What Will Census’ State Population Estimates Say?
The Data Exists to Take a Guess
Every year in December, the Census Bureau publishes an estimate of each state’s population on the previous July 1. Their process is quite transparent and makes use of mostly publicly-available data,, or data that can be roughly-approximated from publicly available data. The result is that, in theory at least, this method can be replicated by others. And, if those “others” are very efficient, it could be replicated before Census Bureau releases their own numbers.
The reason it’s theoretically possible to beat Census to the punch is simply that Census isn’t just estimating total numbers, they’re also trying to estimate by race and age, with the result that they don’t just use preliminary totals, but have to wait on detailed data. Or, at least, I think they do. Take births data for instance. Here’s a section from their official method documentation:
Vintage 2017 data will be released in December. But it turns out that we do actually have vital statistics data in much more up-to-date fashion. We don’t have to wait 2 years to get births data. We can get detailed data up through 2017 Q2, so the end of June, and can get preliminary monthly totals by state through June as well. There’s no reason that I can tell why Census should have to wait 2 years. But that second paragraph seems to say they do get preliminary data and they impute the full detail onto it… but then it’s even more confusing, because Census is forecasting July 1 population. So you don’t need full-year data. You need monthly data from July 2016 to June 2017, and we have that. Maybe that’s what they mean by full year? I’m not sure.
But on the whole, births and deaths are the easiest components to estimate. Migration is trickier.
The trick here is that I don’t have access to the Medicare data they use for retirees, and, also, I’m not sure exactly which IRS years they use. I’m also not sure if Census is using the IRS-produced estimates, or if they’re running their own in-house tabulation. It seems like they’re doing their own thing somehow, because, for example, in their 2016 estimates Census showed North Dakota with a negative net migration balance, despite the most recent public IRS data showing a positive balance. That suggests Census used some other kind of data. But the key point is that I can’t precisely duplicate their migration estimates. I have to just sort of do the best I can. But broadly speaking, my method follows the broad contours of Census’ method.
The result is that I can make a first guess at July 1, 2017 population. It’s only a first guess, and this is the first time I’ve done this. And, crucially, I take their 2016 estimate as “given,” when in fact they do revise back-years. Furthermore, they do revise their method periodically, with a very big revision in 2016. There is no way to foresee these revisions of which I am aware.
I have no doubt that my estimates here will prove ludicrously off-base from what Census produces next month. It’s my first time trying to get the jump on Census’ estimates like this. But my hope is that after a few years of doing this, I’ll get better at it. So, without further ado, here’s my estimate of state-level population change from July 1, 2016 to July 1, 2017:
Now, the truth is, this map means very little to most people, because very few of us have a solid mental estimate of a “normal” population growth rate. The notable thing here is that I forecast the mountain west, Texas, and the Atlantic south continue to exhibit strong growth, while we see continuing decline in West Virginia, Illinois, Connecticut, Wyoming, and Vermont, though in most cases the decline is moderated relative to 2016. My model, by the way, also gives a whopping 2.23% decline for Puerto Rico, a worsening decline vs. 1.81% in 2016. Keep in mind, this is population as of July 1, 2017, so it does not include losses from Hurricane Maria.
Another way to look at this is by showing how growth rates changed from the 2015–16 period to my estimate of 2016–17. Here’s that map:
As you can see, there’s more red than blue. And my estimate seems to suggest that “Sunbelt” gains made in 2015 and 2016 may weaken somewhat in 2017, with faster growth in some slower-growing northern and midwestern states.
However, let me put a caveat here. My method for calculating net migration produces less volatile numbers than whatever Census’ underlying method is. So some of this may simply be that my method artificially suppresses variation. The real test will be once we get the Census numbers, and can see whether I was close or not.
Finally, here’s a chart of net migration by state.
To be clear, that’s total net migration, not just domestic. And as you can see, resource-boom states still show up as negative in my estimation, as do Illinois, New York, and Connecticut. Post-Maria, I suspect Florida, Texas, and New York do better than shown here.
We’ll see next month how my estimates hold up!
Check out my 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 native of Wilmore, Kentucky, a graduate of Transylvania University, and also the George Washington University’s Elliott School. My real job is as an economist at USDA’s Foreign Agricultural Service, where I analyze and forecast cotton market conditions. I’m married to a kickass Kentucky woman named Ruth.
DISCLAIMER: 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.