Better mortgage spreads were a positive storyline in 2026, keeping mortgage rates under 7% all year until this last week. 2026 had the lowest rate curve in many years, but an escalation of the Iran conflict pushed the 10-year yield higher and closer to 5% last week. Mortgage rates, which were as low as 5.99% at one point this year, ended the week at 7.12%.
As we’ve stressed over the past few years, housing data tends to improve when rates fall below 6.64% and move toward 6%, but demand tends to fade when rates rise above 6.64% and move above 7%. Now, we are also on the verge of a Fed rate-hike cycle, partly because this conflict has lasted so long.
Mortgage spreads
Last week, we saw mortgage spreads take their last stand to keep rates under 7%, like King Leonidas in the movie 300, giving a loud roar. The 10-year yield has been rising as the Iran conflict has worsened entering its sixth month, and when President Trump said the conflict might not end until after the midterms, yields almost closed at cycle highs.
The one positive here is that things could have been a lot worse. This week shows the widest gap between where mortgage rates could have been — given where the 10-year yield is — to where they are today, that I have seen since we put the spreads variable in the tracker articles.
Historically, mortgage spreads have ranged from 1.60% to 1.80%. Last week, spreads were at 1.92%, down from 1.94% the week before.
Let’s compare last week’s mortgage rates to where they would have been over the last three years, given the 10-year yield’s current level:
- If we had the worst mortgage spread levels of 2023, mortgage rates would be 8.31% today, not 7.12%.
- If we had the worst levels of 2024, mortgage rates would be 7.94% today.
- If we had the worst levels of 2025, mortgage rates would be 7.74% today.
10-year yield and mortgage rates
In the 2026 HousingWire forecast, I anticipated the following ranges:
- Mortgage rates between 5.75% and 6.75%
- The 10-year yield fluctuating between 3.80% and 4.60%
Last week was probably the most important week of the year, when the bond market basically said: the higher oil prices go, the higher the 10-year yield goes. The bond market just ran right over Scott Bessent’s “I am the House” plan to buy back more long-term debt because the conflict got worse. I talked about how much higher mortgage rates can go in this episode of the HousingWire Daily podcast.
Earlier in the year, when I wrote about how bad mortgage rates could get, I said I could add 0.375%-0.43% to the top of my forecast, which would mean rates at 7.13%-7.18%, and we ended the week at 7.12%. This was all predicated on the 10-year yield not going above 5%.
With markets pricing in a rate hike next week, much is already priced in, but in reality, this conflict is running the show now, as the bond market and oil price correlation is very high.
Note: This week’s tracker got heavily impacted by the labor data weekend on all aspects, so expect a rebound in these numbers, which is very normal. After next week, we are back to our normal channel.
Housing inventory
In a normal setting, weekly inventory would have risen, but last week the holiday impact drove inventory lower, and the year-over-year data was also impacted, as the same week last year had the second-week recovery from the holiday. Inventory growth hasn’t been much in 2026, but as rates go higher, inventory will have legs to grow, as long as new listings data doesn’t tumble more than the trend.
Also, be mindful that year-over-year comps will facilitate better inventory growth, as last year rates at this time were falling and demand picked up.
- Weekly inventory change (Sept. 4–Sept. 11): Inventory fell from 883,683 to 873,978
- Same week last year (Sept. 5 -Sept. 12 ): Inventory rose from 846,529 to 860,233
New Listings
New listings are in their typical seasonal decline, but the bigger-than-normal decline is holiday-related, so look for a bounce back next week. 2026 has been the healthiest year for new listings data since the 2022 sales crash.
Normally, new listings range between 80,000 and 100,000 every week during peak periods. For context, during the housing bubble years, new listings ranged from 250,000 to 400,000 per week for several years.
Here is last week’s new listings data for the past two years:
- 2026: 58,803
- 2025: 64,444
Price-cut percentage
Typically, about one-third of homes see price reductions before they sell, reflecting the housing market’s dynamic nature. Overall, price-cut percentages this year have been lower than last year. Because mortgage rates have risen versus last year, the price-cut percentage should be higher than last year, especially now that rates are above 7%. However, we saw a week-to-week decline because of the Labor Day holiday.
In my 2026 home-price forecast, I called for a national decline of 0.62% for the year. Home-price growth really isn’t going anywhere this year, and my forecast of -0.62% might be hard to achieve, as most home price indexes show price growth between 1% and 2%. However, with rates rising again, I might be right in 2026.
The price-cut percentage for last week:
- 2026: 42.08%
- 2025: 42%
Weekly pending sales
Our pending home sales data provides a week-to-week perspective, though holidays and short-term fluctuations can affect results. This weekly pending sales data typically takes 30-60 days to be reflected in the sales data.
Now that rates have been above 6.64% for some time, we have seen a slowdown in pending sales, and now rates are above 7%. This week was also impacted by the holidays, but without them, sales would have been slower.
We also have to remember that the year-over-year comps here will be very difficult, as rates were falling last year at this time and demand was improving. In two weeks, we will have a better gauge of how much damage rates above 7% are doing.
Here are the pending sales for last week over the last two years:
- 2026: 56,255
- 2025: 62,185
Purchase applications
Purchase application data, which looks out 30-90 days, has shown softness as mortgage rates have risen above 6.64% and now rates are above 7%. Since we are working with higher comps, this area should see some year-over-year weakness, especially now that the year-over-year comps will be harder. We didn’t see that softness this last week, but keep an eye out here.
Here are the stats on purchase apps so far in 2026:
- 15 positive week-to-week prints
- 17 negative week-to-week prints
- 5 flat week-to-week prints
- 10 weeks of double-digit year-over-year growth
- 25 weeks of positive year-over-year growth
- 7 negative year-over-year prints
The week ahead: Iran, the Fed, retail sales and housing data
Despite the fact that it’s Fed week, the Iran conflict is still more important to the bond market and mortgage rates than the Fed. The market has already priced in a rate hike, so the surprise would be no hike now. We will also get retail sales and housing data this week, which can’t be looking good now. Again, I stress that until this conflict calms down, the Fed will play second fiddle regarding the bond market.
