Why aren’t Americans moving anymore?

The Migration Mystery

There is a perception of Americans as an unusually mobile people. We are not content to live and die in the same place as our parents, instead we tirelessly chase opportunities, moving wherever fortune looks favorable and the future looks bright.

But not so much anymore. Americans are less mobile than they have been since the census started keeping track of movement in 1948. Back then, 20 percent of Americans had moved in the previous year, in 2016 only 11 percent had.

A fall in migration could point to a decline of dynamism in the American economy, a benign demographic change, or a positive sign that Americans don’t need to move in order to find the life they wanted.

It all depends on why we’ve stopped moving, and that turns out to be a surprisingly tricky question. What follows are some theories: all plausible, some backed by data, but with no obvious answer.

But first, the decline.

The Decline

The census keeps track of the percentage of Americans who have moved in the last year. The most recent figure is that 11.2% of the population had moved within the last year. Most of these moves, 6.9% were shorter moves that did not involve leaving the county. About 1.5% percent of Americans had moved to a different state in the last year. Both figures are about half of what they were in the 1980s.

Data from here. * indicates a year with multiple values, depending on controls. (In which case, I used most recent set of controls.) See census notes for more details.

The decline is relatively consistent over the years. Which means is not due to some cyclical reason such as business cycles, or even a large one off shock, such as the great recession[1]. Something more fundamental has changed.

Near and Far

People move for all sorts of reasons: work opportunities, family or relationships, school, a simple change of scene, fleeing enemies, or nearly infinite others. Short distance moves are not motivated by the same reasons as longer moves. They tend to be mostly driven by housing related reasons, while long distance moves tend to be driven by employment related reasons.

From: https://www.federalreserve.gov/pubs/feds/2013/201327/201327pap.pdf

While short distance moves are vastly more common, long distance moves are probably a better indicator for overall economic and social dynamism. Also, moving across the country is just more interesting than moving across town. So, if you don’t mind, we’ll focus on that.

Age

There are several demographic changes that have occurred in the last 30 years that could help explain the decline in moves. A big one is age. The US population has been getting older. The share of the population that is between 20-34 has fallen since the 1980s, this segment of the population tends to move more frequently than other segments.

Data for all these charts is from here (table 2). Charts are my own.

As an explanation, the general aging of the population would be more credible if there was less of a decline within each age group. Younger people seem to be moving less frequently than their counterparts of a few decades before, but so are older people.

Homeownership

Homeowners typically move one fourth as frequently than those who rent[2]. Partially because selling a home makes moving more difficult, and partially because the sort of people who decide to buy a home are likely more invested in their community than those who rent.

Between the 1990s to the mid 2000s, the homeownership rate rose, although it has since declined back to a lower level. The overall trend does not bear much resemblance to the trend in migration, so it seems unlikely that home ownership played a substantial cause in the decline of migration.

Education?

Other demographic trends don’t help much either. For example, the changes are not well explained by changes in education levels. As Americans become more educated, they have moved less, even though people with more education tend to move more frequently than those with less education. As with age, the decline in moves has occurred across education levels.

According to several economists [3], homeownership and an aging population are associated with about half of the decline in within county moves, but very little of the decline of between state moves.

From here (figure 2). See paper for full explanation.

The black line shows the change in migration rate. The dashed line controls for changes in age and homeownership-what the change in migration would be if age and homeownership had not changed over the years. If this line matched the red horizontal line at zero it would indicate that all the change was due to homeownership and age. The increasingly dashed line involves a similar approach for more variables (listed in the note). Homeownership and age do not make a big impact on interstate migration. Other variables matter more, but had they not changed it would have resulted in more migration, not less.

Land Use

In the most prosperous American cities the increase in housing has not kept up with the increase in housing demand. San Francisco is probably the most obvious example. Since the mid 90s, the amount of housing in San Francisco increased by only 12% while the price has increased by 341%. More people might move to San Francisco, but restrictions on housing stock could be stopping them.

Still, I’m sort of skeptical about this being a sizeable contributor. The change in migration is over a long enough time period, and for a large enough area that I don’t think changes in specific real estate markets are consistent enough to explain it. If the change was due to changes in house prices, it would have been more influenced by the housing bubble and subsequent housing crash. It wasn’t.

Changing Jobs

Since the early 90s Americans have changed jobs less frequently. Especially for long distance moves, there is a logical link between job switching and moving, as most people who make a long-distance move change jobs.

From: here figure 3.

This link could go either way, a decrease in job switching could be both a cause of lower mobility or a consequence of it. It looks like the relationship is not being driven by the process of people moving. Job switching is lower in states that fewer people are moving to, even for those who have not moved.

Licensing

There has been an increase in the number of occupations that require state based licenses. Twenty two percent of workers have a government license of some sort. The percentage of workers who require state based licenses have increased 28% since 1980. Since these licenses usually do not transfer between states, the cost of moving is higher for people whose profession require a license.

Indeed, workers with a state license are about as likely to move within a state as non-licensed workers, but are much less likely to move in between states.

While licensing almost certainly limits mobility for those in licensed occupations. It’s not clear if the effect is large enough to explain much of the decline in migration.

Why are workers changing jobs less frequently?

If workers have stopped moving due to less frequent job changes, the question then morphs to why aren’t they switching jobs as much. There are several potential reasons for this, all possible, none obviously dominant. It would be especially nice[4] if this reason could also explain changes in movement.

Changes in type of occupation

Over the past few decades, the US labor market has bifurcated away from middle skilled work (think things like administrative, manufacturing, and sales jobs) into low skill and high skill occupations. If a large source of past geographic movement was due to movement from low skill to middle skill occupations, this new gap could contribute to the decline in interstate migration.

This doesn’t seem to be the case. While mid skill work is associated with increased movement, the changes in the proportion of middle skill employment are not large enough to be a sizeable contributor to changing patterns in movement. Movement rates have declined evenly across education levels, which means that the decline is likely not due to changes impacting any particular demographic.

Information on Place

It is possible that the decline in moves corresponds with the rise in the ease of obtaining information about potential places to move to. Information could be more easily obtained thanks to declining travel costs and the increase in information technology. Perhaps people don’t need to move to find or investigate opportunities, but can conduct their research from afar and only move when the opportunity seems particularly enticing.

The ease of obtaining information is not an easy thing to measure. One way to obtain information about how much you like living in a place is to try living there, if you like it you stay, and if you don’t[5] you can leave. Greg Kaplan and Sam Schulhofer-Whol present evidence [3] that this sort of experimental migration has declined since the 1970s. They are also able to incorporate this sort of information into a model of migration.

All of which is fine, but it doesn’t say much about why people might switch jobs less. As most long term movement is undertaken for job related reasons, I am skeptical of this sort of movement being a significant driver.

The same opportunities, everywhere.

If certain kinds jobs are only available in certain locations, it is necessary for most people to move to get those jobs. The more jobs are specific to certain locations the more people will need to move. For the last few decades that the number of different occupations available in a locale has tended to increase.

Local labor markets have become more diverse in terms of the kinds of jobs and opportunities they offer, which means that it is not necessary to move to pursue a particular career. The degree of occupational segregation between states has declined substantially over the past couple of decades.

From here

While this fact might explain a decline in movement, it needn’t explain a decline in job switching. More types of local opportunities might easily lead to more job switching, not less.

The same wages, everywhere.

A similar impact is that the wages within occupations vary less between cities than they used to. If a worker can make more in one location than another, while still doing the same job, they will be more likely to move. Just as American cities are becoming more diverse in terms of which occupations they contain, the difference between wages between locations, within a given occupation, has generally fallen.

Changing employers

There is evidence that the benefits (measured in an increase in wages) to changing employers has declined while the benefit of staying with an employer has remained constant. Fewer benefits from job switching would lead to workers changing jobs less, and consequently migrating less.

This would provide a nice reason for both a decline in job switching and a decline in movement. Unfortunately, it raises additional questions. It’s not clear exactly why this would be occurring-some possible reasons might include a general decline in worker bargaining power or a decline in unionization.

Should we worry?

While I haven’t found a certain reason that is responsible for the decline in movement, I would veer toward pessimism. There are a number of ways in which the American labor market has become less dynamic, not only less job switching, but also fewer new firms being created, and fewer people being hired or leaving their jobs. Whatever the exact cause, the decline in migration is probably not good news.

(Not terribly) Related PARTYSHEEPHATS posts

Why is Flying in Canada so Expensive?

Who are the Self Employed?

Which Unemployment Rate is Best?

Footnotes

[1] The technical term for something that steadily changes over the long term and not cyclical is “secular”. I just learned this.

[2] https://www.federalreserve.gov/pubs/feds/2013/201327/201327pap.pdf -Table1

[3] Raven Molloy, Christopher L. Smith, and Abigail Wozniak

[4] as Molly, Smith and Wozniak point out

[5] If, for example, you have unwittingly moved to Reno.

Sources, References, and Further reading

Papers