Over the last few weeks, I've been reading research on return migration. I didn't expect to discover so many different theoretical models in play. Some of the debates are the garden variety battle of the paradigms. Consider the matter unsettled. That opens the door for the likes of the Martin Prosperity Institute at the University of Toronto:

While it is generally agreed upon that talent is a key driver of economic growth, there is a fierce debate surrounding the optimal set of factors that help to attract and retain individuals with high levels of human capital (‘talent’) and mobility. One camp argues that good quality jobs must be present before talent will migrate while another camp argues that talented individuals are attracted to locations that offer a rich mix of amenities such as theatre, musical venues, restaurants, and other opportunities for recreation. In reality, however, the nature of work and what constitutes a ‘job’ is changing and preferences for work are differentiated by occupation, gender, ethnicity, life cycle, and past experience. As a result, understanding the conditions that attract and ultimately retain talent requires a multi-stage analysis. To move beyond the aforementioned jobs vs. amenities binary, Martin Prosperity Institute researchers Brian Hracs and Kevin Stolarick have developed a conceptual model that includes different stages and allows for the expectations and preferences of individuals to evolve over time while taking into account the draw of both jobs and amenities.
The above passage is rife with misconceptions. The suggested model will result in poor policy decisions. I have to wonder if these researchers even bothered to review the literature. But that's not to say that the work is worthless.
The best model is the one that explains the most migration. The bulk of relocations are short distance moves. Proximity is an important variable. Lumping together the jobs vs. amenities binary still misses a lot of migration. There is a ton of error in that model.
I've argued that knowledge of place is a robust way to model relocation decisions: You go where you know. Hracs and Stolarick can speak to pioneer migrations: Far away and terra incognita. Few moves fit into that category.
What really strikes me is the enormous success of college towns. When I wrote Rise you could see the rise of Austin, Texas. Consider the South by Southwest festival, what it is was then and what it is now. A couple of years later we could see in our data the rise of Boulder, Colorado and it's now one of the leading places for startups. The place at the top of our metrics now is Ann Arbor, Michigan. Here's a city that's in the shadow of Detroit, which many people are saying is collapsing and decaying. It's a large college town that has everything in common with Palo Alto, or Boulder, or Austin. We can add to the college town list Ithaca, New York or Madison, Wisconsin.
I emphasized the part about Ann Arbor. I'll remind you of that place's talent dilemma. In that blog post I compare Madison to Ann Arbor, referencing an article that has heavily influenced my thinking:
And despite Ann Arbor's educated work force, employers here find Michigan's reputation as a failing manufacturing economy can deter potential hires from moving to the state.
At HandyLab, an Ann Arbor firm that makes a DNA-analysis device, Chief Executive Jeffrey Williams says he has had a hard time finding Ph.D.-level workers with highly specialized skills. His company, which has doubled to roughly 60 employees in the past year, has 10 job openings.
"It's definitely gotten much harder with all the stigma around Detroit," he says. "Somebody tries to pigeonhole us as Detroit, we say, 'No, it's Ann Arbor, it's a completely different environment.' "
According to Richard Florida, Ann Arbor should be a talent magnet. It isn't. My model can explain why. The Martin Prosperity Institute model cannot. Boil it down to proximity. Madison touts nearby Chicago. Ann Arbor frets about Detroit's bad reputation. What you'll find when you disaggregate the Michigan population story:
As a region, West Michigan showed stronger population growth between 2000 and 2010 than most other parts of Michigan, which lost population overall, according to U.S. Census figures released Tuesday.
Ottawa County grew the fastest among the state’s 10 most populous counties, adding 25,487 residents in the past decade.
The Lakeshore county’s population grew by 11 percent, fueled by strong growth in its suburban communities of Holland Township, Georgetown Township, Grand Haven Township and Allendale Township.
Welcome to the Michigan part of Chicagoland, a world away from Detroit.