Mortgage rate predictions: who publishes them, how they are built, and how often they miss
Every serious rate forecast carries a published error bar, and it is wider than most borrowers expect. In the Federal Reserve’s Summary of Economic Projections released June 17, 2026, Table 2 puts the average historical error range on short-term interest rate forecasts at ±0.7 percentage points in the current year, ±1.8 points in the second year and ±2.3 points in the third, measured across private and government forecasters from 2006 through 2025.

That is the Fed grading the whole profession, itself included. A rate prediction is a conditional statement about inflation, Federal Reserve policy and bond demand, not a schedule. Align Lending does not publish rate forecasts, and no licensed broker should tell you where rates are going. Learn who publishes mortgage rate predictions, what goes into each one, why two credible forecasts disagree, and how accurate the record has been.
Key takeaways:
- Forecast error is measured, not guessed: The Federal Reserve’s June 17, 2026 projection materials put the historical error range on short-term rate forecasts at ±0.7 percentage points one year out and ±1.8 points two years out, at roughly 70% confidence.
- The Fed does not set your mortgage rate: The FOMC cut its target range a full 1.00 percentage point between September 18 and December 18, 2024, while Freddie Mac’s 30-year fixed average rose from 6.09% to 7.04% by the week of January 16, 2025.
- Plan for a band, not a point: From 2000 through 2025 the 30-year fixed weekly average traveled a median 0.98 percentage points between its yearly low and yearly high, nearly a full point inside every calendar year.
Who publishes mortgage rate predictions?
Four institutions produce the numbers nearly every rate headline recycles, and only one actually forecasts a mortgage rate. The rest project the federal funds rate, project broad macroeconomic conditions, or simply measure what lenders charge today.
| Publisher | What it publishes | How often | What it actually projects |
|---|---|---|---|
| Federal Open Market Committee | Summary of Economic Projections | Four times a year (March, June, September and December meetings) | Each participant’s own view of the appropriate federal funds rate path, not a mortgage rate, and not a committee commitment |
| Fannie Mae Economic and Strategic Research Group | Economic Forecast and Housing Forecast | Monthly | A projected 30-year fixed mortgage average, plus home sales, prices and origination volume |
| Federal Reserve Bank of Philadelphia | Survey of Professional Forecasters | Quarterly; survey running since 1968, administered by the Philadelphia Fed since 1990 | Consensus and individual forecasts from outside economists for growth, inflation and interest rates |
| Freddie Mac | Primary Mortgage Market Survey | Weekly, Thursdays at 12 p.m. ET | Nothing. A measurement of current pricing, and the benchmark forecasts are scored against |
The Mortgage Bankers Association and the National Association of Realtors publish industry-side outlooks too. Both are worth reading, but each carries the perspective of the industry publishing it, so compare several rather than adopt one. Our Mortgage Rates & Market guide tracks what these sources say without adding a prediction of our own.
One point deserves emphasis. Freddie Mac’s survey is not a forecast at all, which is exactly why it is useful: it says where pricing sits this week rather than where somebody hopes it sits next year. For the current figure, go straight to Freddie Mac’s weekly Primary Mortgage Market Survey rather than a rate headline written days earlier.
How is a mortgage rate forecast actually built?
Every mortgage rate forecast is a chain of four assumptions, and the mortgage rate is the last link, not the first. A forecaster starts with inflation, converts that into an expected Fed policy path, then a 10-year Treasury yield, then adds a spread to reach a borrower’s quoted rate. Break any link and the number at the end changes.
- Inflation expectations: Investors funding a 30-year loan need compensation for purchasing power they expect to lose, so an upward revision to expected inflation pushes long-term yields up before any policymaker acts.
- The Federal Reserve policy path: The Fed’s tool is the target range for the federal funds rate, which it defines as the rate depository institutions charge each other for overnight loans of reserve balances. Congress directed it to pursue maximum employment, stable prices and moderate long-term interest rates, but long-term rates are a goal it influences, not a dial it turns. Read how the Federal Reserve sets monetary policy in its own words.
- The 10-year Treasury yield: Because most borrowers move or refinance well before year 30, the 30-year mortgage prices off the 10-year Treasury rather than the 30-year bond. This input tracks mortgage rates most closely.
- The mortgage spread: Loans are pooled into mortgage-backed securities and sold to investors, who demand a premium over Treasuries for prepayment and credit risk. That premium moves on its own.
The spread is the link forecasts handle worst, because it is not stable. Across 1,913 weeks pairing Freddie Mac’s 30-year average with the Treasury’s 10-year constant maturity yield from January 1990 through the week of August 27, 2026, the spread averaged 1.77 percentage points, but 1.67 points through 2021 and 2.44 points from 2022 onward, ranging from 0.97 points in March 1996 to 3.26 points in November 2022.1 A forecaster can call the Treasury yield correctly and still miss the mortgage rate by three quarters of a point.
The 2020 and 2021 record lows show what happens when the chain is pulled directly. In its December 16, 2020 statement the FOMC held the federal funds target range at 0 to 0.25% and committed to buying at least $80 billion of Treasury securities and $40 billion of agency mortgage-backed securities every month. A buyer no private forecast can assume. The 30-year fixed average first printed below 3.00% in the week of July 16, 2020, bottomed at 2.65% in the week of January 7, 2021, and came in under 3.00% in 55 separate weeks, not consecutively, between July 2020 and November 2021. Our guide to the Federal Reserve and mortgage rates walks that transmission step by step. The St. Louis Fed publishes the long-run history of the 30-year fixed mortgage average back to 1971.
How accurate have mortgage rate predictions been?
The Fed publishes the honest answer itself: a ±0.7 percentage point error range one year out, ±1.8 points two years out and ±2.3 points three years out on short-term rate forecasts, at roughly 70% confidence, based on twenty years of outcomes through 2025. Those bands cover short-term rates, not mortgage rates, mortgage rates add spread risk on top.
The same document explains why forecast consensus feels reassuring but is not. The Fed notes that the dispersion of projections across FOMC participants is much smaller than the average forecast errors over the past 20 years. Forecasters agree with each other far more than they agree with what happens, so a cluster of similar forecasts shows shared assumptions, not confirmation.
The clearest recent illustration ran from autumn 2024 into winter 2025. The FOMC cut its target range by 0.50 percentage points on September 18, 2024, by 0.25 on November 7 and by 0.25 on December 18, a full point of easing across three meetings. Over the same window, Freddie Mac’s 30-year fixed average went from 6.09% in the week of September 19, 2024 to 7.04% in the week of January 16, 2025. The Fed cut a point; the mortgage rate rose nearly a point. Anyone who timed a purchase around waiting for the Fed got the policy call right and the payment wrong.
Movement inside a single year is the other reason a point forecast ages badly. Across 2000 through 2025, the 30-year fixed weekly average moved a median 0.98 percentage points from its yearly low to its yearly high, 0.50 points in the calmest year, 2015, and 3.86 points in the wildest, 2022.2 A year-end forecast published in January names one point inside a range the rate will travel nearly a full point through. Our page on mortgage rate volatility covers what that swing does to a buyer mid-search.
Why do two credible rate forecasts disagree?
Most apparent disagreements are not disagreements at all. They are four different questions answered separately, then compared as though they were one. Before deciding a forecaster is wrong, check which of these is going on.
- Different vintage: A monthly forecast from March and a quarterly one from May are not competing views; the later one has two more months of inflation and employment data. Check the publication date first.
- Different variable: The FOMC projects the federal funds rate. Fannie Mae projects a 30-year mortgage average. Those are different instruments separated by a Treasury yield and a spread, and they can move in opposite directions for months.
- Different measurement basis: Even the benchmark changed how it counts. Freddie Mac surveyed lenders until November 17, 2022, then switched to rate data drawn from thousands of applications submitted through its Loan Product Advisor.
- Rate versus APR: Under Regulation Z, the annual percentage rate measures the cost of credit as a yearly rate, computed by the actuarial or United States Rule method set out in 12 CFR 1026.22 and Appendix J, and treated as accurate within one eighth of one percentage point on a regular transaction. It is not simply the rate plus your fees.
That last distinction matters most once you start collecting quotes. Disclosures reflect the terms of the legal obligation as of the day they are made, so on an adjustable-rate loan the disclosed APR rests on assumptions about future index values that may not hold. Our explainer on annual percentage rate (APR) shows what is folded into the calculation, and comparing mortgage lenders covers reading two Loan Estimates side by side.
It also helps to hold the full range in view. Freddie Mac has published the 30-year fixed average weekly since April 2, 1971, opening at 7.33%. The highest weekly reading on record is 18.63% in the week of October 9, 1981; the lowest is 2.65% in the week of January 7, 2021. Any forecast is a small claim inside a very wide range.
How should you use a rate forecast when you are actually buying?
Use it to size a range, never to pick a date. The practical question is not where rates are headed but what a half point does to your payment, and that you can calculate today. Here is the arithmetic on a $300,000 loan over 30 years, principal and interest only.3
| Illustrative rate | Monthly principal and interest | Change vs. the row above | Total interest over 30 years |
|---|---|---|---|
| 6.00% | $1,798.65 | n/a | $347,515 |
| 6.50% | $1,896.20 | +$97.55 | $382,633 |
| 7.00% | $1,995.91 | +$99.70 | $418,527 |
Those rates are illustration points a half point apart, not quotes, forecasts or any lender’s pricing. The pattern is what matters: on this loan size every half point is worth roughly $98 to $100 a month, and a full point about $197 a month or roughly $71,000 of interest over the full term. Once you know your own number, a forecast becomes a planning range instead of a verdict.
Hypothetical example, not an offer. Actual rates, costs and payments vary.
- Lock on your contract, not on a forecast: A rate lock protects the price of a specific file for a specific window, so the time to lock is when you have an accepted offer and a closing date the lock period can cover.
- Price the buydown against the break-even: Discount points buy a lower rate for cash at closing. Divide the cost of the points by the monthly savings to get your break-even in months, then compare that to how long you expect to stay.
- Separate the house decision from the rate decision: You can change the rate later through a refinance; you cannot change which houses were for sale this spring. Our guide on whether refinancing is still worth it when rates are rising walks the timing math, and if a lower start rate tempts you, read fixed vs. adjustable-rate mortgage first.
- Remember prices respond too: Falling rates rarely arrive as a pure discount, because they bring more buyers with them. Our breakdown of the impact of interest rates on home prices covers why a lower rate does not mean a cheaper house.
None of this requires knowing where rates go, only your break-even points and competing quotes in front of you the day you decide.
How Align shops your rate
Your rate depends on you and on the lender. The same file can price differently from one wholesale lender to the next, which is why one application with Align goes to multiple lenders.
In late 2022, Freddie Mac found borrowers who got two quotes could have saved as much as $600 a year, and four or more quotes more than $1,200 a year.
Align is paid by the lender, not by a fee you pay us, and federal rules bar a broker's pay from changing with your rate or fees.
Why Align doesn't publish rates
Sources: Freddie Mac, When rates are higher, borrowers who shop around save (February 2023); Regulation Z, 12 CFR 1026.36(d).
FAQ: Mortgage rate predictions
Here are answers to common questions about mortgage rate predictions.
Will mortgage rates go down?
Nobody can tell you, and the published error ranges are the proof. The Fed’s June 2026 projection materials put the historical error on short-term rate forecasts at roughly plus or minus 0.7 percentage points one year out, wide enough to contain both a meaningful decline and a meaningful rise. What forecasters can name is what would have to happen: cooler inflation, an easier Fed policy path, stronger Treasury demand and a narrower mortgage spread.
What is the lowest mortgage rates are expected to go?
There is no floor anyone can responsibly name, but history frames the question. The lowest weekly average Freddie Mac has recorded is 2.65%, in the week of January 7, 2021, alongside a federal funds target range of 0 to 0.25% and Fed purchases of at least $40 billion of agency mortgage-backed securities a month. That was a policy regime, not a market equilibrium. Inflation is the binding constraint on how low long-term rates go, which makes our page on inflation and home prices a more useful read than any floor estimate.
Should I wait for lower rates to buy a house?
It depends on whether you can carry the payment today, not on where rates are headed. Waiting bets on two variables at once: if rates fall, more buyers return and prices tend to firm up, which can erase the savings you waited for. Buying at a payment you can afford and refinancing later keeps the house and leaves the rate decision open. Our breakdown of affordability challenges in today’s housing market lays out what determines whether waiting helps you.
Does the Federal Reserve set mortgage rates?
No. The Fed sets a target range for the federal funds rate, what banks charge each other for overnight loans of reserve balances. Your 30-year mortgage prices off the 10-year Treasury yield plus a mortgage-backed securities spread, both of which respond to inflation expectations and investor demand. That is why mortgage rates often move before a Fed meeting, and why they rose nearly a full point between September 2024 and January 2025 while the Fed was cutting. Our primer on how mortgage rates work traces the path from the bond market to your quote.
The bottom line on mortgage rate predictions
Mortgage rate predictions are conditional scenarios with measured error bars, not schedules. The Federal Reserve’s June 17, 2026 projection materials put the historical error on short-term rate forecasts at ±0.7 percentage points one year out and ±1.8 points two years out, and since 2000 the 30-year fixed average has moved a median 0.98 percentage points inside a single calendar year. Read forecasts to understand what drives rates, check Freddie Mac’s weekly survey for where pricing sits, and decide on the payment in front of you rather than the one in a headline.
If you’re ready to see what today’s pricing actually means for your budget, talk to Align Lending. We’ll shop your scenario across our lender network and show you the rate, the APR and the costs side by side. Call 248-506-5727 or start online.
Footnotes:
1. Spread figures are an Align Lending calculation from public primary data, not a third-party published statistic. The series pairs each weekly Freddie Mac Primary Mortgage Market Survey 30-year fixed average with the U.S. Department of the Treasury daily par yield curve 10-year constant maturity rate for the same date or the nearest prior business day, producing 1,913 matched weeks from January 5, 1990 through August 27, 2026. It is a historical measurement and is not predictive. Return to text.
2. The intra-year swing figure is an Align Lending calculation, not a third-party published statistic. For each calendar year from 2000 through 2025 it takes the difference between the highest and lowest weekly 30-year fixed average reported in the Freddie Mac Primary Mortgage Market Survey rate history file, then reports the median of those 26 annual ranges. It describes past variability only and does not indicate future movement. Return to text.
3. Example figures are hypothetical and for educational purposes only; they do not constitute an advertisement of credit terms or a rate quote under federal or state lending laws. The illustrative rates shown are evenly spaced reference points, not quotes, forecasts or an indication of any lender’s pricing. Payments shown are principal and interest only and exclude property taxes, homeowners insurance, mortgage insurance and any HOA dues. Your rate and terms will depend on your application and market conditions. Return to text.
Sources: average historical projection error ranges of ±0.7, ±1.8 and ±2.3 percentage points on short-term interest rates, and the note that participant dispersion is smaller than average forecast errors, Federal Open Market Committee, Summary of Economic Projections, June 17, 2026, Table 2. Eight scheduled FOMC meetings per year and Summary of Economic Projections at the March, June, September and December meetings, Board of Governors of the Federal Reserve System, FOMC meeting calendars. Definition of the federal funds rate as an overnight interbank rate, and the goal of maximum employment, stable prices and moderate long-term interest rates, Board of Governors of the Federal Reserve System, monetary policy and open market operations pages. Target range of 0 to 0.25% and monthly purchases of at least $80 billion of Treasury securities and $40 billion of agency mortgage-backed securities, FOMC statement, December 16, 2020. Target range reductions of 0.50, 0.25 and 0.25 percentage points, FOMC statements of September 18, 2024, November 7, 2024 and December 18, 2024. Weekly 30-year fixed averages of 7.33% (April 2, 1971), 18.63% (October 9, 1981), 2.65% (January 7, 2021), 6.09% (September 19, 2024) and 7.04% (January 16, 2025), the 55 weekly readings below 3.00% between July 16, 2020 and November 10, 2021, plus weekly publication on Thursdays at 12 p.m. ET and the November 17, 2022 change from a lender survey to Loan Product Advisor application data, Freddie Mac Primary Mortgage Market Survey and its published rate history file, retrieved August 31, 2026. Monthly Economic Forecast and Housing Forecast, Fannie Mae Economic and Strategic Research Group. Quarterly Survey of Professional Forecasters, in continuous operation since 1968 and administered by the Federal Reserve Bank of Philadelphia since 1990. Annual percentage rate definition, the actuarial and United States Rule methods, Appendix J, and the one-eighth of one percentage point accuracy tolerance for regular transactions, 12 CFR 1026.22; disclosures reflecting the terms of the legal obligation, 12 CFR 1026.17(c)(1). Ten-year constant maturity yields, U.S. Department of the Treasury daily par yield curve rates, January 1990 through August 27, 2026.
This article explains how third-party mortgage rate forecasts are produced and how accurate they have historically been. It does not contain a rate forecast, a rate quote or a prediction, and Align Lending does not publish rate forecasts. Figures reflect data available as of August 2026; check the primary sources named above for current readings. Align Lending is an independent Michigan mortgage broker, NMLS #2041154.

