Today, we released our new Tax Time Series. As someone who both worked on developing the feature and who personally relates to its primary use case, this project is something that I am so excited to see come to life.
I have a mixed background, having studied economics as an undergrad and earning a masters in finance before transitioning into software engineering — I first got excited about engineering during my time as a financial analyst (got very into Excel).
During this time, I also became a small-time, side-hustling commercial landlord.
Real estate is interesting to me not only as a financial asset but, given my love of economics, also, because real estate prices signal the economic pulse of an area; they’re a real-time gauge of how much people want to live there.
Working at Reonomy gives me the rare opportunity to combine economics, finance, and real estate knowledge with engineering, and it’s been a blast!
Additionally, I’ve personally felt the pain of large property tax increases in the past. But, because there’s been no foolproof way to evaluate it, I’ve never really had the means to do anything about it.
Upon joining Reonomy, this immediately jumped out at me as something we had the power to solve.
A bit of background for curious readers outside the industry: real estate property taxes are a local tax (as opposed to state or federal), typically billed twice a year.
This tax applies to all real estate assets (a single family home, office building, parking garage, etc.) though it often excludes property owned by religious or government organizations. The tax amount for a property is determined annually, and there are two key inputs*:
Tax Rate: The real estate property tax rate set by local publicly elected officials
Assessed Value: The current value of your property as determined by the local assessor’s office
*Technically there’s also the market value and assessed ratio. They don’t modify the concept and are relatively simple, thus are excluded to avoid confusion.
Annual Property Tax for an Asset = Tax Rate * Assessed Value
The tax rate is a public policy tool and standard across the locality. The assessed value, however, is specific to each property in the said locality.
And, despite there being a standard equation, it’s definitely not infallible and can often lead to over-taxation.
Which is why I wanted to help. Given our trove of data and tax records, I realized we could harness the power of Big Data and machine learning to create something even more useful for our users. As my first major project at a data company, however, there were a few interesting findings about Big Data uncovered that I think are worth sharing…
Big Data Has Holes
In a simple spreadsheet financial analysis, all data sources can be hand-checked (by folks like my 22-year-old self back in the day). When you are dealing with terabytes of data covering the roughly 52 million assets in the US commercial real estate space, however, it gets complicated.
For example, let’s say you have 10,000 fields of data covering each specific property aggregated from over 1,000 different sources. In this instance, it would be extremely time-consuming to hand-check even a single property… So no way would it happen for 52 million.
Having a large number of missing, or worse, incorrect values coming from various sources must be anticipated, but how does one deal with that?
Big Data Network Effects
The inability to guarantee the correctness of sources means that you will have both inaccurate and missing data coming in from your data sources. You cannot manually check this data, either—you must use technical methods to attempt to automate improvements in completion and accuracy.
This creates both an inter-entity and intra-entity network effect. The explanation for this is simple, if we can assume that the majority of our data sources covering a given entity (e.g. a specific tax year’s data for a property) are accurate, we can find errors and fill in gaps that might exist in a single data source.
Therefore the more data you have covering similar entities, and the more data sources you have covering the same entities, the more accurate and complete your dataset combining these sources can become. Thus a network effect is created by the fact that each individual data point is made better by the presence of additional data points.
The techniques used to improve completeness and accuracy by combining these sources could be its own article, but in short, they broadly fall under interpolation and anomaly detection and can be domain-specific.
So, how will Reonomy lower your tax bill?
Many articles focus on the future possibilities of Big Data and the lofty potential for it to transform our lives in 10 years.
But the Reonomy platform allows commercial real estate owners to leverage big data today to determine if your property is over-taxed.
How? Well, the job of the tax assessor is to determine the current market value of your property, i.e. what is the maximum amount someone would be willing to pay you for it today were you hypothetically selling?
The job of determining the correct price is very hard, so it is acknowledged that assessors will make mistakes, and almost all localities allow one to appeal the assessed value of a property.
This is where Reonomy comes in. You can see if the assessed value has gone up more than similar properties near yours… but how?
You, as an over-taxed commercial real estate landlord, can construct a compelling, fully-informed tax appeal in minutes instead of days by searching for properties similar to and near yours which have not seen a tax increase.
The time you save when building an appeal is critical because there are tight annual time windows (often 30-90 days) in which an appeal must be submitted.
There is also an industry of tax appeal professionals who often work on a contingency basis (they only make money if you save money) and provide a hassle-free property tax appeal experience for owners.
In addition to the benefits mentioned above, Reonomy’s “Year-over-Year Tax Change” search is also a great prospecting tool for these users, as it allows them to find and contact property owners that are over-taxed relative to their peers.
For a side-hustling property owner with a full-time job and other personal matters to attend to, spending days building a property tax appeal is nearly impossible.
A decrease in the amount of time required to build a tax appeal makes a world of difference to these busy small-time investors, like myself, who don’t have employees to help them out. Last year I took a 7% tax increase, and using Reonomy, I can look at the two neighboring properties and see that one had only a 1.76% increase and the other had a 1.86% decrease.
Across the street from me, a similar property had a nearly a 7% decrease — a move opposite to my own! In the often thin-margin game of real estate investing, this makes a huge difference.
The next time tax assessments come out, I will be ready with Reonomy, and hope you will be too!