What Happened to Migration in 2014 (2013?)?
IRS Edition — Redux!
My post yesterday took a detailed look at the newly re-released IRS Statistics of Income migration data. I explained there how the data quality has been greatly improved, and noted the reason for the delay. So if you want cool IRS data showing how migration is associated with changes in income, like this chart:
Then go look at yesterday’s post. It was a good’un.
For today, I’m gonna run through the headline numbers again. I did this before when the IRS released this data the first time, but the re-released numbers are a bit different.
Migration Fell in 2014 (2013?)
Which is What CPS and ACS Told Us
I’ve said many times that the IRS has a biased sample, and should thus be seen as a useful indicator of migration, not an authoritative number. This is often overlooked simply because the IRS data is a “census” not a “survey,” so has no technical margin of error. But the truth is, the IRS migration data tends to understate the volume of migration by between 5% and 30%, depending on area and group. As it happens, the “denominator” for IRS migration, total tax exemptions claimed, is also much lower than total population however, so this under-estimation is largely hidden.
With all that said, let’s look at the IRS’ estimates of migration versus other sources.
The above chart shows that, since adopting the new methodology after 2011, IRS’ migration estimates have jumped upwards. I’ve covered that before.
However, the revisions to the 2013–2014 migration estimates ultimately led to a lower estimate of intercounty migration. What before looked like a small jot downwards now looks more substantial. The decline in interstate migration is less pronounced, but still present.
These declines are mirrored in the fall of CPS intercounty migration in 2014, and by the stagnation in ACS and CPS interstate estimates for 2012–2013 and 2013–2014. So really, they’re just confirming that 2013/2014 were weak years for migration. CPS jumped way up in 2015, and we don’t have any ACS or IRS data yet for 2015, and won’t until the autumn.
But I want to make a note here. It’s not totally clear that IRS 2013–2014 data should be accounted to 2014, rather than 2013. Most of the movements recorded as 2013–2014 probably actually occurred during 2013, though not all. The issue relates to how the IRS tracks migration.
2013–2014 migration compares the return address for “returns filed in 2014" to the return address for “returns filed in 2013”. Returns filed in 2014 will generally reflect 2013 income; returns in 2013, 2012 income. So we get 2012 and 2013 income, reported as 2013 and 2014 filings, in the 2013–2014 data, which reflects a change in address between 2013 filings and 2014 filings.
Most filings happen Feb-May, while migration is seasonally concentrated May-Sep. So much migration reported in the 2013–2014 data will occur in May-Sep of 2013. In other words, 2013–2014 data will largely reflect migration in 2013.
But that’s the same as CPS. CPS is based on migration over the past year as reported in March, most of which will have been during the past year. So for the “2014 to 2015” CPS migration estimates, we’re mostly asking about 2014 migration. This matches up to IRS data, and that’s why they often move together: they’ve got similar seasonal quirks, although because IRS is filing-to-filing, it sometimes asks about more than 12 months, sometimes less, because we don’t all always file in the same month each year.
But then there’s the ACS. The ACS is based on rolling monthly surveys applied to a correct calendar years. So ACS 2014 data refers to 2014. So in a sense, ACS 2014 is roughly equivalent to CPS 2015 or IRS 2015. Or at least, it should be. Except when there are events that cause weird spikes of seasonally-disrupted migration, which, okay, such events could happen all the time.
The problem is if I shift all the years around to line them up with what they should be… yeah it doesn’t work and the peaks and troughs are in the wrong years. So, um, let’s just say that it’s hard to get a good handle on exactly what data refers to exactly which years.
The IRS data seems to suggest that there was a downshift in filer-migration sometime around 2013 or 2014. This was after a move up the year previously. Other, more-recent sources suggest migration has risen since, in 2014 or 2015. We’ll see.
Here’s Your Bread and Circuses
Winners and Losers
Every migration researcher hates the “winners” and “losers.” But I crave your clicks. So here are some brightly-colored maps showing areas that gained or lost net migrants on net.
The trends here look a lot like what we saw in the 2015 county population estimates. So just go read my commentary there.
The map below shows gross migration by county. That’s a little different. For gross migration, I take inflows plus outflows, and divide by the population. This shows areas of high migratory population “churn.”
As you can see, there’s an arc of low-migration through Appalachia, western Pennsylvania and New York. There’s another broad area of lower migration in the northern plains and western Midwest, though Indiana and Ohio have higher migration rates. On the other hand, we see high migratory churn in western rural counties, Texas, Oklahoma, western North Dakota, the Atlantic coastal counties of the Southeast, Florida, greater Atlanta, much of Virginia, especially Washington, DC. If, as many experts believe, migratory churn is itself a positive indicator for regional economies and innovation, then this map may matter as much as net flows.
Now because I know some of you are nerds, I’ve also got the plain inflow (left) and outflow (right) maps below.
If it looks like the inflow map is more volatile than the outflow map, that’s because inflows are more volatile than outflows. The Interquartile Range for inflow rates was 2.38%, versus 1.97% for outflows. Standard deviations were closer, 2.20% versus 2.13%, inflows still had more variation. This, despite the fact that outflows actually had a higher maximum value (and they both had essentially the same minimums).
This is a common enough finding in the academic literature, that economic shocks primarily operate through shocking inflow rates rather than outflow rates. But new research suggests this is true even for some major environmental catastrophes like the Dust Bowl.
Conclusion
IRS migration data has flaws, but is constantly improving, and is a valuable additional source for migration data. In broad strokes, it seems to confirm other migration data sources, but methodological differences in each source make it hard to compare yearly estimates. County-level net migration trends seem to confirm data from the Census Population Estimates series, showing a return to pre-recession migration hotspots. Disaggregation of inflows and outflows also suggests that more of these trends are driven by regional differences in inflows than regional differences in outflows.
See my previous post, also about the new IRS data.
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.
