Read More

Have we become too data-driven?

I anticipated the question, but I wasn’t sure in what form it would come. Turns out it was about the extent to which high school and postsecondary curriculum should be aligned with what employers say they need, assuming they (1) know what they need; and (2) can communicate needs in terms educators can translate into curriculum design and operationalize. My response struck a somewhat philosophical tone:

“The goal of education is not to get a good job. It’s one of the goals. It’s not the goal.”

Such started a thoroughly enjoyable and thought-provoking stay in Coeur d’Alene last week for EMSI’s 2015 National Conference, where I served as the keynote presenter. EMSI posted a rundown of the conference and a selection of memorable quotes from participants. You can find my keynote slides posted on the CA website. Presentations were not recorded, but the slides should give you an idea of the key points – all things covered during the semester in my class at UT-Austin, but never assembled as a single talk in a partially autobiographical format. Thanks again to EMSI for giving me an opportunity to try that out.

So, have we become too data-driven? Given the nature of most of our projects at Civic Analytics, it’s something that is always on my mind when working with clients, as well as with students in my class. Examples include:

What is the appropriate role of data in planning, and how best should it be incorporated into a planning process?

How much insight can we really glean from imperfect data, such as job postings, about supply and demand conditions in the labor market?

Further, to what extent should we rely on labor market information at all to drive curriculum choices in K-12 and postsecondary education? We insist on “aligning” education and workforce development to employer needs in an employer-driven system based on labor market insight, but what good does alignment do if newly minted graduates don’t have the three to five years of work experience required to get a foot in the door for an “entry-level” job interview?

Do we have too much faith in data?

Answering that question depends on how you view data. For me, this is where the line is drawn that separates academic research from applied research. Most academic researchers would, I think, say that the goal of research is knowledge creation. That knowledge may be applied to improve the world in some way, but creation is the primary goal, and, hopefully, we are all smarter as a result.

Most applied researchers, at least in the economic and workforce development context, view data as a means of communication. Applied researchers are responsible for using data to tell compelling stories, create a sense of urgency about needed improvements, and inspire people to act. We are judged not so much on our ability to make sense of the world, but on our track record of inspiring public and private sector leaders to make tough and often unpopular decisions to act to improve it. That requires quite a bit of faith in data, but even more confidence in the ability of practitioner storytellers.

Teaching applied data analysis for planning in an environment dominated by infographics, lofty claims about the power of “big data,” and a growing spotlight on open data in the public sector is increasingly difficult. The gains we’ve made in data accessibility are unbelievable, and planning is better off because of them. Yet, when confronted with such a vast amount of data, it’s easy for early-career planners and students to focus only on the mechanics of dealing with data–methods, statistics, dashboards–and less on the power of good storytelling. Knowing your audience. Avoiding jargon. Creating data-driven calls to action and goals that connect with community priorities. Inspiring people to want to act, and then empowering them with the insight and tools to do so.

Teaching methods is relatively easy. Helping people learn how to use data to ask good questions and inspire action is much more difficult.

But based on the expertise in the room and quality of discussion at the EMSI conference, I think our profession has a bright future.

Photo: Peter Røise Photography

Austin’s takes different tack on hiring tech talent

This story was written by Lori Hawkins and appeared in the Austin-American Statesman. It mentions a study that Civic Analytics completed for the Austin Technology Council on workforce availability in the tech sector.

A recent study by the Austin Technology Council underscored the hiring mindset held by many local tech companies.

Although 70 percent of those surveyed said they were having difficulties filling technical positions, 42 percent said they require applicants to have at least five years of work experience to even get consideration.

Only 12 percent of respondents said they hire recent college graduates or don’t require previous work experience.

At, company leaders are taking a different tack. The Internet job search company is focused on hiring engineering students directly out of college and providing extensive training to prepare them for their first job.

Indeed, which has grown to 500 employees in Austin, is one of the region’s fastest-growing Internet companies

Indeed has grown from 90 employees in Austin four years ago to more than 500 today. The company, which has been one of the more aggressive recruiters of engineering talent, is also hiring experienced developers.

This year, the company hired 54 graduates from 25 schools, including University of Texas, Carnegie Mellon University, Rice University, Georgia Tech, University of Illinois at Urbana-Champaign and MIT.

The new recruits have spent the past 12 weeks at Indeed’s offices in Northwest Austin, where they learned about the company, got to know company executives and broke into teams to come up with new Indeed products that they’ll pitch at an upcoming Shark Tank-style competition.

“There’s a huge benefit to bringing in people who don’t have experience,” said Chris Hyams, Indeed senior vice president of product. “Sometimes as companies get bigger they fall down in terms of new innovation. When you bring in people who are fresh out of college they are unburdened by the experience of what can’t work, and they are full of creativity about approaching problems from a new way that we might not have thought about.”

Indeed’s website aggregates job listings from thousands of websites, including job boards, newspapers, associations and company career pages. Job seekers search for listings on Indeed’s home page and can post their resumes for recruiters. The website, which is available in 55 countries and 28 languages, has 180 million monthly unique visitors worldwide.

In January, Indeed will move into a new 220,000-square-foot space at the Champion Office Park on North Capital of Texas Highway (Loop 360). The new offices can accommodate 1,500 employees, and Hyams said Indeed plans to fill them.

Founded in 2004, Indeed was acquired by Japan’s largest recruitment firm, Recruit Co., in 2012 for an undisclosed price. Under the deal, Indeed retained its name and operates an independent unit of Recruit.

According to financial reports filed by Recruit, Indeed posted revenue of $453 million in fiscal 2014, up 71 percent from the year before.

In addition to its Austin office, Indeed has operations in Seattle, San Francisco and Tokyo, and college hires have the option of working at those offices.

This year’s recruits spent the summer creating new apps and websites that could eventually become part of Indeed’s offerings. The teams developed 11 products, including a site that lets people post and respond to one-time job listings, such as language translation or logo design; and another that helps workers map out new career paths based on their work experience.

Shumeng Gu, who graduated from University of North Carolina at Chapel Hill in May, said she chose Indeed over other job offers because the company gives new hires big responsibilities.

“If you go to Facebook or another big company, you will likely spend your time just fixing bugs. At Indeed, you have ownership over what you’re working on,” Gu said. “The first week I got hired I put code into production, which is almost unheard of for someone at my level.”

Although it costs more to hire and train new workers, building your own talent has advantages, Hyams said.

“In Austin, we’re all trying to hire those same people with three to five years experience,” he said. “We decided this is a long-term plan for us, it’s not a short-term gain. We want to have sustainable growth, and the best way to get people with three years experience is to hire them at zero years experience. We now have a bunch of people with three years experience that we hired three years ago. Soon we’re going to have 54 more.”

Among Austin transplants, Florida is the new California

This story was written by Marty Toohey and appeared in the Austin American-Statesman on 09/05/15.

Among Austin transplants, Florida is the new California

“Californian” has for years been Austin’s shorthand for someone, usually of means, who moves here from another part of the country. But no more. Florida is the new California.

Here are five things to know about the newest migration pattern:

1. Nearly twice as many Floridians moved to Travis County in 2012-13 as did Californians. The Sunshine State sent 13,347 transplants here versus 7,959 from the Golden State,according to Internal Revenue Service data and further analysis by Austin economist Brian Kelsey. That ratio held for 2011-12 as well.

2. When you calculate net migration between Travis County and other states (people coming here minus people going there), Florida still remains the top feeder with 9,695 net arrivals. But New York (4,416 net arrivals) slipped past California (4,278) for the No. 2 spot. This trend held in 2011-12 as well.

3. Think it’s California tech money pushing up earnings? Floridians brought higher gross-adjusted incomes ($128,509 per household) than did Californians ($98,979) in 2012-13, according to the IRS data.

4. It’s not clear whether the Florida influx, which only surpassed California in the two most recent years of data, is a long-term pattern, Kelsey said. The IRS changed its methodology starting with the 2011-12 data by including late filers who previously weren’t counted and by better tracking people as they shift from a dependent to filing their own tax return, for instance.

5. Texans still by far make up the most people moving in and out of Travis County. But there’s a consolation prize for California: It remains the top out-of-state destination for people moving out of Travis County, edging out Florida in 2012-13 by 29 people.


Blame Florida? The Sunshine State Sends More Transplants to Austin Than California or NY

This story was written by Mose Buchele and appeared on KUT on 08/27/15.

Blame Florida? The Sunshine State Sends More Transplants to Austin Than California or NY

Sure, Austin revels in its youthful reputation, but a lot of the people coming here are probably not fresh-out-of-college looking to form a band or a startup.

A new look at income migration from the IRS shows that newly-arrived Austinites aren’t as young as previously thought. What’s more, the highest concentration of transplants isn’t from either of the tried-and-true drivers of Austin population growth, New York and California. They’re from Florida.

“There’s been some speculation lately that more and more retirees are picking Austin and this data certainly implies that that could be the case,” says Brian Kelsey with Civic Analytics, an Austin-based economic research firm.

The recently-released data tracks where people file their tax returns from one year to the next. He says people that are moving to Travis County are bringing a lot of money with them. In fact, the most recent 2012-2013 data show newly-arrived income in Travis County totaled about $2.3 billion dollars. Kelsey says that means many newcomers are probably older, maybe even retirees.

“If you look around the country at the counties that receive the most higher net worth people. It’s usually those destination retirement counties in Florida and Phoenix,” Kelsey says.

That brings us to the whole “California” thing. People are used to thinking California transplants were driving growth in Austin. Actually, Kelsey says, California’s in third place. New York is in second. The data puts Florida is in first place, for which, Kelsey says, there are a couple of reasons.

“Now we’re seeing Florida kind of coming out of nowhere. That could be something in the methodology that changed, where there was historically an undercount [of people] from Florida and that’s been corrected, or it could be that migration is actually increasing,” he says.

Kelsey says one thing’s for sure: People moving here with higher household incomes than the local average are able to pay more for things. That contributes to the local affordability issues we hear so much about.

“You’re going to see that show up in rents, you’re going to see that show up in housing costs,” he says. “And, just generally, it’s going to affect the aggregate amount of income in Austin for everything from discretionary money spent at restaurants to housing costs.”

Kelsey says the IRS plans to release more recent 2013-2014 data before the end of the summer.

Migration Added $2.3 Billion to Travis County Income in 2012-2013

Governing published a stunning statistic about Travis County in its review of the 2011-2012 migration data released earlier this month by the IRS. Mike Maciag pointed out in The Counties Where Wealthier People Are Moving that Travis County ranked second nationally in the amount of net income gained as a result of people relocating–i.e. income of people moving in minus income of people moving out. Travis County at #2 ranked among some of the most popular retirement destinations in the country, including Palm Beach County (#1) and Collier County (#3) in Florida, as well as Maricopa County (Phoenix) in Arizona (#4). Travis County gained net adjusted gross income in the amount of approximately $1.1 billion in 2011-2012, a truly stunning figure, especially when compared to wealthy enclaves for retirees, where migrating net worth tends to be a one-way trip.

Stunning, that is, until the IRS released the 2012-2013 data yesterday.

Net adjusted gross income flowing into Travis County in 2012-2013 totaled about $2.3 billion, more than double the amount for 2011-2012, and nearly all (97%) of it came from people moving to Travis County from other states. I’ll wait for Governing to update their interactive to see where Travis County ranks nationally for 2012-2013, but based on a quick check of Palm Beach County and a few others from 2011-2012, I’m guessing Travis County will be at or near the top of the list again.

A few other observations from the 2012-2013 release:

An estimated 68,664 households and 115,006 people moved to Travis County in 2012-2013, up by at least 35% from 2011-2012. Accounting for people moving out, net migration added an estimated 26,883 households and 41,210 people to the Travis County population in 2012-2013, according to IRS data. Changes in methodology at the IRS prevent us from comparing the 2011-2013 data to previous years. But I’ve been watching this data for Travis County since 2005 and this is the first time we’ve been anywhere near 100,000 movers.

Florida has surpassed California as Travis County’s most significant donor state. Historically, Florida consistently ranked in the top five, but always well behind California. Now, Florida is well ahead, which leads me to believe that there may be something quirky going on with the IRS data. Perhaps the change in methodology has fixed a historic undercount for Florida. Whatever the case, net migration from Florida added an estimated 6,647 households and 9,695 people to the Travis County population in 2012-2013, compared to just 2,092 households and 4,278 people from California. Net transfer of wealth was also much greater from Florida compared to California. Travis County gained approximately $1.1 billion from Florida movers, compared to just $261 million from California. Estimated adjusted gross income per household for Florida residents moving to Travis County in 2012-2013 was $128,509, compared to $98,979 for California households.

Speaking of California, Austinites might have to find another scapegoat, at least when it comes to blaming the state for driving population growth. In addition to Florida, California also trailed New York in net migration to Travis County for both 2011-2012 and 2012-2013, and Georgia wasn’t far behind in 2012-2013. However, you are likely still safe in blaming Californians for driving up housing costs, or perhaps thanking them for fueling Austin’s love-affair with high-priced foodism, whichever way you prefer to look at it. With an estimated average adjusted gross income of $98,979 per household, Californians moving to Travis County have relatively higher incomes than movers from other states sending significant numbers of residents to Austin, with the exception of Florida ($128,509).

Consistent with previous years, in-state movers had little impact on Travis County’s net increase in population due to migration, according to IRS data. Virtually the same number of people moved into Travis County from other counties in Texas as moved out. Travis County’s net gain in population resulting from migration can be attributed almost entirely to people moving to Travis County from other states.

For questions about the IRS data, see my recent primer on migration. I’d be interested in your thoughts on Florida in the comments section. Some have speculated a recent influx of retirees to Austin.

The Californians Are Coming, the Californians Are Coming

This story was written by Terrence Henry and was published on on 07/20/15.

Austin’s ‘Most Invasive Species’ Isn’t the One You Might Think

Central Texas is under attack. No, not Jade Helm, or even the summer swarms of mosquitoes. We’re talking about an invasive species. Zebra Mussels? Nope. Fire ants? Try again. We’re talking about an even more supposed “invasive” species: Californians.

They arrive with their telltale license plates, often heading straight to In-N-Out Burger and Trader Joe’s. As Austin continues to grow at a rapid pace, plenty of anecdotal blame has fallen on people moving here from California. Except … they’re not?

Turns out the net total of Californians moving to Travis County each year is only about 1,000 people.

“So, in a county that’s 1.1, 1.2 million people, that is a tiny, tiny, almost unnoticeable portion of the growth here. You know, in overall terms, it’s a very, very insignificant driver of population growth,” says Brian Kelsey, a principal with the economic research and planning firm Civic Analytics.

Kelsey looked at the most recent numbers available (2008-2012) from the Census Bureau and IRS to see where people are moving to Austin from. Of the people that move here from out of state, about one in seven is from California. But more than half of the people moving to Austin aren’t even from out of state — they’re from Texas. (You can read the full analysis at Civic Analytics.)

“Sixty percent of the people moving to Travis County come from other parts of Texas. So yes, the majority of people who move here every year are from another county in Texas,” Kelsey says.

So, if you want to blame someone for moving to Austin, you’ve got to start with Texans first.

Using Interactive 3D Models For Urban Planning

This article was written by Nathan Brigmon and appeared in 3D Visualization World Magazine on 06/22/15.

Using Interactive 3D Models For Urban Planning

The City of Austin, Texas is using 3D models for urban planning. These models provide the city with an effective and interactive visual approach that results in improved research and decision-making for urban planners. Civic Analytics, an economic research, planning, and consulting firm based in Austin applies this technology which has been used to enhance discussions around height restrictions, set backs, density, and also poor development decisions that may have been missed in previous maps.

Online 3D modeling has made great strides in recent years by showing its power to aid in the research and decision-making of urban planners. While it still doesn’t have quite the buy-in many would hope, web technologies have made great improvements thanks in part to demands in both the gaming and architecture industries. Some companies have been combining efforts of the two fields to create useful civic engagement tools. Civic Analytics has been utilizing ESRI’s CityEngine to show how the modeling of future landscapes and cities can play an iterative role in generating better public feedback and improved data analysis.


In the spring of 2014, the City of Austin working with the University of Texas at Austin (UT) and Civic Analytics set out to measure the economic impacts of building urban rail through the heart of Austin, Texas. The purpose of this analysis was to aid in City Council’s decision to place the initiative on the ballot and also to better inform the public during its potential November vote. The idea of urban rail in Austin had failed before, but many felt it was shot down for the wrong reasons and without proper consideration. In addition to being a smart planning approach to rail, many thought by bringing more data and answering better questions, a more informed populace would lead to a more suitable outcome.

Project Connect Answers

The question to answer in the economic impact analysis was simple: In the future Austin corridor, what happens if you build rail and what happens if you don’t? As urban planners, we were interested in the effects of new development, especially on its impact to land value. We were also interested in finding other differences such as changes in property tax revenue, walkability, water use per household, and a range of other indicators between the two scenarios.

The Analytical Process

Fig 1 – The corridor area for our economic impact analysis















Fig 1 – The corridor area for our economic impact analysis

To measure differences, we started both scenarios with the same base year. We then added projected population and employment data based on transportation impacts provided by a transportation research firm.

Next, we looked to the year 2030 as our comparison year. After adding the transportation impact data, we had a gap between our two scenarios we were now calling “2030 Build” and “2030 No Build”.

Our next step was to add building prototypes with unique data for people, jobs, water use, property value, etc. The route where we would place our building prototypes spanned 9 miles with a buffer of a ½ mile.

Our goal was to add buildings until the cumulative impact of those buildings’ attributes for people and jobs met our projected numbers.

In other words, we were filling in the gap with actual development. Each building was vetted for accuracy and local appropriateness including form, taxable value, and height. Our building library consisted of over 80 individual models with a variety of attributes.

Some of the attributes considered things like people her household, water use per household, and site area impervious cover. So, we were constructing a future where we could measure a number of impacts ranging from household level to citywide.

The software used was ArcGIS and Envision Tomorrow. ArcGIS is well known in the geographic information systems field for approaching any task that deals with spatial data and maps.

Envision Tomorrow is a popular open source scenario planning tool that allows planners to measure land use impacts based on building inputs. It runs within ArcGIS and creates a variety of useful outputs. To sum up the technical piece, we added building data to parcels in ArcGIS and Envision Tomorrow showed us the results.

The workflow for the analysis was a back and forth process intended to calibrate the model and improve accuracy. Researchers from Civic Analytics and UT would run the analysis and then pass the results on to UT professors and a Citizen Advisory Group for feedback.

Once feedback was received, the researchers would take this feedback, recalibrate the analysis, and pass the results back for more feedback.

By the deadline of the analysis, our model would have been tried and tested over and over. At least, this was our goal in theory.

Generating Feedback

When looking to the future you are always going to make assumptions. The best way to improve the accuracy of a future model is to come to an agreement over those assumptions.

Our original process for challenging the model was to use maps to show the land use changes and development impacts. This works great in theory, however, we were working with a non-technical citizen committee and an extremely large dataset, so we ran into a few issues.

First, we had the issue of complexity. Each mapped corridor of our analysis involved using over 25-30 building types. This meant we had an extensive legend that became hard to interpret. It also meant we were potentially overwhelming our non-technical audience.

Second, we had the issue of form. Showing building impacts by adding them to parcels mean we lose the opportunity to examine density and height. This is of course a natural limitation to many two dimensional visualizations.

Fig 2 – An early draft map showing the extensive legend, which creates data complexity.

Fig 2 – An early draft map showing the extensive legend, which creates data complexity.

To be fair, this did not mean we had no feedback, but there was much to be gained as we found with the introduction of our 3D modeling tool, CityEngine.

Enter CityEngine

The 3D component of the project didn’t have a lot of buy-in from management at the initial stages of the project. It was actually referred to as “a bell and whistle” as some couldn’t visualize a 3D model benefiting an economic development process. However, after the first round of modeling, it quickly proved to be a very useful tool.

CityEngine is a 3D modeling software originally built for creating city landscapes very quickly and with mass randomization. ESRI bought the software in 2011 has worked to better integrate it into the GIS world, which means getting it to focus less on randomization and more on real world infrastructure. Since CityEngine has yet to cross that threshold into seamless integration, I had to take some interesting paths to create the model, but it was well worth it.

First, since we were adding building data to parcels, I needed a method to create building structures from those same parcels. So, I wrote a python script to read the land use data of the parcel and then split it based on the attributes of the building site’s area. Next, I wrote a CityEngine script to add setbacks and height according to the building’s attributes. The next step was to apply this to every single building throughout the 9-mile corridor. Using the iteration power of CityEngine, I was able to apply my code in seconds!

The result was two models that represent the visual impacts of a build and no build scenario. However, this only got us part way to our model since in this current form it only existed on my computer. The next step was to use ESRI’s cloud service to share this model online with the public. So, I uploaded the model and now could share not by file but by link.

Fig 3 – ESRI’s ArcGIS Online provides cloud storage for 3D CityEngine models

Fig 3 – ESRI’s ArcGIS Online provides cloud storage for 3D CityEngine models

At our next meeting, we had discussions around height restrictions, set backs, density, and also poor development decisions that may have been missed in previous maps. We were also able to remove the color scheme thanks in part to the interactivity of the model. If someone had a question about a particular building use or purpose, we could select the building and learn more information.

Fig 4 – Having the option to interact with buildings aided feedback discussions

Fig 4 – Having the option to interact with buildings aided feedback discussions

The data became less complex, and therefore easier to ingest and interpret for errors. For example, as a researcher I had little knowledge about local neighborhood master plans that had height restrictions.

Another great feature from CityEngine, is the ability to compare both Build and No Build datasets side by side and use the “swipe” tool. Instead of putting two maps side by side, the swipe tool allowed us to spot differences in a much more intuitive fashion.

Fig 5 – The swipe tool empowers users to visualize data side by side

Fig 5 – The swipe tool empowers users to visualize data side by side

Finally, since the model exists in the cloud, it can be visited by anyone during his or her own free time and not during scheduled meetings or within short time frames. This gives people time to digest and revisit things they feel might have been overlooked. After our first meeting with the model, it was clear that it had been a success, but it’s important to understand that it was not being used as a finished product but as a tool to generate feedback and improve the analysis. As we moved on to our next iteration, we had better feedback and also a new understanding of our new 3D modeled dataset.

Next iteration

For the next iteration, we decided to add some texture and life to the model. In addition to some of the changes made from our last round of feedback, we added rail cars, terrain, and building form.

Fig 6 – A view of the finished transit stop

Fig 6 – A view of the finished transit stop

This allowed us to put ourselves in the potential transit corridor and question assumptions about how we got there. Again, the model is not a finished product but a means to interact and inform. It became clear to the research team that this tool had a lot of potential and its future was bright. For a full look at the model, visit it in ESRI’s cloud.

Future uses

As CityEngine and other interactive 3D modeling software gain acceptance, we will begin to see them used more and more. At conferences, ESRI is famous for incorporating Oculus rift into their 3d web models and allowing users to run around and even fly through new developments. Is there a better way to get feedback than placing an individual inside a model and enveloping their senses?

Also, as mobile technologies improve, some companies are taking this idea to our cell phones. Both Google and InsiteVR have created 3D technologies that enable you to iterate through cities and other worlds to explore and also play via a mobile device.

From an urban planning standpoint, this technology is a game changer. Imagine local governments voting on new development corridors with the power to model impacts at the click of a mouse or swipe of a screen. It simplifies discussion and empowers anyone to be able to view future changes and how it could potentially affect them.

In my opinion, as 3D technologies gain momentum, it will become as expected to have a 3D model during master planning projects in the same way 2D maps are considered a requirement today.

Nathan Brigmon is an analyst at Civic Analytics, an economic research, planning, and consulting firm based in Austin, TX. He specializes in 3d modeling, GIS, and data analysis.