Luke Metcalfe on What AI Can and Cannot Tell You About a Suburb

It seems that everyone is talking about AI at the moment and platforms like ChatGPT have put artificial intelligence into the hands of all of us.

This can in turn be both great and disastrous as it really takes a subject matter expert to know whether what the bot can tell us in a matter of seconds is true or not. And it really takes a data scientist to know whether we're even asking the right questions in the first place.

How will the rapid advancements in AI change the way we understand and interact with the property market?

Welcome to the elephant in the room. This is the podcast where we love to talk about the big things in property that never usually get talked about.

I'm Veronica Morgan, real estate agent, buyer's agent and buyer's agent mentor, co-host of Foxtel's Location, Location, Location Australia, author of Auction Ready and co-host of Your First Home Buyer Guide.

And I'm Chris Bates, mortgage broker, recently ranked number five in Australia out of over 18,000 brokers in the annual MPA Top 100 Mortgage Broker Awards. Before we get started, I need to let you know that nothing we say here can be taken as personal advice.

We always recommend you engage the services of an appropriate and experienced professional.

Today we're joined by Luke Metcalf, an entrepreneur and data scientist. Luke is well known in the area of forecasting location analytics with a recent focus on the real estate market.

He began his career with a degree in computer science and quickly made a name for himself through Nation Master, an international statistics website that drew in 10 million unique visitors monthly and leveraging the success of that He founded MicroBurbs, an Australian data company best known for hyper-local livability guides and public housing data.

Great website if you've never played around with it. Now, this innovative platform delivers in-depth neighborhood-specific information, revolutionizing decision-making processes of buyers, agents, and investors, and known for his unique blend of first-principle thinking and AI-driven analytics processes.

Luke has been successful in synthesizing data from diverse sources to uncover critical insights. Now, we last spoke with Luke way back in episode 16, so we've got a lot of catching up to do. And thank you so much for joining us today, Luke.

I'm very happy to be here. Can I tell you, actually, you just read what GPT said. So you asked for a bio. I edited it a little bit, but it went a little bit too far saying I was Australian born as if I'm an international superstar.

I had to take a couple of things out. But otherwise, I was like, yeah, that works.

And you know what's funny? Because I very rarely read them out verbatim. There was a few little tiny tweaks in there, but I basically read the whole thing out, which I very rarely do. So go chat GPT. That's all I can say.

Sometimes it's awesome. Well, let's stay on this trend, I guess, with AI and property data in particular. How do you think... I mean, back in episode 16, that was a long time ago, Veronica. That was over five years ago. And...

You know, I remember somebody on your website even many years before that, I think, I was just blown away with how you were looking at things, not so much at the, you know, just very basic level, this is the median price, but all these other factors that go into making a place livable and desirable and livability.

And I actually went on your website recently and saw that it was a whole new, many different data points, et cetera. So you can get blown away with data. I guess, what's your view?

harness all this data to actually get some meaningful insights that when we're looking at i guess investing or buying or buying a home in the property market we're actually making better decisions rather than just overflowing a busy mind already Yeah. Excellent question.

And by the way, thank you very much for having me here. I, this is, uh, I'm in, as you said, I, I'm in great company being here. So, uh, and this is by far the best real estate podcast. I love you guys straight talking.

So I don't even know who is, who's second, like this, this, you guys are just number one. So great question. So I guess back in, it was about 2016, five years ago, then AI was quite different to today. So I think it's worth just talking about how it's progressed.

So back in the day, it was a bit more like maths as you would recognize it. So it was more like closer to most things were like, just get a formula. It might be a complex formula and hard to read, but it was like boiling it down. So if you imagine like,

Um, to oversimplify, you might say that the, uh, base price for a suburb is 600 grand and then it's 200 grand per bedroom or something like that. Right. It would never be that simple, but that's just to illustrate, uh, how a simple valuation model might work. Yeah.

Um, so back in the day, it would take in lots of different factors and I've always been about getting a very large number of, uh, features that what they're called like columns, uh, data points to, uh, give the AI as much context as possible.

But what we've seen since last time is that AI takes in way more information and can understand it in a much more subtle way, in a much more nuanced way, more of a vibe. So you can go into ChatGPT and ask it a question that is like,

not strictly how a literal person would answer it, it can get the vibe of what you're saying and answer it. So ChatGPT and all of these large language models, there's heaps of them, obviously ChatGPT gets all the attention, but what they're doing is just trying to predict the next word.

over and over again. So it's just incredible. So in the same way that you have an AVM CoreLogic and us to a lesser degree, we come up with that. We're trying to predict the price. If it sells, they're just trying to work out what the next word is.

Now, of course, it's a bit more complex in a sense because there's so many words to choose from. And so it's got all behind the scenes. In one way of thinking of it is it just has a whole lot of percentage chances of each word coming up.

But what's actually, yeah, so you can't go to the CEO of OpenAI who produces ChatGPT and say, what's your formula for coming up with the next word? What are the five things you need to know to work out the next word? Yeah.

In the same way that all of us real estate people get asked this, you know, what are your five things? It's just way too complicated. There's way too much. They themselves don't know exactly how it works.

So, yeah, I would imagine a lot of this conversation will be about like, what do you do when computers are starting to get the vibe? They're not perfect by any means. And I'm of the view that there's plenty of room for humans, including buyers agents and sellers agents and mortgage brokers.

Um, but, uh, yeah. So how does it affect the real estate industry is a big thing. I think.

uh obviously i guess people's first thought is um this thing's gonna might take my job i don't want to know about it um it's scary there's a lot of pushback um but the reality is that if you're a sort of a generalist which i think a lot of buyers good buyers agents like yourself and good mortgage brokers where you're if you're seeing the whole picture

Uh, then you're the kind of job that's safe, even if it might mean that your job changes a little bit. Uh, the fact that you've got so many things that you can do, you can negotiate, you can sell, you can create an engaging podcast.

Um, all of these things mean that if one of those things gets, uh, really done better by AI, like maybe the title for this podcast is best, best done by an AI because it can try out a hundred different titles. Yeah.

Um, or maybe it just, you have a conversation with it and it tells you which is the best title to use. Yeah. But, uh, so those, those kinds of roles, I think are much safer than say someone who is like a translator. Yeah.

So if they've just got a very set task and they're just brilliant at going between English and French, uh, they're in a lot of trouble. I studied law.

Thankfully, I didn't go into law because even years and years ago, a whole lot of paralegals lost their jobs from major law firms because the act of scanning through very large amounts of email suddenly got doable by an earlier kind of the same model.

as what you see with GPT, but it's like it can scan for the meaning to find. It's sort of back in the day, they would search for keywords and then have to read it manually. So a whole lot of people in law lost their jobs. So being a one trick pony,

is not a great thing in general but i think uh good real estate industry professionals particularly ones that have been in the industry for a long time they've got a fantastic ability to adapt and grow and um in a lot of ways give themselves a promotion so for for us in

in micro burps, uh, our big thing is going to be at the moment, it's a big data dump. There's loads of stuff. Um, so smart buyers, agencies are picking out the things which we've tried to present it as well as we could.

Um, this fantastic indicators in there, but the next step is to create beautiful, um, custom, uh, PDF reports for you guys. So, uh, it's, it saves a lot of the legwork in having to go in and look up a whole lot of government documents. and it will be relevant to the actual clients.

So my vision personally, and I'm putting my money down on this, so I've invested, I'm not building a chat bot to replace anybody. Yeah, my big thing is it's going to be much more a helper that will be able to do a lot of the legwork.

Another thought I had was that obviously all real estate industry professionals, their customers tend to not want to pay them per hour. So that leads to a problem that you negotiate a fixed price.

Um, uh, maybe it's a percentage, 20 grand or whatever, and you can't just keep on serving them forever. And real estate is a very, uh, stressful business.

So for the, um, the buyer and seller, so I can imagine that there will be something like a chatbot talking to them at 3am when they can't sleep. And, uh, that context will come back to you.

And all these little questions that you don't want to deal with, obviously at some point they have to actually make a decision and buy the property.

So yeah, the reports that we're building are also very much about give people the confidence, give them the information they need to get them to that next step.

So you're suggesting, because I mean, I haven't actually checked out the most recent iteration of microburbs, I confess. The old version, we had the sort of 10 metrics that your computer had found for you in terms of the important things, factors in a location, correct? Yes.

I love the hipster score, for example. And I remember back in episode 16, a few things really stood out in my mind. I remember the revelation that Tranquil, so the Tranquility score, was inversely related to capital growth. I thought that was really interesting because everyone just thinks quite location.

We're not talking main roads are good for capital growth. What we're talking about is location.

areas without lots of amenity right but also so you know you set me off because back then this five years ago and I was in this quest to try to work out the secret to long-term capital growth which at that time nobody was willing to commit to to say I can say that these things will point to anything working for more than two years and

Before you say anything, I would add, I've since learned that beautiful natural settings can have very large capital growth. So there's a whole lot of nuance to all of these things, right?

Which is precisely where I'm going with this question, because what I want to know is what's changed? What have you learned since?

And the other thing, just a bit of a confession here, that while I was in this process trying to work all this stuff out and I was doing all my research and learning from people like you, and you said zoning is the number one impact on capital growth.

And I was like, right, that's it. We've got to go out there and look for all the areas where there's zoning. It was like me and everybody else. It was like, yeah, I mean, you didn't actually need AI to work that out in reality. I was thinking, that's just brilliant information.

But the simple fact is that what I've since learned is trying to hotspot is just a mug's game. If you're going to invest in property, you've got one shot at it.

Trying to gamble and pick the next place is really fallacy because there's other ways to bet your money safer or to invest your money safer. So leading to that question, which you've just alluded to, you're going to give us lots of answers. What have you learned?

What has AI been able to change your opinions in that intervening time? What have you learned was necessarily ill-formed and potentially still ill-formed with the accelerations and progressions in this area?

Yeah, so a big thing is when you get more longitudinal data. So I have data going back way further. So a big problem with real estate is like other markets, like the stock market, it has phases.

So if you base your decisions on a short period, so for example, during COVID, you might say that regional Australia is awesome. Uh, I found the models have found that the late nineties are a particularly unusual time.

So there's this fun, um, uh, graph I have, which is correlation of postcode to capital growth. And it's like this sine wave. Yeah. So it's like, if you picked a particular period, like, um, four to six years.

And you would, you'd think that you would know, and very often the model will just go for postcode. Um, it'll just like, if you feed it everything, it'll say, oh yeah, it's just this area here, but then that might not be the case further on.

Are you saying then that basically every postcode will at some point experience the same period of growth? Is that what you're saying?

No, what I'm saying is that the correlation to postcode and capital growth... So I should mention, by the way, when I'm talking about capital growth, it's always relative to the national average. So I'm not a macro guy, so I don't have a view on like...

I don't know what's going to happen in the war in Ukraine or...

um the chinese shadow banking market so my focus is very much on uh how does it compare to uh the rest of the country so the relativity of it i love it yeah yeah yeah okay so luke i mean i guess if you think fast forward right like ai is meant to know you better right like it's meant to know what your what you like what your preferences are what is you know it's kind of almost predictive of what you want in the future right i think that's like the more that that and that

AI knows about you, more should be able to read what's the best next thing for you, right? I mean, is that sort of the direction we're going in as a human? So, like, they're basically saying the next thing you need to do is drink water or eat this food.

Like, it starts to almost, like, predict your life, what's the next best move. Do you see that it could also...

become that knowledgeable around what drives prices right and it almost could have like a property score right like so this property is a 99 out of 100 or a 72 out of 100 and or it also and then to a point where

A property comes on the market and it says, look, don't even look at that property because it doesn't tick what you're looking for, right? And because you're telling me you're a family, you want to be walking to the school and all your preferences. So this is a 60% match to what you want.

But a property comes on, it's a 99% match to what you're looking for. But then it could be overlaid with, it's actually a 99 type of property for capital growth. Because it looks at traffic, it looks at amenity, it looks at the location.

Do you think we're going to go to this level where each individual's preferences are going to be loaded in and we're almost going to be delivered properties that are matching to what we want? And then properties are going to be matched to capital growth drivers.

So we're going to have enough data to know, actually, this property is a bit of a dud. It's traded three times in the last 10 years. It's always underperformed growth. It's on a rat run. It's got aspect issues.

Are we getting to this level of data where not do we have to go and find it, but almost it kind of finds us the solution?

Personalization of data, finding the right property, that's one thing. And then there's also, is the individual property right for anyone? It might be a dud property that won't give capital growth. So yeah, there's big tech companies, obviously, not me, but big tech companies are gathering a huge amount of data on people.

And so they could be employing that in the same way that you have internet advertising where you go look for something and then it shows up as something you should go buy a fridge. It thinks that you want a fridge. So I definitely see that is possible.

I think it comes down to a great degree on how we regulate privacy and what people are prepared to accept. So it may be that there is a September 11 of privacy, for example, a big dump and people get shocked and they push back.

And they say, no, no, no, I want to control everything myself. So one model is that the individual on your computer, you hold all your data and you give it to people when you want, or even the decision gets made on your own computer. So that's called federated machine learning.

So yes, there's definitely the ability to take data and personalize based on others. But having said that, I think there is a lot about the human interaction. So we know that it gives you joy to interact with other people.

So I don't think that it's a simple matter of the AI simply handing you a property. I think people will want to talk to someone about the problem about it and have them give them confidence.

I mean, the mere fact that a investor will pay, if I'm lucky, they'll pay me a hundred dollars. Well, they'll pay you guys 20 grand. That speaks to how much value they will often perceive in a buyer's agent. And obviously there are more buyer's agents coming on the market.

So there's more of them at the moment. You might've expected less. Uh-huh. Yeah. So, and then in terms of the property, finding a dud property. Yeah, absolutely. So over time, there will be more, there's ever more data that comes on.

You see different data sources coming online, both in terms of government data and what's out there in general listings and things. So that does give ever more predictive ability.

But what's happening now with the generative AI revolution is increasingly there's a democratization of access to data.

So instead of you have like a Boffin data scientist with his tools, like me, writing code, and then you have someone who's out in the field, who's been in the industry for a very long time, who knows everything. how people react and has seen it all.

So increasingly there's the ability for you guys to directly talk to an AI in English and also being sophisticated, you'll be able to ask sophisticated questions of that

So let's get into the questions, actually, because equally, yes, you're right. Someone with lots of experience can harness all of their experience and augment that, if you like, with AI. Certainly in the property space, there's a marriage of data and local smarts.

You know, so data will get you so far, but individuals make a decision on an individual property and you can't work with aggregated data to make that decision, right? You need to, to bring in, to play other elements there, but equally someone without that experience, someone without all of that knowledge.

can then, for the democratisation of data, get access to data but not actually know what to do with it or what to wait and what to actually pay attention to. And equally, the experienced person could be biased as well and not be paying attention to things they should be paying attention to.

So I guess these are some of the problems, though, with this easy access to data. And that feeds back into knowing what questions to ask, right? So chat GPT-LI is very convincingly

Yeah. So a really big question of AI is what's your target variable? Yeah. What are you aiming for? Yeah. So if you're aiming for capital growth, that's a pretty good one, but we see other AIs that we're already, so we're already in a very much an AI era.

So probably people who see this podcast. It's because AI has recommended two of them, whether it be this particular episode coming up on YouTube, or is it the podcast in general? There's some clip they discovered from the recommendation engine. So those models are optimized for attention.

And these attention-based models, and you could have other models like advertising models that are optimized for you parting with your cash. Yeah. So it can be, I think it's really good to, it's not in the debate enough to ask, what are you trying to optimize for?

So increasingly it becomes opaque how it's choosing. So what are your goals? I think regulators need to move to that. So like, what are you trying to predict here? However you do it, it's going to be very complicated. It's hard to write legislation for it. Uh, so, uh, yeah, absolutely.

So, and a big way, uh, so as if the actual target is something that is good for the person, whether it be yields, capital growth, uh, I'm, I'm personally very passionate about happiness as well. So if that can become a target, then that's great.

Uh, but there is a potential for it to be optimized for other things. It is also true of the real estate industry. So you can be a sales-oriented person who is optimizing for short-term revenue, dare I say.

So yeah, I think thinking of things in that way is more helpful than, and in that way also, there's a lot of, I'd love to see there more, like if we have the same metrics, AI and humans, then that would be useful. Yeah.

So like if you have, so you can ask a data scientist, what is the R squared? So that's like, or more simply correlation. Yeah. So that's like, what's the percent of the difference between some variable that you have? And then you can go, aha, you don't have this other bit.

And I help with a lot of this other bit.

So just on that happiness part, is there anything in your data points where you can relate that to prices? Like, are people in more expensive suburbs any happier than people that live in cheaper suburbs, for example? Or leafy suburbs?

Or yeah, or suburbs surrounded by water or closer to the city or further away, or, you know, is, is there, is there a way to measure this sort of thing?

I, yeah, I want, uh, so I've got a whole lot, this big research I'm working on at the moment. And there's like, there are many pieces to the puzzle and I'm sure there'll be revisions on things that I said five years ago. Um, so, uh, I haven't fully pieced it together yet.

Like there's so many things that I want to control for. And, uh, so I won't have all the answers that are of things I've studied today. I'd love to do a follow-up podcast if I'm invited back.

Uh, one thing that is interesting though, if you look at the, um, census, you'll ask for anxiety. They asked for anxiety in the last census and what you'll see, if you go to under microburbs, uh, you'll, you'll see for Sydney, um, there is this huge blob around Marrickville. I'm sorry to say Veronica.

So the inner west not is a rather anxious place.

Do you know that?

Hang on. It's very distinct.

There was a show, I think it was on SBS, with Dr. Tim Sharp, who we interviewed some years ago. Oh, yes. Who is Dr. Happiness or whatever, a Dr. Happy, and that was set in Maryville for that very reason. So this is not a recent phenomena. Yes. Do you know why?

Did the bot tell you why?

Um, I, yeah, so there is research on this internationally. Um, so, uh, yeah, it gets into politics. I don't know if, if we want to go there, but, uh.

We might as well. Isn't that where Albo lives? Our prime minister?

Well, I'm hoping, I think a certain level of anxiety is a good thing. Like you want to be, as a prime minister, you want to be like, uh, um, not completely taking it easy.

I'm on a personal mission to help more people make better property decisions. And you can find out all about what I'm working on at veronicamorgan.com.au.

And there you'll find resources for first home buyers, details about my buyer's agent mentoring program, access to suburb help for investors, or if you're looking to buy your dream home or an investment property in Sydney's inner west, eastern suburbs, or lower North Shore, you can connect with my team at Good Deeds Property Buyers.

If you're thinking about buying your first home, upgrading to a new one or purchasing an investment property anywhere in Australia, we would love to carefully guide you through this journey and importantly, get the finance right. Please reach out via our website, wealthful.com.au.

Don't forget that you can download our free full or forecaster report. Which experts can you trust to get it right?

Theelephantintheroom.com.au But do you reckon, look, I've just, are you able to get access to like, I don't know, like Google Maps data, right? Like that's amazing data. You'd know where people are driving and, you know, is there any data sources that you think that,

you know, we aren't factoring into our property decisions, that if we did get access to it, we would change our view on, you know, how we invest or, you know, because rat runs are something that, you know, you only really figure it out years after living somewhere, to be honest.

Like it's not even initially, you just, you start to find, oh, actually, you know, this is a bit of a cut through, right? I didn't actually know people took this road, you know? So is there data that you wish you could get access to that you don't, that you think is really crucial?

There's certainly, yeah, amenity data is definitely useful. It's definitely interesting. So proximity from different things. So I think I mentioned lifestyle school last time was predicted subsequent growth.

And that was like proximity to a whole lot of things that are useful for mainstream people, RSL clubs and shops and things like that. So yeah, amenity, from my point of view, it's all useful.

So I'm building a general AI and I'm much more interested in its ability to help people out and do the legwork and democratize this kind of analytics than always being the expert on everything.

I mean, here's an important point. So there's lots of researchers that do research on very specific things, right? And they write papers and they consider. And then you have someone like me who just comes in and goes like, what's a leading indicator? Yeah.

So it's like this thing happens and then that thing happens. Causation is a harder thing. Question to answer. Yeah. So there's, so there's like, there's like a data point, a story, and then there's like, um, many data points somewhere in there.

It's the, uh, the expert professional, and then there's like a leading indicator and then there's like. a reliable leading indicator. And then that causation piece is the ultimate one. It's actually quite hard to prove causation.

And very often from the point of view of real estate, you might be like, look, as long as it's a reliable leading indicator, let's keep moving.

Have you done any work on incomes relative to growth? Because we had a recent guest on that said that there was no correlation. We haven't tested the data search. I saw that episode. Yeah, okay. So we haven't tested.

We weren't equipped with any data on that to be able to sort of further that discussion other than what our brains tell us. But what are your thoughts on that?

I've definitely looked into it, but I'm not going to say today.

So is it work in progress? Is that what you're saying? Yes, it is.

Yeah. So Jeremy and I will definitely have things to say about this. So yes, I mean, very often what happens is one person says, oh, go for low income. Another person says, go for high income. Another person says it doesn't correlate. Very often it's nuanced. Yeah.

So, uh, is it, one question is, are incomes going up? Is it good to have low income, but things are getting better? Maybe that's more important. Uh, maybe it's not so good if incomes are going up because it's a mining town. One thing I can tell you is mining sucks. Don't go mining.

No.

And this is all back to the question though, isn't it? So this is about, you know, we're always about the nuance on this program. And, you know, you're a data scientist. You're selling data. You're selling information. That's your product, right? Tell me if I'm wrong.

Yeah, reports I would say increasingly.

Yeah, but it's information.

You're pitching that, right? Where I have an issue is with people that are pitching their particular property magic pill or their silver bullet and they've got their particular pitch and they cherry pick data to support whatever their marketing pitch is. And that's something that we do take to task on this podcast.

But we're curious. We want to understand where people get their information from and why they're using it in a certain way. And you're just the information guy. So that's why we're just like quizzing you.

Yeah. I reckon our next podcast, if you'll have me on again, will be awesome. We'll run through the stuff. I'll show graphs.

I love that promise. God, talk about, you know, big fat carrot. Yeah.

Yeah, I mean, this one hits the cutting room floor. I don't know.

No, no.

This particular podcast, I know you're working on something. And so, you know, this is maybe slightly preemptive. However, you know, you've been in this game a long time. You're not a newcomer to the property space. you are a different thinker, you know, that you're a bit contrarian in your own ways, right?

And you're curious and you're also an expert in your field. So that's why we're talking to you because we know this is a changing landscape. We know it's changed a lot since we last spoke to you. We know that you're working on a project. We will get you back.

We want to find out more about that. But I'm really keen to know in the intervening five years between then and now, what are you having to admit you got wrong, you know?

Um, yeah, I don't remember exactly what I said in that podcast. No, I mean, you know, you must know the iteration of your- I reckon that would be, that would be great content to say, look, uh, this is what I got right. This is what I got wrong.

I based it on, so obviously the more longitudinal and the more data you have. So if it covers, I'm very much focused on cover the whole country, go back as many years as possible. So the old microberbs was one point in time. I did have some longitudinal data in the background, but, uh,

Now it's much more like, okay, can we rely on this? So the ideal is what can you rely on in every market? So that's like, what's a universal?

On that though, because there are some universals Australia wide, some universals that might be say capital city, you know, applicable. Some of them might be regional applicable. Some that are suburb applicable. I mean, what is a universal?

I guess, because a lot of people think that they can apply the same sort of the same metrics to every market and you can't. Because what's valued in one market is totally different to what's valued in another one.

Yeah, so I will build a model that predicts overall, and then I will ask, where does it work and where does it not work? And then you might say, oh, okay, if it doesn't work there, that's weird. I would have expected it to.

Or maybe this model applies for this market and not for that market.

luke i was just wondering if you've um so what i mean what i guess looking at suburb data is great because you get this macro view of what a suburb exists right now right but what a suburb exists of the people who live in a suburb doesn't really matter per se what matters to in my mind is who's moving to the suburb right now because property is basically is a marginal buyer right it's the price of something is not what

everyone in a suburb wants today it's what who's buying in that market say what they can afford and what they what's their situation what's their income what's that and so that's really stuff you want to be reading right you want to be reading like like who's this is what gentrification is and this is what changes and this is what changes prices right because if the same people were earning the same amount of the same people 10 years ago that have the same means and so prices couldn't go up and so

I mean, my view is the data you really want to know is like, it's kind of like search data, right? So REA search data, who's looking in this suburb this year versus last year? Like, what do they do? Is like, are their incomes higher? Are their qualifications higher?

You know, have they got other assets, et cetera? And is there any way to sort of get that real data that, is the people who are transacting and how that is shifting over time.

It's almost like you could look at a suburb at a whole level and say, I don't know, 30% are earning over a certain income, right? And that could go from 30% to 32%, which doesn't look like a big improvement.

But if that's the 2% that basically bought into the suburb, and it's actually as a whole, it's actually quite high. So How do we get that type of data that is the people that are actually transacting today and how that is different to what it was in the past?

Is there any way in your mind that you can get that? Yeah.

So that's something that real estate professionals obviously see firsthand. And it's something that...

So domainandrealestate.com.au, obviously they are seeing who is coming through. They might know more about people than that. So they do represent, like, they may represent a threat to the industry in that. I'm not an expert. I don't know what their plans are, but obviously, conceivably,

If I was them and my goal was I was following your line of thought of who's buying in, then that would make sense. Where is the market? Obviously, there are metrics, online search interest that are already out there. So obviously, the bigger one is who's likely to be there later?

So where are people going to be? And I've worked with data sets before outside the real estate industry where you have lists of where people have lived. So working in insurance, for example, right? So you can actually predict, okay, they're living in Newtown, a two-bedroom terrace in Newtown now.

They're going to need, they've got kids, they're going to have to move to Merrickville or somewhere else. So you could actually, this is what's called an agent simulation.

So this is where you actually are trying to predict for every single person in the country as a very sophisticated kind of model, if you can get it right. So if you can build it on a low level and then aggregate that up, then that

can, so you can have certainty on, if you can get certainty. So what I'm always trying to get is like more data. Yeah. So more rows, more columns so that I can be more sure that's my approach.

It's a different approach to you'll say like, it's logical that people who earn more, they can pay more. Um, so you'll, you'll be interested in that. Um, and you know, it, it makes sense and you've, you've seen it in your own, uh, experience. So yeah, it's a different kind of approach.

And at the end of the day, we should hopefully be able to agree on these things and know what it is. And it won't be so much of a debate. And that will give investors a lot of certainty.

I think too, I mean, back to, this is my big bugbear, is that we make individual decisions and yet there's, you know, there's 11 million properties in this country, if I got that figure right.

We only buy one at a time, generally. And also even just Chris talking there about, you know, is it possible to work out where those movements are, you know, where are the suburbs with the greatest degree of movement of people with increased incomes moving into the rest of it?

And that's in a little, in a sort of way, it's sort of grabbing and understanding what's happening as it happens, as opposed to waiting for the lag indicators, right? And it's also a bit predictive, right? But it's also a bit hotspotting-y.

It's also a bit potentially short-termism in the sense that it's happening now or it's about to happen and I want to buy there. Isn't there more value in working out where the underlying elements that are in place are really solid and it's just going to continue performing well?

So the places that necessarily aren't changing but are just solid performers and there's nothing on the horizon that suggests that they're going to change in any way, isn't that more valuable?

Mm-hmm. There are a lot of different ways to make money, Ronica. I know of more than one. Unfortunately, I'm going to tell you zero in this podcast.

I'm not asking you for that. I'm just talking about the types of way that we could use data and look at data.

But unfortunately, in the property space, there is a lot of short-termism in terms of promotion of services, in terms of promotion of value adds for buyers' agents, right through to spruikers, right through to, I don't know, maybe mortgage brokers even. And it's unfortunate because it is a long-term hold asset.

I'm a fan of the long-term hold. I'm a fan of buying A-grade assets and fewer of them. I'm unashamed of that. Yes, people can make money doing it other ways, but I just find that they're more risky. And so that's for me, I like to avoid risk.

When it comes to using data and looking at data to plot out and make decisions around property, some people are going to get all excited about the stuff that points arrows towards what's next to take off and where they're going to make money where others haven't.

And people like me are going to look for places that are like, great, if I buy there, it's less risky and more of a safe bet than if I go out trying to work out which is the one place that's going to make me rich, right?

Is there a market for the boring data? Is there, you know, are people interested in that or is there much more of a market for the shiny stuff?

I can tell you I'm interested in all of it. So I'm interested in, I'm interested in make money when you buy from an individual property. I'm interested in short term, like what is the, what are the short term opportunities? Are there very long term opportunities?

You know, should you hold onto a property the whole time? Like, Yeah, I'm very, um, I like to think I'm very rigorous about it, so I don't have any skin in the game. I'm an entrepreneur, so when I get money, I put it into my businesses. I don't even own investment property.

Uh, so, uh.

That's why we're asking you these questions, because, you know, you don't have skin in the game. This is why it's, I find what you, you know, your take on this really interesting.

Yeah, I would very, yeah, I'd love to give it next time.

So Luke, on the, okay, so let's just move on to, I guess, a different way that data could be used, right? So, you know, you mentioned zoning. So there's a company called Archie Star. So I don't, we had them to come on the podcast, actually. I don't think we've rebooked them, but.

um it's a very interesting business so what they can do is they can um overlay uh suburbs and then basically do feasibility suburbs on development potential right current land what they could build on it with zoning um and then you could basically predict how a city could expand right like and you could say look yeah and then you could overlay it with sort of transport and um you know things like the metro and

You know, and then you could basically envisage how a suburb could change. So if Sydney's population went from 5 to 10 million, where would we fit that extra 5 million people, right? Like, and, you know, is there sort of data along those sort of lines?

Like, you know, because you've got a lot of development applications and, you know, is there some type of way of predicting how suburbs are going to shift over time based on their zoning and

Absolutely.

Yeah.

Yeah, you definitely can. Uh, so development applications, we have that. So we, we, the new dwellings, um, we track that over time going back many, many years. Uh, so yes, we're very interested in these supply indicators, uh, population projections. We have those as well.

Uh, and, uh, yeah, I mean, primarily microbirth is about giving people the data so they can make their own decisions about this. Um, so yeah.

uh yeah infrastructure is a tricky one from a data point of view because it's like when do you count it what's the date that you put in so it's like people are talking is someone that has the good idea then someone prominent has the good idea uh and then it is um

y'know, it's discussed and people saying it's going to happen, and then someone announces it, then there's an election, and then there's a new government in. So it's actually not a simple thing to work out, from an empirical point of view, to work out.

Um, cause obviously at any point in that, people could act on the information. And the other problem is that infrastructure often accompanies supply. So you really have to take supply into account. And sometimes it may be that there's more supply, say, for example, a shiny new metro.

It might be more of an opportunity to put units in than it is to get people into the city.

On that sort of same line of thought, I mean, we chatted to you pre-COVID, right? And, you know, I guess these... COVID happened, right? And that changed, obviously, hybrid work, work from home. It changed the regions, et cetera. So there's these, like, you know, black swan events, right? Where...

You know, we're not thinking because the longitudinal data actually says this is what's happened over the longer term. But if one of these black swan events happened where we all worked from home or I don't really know. That's what black swan events are hard to predict.

These things are hard to know what they could be. But, you know, you could come up with some right. And.

Is there some type of way of putting that into your data and saying, look, if something like this happened, if a population went at 3% annual versus 2%, you know, or population was 1% because you had to change a policy, that could have a huge impact, right?

So is there some things like that that you can overlay in your data set or have you thought of doing things like that?

Yeah, so those really long-term things, long-term economic cycles are harder to model. So there is an incentive on people like me. If I can find a short-term model, it's better. So if you could buy and sell within one market, then that would be better.

Or anyone who is using the kinds of data that I do to do that. So yeah, no one knows. A meteorite might hit the Earth tomorrow.

There might become a superintelligence where AI takes over and it'll be something like the Christian rapture, except everyone is going to hell or everyone's going to heaven. We don't know which. It'd be great to know. So, yes, I don't have yet. It's definitely outside my knowledge.

It's great that people are worried about all kinds of different issues like climate change and AGI. Obviously, you need to. But from a...

Yeah. So I'm not, I'm not a spreadsheet modeler, so I don't, I don't have like, so I'm not, I'm not like I'm one of these guys that has like the three yellow boxes and it's like, oh, this one seems about right.

And we'll project that forward and look, everything's going to be, there's gonna be a kilometer higher sea levels in a hundred years or something. I, yeah, I don't, I don't do that kind of modeling.

So, and it is obviously, I think there's a lot of brilliant people who do that, but it is also very hard to do obviously. I like to do easy stuff.

But getting access to that data is quite interesting, right? So, like, you know, climate change has gone. We've asked multiple people to come on. We've had multiple people already, but there's some big ones that I've asked to come on. And, you know, that'd be something to easily, you know, put into microberbs, right?

Like, what if we do see a rise in sea level? How would that affect price? If we did see an increase in rainfall, if we did see a...

increase in heat um and you know that someone's making a 30 40 year property decision they could say look that's great but maybe i shouldn't buy this waterfront in coloroy you know like um and you know this place in where there's very little tree cover or this place that's known to infrequently flood um

So yeah, we have landslide risk.

I think we have, uh, um, flooding, uh, we have energy efficiency. So we're, we're very much about provide whatever data we can that, uh, will empower, uh, people to make decisions. We're absolutely put it out there.

So it's one thing I can make data to build these models, but these models are going to feed back into just indicators for humans to use.

I mean, you mentioned earlier that obviously it's easier if people transact within a market, the same market conditions, right? So if it's one rising market in that time or if it's a falling market in that time, but not if they transact in one of the market and then move into another.

My observation post-COVID is that our markets seem to be increasingly volatile. You know, that massive gains of 2021, massive losses in 2022, and now we're midway through 2023 and we're looking at gains again.

In my 20-odd years in property, I've never seen this sort of zigzagging of market behaviour of prices, which is the end game, right, the end result of market behaviour. Do you have anything to add to that?

Have you been observing that, that there's been a shortening of cycles and any sort of theories as to why that might be?

covid was a very distinct period so that was obviously a very strange period um and yeah so i've seen a bit of a reversion in terms of market dynamics since then like drivers of things um so uh yeah i mean i think broadly so speaking outside my analysis here it's just like i think broadly uh

The world is speeding up, so technological change is increasing. You're having less certainty than you did in the 1990s. You know, we're living in a unipolar world dominated by the United States, and now we've got geopolitical competition. We have shorter attention spans.

And, uh, uh, identity is more fluid as well, I would say. So I remember being a teenager in the nineties, you know, you're a raver or a golf and that you would just stick to that at least for six months or 12 months.

But it seems these days less like it's just a meme and you're about invested in that. I've heard about this, like VSCO girls is I'm trying to get subcultures out of my 13 year old daughter and just say VSCO girls, they're girls that go and say, save the turtles. And that's about it.

That's a subculture.

maybe there's more to it than that that's right yeah so people have a lot more i guess people have a lot more at their fingertips now so back in the day you might if you're a raver you go get 3d world and you'll have all these um dance records and that's that's you you have those clothes whereas now you got fast fashion so uh yeah i do wonder about maybe you guys would know like i wonder

would there be higher turnover rate and more young investors getting cold feet? Also, they have more options internationally. So you have now much more of a global middle class where someone earning a hundred grand a year can just move, not just to London, but to anywhere. And now like Asia is livable.

You might even take a pay cut to live in Thailand. Whereas in the past, it would always be some expat that's earning way, way more. So Yeah, I do wonder about that thing, everything speeding up. Obviously, in real estate, there's massive transaction costs.

So you have to be very careful before you try and you can't do it like the stock market.

No, you can't. But the problem is that people... My observation, once again, I'm curious on this and that's why I asked you and

You've got some similar thoughts to what I've got, which is the speeding up of our world, the increase in access to information, and the reactivity of us with that sort of news cycle as well, and also the access to information, particularly through social media, and different news sources as well, or information sources.

Yeah. So it's a little bit of a concern because property, it remains a long game because of those in and out costs and because of the risks associated with it, et cetera, et cetera, et cetera.

So as an asset, it hasn't changed, but we as consumers of that asset are changing and we're behaving differently. And so this is just something that I guess I've got

In the back of my mind, I think I have to do a PhD on this or something and go back to uni and learn how to research.

Yeah, so a few things in favor of real estate as an asset class is obviously we need migrants structurally to grow the economy. We're taking in ever more. We're not building enough to house them. So you've got a demand-supply imbalance.

You've got trillions of dollars of superannuation that hasn't been used for real estate yet.

uh you've got demonstrated um ability for the government to step in and provide job keepers and mortgage holidays to and so there's there's and obviously it would be rather bad to lose a the property market you probably subsequently lose an election so there's definitely and australia is obviously a desirable place so this is none of this is new original thinking here i guess but uh

Yeah, so, yeah, maybe the counter argument may be like, maybe people won't even care. They'll just have their headsets on, the new Apple headsets, and they'll have vast farms, vast ranches.

And similarly, I mean, so I guess an analog for that would be that kids... These days, you know, they stay with their parents way longer than we did in the 90s. And I think the big change, looking at the social research, is that the big change was around the incident of the smartphone.

Kids started behaving very differently. So to a degree, they're already kind of in their own world and they're prepared to stay with mom and dad longer. So maybe that's an argument that there'll be more densities. But again, this could just be an argument for more dense cores and less suburbia. I don't know.

Yeah. All right. We got to that point in the episode where you need to ask if you've got a property done before us. Do you have an example of a mistake that somebody's made, perhaps yourself, that we could all learn from?

I learned this one, I heard this from a buyer's agent yesterday. Uh, so it's not exactly mine, but I thought it was really good. So, uh, they were saying that, yeah, so they really had these high expectations for a house with a pool and such.

Um, and it was, uh, you know, at the higher price for the suburb. And, uh, but they were saying, look, I won't pay that. It's not the median price. So as if the median price was a recommended retail price for the suburb or something.

That's gold.

So hang out for that.

Not understanding that that's actually the bill curve and that's the margin. That's gold. So people, be that a warning against a buyer's agent that suggests you should only pay the median.

That's it. If you have a question that you'd like us to answer in an upcoming Q&A episode, you can send us a voicemail or written question via the website, theelephantintheroom.com.au or you can email us directly at questions at theelephantintheroom.com.au.

If you like what you're hearing, please share this episode with others you feel would benefit. And while you're at it, why not leave us an iTunes review? Five stars would be great.

I know that sounds a bit cringy, but we have it on good authority that every review helps make it easier for other people to find out about us and hear what our amazing guests have to say.