Developers, real-estate investors and homeowners can plan for known costs, but they cannot plan for uncertainty. In St. Louis County, Mo., where I serve on a local school board, our recent reassessment cycle saw valuations jump sharply. While much of this increase was caused by market inflation, the public outcry was made worse by a failure to effectively communicate the rationale behind the numbers.
When a “black box” algorithm says a home is worth 20 percent more without providing adequate context, it creates sticker shock that degrades trust in the competency of local government. But the problem goes deeper than bad PR. Dated assessment systems aren’t just lacking in transparency; they are often structurally biased.
As jurisdictions nationwide face similar challenges, public administrators must move beyond the era of opaque legacy systems. Local officials need a procurement mandate for “glass box” assessment tools — systems that are transparent as to how they come to their conclusions about property value. This is particularly important as assessment systems increasingly employ artificial intelligence.
Outdated mass appraisal systems suffer from a statistical flaw known as “over-smoothing,” or trying to pull properties toward the average. This works well enough for a subdivision of identical homes. But across mixed-income neighborhoods it creates an invisible tax shift.
The legacy black box system effectively says, “I don’t have enough data to prove this distressed home is only worth $30,000, so I’ll value it closer to the county average.” The result? The $30,000 home gets assessed at $60,000. Meanwhile, a house with a higher market value gets pulled down toward the average.
This distortion is frequently compounded by a reliance on outdated “cost approach” models that default to calculating replacement costs when sales data is thin. These systems see a house that would cost $100,000 to replace but miss the reality that it sits in a disinvested neighborhood where it would sell for only a fraction of that price.
This inequity is further cemented by the appeals process. Wealthier homeowners are more likely to hire representation to appeal their assessments, and win. The legacy system then “learns” from this new, lower data point. Meanwhile, lower-income homeowners appeal less frequently, so the system “learns” that their inflated value is correct. Year after year, this “data rot” compounds. The way to break this cycle is by moving to a glass box model that invites the taxpayer to correct the data before it becomes a permanent record.
The core of the glass box model — otherwise known as explainable AI — is simple: The system must “show its work.” In a glass box model, the assessment office presents a clear narrative: “We rejected comparable property A because it lacks a finished basement. We selected comp B because, like your property, it is along a busy road.”
This transparency can transform local government’s relationship with the business community as well. Commercial property owners often receive assessments based on generic assumptions that bear no relation to their actual net operating income. A glass box system explicitly details the capitalization rates and vacancy assumptions used, transforming an adversarial dispute into a data-enlightened conversation.
This shift isn’t just about fairness; it is a matter of aligning with the regulatory standard. As of October 2025, new federal rules require lenders to rigorously test automated valuation models for discrimination.
If the private sector is now held to this standard, the public sector should hold itself to it as well. As banks abandon opaque models to comply with federal law, local assessment offices that rely on dated black box vendors will find themselves increasingly out of step — and legally vulnerable to challenges under the Fair Housing Act.
Another practical, high-impact innovation that more local governments should employ is the “pre-appeal” portal. This tool treats taxpayers as partners in solving the dirty-data problem. It allows property owners to upload interior photos — showing, for instance, the deferred maintenance that an exterior-only inspection could not have seen — before the formal appeal deadline. Intelligent processing tools can streamline this review, allowing assessors to fix the condition rating in the database without a monthslong hearing, significantly reducing the administrative burden on staff.
For city councils and county commissioners, the path forward is clear. Do not give your assessment office a blank check for technology. Rather, use the power of procurement to establish a targeted mandate: We will fund best-in-class tools, but only if they explain themselves to our taxpayers.
Trust in government is not built through grand promises but through pragmatic, incremental improvements in service. If local governments want to regain public trust, stabilize revenue and stop the invisible tax shift onto their most vulnerable residents, they must upgrade to a system that is as transparent as it is precise.
Peter Gariepy is a CPA serving on the Ladue Schools Board of Education in St. Louis County, Mo.
Governing's opinion columns reflect the views of their authors and not necessarily those of Governing's editors or management.
