The Shopping Study Measures the Smallest Number in Your Deal

Every source a buyer is likely to consult says the same thing. Get three quotes. Get five. Compare the Loan Estimates. The CFPB says it, the personal finance columns say it, and every AI assistant will say it within two sentences of being asked how to get a good mortgage.

The number behind the advice comes from Freddie Mac. Five quotes saves about $3,000 over the life of the loan. One extra quote saves about $1,500.

Spread across 360 payments, $1,500 is four dollars a month.

A real file, and where the money actually was

A seller agreed to $12,108 in concessions. By default that money disappears into closing costs and nobody makes a decision about it. On this file it went into a 2-1 buydown instead. The first year payment went from $3,432 to $2,765. Same buyer, same house, same lender, same day. $667 a month.

That decision was not available to anybody comparing rate quotes, because it was never a rate question. It was a question about what the seller had already agreed to give up, and what to do with it once it arrived.

Nobody on that file saved money by shopping harder. The agent knew what to ask the seller for and the lender knew what to do with it.

What actually decides what a mortgage costs you

What the offer says, not only what it pays

A seller choosing between offers is guessing at which one closes. Price is one line on the page. Closing date, contingency deadlines, appraisal gap language, who pays title, whether the seller can stay in the house for two weeks after closing, all of it is negotiable and all of it is worth money to somebody on the other side. An agent who knows this seller needs a June 30 close can trade for that. Every dollar you do not overpay is a dollar you never borrow, never pay interest on, and never have to refinance your way out of.

Where a concession goes

Seller-paid money can cover closing costs, buy the rate down for two years, buy it down permanently, or fund the prepaid taxes and insurance. Those are four different payments out of one pot of money. The default is usually the weakest of the four, and it happens by omission rather than by choice.

Which loan you are actually in

Conventional mortgage insurance comes off. FHA mortgage insurance on a loan above 90 percent of value stays for as long as the loan exists. Two lenders can quote you an identical rate on those two loans and one of them costs many thousands more over the years you keep the house. Program choice also sets your minimum down payment, your reserve requirement, and whether down payment assistance is on the table at all.

Where your cash sits

Down payment, points, reserves and repair money are one pool with four uses. Crossing 80 percent of value drops mortgage insurance and improves pricing at the same time, so a few thousand dollars moved from one column to another can change the payment more than every quote you will ever collect. Sometimes the right answer runs the other direction and you put less down and keep the cash. A rate comparison cannot evaluate any of that, because the loan itself changes.

Whether the seller believes you will close

Sellers take less money for offers they believe. Underwritten instead of prequalified, a closing timeline the lender will commit to in writing, an appraisal ordered the day the contract is signed. That belief is worth negotiating room, and negotiating room is worth more than an eighth of a point.

The part that never shows up in the advice

You do not negotiate with the listing agent. Your agent does. If they are any good they already know what is available to ask for on that particular listing and what this seller will trade, because that is their job and they have been reading these contracts all year.

What they need from the lender is the answer to one question: what can this buyer absorb? How large a concession is worth asking for before it stops doing anything. Whether the file survives a low appraisal, and by how much. How fast it can genuinely close, not the number on the marketing. Whether $6,000 moved into the rate beats $6,000 off the price on this specific loan. An agent holding those answers negotiates for the right things. An agent without them negotiates on price, because price is the only variable they can see.

Now drop a five-lender auction into the middle of that. The buyer becomes the only line of communication between the lender and their own agent, and the buyer has done this once. Whoever won the auction by eight basis points has no relationship with that agent, no stake in this contract, and no reason to answer the phone on the Saturday of inspection week. The agent ends up negotiating blind. The money that was sitting in the contract never gets asked for.

That is what the advice costs. Comparing prices is harmless on its own. The damage is that it quietly hands the buyer the job of coordinating the two people who were supposed to be working the file together, in the exact week when every decision that matters gets made at once.

What the study actually measured

None of this makes the research wrong. Freddie Mac measured something real and measured it honestly. It is worth knowing what.

They did not follow a group of borrowers who shopped and compare them to a group who did not. There was no such experiment. They took rate quotes from applications submitted through Loan Product Advisor, their underwriting system, filtered them down to similar borrower profiles on the same day, and then ran a simulation. From that pool they drew 500 rates at random and averaged them. Then they drew two and kept the lower one. Then three, then four, then five. The gap between one random draw and the best of five is where the $3,000 comes from.

So the finding is that on a given day, for the same borrower and the same loan, price varies between lenders more than most people expect. That is true and worth knowing.

Four things follow from how it was built. The draws are random and independent, which no real shopper is, so the figure is a ceiling rather than a typical result. The loan is held identical on purpose, which is good method and also means the simulation cannot see a single decision listed above. The quotes are quotes, and Freddie Mac says plainly that an application is “neither a pledge by the applicant nor a commitment by the lender.” And Loan Product Advisor is a conventional system, so FHA, VA, USDA and down payment assistance never enter the pool, which are exactly the files where structure decisions are largest.

Why this is the advice everybody gives

Shopping advice scales. It applies to everyone, it requires knowing nothing about your income, your seller or your appraisal, and it cannot be wrong. That makes it the only safe thing for a federal agency, a columnist or a chatbot to say to millions of people at once.

The decisions that move real money cannot be delivered that way. They are specific to one contract, on one house, with one seller who wants something. Somebody has to read the contract. There is no general-purpose version, so it does not appear in general-purpose advice, and the gap has nothing to do with anybody hiding anything.

There is a second cost that never gets mentioned. Doing it yourself puts the entire transaction on you. Five sets of documents, five sets of quotes that are not comparable because nobody has designed the loan yet, and at the end you still own every decision, because none of the five ever saw the whole picture and none of them was accountable for how it turned out.

Decision	What it is worth
One extra rate quote (Freddie Mac's figure, spread over 360 payments)	about $4 a month
Five rate quotes (same basis)	about $8 a month
A $12,108 seller concession put into a 2-1 buydown instead of closing costs	$667 a month in year one
Your agent getting the contract accepted $10,000 lower	$10,000 on day one, over three times the study's entire life-of-loan figure

If you want to do it yourself, this is the bar

Plenty of people run their own transaction and it works out. That is a legitimate choice and nothing here is an argument that you cannot handle it. But if you are going to do it, do the part that actually pays, and hold whoever is helping you to the same standard we hold ourselves to. This is what we expect on every file, and it is what you should expect from anybody, us included.

• The lender reads the contract before it is signed, and says what they would change while the terms are still moving. Reading it after the fact is not the same job.
• Every seller concession gets a decision, not a default. You should see what each use of that money does to the payment, side by side, before anybody picks one. If the answer is closing costs and no discussion, that is a miss.
• You are told which program you are in and why, against the ones you are not in, and what the mortgage insurance is doing in year six.
• The closing timeline is real and goes in writing. A date the lender will stand behind in the offer is a negotiating asset. A date nobody will commit to is marketing.
• Your agent finds out what this seller wants besides money, and the lender answers whether the file can hand it to them. That exchange happens before the offer goes out, not after it is countered.
• Compare prices if you want to. Use Loan Estimates, on the same day, since rates move and the study holds the day constant for that reason. It takes an hour and it is the smallest item on this list. Do it last.

Our stake in this, stated plainly

We are a mortgage broker. If you conclude that structure matters and that it takes a lender working with your agent rather than an auction, we are one of the people you might call. Weigh that the way you would weigh any argument from somebody with an interest in the answer.

So here is the version that does not require trusting us. The Freddie Mac research is published and the methodology is inside it. It simulates random draws across similar borrower profiles, holding the loan identical. Read it and decide whether that is the question you are trying to answer. If it is, the advice costs you nothing. If it is not, you now know what nobody told you to ask.

Freddie Mac, April 2018 Insight · Freddie Mac, 2023 follow-up on shopping when rates are higher