Over the past four decades, the United States suffered its first prolonged period of economic divergence. This type of divergence occurs when rich places grow at a noticeably different rate than poor places, which causes the standard of living to diverge. Roughly around 1980, the nation experienced broad convergence, and capital (businesses) and people migrated from poorer (lower-income, lower-cost) areas to richer (higher-income, higher-cost) areas.

Understanding why this growth happens is important because increasing differences in prosperity leads to fissures in American culture and politics. Increasingly poor places differ substantially from growing places. In these poor places, far fewer families thrive, social institutions such as churches and fraternal organizations weaken, economies languish, and health is worse.

This is a growing problem nationwide, one that far too few Americans take seriously. To address the problem through policy, we must first understand what causes this divergence. Whether or not divergence is happening is not in dispute. However, the details of why this is happening are not fully understood. A few general facts do give us some clues.

First, the places that are shrinking in population uniformly have much lower levels of educational attainment than those that are thriving. Nationally, college graduates make up approximately 38 percent of the workforce. We can assume, as a rule of thumb, that no place with a workforce of at least 40 percent college graduates is doing poorly. Conversely, no place with a workforce of less than 36 percent college graduates is thriving.

Whether educational attainment is a causal factor in regional divergence isn’t in question. Rather, the lingering unknown is understanding precisely what within the measured educational attainment contributes to regional divergence. One hypothesis is that regions with few college graduates lack the innovative capacity to adjust after an economic shock—such as a recession or plant closing. I’m doubtful of this explanation. Business process and technological innovation occurs very heavily on the shop floor of factories and offices. Indeed, most productivity-enhancing innovation seems to happen at the worksite from employees across the business. It’s not exclusively a college graduate phenomenon.

A second hypothesis is that higher levels of educational attainment provide a higher share of entrepreneurial talent in a region. I am very dubious of this explanation for two reasons. First, there’s very little evidence of important differences in entrepreneurial talent among regions. Second, there’s not much evidence of major differences in entrepreneurial capacity across educational attainment levels.

I’d be happy to be wrong on both of these hypotheses, but I suspect there’s a much more believable pair of related accounts for the role of educational attainment in regional divergence. I say ‘believable’ because these hypotheses closely correlate to the facts we already know about the places that are in decline. That alone makes them a far more powerful explanation than the innovation and entrepreneurship arguments.

The first of the two related arguments is simple—job automation and digitization are more likely to replace the tasks performed by less well-educated workers. We know this is true from a number of studies, including some that I’ve authored with colleagues at Ball State. We also know that these at-risk jobs are more likely to be clustered in places with lower levels of educational attainment.

To be clear, automation doesn’t cause a net decline in the demand for labor; it creates some jobs and destroys others. Automation and digitization create new, higher-paying, higher-skilled jobs in some places and destroy lower-paying, lower-skilled jobs elsewhere. If you have the right skills and live in a place with lots of other workers with those desired skills, automation can be a great economic boon. If you lack the desired skills and are surrounded by lots of other people who also lack those skills, automation can lead to job losses and significant family disruption.

This story is as old as the Industrial Revolution, but the global economy has been growing robustly for the past 300 years, and it will continue to do so. In order to participate in the new economy, individual workers must possess the ability to absorb new skills that may be vastly different from what they currently know.

For most of American history, we’ve been pretty good at this type of adaptation. From 1880 until about 1950, four generations of boys raised on farms built the Midwestern manufacturing economy into the marvel of the world. Farm skills and an elementary/middle school education easily transitioned into the skills and education necessary to work in steam- and electric-powered factories.

Modern automation and digitization requires a much higher level of educational attainment. We aren’t talking about moving from an eighth-grade skill to an eleventh-grade skill to accommodate new machinery. Today’s technological changes eliminate whole categories of tasks that in turn reduce the number of jobs in many low-skilled occupations. The jobs created by automation typically require several years of post-secondary education.

This cycle doesn’t just happen once, but several times over a working career. This sets up the second, complementary reason why places with a less-educated workforce now suffer long-term economic decline. College isn’t just a gateway to a new job; the college education itself insulates workers from job losses by making individual workers more adaptive to technological changes.

When we put together these two explanations, we are left with a story of regional success and failure that is pretty straightforward. Technology changes favor better-educated workers, and at the same time, education makes those better-educated workers more adaptable to new technologies.

Thus, places with lower levels of educational attainment are more likely to experience job losses. At same time, those displaced workers are less likely to have the ability to ‘upskill’ into new jobs. As a consequence, better-educated places grow richer, while less well-educated places grow poorer. I believe that over the coming years, this will be the accepted explanation for regional divergence.

These matters are well within the capacity for states to address. We just need more people to take seriously an economic problem that risks affecting America’s politics for several more decades.