What the rate years did to North Fulton: the lock-in, the drought, and the twelve months ahead
Four years of rate swings changed what moves in Alpharetta, Roswell and Milton. A look at the Freddie Mac curve, the inventory drought and who can still sell.
Noted by Dana Whitaker, editor, towns field bookverified by Marc Ellison6 min read2 sources
If you are buying in Roswell this month, selling in Milton, or working out whether Johns Creek is worth the premium over Cumming, the number that decides your year is not printed on any listing. It is the one the Freddie Mac weekly survey puts out on Thursday afternoons, and over four years it has rearranged this part of the map more than any zoning vote. The journal walks these streets: Crabapple Road on a Tuesday, the stretch of Highway 9 where the car lots give way to pasture. This is not a crash and not a boom. It is one market split in two by the year a household signed its note.
The arc, in four readings
The Freddie Mac Primary Mortgage Market Survey is the longest continuous record of what a mortgage costs, and it tells the last four years in four numbers. In January 2021 it put the 30-year fixed at 2.65 percent, the lowest reading in the survey's history to that point. It crossed seven percent in the autumn of 2022, the first time in two decades. In late October 2023 it touched 7.79 percent, the highest since 2000. It has not been back to the threes since 2021. A two-year window when money was nearly free, and three years since when it was not. Everything below follows from that.
What does a rate actually do to a payment?
Run the arithmetic on your own numbers and it stops being abstract. On a $500,000 mortgage, principal and interest at three percent comes to roughly $2,100 a month. At seven percent the same loan runs about $3,330. The house did not change. About twelve hundred dollars a month changed hands between the buyer and the lender, and that is the whole distance between the four-bedroom on Bethany Bend and the three-bedroom townhome off Windward Parkway. Buyers here were not priced out of North Fulton so much as priced down inside it: same zip code, less house, longer walk to the mailbox.
Why won't the house down the street come to market?
Because the loan cannot travel with them. A household that refinanced in 2021 is carrying a note in the low threes, and the same money borrowed today costs more than twice the interest. Selling on Rucker Road and buying the identical house two exits north raises the payment by a third even if the price never moves. So they stay. They finish the basement, they put up the fence, they screen the porch. The lock-in effect gets described as a national statistic, but in Milton and Roswell it is a specific fence and a specific porch. The journal cannot count the households that stayed, because staying leaves no record at the courthouse. What leaves a record is the other side: the estate, the relocation, the job in Charlotte. Those houses come to market on life's schedule, not the market's.
What do the inventory reports actually show?
First Multiple Listing Service publishes the monthly counts, and they are worth reading, provided you read them at the right scale. Months of inventory is the figure to ask for: how long it would take to sell everything listed at the current pace. Across the northern Fulton submarkets in 2021 and 2022 that number sat under a month, which is not a market so much as an auction. It has widened since, unevenly, and not back to where it stood in 2019. The metro figure is the wrong one to use, because metro Atlanta includes places that are not Milton, and these towns move more like each other than like the average. The FMLS reports through the spring and summer of 2026 do not agree with one another month to month, which is what a sideways market looks like in a monthly series.
Who bought anyway, and what did they give up?
Plenty of people. They bought differently, and three mechanisms did the work a lower rate used to do. The first is the seller concession: in the Georgia Association of Realtors contract, the seller's contribution to the buyer's costs is a line item with a number in it, and for three years that number has carried more than the asking price. Two points of a buy-down changes a payment every month, and the seller pays it out of proceeds without touching the list price. The second is the assumable loan: a 2021 VA mortgage in the low threes can be taken over by a qualified buyer, which gives that listing two prices, the house and the note. The third is new construction, where the builder has margin to spend. Drive up 400 into Forsyth or 575 into Cherokee and the signs advertise the rate, not the square footage.
What does the county record show that the listing does not?
Georgia is a non-disclosure state. The deed recorded at the Fulton County courthouse does not state what the house sold for. It does not have to. The transfer tax is a dollar per thousand, and the stamp on the deed is public: divide it and you have the price to the nearest thousand. When a listing claims a number above asking, the stamp is how you check it. The same courthouse holds the number that matters to an owner: the assessment. Georgia assesses at forty percent of fair market value, the notice arrives in the spring, and the appeal window runs forty-five days from the date printed on the notice, not from the afternoon it reached the mailbox. If your value jumped and your neighbor's did not, that window is where the argument happens.
So what happens in the next twelve months?
Nobody at this journal knows where the survey goes, and anyone who says otherwise is selling something. What can be said is what does not depend on the forecast. The lock-in thaws on its own clock: every year a slice of those three percent notes retires, not through refinancing but through estates, divorces and job moves, each of which puts a house on the market anyway. The price bands are moving separately too. In most of these towns the scarce house is the one under the median; the plentiful one has five acres and a long driveway. And the towns are diverging faster than the market is: Roswell prices like Roswell, Cumming like Cumming, and the metro average describes neither.
What the journal would do this week
Pick one street. Not a town, a street, ideally the one you would live on. Pull the last three deeds from the county's online records and read the transfer tax stamps: you will know what those houses truly sold for in a state that does not publish it. Then ask a lender for a written Loan Estimate, not a rate quoted over the phone, on the house you would really buy, and set it beside the payment you make now. That is the market on one page, and it will tell you more than any forecast, this one included. It walks, it reads the record, and it writes down what it finds.
