Search
  • Ben Piper

How Accurate are Agents’ Estimated Prices ?

Warning: This section is only of relevance to Melbournites. I eventually hope to write a similar article for Sydneyites.

Survey overview


In my book I thought it would be useful to make a comment on how reliable estate agent price estimates were now that the 2016 Victorian laws had been in force for a while. I therefore resolved to do a short survey. Unfortunately the results of that survey were fairly ambivalent, so I decided to try a more prolonged survey. Initially I was just going to do a survey over 7 weeks. However, the results of that survey were so unusual that I thought I had better do a quick follow-up to see whether those survey results were an aberration. They were not. In fact the results were even worse.


I have now combined the results of these surveys into a single survey. The results are definitely not ambivalent - in fact they are eye-openingly astonishing.


Over the 12 weeks of the survey, agents underestimated the prices at which houses would be sold at auction 82% of the time !


And the average amount of the underestimate was 11.6% !


This is not good !


Survey - 1st period


For the 1st period of the survey, I looked at properties featured or listed in the Auctions list in the “Open this Weekend” section of the Domain lift out section of The Saturday Age that were auctioned on one of the 7 Saturdays between 22 June 2019 and 3 August 2019. Further, I only looked at houses that had a price estimate shown; that sold on the day of the auction; and that had a price reported to Domain. In the relevant period, 30 properties met these criteria, and there were also 8 properties passed in with interesting results. The results are as follows:

0 of the houses were sold for a price lower than the bottom of the price range
7 of the houses were sold for a price in the price range (4 of the houses were sold at about the middle of the price range, 3 near the top, and none at the bottom)
23 of the houses were sold for a price higher than the top of the price range!!!
(I note that the last category includes the curious case of one apartment with an estimated price range of $500,000 to $550,000 that was sold after the auction for $575,000!)

With respect to the houses that sold for a price higher than the top of the price range:

7 of the houses were sold for a price within 5% of the top of the range
8 of the houses were sold for a price between 5.1% and 10% higher than the top of the range
5 of the houses were sold for a price between 10.1% and 20% higher than the top of the range
3 of the houses were sold for a price more than 20.1% higher than the top of the range. (23.5%, 31.1% and 36.5%).

Of further interest were some of the properties that were passed in during the period I looked at.

5 of the houses were passed in even though bidding reached the lowest figure in the price range
3 of the houses were passed in even though bidding was above the lowest figure in the price range (1.6%, 1.7% and 6% above that figure).

I note that the properties whose results I have reported included a pretty good mix of houses and apartments, price ranges, suburbs and agents.


Analysis


Clearly the results of this survey are concerning. It is not a good look if the vast majority of agent price estimates are underestimates, and it is clear that the bottom of the price estimate ranges are not amounts at which houses can be bought.


Of particular concern is the fact that if you look at the percentages by which the top of the estimated price ranges have been exceeded, they seem to cluster near the 10% mark. (The average of the percentages, which of course is, strictly speaking, a mathematical no-no, is 10.6% . That seems very much like 2016 figures in that there is a strong suggestion that the estimated price ranges are a deliberate underestimate.


By the way, given that there are not a great many results overall, I thought it might be useful to also mention that if the 3 outlier results (that is, the estimates over 20% wrong) are removed, the average over-estimate is then 7.6%, which is still a pretty high figure.


Survey - 2nd period


The results I just gave were disturbing. So, as mentioned, I thought it was worth doing a follow-up survey. Using the same methodology, I conducted a further 5 week period survey.


I looked at properties listed in the Auctions list in the “Open this Weekend” section, or that were featured, in the Domain lift out section of The Saturday Age that were auctioned on one of the 5 Saturdays between 26 October 2019 and 23 November 2019. Again I only looked at houses that had a price estimate shown; that sold on the day of the auction; and that had a price reported to Domain. I note that one of the Saturdays in this period was on the Melbourne Cup weekend, and that there were very few auctions held that weekend.


This time, 26 properties met the criteria listed above, and there were also 3 properties passed in with interesting results, and 5 other results of interest. The results are as follows:

0 of the houses were sold for a price lower than the bottom of the price range
3 of the houses were sold for a price in the price range (2 of the houses were sold at about the middle of the price range, 1 near the top)
23 of the houses were sold for a price higher than the top of the price range!!!

With respect to the houses that sold for a price higher than the top of the price range:

3 of the houses were sold for a price within 5% of the top of the range
8 of the houses were sold for a price between 5.1% and 10% higher than the top of the range
9 of the houses were sold for a price between 10.1% and 20% higher than the top of the range
3 of the houses were sold for a price more than 20.1% higher than the top of the range. (21.7%, 30.5% and 31.6%).

Of further interest were some of the properties that were passed in during the period I looked at.

3 of the houses were passed in even though bidding reached the lowest figure in the price range
1 of the houses were passed in even though bidding was above the lowest figure in the price range (2.1% above that figure).

Also:

2 houses were sold after being passed in, but for prices that were higher than the top of the price range (5.9% and 18.4% higher)
3 of the houses were sold before auction at prices 4%, 6% and 10.9% higher than the top of the price range

Again, the properties whose results I have reported included a pretty good mix of houses and apartments, price ranges, suburbs and agents, although over time one notices that it is only a fairly small group of agents (15 – 20) who advertise or feature in Domain.


In the previous survey period, I noted that the average percentage by which actual prices exceeded the top of the estimated price ranges was 10.6%. In this survey period, the equivalent figure was 12.5%. If one combines the results of the 2 periods, actual prices exceeded the top of the estimated price ranges by an average of 11.6%.


Again, in the interests of fairness, if one takes out the 3 outlier results of the 2nd survey period, the overall average underquote falls to 10.2%, which is still pretty big. The overall average underquote over the 2 survey periods without the 6 outlier results is 8.9%.


If one combines the results of the 2 periods, 46 out of 56 properties sold for a price above the top of the estimated price range – that’s 82%!!


I note that auction clearance rates during the 2 periods of the survey were 10-15% above those of the clearance rates that applied up to about May 2019.


There are, however, 2 big differences between the survey periods. During the 1st period, there were each week 2 – 3 times as many auctions each weekend as there were during the 2nd period. Of perhaps greater significance, the median sale price of houses during the 2nd period was roughly 15% higher than the median sale price of houses during the 1st period, although from week to week there was sometimes considerable volatility in movements of the median price.


Analysis


Taken over the 2 periods, my survey provides hard evidence the 2016 auction laws are not having the effect that they were supposed to have.


In fairness to agents, it is quite foreseeable that in a time of rising prices it might be very difficult to accurately forecast prices. While agents would quickly be aware that prices are going up once they start going up, the difficulty is in knowing whether in 5 or 6 weeks time prices will still be rising, and also whether the prices in a particular area are part of a general trend, or will move to their own drummer. And the outlier results are a timely reminder that strange things can happen in individual cases.


However, that said, one would hope, even in times when house prices are dynamic and volatile, that over time an increasing proportion of agents would be able to improve the accuracy of their predictions. At present there does not seem to be any sign of that happening.


While, if I was a lawmaker at present, I wouldn’t be reaching for the panic button, by the same token I would be concerned. Ideally a law trying to prevent underquoting should not be rendered meaningless by market conditions that are quite foreseeable, and that are bound to happen from time to time.


A further concern I would have as a lawmaker is that there is no easy way to improve things. In my view, based on my experience with earlier laws, the problem is not with the laws themselves. Rather, the problem is with the enforcement of the laws. That creates 2 problems: there is now not much room to move in terms of changing the laws - the laws are pretty clear and have no obvious defects, again, in my opinion. That being so, better enforcement action is the only way forward, but where estate agents have effectively been allowed to flout the laws pretty much across the board for a period of time, it is difficult to take enforcement action in a way that does not seem to be unfair ("I was only doing what everyone else did.")


Regardless, the message from my survey for prospective house buyers is loud and clear: you need to work out the likely market price of a house you are interested in by yourself. Estimated price ranges by agents only seem to be useful at present for establishing a “not below” price.


Finally, the obvious question arises: how representative is my survey sample of the general situation in Melbourne.


Representativeness


As to representativeness, I note that the results reported in my sample involve no element of subjectiveness on my part. If the qualification criteria I listed above were met, the result is included in my sample.


Therefore, if there is any lack of representativeness in my sample, it must be found in the nature of what finds its way into Domain with a price estimate. In that respect, I note that for a property to be featured in Domain, it has to be shown with a price estimate. That is not the case with the properties listed in the Auctions list of the “Open this Weekend” section of Domain (those are properties that are to be auctioned on the day the list is published). The majority of properties in that list do not show estimated prices. Instead, the price is shown as “POA” or “NPD”. Neither of these abbreviations is explained. “POA” is presumably “Price On Application”. “NPD” is presumably “No Price Disclosed”.


I should also note that the properties listed in the Auctions list represent only a very small number of the properties that are auctioned on the relevant day. To take a fairly random example: in the Domain published on 29 June 2019, 21 properties appeared in the Auctions list (and of those, 7 showed an estimated price range). The following day the Domain website stated that there were 463 auctions in Melbourne on 29 June 2019.


Clearly the Auctions list is a very select group. But presumably the selection is made on the basis of agents who have a relationship with Domain, and presumably the relationship is with agents who advertise with Domain. That in turn presumably depends on the advertising budget that the agents negotiate with property sellers. Now, unless one is prepared to countenance a theory that agents who negotiate higher advertising budgets tend to be agents who engage in under-quoting, there is nothing in any of that that suggests that a Domain sample is unrepresentative with respect to likely under-quoting by agents.


With respect to properties that Domain features, again presumably Domain only features the properties of agents who advertise with it (it is also possible that the “featuring” is simply another form of advertising – I do not have information on this). I also presume that if a property is the subject of a feature, the price range shown has to be the price range that the agent gave the seller (or else some serious law-breaking is going on). In other words, I am suggesting that the price ranges shown for featured properties are not “special” price ranges that the agents have determined because the property will be featured.


There is one other factor that might play a part. In only looking at houses with estimated price ranges that are quite publicly published, in a time of rising prices and high clearance rates, it is possible that if a such a house is passed in, or only realises a price in its estimated price range, there may be some incentive to try to ‘bury’ the result as much as possible by not reporting it. This incentive would not be as strong in the case of houses that have not been put in the public eye in the way that houses listed or featured in Domain have been. If this is so, it means that houses that are sold for prices within their estimated price ranges will be under-reported in my survey results. Food for thought.


In any event, my survey shows that there were, in the periods of the survey, quite a few houses that were sold significantly above their estimated price range. And, as mentioned, all of these examples occurred in a very public space. That is, they are the cases that one would least expect agents to deliberately engage in under-quoting.


In summary, I can’t see anything that suggests that my sample is significantly unrepresentative of the whole auction market. However, I am very open to the possibility that I have missed something, so if anyone has any ideas of what that I might be, I would very interested to hear about it.


And, of course, I have to finish here by repeating, for emphasis, the point that prompted this whole exercise. In my book I very strongly recommend that you not rely on agents’ price estimates when making any decision in relation to a property that you might be interested in. Clearly my survey strongly supports that recommendation.



51 views
© Ben Piper 2019
Ph: 0412 469 824      Email: buyhouseatauction@gmail.com