Transcript of Question & Answer Session The Current Global Expansion

Is there a microphone?

No, I don't think so.

You'll just have to shout.

My question is, you mentioned in the global perspective, that monetary policy needs to be stimulatory if inflation remains low before it picks up. How would you explain that in the Australian context?

Just for the recording, the question was, I stated that globally, monetary policy needed to stay expansionary while inflation remained low, and then I was asked to describe what that means for Australia. My answer is that I'm not going to speculate about the forward path for monetary policy. I don't have anything to add beyond what was in the minutes yesterday.

Tanya's just asked for me to talk a little bit about the word clouds and how we did the sentiment analysis. And I'm going to demonstrate something Pip said. We did this in Mathematica. There is actually a sentiment analysis package and a word cloud function in Mathematica, so this is only about four lines of code. It's disgusting how easy it is nowadays.

One of the things we've learned about looking at sentiment analysis for various things is it doesn't work … these things are all calibrated on social media posts, so the standard packages that are out there do not work well for central bank communication, which is far too nuanced. [Laughter] I don't know why you think that's funny.

So what we actually did, is we went through and broke up … You strip out all the simple words like "and" and "but". There's a footnote in the written version of the speech that talks a little bit about it, but I have to admit, the way we actually coded for this particular word cloud, is we just coded the words individually because it was all pretty obvious. I dithered on whether inflation was good or bad in the current environment. I decided to code it in red rather than blue. But you can kind of see, yes, there's words like "recovery" and "increase" and "growth" in the April 2016 chapter, which are used quite frequently, but "weak", "decline", "low", just even the fact that some of the top words, "decline", "low", and "growth" in April 2016. Then in April 2017 it's "growth" and ‘increase’ and ‘strong’ and "inflation" and "recovery", "high", "support". It was pretty obvious how to code it manually for a small data set.

I hope that over time … There are some other central banks doing more on this. There's a great Bank of England paper on text mining. So I think these tools are going to start getting used more frequently as people gain experience. This one was our first dip our toe in the water, but it is something we're doing a little bit more of and experimenting with to see if it … We wouldn't use this for predictive content, but I thought it was a really good way to illustrate the point. We tried a couple of different ways to do sentiment analysis to demonstrate that the rhetoric really had changed over those twelve months, and I thought this was a good way to show it.

Up the front here.

I'm wondering, just on that topic, whether you've actually made use of search terms. That's sort of a similar idea, but in a sense, crowdsourcing what are people interested in. It may not be quite as simple because you might not really think of people searching growth does not necessarily mean that the economy's growing. Is that something that you are thinking about?

We've been dipping our toe in the water. It's not something that we've integrated in an ongoing way into our normal Board briefing or anything like that. I haven't showed the Board … I don't think I'm giving away too much to say I've never showed the Board a word cloud, but I think what we're generally seeing is here, there are uncertainty indices that various other analysts have put together. There's a range of different things that people are now thinking are possible. We certainly have used the Google Books word searches and the Google n-grams, but again, it's not something that we'd ever put in a regression. It's more like: here's another illustration. It's for exposition rather than for analysis at this stage.

(off microphone)

For the benefit of the recording, someone asked why wages growth globally hadn't really picked up even though economies are looking a lot better. The Governor talked about this in some detail in the most recent Parliamentary testimony. There are a number of things going on, and I don't think we have a completely nailed down answer. I think this is the thing. What we do involves uncertainties. We'll never be 100% sure, but a couple of things are going on that we think are relevant. One is, most of these economies, if they've gotten to full employment, it's only really recently happened and lags matter.

We have actually seen wage growth pick up a touch in the US and in Japan, which is the other country where they're obviously closer to full employment or at full employment. You don't see it in the regular workers who are ongoing, but you are seeing a reasonably noticeable pick-up in wage growth in part-time and non-regular workers. So there's a real two-tier labour market in Japan, and we are seeing that. And certainly, talking to our counterparts in Japan, they regard the economy there as pretty hot now in terms of what it's doing to the labour market. Which is great, because if you look at their age-adjusted participation rate, they're bringing people into the workforce relative to what you would otherwise expect. It's a great outcome after such a long period of quite low growth there.

There is a bit of pick-up in wage growth, just not very much. Lags probably have something to do with it. And, as the Governor has said before, there probably is also just the psychological thing of a sense in which the world has become more globalised, your competition could be all over the world with crowdsourcing and all of these other things. A lot of people may well feel a bit more insecure in their jobs. We know that in Australia, voluntary quit rates have actually come down. So there's something going on, and it's something that we're looking into on an ongoing basis.

For the benefit of the recording, I'm not going to be able to repeat back all of that question, but essentially, I was asked for my views on the property market, particularly in Sydney, and whether the increase in supply of apartments would weigh on prices. You're quite right that we're now at a point where, after a long period of population growth having increased significantly in the mid-2000s, beyond housing supply's capacity to meet that, housing supply has now come on and is now very high and, if anything, we are now either eating into an under-supply or building up an over-supply, depending on your perspective. And that's particularly concentrated in apartment markets. Over the years, we've pointed more to Melbourne and to Brisbane as where that's more likely to cause trouble in terms of concentrated increases in supply, resulting in lower prices and therefore potentially, distress. Because they were much more geographically concentrated increases in supply, whereas in Sydney, it's all over the city. So it's much more evenly spread, and it's less likely to create a really concentrated local price effect.

A couple of things that have happened more recently that we've now got to the point of peak building approvals. Building approvals have stepped down to a new level. The backlog, or the pipeline of work yet to be done in residential property is now starting to decline, even in New South Wales and Victoria, but particularly in Queensland. So in some sense, this is a phase that will eventually end and we won't be building more than the population requires. I think one thing that probably at the margin makes me feel a little bit more confident about Melbourne than I would've say, three years ago, in terms of this risk, is that population growth into Victoria and Melbourne in particular has been really strong and that's not something that was envisaged earlier. So, in some sense, the demand's increased and so there's less risk of over-supply there. Similarly, population growth remains strong in Sydney, so it's not clear to me that we'd end up with a serious problem here because of the geographic spread.

Where we've been most concerned is Brisbane, because again, it was a concentrated geographic spread of the apartment market. A lot of that new supply is coming on this year, so this year is crunch time. The pipeline of work yet to be done in Queensland has actually come off quite a way. It's a bit more in balance there, but the properties that are more likely to fall in price there are not the ones that have newly been built, because what will be happening is tenants will move from the five-year-old apartment to the nice new one that's even nicer, because rents are staying low. So it'll be that five year old, not new anymore, second-hand apartment market where you might see some people make losses. But again, housing prices haven't been growing particularly strongly in Brisbane for a while, and particularly apartment prices. So, yes, it's a truism that the more supply you have, the weaker the prices are going to be relative to what would've happened if supply had been lower. But at this stage, we're not seeing that as being a serious … that supply in and of itself is going to create a big fall in prices and certainly not in Sydney.

More generically, price growth had been very strong in Sydney late last year, and also in Melbourne. And as we've said in our public statements, both the housing and apartment markets have slowed a bit in both Sydney and Melbourne. More so in Sydney than Melbourne, but prices still rising at seemingly rapid clip nonetheless.

I was just wondering if I could ask a conceptual inflation targeting question. How would you think about the appropriate stance in policy in a world where inflation continued to undershoot target but unemployment was low and was falling and became quite low?

Well I think that kind of depends on what that tight labour market is doing to labour costs, and therefore to prices. That's the clear nexus. Look, full employment is in our Act, so if we have inflation and full employment, that's great. I think the other conceptual point I'd make is that the inflation target that we have is an on average over the medium-term target. So the Board can make a choice about, if it's outside of the band, it has a choice and can make a judgement about how quickly it wants to get back. So I think the answer would depend on where you thought your forecasts were going.

If, in that scenario you described, our inflation forecasts were, you know, we were starting below two and it was going down, obviously that would be a very different policy scenario to one more like the forecast that we've recently published, which is where actually, slowly, they're going up. So the question is, do you want to go faster than that or not? And as the Governor said in the past, I think they're comfortable with the speed at which we're getting back.

(off microphone) [You say you shouldn't wait too long, but what are you waiting for?]

Wait too long to decide that the global economy is strong. I think that's the thing. That's what I'm talking about here about waiting too long. It's waiting too long to form a view about something. It's funny you should mention, that comment by Ian Macfarlane, because that was actually one of the things that I was thinking about when I decided to pick this topic for the speech, was what he said. I probably should've put the quote in but I couldn't remember where it was.

I couldn't find it either.

Yeah, I couldn't find it either, but I remember it too. But yeah, I think we need to be mindful about what's going on in the global economy, but I guess what I would respond to that is, if I had two pieces of information, the other one would be the amount of slack in the Australian economy. Our assessment is, there's still a fair bit of slack in the labour market.

One of the things we've been wrestling with over recent times is exactly how much there is. Is the unemployment rate all you need to look at, or how important is underemployment? Does it matter now that the labour market's become more flexible? Does that mean that the lags between a tightening labour market and labour costs, and therefore prices, will be slower and weaker? They're the things that we're wrestling with at the moment. So yeah, I think the first point I make is, my topic today I chose because the global economy is a really important influence on the Australian economy. And of course, the previous Governor was talking quite some time ago …

Two Governors ago.

Yeah, two governors ago, exactly. In that respect, I think one of the things that has changed in recent times has been the increase in the role of the resource sector in the Australian economy. That wasn't the case when Ian Macfarlane was speaking on that occasion. Firstly, things have changed. Secondly, the amount of slack really matters. And I think that's also an important consideration.

(off microphone)

Absolutely. It's an excellent question, that given all the economic community, more generally, consistently make certain mistakes, and turning points are sort of impossible. The Bank does have a process of annual forecast reviews. I think there have been some things lately that we've kind of been getting reasonably. If you look at particularly in that sort of now-cast, and very short-term forecast, we've been reasonably okay on that, just lately. But one of the things that is very difficult is things like investment. I think there was a long period where we were forecasting a much stronger increase in non-mining business investment than actually transpired. Although, it is now increasing. I feel more comfortable forecasting a recovery once it's actually started, I think is part of the issue. And that's the thing. You might not be able to forecast the recovery or the turning point adequately, but at least if you notice it quickly … And that's really the moral of my whole talk. So we've certainly learnt a bit about that.

And I think one of the things we've learned, and as you know, I spent ten years away from monetary policy and came back at the end of last year. So the things that Rip Van Winkle has learned, among them just how important the mining sector and the spillovers from mining to other sectors turned out to be particularly for certain parts of the business services sector. I think that was probably something that we initially underappreciated and have a better appreciation of this time around. And that judgement of all the terms of trade did come up a bit more than we expected. Forecasting commodity prices is at least as fruitless as forecasting the exchange rate. That's always going to be a big swing variable that we are going to struggle with forecasting. At the moment, we do expect that iron prices will step down from where they are, but I don't think we were really expecting for them to swing up quite so much.

So understanding industrial structure, understanding what's really going on with the Chinese steel market, understanding those industrial organisation things is something that I think central banks and other economic forecasters are doing much more of than they did when I was last doing monetary policy things. Up the back, there was a question.

(off microphone)

Thanks. That was Sophia Rodrigues asking about the importance of India, and particularly to the iron ore and steel markets in the future, relative to where it is now and where China is now. The reference is to a passage in the minutes where we described a little sequence of graphs that I showed the Board on the iron ore market and the steel market, and what we think some of the drivers there are. As well as having special papers, sometimes we'll have a bit of a special theme in the presentations. There's nothing particular to infer from the timing of that, other than it was a good time to do it and the team had done the work. So I showed it to the Board earlier this month.

The essential message is that countries tend to go through a phase of development where the amount of steel per capita they consume increases. And that goes up as income per capita goes up. Then there's this sort of inflexion point where it flattens out. China is exactly at that inflexion point. China has been completely typical relative to many other countries, now industrialised countries, in terms of its steel production or consumption in particular. What we infer from that is, China is the overwhelming customer for exported iron ore. It has some of its own, but nowhere near enough to meet its domestic needs. So if you want to know about the export iron ore market, you need to know about Chinese demand. And we infer from this sort of relationship that we're quite likely to be close to the point of peak steel, in terms of Chinese demand. They're now at the point where their steel production per capita is likely to level off. And given their demographics, that means the absolute amount will probably also level off. Maybe in the next couple years. That's a medium-term forecast, not a short-term forecast. At the moment, I think, the dynamics are actually a bit more positive than that.

Turning to India. India doesn't produce anywhere near as much steel. It's actually a little bit below the curve that you tend to see in other countries, given its current income levels. I think that's partly because India went straight to a services-oriented economy rather than having that real industrialisation phase that many other countries went through. So it may well end up being a bit below the typical curve of steel production per capita. Then again, it may change that, because a lot of the steel is in infrastructure and residential construction and things like that. That's where it goes into, so you could see a bit of a fillip there, but I don't expect it to be quite as heavy industry-oriented as China has been. Given its income levels, it's nowhere near that peak steel per capita production.

On the other hand, it has a lot of its own iron ore. I don't know enough about its iron ore reserves to know how much it would seek to import iron ore, should its steel production rise a lot. It seems likely, given that the two low-cost producers in the world are Australia and Brazil, that if it does, Australia will be an important supplier. But I think a lot depends on how India's industrial structure evolves over the next couple of decades. It's growing, it's population's still growing, so it's still a growing market. But it does have a lot of its own iron ore, and I think that's an important difference with China, which has some iron ore, but not anywhere near enough to meet its domestic needs. Thanks for the question.

(off microphone)

Yeah. Well, this is just simply the first chapter out of each of the IMF World Economic Outlooks. The next one comes out in October, so I don't have an update. Instead of just taking whatever everyone's views are, we wanted to do something simple for exposition purposes today, so we just took the first chapter of the World Economic Outlook. And that, then, means that you're not getting messed up by different people speaking, and someone's bullish and someone's bearish. You've got to worry about who your sample is. We're just saying, "This is what the IMF was saying." You get similar things if you had a bigger sample, but for the purposes of today, I just wanted to do something simple. But as I said, we're dipping our toe into the water in this, and experimenting a little, and that's how adoption of new technology works. It takes a while for us to really integrate it into our work.

(off microphone)

That is a fantastic question. For the benefit of the recording, it was how we integrate social outcomes, and particularly housing related and intergenerational housing outcomes into our economic modelling. I've got a two-part answer to that, and the first is, although it's not something you can easily quantify in an econometric equation, those sensibilities really do matter for our financial stability policy. One of the reasons why we regard it as a concern, the household balance sheet situation, is it's not just, well if there's an economic shock, people who've got a lot of debt will spend less than if they had less debt and that's therefore bad for the economy in a macro sense. We also believe that it matters if that 1 percent fall in consumption, just to take a random example, is spread across everybody, or whether it's because 2 percent of the population are in serious financial distress. So that's the social element. The third part of our charter, it's stability of the currency, full employment, and the welfare of the Australian people. You can't always put that into econometrics, but it is a sensibility that we have, particularly on the financial stability side.

The difficulty we have on the monetary policy side is that we have one interest rate. Other than our open mouth operations, all we can really do is set an interest rate. So a lot of those particular policies are beyond our remit, and there's a fine line in terms of opining on things that are outside our remit. But what I will say is again, one of the things that the monetary policy Rip Van Winkle has learned over the last decade is just how … Over the last ten years, microdata, that really disaggregated data, has become much more important in developing our macro understanding. You don't just look at the macro aggregates anymore. That's 90s thinking. Nowadays, a modern central bank will be looking at the macro data, will be going over the word clouds, but also will be looking at microdata on businesses and on households. And you know, we've got the HILDA survey. The ABS is doing some amazing things with data linkage at the moment. They've got their BLADE database, which is like a longitudinal database of businesses. The census is actually much richer dataset than a lot of people appreciate. All of these bigger datasets, are now becoming integrated. I wouldn't say that journey isn't complete, but they're now becoming integrated into our general analysis. You see a lot of microdata.

In that respect, you might've seen one of our research discussion papers that just came out last week, was about using microdata to think about people's entry into home ownership, and how that's changed over time. And it's looking at quite fine-grained distributional concepts. So absolutely we think about it, but it's not necessarily in our core forecasting equations.

Okay, there's one up on … How are we going for time?

I think we've got one for … Okay, so we've got a couple of students. We'll start there, and then the last one over there.

Okay, great. Fantastic. Great.

(off microphone)

Not specifically. The team are currently poring over some of that data at the moment. One thing that the household expenditure survey does provide is new weights for the CPI, so I think the core thing for us is to see how the CPI weights will change. That will be announced in coming months and will take effect for the December quarter CPI release, and I think that's the thing that we'll be paying first attention to. But again, often you can rummage through this data for months and find more interesting stuff. That's the nature of the microdata. And finally over here.

(off microphone)

Sure. Someone's been reading this speech online. He just asked me to talk a little bit about the slack in the labour market here. Essentially, it's just that unemployment's 5.6, and our current best estimate of full employment, in the sense that it's the level of employment at which wage growth doesn't pick up, is about five. Plus, there's probably a bit of underemployment in the sense that there are people who are currently working part-time who would like to work more hours. Not necessarily have a full-time job, but at least have more hours. Plus, one of the things we've noticed is that we've had very strong employment growth in the last six or so months, but a lot of that … The participation rate has come up more so than the unemployment rate has come down. So clearly there are other people who, if the jobs are available, they'll be willing to take them even though they're not officially unemployed when the Bureau of Statistics comes around and asks them have you been actively looking. But they're available if something comes up.

So it seems to us that there is a fair bit of spare capacity, particularly in the labour market. As a consequence, businesses don't generally find they have to start paying higher wages to people until that spare capacity is pretty much eaten away, so that's why we would regard wage growth as being likely to only rise fairly slowly from here, and that in turn is an input into our inflation forecast.