RALEIGH – The tax man is getting more high tech in Wake County with Jim Goodnight, the multi-ibillionaire and richest man in North Carolina, providing a lending hand directly through his own efforts as well as that of his software giant SAS.

Wake County’s Revenue Department has turned to Cary-based SAS, a global leader in making sense of data, for help in dealing with the booming real estate market that’s producing a growing number of properties for tax assessment. Making the data challenge more complex, the assessment cycle speeds up to four years from eight with the next one hitting Jan. 1 2020.

The county is now using an analytics program from SAS called SAS Viya to track property sales, incorporate data from more than 20 variables, and determine property values. This information, compiled via emerging technology known as “machine learning,” will augment information compiled by its own staff in compiling assessments for tax purchases.

The program, for examples, provides human assessor predictions about future property sale prices, its color-coded results reflecting green for what believed will be accurate values, yellow for “moderately accurate” and red for properties for those likely deviate “significantly from expected behavior.”

For any property owner, “assessment” is a billfold issue.

Tax rates are already going up $39 per $100,000 assessed valuation of a property due to the increase approved by Wake County or $117 a year for a house valued at $300,000. The increase is expected to produce some $57 million in additional revenue.

Wake’s goal: Fair valuations, not tax revenue

In announcing the deal last week, Wake County and SAS noted that the revenue department needs to reassess more than 400,000 property values each time a property is sold. And more than 3,000 properties are being sold a month, thus generating a tremendous amount of data.

SAS and the county noted that Wake County’s booming population – well over 1 million from around 300,000 in 1980 – has led to a likewise boom in housing construction with 20,000 properties added in the last three years alone. Making the assessment task of some 400,000 properties more demanding is the fact that Wake County is responsible for assessments spread across 15 municipalities – each “unique” in its own ways as well as unincorporated areas.

Adding in the compressed of timelines to four years from eight makes the assessment challenge even larger.

However, for those who might believe the SAS software is being added to generate more revenue, Marcus Kinrade, Wake County’s revenue director, stressed that is not the case.

“This project isn’t about generating property tax revenue,” Kinrade told WRAL TechWire.

“Our goal is to establish the fairest and most accurate assessed property values that we can while working within the current constraints of our staffing and operating budget.

“Without the technology we are utilizing to assist us, we would have to increase our staffing more significantly (if even possible).  The cost of a large staff expansion would far exceed what we have spent on the SAS tools and models.”

SAS image

This image shows predictions of home sale prices by SAS Viya machine learning. Green represents a highly accurate prediction, yellow a moderately accurate prediction, and red a property sale that deviates significantly from expected behavior

Financial terms weren’t disclosed, but the deal is important enough for SAS that Goodnight, the company’s co-founder and CEO, got involved in creating it.

“Under the supervision of Dr. Goodnight, two of our R&D directors led a team of five SAS researchers that explored the property assessment issue,” Jennifer Robinson, SAS Director of Local Government Solutions, explained. “Over a course of five months in late 2017-early 2018, they developed an automated machine-learning system for determining home values.”

At the core of the SAS technology selected by Wake County is machine learning – a big buzz word in emerging technology as is artificial intelligence. Machine learning, explains Wikipedia, is a “subset of artificial intelligence in the field of computer science that often uses statistical techniques to give computers the ability to “learn” (i.e., progressively improve performance on a specific task) with data, without being explicitly programmed.”

According to SAS, the machine learning enables its program to provide “deeper, revealing insights that were previously out of reach. For example, machine learning use can include facial recognition in security systems, speech recognition in customer service applications, accurate product recommendations in e-commerce, self-driving cars and medical diagnostics.”

‘Fair share,’  avoid ‘shock’

The Wake County project is one in a growing number of SAS efforts to help modernize government technology. And Robinson noted that SAS aims to use its prowess in data analytics to help governments determine accurate valuations, thus helping to avoid “shock” for property owners hit with unexpected jumps in taxes due.

“Property valuation is a critical function for communities that levy a property tax. Properties need to be assessed fairly so that every owner is paying his fair share – not too much and not too little – of the total property tax,” Robinson said.

“In most states, property tax is the greatest source of income for a city or county.  And, for individuals, it is one of the biggest expenses of owning a property. In all communities, but especially those going through economic expansion or recession, the value of properties changes over time. When a community assesses a property, the assessment is accurate at the time it was performed.  However, that assessment grows stale as time passes. It can be a shock to property owners when a new assessment is performed and the value of their property dramatically changes.”

Second-source fairness

The agreement with SAS is part of a strategy Kinrade has put in place since the county decided in 2016 to adjust reappraisals to four years from eight.

“What we have seen is with a smaller value change it would tend to be better accepted by property owners and they would be less inclined to appeal,” Kinrade told WRAL at the time of the vote. Appeals of 2016 assessments dropped from the 2008 assessment, according to The News and Observer,  and he hopes that trend continues not only due to a quicker cycle but also the analysis provided by SAS.

Kinrade said he believes the SAS analytics will help ensure fairness when it comes to assessments, providing a second source of data for taxes – higher or lower.

Photo courtesy of Marcus Kinrade

Marcus Kinrade

“There are a lot of subjective factors involved when appraising a property so it’s helpful that we now have a tool that only performs objective analysis on the data providing a check and balance,” he explained.

“An appraiser may have worked in an area of the County for years and tend to rely more on their experience rather than what the data may be saying.   Sometimes this may be fine but other times it may create some inequities.  This is where the second source is very useful. ”

Kinrade added that assessments are just that. Tax rates are set by elected officials.

“The work our department has done with SAS has only been to assist us with one of our core missions of determining fair market property values while administering the tax base,” he explained.

“Services and the service levels that the majority of Wake County residents demand determines tax rates through the annual budget process.  The work we have done with SAS does not address the tax rate side of the equation, but I’m sure that would be a challenge SAS would love to tackle.”

Humans remain in equation

He also cautioned that SAS is just a “tool,” not a decision maker.

“The property values that we ultimately assess must be based on our adopted schedule of values, standards and rules, not the SAS models.,” Kinrade said.

“SAS built another model which analyzes property characteristics of a given neighborhood and then identifies the most comparable neighborhoods in the county.  The most comparable neighborhood might be directly adjacent to the subject neighborhood or 15 miles away. This model may very well help both staff appraisers and property owners working on an appeal to identify comparable sales or comparable assessed properties to include with their evidence, particularly if their neighborhood has very few sales.”

Kinrade added that Wake County will continue to add and train new staff for making the assessments. SAS, however, didn’t seek people with real estate experience in creating the program. Rather, it relies on data and the SAS executive said it is thus free of any “assessor bias.”

“SAS did not use any real estate experts or professionals,” Robinson pointed out.

“The system is based solely on sales, and is designed not to require any substantive prior knowledge or experience in property valuation.

“You enter a sale price and characteristics of the property that are public record and maintained in the Wake County database, and the SAS valuation system processes the sale and updates its valuations for all residential properties in Wake County. By analyzing thousands of sales using machine learning and decision trees, the system became more intelligent with every new sale. And with such a large number of properties analyzed, the system smooths out any ‘assessor bias,’ making it a truly fair system.”

According to Kinrade, stressed that new technology will boost the department’s efforts for accuracy and productivity.

“Our goal is to keep it [effective and efficient] by investing in more staff when needed and investing in cost-effective technology whenever possible,” he pointed out.

“We try to increase our productivity by automating repetitive tasks and increase our accuracy through quality data analysis, sound decision making and the elimination of errors. We try to benefit our Wake County taxpayers by offering quality, accurate services for the lowest cost of operation possible.”

Some man see lower assessments in 2020

In fact, the SAS program may help some property owners ee their taxes go down due to more timely analysis and the use of wide-ranging data. Overall Kinrade expects the SAS data will lead to fewer revisions made than in the previous assessments in 2016.

“The SAS models may help us identify some errors in our data that may yield a reduction, so this is possible,” he said, citing the possibility of lower assessments.

“But any data correction would also have to outweigh the market which has been appreciating significantly since Jan. 1 2016 in Wake County so while I’m sure there may be some properties that decline in value for Jan. 1 2020 based on some type of unique situation; but we most likely won’t see nearly as many that decrease as what we saw when we did the 2016 reappraisal.”

Kinrade also pointed out that property values will not be assessed in “real time” even though the SAS program calculates data as properties are sold.

“The SAS models are a tool that update nightly to give us perspective on what the market is doing now and gives us some insight on where it may be going if market conditions remain steady. ,” he explained.

As for whether the 2020 reassessment will raise taxes further, Kinrade said tax rates are assessed every year.

“In terms of increasing taxes, that will always be the product of the tax rate adopted annually by elected boards to fund budget priorities applied to the market value of a property.,” he pointed out.

Humans will continue to play the crucial role in assessments, Kinrade added.

“We have added two appraisers per year since the reappraisal cycle was shortened from eight to four years and the plan is to continue to do that for at least the next two years and then evaluate our work load,” he said.

“Investing in technology at the same time to assist with repetitive tasks and analysis has allowed us to increase our staffing in a planned way with no surprises and offer quality training to the new staff; rather than attempting to add a big group of staff at one time which would be difficult to handle.”