Black gold.

Oil Booms Drive Huge Migrations

And We Have Good Data to Track It!

It’s widely known that North Dakota has been one of the biggest migration-gainers in recent years. I’ve written about it quite a bit myself. But I was asked earlier this weak if I could do a post talking about mining booms and migration more generally. So I sniffed around for some data, and found that, lo and behold, there’s a lot of county-level data out there about oil and natural gas drilling.

So for this post, I’m using the Energy Information Agency’s drilling report data, found here. I also looked at USDA ERS’ county-level oil production data found here, but this data isn’t as useful, partly because it ends in 2011, and partly because it only shows oil/gas production. And the amount of oil or gas produced is irrelevant to migration. What matters is the amount of drilling activity. So for that, I need rig data. EIA provides rig data by “drilling region.”

Here’s a map of the drilling regions:

So that looks like a lot of the major oil and gas producing regions. But… but it’s actually missing a ton of big producer areas. Here’s USDAs map of production in 2011. Since it’s from USDA, we know we can trust this data, because USDA analysts make no mistakes, and are always correct, and should be trusted without question:

So my oil and gas drilling map is missing all the oil production in California, in southern Illinois, in Michigan, in Kansas and Oklahoma, in western Wyoming, in Louisiana… it’s really missing a whole lot of stuff.

But while I’m missing a lot, the quality of the EIA data is good, and it gives me enough different areas that I can test and see how well any relationships hold up under different circumstances.

So I’m going to use the EIA’s drilling regions. But keep in the back of your mind that large amount of oil and gas production occurs in other areas.

Drilling Matters More to Some Places

So We Shouldn’t Expect the Same Effect Everywhere


The above chart shows drilling rigs per 100,000 people by major drilling region. As you can see, drilling is very prominent in the Bakken, Permian, and Eagle Ford regions. In the Marcellus, Niobara, and Utica regions, drilling is much, much less prominent. We also see two regions with big changes in relative prominence: Haynesville declines dramatically, while Eagle Ford rises from a low-prominence region to a high-prominence region.

So we should expect to see the strongest drilling/migration relationships in the Bakken, Permian, Eagle Ford, and Haynesville regions. In the Niobara, Utica, and Marcellus regions, we can see that drilling isn’t really that big, so we will likely see noisier migration patterns. The Marcellus region is particularly notable, as it has over 8.5 million residents, including several major metro areas like Pittsburgh, with their own economic trends that may distract from any oil-driven migration shocks.

Drilling is Very Strongly Correlated With Migration

Well, In Places Where Drilling Is a Big Deal


The above chart is pretty wonky. But it’s important. The left shows the raw correlation between the prominence of drilling (rigs/100,000 residents) and net migration (net flows/1,000 residents). As you can see, some regions have fairly high correlations, but not all. Broadly speaking, high-drilling-prominence regions do seem to have higher correlations than low-drilling-prominence regions.

But associating, for example, April 2007 rig counts with 2006–2007 migration (which, by the way, is generally April 2006 to March 2007 migration) may be a bit odd. That captures only people who moved before and during any increase in drilling; i.e. more likely to be people “hired directly” by new rigs. But what about people who hear about drilling jobs, and then move to look for work? Those people we would expect to migrate after any initial increase in drilling.

So my lagged correlation coefficient associates April 2007 rig counts with 2007–2008 migration. If you find the idea that rigs hire people from far away, and bring’em in to work most plausible, the raw coefficients are what you should focus on. If you find the idea of unemployed job-seekers showing up to look for work more plausible, the lagged coefficients are what you should focus on. Or if you have expert knowledge of how the employment profile of drilling changes over the lifespan of a rig, you could make your own decision. I don’t have that knowledge and am not interested in taking the time to acquire it.

I prefer the lagged coefficients, and they yield very high correlations between migration and drilling activity for most areas. As expected, Utica and Marcellus regions don’t show much association. No big deal. Bakken, Permian, and Haynesville all show really strong associations, and Eagle Ford isn’t too shabby. Eagle Ford’s behavior is odd to me I’ll say, because it is a small region (under 1,000,000 residents in most years) with many rigs, and should have given me a good correlation. But it didn’t. So that’s odd.

See the Relationships For Yourself

Chart Phalanx, Advance!


Obviously, I’m highlighting the Bakken region, since it’s my golden child. The Bakken region’s explosive growth in drilling combined with its very low population and lack of overlap with any major metro areas makes it the boomiest of the booms, and the easiest to study. It’s a simplified case that shows us the roll a local economic boom can have.

Let’s be clear here: 94% of the change in migration in the Bakken region can be explained by changes in drilling activity. Which means without an increase in drilling activity, it is reasonable to suppose that the Bakken would likely have migration rates similar to what we saw in 2007 today; i.e. negative. Now this counterfactual claim is usually pretty hard to prove and people give you lots of pushback on it.

But seriously, is there anyone who wants to claim the Bakken region was poised for a migration boom without reference to drilling? Bueller? …Bueller? … Okay then.

Now, other regions are less clear. We don’t know exactly what would have happened in the Permian basin without oil and gas, because it’s bigger, more diversified, and even the correlation we have is much weaker, and the theoretical underpinnings aren’t as strong here either due to metro area overlaps.

But let’s make a fun, simplifying assumption that’s totally theoretically invalid, but may help us get a since of scale. Let’s assume my lagged correlation coefficients precisely reflect the impact of drilling for all the regions except Marcellus. This is equivalent to making 3 silly assumptions: (1) that my method is flawless and my correlations are actual causation in all cases exactly, (2) that Marcellus is a total outlier with no impact of drilling, and (3) that there are no other effects; i.e. without oil the residuals I find would not be impacted by any other shocks.

All three of these assumptions are wrong and silly, and so the exercise I’m about to conduct is not in any sense authoritative, and the numbers I arrive at should be seen as “ballpark” in the Major-League-Baseball sense, not the tee-ball sense. We’re talking about a seriously big ballpark.

What’s the Total Net Distortion of Migration Into Oil Drilling Regions as a Result of Drilling Changes Since 2007?

78,962. Or 145,772.

So I did a simple calculation to get my ballpark estimate. I took the annual change in regional net migration versus 2007, then multiplied by the lagged correlation coefficient for each region. This tells me the effect of changes in drilling in that year, assuming my correlation coefficients are divinely inspired causal instruments.

Add up all those increases and decreases, and you get 78,962. On net, boosts in drilling may have caused 78,962 “net” people to move to drilling areas. This isn’t the total number of people who moved to drilling who would not otherwise have moved. It is the number of people moving in, minus whomever they may have driven away (say, by buying up land or driving up local prices or increasing local crime).

Kind of a puny number, actually. If you divide each region’s sum by its 2007 population, you find that drilling boosted the Bakken region’s population by almost 22%… but didn’t boost any other region by more than 4%. The only other region to crack 2% was the Permian region, at 3.1%. In other words, the effect of nearly a decade of drilling has been fairly small for most regions’ population.

If we take this boost in migration as a share of 2007–2015 growth, drilling looks more impressive. Drilling drove 62.5% of the Bakken region’s growth, and in Utica, it drove was basically the only good news for growth: population shrank by 17,000 even as drilling brought in over 4,000 people. In the Permian region, drilling was 22% of population growth; in Eagle Ford, it was 10%.

There’s another way we can look at this, too. Drilling doesn’t just attract people. Declines in drilling lead to people leaving (and even increases, if people are driven out by changing local costs or amenities). This is a form of drilling-associated migration. So if we take drilling-tied changes in inflows and outflows, we get 145,772 “net migrations” distorted by drilling, either drawn in or pushed out of a region. That’s equal to 1.6% of the population of the oil regions (less Marcellus) relocated due to changes in drilling activity. Not huge, until you realize that total changes net flows (net inflows + net outflows) were just 197,000 for these regions. In other words, my estimates here suggest drilling activity drove 74% of the volatility in net migration in these regions since 2007. That’s substantial.


Extractive industries have a huge impact on migration. It should be no surprise that miners have among the highest migration rates of any occupation. In areas where drilling activity has been particularly prominent in recent years, increased extractive industrial activity can account for between 50% and 95% of migration increases. These boosted migrations account for between -5% and 120% of total population growth for impacted areas. All the same, the extraordinary experience of the Bakken region is indeed truly extraordinary. Most oil- and gas-rich areas have not experienced as pronounced a population boom as the Bakken region. This is likely due to the Bakken area having few other competing economic concerns of similar employment and revenue volume, unlike in regions that include large metro areas. Finally, the decline in the Haynesville region suggests that oil-migrants won’t stick around. The booms in state revenues will not last forever, nor will the populations consuming services and housing. This should impact decisions made about long-term investments in areas with high drilling activity.

See my previous post, about brain drain in Puerto Rico.

Check out my new Podcast about the history of American migration.

If you like this post and want to see more research like it, I’d love for you to share it on Twitter or Facebook. Or, just as valuable for me, you can click the recommend button at the bottom of the page. Thanks!

Follow me on Twitter to keep up with what I’m writing and reading. Follow my Medium Collection at In a State of Migration if you want updates when I write new posts. And if you’re writing about migration too, feel free to submit a post to the collection!

I’m a graduate of the George Washington University’s Elliott School with an MA in International Trade and Investment Policy, and an economist at USDA’s Foreign Agricultural Service. I like to learn about migration, the cotton industry, airplanes, trade policy, space, Africa, and faith. I’m married to a kickass Kentucky woman named Ruth.

My posts are not endorsed by and do not in any way represent the opinions of the United States government or any branch, department, agency, or division of it. My writing represents exclusively my own opinions. I did not receive any financial support or remuneration from any party for this research. More’s the pity.

One clap, two clap, three clap, forty?

By clapping more or less, you can signal to us which stories really stand out.