Homestead Exemption Behind the Numbers – Color Me Exempt

This post provides additional information and commentary about a local Austin 20% homestead exemption and its effects. Read Color Me Exempt in the Austin Chronicle and also Cui Bono – Who Benefits in the Austin Chronicle in order to have context for the information in this post.

Note that the Chronicle article published ‘average’ values, which is perfectly valid. Another way to look at it is to use ‘median’ values, which is also valid. The difference between the two is relatively minor – a few bucks on the high end, but nothing that makes much difference.

Appraisal Data (TCAD, WCAD)

The underlying data is provided by the central appraisal districts. The data files they provide are complicated and their documentation is weak. There are many exceptions within the data, as well as fields that have been added, changed, or removed over time. There is not a 100% accurate & reliable way to easily link the parcel file with the tax data, due to various exceptions and strange corner cases in the data. While those challenges are frustrating, most can be overcome.

As someone who has been writing real-estate software, off and on, for the last 7 or 8 years, I am very familiar with the challenges in the data files provided by TCAD and WCAD. If you are attempting to recreate my results and you run into issues, I would be more than happy to meet with you and help you with your database.

The data in my system includes about 99.4% of the homestead-exempt properties. There are only a few hundred missing properties (out of more than 130,000). The error rate is so small that it is insignificant for the purposes of our discussion. Furthermore, it is the most accurate number produced so far — previous estimates, by others, under-counted the number of homestead-exempt homes.

Plus, since we are dealing with aggregate numbers (typically by district), rather than dealing with homes on an individual basis, any errors on one or a few homes are smoothed out in the resulting aggregation and are essentially insignificant.

Data Files:

Travis Central Appraisal District tax roll database 9/9/2014 (supp166) and parcel shape files

Travis Central Appraisal District tax roll database 8/28/2013 (supp150) and parcel shape files

Williamson County Appraisal District tax roll database  9/5/2014 (certified) and parcel shape files

Williamson County Appraisal District tax roll database  7/18/2013 (certified) and parcel shape files

Austin 10-1 shape files (available from city website).

Note that I used 2013 data as an additional way to verify that the 2014 calculations were reasonable (correct). 

Make sure that you have your various shape files in the same projection. You will likely need to reproject at least one of them.

Notes about the Data Files:

If you would like to try and recreate my findings, I used the following data files:

A copy of the TravisCad tax roll database costs $55. Shape files are $25.

http://www.traviscad.org/faq_request_info.html

Williamson CAD data and shape files are free to download

http://wcad.org/public-data/downloads?id=343

General Instructions:

There is more than one way to go about doing this, but generally, you want to do something like:

Make sure all your shape files are in the same projection. If you don’t get them in the same projection the you will end up with properties on the edge of a district being counted in the wrong district (they will be in the wrong location).

Use the 10-1 shapefile as a “cookie cutter” to the underlying parcel data from the CAD (trim the map to the district extant). Your spatial software should have a couple of different ways of doing this. I used Cartographica on the Mac, which is commercial, but there are plenty of open source options out there as well.

The 10-1 district boundaries [in the 10-1 shapefile] do not line up with parcel boundaries (or with any boundaries). Thus, the 10-1 district borders cut across the middle or edges of various parcels. For those properties, you can either use the centroid of the parcel as a best estimate, or you can try and geocode the street address to see if you can get a more accurate coordinate (though they just give you the parcel centroid). Alternatively, you could write a script to run those edge properties through the city’s online district-finder, but it basically does the same thing (best estimate). As a result of this, there may be a handful of properties that end up in one district that should be in another, but it is minimal (perhaps less than 20 would be my guess – most people don’t put their homes on the very edge of their property).

Special Concerns in the Data

If you are trying to recreate my results, you should be aware of a few challenges that you will run into. Initially, people online tried to claim, based on my statements below, that I did not include condo data in my results, but that is not correct. I did include the condo data. But, if you are trying to recreate my results you need to be aware of the mapping between parcel shapefile and multi-family/condo data so that you can get correct results. That is why I listed the concerns below:

Multi-family residences (duplexes, condos and such) exist on a single parcel in the shapefile, usually, but in the data there are multiple owners, with a single street address, but different unit numbers, different property ids, etc. This can complicate things when trying to match IDs across the appraisal district data to the parcel shapefiles (depending on exactly how you choose to do things). Condo data, in particular, generally requires extra work.

The City of Austin extends beyond the border of Travis County, which means you have to use data files from multiple counties. In my examination of this issue, there were only a handful of Austin properties in Travis-adjacent counties – so few, in fact, that I left them out, except for the ones that exist in D6 (a few thousand that exist in Williamson County).

As a side note, it is unclear to me if Van Eenoo’s group was given only Austin homes existing in Travis County, or if their data pull [which was done by TCAD] also included Austin homes in adjacent counties (the Williamson County homes). Van Eenoo’s group came up with a $35.6 million yearly cost for a 20% homestead exemption and I came up with $36.6 million. However, if I don’t include Williamson, then I get about $35.6 million. So, that could be the reason. Or, it could be something else. For one, we did not use the same data files — the ones I used are several months after the ones they used, so that could also be the reason for the difference.

Data Accuracy

I spent several weeks constructing the data for this analysis. 99.4% of all Austin homestead exempt properties are included. The data is correct. Accuracy is very high.

Shortly after the Chronicle published an article using my data, a number of online trolls intentionally misread statements on this post in order to make it sound like I left out the condo data in my assessment. That is absolutely false. They also tried to claim my numbers were incorrect – the numbers are correct. In fact, the numbers are more exact than any prior analysis of a local Austin 20% homestead exemption, in my opinion. Furthermore, at the time of this post, I am the only person that has constructed a by-district analysis of the data, so if anyone online tries to claim otherwise, they are simply making that up — and that’s especially true for the person that sent an email to the Austin Neighborhood Council only a few hours after publication of the Chronicle articles.

It was also claimed that because I am software developer, and not a tax expert, that I have no business doing an analysis like this. The person claiming that is incorrect. This is almost entirely a software project – you don’t need to know much more about homestead exemptions than the fact that they exist & the ability to read some basic information. You do not need to be a tax expert. That said, I am a consultant and I have worked on various projects in the past that involved financial information (including working for and co-founding financial software firms). I have the knowledge and all the credibility necessary. Furthermore, unlike the people online, I have used my real name, I have provided concrete data, and I have no stake in the matter — if a 20% homestead exemption passes or doesn’t, it makes no difference to my bottom line whatsoever. That is most definitely not the case of the people online constructing lies to discredit me or this data. If any reporter would like to look inside the numbers, I will be happy to oblige – contact me and I will walk you through my mysql database, you can runs queries, etc. I stand by this data 100%.  It is accurate.

Dylan Tynan

Consultant, software developer, political wonk, etc.

Many years ago I worked for the Lee Cooke for Mayor campaign, but since then I have not worked for, nor given any money to, any political campaign.

Austin TX

2 Comments

  1. Great work! I think people may not be surprised that the richer districts will benefit more, but it is eye-opening just how big the disparity is. Do you have data on the owned vs rented tax burden per district? Not sure if you can differentiate residential vs commercial in the tcad data… if not maybe homeowner tax burden as a percentage of total tax burden from all properties? Might be interesting. I always wanted to play with the tcad data but am too cheap to pay their stupid fees 🙂

  2. Your article failed to mention people like me, that moved to what is now District 10 in 1999. My property taxes annually have gone from 5000/yr to 9000/yr. By the way it is the same 1-story home that was built in 1939. I guess we both feel that the price to live in Austin is unfair. By the way that is my email and real name. Not a troll and not a Tea party operative.