Transcript of Question & Answer Session Shedding Light on Uncertainty: Using Scenarios in Forecasting and Policy

Thank you, Sarah. Brenton Caffin from Premier and Cabinet. This might be a slightly odd question, given you’ve just spent the last 20 minutes talking about looking forward. I wanted to talk a little bit about looking backward and the role of scenarios in retrospective reviews. How often or how does the bank go back and look at the choices you made at the time, how it played out and what you might have done differently?

Excellent question. So the answer is that we do and we do it through a few different lenses. So maybe starting with a very big retrospective. The RBA review did a lot of that. So it was a wide ranging review of the whole institution but the monetary policy part of it was an exercise in - partly an exercise in doing that, going back and assessing different points in time. It particularly focused on the pre-COVID years, so around 2016’ish through to 2019 and then also the policy response during COVID and really we learnt a number of lessons from that. So big reviews like that are one way. Other ways that we do it though, every year we do a forecast review. We publish it on our website. It was published in November. So that’s a way for us within economic groups, really assess how we’ve seen things playing out, so the economy, what have we learnt. What did happen the way we thought it broadly would, what didn’t, so that’s part of it. And the Board themselves have dedicated time where they reflect, I think that’s actually really critical. These policy decisions are hard. They’re made, as I said, with uncertain, incomplete information. So think of driving a car looking in the rear view mirror. Sometimes you’re not going to get it quite right, for reasons that might be beyond your control, maybe you should have read something a bit different. So we’re actually very conscious of needing to do that retrospective exercise. It’s really, really critical. Taking it forward and learning from it, improving the models, improving the analysis and hopefully getting better, time after time after time, but it’s a constant process. I don’t think you can ever think that you’re done. I think as soon as you do think you’re done, maybe time to have a go at something else. So - that’s a good question. I think it’s learning - it’s different things at different points in time. I think from the pre-COVID years, the lesson there was - particularly for economic group around being open to different interpretations of what’s happening and different understandings of the economy. So we’re very alive to that. We have a very robust base internally, I can promise you, and we’re also now active in seeking out opinions and inputs from outside the building. So, for example, we’ve established regular roundtable discussions with market economists and with academics so they come into the building, we have the discussion so we get to learn from that. They can perhaps learn a bit about our thinking too. We weren’t doing that until relatively recently. So that’s one thing, we’ve certainly learnt from that period. And I think from the COVID years, it probably won’t be a surprise to say that the Board certainly took on a lot of learnings around communication of policy and how we get our message across. I’m very conscious that, as an economist, there’s a lot of language and words that we use that are just not the lexicon of everyone else and we have to adapt to that because we need to communicate with everyone and everyone needs to understand it. That’s been a particular focus of the Governor this year and I know she’s always keen to hear feedback on how people are receiving her press conferences and her other engagements but that’s something we’re really focused on too. So different things at different times.

Thanks, Sarah. Genevieve Knight from SACES. Just wondering, you did speak about how you’ve reviewed your COVID responses, et cetera. At the last briefing the Vice Chancellor mentioned that they expect another pandemic. So how have you been preparing forecasts for that?

So I can’t say that we’re - so I would say that seems like a reasonable expectation. If you look back through human history, pandemics are not unique events. They happen with - no, I won’t say regular frequency but they certainly happen again and again and again. So it seems reasonable to think that in the forward looking view of human kind, we’ll have at least one more to deal with. In terms of thinking about - we have our forecast and policy response. So we haven’t prepared a set of forecasts, mostly because that would really depend on what type of pandemic we were talking about. What the response would be in other countries and there’s so much uncertainty about how it would actually play out but in terms of my team’s time and where I think we can get most bang for buck on the work that we do, I don’t know that thinking and producing detailed forecasts for things like pandemics is necessarily the best use of time. But what we have done, in terms of the policy response, we’ve done a whole series of reviews of all of the various different unconventional policies that we use through COVID. So you can find them on our website. We published the last of those, which was the term funding facility. That was the funding we gave to the local banks at very, very low rates so that they could then pass it on to their borrowers, so have soles and businesses. We just did that last one. So we have reviewed all of our unconventional policies, what worked, what didn’t. Those reviews will go through in some detail, the lessons we take forward into the future, how we would do them differently if we had our time again, going back to the start of COVID. And so for the Board that’s been a really - I’d say a good reflective, healthy exercise so that they can learn. So when they have to make that decision again, which as you point out I would think probably a Board in the future will have to do that again, they’ll be armed with more information. Equally though they’ll probably make some missteps the next time around because it’s never completely clear and hindsight is a wonderful but as long as you’re learning, I think that’s the main thing to take forward.

Thanks. And I had one small additional, so we had a person mention the issue about regional forecasts, we don’t do them at SACES. And I guess the RBA really has a national focus. How much do you account for regional or a state variation in your forecasting?

Really good question. So we are an … national in what we’re trying to do. We operate at the aggregate level. So we certainly don’t do very detailed state level forecasts and other institutions do do that, that have that more local focus such as SACES for South Australia. Though having said that, we do care very much about how different parts of the country are evolving and playing out over time and we gather that information in a few ways. We do look at state level data. We have our fantastic liaison program. I’m joined by some colleagues from our Adelaide office here today who will go and talk to organisations all around the country, so from South Australia. We cover Adelaide of course, but the guys also get out to the Barossa. They get up to Darwin every so often. So we have people in Perth, in Brisbane and in Melbourne as well, as Sydney where headquarters are. So we’re trying to cover the whole country. We interview around 900 organisations a year. We’ve been doing it for over 20 years, so as Jonathan said at the start. We’ve now got 20,000 interview notes that we can pass to really get some insights from. So that’s an incredibly powerful information set that we have access to, that we factor into our forecast. And more - just generally in a very aggregate sense, what’s really interesting as an economist but so important and vital for how this economy aggregately responds is how the states will often be performing in different ways. So right now Western Australia is very strong. South Australia here, you guys have had a really good patch. It’s really - wonderful to see. Other parts of the country in relative terms are performing a bit more slowly and in a relative sense are sort of falling behind a bit, Victoria is a good example of that. That’s not always the case. Pre-COVID it was actually almost the opposite. So perhaps a little bit softer in terms of growth momentum here. Victoria was actually the leader. And so seeing how those ebbs and flows come together and come through in the aggregate data is really important for our understanding of the economy. So I don’t have a set of state forecasts at my fingertips but I do have quite a lot of information about what’s happening and, as I say, it’s vital for the outlook.

Hi Sarah, thanks for being here today. My name is David Lambert. I’m from the LGFA. The Local Government Funds Authority here in South Australia. I’m just wondering with your baseline forecast, how much do you look at the - I guess the potential fiscal spending from the elections coming forth, do you get into the political side and have debates about where you think the different political parties will spend and so forth as well?

So we’re obviously aware when State and Federal elections might be. And so, you know, that’s sort of known. But in terms of our forecast, they’re based on announced policies and legislated policies. So as we are right now, we factor it into our forecast. The Federal Budget that was handed down in May, the state budgets that came through about the same time, which I think the Tassie budget was the last one in September/October, I believe. Those settings as they are today are what we include in the forecast and we run those through. As new announcements, new policies get legislated, then of course we’ll update the forecast. And indeed it’s one of the sources of uncertainty for us. Governments, by their very nature and of course this is what they’re there to do, will change their policies over time. We don’t try and forecast those. We take them in as assumptions once we know about them.

Hi Sarah. Yeah. I’m Furman from the School of Economics and Public Policy. So you often share the forecast modelling so I’m wondering whether you are concerning the frequent list forecasting or division forecasting in your scenarios. So the reason I’m saying that is that because you are saying we high in - in uncertainty world and obviously how can you quantify the uncertainty, if you are doing frequent forecast.

It’s an excellent question. So we don’t try to do that in an aggregate sense. So we have the fan charts that I showed you earlier on and then when we consider alternative scenarios they’re very discrete and we’re not trying to attach probabilities to them for basically the reason you just said. It would be incredibly complicated and hard because we’d have to have those … probabilities for each of the individual series that we’re forecasting that sit within our models and then we’d have to interact them all. It would get difficult and complicated very quickly. And I’m also not quite sure in a scenario world how we would attach distributions to outcomes and probabilities to outcomes as well, because we only have one run of history. So we don’t - it’s not as if we can roll the dice, as it were, a thousand times and see what comes out. We can’t. We only get one run. And so trying to figure out what those probabilistic distributions look like and the interdependencies look like across all the different series that we model in a general equilibrium sense gets pretty tricky and pretty hard. We don’t have it. Other organisations have certainly tried to do that with more general forward looking stochastic deviations, that’s a possibility. What you end up with actually is something that looks akin to a fan chart. There’s a lot more underneath it and so on you can make an argument that it’s a bit more rigorous than the fan chart approach that we do. But in - I suppose in a practical sense you end up in - about the same place. So we’ve not tried it. It would be very hard. If you’re trying it I’d love to hear about that and see how you get on.

Ralph Bayer, also School of Economics, Public Policy. It goes a little bit in the same direction. So you have multiple uncertainties and ambiguities in there. So you have model uncertainty. So is the model right. You have uncertainty about assumptions you make and you have ambiguity where something could happen where you have no idea beforehand, that is it could ever happen. If you look backwards, have you tried to attribute fractions to forecasting error through those different certain ambiguities?

Yeah. Really good question. So it’s certainly the case, in terms of model uncertainty. We do work to review our models and then to update them. To update the relationship. So the time. So maybe to give an example, we’re looking very closely at the moment at the relationship between job vacancies and the unemployment rate, which for economists is the beverage curve. And so we’re trying to work out what level of vacancies is consistent with the labour market being at full employment. So that’s really important. So we want to try and gauge how much of the current vacancies is because there’s really strong demand for labour and how much of it is because there’s just natural churn in the labour market. There’s always people moving around, job to job, students graduating, getting a job or looking for a job and then getting a job and so on. So we do it for that model uncertainty and that way we can identify that. Identifying all of those - unpacking all of those other things you mentioned though, and I would throw into that data uncertainty, because we’re only ever working with the current vintage of data and we do get revisions, sometimes quite substantial ones. That makes it really, really hard then to look and to think, how much of it was because of something we didn’t know about, how much of it was something that we couldn’t see and then how much of it was an actual error on our part. We try that on a very partial basis, so we might look at one or two indicators. We’ve not tried it for our entire general equilibrium model because that would be quite an undertaking. But you’re right. The more we can do the better. So it is part of our learning but it’s a tricky exercise.

We likely have time for another question.

Nathan Petrus from the Local Government Association. I’m interested in any insights you might be able to provide on how your Board might use this information on scenarios to make decisions and the thought that comes to my mind is, as kind of diverse people on the Board they all have different thoughts around each scenario, do we want to act to minimise the downside, do we want to act based on the most probable of those scenarios. Everyone probably brings a unique perspective on that and that collective decision-making around that. You may not be able to comment on how the decision-making process unfolds but that’s an interesting part to me of how this information is actually used to make decisions.

Yeah. So I certainly won’t comment on individual Board members and I should disclose I sit with the Board as they make their decisions so I have a lot of information but I wouldn’t do that. But I would say that, in general, you’re absolutely right. The idea behind having a group of people that make this decision, rather than just one individual, is that you do get those different perspectives. That you get challenged to different views that people come in with a different idea or they put different weights on different scenarios, to take it back to where my talk went. And so the Board do that. They talk through those alternatives. They do talk about the risks around the outlook and where they might see the balance of risk but also where they would be more concerned or less concerned about a deviation from our baseline forecast. So would they be more concerned about inflation looking different to what we expect it to, versus labour market outcomes looking different to what we expect them to be. And that all comes together then in the collective decision-making process. So it is very considered. I mean, the meeting now runs across a Monday and a Tuesday, so there’s a lot of time that they spend doing this and then obviously they get the papers ahead of that. They will have read those and produce their own man hours that go into it each meeting is not insubstantial. And then you get the decision from that. And it is very much a collective decision. So, yeah, I would say that that is how they operate. What you described is what happens on the ground, as it were. And what we’re trying to do and what the RBA review called out for us to do as bank staff was just to provide more structure around - particularly around this uncertainty and alternative scenarios piece. So that’s why we are now doing - publishing in the SMP but also providing to Board members, specifically those alternatives. We refer to how the data is played out, relative to what we thought and what we’ve learnt when it’s the same, and what we’ve learnt when it’s been different to what we thought. And then we have a discussion and challenge around that. So it’s more or less what you described. I personally think that’s very healthy and that’s what it should be and that’s why you have a group of people, rather than an individual. It is to get that diversity of perspective.

Thank you so much, Dr Sarah Hunter. I now invite Professor Jodie Conduit, Executive Dean of the Faculty of Arts, Business, Law, Economics to provide the formal thanks and presentation of guilt.

Thank you, Fiona. On behalf of the University of Adelaide and also the South Australian Centre for Economic Studies, I firstly want to thank all of you for coming and joining us today. I’d particularly like to express our gratitude to our two distinguished speakers who have provided us with some valuable insights. I’d like to acknowledge Jim Hancock from SACES, who has given us a comprehensive overview of the economic conditions at the state and national level. His analysis, of course, helps us to better understand the complex interplay of so many factors that are shaping our economic landscape. We’ve been very privileged to hear today from Dr Sarah Hunter, the Assistant Governor and Chief Economist at the Reserve Bank of Australia. Her presentation on uncertainty and forecasting and the use of scenarios in policy development was particularly timely and enlightening as we navigate through evolving economic conditions, understanding how to better manage them and account for this uncertainty becomes increasingly crucial for both policy-makers and business leaders that are in the room today. So to Sarah, before you return to Sydney we would like to present just a small token of our appreciation with some goods from South Australia.

Thank you so much. That’s very lovely. Thank you.

So just in closing today, I just note that in the room we have parliamentarians, people from the public sector, business leaders, NGOs, who all have a shared commitment to understanding and shaping South Australia and Australia’s economic future. So these briefings that we hold serve as vital forums for our knowledge sharing and for discussion of critical economic and policy challenges shaping our state. Today, as I said, I thank you all for giving up your time and I believe that you’ll leave here far richer for the experience and for the words of wisdom from Dr Sarah Hunter and from Jim Hancock. So as we approach the holiday season I encourage you to plan a scenario - did you get what I did there - in which you have good times spending with family and friends. Thank you once again for your participation and I wish you a safe and happy holiday period.