$549,999 is A Better Price for Your Home Than $550,001 (5% Better)

With more than 70% of home-buyers looking on the Web for real estate to buy, we wondered if it made sense when pricing a house to take into account the parameters used on most real estate search sites. For example, since every site lets folks filter on price in increments of $25,000 at lower price ranges and increments of $50,000 at higher price ranges, wouldn’t a property priced at $549,999 get seen by more Web shoppers than one priced at $550,001?

The answer is maybe, just a little.

How so? Enter Mose Andre, Redfin’s ace statistician, who analyzed the logs of the Redfin site to determine how often Seattle users of our site see properties in different price ranges, between September 10, 2007 and September 24, 2007. His findings:

• About 30% of searches don’t even filter on price. But the number of searches that don’t filter on price is exaggerated on Redfin’s site because Redfin.com price filters aren’t easy for users to find.
• For most neighborhoods, the maximum percentage of Redfin searches you are likely to lose by moving from one price band to the next is 6.5% . For most Seattle neighborhoods, this band occurs for homes costing more than $550,000.

Based on these findings, we would only recommend taking into account how search sites filter on price in cases where a property is priced very near one of the popular threshold amounts. In other words, if you were going to price a house at $570,000, you shouldn’t price at $549,000 just to have it show up in 6.5% more price-filtered searches; but we would consider it if you you were going to price a house at $551,000.

You can see how this plays out on Mose’s graph of search exposure and listing count for Bridle Trails:

The red line represents the percentage of Redfin’s Bridle Trails searches filtering on price that include Bridle Trails properties at different price points; use the numbers on the left axis to measure the percentage of searches that return a result at the prices appearing along the bottom axis. As you can see, less than 20% of Redfin’s Bridle Trails searches filtering on price include properties costing more than $800,000.

The black line represents the density of listings in the area; more precisely it is a curve fitted to the shape of a histogram representing the number of listings at different prices. You can use the numbers at right to track the number of listings at different price ranges. The most common price is the one where demand becomes scarce: $800,000.

The biggest drop in buyer exposure in Bridle Trails occurs at $550,000. One reason drops tend to occur at this point is that Redfin, like many other real estate search sites, only allows price filtering at $50,000 increments for prices greater than $500,000. So the first $50,000 steps are doozies.

Let’s look at a few more graphs, this one of Capitol Hill:

Here most of the inventory is clustered at a price just below $400,000, probably because there is a glut of condominiums on the market, and most of the price-filtered searches are in that range. There is a little hump around $700,000 for houses and townhouses in the neighborhood.

One more graph, this time for stuffy, old Queen Anne…

And here is a table of the price-points where the biggest drop in search activity occurs, and how large that drop is:

If you want to see how demand compares to inventory for your neighborhood, download a package of all our graphs for the Seattle area. If you want these graphs for another market like San Francisco or Boston, just let us know. Thanks to Mose Andre for the stats and analysis; if there are other analyses you’d like to see us perform, just leave a comment for that too.

Update: Mose cranked out some San Francisco graphs.

First Freakonomics, Then the Redfin Advantage, Now An Academic Study Spanning Six Years

Last February, when the rain wouldn’t stop and we were bored out of our minds, Redfin released a year of sales records indicating that our buyers on average got a better deal than customers of other brokerages, on top of the commission savings.

Mose Andre, Redfin’s compulsive stats man, has only recently recovered. Hundreds of bloggers, commenters, e-mailers and callers raged against the idea that Redfin customers got a better deal, or that our agents had any part in our customers’ success. But the data held up.

In childish, tearful rants, I defended our agents. Our CTO, Michael Young, poked his head into my office to ask, “Who cares why our customers win, if they win?” And then shrugged (he has a two year-old). Mose nearly had a nervous breakdown calculating and re-calculating the numbers, then slept for two days straight.

But ever the kinky masochist, last week Mose called me out in the hallway to ask why we hadn’t tallied up the Redfin Advantage for our listing customers.

“Too hard,” I said, turning around. “We could intentionally set a low price than claim a big mark-up. What’s the right number to compare ourselves against?”
“The assessed value,” Mose said. “The Zestimate.”
“People would question those numbers, too,” I said.
“It doesn’t even matter if the baseline number is wrong,” Mose said. “As long as it’s consistently wrong for everybody.” He was now surrounded by his math nerds, and I was all alone.
“Try explaining that in a blog post,” I said.
“Just because it’s hard to explain doesn’t mean it isn’t worth doing,” Mose said.
I started to back away. Mose smiled and said he would come back from vacation with a new way to figure out how our listing customers really did.

Well, it turns out that somebody beat him to it (hopefully Mose will realize he should never go on vacation again). A Northwestern economics professor bet his colleague that a traditional listing agent increases the price of a home, and then spent the next three years analyzing Madison, Wisconsin data from 1998 – 2004 to prove his point. Today, that professor is taking his colleague to lunch, because he was wrong. The traditional agent often doesn’t get a higher price, and consumers know what their home is worth better than anyone in traditional real estate has admitted.

According to a review of the study published in this morning’s New York Times, people in Madison, Wisconsin “who sold their homes through real estate agents typically did not get a higher sale price than people who sold their homes themselves.” In fact, the study found, the agent-sold homes actually sold for slightly less (the difference though was within the study’s margin of error).

The study pointed out one bright side for the traditional industry, reporting that Realtor-listed properties sold more quickly (105 days vs 125 days), but we’re not sure this is such a simple advantage. According to another study by Freakonomics professor Steven Levitt, when Realtors list their own properties, the properties are on the market longer because the Realtor is holding out for a better price. Perhaps Madison home-owners took the same approach.

The Northwestern study worked because Madison is a kind real estate of Neverland, where more than 10% of all the homes for sale are available on a single For-Sale-By-Owner — FSBO — site, FSBOMadison.com, which still allowed owners to offer the buyer’s agent a commission. So the data set of FSBO sales in Madison was large enough that the professors could correct for all sorts of skewing factors, like lot size, neighborhood and time of year — and compare it to Realtor-listed sales.

Everywhere else in America, FSBO marketshare has declined (14% to 12% from 2002 to 2006, scattered across many sites) at the same rate as traditional brokerages (74% to 70%), with alternative brokerages like Redfin taking up the slack. One reason for the decline is that through services like Redfin Direct and many others, consumers can now list their home in the MLS without paying their listing agent a traditional commission.

Which brings us to the final twist: we feel kind of weird promoting a FSBO study. It drives us crazy when traditional agents claim we’re a FSBO type of service. Redfin agents work with clients to price and promote their homes, to negotiate a deal and to handle all the paperwork associated with the sale. So it cheered us to see one of the study’s authors, Aviv Nevo, acknowledge that you do of course want to pay a listing agent for the work he does, so long as you don’t give him a piece of the action based instead on the value of your house. Which is how we’ve paid Redfin agents all along.

Pricing Advice: Make the Last 3 Digits -500

There are very few people with Matt Bell’s zeal for negotiating. He is 6′5”, with a large, slow smile that seems to bespeak an unused capacity for terrific violence, and he is faultlessly congenial. The best way to summarize our friendship is to say that he taught me to shotgun a beer for the first time at the age of 34. I wasn’t very good at it.

Working together at Plumtree, we once took an elevator to the penthouse floor of a massive bank’s headquarters to ask for a $4 million deal. We rode in silence, hands in our pockets for the first 40 floors. When the elevator was about to ding, I opened my mouth to say, “I hate asking for money.” Before I could, Matt said, “Let’s make it $5 million.”

It turned out to be the largest deal in Plumtree’s history, triumphantly negotiated by the bank back to $4 million, and it helped Matt buy the house that he just sold through Redfin. While he was still haggling last week over the cost of roof repairs, we went to lunch, and Matt began speculating on the list price most likely to result in the highest offer. It is the kind of conversation that makes me wonder if my friend is from another planet.

“Does a price that ends in -000, like $490,000 seem casual? Is $499,999 too blue-light special?” I stared into my salad. Little did I know that Redfin’s mad scientist, Mose Andre, was working on that very problem, crunching statistics on the data-set we pulled to calculate the Redfin Advantage.

To do the analysis, Mose took all the houses that sold in King County, Washington last year and grouped them by the last three digits of their list price. For example, one group would consist of all the houses whose list price ended with “-500,” like $499,500, $387,500, $831,500, and $1,230,500. The four most popular endings for list prices of houses in 2006 were “-000,” “-500,” “-900,” and “-950.” Less than 7% of properties were listed at prices that did not end in those four numbers.

Then we threw out new construction, which tends to sell at list price even if other incentives are involved; we also threw out some records where we couldn’t easily tell if it was new construction or not.

And then for each group we calculated the ratio of list price to final price. And it turns out that certain list prices did in fact tend to result in a higher premium over the list price.

The ending that resulted in the highest final price as compared to list turned out to be “-500,” as in $499,500 or $530,500. And the difference was significant: listing for $500 less than an -000 ending seemed to result in a final price that was $3,000 more.

Maybe rounding a list price to a nice, even “-000″ is like putting a big “negotiate me” sticker on a house’s back. Or, as Matt speculated, “A -500 ending sounds like you really thought about it, but it’s not a nickel-and-diming gimmick like -999.”

The column labeled “$ Over -000 Price” compares the final/list ratio for each ending using the -000 final/list ratio as a baseline since it was the lowest; we came up with a dollar difference by using a hypothetical final price of $600,000. The data for condos is also interesting, although there was only one price ending besides -000 that was popular enough to report on, -950. As you might have guessed, it was better than a price ending in -000:

The “$ Over -000 Ending” was calculated using the “-000″ final/list as a baseline, just as before, but assuming a $400,000 average price for condominiums.

Even though it makes me feel like a mutual fund to say it, Mose wants everyone to know that these numbers reflect what happened in 2006, not necessarily what will happen in 2007. Had we world enough and time, as well as more data, he says we would compare listing prices in which the first three digits were constant, and the last three varied. Mose is still a little traumatized by all the trouble our last report on MLS data created, which wasn’t even his fault… but he signed off his e-mail to me tonight by asking “why is this stuff so fun?”

OK, We Can Be Moved .01%

Thanks to McKinsey-trained Kevin Boer, one of several brokers who reviewed Redfin’s NWMLS data, Redfin has discovered that we screwed up the accounting for one transaction in our analysis of Redfin’s negotiating advantage.

Except for this transaction, Kevin seems to have corroborated our analysis.

We originally reported that our King County buyers got a final price of 99.329% below list, whereas King County customers of other brokerages paid 100.233% above list. This is factually correct. But one transaction should have been adjusted to account for a commission refund applied to the purchase price, so as to isolate the negotiating capabilities of Redfin and its customers. Making this adjustment leads to an average final price 99.340% below list.

With this adjustment, the negotiating advantage we claimed to be .904% is .893%. This reduces Redfin’s negotiating advantage by $54, from 4,474 to $4,420. This advantage is still financially meaningful and statistically significant, but we are nonetheless unhappy with ourselves for the error.

The source of the error was obscure: one, and only one, of our 170 King County customers offered to allow the seller to keep Redfin’s 2% commission refund if the seller would lower the price an additional 2% (on top of the 2% commission savings factored into the price, the house in question still sold for nearly 2% below listing price). The MLS # for the transaction was 26136978.

The NWMLS thus recorded the final price as being nearly 4% below listing price, but roughly half of this advantage came because the seller essentially received Redfin’s commission refund. Had we realized this when performing the initial analysis we would absolutely have added the commission refund on top of the final price.

And we would have identified this transaction earlier but for an error in our customer database indicating that the customer had qualified for a 2% commission refund. Before publishing our analysis, we double-checked this customer database against our financial records, but this only confirmed the amount of the commission refund, not that it was or was not offered to the seller to reduce the final price.

When Kevin inquired about the possibility that a commission refund was applied to the purchase price, we were about to e-mail him that this had not happened. Before we did, we decided to review the official HUD-1 forms for King County deals with exceptionally low prices as compared to list, and then, when we found a problem, for every deal included in the study. While I was wining and dining a college recruit in Berkeley, Rob and Cynthia were in the office Wednesday night pulling every file from last year.

We spent Thursday double-checking and re-calculating the data based on an error in one transaction, e-mailing Kevin later that day to explain the error.

Once we recognized the problem, we could have actually accounted for it in one of two ways, either by lowering our negotiating advantage or lowering our average commission refund. Lowering the commission refund amount would have allowed us to avoid making an adjustment to NWMLS data, which was appealing to us because the NWMLS data is a matter of public record for other brokers and agents.

We decided against this. Since the buyer got an extra 2% reduction in price only by using Redfin’s commission refund in the negotiation, we decided to reflect the change in a lower negotiating advantage. We also thought that lowering the negotiating advantage was the most conservative approach, since the negotiating advantage has been most hotly disputed.

Over the course of the morning, we will update our website, issue a corrected press release, and contact journalists and bloggers to whom we had sent the numbers that included this error.

In other news, the NWMLS report cited by various bloggers as contradicting Redfin’s data turned out to be wrong by a large margin. The NWMLS adjusted the ratio of the median final price vs. the median list price from 81.61% to 99.52%. When the NWMLS realized that a further adjustment upwards seemed to be in order, it re-published the report a second time with the table in question entirely removed.

Thanks again to Kevin Boer for finding Redfin’s error. After a week of intense scrutiny, the basic conclusion that Redfin’s King County customers got a price significantly better than customers of other brokerages still stands, but we are exhausted from having to re-analyze an already exhausting analysis, and apologetic to everyone we let down by making this mistake.

Here We Stand, For We Cannot Be Moved

Publishing MLS data that shows that Redfin got a better deal for buyers than agents at other brokerages sparked a riot yesterday: here, here, here, here, here, here, here, here, here, here and here. We also showed up in Freakonomics (holy cow!).

There has been a healthy discussion about how to interpret the data: whether Redfin agents negotiate better, for example, or its customers tend to seek better deals. We think both factors contribute to our success, and we love the debate.

But attacks on the main finding, that Redfin customers tend to pay less for properties above and beyond the commission refund, have been flat-out wrong:

The date range was arbitrary (see comment #5): the date range we chose was from February 6, 2006 to February 5, 2007; exactly one year from the launch of our home-buying service. If the date range had been January 1, 2006 to December 31, 2006, as critics suggested it should have been, our negotiating advantage would still have been .769%. It seemed less fair, not more fair, to include January 2006 data when Redfin Direct was not available in January 2006, but the overall result still favors Redfin.

Redfin sales increased as the market softened, skewing its advantage (see comment #6): if we analyzed only the last 90 days of the time period studied, the Redfin negotiating advantage would have been 1.10%, as opposed to the .904% we calculated for a full year. Redfin’s negotiating advantage over other brokerages actually increased, not decreased, with deal volume.

The data are not statistically significant: based on a p-value calculated from the MLS data set, the likelihood that Redfin’s advantage is entirely due to a small sample rather than a legitimate difference is less than 3%.

An NWMLS report contradicts the NWMLS data Redfin cites: an NWMLS report states the median final price of King County homes sold in 2006 was 81% of the median listing price of King County homes listed in 2006. No one believes that a typical home sells for 19% below its list price; to verify this, Redfin retrieved from the MLS every record of a house or condominium sale that closed in 2006; of the 37,185 transactions, only 49 (.13%) closed at a discount of 19% or more. If that report were correct, a discount of this size would have been 385 times more prevalent than it actually was. We have called and written the NWMLS, which is verifying its own report; we will notify you when the source data or the methodology becomes available. If this report proves us wrong, we will say so.

The data are impossible to replicate: we published a methodology for replicating the data that several complete NWMLS neophytes were able to follow. Nonetheless, we are now offering to share the data, with addresses and other private information removed, until the NWMLS objects. We already have sent the data to folks from Rain City Guide, Three Oceans Real Estate, Bloodhound Blog and 360 Digest, all of which have been strongly critical of Redfin in the past; if there is an error, one of these bloggers will find it.

Having challenged us, we would ask at this point that our critics report their findings, whether there is an error or not.

A Year’s Numbers Come in From the MLS: Redfin Agents Negotiate Better

Redfin announced big news today, publishing MLS data that indicates our agents negotiate significantly better than their counterparts at traditional brokerages.

Last month at Brad Inman’s big real estate conference in New York, Realtor.com President Allan Dalton accused Redfin and other critics of a “massive level of disingenuous communication,” because we ignore the likelihood that traditional agents offset higher commissions by negotiating a better price (skip to minute 9:00):

Already our surveys had established that most of our customers get service that they believe is better or much better than what they got from a traditional real estate agent. But forget the touchy-feeley stuff: every day we heard that our customers would lose our commission refund and more at the negotiating table; it is the centerpiece of the traditional industry’s argument against Redfin.

And we agree, that the price of a home fluctuates with market conditions, increasing the importance of a real estate agent’s pricing guidance and negotiating ability.

But our problem with commissions is not simply that they’re too high; our problem is with the commission itself, because it pays the buyer’s agent more when his clients pay more. In other words, rather than being offset by better negotiations, the buyer agent’s commission actually causes worse negotiations.

This is why we decided to pay Redfin agents a salary with a customer satisfaction bonus, not a commission. Agents do what you pay them to do, we reasoned, and we believed our agents would be more likely to get the price our customers wanted.

After a year in the market, we decided to put our theory to the test, by querying the Northwest Multiple Listing Service for data on every home or condominium sold via a brokerage from February 6, 2006 (the date of Redfin Direct’s launch) through February 5, 2007. Since we didn’t offer a service for sellers or support areas outside King County until much later in the year, we limited the data to King County and we only evaluated our capacity as buyers’ agents.

But we still had the problem that Allan highlighted, namely that there is no “set base” price for a home.

So we compared what buyers’ agents negotiate for — the final price — to what the sellers’ agents ask for — the asking or listing price; some sellers’ agents may ask for too much, others for too little, but, since all our customers are all shopping in the same store, looking at the same listings, all King County brokerages are negotiating against the same set of asking prices (note that evaluating a seller’s agent is problematic, since the seller’s agent only competes against the prices she sets herself.)

The results were striking; Redfin customers paid on average under asking price, whereas customers of all other brokerages paid on average over asking price. The difference in negotiations was .9% of the home price, equivalent in King County to over $4,000, on top of a commission refund of nearly $10,000.

What makes this noteworthy is that the data did not come from Redfin, but from the MLS, from the brokers themselves who contribute to the MLS. Any brokerage can validate the data by following the instructions available in the appendix of our report.

Already, the Seattles Times reviewed the report and picked up the story in yesterday’s big Sunday spread.

Perhaps there is another way to evaluate whether traditional agents negotiate better than Redfin agents; until there is, the most likely conclusion is that Redfin agents negotiate better than their more expensive counterparts.