Luke Metcalfe on Why AI Cannot Pick Property Winners

Imagine you asked seven different analysts to pick Australia's top growth suburbs and they all came back with the same list, not because they compared notes or even used the same methodology, but because they'd all inadvertently been trained on identical source material before they'd even started.

If that could happen, every investor using an AI tool to find the next growth opportunity is essentially asking the crowd what the crowd already believes and paying a premium to buy into it. Luke Metcalfe made exactly that hypothesis on this show last year and then went away and tested it.

17,000 suburb predictions, seven GPT versions, eight states, up to 38 months of actual price data. We'll get into what the results showed and what it means for investors using AI to pick suburbs and do their due diligence.

And then we'll land on what actually does drive capital growth when you strip the consensus noise away. 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.

Hi, I'm Chris Bates, ex-financial planner and mortgage broker, currently ranked number three in the annual MPA Top 100 Mortgage Broker Awards. Before we get started, everything we talk about today is not personal advice, and we recommend you engage the services of a licensed and experienced professional.

Our guest today is Luke Metcalfe, who we've had on the show a number of times. I do remember our first conversation about machine learning eight years ago now, if you believe it. And Luke, you've been in this space for a long time.

So when you talk about AI capability around property investment, we definitely listen. It is really good to see you again. We cannot wait to hear what you discovered through your research.

Thank you very much for having me, guys. It's always a pleasure. Luke, always good to have you on here. Let's get into the research and you putting, I guess, what the AI models thought could predict price growth. I mean, what was some of your takeaways from going through that process?

Yeah. People are increasingly going to GPT for research on everything. I do it even to get a nice puree in the morning for a nice recipe. So we have customers, everyone's finding customers using GPT. So I thought, well, let's put GPT to the test.

Let's see what's its actual track record.

And we can do that because there are previous versions of GPT that were stopped. So every three, six months, there's a new version and then it doesn't learn anything more. So it's like talking to someone from the past.

So it's a bit like if someone woke up from a coma after three years, their mindset is still in 2023. So we can ask a 2023 model, where do you think suburbs are going to be in 2026?

And the beauty of this research is it's completely replicable. So we describe exactly how we did it. We put out the CSV files like you can actually look. See, we give the exact prompts that we used and anyone can go talk to these old models and see where these suburbs actually ended up.

So we can hold these AIs to account in a way that- It's like a time machine.

And it's like you go back and it literally does not know what actually happens, but we know what happened, but it doesn't know what happened.

That's right. You can't do that with guests on Elfin in the room. You can't say, now answer as if it's 2023, and then we'll look up the data for 2026 and see how you went. Yeah, so the upshot is that it underperformed the market.

So you would have been better off choosing at random. So if you threw a dartboard at Australia, if you threw a dartboard at each capital city across the country, across time, it underperformed the market.

So I think on average, you would have lost about 13 grand listening to ChatGPT versus just randomness, pure luck. So it was really bad at forecasting. And of course, as I mentioned in that last show, it's not designed to do forecasting.

It's not trained on the future like that. It's trained on the next word. So it's coming up with things that sound good,

things that fit, commentary that fits your question, that answers, but it's trained on a whole lot of stuff, which is all these top lists that are coming out, all this promotional copy, and it comes up with very generic answers.

So, you know, it'll say Paddington has a great vibe and is connected to the city and, you know, Brisbane's getting an airport, all this really generic stuff. but it's not actually going through the process of reasoning like even a pundit forecaster would do, maybe to a degree, and it's certainly not trained.

So when there's an earthquake like the budget, it's certainly not suddenly changing its worldview based on that. It's probably going to get even worse. Sam Altman, the CEO of OpenAI, he doesn't wake up in the middle of the night worrying about the Australian property market.

He's trying to create a really smart, intelligent, generic thing, but it doesn't look at property data. And to your point in the intro there, so... Even other forecasters, they're all just training on core logic data.

So you can end up with the same answers because you're trusting their medians, which is something that we avoid at Microburbs. So we don't end up with our own. We don't end up with the same as others. So, yeah, it's really got to be careful.

And now you're having this era of buyers agents using AIs as well. And if they're not fed, if the AI isn't fed great data to start with, they are also charging a lot more for just these reports that are easy to receive. But can you actually trust it?

Not really, certainly from a forecasting point of view.

You've got a few issues there. You've got a lot of buyers agents who haven't been around for long and so therefore they don't really understand the property market. They sort of haven't been through cycles. They haven't understood multiple markets. So therefore they're not subject matter experts necessarily.

And so then they're feeding the AI the data. But again, if you don't know how to interrogate the data or work with that data, then you're not necessarily going to know what output you're going to get or you're not going to understand how that output was even arrived at.

And so you sort of have to be, you have to understand not just data, but how data can be misapplied or misunderstood or, you know, how to lie with statistics, basically. That's what gets me going.

I think you need to understand what you're expecting to get in a way so that you can interrogate that and think, does that sound right or not? And then go back to find out more, right? But if you just...

take what's given to you and go, right, well, that must be good because it was an AI that worked it out, not me, then that's highly dangerous.

Yeah. And yeah, so a lot of people in the property space are just following trends. So something's hot, it sells well, you copy it.

These trends, if you've been around for a long time in the property space, or you've got property data going back 25 years, then you see that these trends go into reverse. So something is hot for a while, and then it goes from regional to the capital cities,

Sydney and Darwin are inversely correlated. So there's a whole lot of things that it's like, yeah, if you had a bit more experience, if you had access to more, you would see that it's not. But the incentives on an individual buyer's agent is obviously primarily to sell. People love trends.

People love following the herd. There are certain narratives that make sense. But it's the narrative followers that are often typically the ones that underperform the market. And they're often getting in too late as well. So by the time that the story has got out there, you know, Brisbane's having the Olympic Games, right?

It's like, yeah, that's great to know that before anyone else does. But is it that good? When everyone does, like, isn't it already priced in?

So there has been this explosion of like data-driven, you know, AI, buyers agency sort of, you know, world and sort of taken off. Like, what are some of the big challenges with those like models though? Because you mentioned they're around the median prices. I hadn't thought about that.

You know, it's not suburb individual property levels and what drives one over the other. And I think what some of the other, the challenges of sort of building this data-driven model sort of buyer's agency. Like, you know, it sounds great, right?

I buy on the data, but ultimately it's what data, I think, they're kind of creating a story, aren't they, somewhat?

Yes. I have a competitor who says every data point is a story. He said it on this podcast, no less. Yeah. What we've found in the data is that it's really small. The smaller the market, the more of them there are, the more you can actually predict it.

So we've got research out that's shown that at a national level, two out of three properties move in the same direction. They either out or underperform the market. And the other third go in the opposite direction.

And even down to a suburb level, one in 10 properties underperform the market when the other nine outperform or vice versa. When you get to a street level, you get to 97% agreement. So streets really move together.

So when you're dealing with median prices on a suburb, it's a bit of a mixed bag. So it's like, what's the median age in your family? It's like, well, you've got three kids. It's going to be pretty young. If you're a single mom, it's going to be pretty young.

But if you've got grandma in there, Like it's going to be very different. And these are quite small. Like there isn't that much turnover in a particular suburb. So this whole median thing is all over the place.

So if you just look at what a median property is in a suburb from year to year, so, you know, North ride, it can go from being a, you know, a veteran's home or like a, you know, a humble post-war shack up to, you know, increasingly then 70 stock.

And then, you know, up comes the new stuff. So you can go from a 70s home that's a knockdown or a 50s home to a really big one. And so about, yeah, so about 30, 40% of all growth is just improvement in the home. Yeah. And the property industry takes that credit.

But of course, that's no help to the average investor. So people are building up around you. Let's say everyone in your suburb drives a Datsun 180B. So the median price of a car is $50. Then everyone else goes and buys a Lamborghini. That doesn't mean your Datsun 180B is worth $200,000.

I love your analogies. They're really funny.

Well, Datsun 180B is probably going up in value, to be honest. It's vintage now. I mean, it's heritage. It's vintage, right? It's probably like the old established house in the suburb.

It probably is going up, but the guy who's got the brand new sort of flashy house is probably depreciating in value a lot as well.

That's true. So yeah, that's a huge thing. So much of our research, increasingly, I try to do all of our research now looking at resales and not median values. So it's like, what is the actual return?

So individuals have bought one home on this date, they sold later at this date, what's the growth? What are their costs as well? And on that basis, it's much better anyway in that you can look at the actual fine-grained locational context, yeah?

So we'll have a competitor that will say, oh, that's a bushfire suburb, right? Your bushfire zone is only on like a third or a quarter of it. It's like, what's the actual effect in the bushfire zone versus elsewhere? And we've put a lot of work into taking out renovations as well.

So it's like make the assumption of the typical investor that's just going to buy, not do it up, and, you know, 1% maintenance or whatever it is. And that way we can see much more fine-grained, much more empirically.

And given statistics, when statistics you need, the more you have, the better as well, the more confident you can be on the outcome. So this whole like –

spreadsheet modeling based on these woolly medians it's like the median you can debate endlessly what a median price is anyway so there's like to what degree do you take into account how much what the stock is that you're that's actually selling versus how you value the stock that's there different kinds of stock sell at different times so we have to talk about medians on microberbs because people always ask but if everyone listened to me we'd only be talking about growth on resales

I love that. And that's exactly what, you know, one of my favorite reports is the pain and gain report from CoreLogic, but it doesn't go deeper than nominal, you know, gain or loss, right?

But if they did, you know, were able to get in there like you're doing and really dig in and say, take out the renovation costs, take out all these other, I think we'd have a really clear picture of exactly how much of Australia is worth less than they paid for it versus how much is worth more than they paid for it as well.

Because when you extrapolate that out, you think, you know, say on average, plus or minus 10% of properties that sell, sell at a loss, right? And I know it's gone up to as high as I think 14%, a really bad market.

And it's gone down as low as about 8% in a really good market. So it's probably more like 11% plus or minus. Then you dig into that and you go, well, there's holding costs. There's renovation costs. There's a whole bunch of things.

And you think, well, how many people are sitting on a property that if they did put on the market would actually sell at a loss? And it'd be really interesting to predict or to really look at, well, all the stock in the country.

Could you sort of extrapolate that out and make a guess on that one? But we're sort of diverging here. Go on.

Yeah, no, I'll pick up your point on sold at a loss. So looking at individual property sold at a loss, that is predictive of subsequent capital growth.

so you can see in the data quite clearly that there's a lot of hesitation for regular people to sell at a loss even if it's a just a dollar value just psychologically it was like uh when telstra the telstra shares yeah came out but it went down massively and then it came back up then then it like flatlined at the point where it was like where they bought it for even though like there's been inflation since then you should be thinking about relativity

to other investments. Real costs. But still, people are like, oh, I've got back to where I want. I'm even. I'm square. I can do that now. I can't. People don't like the concept of taking a loss.

Loss aversion.

Yeah, so that actually is a good predictor. So if people are so desperate that they're prepared to lose face and sell for a lower amount, that hurts the subsequent suburb growth.

Wow. But that report doesn't come out in that micro level. That report only comes out in LGAs. So it's impossible to take that particular report.

Yeah, we have it on the micro level. So we're constantly going, okay, what radius and over what time period? And we can change all those variables to find where the optimum is for maximizing capital growth.

Let's go back to your research here because I want to sort of get back into the AI side of things in terms of non-data scientists, non-property experts using ChatGPT, Claude and the like to make property recommendations and property strategy recommendations as well.

And it's fascinating, you're talking about the different versions of GPT. So you went through seven versions and so at seven different points of time, you go, what would you predict? and six out of the seven underperformed the market.

So this is pretty alarming for anyone who is turning to chat GBT and the like. It's not just chat GBT here. This is the large language models, right? And the only one that didn't underperform the market, only beat it by 0.4% over a nine-month period, right?

Yeah, it's margin of error. That doesn't mean it's a good model for that. It's just luck.

So it's like, that's just hands down.

Actually, yeah, that was a more recent model. And I think the reason why it wasn't as bad is just a bit of momentum. So sometimes they can be okay on momentum, but totally crap after a year. So like totally crap further on.

But this has always been the problem. Predicting what's going to happen in the property market or predicting price growth or price falls, for that matter, in any period more than a year for a human being has been almost impossible.

Most forecasters will say, look, I'll hang my hat on two years, but it's a bit shaky. And we wrote, what, five fuller forecaster reports, Chris, where we reviewed... Year on year on year, you know, the forecasts that have been made, and I think the best year, 15% of them was somewhat accurate.

So that means 85% on a good year are way off beam. And we also, every year, we did a review on older predictions and really saw, okay, 10 years has gone by, how did it go? And most, hands down, most of them were way off beam.

So we know that humans can't predict into the future with property. And what we now know is that AI can't do it either. Yeah.

Yeah, so macro is very complex. So macroeconomics, there's too many things going on. What happens in the left-wing faction room in labor, that has an impact. They have the power.

So macro things are very hard for a computer to take into account, and humans aren't good at it either. So we have research out there that shows that The forecasters altogether, they're only just within the margin of error beat simply being a bull, simply always saying things are going to go up.

Um, Yeah. So, you know, you can take Louis Christopher, his scenario-based approach works a bit better, but of course, by having scenarios, you are taking out some of the risk yourself. Yeah. And these people can be smart. It's just extremely hard problem.

So if there's just too many factors and there's not enough data, but as you go more and more fine grained, So a suburb level, nine out of 10 properties move the same way. There are lots of suburbs. There's been a lot of time gone past.

You can construct a model that can predict it in the same way that you have models that predict stock market. So we've done that at Microburbs and it very much outperforms. It doesn't know exactly, but like, so the top five picks 15% of the time, it gets it wrong.

So it goes down, doesn't go down totally in dollar value, but 15% actually underperform the market.

Do you mean when you get it wrong, 15% underperforms instead of 10% underperforming? Is that what you mean?

As opposed, what do you mean the 10%?

Well, you said before that nine properties out of 10 will move in the same direction.

Okay, so yeah, there's the nine out of 10. So that's referring to suburbs actually observing them in the past. What we did was we looked at for each suburb and each sell year and buy year, did they outperform or underperform the market? So that one, 90% of them moved together.

So which is another way of saying that even if you have a good suburb forecast, there's still a reasonable chance that the individual property that you bought will underperform. But if you've got a street level forecast, so on a street level, houses almost always move together.

So 97% of the time, because these are really genuinely separate submarkets. And at that street level, you've just got way more for the AI to work out. So in the same way that GPT is smart because it's read an extremely large number of words. So all it's done is predict the next word.

But because it's read trillions of words, AIs get very smart when they have lots to train on. So in the same way, if you have prices, median prices by street, over a very long period of time, we have way more streets and we have suburbs, then you get yourself a better model.

But yeah, that's not the reality of how most people are doing forecasting. So at best they're doing, most people they're doing at best suburb or they're doing like these macro forecasts, which are, so even if you were the perfect macro forecaster,

there's still one chance in three that the suburb that you chose underperforms the market. The Australian property market is actually a very large number of sub markets. And you can experience this as someone looking for property. If you see the same people at inspections, even if it's not the same suburb,

you know, they're your market. They're who you're competing against. If you suddenly within the same suburb looked at a five bedroom house instead of a three bedroom house, you're not going to see those people. It's a different market.

You know, the five bedroom people are going to have shinier cars and, you know, the benchtops shinier and they're going to, you know, they're going to want to pool. It's just different people. So don't put them together. So that's the problem.

Fundamentally, in my view, that's the problem with data is that it's like a, bad soup. So it's just all this crap is put in together and it doesn't taste good.

I mean, do you think that that's where we're at though? Like in the models though? Like we are, because once you get to a street level, you can basically, and this is why when we first found your sort of website, it was probably a decade ago.

And I was like, oh wow, someone who's actually truly gets livability, right? Someone who actually understands sustainability. Suburbs are different and parts of suburbs are different. And this is attractive to people for these lifestyle reasons. And within that suburb, this pocket here is really attractive because it ticks all of these.

It's got the walk score. It's got access to the amenity. It's not a rat run. But are we able to even do that yet? Is anyone able to sort of, you know, because there'd be millions of streets almost across the country, maybe not millions, but are we at that level yet? Yeah.

The tech is perfectly good at processing very large amounts of data, even back in 2016. So yeah, you can feed in every single street. You can feed in every single transaction, every single resale. So there's no problem with the computers scaling to do the number crunching. Streets are great. Yeah, absolutely.

So yeah, we work primarily on resales now. So we're looking primarily on an individual property level. And yes, there are things that we don't know in data. So there's things that you know as a buyer's agent, as a real estate agent.

These things will often show up in things like turnover rates. So once you're looking at that individual level, so if properties have turned over lots of times, that comes up in the data as well.

It's kind of interesting where the ASX goes from week to week, but what you really want to know is what's going to happen to a particular stock. You will buy into a particular stock if you want to get a massive return, not the overall market.

And so you've got this new data coming in, right, which is a shift to investment fundamentals or access to credit. Obviously, this post-budget, how are you... feeling like just someone, a big shift like that, right? Because it changes human and both buyer and seller behavior.

When your big force driving prices up is someone's ability to or desire to sort of leverage, how are you thinking that's going to change things? Like, you know, someone who's just looked at this, I mean, you've written, what, 30-odd page research reports on this.

When you do a research report on this in a few years time, what do you think it's going to do? What do you think it's going to highlight?

Yeah, so we calculate median prices every week, unlike everybody else that do monthly. So we're very interested. I'll be watching it very closely. It's a very exciting time. As you guys have said, it is a major earthquake, the budget.

So I don't pretend that things are the same by any means as they were before. The incentives have massively changed. So, yeah, we're looking very carefully at how many people are negatively gearing. What's the investor concentration?

There's going to be a whole lot of suburbs that are simply no sense in coming in as an investor now. And you've got a massive trend towards, you'll see a massive trend towards investing in the family home.

You probably have seen loads of clients there where their best investment has actually been the family home anyway. We know that even before, there's so many things about where things are going that are consistent with what we've already found. And it's just going to increase that.

People are going to be more aware of it.

so investors underperform home buyers anyway even before any of this came so just on capital growth investors get more they earn like one percent more per year and also what we found is the further away the investor is from the property the worse the performance you know the properties in darwin don't do nearly as well as if you're buying in the same city as you yeah

Okay, but Darwin has, with recent history accepted, Darwin's done pretty bad for the last 14 years.

Yeah, you can make other arguments.

Let's pick another suburb, another city, I should say. You know what I mean? Because if you use that, I can see that everyone's going to perform badly if they've owned it long enough.

Yeah, yeah, yeah. So there's definitely other factors going on there. So there's obviously big capital cities outperforming and all that sort of stuff. But yeah, so we have seen in our data over and over again, there's lots of things that I wish weren't true. So we have awesome infrastructure data, but

infrastructure, I can't find the relationship to capital growth. It's not something that's shown up. I wish it did. I've tried all kinds of ways. I've asked people who talk about infrastructure. So I think you could probably get some of the way if you look much more carefully at supply. But the problem...

There's a whole lot of things like infrastructure that if they're building that, they're also adding lots of supply. And the supply, the infrastructure might be overstated in people's minds. Oh, there's a train station. Yeah, so what? We're a car-driven society. Most people are still not going to take the train, right?

Infrastructure, no, no.

I would love it if our hip score that we're most famous for, I would love it if that was a predictor of capital growth, but it's not, not consistently. What we found, now we've been doing this for longer, another 10 years, is tranquility is a really good one.

Because I remember you saying eight years ago that tranquility actually was a predictor of low capital growth.

Yeah, so I didn't have enough data back then. Right. So I was looking, I had data going back maybe 10 years and now there's more time to look at this stuff and we've got ever better models for this stuff.

And this is a Tranquility that is actually, so previously I was making them for the home buyer market and that was like what people consider to be Tranquility. These new ones are actually honed to capital growth. So they have that as the goal. And so basically there's all these things we're finding.

It's just like anything that impedes supply is really associated with capital growth and things that people, home buyers actually want are the things that drive the capital growth.

So even though it's very trendy to talk about inner cities and amenity and diversity and density and trans and all these beautiful slides and things,

The data shows quite consistently that freestanding homes, communities that know how to fight restrictions, so communities that get together, close-knit communities, so that's not your inner city, you know, students renting for a year or two. It's not those kinds of places.

Now we're seeing over the long term, another thing is we're breaking it down and checking it works in each market. And this is more important now after the budget because the actual utility for what homebuyers actually want is ever more important.

I think investors understated the importance of who the next buyer is going to be because broadly speaking, two chances in three, the next buyer is... is going to be a home buyer because two-thirds of the market are home buyers, right? Yep. So you should have always kept them in mind.

And that's going to be even more because there's going to be fewer investors coming into the market.

It's so interesting you say that, Lou, because for years we've been saying to our investor clients, you have to think like an owner-occupier. You have to be thinking and buying with owner-occupiers in mind because they're the ones you want buying this property when you go to sell it.

And we say to owner-occupiers, think like investors. And again, don't just buy it for you and your unique needs. You want to solve those problems, but you also want to make sure your future buyer is also going to be a nice broad pool.

So now you're saying that you actually have data to back that up.

So we know that owner occupiers outperform. We know that they outperform investors for capital growth. The one thing that investors tell you to their blue in the face is what they care about. They're not that good at it because they're just following trends, right?

And they're looking at numbers and they're divorcing themselves from what it's like to live next door or that public housing or in that crime ghetto. Yeah.

It's exactly right. It's like, oh my God, I love, I mean, we didn't prime you for this, but it's so interesting that it's, you know, you've got all this additional years of data and it's just coming out loud and clear because, you know, we've known this to be true.

We've seen it and felt it on the ground. You see who's competing, competitive auctions, forget the buyer's agent led frenzies in regional towns, you know.

At an actual auction where the actual buyer is there competing for themselves, the investors will not be fighting hard against owner-occupiers when they've all got the same amount of money, that's for sure. The owner-occupier is going to be the emotional buyer.

Yeah, absolutely. So stop thinking about millions of new renters coming in and all of that. They come and go. Another thing we've found in the data is that houses that have a home office in them perform much better. So potential to upgrade.

This would have changed since COVID though, right?

And a bit by tranquility as well, I suspect. Yeah.

Tranquility is big over a long period of time. As I said, the specific tranquility index we have is deliberately designed to predict capital growth. So it's a mix of settings. So it's not a simple matter of people's subjective view of tranquility, which is what our old scores were about.

So these are specifically designed to hit capital growth, and we don't reveal exactly how we do it, but there are certain combinations of factors in that. And I think the big underlying thing is just a community that can defend against supply is way better.

So people's revealed preference will be to go for the home where they can have their own castle. It's less trendy to say, but you've got your own place, you can defend it, and it's less... less about amenity. I mean, there are certain amenities that are bad.

So one thing we found is fast food hurts your capital growth, proximity to fast food.

Wow.

Yeah, no one wants a KFC across the road, right? But I think, I mean, Or even within two kilometers. Yeah. And what?

How many?

Even within a couple of kilometers. Yeah. Like in the same suburb.

Like fast food. So yeah, we've got the ability now. Maybe it's a temptation thing because if I can walk two kilometers and get the food, I'll probably just pop out and get it. So maybe people are like making a strategic play to live away from the fast food. But I mean, the tranquility,

Is the suburb level, you've got obviously the trees and in climate change, might as well be playing to this as well, right? The cooler suburbs, I feel like people are more aware than that than they probably ever have been. But how have you dealt with the zoning shifts?

Because in recent years, what you would have thought was pretty secure zoning, like you said, fighting developments, right? But then we've seen some major shifts with state governments, particularly in New South Wales, because it shifts the tranquility. You thought this suburb was tranquil.

You know, you thought that somewhere like Mossman or Rose Bay right now in Sydney, they thought that it was always going to be quiet and you'd have a little high street, you know, 4Ks from the city and all of a sudden, bang, there's more apartments coming.

So how are you sort of dealing with like planning controls and... You know, particularly now, if there's a bit of an understanding of how it's going to shift over the coming decades around sort of key train stations, are you now putting that into your models as well?

Yes, so I already had it on my to-do list pre-budget to dust off the old rezoning forecasting model on a property level. So looking at zoning over time and there is definitely demand from the big end of town, the people who quietly are going to benefit from this stuff, from the changes.

So yeah, absolutely looking at it. And yeah, watch this space. Lots to say they're coming. Obviously, in general, though, for your average investor listening to this, you just look at how much amenity is there. And the more amenity that's there, the more chance that there'll be high density.

I can tell you, having consulted to fast food, that they do look at how much new density is coming in. Yeah. So right near me in East Chatswood, there's an Anaconda and a Woolworths in the middle of an industrial state. No one's there.

But there's going to be a whole lot of new units there, yeah? Yeah, yeah. So when other people see convenience, I see supply, and supply is the killer. That's a really good way to think about it, right?

Yeah, exactly. You've got great trams. You've got great buses. You've got a train station there. you know, actually you might feel is really convenient. What a great little pocket. No one knows about this, but it's actually also highlighting that it's actually a great place to build more housing.

And it's going to be very hard to defy, you know, the planning controls because it is just so connected to, say, public transport.

I think that's where, you know, in the Northern Beaches, for example, like not much has been built here or is likely to get built here because the access with public transport is really bad. But if you go to somewhere like Warringah Moor where there's lots of bus services, there are D.Y.,

all of a sudden, bang, they can't stop 2,000 apartments getting built because it's a very convenient place to get in and out of. So that's a really good little tip for people.

Yeah. Yeah. So with the massive pressure on supply now, so you've got the spruikers out there pushing new builds, We're going to see a new build bubble, I think. But there's also going to be pressure from these corners as well as the big end of town.

There's going to be more money rolling into councils to try and get rezoning to happen. And it's going to be a big fight. And yeah, so you've got to be ever more on the watch out for new supply, particularly supply of units. Houses supply in a town is good news.

So I should preface by saying that out in towns, if you can actually grow a town and get amenity, awesome, particularly houses. But units in a city is really bad news for capital growth.

So the unit market hurts the housing market as well as, of course, totally cratering the unit market.

So you're giving lots of ammo there to NIMBYs. So look at Melbourne's urban sprawl. You know, when you look at aerial shots of outer Melbourne, it's brown in a lot of places and then you've got these little patchwork quilts of grey roofs.

You know, and there's not a lot of infrastructure out there either. How is that good for a town or is that different?

Yeah. So Melbourne, big capital city, high demand place. You've got massively oversupplied places like Tarnate. I don't have a specific Melbourne model. So I won't, you know, as a researcher, there's just some things I've, some questions I've asked and others aren't.

I'm not a pundit, so I don't have like ready answers for particular places. Yeah. Fair enough. But broadly speaking, as a capital city, yeah, there's a lot of danger.

Yeah, Melbourne is interesting because, as you say, there's just so much more infill to go and what happens to that. Whereas somewhere like Sydney, it's just going up in so many different places. So polycentric. It's a very different experience flying into Tullamarine as it is to Kingsford Smith.

Also, where the airport is located, Kingsford Smith is a lot closer to the city than Tullamarine is, but also Melbourne's already had their vertical growth, and it's really damaged their unit market for years, whereas Sydney has had it in areas, but not sort of probably as widespread.

It certainly hasn't seemed to have the same impact on the entire unit market than the Melbourne area.

One thing I have found is that the unit market, so supply in the unit market is quite broad. So seven kilometers out, it can still affect prices. So we're always looking at like how far back. So it seems that unit people are prepared to really jump. Maybe they're less likely to be locals.

They might be foreigners. And so they just think, oh, it's still on the train line. So it's okay. Yeah, so there's definitely like... So suburb level suppliers for unit is not the full picture. You've got to look at way further. That's really interesting.

Yeah, they're not building much here, but they are two stations down. And then other potential buyers saying, well, it's a bit cheaper and it's only two train stops. I mean, we are going to see a massive... you know, after this budget, a shift in, you know, what people were trying to sell, right?

Like it's very hard to sort of sell established property as a, say you're an investment buyers agent, or you want to sort of be making money off people buying property, let's say. It's really hard to recommend that strategy. Like it just, it is, it's only a

a small handful of people are going to be able to afford the negative cash flow and borrow the money from the bank, to be honest, and let alone then have to pay more capital gains tax. So there's this real incentive pushing people towards new property, right?

They can borrow a lot more, they can get negative gearing, they can depreciate it, and then they can save on CGT down the line. And even in the last week, even though I was unsubscribed from those things, I've been hit with all these new developments. Somehow I'm on a mortgage broker list, right?

And I'm getting flooded by duplexes and townhouses and regional sort of growth corridors. And so I absolutely was seeing this. This is where all the investors are going to get sort of attracted to. When you look at your research and you look at

The people who've invested in house and land packages, in new townhouses, in sort of high density apartment, what's your sort of take on the data behind their performance compared to the person who you say, you know, just bought a house or bought an older apartment or something?

Absolutely. There's a huge amount of this and there's going to be a bubble. We do have data coming out soon. So we've got a model now on build year. But the research hasn't come out. So again, I'm not going to be pundit and say something that won't be in the report.

But yeah, we will have more to say about what year and how much capital growth you get out of it. But of course, even if it is okay for recent, you've got to take into account depreciation. which is much more harsh on the recent properties. But yeah, absolutely to your point.

Yeah, it's gonna be very hard for investors, new investors to come in on markets where you're competing against people who are negative gearing. The rents will be artificially low. The yields will be terrible.

And, you know, you're having to pay all the capital gains tax as against the incumbents that for the time up to June 27, they weren't.

One thing I'd like to add is it was not your question, but that's one thing I'd love to insert for people to think about is that another thing we found in our data is that investors don't hold on to properties nearly as long as people think. I see your mouth opening there, Chris.

Can you just explain that? Because I was doing research on this last week because I was trying to figure this out because NHIF, it's basically, they said that it was like 50% or something only had it for a year or something stupid. National Housing Australia, I think it was.

And I was like, surely that can't be right. And I was trying to find some better data on this because I think you're right. Like a lot of investors, they go in with this long-term mindset or they go there just to cash and grab the money.

And I think that's a lot of regional investors in recent years. That's their plan. It's like, I know this is a momentum play. The buyer's agent often doesn't tell them that. They say, look, this is a buy and hold, buy and hold. It's great long-term.

Because they don't want to feel like they're trying to time markets. It's by good assets. But a lot of those investors are going to just as quickly as they bought they're just as quickly likely to sell if they feel like there's no longer returns there.

And I think a lot of other investors typically just haven't been able to hold properties.

They've gone in there and they've had the idea of this is going to be a long-term hold, but then their life plan changes and now they need to upgrade their home or they're getting a divorce or they can't refinance.

So there's this idea that every investor is just going to buy a property and it's going to sit there long-term. This isn't the case, right? Yeah.

Yeah, my guess would be how they got to 50% within a year would be that they looked at subsequent listings and assumed there were different owners when actually it might have been just a slightly different listing on the portal. Development stock or something?

Yeah.

Yeah, but I don't know their research. But I can definitely tell you that the investor who's bought most recently on the street is the most likely to sell next.

Oh, wow.

It's actually – you would think that it's like, oh, yeah, on average they hold for 10 years. So, you know, forgive my statistics here, but imagine a bell curve. That's what I thought it would be. But, no, it's actually like that.

So the shorter – yeah, the less time that you've held it, the more likely you will let it go, which is true in general in my data science for just –

use of products for example how long do you hold something i would guess marriages would be the same so you know a marriage is most likely to divorce early on i don't know if it's right at the very beginning um so yeah it's a phenomenon that's not unique to real estate

And yeah, so we found that out of people who sell their property, obviously there are some that have still held it from when they bought it and we don't know when they're going to sell.

But out of those, when we knew that they were going to sell based on long-term, the median hold time was 5.5 years.

Whereas for owner ox, if I remember correctly, it was 6.5 years.

So they're actually, the owner ox are actually holding on longer, even though they're going through all these life changes, you know, they're getting, first they pair up and they get the awesome terrace in town if they've got money and then they move out to the burbs and then they empty nest and they,

All of that. Yeah, I agree with your overall thesis that this is a really big thing and people may be underestimating how many investors are going to get out anyway just for other reasons. Just because they do, God knows why. Probably your surveying with PIPA could find out. You may already know.

So that is going to lead to there being less stock for renters.

Yeah, because that's what I've sort of been trying to get my head around. Okay, so now this is the shift.

If they go, if we're not getting more investors into the market, then because of it's not as attractive and we really need the current investors to stay in the market to keep the rental supply right. But when you look at the current rental supply, let's say it's over 3 million properties.

50% of those are owned by people over the age of 50 who are going to be more attractive potentially to sell and get money into a tax-free super account. Bearing in mind that 80, I think it's 85, 90% of properties are owned by people who have one or two properties, right?

So they're not the big property investor. The big property investor can't offset that supply. They're already capped out. Yes, they can come up with some creative, you know, are buying companies, but they're already capped out.

They've been buying heavily for the last five years and they're finding it harder than ever to get finance. So 50% are owned by the over 50. The other 50% are owned by 30 and 40 year olds.

And the problem with them is they went in there with a bit of a shorter term mindset, a lot of them, because they went in there with interest rates are going to fall and I'm going to be able to hold this property.

And often our rent investors buying it, well, they're going to say, well, I'm going to bail. And so... The issue is you've got a lot of people incentivized to sell, but then a lot of new people not entering the market.

And so that's great news for first-time buyers potentially, right, over the next coming years.

And I think it's great news for the overall property market because you're going to find that, you know, even houses in good parts of capital cities are going to start to come on the market because, you know, an investor is going to say, well...

Yeah, I'm going to start paying a lot of CGT from July next year, but it's going to be a lot of problems in our rental market, I feel, because a lot of people are going to bail. And rents have already gone up 50%.

So this whole idea that rents are going to go up another 50% is a bit of a fallacy. I mean, people are already... There's a natural affordability sort of curve where people just can't keep paying more and more for rent, right? So there's going to have to be a ceiling.

And so yields aren't going to likely go up much because rents are going up. And then... The cost to hold is actually, if anything, going up. Everything from tenancy laws to interest rates to land tax, it goes on.

And so a lot of investors, they won't be able to sustain that cash flow for very long or recent investors. So I don't know, is that sort of along your thinking? Is anything that not makes sense to you or you want to challenge, I guess?

No, I think everything you said made sense to me. One thing to note, though, is that talking about investors as a group, they have massively different outcomes.

So if you make it just to the very bottom of the top 10%, you're looking at 13% capital growth per year for the first four years of the hold. So that more than makes up for any change the government does.

But if you're in the bottom 10%, then you're looking at really going backwards fast.

I think you started saying at the bottom. Did you mean at the top? At the top, you're looking at 13%.

At the top, you've got massive capital growth. So when we start looking at resales, it's not like it's all clustered really closely at a single thing. There's a wide distribution. So there were investors. So at the top, they're doing things that you're not supposed to do, right? They're not following the herd.

They're doing things out of cycle, yeah? So there's still people making lots of money. There's always people making lots of money. You can hold me to this, guys. There'll be people who bought today who make loads of money when they sell in five years' time. It'll be a sizable proportion. Of course.

I've had quite a lot of people come to me. This has been one reaction to the budget announcement last week. A number of people that have come to me because they actually have rental income from property and they're in the lovely position where they can use negative gearing.

And so they are already wealthy in property and they are going to get wealthier because they're seeing an opportunity to build on that. And potentially if there are price falls and there's opportunities to buy with less competition and there's a general, general sort of lack of confidence in the market.

Look, Sydney's clearance rates down 55%. I don't think I've ever known it to be that low and it's been like that for weeks. you know, they're seeing opportunity, some of these people, but they're already, you know, high net worth individuals.

That's another thing that I was thinking, I was talking to Kate Bakos this the other day, and I'm like, now I'm actually seeing that, you know, that's another disadvantage or widening the gap, really. You've closed the door on rent festers.

You stopped first home buyers with that opportunity to get into the established property market. And in fact, there's the 5% of first home buyers are rent festers. So last year, there's like 120,000 first home buyers. So that's... six grand, 6,000 rent investors.

And then you've got seven and a half thousand new buyers that the federal government thinks are going to come in as a result of this budget. So there's a gap of, you know, one, one and a half thousand extra first home buyers.

That's the net effect using their numbers of, you know, new first home buyers into the market. But then you're actually seeing a situation where people who are already wealthy can get wealthier.

So if you own lots of properties, you don't care about whether one is negatively geared versus another. It all balances out. It's like it matters for cash flow for the ordinary working man and woman.

Yeah, that's the point. Yeah, the big end of town wins. It's like people using it.

Yeah.

Yeah.

They were using it. Yeah. And boomers that are already established, they can just offset. So they're fine. Renters, yeah, rents will go up. Yeah, there's going to be less investment stock. I reckon it's going to be uneven.

So I think that some places are going to have, with very high investor concentration, poor yields, high proportion negative gearing. So relying on negative gearing as a strategy is, So, you know, your surface paradise kind of place that is going to be.

Yeah, it's going to be good for home by first time buyers. So it definitely will lower the price. They have to lower the price to let them in. So, yeah, for people listening to this, you put more money into your own home and invest

in that home so we'll be doing lots more research on renovations what are worthwhile doing and or you be prepared to be a more researched more sophisticated investor that's going to do things that everyone tells you at the barbecue that sounds a bit weird so we'll be doing more work on what these people are doing interesting yeah it's gonna be interesting also the um

you know, holiday market, you know, like you got your city pad or you live in the city and you want to buy a sort of a, a house that you're going to Airbnb, you know, a bit of the year and you're going to use it a bit like you can't negative gear it.

Right. Like, so the cost to hold it is just so much higher now. And, you know, you didn't really care how much you Airbnb'd it because you would just say whatever the cost is and, you know, and I'm just going to, Use it occasionally.

That together with land tax, you know, I think it's a place like Mornington Peninsula, right? Like, you know, I've got my house in the city paid off and I've got this place negative geared down Mornington Peninsula. That's just another hit to that market. Yeah. Even like Palm Beach in Sydney.

That was already a market. Yeah. Like that's another one that's really driven by second home market. Yeah. Unless you've got a lot of wealth and it's paid off and you just go, I don't really care. Like that's a bit different, but yeah.

Any of these sort of typical sort of second pad locations, unless they can get the true owner-occupier demand moving there, which just isn't really, you know, as strong, you know, particularly because they're just hard to, you know, there's a lot of second homes in a lot of these markets.

Yeah, so there's a movement towards company structures, and really an Airbnb is a little business in a way that a landlord isn't. So I would imagine it'll be more company-owned, company-run for Airbnb markets over time.

Yeah, because they'll pay lower tax.

That's what you're saying, like lower income tax rather than 30%. Is that what you mean?

Yeah, yeah, so being able to offset non-essential repairs and things. You can do it in a company that you won't be able to do, as I understand it, under the new system, which will also be bad for – not good news for renters.

So it's going to be harder to persuade the landlord to make non-essential repairs because if he's negatively geared, he won't be able to offset that. That's definitely true, yeah.

Yeah, won't be able to afford to. Yeah, that's actually really interesting. Yeah.

Whether it's really going to help labor, I don't know. If people are aspirational enough to want a home but not aspirational enough to earn money off assets, so if they're at just that level, it's like, yeah, one day I will be able to, but life as a tenant is going to get worse.

Yeah. It's also the pressure to increase your rents is higher because you're like, well, I'm not going to get, I've got to get every dollar I can back on this property, you know, because it goes straight into my pocket. Yeah. I can't just offset against my income.

So I think that's going to be a real pressure as well, because also your property looks more attractive if your yield's a bit higher as well. It doesn't really, you know, particularly... So I think everyone will be trying to increase their rents as much to sell. So...

They'll be, you know, if they're thinking about selling in recent, you know, coming years, they'll really care about what they're renting it for. And so I think that'll be everyone will be pumping.

Yeah. Yield is awesome. There's people going around saying yield is bad for capital growth. It's never shown up in my data unless you're at the very top.

Is it bad for capital growth or is it mutually exclusive? Because like, for example, in Sydney, we've been sitting on 3% yield for years. It's like 3%. It's pretty shit, right?

Forget this most recent period, but we've had pretty stable capital growth over the last couple of years and before that with ups and downs. But generally speaking, we get pretty stable capital growth. But yield has been garbage. So are you saying there's no relation at all?

No, there is a relationship. Unless it's at the real extreme end of yield, like mining towns and sudden massive need for worker supply or something, yield positively predicts capital growth. I've always found it. Yield growth, rental growth is a great one. And it's obvious why.

So obviously, you can get more returns on the investment. So it attracts investors in. But also, if rents are going up, then it's getting more attractive in general. So, yeah, so it's actually a reverse N shape. So the very bottom yield has good capital growth for some reason.

I need to look further into that. But, yeah, so it's... But for most of the market, it's actually positively correlated. Yield's a good thing.

Can I ask some more questions about this, though? Because, like... Rents can't move as fast as prices in many cases, right? And it's even been slowed down further because you've got legislation, at least along the S&C board, which means you can't put the rent up more than once in a 12-month period.

And so you can't have a rapid increase of rents to follow. And if price growth is really fast, then yield is going to fall because you can't put rents up as fast as you can put prices up. Yes, yes.

So you'll have... If your yields are good, then the price, you're right, the elastic part is the sale price. So that will go up to level it out.

So you have a mean reversion kind of thing happening where it then stabilizes on a, you get growth and then it stabilizes back to a more standard yield. Yeah, so the market is correcting so that you end up with more similar returns over time.

People are going, well, there's better yield to be had there, so I'll buy that. So that increases the demand for those properties.

but it's short-lived that yield though.

Well, I'm just focused on that. Our primary focus at MicroBurbs is always where to buy at the beginning. So I'm not talking about cashflow ongoing.

So yes, by definition, if you have capital growth, all other things being equal and you're right, rents don't move so easily as quickly as the sale price, then you actually are going to get less yield.

So, yeah, it's not like you can maintain that yield as you experience the capital growth.

No, yeah.

Yeah. But you get the capital growth.

Have you got any probably Dumbo, Luke, to finish this off?

Oh, good. Yeah, 15% of our predictions get it wrong. How about that for an empirical one? So yeah, one in seven times our forecasting model gets it wrong. We're always working on reducing that number. So a seventh of me is a dumbo. How's that?

Well, I love that you're honest about that because there's a lot of certainty and hubris that comes into the property market and a lot of people will not entertain the idea they could be wrong.

So I think that's fabulous and that, in my view, doesn't make you a dumbo, but maybe trying to predict the property market does. But if you get it right, what would you say, six out of seven times? Over what period of time are you getting it right, though?

That's what I want to know. So over the next four months, two years? Four years. Four years.

Yeah. So yeah, the further you try and forecast into the future, the harder it is, like the more things equal out. There are certain things that last for a long time, like tranquility, being coastal works really well.

So there are certain geographic drivers that we're going to be talking about more that are obviously things that are permanent. But, yeah, so fundamentally the model is about mispricing. So it's just finding things that aren't priced correctly. An example of that is yield, right?

So that's how the model can find what's true. And AI models, machine learning is better at trying to work out what the price should be rather than humans, which are much more like, follow the herd or that herd is headed for a cliff, more that mentality.

So it's like value-based investing. You're sort of looking in the stock market. You're sort of looking for areas that are undervalued.

Yeah, it's mispricing. That model specifically is not like a fundamentals model. It's very much looking at relative prices. So another example, the model will do things like there'll be a small country town and it will be like –

not performing for ages, the region is going really well, and then bang. So that's the sort of time, and then you suddenly get this doubling, right?

And those are the sort of things that the top 10% of investors are doing, looking in places that other people wouldn't look, buying in hamlets that would never go onto the radar of a spruiker or a buyer's agent, because there's just no stock to, there's only one house to buy there. Yeah.

Luke, it's always fun chatting to you. Your brain works in a particular way that is just fascinating and I love it. And we really appreciate you coming along and sharing what you've been discovering. And it makes me want to go back and have another look at what microberbs is.

It's a while since I've been on your website. So it sounds like there's lots of new and interesting stuff in there. Thank you.

Lots of papers. So we put out lots of white papers these days. It's all about the transparency. So you can actually see like when I make these claims, you can actually go on the site and say, yeah, but what about Melbourne?

Like that point you made before, like we'll actually disclose, you know, to what degree did it work for different capital cities? How did it work over time? And we're always looking for things that are consistent over geography and time. Love it. Thanks so much, Luke. Great. Thank you very much, guys.

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.