The Zestimate Part II

zestimate_noAs a realtor in Southern New Hampshire I get a lot of questions from buyers and sellers about the Zillow.com “Zestimate”. The most common questions include: How accurate is the Zestimate? Why is the Zestimate lower than the seller’s asking price? Why is the Zestimate higher than the seller’s asking price? Why is the Zestimate so intent on preventing me from selling my home?

The simplest answer to those questions is that the Zestimate is flawed. It is only accurate, or close to accurate, when the comparison properties it uses to price a particular home are 1) good comparisons, 2) recently sold comparisons, and 3) comparisons with a reasonably accurate Zestimate. You see, all too often the Zestimate of a home is based on already existing Zestimates of other homes that it has determined are good comparisons. For the system to work, those Zestimates have to be right. It is a problem when the comparisons used to estimate a home’s value are based on other estimated values that do not really reflect true activity in a housing market.

Last May I wrote about some of these issues and said I was going to track some Zestimates for homes on the market to determine how closely they mirrored the home’s sale price later on. As an initial example, I used one of my own listings to show what most realtors already know: the Zestimate is more of a nuisance than a helpful resource. Zillow’s automated home value estimate is not exactly the gold standard for determining a property’s worth. Moreover, these estimated values have the potential to impact a buyer’s perception of a home’s market value.

Why is this a problem? Well, if the Zestimate is lower than the asking price then a buyer might interpret it as confirmation the seller/s (and/or the seller’s real estate agent) priced the property too high. That’s not outside of the realm of possibility, right? Sometimes sellers and realtors do over price homes, and it is not always done intentionally either. Depending on the market and variables influencing it at the time, a price determined by a thorough market analysis may or may not reflect what a buyer is willing to pay for the home.

The Zestimate being low is an understandable problem, but what happens when it is too high? If the Zestimate is much higher than the asking price it may prompt buyers to ask what is so wrong with this house that the seller has discounted it so significantly. Realistically, the home may be priced appropriately when compared to recently sold comparison properties, but how would a buyer know this without being told? How would they know the Zestimate is potentially false without seeing evidence showing this? The sad fact is they wouldn’t. It is troubling for buyers when they are faced with a Zestimate that varies considerably from the asking price – especially when there is no clear explanation as to why.

When I wrote about this topic before, I used 1 Grove Court in Litchfield, NH as an example. I selected it mainly because of how large the gap was between the asking price at the time and the Zestimate. At the time my original post on this topic was written, the asking price was $460,000. The Zestimate, however, was $387,805. That is a big difference.

The above Zestimate was based on two specific properties: 47 Garden Drive and 2 Garden Drive. I described these properties in detail previously to demonstrate the reasons why they were not good comparisons. One glaring problem with these two comparisons is that when they were used for comparison purposes by Zillow they were still on the market. Many real estate agents use active and pending listings to evaluate a home’s worth, but it is important realtors use recently sold comparisons in a market analysis to show what buyers have paid for similar homes in the market. If a comparison property is still active it is helpful in terms of indicating what buyers probably won’t pay for the home, but it does not show how much lower the seller would need to price their home to sell it.

Getting back to the property I used as an example, 1 Grove ultimately sold for $455,000. According to the buyers’ lender, the home appraised at that value. The Zestimate was wrong and not just by a little. It was off by $67,195.

Even though the home sold for much higher, the Zestimate can’t or won’t admit when it is wrong. It does not seem to be coded in a way that takes real market activity pertaining to the property into account. I say “seem” because perhaps it does and it just does not attach enough weight to make a difference in the Zestimate after the property sale is complete. For instance, as I am writing this Zillow reports that 1 Grove is worth less than the final selling price of $455,000. However, I should point out that instead of the original $387,805, the Zestimate has increased ever so slightly and currently values the house at $389,889. That could change though. Tomorrow it might suggest the value is more or less. The bottom line is that it’s wrong today. 

Putting aside the home on Grove Court, what about the properties Zillow used as comparisons for that home when it was on the market in May? What became of those properties? Did they sell for as much, or even near, the same price as 1 Grove?

Let’s take a look at that.

47 Garden was the first comparison included in Zillow’s estimate. When I wrote last about the Zestimate this home was being sold by the owner. The owner eventually got a realtor and then the home sold for $330,000. The current Zestimate values the home at $347,906 (as of the time of this writing). If accurate, this home has increased in value substantially in just a matter of months. Such a massive improvement does not seem likely, does it?

Then we have the other comparison Zillow used for Grove Court: 2 Garden. This property sold shortly after my original post about the Zestimate. The actual closing price in June of 2016 was $290,533. Right now the Zestimate considers the value to be $337,500. Does it matter though if buyers will not really pay that much for the home? Apparently not. 

Since it does not seem as though the Zestimate is going anywhere soon, realtors and sellers have to find ways to overcome the estimated values that are not in line with the market. One way of addressing the issue is to have accurate comparison sales ready to show buyers and/or their agents why the selling price is different than the Zestimate. It also helps to have currently active and pending comparisons to illustrate what is happening in the market at a particular moment in time. Another means of breaking down this barrier is for sellers to claim the property through Zillow as their own and update the description and features. This action comes with a warning though: it may or may not help to bring the Zestimate in line with the price the owner is asking. It may increase or decrease the value, depending on how accurate the description of the property was before the changes. And that value is still based on other Zestimates.

Ultimately we are all at the mercy of Zillow and it’s seemingly arbitrary selection of properties it deems as comparable, and so it helps to understand how the Zestimate works, and doesn’t work.

The Zestimate Project: An Introduction

A bane of a realtor’s existence is the Zillow.com “Zestimate”. The Zestimate is the website’s estimated market value of a property – whether it is for sale, sold in the past, or off the market. When a home is offered for sale on Zillow the website automatically provides an estimated value of the home and places it near top of the page, just below the asking price of the property for viewers to see. It appears as follows:

1 Grove

In the example above, the Zestimate values the property on 1 Grove Court at $387,805. The house is a 4 bedroom colonial, with 2.5 baths that sits on just over an acre of land. The value Zillow has provided is considerably lower than what the sellers are asking for the home. $72,195 lower to be exact. Why though? Is the home overpriced? Is Zillow using poor sales comparisons in its estimation? What gives? And more importantly, is the Zestimate in this case, and others, even accurate? 

The above questions, combined with endless conversations and debates over this seemingly benign valuation tool have prompted me to examine this matter further and decide – at least for myself – once and for all whether this tool is helpful to buyers and sellers, or harmful. Thus this post begins what I am now calling The Zestimate Project. My project will consist of research into past and present sales in Southern New Hampshire to track the accuracy of the Zestimate, shed light on the strengths and weaknesses of the tool, and propose recommendations for improving it if applicable.

Going back to the first home example above on Grove Court, I scrolled down to the bottom of the listing page to determine what comparisons (comps as we call them in realtor speak) were used to calculate the estimate. The homes used are as follows according to the tiny print at the bottom of the page:

47 Garden Drive in Litchfield

2 Garden Drive in Litchfield

Let’s discuss these two properties for a moment, from a valuation perspective. 47 Garden Grove was last listed on the MLS (multiple listing service) in 2014 for $329,900. The listing did not sell and subsequently expired. The home is presently offered for sale on Zillow by the owner. It still has not sold and therefore is not a valid comp. However, it is also not a valid comp for a couple of other reasons. A look at the expired listing from 2014 shows this home was built in 1983. The one on Grove Court was built in 2002. 27 Garden Drive has a 2 car garage and is a cape style home. 1 Grove Court has a 3 car garage and is a colonial. A look through the photographs on Zillow for both properties demonstrate the many differences between the interior and exterior aspects of the homes. They are not reasonable comps and 27 Garden should not even be used as such by Zillow because it has not even sold.

The next property Zillow listed at the bottom of the page, and claimed was comparable, is 2 Garden Drive. Zillow states, “Another comparable condo, 2 Garden Dr recently sold for $290,533 on 06/05/2015”. None of these homes are “condos”, but other than scratch my head over that description I don’t know what to do with that information and have no knowledge of where Zillow came up with that. The homes discussed herein are single family residences. Not condos. 

In reference to 2 Garden Drive, a home that sold almost exactly one year ago is not a good comparison property. Second, the home itself is not comparable for other reasons. 2 Garden Drive was built in 1973, instead of 2002. It states on the MLS that total finished square feet is 2,366. The total finished square feet for 1 Grove Court is 3,021. The list goes on.

The difference of 655 in square feet is important to note. I will preface my next statements with an acknowledgement that I am a realtor, not an appraiser. I am well versed in providing valuations for sellers in the form of a comparative market analysis, but a true appraisal is exceptionally detailed and thorough. That said, square footage is an important aspect of a home’s value. The difference of 655 square feet might seem minor, but if an appraiser were determining the value of either property he or she would choose recently sold properties that were as comparable as possible in terms of size, age, condition, square feet, etc. and then make adjustments to the value of each comparison property to bring it in line with the subject property. In simpler terms, if the comparison property for example has 200 more in gross living area (square footage) then the property being appraised, the appraiser would deduct an amount of money from the comparison property based on a determination of what each square foot is worth. I have reviewed a wide range of appraisals completed on properties I have sold and found there is some variation in terms of the value attached to square feet that depends on many factors, including locality. An average, based entirely on what I have seen in local appraisal reports I have viewed is about $25 per square foot that is above grade and finished. Sometimes higher and sometimes lower. If we use this average to determine the difference in value between 1 Grove and 2 Garden, merely going by the square footage difference, then we get $16,375. That’s no small chunk of change.

The above does not take into account the age difference between the properties, 2 Garden Drive having one less garage bay, the differences in exterior and interior updates, and so much more.

The Zestimate in this case only refers to the above two homes as comparison properties. If it is using others it does not state as much, but realistically the website should disclose all homes used for comparison purposes to allow readers to make decisions about the Zestimate’s reliability in any given situation. The comparisons used for Grove Court consist of a home that sold a year ago and one still on the market. But now let’s look at recently sold comparison properties on the MLS to see how far off the Zestimate is from actual comparative properties. I will use three properties for this purpose.

Property 1: On May 5, 2016, 9 Riverview Circle sold for $410,000. The home has 3,116 square feet, 4 bedrooms, 2.5 bathrooms, and a 3 car garage. However, this home was built in 1986 instead of 2002 and lacks the multitude of updates found in 1 Grove Court. That said, 9 Riverview is a lovely home, with plenty to offer, and is a closer comparison than the ones chosen by Zillow. But here’s the kicker. The current Zestimate for 9 Riverview Circle is $494,876, even though the home sold for $410,000 this month. If the Zestimate is truly accurate then the buyers made out well and the sellers…not so much.

Property 2: On April 1st, 2016, 21 Moose Hollow Road sold for $455,000. It has 3,884 square feet according to the MLS, 4 bedrooms, 2.5 bathrooms, and sits on just over an acre of land. It has one less garage bay than 1 Grove Court, but is updated throughout the house and was built in 2005. The home does not appear to have as many updates as 1 Grove, and is lacking some of the structural architecture that makes 1 Grove unique and aesthetically appealing, but 21 Moose Hollow is a fabulous house and would serve as a good comparative when valuing a home like 1 Grove Court. The current Zestimate is $470,283 despite the fact the property sold almost two months ago for $455,000.

Property 3: On December 11th, 2015, 15 Moose Hollow Road sold for $457,500. The home is comparable in that it is extremely similar in square feet, has 4 bedrooms, 2.5 bathrooms, is a colonial style home, and sits on over an acre of land. This property only has a 2 car garage and is lacking some of the cosmetic and structural features that make Grove Court an appealing property, but nevertheless 15 Moose Hollow is nothing short of a gorgeous home that appears well kept and cared for in the photographs. The current Zestimate for this home is $464,659, even though it sold for $457,500.

In summary, the three recently sold properties I found on the MLS are more in line with the price point of 1 Grove Court when one takes into account size, land, updates, special features, number of garage bays, location, and other factors than those used in the Zestimate (or at least the two the website suggests it used). It will be intriguing to see the final sales price for 1 Grove Court in light of all this information.

Over the course of the next few months I will periodically track the Zestimates of properties in the Southern New Hampshire region, including Litchfield, Merrimack, Amherst, Bedford, Nashua, and Manchester (perhaps other towns, but right now I am limiting it to certain markets) to make estimations about its accuracy within those markets. I would also like to conduct a survey at some point of buyers and sellers to find out how much faith they place in Zillow’s Zestimate and learn the ways in which the Zestimate impacts people’s perceived value of the properties it valuates. Because maybe, just maybe, the Zestimate could use some improvement in areas of consistency and accuracy.

Time will tell!