Philly property assessments are systemically inaccurate in Black and low-income neighborhoods

[view original post]

Two years ago, as he was approaching 50, Paul Hall bought his first home.

He found a modest two-story brick rowhouse in the Cedarbrook neighborhood for $175,000.

But this year, the city drove up the taxable value of his home, saying it’s worth $227,600. Even though he’s enrolled in a tax relief program that will provide some savings, the spike will increase Hall’s tax bill by at least $611 next year.

“That’s $50 a month,” said Hall, who works as a cement finisher. “That’s $50 less in my truck’s gas tank. That’s lunch meat. That’s $50 towards the water or electric bill.”

Mere blocks away, another similarly sized home sold for a similar price to what Hall paid in 2020. But Hall’s assessment — and therefore his tax bill — are much higher. By the city’s own estimates, his property is over-assessed by 24%.

An assessment is supposed to reflect the amount homeowners could sell their homes for and is used to calculate a property owner’s tax bill. Accurate assessments are the bedrock of a fair property tax system — everyone is supposed to pay an equal share of their property’s market value.

And from a bird’s eye view, Philadelphia’s residential property assessments do fall within industry standards for accuracy and equity on the citywide level, and have been steadily improving on those counts for the past few years. But a more granular Inquirer analysis, including applying the city’s own analysis to the neighborhood level, finds Philadelphia’s property assessments are systemically inaccurate and distribute the tax burden unequally:

  • The city most accurately assesses homes in neighborhoods with higher proportions of white residents.

  • The more Black residents in a neighborhood, the less accurate Philadelphia is in its property assessments.

  • The city less accurately assesses neighborhoods with lower median incomes and higher poverty rates.

  • Most of the lowest-priced properties that sold in recent years are over-assessed. By contrast, most of the highest-priced properties are under-assessed.

These biases have been baked in for years. Independent researchers and auditors have repeatedly found similar patterns in the city’s previous assessments. Despite progress, the disparities remain.

“Estimations should be really close, and when they miss, they shouldn’t systematically miss,” said Ira Goldstein, chair of the Philadelphia Home Appraisal Bias Task Force and policy director at local policy think tank The Reinvestment Fund.

» READ MORE: Philly property assessments double in some working class neighborhoods

“OPA’s goal is always to assess properties as uniform and equitable as possible,” Philadelphia assessments chief James Aros Jr. wrote in response to questions. In general, very granular studies on a neighborhood level are hampered by “an over-concentration of sales or a small number of sales, either of which can produce unreliable results,” he wrote, although the city can’t confirm or dispute findings because it’s barred from considering race and income in assessments.

OPA’s reassessments in recent years have resulted in significant spikes in values. Hall’s new property valuation is a 52% jump, compared to the 31% average increase that hit residential homes after the city’s first reassessment in three years.

In practice, some level of error is inevitable. The Office of Property Assessment studies recent sales to try to figure out how home prices are affected by a handful of features, like square footage and the age of the building. It then tries to estimate market values for all properties.

The goal, therefore, becomes to minimize error — and, crucially, to have those errors distributed randomly, so that no one demographic of people bears a greater or lesser share of the tax burden.

Experts pointed to three data issues that are likely driving the inaccuracy and inequity of Philadelphia’s assessments: not having enough data on sales for the highest- and lowest-priced properties, relying on old data, and not being able to consider demographic data in understanding the impact of their assessments.

» READ MORE: Philly officials touted tax relief efforts. But some homeowners stand to lose money.

In an external report released in May by the International Association of Assessing Officers, auditor Larry Clark said OPA staff had concerns about the quality of the sales data they use to build their models. Clark and his colleagues hypothesized that inaccuracies on the extremes of home prices are caused by sparse data.

For the lowest-priced properties, sales may not even enter OPA’s database because “people in these neighborhoods had to devise ways of transferring properties without mortgage assistance”; on the higher end of the spectrum, properties sell based on amenities OPA may not be tracking.

“High-priced homes may feature amenities that are atypical in the market, such as gold-plated fixtures, professional grade kitchens, elevators, etc. These amenities may not be captured as part of the mass appraisal process,” Aros wrote.

Carl Gershenson, project director at the Princeton-based Eviction Lab and a Point Breeze homeowner, noticed that two near-identical, adjoining homes in his neighborhood were valued at $220,000 apart. On other blocks, he saw some newly built townhomes assessed less than older, more affordable housing.

“I’m very happy to pay my property taxes, but I’m not happy that my tax bill doubled and my neighbor who lives in an objectively nicer home did not,” said Gershenson. “It’s no way to build a city.”

Because assessments are built on past data, they are already out of date by the time they are released, which means assessors are always playing catch-up — which can leave them blindsided.

“The shorter the cycle the better,” said Syracuse University professor Yilin Hou, who has studied the city’s assessments with the Federal Reserve Bank of Philadelphia.

Over-assessment and under-assessment both present distinct challenges for homeowners.

Even though the city sets a flat tax rate for everyone, over-assessed properties effectively face a higher tax rate than everyone else, Goldstein said.

And the higher tax bill associated with an over-valued home can make it harder to sell, Hou said.

As for under-assessment, the city is directly undervaluing those properties, which can harm people’s ability to use their homes as collateral to take out loans. Clark said he has seen banks consider assessments in second mortgage decisions.

Brett Mandel, a former deputy city solicitor, said his home in the affluent Fitler Square neighborhood was assessed well below its market value, which amounted to the city missing potential tax revenue.

“Over the past 20 years, the city has made fits and starts at fixing this problem,” said Mandel, who once sued the city over its assessment system. “But every time, people start looking around to see what their neighbors are paying, and they go nuts.”

Hou and the Philadelphia Fed found that the city’s improvements in accuracy and equity since 2014 were attributable to more regular assessment cycles. Aros agrees — and while OPA hasn’t always met a prior goal of annual assessments, he said the city will conduct them regularly.

OPA could also collect more data on property features to better understand how they all contribute to a home’s market value, though there’s no industry consensus on the specific features assessors should consider. The IAAO has a few recommendations, but further variables are left to the discretion of the assessing authority, Clark said. As a rule of thumb: More data helps.

When OPA commissioned its own study of assessment practices in 2019, independent auditor Robert Gloudemans recommended the city consider the quality of construction in evaluating buildings. “I have never seen a single family residential model where it does not contribute to value,” he wrote.

» READ MORE: What we know about how Philly conducted the citywide property reassessment

The city hasn’t included construction quality in its assessment models and OPA isn’t currently planning on adding any new variables to its data, Aros said, noting that “if a variable cannot be captured consistently, OPA is unable to use it in modeling.”

In the scale of a city like Philadelphia, keeping up-to-date data is a major challenge, Clark said. “One of the most, if not the single most, expensive part of assessors’ responsibilities is to collect data in the field, door to door, what we call boots on the ground.”

As for demographic variables that could help identify and address disparities, the city is constrained by ethics codes. It can’t consider household race, ethnicity, or income in its modeling. In their methodology, OPA wrote: “In property assessment there is no such thing as a ‘poor’ or ‘low income’ or ‘wealthy’ neighborhood — only low-priced or high-price properties.”

Nonetheless, Hou had a recommendation: the city should start tracking income characteristics of households to understand the impact of its assessments.

But until improvements are made, homeowners like Hall of Cedarbrook could continue to get unfair assessments and tax bills.

”It makes me feel angry,” Hall said, “that the little guy is always getting the shaft.”

About the analysis

The Inquirer used the five years of sales data that the Office of Property Assessment uses to design and test its property assessment algorithm. Using OPA methods, that dataset was filtered to keep only arms-length transactions representative of an open market — that is, cases where a buyer was entirely free to buy and the seller was entirely free to sell.

To measure the accuracy of assessments, The Inquirer used “coefficients of dispersion,” the industry standard method and the one used by OPA at the city level. Properties were geocoded to census tracts and coefficients of dispersion were calculated on the tract level. The Inquirer found moderate correlations between tract-level COD numbers and the racial and economic make-up of the tracts.

The Inquirer also studied the uniformity of the city’s modeling across properties of different value, and found that overall, the city’s numbers fell within generally accepted standards. The city’s analysis found that for multi-family housing specifically, its modeling was considered unacceptable, erring on the side of undervaluing high-value properties.