2019 Survey on the Use of Automated Valuation Models (AVMs) in Government Assessment Offices

Accurate and consistent assessments are critical components of an equitable property tax system. Automated valuation models (AVMs) are programmed statistical and mathematical models used to estimate the market value of a property. AVMs have been demonstrated to help governments promote more equitable ad valorem property tax assessments at reduced costs and higher efficiencies than do traditional appraisal methodologies. To gain insight and provide industry support to government employees tasked with valuing property for taxation purposes, the Lincoln Institute of Land Policy, in partnership with the International Association of Assessing Officers, facilitated a survey in 2019 among 819 individuals working in at least 400 government assessment offices to learn whether and how AVMs are being used in the assessment process. Responses of employees from assessment offices using regression-based AVMs demonstrate widespread benefits of AVMs to the assessment process.

These AVMs are largely reported to offer increased accuracy, uniformity, and equity compared to non-regression-based valuation approaches, at lower costs and higher efficiencies. The adoption of AVMs has been much more common among larger jurisdictions, with respect to both parcel count and staff size, as well as jurisdictions with more frequent valuation cycles. Still, with only 16 percent of respondents reporting regression-based AVM use, there exists a great opportunity for assessment offices to benefit from their implementation. The most common reason for not adopting AVMs is the belief that a current valuation approach works just as well, if not better. Additional challenges to AVM adoption include difficulty explaining results to stakeholders (e.g., taxpayers, staff of assessment offices and other government agencies) and a lack of both budget resources and technical expertise to implement and maintain AVMs.

To overcome these concerns, research and educational efforts should focus on increasing the overall transparency of AVMs, including explainability and interpretability. Courses, workshops, and conference sessions should teach assessment staff how to create, test, and maintain AVMs using free open-source software (e.g., R, Python). Greater familiarity with AVM technology will lead to more widespread use, ultimately resulting in greater efficiency and equity in municipal property tax regimes.