Tonight, I’ve been playing engaging the inner real estate geekiness along the likes of Tom Dozier’s Seattle Property News blog… I’ve been following (from a distance) Tom’s “Home Tracking” posts and not exactly sure where he was going until today. When he said “since one of the signs of the strength of a market is whether high demand is causing people to bid prices above the original amount the seller seeks,” I decided to test out his theory with a larger sample size…
So here is my method. I took all the homes that sold in my neighborhood (Loyal Heights) over the past year and calculated the “Net (Sale minus List Price) as a Percent of List Price”, or this calculation:
(Average Sale Price – Average List Price)/Average List Price- Average((Sale Price – List Price)/List Price)
The result is summarized in this chart:
(Click on the chart to see a larger version!)
The first thing I found interesting is the extent to which the “Gross Net as a Percent of List Price” varied seasonally… For two years running the % of List Price bottomed out around Feb/Mar and then quickly picked up to peak around May (with homes selling for 4% and 8% over the list price!). Note that February (the bottom) is also when the number of sales bottomed out over both of the last two years.
For reference, there were 182 homes sold between October ’03 and August ’04 in Loyal Heights with the average home selling for 1.2% over asking price. (I threw out September 03 and September ’05 data since I only downloaded data for the partial months…)
My conclusion is that Tom is wasting his time if he expect to see the health of the market by looking at whether or not homes are selling above or below the market price. As my chart demonstrates, there is way too much seasonal variation for a one or two home snapshot to be valuable. Even in the last two years where the value of homes in my neighborhood have consistently risen (quite substantially), there have been long periods (up to six months) where the average home has sold for less than the asking price. However, with that said, I want to say thanks to Tom for raising this intersting issue because you’ve given me an opportunity to learn about (and demonstrate) the huge seasonal variation in the local market!
I welcome anyones comments on my method if you have ideas on improving things (including the obvious improvement of adding more neighborhoods).
Update: I realized this morning in the shower that I was calculating the net (sale price minus list price) as a percent of list price, so I’ve corrected the text in the post (but not the chart!).
It would be useful to factor in interest rates and inventory. Were there fewer sellers or more buyers?
Also, What other items of value are masked by the selling price? Do repairs, closing costs, furniture, waived inspections, waived financing, larger earnest money, etc get rolled into/out of the sales price? Are these amounts significant? Are they common? That probably depends on who is being courted, so it may not change the dynamic.
I like the graph:)
Wow, this is really great. Thanks, Dustin. Some of my thoughts and questions:
1) Where are you getting the underlying data for average sale price vs. average list price? Is this something that’s publicly available somewhere?
2) If this ends up being representative of the region, does that mean buyers have more leverage in the February/March time frame? (Along the same lines as Joe’s second question: Are there fewer buyers and/or more listings in that time frame generally?)
3) The seasonal trend is really clear, but the nearly 4 percentage point difference in peaks between spring/summer ’04 and spring/summer ’05 also seems notable. It may not be statistically significant at this level, but it would be interesting see what happens if you expanded the geographic scope.
4) This is a minor thing, but it took me a minute to figure out what the percentages meant. Intuitively, as someone looking at the graph, I would almost prefer a straight division — average selling price/average list price. That way, for example, you would get a figure like 105% if it sells for more than the list price or 95% if it sells for less. But that’s just me.
Thanks again for picking up on my post. That 4 percentage point swing I mentioned makes me think you may be underestimating what you might have here. Personally, if you have access to underlying data that’s not otherwise available publicly, I’d be interesting in seeing you expand it to the regional level and maybe even make it a regular feature of your site. You could call it the “Rain City Home Price Index” or something like that. 🙂
Wow, this is really great. Thanks, Dustin. Some of my thoughts and questions:
1) Where are you getting the underlying data for average sale price vs. average list price? Is this something that’s publicly available somewhere?
2) If this ends up being representative of the region, does that mean buyers have more leverage in the February/March time frame? (Along the same lines as Joe’s second question: Are there fewer buyers and/or more listings in that time frame generally?)
3) The seasonal trend is really clear, but the nearly 4 percentage point difference in peaks between spring/summer ’04 and spring/summer ’05 also seems notable. It may not be statistically significant at this level, but it would be interesting see what happens if you expanded the geographic scope.
4) This is a minor thing, but it took me a minute to figure out what the percentages meant. Intuitively, as someone looking at the graph, I would almost prefer a straight division — average selling price/average list price. That way, for example, you would get a figure like 105% if it sells for more than the list price or 95% if it sells for less. But that’s just me.
Thanks again for picking up on my post. That 4 percentage point swing I mentioned makes me think you may be underestimating what you might have here. Personally, if you have access to underlying data that’s not otherwise available publicly, I’d be interesting in seeing you expand it to the regional level and maybe even make it a regular feature of your site. You could call it the “Rain City Home Price Index” or something like that. 🙂
Joe and Tom,
You both make some excellent points… Here’s my response to Tom’s comments (I think this will cover Joe’s questions as well):
1) The data is not publicly available (as far as I know), but rather, Anna just downloaded it for me from behind the NWMLS firewall… I’ve had trouble in the past downloading large amounts of data from this site, so I purposefully limited my study area to a pretty small region. However, last night I didn’t have any problems downloading the 182 sale points, so I could definitely push the site a little harder and download some more points.
Also, I remembered in the shower this morning that the source of my data listed two “list prices.” One was the “original list price” and the other was the “adjusted list price.” I used the “adjusted list price” because that is the most common field that is quoted, but it would probably be more interesting to use the “original list price” in order to capture the change in people’s expectations.
2) I can’t think of a way to get the “number of buyers” based on my data source, or even an approximation of that… It would be possible to get an estimate of the number of homes on the market (supply) at any given time, but the only way I can think to do that would be to put a flag for each month that determines whether or not the home was listed and then fill in the flag (1 or 0) based on the “list date” and the “sale date”. This wouldn’t capture homes that were taken off the market or re-listed, but it would give an indication of supply.
3) I agree that it would be interesting to add some more data points… I’ll look into that, but I’ve got so much on my plate right now that I don’t want to promise anything.
4) I agree… Next time, I’ll simply use the selling price divided by the list price. Not only will that be easier to follow, but it will be easier to describe!
Thanks to both of you for your comments. It is definitely a fun little index, and I look forward to diving into it some more.
Joe and Tom,
You both make some excellent points… Here’s my response to Tom’s comments (I think this will cover Joe’s questions as well):
1) The data is not publicly available (as far as I know), but rather, Anna just downloaded it for me from behind the NWMLS firewall… I’ve had trouble in the past downloading large amounts of data from this site, so I purposefully limited my study area to a pretty small region. However, last night I didn’t have any problems downloading the 182 sale points, so I could definitely push the site a little harder and download some more points.
Also, I remembered in the shower this morning that the source of my data listed two “list prices.” One was the “original list price” and the other was the “adjusted list price.” I used the “adjusted list price” because that is the most common field that is quoted, but it would probably be more interesting to use the “original list price” in order to capture the change in people’s expectations.
2) I can’t think of a way to get the “number of buyers” based on my data source, or even an approximation of that… It would be possible to get an estimate of the number of homes on the market (supply) at any given time, but the only way I can think to do that would be to put a flag for each month that determines whether or not the home was listed and then fill in the flag (1 or 0) based on the “list date” and the “sale date”. This wouldn’t capture homes that were taken off the market or re-listed, but it would give an indication of supply.
3) I agree that it would be interesting to add some more data points… I’ll look into that, but I’ve got so much on my plate right now that I don’t want to promise anything.
4) I agree… Next time, I’ll simply use the selling price divided by the list price. Not only will that be easier to follow, but it will be easier to describe!
Thanks to both of you for your comments. It is definitely a fun little index, and I look forward to diving into it some more.
This question seems heavily dependent on the ratio of buyers to available homes. 100 buyers for 1 house means that if it is priced low enough initially to inspire multiple offers…it is more likely to escalate. If demand is lower, sellers that price it higher and wait will sometimes do better.
I also agree with Tom that a wider snapshot would be interesting. However, beyond first-time buyer pricing, I suspect that important local factors are obscured by “region.” Nobody buys a house in the “region” except for relocators that are saying “Seattle versus Austin.” Applying the index to neighborhoods is a good starting point.
To calculate inventory, perhaps a sales/listings ratio? Maybe pending to listings would be better. It takes so long to actually close that it might not effectively capture market conditions when the offer was made.
#sales/#listings
#pending/#listings
Interest rates also could be useful to predict when buyers will move. Some other site must have data on this. That could explain the larger effect of IR on the amount of buyers in a market.
I also think that the swing would be even wider (higher escalation and bigger drops below asking price) if concessions (contingencies, closing costs, etc) not captured by the closing cost were considered. That information is not readily, publicly available, is it?
I’ll let it go. Those factors describe the same phenomenon that sp/lp describes.
Interesting stuff you guys.
This question seems heavily dependent on the ratio of buyers to available homes. 100 buyers for 1 house means that if it is priced low enough initially to inspire multiple offers…it is more likely to escalate. If demand is lower, sellers that price it higher and wait will sometimes do better.
I also agree with Tom that a wider snapshot would be interesting. However, beyond first-time buyer pricing, I suspect that important local factors are obscured by “region.” Nobody buys a house in the “region” except for relocators that are saying “Seattle versus Austin.” Applying the index to neighborhoods is a good starting point.
To calculate inventory, perhaps a sales/listings ratio? Maybe pending to listings would be better. It takes so long to actually close that it might not effectively capture market conditions when the offer was made.
#sales/#listings
#pending/#listings
Interest rates also could be useful to predict when buyers will move. Some other site must have data on this. That could explain the larger effect of IR on the amount of buyers in a market.
I also think that the swing would be even wider (higher escalation and bigger drops below asking price) if concessions (contingencies, closing costs, etc) not captured by the closing cost were considered. That information is not readily, publicly available, is it?
I’ll let it go. Those factors describe the same phenomenon that sp/lp describes.
Interesting stuff you guys.
This might sound off-the-wall… but I’m not sure of the question we’re trying to answer anymore…
My original analysis was really just trying to see if we could use the difference between the sale price and the list price to measure the health of the market. And I think I’ve concluded that we cannot…
Nonetheless, I’d agree that we’re going down an interesting path and we might be able to uncover some gems of information if we keep going. What questions would are you looking to answer? Are we simply looking for a better way to gauge the health of the local market? A new index? What is the index suppose to be indicative of?
I’m pretty sure we are not going to get a good indicator of the health of the market out of this analysis for the simple reason that I don’t have any good data for a time when the market was not rising at unsustainably high rates! In other words, a count of the days that my five-year old daughter has been alive would have a pretty good correspondence to the change in housing prices over the past five years! (and as crazy as this sounds, this indicator (my daughter’s age) this would probably have a higher statistical significance than looking at something like the Days on Market (DOM) for each home sold!!!)
Nonetheless, we may be able to pull something interesting out of the data (at least a good indicator of seasonal variation!) and looking at the # of sales vs. the # of listings could definitely help get at this… but the # of listings (or the # of pendings) is not nearly as easy to calculate.. One bit of data that is readily available is the Days on Market (DOM) for each home that has sold. I imagine this will also have a large seasonal variation.
And Joe, I’d agree that keeping things sorted by neighborhood would probably make for an interesting analysis… Here is what I propose doing tonight…
I’ll graph out the Sales Price/Original List Price for all five or six neighborhoods in the MLS definition of Ballard (I’ll separate out each neighborhood so that we can see if all the neighborhoods follow similar patterns). I’ll also go back at least one more year to give us one more annual cycle. In addition, I’ll look into the DOM for each of the neighborhoods to see what that might indicate…
Do you have any other ideas on calculations I could perform with available data?
This might sound off-the-wall… but I’m not sure of the question we’re trying to answer anymore…
My original analysis was really just trying to see if we could use the difference between the sale price and the list price to measure the health of the market. And I think I’ve concluded that we cannot…
Nonetheless, I’d agree that we’re going down an interesting path and we might be able to uncover some gems of information if we keep going. What questions would are you looking to answer? Are we simply looking for a better way to gauge the health of the local market? A new index? What is the index suppose to be indicative of?
I’m pretty sure we are not going to get a good indicator of the health of the market out of this analysis for the simple reason that I don’t have any good data for a time when the market was not rising at unsustainably high rates! In other words, a count of the days that my five-year old daughter has been alive would have a pretty good correspondence to the change in housing prices over the past five years! (and as crazy as this sounds, this indicator (my daughter’s age) this would probably have a higher statistical significance than looking at something like the Days on Market (DOM) for each home sold!!!)
Nonetheless, we may be able to pull something interesting out of the data (at least a good indicator of seasonal variation!) and looking at the # of sales vs. the # of listings could definitely help get at this… but the # of listings (or the # of pendings) is not nearly as easy to calculate.. One bit of data that is readily available is the Days on Market (DOM) for each home that has sold. I imagine this will also have a large seasonal variation.
And Joe, I’d agree that keeping things sorted by neighborhood would probably make for an interesting analysis… Here is what I propose doing tonight…
I’ll graph out the Sales Price/Original List Price for all five or six neighborhoods in the MLS definition of Ballard (I’ll separate out each neighborhood so that we can see if all the neighborhoods follow similar patterns). I’ll also go back at least one more year to give us one more annual cycle. In addition, I’ll look into the DOM for each of the neighborhoods to see what that might indicate…
Do you have any other ideas on calculations I could perform with available data?
The difference b/tw sale and list price is a very interesting dynamic. (I probably have been sloppy about answering the question:) But I was thinking, hmm, why is there a gap? I like Tom’s idea of tracking a property and trying to explain why the price dropped/rose.
What factors conspired to drive multiple offer escalations for a few months that seem to have tapered off? Or have they? Could it aid pricing?
I don’t mean to call for you to do a bunch of calculations Dustin. Thanks though:) I look forward to the number crunching. It’s great to see data in this way.
The difference b/tw sale and list price is a very interesting dynamic. (I probably have been sloppy about answering the question:) But I was thinking, hmm, why is there a gap? I like Tom’s idea of tracking a property and trying to explain why the price dropped/rose.
What factors conspired to drive multiple offer escalations for a few months that seem to have tapered off? Or have they? Could it aid pricing?
I don’t mean to call for you to do a bunch of calculations Dustin. Thanks though:) I look forward to the number crunching. It’s great to see data in this way.
I am trying to find out the sale price of a home sold in Seattle this past summer/fall. I live in California. How can I find this information out? I tried googling a few search words, but couldn’t find anything. When a house gets sold in the Bay Area, where I live, the asking and sale price are listed each month. I’m hoping there is some public access source that will give me this information through the Internet.
Thanks.
I am trying to find out the sale price of a home sold in Seattle this past summer/fall. I live in California. How can I find this information out? I tried googling a few search words, but couldn’t find anything. When a house gets sold in the Bay Area, where I live, the asking and sale price are listed each month. I’m hoping there is some public access source that will give me this information through the Internet.
Thanks.
Roxann,
There are a couple of different places that give the sold information for the Seattle area. The ones I’m aware of are:
1) http://www.homepages.com
2) http://www.shackprices.com
3) http://www.redfin.com
All three of them have very different interfaces… and I’m not sure that any of them are going to have prices that are up-to-date. It kind of depends on where they get their data.
I know that ShackPrices gets their data from Kind County records which means that theirs will include for-sale-by-owner (FSBO) properties. But that also means that a home sale will not go into their database until the sale is recorded by King County and then published on the internet (maybe a month or two?).
The other two sites include home sale information so they are more likely to get their sale data right after a home is sold (assuming the home was listed in the MLS). I’m not sure if either of them supplement the data with FSBO data.
Hope that helps!
Roxann,
There are a couple of different places that give the sold information for the Seattle area. The ones I’m aware of are:
1) http://www.homepages.com
2) http://www.shackprices.com
3) http://www.redfin.com
All three of them have very different interfaces… and I’m not sure that any of them are going to have prices that are up-to-date. It kind of depends on where they get their data.
I know that ShackPrices gets their data from Kind County records which means that theirs will include for-sale-by-owner (FSBO) properties. But that also means that a home sale will not go into their database until the sale is recorded by King County and then published on the internet (maybe a month or two?).
The other two sites include home sale information so they are more likely to get their sale data right after a home is sold (assuming the home was listed in the MLS). I’m not sure if either of them supplement the data with FSBO data.
Hope that helps!
I love the mathematical genius behind your chart. I was a pretty interesting comparing % of list and # sold. I look forward to reading though all of your other posts.
Walt