Census Population Estimates 2020-29

[Vosem]

Anyway, demographic patterns will be back to pre-Covid levels again.

Well, the real date at which growth patterns in the US shifted is 2015-2016.



Also, inasmuch as there's been a rise in fertility levels in the 2020s, it really doesn't seem to have similar patterns to the fall from the 2010s, right? The 2010s decline was concentrated in minority communities which had previously had especially high fertility (like Hispanics) and very religious communities experiencing secularization (like Mormons), whereas the 2020s rise has been described as a "work from home" boost and seems predominantly concentrated among wealthy people first and foremost. (This seems underscored by rises in the fraction of births which are to white mothers -- which hit a trough circa 2015 and have since risen -- and growth in the difference between conservative and liberal fertility.) Even if fertility rates do go back to their mid-2010s levels -- and to be clear there's very little evidence that they'll do so -- the actual people having children are very different ones.

Immigration patterns are really different today too -- a much greater fraction of immigrants are from East Asia or sub-Saharan Africa compared to Latin America.

Idk if this is just me, but these estimates seem to have a pretty universal theme that D communities are shrinking or at least stalling in population relative to previous census, while R areas are doing better. I wonder if there's some sort of reason for this because these estimates aren't politically biased, but if there was truly this theme of folks fleeing "Democratic cities", surely it would've shown up in the 2020 census.

Not if it was outweighed by early 2010s migration patterns (crime didn't hit a trough until 2013-2015, depending on the city -- for most of the Obama years, urban areas in America were still getting safer), and not if cities were generally overestimated on the 2020 Census. It's kind of underrated how much things have shifted over the past decade -- Oregon gaining a seat was fueled by bonkers first-half-of-2010s growth numbers, while their fortunes have shifted so much that it looks plausible they'll lose a seat in 2030. Florida's large COVID-associated gains aren't going to be reflected in the Census until 2030, either.

[Vosem]

It looks like Trende has added an extra seat? I'm trying to compare his numbers to Leon Sit's linear extrapolation from a year ago, and it looks like he's exactly the same, except that he has MI losing a seat and CO and SC gaining.

By the Census Bureau's own estimates, CO only gained a seat in 2020 in error, so it gaining another seat in 2030 would be the equivalent of it gaining two seats in a world where the 2020 Census wasn't plagued by errors, which would just be an absolutely insane outcome for such a small state.

We're also still on track for the "epic" results in other ways (among states gaining or losing more than 1 seat, California is projected to lose 5, New York 3, and Illinois 2; and Florida and Texas are projected to gain 4 seats each).

[Vosem]

After the 2020 Census I take these estimates with a giant grain of salt.

The $1M question is was there a bigger problem with the 2020 census itself or the annual estimates?  Did the census undercount the South or do the estimates systematically overcount the South?  In other words, would we expect the deviation to be a one-time thing or is this a persistent result of the difference in methodology where something about the literal census process/funding favors the North in perpetuity?

There's a good argument that the marginal dollar spent on census outreach went further in 2020 due to the pandemic conditions, and Northern/Dem states were more likely to fund outreach.  However, there's also a good argument that Northern/Dem states accustomed to a larger social services budget will always be willing to spend relatively more on this than Southern/GOP states...

The Census's own estimates are that its 2020 numbers were inaccurate, with CO, MN, and RI all having obtained seats they were not entitled to, and FL and TX each being shorted at least one seat. The last seat is very close between FL being shorted two seats and TN needing to gain a seat. (TN gaining in 2020 would've been an out-of-nowhere event, since none of the ACS estimates had it actually gaining).

After the 2020 Census I take these estimates with a giant grain of salt.

The $1M question is was there a bigger problem with the 2020 census itself or the annual estimates?  Did the census undercount the South or do the estimates systematically overcount the South?  In other words, would we expect the deviation to be a one-time thing or is this a persistent result of the difference in methodology where something about the literal census process/funding favors the North in perpetuity?

There's a good argument that the marginal dollar spent on census outreach went further in 2020 due to the pandemic conditions, and Northern/Dem states were more likely to fund outreach.  However, there's also a good argument that Northern/Dem states accustomed to a larger social services budget will always be willing to spend relatively more on this than Southern/GOP states...

Also its odd that so many Republicans went out of their way to discredit the idea of counting undocumented immigrants, when that would only go to further help in states like Texas and Florida.

While it marginally helps Reps in the EC, it helps Dems tremendously in the US House.   Any additional districts gained from this would legally have to be drawn as Dem districts in most cases, and with a tiny VAP for a US House district. 

This seems non-obvious to me, since the areas that are likely undercounted are quite red (fast-bluing, but quite red) suburbs, and Allen set a pretty high standard for when minority districts are required to be drawn. Also, if we're discussing the 2030s, then it should be noted that a Republican victory in 2028 (and possibly one with a strong enough trifecta happening in 2024) would push for a Census citizenship question, which would itself enable red states to pursue CVAP redistricting (as in the Evenwel case). Such would constitute a large internal reapportionment in TX and FL (...also CA and NY, I guess, but I doubt this would ever be implemented federally) from urban areas to rural areas.

Also, on your map the easiest additional large state for Biden to carry is GA, right? If FL and TX remain red then GA starts to have enormous importance for the Democratic party.

[Vosem]

After the 2020 Census I take these estimates with a giant grain of salt.

The $1M question is was there a bigger problem with the 2020 census itself or the annual estimates?  Did the census undercount the South or do the estimates systematically overcount the South?  In other words, would we expect the deviation to be a one-time thing or is this a persistent result of the difference in methodology where something about the literal census process/funding favors the North in perpetuity?

There's a good argument that the marginal dollar spent on census outreach went further in 2020 due to the pandemic conditions, and Northern/Dem states were more likely to fund outreach.  However, there's also a good argument that Northern/Dem states accustomed to a larger social services budget will always be willing to spend relatively more on this than Southern/GOP states...

Also its odd that so many Republicans went out of their way to discredit the idea of counting undocumented immigrants, when that would only go to further help in states like Texas and Florida.

Not to put my total-cynicism hat on, but it would help the wrong parts of FL and TX, and if you are a Republican in the FL/TX legislature maintaining control of that is the priority. Off the top of my head, Texas has a congressional seat -- TX-33, covering poor parts of inner-city Dallas and Fort Worth -- which verges on rotten-borough status due to the high non-citizen population, who don't vote, and giving areas like that more representation in state legislatures might allow Democrats to win even if they lose the popular vote (in Texas in particular I think this is a concern, because the rural areas are just so red). You ideally want a count within the state permitting you to draw the lines you want, and for large states this concern is probably more relevant than a single marginal congressional district.

[Vosem]

If this actually played out as Trende modeled with CA going full MI in the 2000's (and making the CO adjustment discussed upthread), the new map would dangerously close to "Dems must win TX or FL to win the presidency."  Needless to say, this would be a future where the South dictates terms to the rest of the country by midcentury.  For example, this would be a Republican EC win:



However, I'm dubious this will actually happen.  Between COVID, WFH, and mortgage rates, I suspect most of the people who were going to move between states this decade already moved, making a straight line projection unrealistic.

I have all 4 of CA/NY/TX/FL becoming swing states by 2040. If we look at current polling, CA and NY seem to be trending dramatically right, while TX and FL trend left.

Man, the four biggest states being competitve would make campaigning so unbelievably draining and even more expensive than it already is (though it would make things much more spicy for election junkies like us, lol). When's the last time the biggest state in the Union was a swing state? Was California considered a swing state during the post-VRA era?

New York single-handedly determined three consecutive elections in the 1880s (1880/1884/1888), with other swing states existing but being irrelevant. It was also a swing state during the 1940s, but after that tended to be at least Leans D.

California only became the largest state after the 1970 Census, but there was never really an election at which it was one of the central swing states. It was pretty clearly more Republican than the nation at the 1972-1980 set of elections (although McGovern demonstrated Democrats had a high floor there), and already more Democratic than the nation, and a sine qua non for Democrats, by 1988. It would've been a swing state if 1984 were closely contested, I guess.

Elections decided by a single state (if that state flips, it delivers the election by itself; all states closer than it together fail to flip the election):
1796: New York
1800: New York
1844: New York
1848: Pennsylvania
1876: South Carolina
1880: New York
1884: New York
1888: New York
1916: California
2000: Florida
2004: Ohio

New York was the ultimate swing state under multiple different party systems in the 19th century.