Redfin Saves You Money!!! So they say…
Redfin announced their first year performance statistics yesterday. The gist of their announcement was that “
My first impression was, ..wow! That’s definitely interesting. However, never having been a person to just take someone’s word blindly, I did my own number crunching. I originally posted these findings over on Rain City Guide earlier today. More discussions on this are on 360Digest, Bloodhoundblog, Freakonomics, and another one on Rain City Guide.
According to the NWMLS, there were 69,048 closed transactions for residential and condo homes in 2006 (in King, Pierce, and
Redfin may be perfectly correct in their analysis of their data …that a Redfin buyer paid $4,474 less on average, but I’m going to argue the point that using 224 transactions out of 69,048 to draw any kind of conclusion about the other 99.7% of the transactions and the agents that represented those transactions is not only silly, but very misleading and not statistically sound.
To emphasize my point that Redfin over-generalized the other 99.7% of transactions, I did some number pulling from the NWMLS. I wanted to see how Redfin agents on the selling (buyer’s) side compared directly to other brokerages and compare them straight up rather than every brokerage lumped together. I first wrote down the first 5 brokerages I could think of (I just pictured driving down a local road and which brokerages were there). Since Redfin is listed in the NWMLS as being based in
Century 21 North Homes
Windermere Real Estate Co.
John L. Scott Inc.
Coldwell Banker Bain
ReMax Metro Realty
I pulled closed Residential (no condos) sales with Redfin’s first year time frame of 2/6/06 to 2/5/07. I did not limit it to
Median Sale Price to Median List Price comparison:
100.97% - Redfin - based on 149 closed sales
100.31% - Century 21 North Homes based on 575 closed sales
99.72% - Windermere Real Estate Co based on 1032 closed sales
100.00% - John L. Scott Inc. based on 5274 closed sales
99.43% - Coldwell Banker Bain based on 4093 closed sales
100.19% - ReMax Metro Realty based on 940 closed sales
Let’s translate that into dollars. Based on the same data, I averaged the Median selling price of all 5 companies and got $412,850. Round up to $415k for easier number to use. So, based on the above percentages and a house that’s listed for $415k:
Redfin buyer pays $419,016
Century 21 North Homes buyer pays $416,306
Windermere Real Estate Co. buyer pays $413,828
John L. Scott Inc. buyer pays $415,000
Coldwell Banker Bain buyer pays $412,646
ReMax Metro Realty buyer pays $415,776
…continuing the number analysis…I broke down the Sale Price to List Price percentage based on Cumulative Days on Market. For homes that were on the market:
between 0-30 days, Redfin was 1st at 99.86% while the others ranged from 100.07% to 100.90%. This means that a Redfin buyer paid the least (99.86% of list price).
between 31-60, Redfin’s 98% was 2nd to Windermere’s 97.58% with the most at 99.17%.
between 61-90, Redfin’s 98.74% was 5th with only Century 21 North Home’s 98.89% higher than it. Lowest and best was Windermere’s 96.44%.
between 91-120, Redfin’s 99.08% was again 5th with again Century 21 North Home’s 99.32% higher than it. Again lowest and best was Windermere’s 97.48%.
From 120+ days, Redfin stole the show with 93.45%. The others ranged between 96.13% and 98.04%.
Wow, ..93.45% ..something is way out of line. So I looked at the 149 closed homes Redfin had and filtered out the ones that were 120+ CDOM. There is an element of error here that I must mention …CDOM is based on the listing agent accurately entering when the house actually went pending. Those in the business knows…this does not always happen. Which may explain when I actually looked at the individual listings ..there were 18 homes showing 120+ days in the CDOM with an average SP/LP of 95.93%. Now…of these 18, four of them were $1 million + homes. For homes that high in price, huge price reductions in final sale price is not uncommon. So I simplified it. I eliminated all four of them. The remaining 14 homes now had an average SP/LP of 97.42%. At this number, it would put them at 4th, with the high and low of 98.04% and 96.13% respectively. But if you don’t like how I eliminated the million+ dollar properties, then fine,.. Redfin wins the 120+ day category with their 18 sales whereas say…John L. Scott had 605 sales of 120+ CDOM and Coldwell Banker Bain had 411.
So… looks like brand new listings ..Redfin agents did slightly better. However, as the listing ages, ..the Redfin agents did progressively worse.
I’d be happy to share my spreadsheet that I have all this data in.
I read that you were a math major in college. No surprise there. I loved your analysis of the numbers.
Thanks! The general public will take most published things (the Seattle Times Sunday paper had a full spread on Redfin) at face value and never think to question it. I’m glad that I can at least verify or deny real estate stats on my own. Hopefully people will look to find the truth or at least come to their own conclusions themselves.
First, a few disclaimers:
(1) I work for Redfin, although I’m a programmer, not an agent, and I was not involved in any of the data crunching for the report. I haven’t actually even seen the data.
(2) This post is my personal reaction and does not reflect any kind of official Redfin opinion. I speak for myself alone.
So, with that out of the way, I wanted to commend you for getting down into the numbers while taking issue with two parts of your analysis.
First, you say that Redfin’s use of around 200 transactions is “silly” and “not statistically sound”, but then you use start slicing up the Redfin data by CDOM and drawing further conclusions. I may be missing something here, but isn’t this a contradiction? It seems to me that either 224 transactions is too small an amount from which to draw statistically significant inferences, or it isn’t. I don’t think you can have it both ways.
The second issue I have here is about the methodology, and it may be that I’m just misunderstanding your description of how you crunched the numbers. You say that you got your percentages by comparing “Median Sale Price to Median List Price”. I assume this means that you calculated the median sale price for each brokerage and then divided by the brokerage’s median list price.
My problem is that median sale price divided by median list price tells you almost nothing about the ratio of list and sale price on an average deal. The median of sale price divided by list price is much more indicative, and the average of sale price divided by list price is probably even better.
To see this, take an example of two theoretical agents, A and B, who each sold a total of three homes: one listed at $100K, one listed at $1 million, and one listed at $10 million. Agent A negotiated every house down to half its price ($50K, $500K, $5 million, respectively). Agent B got a good deal on the $1 million house, negotiating it down to $450K, but he did a horrible job on the other two houses and got negotiated up to three times the price, buying them for $300K and $30 million. Let’s look at the stats:
Agent A
——-
Median Sale: $500K
Median List: $1 million
Median Sale/Median List (%): 50%
Median (Sale/List) (%): 50%
Average (Sale/List) (%): 50%
Agent B
——-
Median Sale: $450K
Median List: $1 million
Median Sale/Median List: 45%
Median (Sale/List) (%): 300%
Average (Sale/List) (%): 215%
The metric you used, median sale/median list, says that Agent B did a better job, but that’s clearly not the case. Agent B did better than agent A on one deal by a small amount and got absolutely pummeled on the other two deals. What we’re really interested in here is the average ratio of sale and list price, and in that stat, agent A blows agent B out of the water (as well she should). Average of sale/list is, I believe, what Redfin used in their report, and I think they were right to do so.
As I said before, I want to thank you for getting down into the data. Personally, I’m excited to see people making arguments based on hard numbers, and I thank you for your contribution to the lively discussion.
Hi Sasha,
Thank you for your well thought out argument. If I said that because the number of closed transactions redfin was involved in was statistically insignificant, and left it at that…that wouldn’t have been very useful. So, if we accepted that 224 transactions was fine, then we could try to analyze the performance in another way (my head to head comparison with other brokerages). My apologies for not making that transition of my argument clear.
I’m a bit confused with one of your paragraphs. Starts out “My problem is that median sale price divided by median list ….” In there you say that median of sale price divided by list price is much more indicative. I don’t agree cause you’re mixing the two, but I’ll entertain your point and and see what happens to the numbers (taken from my spreadsheet which I already have all the numbers in question). Here’s what happens (i’m shortening the brokerages names here to save space, refer back to the original post for full office names):
Brokerage: Median SP / Mean LP
Redfin: $469,500/$558,617 = 84.05%
Century21: $305,950/$329,772 = 92.79%
Windermere: $485,500/$615,899 = 78.83%
John L. Scott: $375,000/$453,043 = 82.77%
CB Bain: $425,900/$554,593 = 76.80%
Remax: $415,250/$508,807 = 81.61%
This is what I don’t like about mixing numbers derived from means and medians…you get funky results. If I was a seller looking at these, I wouldn’t hire any of them.
So let’s look at the other metric you wanted to see. Avg SP / Avg LP: …If I may, I’ll opt out of writing the actual figures and just report the percentages:
Redfin: 98.51
Century21: 99.44%
Windermere: 99.37%
John L. Scott: 99.28%
CB Bain: 98.94%
ReMax: 99.58%
Redfin bests everyone but only by 0.43% ..so roughly half of one percent. So for a $500k house that’s a $2150 redfin advantage.
If I left it at that, I’d be disappointed in myself. The closest brokerages was CBBain. Only 0.43% higher. However, CBBain had nearly 25 times more volume (4093 sales vs 149).
Can you say for sure that if Redfin’s sales volume increased to the same range as CBBain’s that their “negotiating advantage” will still be the best? That’s anyone’s guess. I’d say ..it’d be tough….not impossible, but tough.
Ultimately, to me, …it boils down to this. Redfin is a competitor. I welcome competition. It keeps everyone honest and striving to perform their best. I’d say let’s have a friendly hand shake and may the best person win.
On the other hand, I kind of chuckle. Agents giving rebates to their clients is no new thing. Agents often use their commission to solve a sticky problem between the parties, or to say thanks for their continued business, or whatever. It’s said (at least I’ve heard it many times) that agents who feel necessary to automatically undercut their price (as opposed to a case by case situation) do it because it’s the primary way they can draw clients in. Their work alone isn’t enough to stand on its own. …now entire companies are basing their business model on this discount approach. Okay, fine.
My biggest beef is with the Redfin marketing approach. The Redfin Marketing Machine chose to take one metric that worked in their favor and broadcast that message. The average person will never question things in print, and thus in a way, …it’s false advertising. At least it’s not giving people all the facts, and that is what this online community is trying to do. Provide the other sides of the story that people usually will not see.
The best way to resolve this debate (and it will never happen correctly) is to have several agents go and negotiate on the same house with the same seller and see who gets the lowest price. Unfortunately, if this happened in the real world, that’s a multiple-offer situation in which the highest bid usually gets the house.
I appreciate your insight and good luck.
James,
Thanks for the response. I have to run to a volleyball game, so this will have to be brief.
I think you may have misunderstood my point about your analysis. I wasn’t saying that median sale price divided by average list price is interesting; I agree with you that it’s meaningless.
I was trying to point out, though, that the ratio of two medians is NOT the same as the median of the individual ratios. Similarly, the ratio of two averages is not the same as the average of the individual ratios. Put more mathematically, if we have paired sets of data x and y:
1) Median(x)/Median(y) does not necessarily equal Median(x/y).
2) Average(x)/Average(y) does not necessarily equal Average(x/y).
In the data set I gave for my hypothetical agent B, the median list price was $1 million and the median sale price was $450K. Thus the ratio of the two median is 45%. When we look at the individual ratios, though, we see that the median is 300%, because the ratios are 45%, 300%, and 300%. The latter median is at least somewhat interesting, as it tells you what kind of deal the median client got.
As I said in my post, I think average is more helpful here for answering the question we’re asking, but once again average sale price divided by the average list price is different than the average of the sale price divided by the list price. In agent B’s data set, (average(sale)/average(list) is 277%, whereas average(sale/list) is 215%.
It looks to me from what you wrote that you used Median(x)/Median(y) and, in your last comment, Average(x)/Average(y). I think that those numbers aren’t the ones we are interested in. Do you disagree?
I really have to run, and I don’t have time to respond to your other, very interesting, comments, but thanks for keeping the discussion lively and civil.
A Refin representative recently spoke at a big real estate technology conference in Seattle. It was called the Luxury Real Estate Tech Exchange. I can’t remember his name exactly, but he was very hopeful about his company becoming very successful in the future. I think it’s really cool that Redfin is trying to find a new niche in the real estate market for regular home buyers. But no company is better at offering the best homes in the market than Luxury Real Estate.
Check out http://www.LuxuryRealEstate.com for the most amazing homes in the worldwide market.
In our local market, I know that an agent by the name of Wendy Lister of Coldwell Banker Bain has a large chunk of the luxury (aka…super high end) market. I’ve actually never heard of luxuryrealestate.com. Looks like they’re trying to be a global presence. I don’t know what to think of that.
Redfin can aspire to be whatever they want to be. I just don’t like it when them, or anyone else for that matter, makes blanket statements that has so many holes in it. Or it should be accompanied with a bunch of asterisks.