Economists have been seriously studying migration for a couple centuries; for me it’s been a couple of decades. Over that time, we’ve learned a few useful things that should help elected leaders think about why places grow or shrink.

First, most migration is concentrated among younger people with high human capital. Yes, retirees move, as do folks in mid-life, but most don’t. One result of the age concentration of migrants is that this movement of people also drives natural population change of births minus deaths. So, places with net in-migration tend to thrive over the coming decades, while places that lose folks do not.

Migration of people is driven by three factors; economic opportunity, quality of life and housing elasticity. Housing elasticity is simply whether the supply of housing adjusts to demand. With the exception of a dozen or so large metropolitan areas in the U.S., housing elasticity plays no meaningful role in household migration. In fact, the Midwest currently benefits from bad housing policies in other regions such as the West Coast. Thus, migration in the Midwest really comes down to economic opportunity and quality of life.

Until the 1960s, almost all migration within the U.S. was driven by economic opportunity. By every measure, Americans were much poorer in the 1960s, and thus more willing to move long distances for jobs. For most of American history, we’ve moved often for better farmland, better jobs and better places to start businesses.

During the 20th century, economic opportunity was clustered in urban places and, until the 1970s, in manufacturing. Here workers could anticipate a very high wage premium. By the 1970s, the biggest opportunities shifted to urban areas with large shares of college graduates. Manufacturing employment began its long decline, and so was no longer a destination industry. It is still productive today, but job growth has been a declining share since the 1940s, and will continue.

Today, the role of educated workers is even stronger, with the share of college graduates explaining close to 80 percent of the growth and earnings in a city. Never before has education mattered as much as it does today. But, as we grew richer, we began to value more than just economic opportunity in our location choices.

Over the past 50 years, the strongest predictor of population growth in the U.S. was climate. January temperatures alone accounted for a third of the variation in migration. This was the start of several decades of the rise of the Sunbelt. But, there is more to the story as well.

Economic opportunity in northern states languished. They recovered in the Northeast, but were much slower to in the Midwest. Real economic opportunity was clustered in large urban places on the coasts and in the South. During the 1990s, several Midwestern states awakened to the problem, and by the early 2000s began worrying about economic opportunity.

Over the past couple decades, ‘business friendly’ policies designed to weaken labor unions, reduce taxes and regulatory burdens swept across much of the Midwest. But, these came too late to affect the movement of people. By the start of the 21st century, most Americans, especially educated and mobile workers, had occupations that could be performed anywhere.

The lure of economic opportunity no longer followed the factory or farm, but was tied to a service sector job in finance, healthcare, or personal services. These jobs follow people, not business-friendly tax climates. In 1980, few places enjoyed both economic opportunity and high quality of life, but as of 2019, they are highly correlated. This is changing where we live.

Over the past couple of decades, families found they their location choices were vastly expanded. Economic opportunity was tied to the places where people clustered, and people clustered where the quality of life was good. But, the factors that defined quality of life changed rapidly. The Sunbelt couldn’t rely on weather alone.

Natural amenities, such as climate, mountains or oceans dominated the early quality-of-life migration. To anyone who remembers the ’60s or ’70s, that makes perfect sense. Much of the obvious differences between places were driven by nature, not government. There were two obvious exceptions to that. Parts of America were unwelcoming to minorities, particularly African-Americans. Second, cultural norms towards women differed by region.

The empirical evidence is now extraordinarily clear. Places with restrictive social policies in the United States fail to become destinations for economic opportunity. They struggle to attract and retain their share of well-educated people. That trend is sure to continue, if not accelerate.

However, by the 2000s, something else arose that mattered far more than economic opportunity. Our national focus on school quality made available actual data at the school level. This data revealed surprisingly vast differences between schools. At the same time, labor markets began valuing education far more heavily. So, for the past couple of decades, it has become obvious that the quality of a K-12 and college education were prime determinants of economic opportunity for individuals.

In the post-COVID environment, the role of quality of life is even stronger. Today a quarter of all young, educated people have full-time remote jobs, and half work at least partially remote. The certain effect of this is that the amenities (and dis-amenities) of a region will weigh more heavily on prospective residents than ever before.

Though people are wonderfully heterogeneous beasts, we know from recent research what the most popular attributes of a region are. First, they are superb (not merely good) schools and an absence of crime. A few natural amenities matter, but they are less important than local recreational opportunities. Healthier populations lead to higher quality of life in a region, which is likely connected to recreational opportunities. None of this should be news to anyone.

What is new is the fact that the effect of quality of life on population growth is close to four times larger after COVID than in the decade before. Much of that is due to remote work accelerating the existing trends. We don’t yet know how long that will last, but my guess is for at least a generation. We also know that a welcoming social climate matters.

States, cities and towns who make themselves unwelcoming towards groups of people have and will find themselves with fewer of those residents, and fewer of their friends and family members. With economic opportunity wrapped so closely to quality of life, we should expect preferences towards state and local policies to play an increasing role in local choices. None of this can be ignored.