSD’s enforcement of the job-search requirements.

---

In this post I showed that the unemployment rate is our best indicator in real time of the state of the economy. At the time, the unemployment rate was 3.9% - a touch stronger than a ‘neutral’ setting, which I put at a range of 4% to 4.5%. That was also about where the RBNZ saw things, and quite different from the IMF’s estimates which had us as the most overheated economy in the developed world.

As we learned last week, the unemployment rate is now 4.7%, and rising at a faster pace than a year ago. That puts it above my estimated ‘neutral’ range, so if I’m right about the latter, we should be seeing an easing in domestic inflation pressures now.

That is indeed what’s happening. Non-tradables inflation has come off its highs, although it has been held up by some items, such as insurance, that have little to do with the strength of the economy. It’s a similar story with wage inflation, which has clearly slowed if you exclude the impact of collective pay agreements in health and education, which were set some time ago and don’t reflect the current balance of power in the labour market.

One reason to pay attention to IMF estimates is that they serve as something of an impartial observer, applying the same method consistently across countries. But the downside of a one-size-fits-all approach is that it can miss the nuances of an individual country’s economic structure, its institutions, and its data. In the case of the output gap, I think we can chalk this up as a win for, well, local knowledge.

---

Surprisingly, this is my most-read piece to date; it seems that my readership veers more towards ‘data nerd’ than I realised. (Less surprisingly, this one on migration is a close second.)

Tying in to the previous piece, one of the difficulties of gauging the strength of the economy in real time is that the GDP figures are always subject to future revisions, sometimes for years after the initial release. These revisions are unavoidable and even welcome, but they do mean that even our premier economic indicator needs to be treated with a dose of caution at first.

There isn’t much to update here; the really big revisions to GDP come with the September quarter results, which will be out in mid-December. I do wonder if there’s something in the way that GDP is measured that hasn’t picked up on the record population growth over the last year or so. If so, that would go some way towards explaining why our per-capita performance seems to have been so awful lately. But at this stage, we have no real way of knowing whether this year’s revisions will be up or down on balance, and I know better than to make a forecast based on hope.