Property Hub - Investment Insights & Inspiration - Realty Talk Vault: Misleading Medians

Episode Date: May 10, 2023

We bring you another classic Realty Talk episode from the vault, which was originally published on October 29, 2022, but is still just as relevant today. Are median prices an appropriate metric to make informed property decisions? Kent Lardner from Suburbtrends discusses this and opens our eyes to more useful measures. RealtyTalk is part of the Property Hub podcast channel, your home for property investment insights, inspiration, and stories from Australia’s top property experts, investors, leaders, and analysts. Subscribe now to get every RealtyTalk episode delivered to you each week for free, and also get full access to Get Invested, the leading podcast for Australians who want to unlock their full ‘self, health, and wealth’ potential and get inspired by the stories of investors, founders, and entrepreneurs. Subscribe to RealtyTalk on the Property Hub channel: Apple Podcasts | Spotify | Google Podcasts | Email Property Hub is a collaboration between Bushy Martin from KnowHow Property, Kevin Turner from Realty, Andrew Montesi from Apiro Marketing and Apiro Media, and Australia’s largest independent podcast network DM Media.  Business and partnership enquiries: [e-mail] omnystudio.com/listener for privacy information.

Hi, and welcome. Now, for some time now, we've been opening your eyes to the pitfalls of a mainstream media's fear-driven obsession with national and state-based property markets that actually don't exist, along with a reliance on aggregated median property prices that are at best misleading and at worst downright dangerous because as we know the devil's always in the local detail when it comes to property condition trends. So where do you turn and what property information do you actually need to access in order to make better informed property decisions based on more relevant and more useful data? Well to discuss this we're joined by one of Australia's best and most respected property data analysts Kent Lardner, the founder

of Suburb Trends, who has recently released a report on the hardest five markets to measure house prices right now. So welcome back to the show, Kent. Thank you, Bushy. Thanks for the invite. Mate, this is a great subject that's close to both of our hearts. So we're pretty keen to dive into this one. Let's kick off with your opinion on the usefulness or otherwise of the most commonly used suburb medium to measure property price trends. Yeah, one of the biggest problems I've always found is the use and abuse of a suburb median. One of the, probably what really kicked this off were, there were a number of suburbs that were on these top 10 lists, the top growth, so we've all seen them.

And a couple of them were areas where they were old farms and old shacks and whatnot on main roads, and it was a redeveloped site. And it was Fern Bay, just north of Newcastle. and it made the top of the list for a long time and it was because it came off a base that were you know a number of properties purchased for 200 or 300k and then were selling for 500k and there's a number of these happening around the country so that was the catalyst for me so what's the problem the problem is we don't really have a true measure of what it was a year ago or even a month ago and what it is now and we've got this problem of compositional bias now i'll start at a suburb level the problem we've got at a suburb level is uh the assumption is it

needs to be normally distributed it needs to have most of its sales in the middle and you know tailing off either side the old bell curve the old bell curve and the problem you've got and it's most perfectly illustrated along the beaches so if you go to a beach-sized suburb you've got one market that sits along the beach and then you've got a market that sits walking distance to the beach and then the other and a lot of these suburbs that are long suburbs i.e they don't stretch along the coastline but they've got that little patch that sits near the beach and then stretch back one or two or three kilometers backwards it's pretty easy for you to imagine that there are three distinct markets when you look at the price distribution of these

suburbs you can clearly see market one market two market three in three different distributions so when you're measuring the median there it only takes a little bit of a shift in what's listed for sale at the bottom end or a little bit of a shift on what's listed for sale at the top end to have a dramatic impact on the median so that's that thing i call the compositional bias the problem with that compositional bias is you can jump at shadows yeah spot on and if you go through a period where there's very few sales then again you can get distortions that are coming out of that when you're trying to join the dots and draw lines across a period of time exactly and and there are measurement systems that try and control some of those variable things you may have heard of

hedonic indices etc you can control for things such as a bedroom you know pulling in comparable sales and adjusting for how big a house is and whatnot you can apply a coefficient to adjust for a bedroom count and effectively normalise the whole lot. But you can't control for the fact that there's going to be listings that come in that are at the top end of town and then they're not there again. So you can't control for everything. So compositional bias exists no matter what you try, it exists,

and you need to just call it out. I think the biggest thing is to kind of appreciate it exists and then look for ways to remedy it without only relying on a suburb median. And what I've attempted to do is to find the Goldilocks, which is what I call, what the ABS call a statistical area three, SA3. So the Australian Bureau of Statistics created its own set of geographies. The smallest one's around 200 homes, and they call that an SA1. And like a Lego, that plugs into an SA2, which is about, give or take, about three suburbs big.

And then the SA2s plug into what's called an SA3. And across the country, give or take, there's about 350 of those. So I use that as my measure because we used to all use LGAs, local government areas, in the day. But the problem is Brisbane turned into the blob and aggregated and kept on going. So it became the biggest, you know, LGA in history. Some of those LGAs are as big as states.

It's a monster, right? And then the Gold Coast did the same thing. So it almost became irrelevant to measure at an LGA level because of Brisbane. It's too big. And you know, anyone who knows Brisbane, it's a lot of markets. Whereas the SA3 carves it up in, I believe, a fairly perfect size. So I use SA3s for my median.

In most cases, they're normally distributed. But not always, not always. And there are problems there, which is the foundation of that report. Yeah, okay. Well, let's sort of drill into some of the other approaches that can be adopted to measure house price trends. and talk about some of the pros and cons of them. Yes, probably the one that stands out,

and I tried building this, or we did build it back in the day, 15, 20 years ago, called a repeat sales index or the Case-Shillant methodology. And this is widely used in the United States. What it does is it takes a property and then tracks its history through time. And so what you need for it to enter the data set

or the sample to be used and measured is two sales. sale one sale two and time in between so what that tells you is property a grew by a certain size over a certain amount of time once you put them all into a bucket you can slice it and dice it come up with an average growth rate by a given geography yeah so in principle that sounds great because it does control for a lot of that compositional bias you're measuring the same thing exactly the same thing through time in theory there are a couple of gotchas and a couple of reasons why it hasn't worked that well in australia our whole period's been getting longer and longer and longer good point so and our sample sizes get quite small in certain geographies so

in america it seems to work quite well for rentals it works quite well in australia because rentals come back on every three or four years but for sales it hasn't worked that well yes it can work well for units because units have a lower hold period right but by and large we don't have this type of measurement problem when it comes to units. Most of this measurement problem pertains to houses. Good point. Good point. Any other approaches then in the pros and cons? Well, the other approach is the hedonic index or hedonic approach. Now, I won't mention any particular brand or any company, but the approach there is similar to the way an AVM works. Effectively, if you pick the same sample of properties is one method and you use an AVM

to value that same sample of properties through time, and then you measure that median. That controls for the fact that there might be sales that come in that are bigger or smaller than that house that you're valuing with an automated valuation model, and that is automatically adjusted for. So that's that hedonic adjustment, and then if you control that sample through time and you hold the same properties through time, you get a little bit of the best of both worlds.

You get the best of the hedonic, you get the best of the repeat sales, but none of these are perfect. Yeah, okay. So picking amongst those, which do you believe is the best and most useful method to apply? Look, I like the simple SA3 medium. The reason why I like to use it is because it's easy to explain, so there's no smoke and mirrors.

And also you can capture and use agent-advised sales, so you get an early reading on the market. So with a lot of models that rely very much on data that's processed through the state governments, you've got a significant lag there. So typically what you'll find, the law of large numbers traditionally means that as you collect your sales, you land on the median pretty quickly and it shouldn't vary from that. So if it's normally distributed, you don't need 100% of sales

to land on the median. You need a statistically relevant sample, and in most cases, agent advised sales do represent more than 50% of total sales. So agents will push their sales out there quickly. You can collect that and measure that at an SA3 level and arrive at the median very, very quickly. Yeah, I like it. I like it. Well, let's have a look at the report that you've recently released. What are some of the hardest housing markets to measure this year and why, Ken? Yeah, so this is a snapshot in time. So I did this report as of listings in August. So what the call out here too is that this volatility varies month to month so what was relevant and

what is relevant for august is not always going to be guaranteed for the future months so as of that particular time here's a couple of call outs um i'll list down the areas and i'll drill into a couple of them specifically so got brisbane inner city broad beach burley so in queensland eastern suburb south which is sydney port phillip down in melbourne and the southern highlands so you know in you know the the nice area of the southern highlands uh in in probably i don't know about an hour and a half uh west of sydney got it maybe two hours um so the one i wanted to focus on specifically broad beach burley this is really interesting what i tried to do is measure i come up with a measurement system to tell me if that compositional bias existed

or exists at the SA3 level. And the best I could come up with was to say, here is the SA3 median. Let's split the suburbs into above the median and below the median and then count their listings. And therefore, if I see a significant shift above or below or into those suburbs that are above the median or below the median, that will give me a reasonable proxy

for what type of bias might be in the measurement in the months to come. So here's Broadbeach, Burley, Mermaid Beach and Broadbeach waters were above the SA3 median, and they had a reduction in listings against their, you know, so effectively the six-month average. They had a reduction of 10, but the suburbs of Mermaid Waters, Miami, Burley Waters and Burley Heads, they had an increase of 19 listings. so it's quite a significant shift there so quite an imbalance between the above and the below so i'd expect that once those listings end up selling there's going to be a compositional bias down to give me some level of artificial decrease in that price for that particular market yeah it's interesting and the sort of uh buyer and or seller intentions behind the properties in terms of their

values you might you might get those at a higher price point who are thinking well we'll just we'll just sit tight a whole minute and then others at the lower that are that are more active is is that having an influence do you think there's some i have to make some assumptions as to that because you know the idea would be that i could get on the phone and call the agents that was what i'd love to do i just you know don't have that time to do it but that would be the ideal because i can describe the what and what's going on but the why ultimately is really the the the the discussions with the buyers and the sellers. So, yeah, I have an assumption.

I've always held the assumption that the blue-chip markets especially, people will hold. They don't have to sell. They're smart people. They're in a multimillion-dollar property for a reason. No, it's a very good call, mate. For those that have, you know, you've really captured their interest, where can they get a copy of the Hardest Five Markets to Measure

House Recipes Report? Yeah, it's a free download on suburbtrends.com. Just go to the homepage and scroll down. Yeah, awesome, mate. Look, I really want to thank you again for these very timely insights, Ken, and thanks again for joining us on the show today. Thank you, Bushy.

Awesome, Ken. Well, here is yet more evidence that trying to read the tea leaves of property trends and basing your property decisions on non-existent property markets that revolve around suburb median price movements is likely to be close to meaningless and misleading. So if you want to tap into much more relevant

and useful data, take advantage of Kent's complete suite of property reports that you can access now at suburbtrends.com. Stay with us for more here on your Place for All Things Property Realty Talk.