Improving Property Tax Assessing with Machine Learning
Jul 01, 2019
For many local, city, and county governments, property taxes are primary source of revenue. For homeowners, property taxes can be one of the biggest expenses. Accurate assessments of property values are key to guaranteeing that property owners are paying their fair share and benefit both taxpayers and their community. However, keeping track of changing property values can put a strain on already overstressed property assessor’s offices. Sudden jumps in value, after an appraisal in a booming market, can be a shock to homeowners. Local government property assessors are struggling to be more transparent and reactive to taxpayer’s concerns. When Wake County, in North Carolina, decided to speed up their assessment cycle and do new assessments every 4 years as opposed to every 8, the county’s assessment office felt the squeeze. They couldn’t hire and train enough new staff to fit the new timeline. Instead, the county turned to machine learning to help lighten the load.
Machine learning allowed Wake County to build a statistical model to estimate the value of a homeowner’s property. The model looks at over 100 variables and learns which variables have the most impact on the price of a home, for example a finished basement adds value, unlike a second fireplace. The model also finds similar homes across neighborhoods, helping generate more accurate estimates for houses, even in areas where not many properties are being sold. Currently, there is a lot of room for subjectivity and bias in property assessments. With the introduction of the model, assessors still make the final judgements on property values, but the model allows them to make faster and more data driven decisions. Assessors can compare their initial analysis to the data and let it inform their final valuations.
The data driven approach to property assessment should be beneficial to both taxpayers and the county. Taxpayers get more regular and accurate assessments. The county is hoping that the process will produce fewer appeals from taxpayers who feel their property has been improperly assessed. Appeals are a time consuming and labor-intensive process for the assessment office. Though some homeowners will always try to appeal their initial assessment, producing accurate valuations both cuts down on valid appeals and increases public trust in the work of the office. Wake County’s appraisals will start using the model for their assessments for 2020, but in the longer term, machine learning could help speed up the appeals process as well, helping sort out legitimate complaints and recommending next steps. By leveraging data analytics and machine learning, Wake County’s assessment office has made their process more efficient, saving time and money for both their employees and their constituents.
How can machine learning help your organization work more efficiently? Contact ASR Analytics and allow our team to help you harness the power of machine learning.