How can rural areas in Texas move from dependence upon a single industry or employer to broad-based and sustainable economic growth? Is it by incentivizing big-city companies to set up satellite offices in small towns? By supporting local entrepreneurs to build up economic assets, or by recruiting young immigrant workers and families?
The answers to these questions are not straightforward. The number of factors and amount of data required to answer them are huge, and they affect each other in difficult-to-analyze feedback loops: Economic growth leads to more entrepreneurship, for example, which in turn leads to more economic growth.
This complexity outstrips the tools and categories that researchers have historically used to answer these questions. The geographic categories, for example, often do not match reality: Economic activity seldom stops at the county, municipal or state boundary, even where those boundaries line up in intuitive ways. This is part of why the U.S. Office of Management and Budget uses metropolitan and micropolitan statistical areas, formed of counties with employment and commuting connections.
The second complicating factor in answering questions about economic development is that the amount of data involved is so large that it renders useless most statistical methods. Today we have access to an unfathomable amount of data, from purchasing records to social media links to mobile phone location data. Loading that much data into the rigid models used by most economists and political scientists snaps those models in two.
Thankfully, an emerging set of analytical methods, called statistical or machine learning can help us create and analyze large amounts of data in new ways. For the last three years, I have taught masters students in public policy, engineering and information systems how to apply these methods to complicated policy problems of their own choosing, from predicting armed conflict to analyzing attitudes toward fracking. These methods are designed to build and test multiple models in rapid succession, until we arrive at a model that most accurately classifies new data as we feed it in.
Thanks to grant funding from the “Innovation and Entrepreneurial Ecosystems in Rural and Small City Environments” at IC2, graduate student Mark Hand and I have the opportunity over the next two years to apply these statistical learning methods to the sprawling, complex economy of Texas. We will explore relationships among economic booms (often driven by energy development), entrepreneurial activity, and economic growth in Texas. Our hope is not only to better understand the economy of rural and small-town Texas, but to make recommendations to policymakers, energy producers, and local communities about how best to capitalize on rapid economic growth.
Click here to read more blogs from the IC2 Institute.