Ask five economists for a housing interest rate forecast and you’ll get seven answers. That isn’t because they’re bad at their jobs. It’s because mortgage rates are set by a market that reprices daily, and any forecast longer than a few months is really a bet about inflation, bond investors, and politics all at once.
Still, a forecast has value if you use it correctly. Not as a promise, but as a set of probabilities you can plan around. Here’s what the next 18 to 24 months most plausibly look like for mortgage rates, what has to be true for those numbers to hold, and where the whole thing could break.
The Fed doesn’t set your mortgage rate, and that trips people up
The single most common misunderstanding in this space is that the Federal Reserve controls mortgage rates. It doesn’t. The Fed sets the overnight rate banks charge each other, and that feeds into credit cards, auto loans, and home equity lines. A 30-year fixed mortgage is priced off the 10-year Treasury yield, which reflects what bond investors think inflation and growth will look like over the next decade.
The two often move in opposite directions at exactly the wrong moment. In September 2024 the Fed cut rates by half a percentage point, and mortgage rates climbed over the following weeks because bond markets had already priced the cut in and were reacting to stronger economic data. That kind of divergence is why a closer look at what actually moves mortgage rates is more useful than tracking Fed meeting dates on your calendar.
Three inputs do most of the work
- The 10-year Treasury yield. This is the anchor. When it moves 0.25%, mortgage rates typically follow within days.
- Inflation expectations. Not just the latest CPI print, but what bond buyers think inflation averages over ten years. Sticky services inflation keeps this elevated and puts a floor under rates.
- The spread between Treasuries and mortgages. Historically about 1.7 percentage points, this gap widened to over 2.5 points in 2023 and has been slow to normalise. Every 0.1% of spread tightening is worth roughly 0.1% off your rate for free, with no change in Treasuries at all.
What the baseline forecast looks like right now
Strip out the noise and most credible forecasts cluster in a similar place for the next two years, assuming no recession and no inflation shock:
- 30-year conforming fixed: mostly 6.0% to 6.75%, with occasional dips toward 5.875% on weak jobs data.
- 30-year FHA: roughly 5.75% to 6.4%, sitting below conventional because government loans carry different pricing and mortgage insurance instead of a rate premium.
- Jumbo: 6.25% to 7.0%, wider than usual because bank balance sheets are tighter than they were pre-2022.
- 15-year fixed: 5.375% to 6.0%, consistently 0.5 to 0.75 points under the 30-year.
That baseline assumes core inflation drifts toward 2.5%, the 10-year Treasury settles between 3.9% and 4.4%, and the mortgage spread narrows modestly as volatility fades. It is not a prediction of smooth sailing. It’s a centre of gravity, and rates will oscillate around it.
Three scenarios, and what pushes rates into each
The base case: slow grind lower
Inflation cools gradually, the Fed cuts two to four more times over a year, and the spread narrows as bond volatility subsides. Mortgage rates drift down 0.5 to 0.75 points over 12 to 18 months, in a jagged line rather than a straight one. This is where most forecasters sit, and it’s the scenario where patience pays.
The higher case: rates stay put or creep up
Tariffs, energy prices, or a tight labour market keep inflation above 3%, and the 10-year pushes back toward 4.75%. Mortgage rates retest the mid-7s. This scenario gets underweighted in most public forecasts because it isn’t fun to write about, but it happened for most of 2022 through 2024.
The lower case: a genuine break
A weakening jobs market forces faster cuts and bond investors front-run them. The 10-year falls under 3.75% and the spread compresses at the same time. Rates touch the high 5s within a year. It’s possible, and it usually arrives faster than anyone expects, but it typically comes with a softer housing market and less competition for the house you want. Understanding which forces genuinely drive mortgage rate drops helps you tell the difference between a real shift and a two-week blip.
The specific things that have to happen before rates fall meaningfully
Vague optimism about rate cuts is not a strategy. Three conditions need to line up, roughly in this order: inflation prints that consistently land at or below expectations, a labour market that loosens without collapsing, and a mortgage spread that narrows as MBS volatility calms. Miss one and rates stall. The mechanics are spelled out in more detail in what has to happen before mortgage rates go down, and it’s worth reading before you build a budget around a hoped-for number.
Looking further out, the 2027 picture is even fuzzier. Reasonable people put the 30-year anywhere from the high 5s to the low 7s by then, which is a wide enough band to be almost useless as a decision tool. The 2027 rate forecast is honestly a range, not a number, and treating it as a single figure is how people end up waiting three years for a rate that never arrives.
What this means if you’re actually buying a house
Here’s the part forecasters rarely say plainly: a 0.5% difference in rate on a $400,000 loan is about $120 a month. That matters, but it’s usually smaller than the swing in asking prices and competition you’ll face by waiting a year. If you find the right house and the payment works at today’s rate, waiting for 5.9% is a bet that costs you real money in rent and missed equity.
The exception is if you’re stretching to qualify. Then the loan type matters as much as the rate. FHA pricing behaves differently from conventional, and the advertised headline number often hides mortgage insurance and upfront costs that change the real comparison, which is why it’s worth digging into what a 30-year FHA rate quote leaves out before you assume it’s the cheaper option.
If you already own a home
Run the refinance math on break-even, not on the headline rate. If you’re at 7.25% and could get 6.25%, you save roughly $260 a month on a $400,000 balance. If closing costs run $4,500, that’s about 17 months to break even. If you plan to move before then, the refinance probably loses money. If you’re staying put for five years, it’s a straightforward win. The trigger point for most people is a full point below their current rate, plus a plan to stay at least two years.
The four numbers worth checking every month
Skip the daily rate headlines. These four data points tell you which scenario you’re drifting into, and they’re all published on a schedule you can set a reminder for:
- Core PCE, released monthly. The Fed’s preferred inflation gauge. Three consecutive prints at or below 0.2% month-over-month is a real signal.
- The 10-year Treasury yield, weekly. A sustained move below 4% is the precursor to mortgage rates starting with a 5.
- The spread between the 30-year mortgage and the 10-year. Track it roughly by subtracting one from the other. Anything under 2.0 points means lenders are competing again.
- Weekly jobless claims. A four-week average trending above 250,000 historically precedes faster rate cuts.
The question that beats any forecast
Instead of asking when rates will fall, ask what rate has to be true for your plan to still work. If the answer is 7.5%, you’re insulated from almost anything the market does and you should buy when you find the house. If the answer is 5.75%, you’re not really buying a house, you’re betting on a forecast, and that’s a much riskier position than it feels like from the outside.
Build in a cushion. Keep the payment under 30% of gross income at today’s rate, not at the rate you hope to refinance into. Get a written loan estimate from at least two lenders, because the gap between the best and worst quote on the same day is routinely 0.5% or more for identical borrowers. And if rates do drop a full point within two years, treat the refinance as a bonus rather than the foundation of the plan. That way the forecast becomes what it should be: background noise you’ve already accounted for.