Appraisers’ case for local judgment, and what federal data say about itAppraisers’ case for local judgment, and what federal data say about it

President Trump’s October 22 housing order makes it federal policy to “embrace innovative risk and value assessment technologies that incorporate Super Intelligence,” the administration’s term for artificial intelligence. It also sets up a test in which an automated valuation model, or AVM, can establish the equity needed to cancel mortgage insurance, and it tells the Federal Housing Finance Agency (FHFA) to consider dropping appraisals on refinances that lower a borrower’s payment. Our news story on the order covers each section.

The order is the latest of several steps by the White House, FHFA, Fannie Mae and lenders toward automated valuation. Over the same years, appraisers have given hybrid appraisals low marks, ranked waivers and AVMs last among ways to speed up closings, and written cautious guidance on AI. Both records are below.

The National Association of Realtors’ October 2025 survey of appraisers gives the clearest current numbers. Appraisers rated their comfort with traditional appraisals 4.56 out of 5 and with hybrids 2.07. Asked to rank ways to cut transaction delays, they put AVMs, waivers and data collectors last of ten options.

The survey does not ask about AI, and we found no current survey that measures how many appraisers use mobile inspection apps, laser measures or 3D scanning. Trade group statements on AI accept the tools with conditions: the American Society of Appraisers calls AI “unavoidable” and tells appraisers to verify and proofread its output, and Advisory Opinion 41 under the Uniform Standards of Professional Appraisal Practice keeps the appraiser accountable for what a tool produces.

The lending side has been slow in places too. Desktop appraisals stayed under 1% of Fannie Mae and Freddie Mac volume, and one large lender kept ordering full appraisals because its systems could not order a property data collection. The two main reasons appraisers gave in 2020 for refusing split assignments, liability and fees, are set by lenders, appraisal management companies and insurers.

Fannie Mae and Freddie Mac already route part of their volume around appraisers. About half their loans had an appraisal waiver in March 2021, and 19% did in June 2026. Freddie Mac’s research found its waiver loans had about 8.9% lower 90-day default rates than comparable appraised loans, though most were refinances. Where an appraisal is required, Fannie Mae’s June 2026 Quality Insider says “Appraisal-related issues made up four of the top ten findings.”

Trust in the product also depends on how poor work is policed. States reported 755 disciplinary actions against appraisers in 2025, about 0.8 per 100 credentials. When FHFA found appraisals that openly referred to race or other factors fair lending law prohibits, it filed no complaints with state appraiser boards; it has agreed with its inspector general that it should. The evidence board below sorts the figures for both sides by type.

Lenders, Fannie Mae, Freddie Mac and their regulators are adopting AI, AVMs and new ways of measuring markets and value to manage collateral risk. Appraisers argue that their local judgment is what those tools cannot replace. We test that argument against the profession’s own standard, the Uniform Standards of Professional Appraisal Practice (USPAP), which opens with this:

USPAP does not define “public trust,” and appraisers have argued for years about how far it reaches. The Preamble does say whom the standards serve: the Appraisal Standards Board issues USPAP “for both appraisers and users of appraisal services,” and appraisers must communicate clearly to “intended users.” On a loan sold to Fannie Mae or Freddie Mac, the lender is the client and intended user, and its credit and collateral risk staff are the people who read the report. Fannie Mae’s Selling Guide adds that in certification 23 of the report, the appraiser accepts that other parties, including “mortgage insurers, government-sponsored enterprises, and other secondary market participants,” may rely on the report even though they are not intended users.

A collateral risk analyst reading the report has to decide whether its market analysis and highest and best use analysis are good enough to judge the loan by. The evidence below suggests they often are not, and lenders have responded by building their own tools. That leaves the profession a gap to close: judgment is appraisers’ strongest argument, yet the parts of the report that carry that judgment are often incomplete or out of step with the market.

The push for automated valuation comes from the institutions that order and buy appraisals. On May 29, 2026, eight trade groups, including the Mortgage Bankers Association, the American Bankers Association and the Independent Community Bankers of America, wrote to FHFA Director Bill Pulte that the president’s March order on mortgage credit “presents an opportunity for appraisal modernization.” They asked for more use of AVMs, desktop and hybrid appraisals and AI valuation tools, and fewer appraisal requirements on low-risk loans. In its October roadmap, the Mortgage Bankers Association also called to “Expand Responsible Use of Automated Valuation Models and AI-assisted Underwriting” and proposed a Fannie Mae and Freddie Mac refinance with no appraisal. The October 22 order took up the refinance idea as something FHFA must consider.

Inside the profession, the debate is still over whether remote and app-based methods are acceptable at all. In the Realtors’ 2022 survey, 63% of appraisers said they were “very uncomfortable” with desktop appraisals and other remote methods, and the most specific estimate of how many appraisers use mobile inspection apps is in the single digits. When lenders want faster and cheaper options that appraisers do not offer, the work goes to AVMs and data collectors. Fannie Mae counts 5.34 million loans with appraisal alternatives since 2018.

In a 2020 survey, 75% of appraisers said they would refuse work that split the inspection from the analysis, and the Appraisal Institute asked FHFA in 2025 to reconsider expanded waiver and hybrid eligibility. The objections have real grounds, chiefly liability for another person’s data, fees and control of the inspection. Each of those can be handled in contracts and errors and omissions insurance. If appraisers do not help shape the products, lenders and appraisal management companies will set the terms, or the loans will go to AVMs.

The redesigned appraisal report, known as UAD 3.6 and required for new Fannie Mae and Freddie Mac appraisals from November 2, 2026, asks for more structure: it replaces many checkboxes and free-text boxes with structured fields, and its market section asks for counts of active listings, pending sales and closed sales. Our UAD 3.6 series covers the change field by field.

Every appraisal for a Fannie Mae or Freddie Mac loan has a checkbox for whether one-unit home values in the neighborhood are increasing, stable or declining. Fannie’s Selling Guide says the conclusion must include “factual data from information sources such as, but not limited to, market data, home price indices” and that the trend “must reflect the overall movement of the market.” FHFA publishes how often each box is checked, and its own House Price Index shows what prices did.

From 2013 through 2020, the index rose between 4.3% and 6.3% a year nationally, and appraisals marked values stable 78.7% to 84.5% of the time. In 2021, when the index rose 18%, 52.5% of appraisals still marked values stable.

The checkbox describes the neighborhood around the home being appraised, and a national or metro index can rise while some neighborhoods stay flat. These comparisons show a pattern across millions of reports; they cannot show that any single report was wrong.

The national index never fell year over year in this period, so the clearer test is in metro areas. We matched FHFA’s metro appraisal files to its metro price indexes for 899 metro-years (one metro area in one year) in 97 markets. Where prices rose 15% or more in a year, the median share of appraisals marking values stable was 44.5%. Where they rose 5% to 15%, it was 76.5%.

Where prices fell, stable still dominated. In Austin in the second quarter of 2023, with the index down 9.1% from a year earlier, 81.1% of appraisals marked values stable. By the third quarter, with prices down 9.2%, 89.2% said stable. San Francisco’s index fell 4.0% in 2020, and 92% of appraisals there marked values stable.

The declining box is the one a lender most needs appraisers to get right. A loan made near a price peak is secured by a house that may be worth less by the time a borrower defaults, and the trend call is where the appraisal is supposed to warn of that. In the 23 metro-years where prices fell, FHFA’s files count 396,815 Fannie Mae and Freddie Mac appraisals. Of those, 14,478, or 3.6%, marked values declining. The median metro-year was 1.8%, and in none of them did declining outnumber stable. Nationally, the declining share was under 1% in 35 of 47 quarters.

A market value opinion rests on two analyses: the market analysis and highest and best use. USPAP Standards Rule 1-3 requires an appraiser, when necessary for credible results, to analyze land use regulations and their likely changes, supply and demand, the property’s physical adaptability and market area trends, and to “develop an opinion of highest and best use of the real estate.” Standards Rule 2-2 requires the report to state that opinion and summarize its support.

Minneapolis gave regulators a clean test. On January 1, 2020, the zoning description for the city’s R1, R1A, R2 and R2B districts, which had been single-family and two-family districts, changed to “Multiple-family District.” The Minnesota Department of Commerce had told appraisers and appraisal management companies about the change on December 20, 2019. Commerce then collected 135 appraisals of Minneapolis properties completed after January 1 and checked how they handled the new zoning.

Commerce wrote that the other 132 reports “seemed to stop their analysis at single family residential, which is what their current use was, but did not account for the multiple family uses allowed by the new Minneapolis zoning law.” Commerce allowed that some appraisers may have done the analysis and left it out, and noted that USPAP requires them to summarize it in the report. The state’s Real Estate Appraiser Advisory Board formed a work group on highest and best use in June 2020; the work group presented to the board in December 2020, and the minutes do not record its recommendations. Our highest and best use feature covers what Fannie Mae’s guide asks for.

When a report has no market analysis that tracks the data and no highest and best use analysis, the lender gets a value and a list of comparable sales, but not the reasoning that is supposed to set an appraiser’s opinion apart from an AVM’s estimate. That reasoning is the part of an appraisal a model cannot supply, so it has to be in the report.

A risk manager who cannot rely on the checkbox has to find the answer elsewhere, and Fannie Mae, Freddie Mac and their regulators have built those sources. Fannie Mae stopped requiring its Market Conditions Addendum in August 2018, saying Collateral Underwriter, its automated appraisal review tool, “now provides robust market trend information.” The redesigned report drops the checkbox. Fannie Mae and Freddie Mac deleted the Property Value Trend field from the form’s reference guide in September 2024, so the appraiser now describes the trend in commentary or an attached graph. The new market section’s counts of listings and sales let a reviewer work out months of supply directly. The Mortgage Bankers Association, which supported the waiver that gives lenders and appraisers until May 19, 2027 to adopt the new form, says it “significantly improves the appraisal product by ensuring greater consistency and better readability.”

Fannie Mae also checks appraisers’ condition and quality ratings with image recognition against listing photos; in a test on more than a million appraisals, its reviewers found the model’s flags “98% accurate.” Since October 1, 2025, a rule from six federal regulators has required mortgage originators and secondary market issuers that use AVMs to keep quality controls on them, including “random sample testing and reviews.”

In each case, a lender, Fannie Mae, Freddie Mac or a regulator is building its own check on information the appraisal is supposed to supply. Those checks cost money. Once they exist, the appraiser’s conclusion becomes one input to a model-based risk decision, which is the arrangement the order’s “Super Intelligence” policy describes. An appraisal whose market call matched the price data would carry more weight in that decision.

Part of what makes a profession’s work reliable is that poor work gets corrected, and the public record shows little enforcement. The Appraisal Subcommittee, the federal body that oversees state appraiser regulators, says states “must process complaints of appraiser misconduct or wrongdoing in a timely manner” and decide them within a year. In its 2025 reviews it rated 10 of 37 state programs below Good, including New Hampshire and Puerto Rico at Not Satisfactory.

Texas, whose program the subcommittee rated Excellent in 2024, shows how complaints usually end. Through June of its 2026 fiscal year, the Texas Appraiser Licensing and Certification Board resolved 150 complaints and dismissed 97% of them. Consumers filed most of the complaints, and lenders, Fannie Mae and Freddie Mac together filed 19%.

The public record points to four reasons enforcement misses the problems above. First, the system runs on complaints about single reports, while the patterns in this article, such as a stable box in a falling market or a highest and best use analysis that stops at the current use, only show up across many reports. Minnesota’s review is the exception: a regulator pulled reports itself. What followed publicly was a notice to appraisers and an advisory board work group.

Second, the people who see the most reports file few complaints. FHFA’s inspector general found that FHFA filed no complaints with state boards after finding appraisals that openly referred to prohibited factors, and had no policy on doing so. Third, no one tracks outcomes nationally: the subcommittee’s hotline “does not collect data on the resolution of complaints,” and states do not report complaint data to it. Fourth, funding: the ROAD to Housing Act restructured registry fees so that, in the Appraisal Institute’s summary, the subcommittee would have “sufficient resources to oversee state regulatory programs.”

The people who already hold the evidence could address each one. Collateral risk staff who find a defect in a review can file it with the state board, regulators can run data calls like Minnesota’s, and states can publish what happens to complaints. Enforcement that lenders can see would also give them a reason to trust the appraisals that pass it.

Appraisers explain much of this record with pressure and fees. They say lenders, appraisal management companies (AMCs), agents and borrowers push them to hit a value; that a report calling a market declining or a value short of the contract costs them the next order; and that fees have fallen so far that a full analysis does not pay. In the Realtors’ 2025 survey, 48% of appraisers named fee pressure a top business challenge, and when asked how to cut transaction delays, they ranked reducing or eliminating AMCs first. The Government Accountability Office (GAO) reported industry estimates that AMCs take “at least 30 percent” of the fee.

The pressure is older than the AMC model. By the industry estimates GAO collected, AMCs handled “less than half” of appraisal orders before the Home Valuation Code of Conduct took effect in 2009 and 60% to 80% afterward. GAO also reported that in the mid-2000s, some appraisers said “loan officers and mortgage brokers pressured them to overvalue properties to help secure mortgage approvals.”

An October Research survey of appraisers in 2007 found 90% felt pressured to “hit the number,” up from 55% in 2003, when the survey traced the pressure to mortgage brokers, lenders, real estate agents and homeowners. Studies from 1997 to 1999 found appraisers described themselves “more as price validators than as third party evaluators as a result of lender pressure.” And FHFA researchers comparing AMC and non-AMC appraisals from 2012 to 2016 found “no clear evidence of any systematic quality differences.” By these accounts the pressure came from whoever placed the order, before AMCs and since.

USPAP places the responsibility with the appraiser who signs. The appraiser is the licensed professional on the report, and USPAP does not scale its requirements to the fee or the client. Its Scope of Work Rule says “The scope of work must include the research and analyses that are necessary to develop credible assignment results.” Its Ethics Rule bars agreeing to an assignment whose compensation is contingent on “the reporting of a predetermined result” or “a direction in assignment results that favors the cause of the client,” and the appraiser certifies in each report that the assignment was not contingent on either. Standards Rule 2-1 requires every report to “clearly and accurately set forth the appraisal in a manner that will not be misleading.” Federal law also bars anyone with an interest in the transaction from coercing or pressuring an appraiser (15 U.S.C. 1639e), though we found no federal enforcement actions under it in public reporting for 2020 to 2026.

Appraisers bear the cost of pressure they do not control. Under these rules, a fee too low for credible work is a reason to decline the assignment or raise the price, and the market trend call and the highest and best use conclusion remain the appraiser’s own, whoever ordered the report and whatever it paid.

On many loans, the competition is a data product with no appraiser in it. Freddie Mac’s ACE+ PDR pairs its automated collateral evaluation with a property data collector’s visit to the home. Freddie says borrowers “can save on average $400 per transaction compared to loans that use appraisal reports ($200 vs $600)” and that, as of the second quarter of 2025, those loans closed 12 days faster on purchases and 10 days faster on refinances.

Home equity lending has gone further. The Mortgage Bankers Association’s 2024 Home Equity Lending Study, which MBA told us holds its most recent figures, reported that AVMs or desktop valuations covered more than 75% of home equity loan and line of credit originations in 2023, and that AVM use alone “nearly doubled” from 22% of originations in 2020 to 43% in 2023.

The same kinds of tools are sold to appraisers. In 2023 Bradford Technologies announced that Restb.ai’s computer vision would be built into its report quality control to flag “out-of-focus images and other problematic content.” Advisory Opinion 41 sets out how an appraiser can use such tools and stay responsible for the result, and the Appraisal Institute’s 2026 conference program “treats artificial intelligence as a professional competency.” No current survey measures how many residential appraisers use these tools.

The profession’s clients are moving quickly. Fannie Mae has checked condition and quality ratings on more than a million appraisals with image recognition, and since 2018 has pointed to Collateral Underwriter for market trend data. Within four weeks this fall, Executive Order 14434 renamed AI across the executive branch (September 29), Fannie Mae wrote “Super Intelligence” into its rules for how lenders use AI (October 20 or 21), and the housing order made it federal policy for risk and value assessment (October 22). The redesigned appraisal report becomes mandatory on November 2. Over the same years, appraisers rated their comfort with hybrid appraisals 2.07 out of 5, 63% said in 2022 that they were very uncomfortable with remote methods, and the one estimate of mobile inspection adoption is in single digits. Advisory Opinion 41 gave appraisers a standard for using AI in April 2026; we found no measure of how many appraisers use AI.

In our analysis, the profession needs to catch up, and the best way to compete with data products on price and speed is to use the technology they run on. Those products are cheaper and faster because software does the data gathering, measurement and photo review. An appraiser who automates that work can charge closer to their price and spend the saved time on the analyses where an AVM is weakest. The same tools can check for the two gaps in this article before a report goes out, by comparing the market trend call with listing and sales counts and a price index, and by running a zoning lookup that flags a property whose permitted uses have changed and needs a highest and best use analysis. Advisory Opinion 41 leaves the conclusion and the signature with the appraiser. Appraisers who adopt these tools would also have standing to help shape how lenders, Fannie Mae and Freddie Mac use them, through comment letters, pilots and product design.

Appraisers’ case against automated valuation rests on judgment: in the Appraisal Institute’s words, “an appraiser’s expertise, judgment, and ability to assess market nuances.” In a report, that judgment shows up most clearly in the market analysis and in highest and best use. FHFA’s data show that the market trend call seldom reports a decline, and the one state review that checked found a conforming highest and best use analysis in 3 of 135 reports. Meanwhile, lenders, Fannie Mae and Freddie Mac, which appraisers say undervalue judgment, are building their own market, condition and value checks, and in Texas the complaint process dismissed 97% of the complaints it resolved. These are aggregate figures. They do not show how any one appraiser works, and the reports that did call a decline or did analyze the new Minneapolis zoning show that the work can be done.

Pressure and fees explain part of this record, and the lenders, brokers and AMCs who apply that pressure share the responsibility for it. The appraiser’s name is still the one on the report, and the data suggest that the profession’s strongest argument, its judgment, often does not show in the reports lenders read. As long as that is true, the people USPAP says appraisers serve will see the lending side’s own tools and the data products described above as the safer and cheaper way to manage collateral risk, and the order’s policy will keep moving work toward them.

In our analysis, the answer is for the profession to modernize its product and its enforcement together. A modern appraisal uses the same data capture and AI its clients use, keeps pace with a data product on turn time where the assignment allows, and shows the market and highest and best use analysis a model cannot supply. Enforcement that catches and corrects unsupported conclusions is what would let lenders rely on that analysis without rechecking it against their own models. Doing both would meet USPAP’s opening commitment to public trust and keep the appraisal’s advantage over an AVM: an accountable professional who inspects the property, knows the market and can show the analysis behind the value. That takes seven steps:

We asked The Appraisal Foundation, which writes USPAP and the appraiser qualification criteria, for its view of the order and of the steps above. It sent this statement: