Population
Undertow

Start from the observation that prompted this brief: voter rolls show a dip among younger cohorts, then a rise, and the rise never climbs back to the height of the baby-boom peak. That shape isn't a quirk of who registers to vote — it's the actual shape of the U.S. population by birth year, and it's produced by two separate mechanisms.

First, a wave passing through. 76 million Americans were born between 1946 and 1964. They are now between 62 and 80 years old, they own a disproportionate share of the country's housing and financial assets, and over the next two decades they will die at a pace no other cohort in U.S. history has. That's not a prediction — it's arithmetic on a cohort whose size and age are both already fixed.

Second, the replacement isn't showing up. Every generation born after the boom has arrived smaller, relative to what would be needed to hold the population steady, because the fertility rate that produced the boom never returned. The dip after the boomers wasn't temporary — it's been the baseline for over fifty years.

Together, these describe a population structure that is top-heavy today and will be bottom-light for decades. What follows is the data behind each half, a real-world instance already visible in the pipeline's own voter data, and what it plausibly means for where jobs, housing, and working-age people will and won't be.

Line up the generations by birth year and the shape the user spotted in the voter file shows up at the national level too: a peak, a trough right behind it, a partial recovery that doesn't reclaim the peak, and a renewed decline heading into the present.

The mechanical result: the cohort with the most money, the most home equity, and the most political participation is also the one about to shrink the fastest, and there is no cohort behind it large enough to fully backfill either its spending power or its labor-force exit.

Before going further into national projections, it's worth anchoring this in something already measured directly rather than modeled: the age distribution of Montana's 672,321 active registered voters, pulled from the current statewide voter history file (20260909_VIH.csv).

A question worth answering directly: do these figures include undocumented immigrants? The two data sources in this brief answer differently, because they're built for different purposes.

Practically: the Montana voter-age chart above is unaffected by illegal-immigration questions entirely — it's already a citizens-only dataset. The national population, fertility, and crossover figures elsewhere in this brief are not — they include an estimated undocumented population, which is now shrinking for a different, policy-driven reason covered in Section 05.

The "dip" isn't a one-time postwar correction. The U.S. total fertility rate (average births per woman over her lifetime) fell out of baby-boom territory in the late 1960s and has been below the population-replacement level of 2.1 for all but a handful of years since — most recently for sixteen straight years.

The rate briefly climbed back toward replacement by 2007 — the very years that produced the Millennial-echo bump in the previous chart — before falling again. It has not touched 2.0 since 2010, and 2024's rate of 1.60 is roughly 23% below the 2007 peak. Every year the rate sits below 2.1, the generation being born is, on paper, too small to fully replace the one it follows.

Europe got here first, and fell further. The U.S. decline looks less like an American anomaly once set beside Europe's — Southern European countries in particular dropped below the 2.1 replacement level in the 1970s and 80s, a full generation before the U.S.'s current sustained sub-replacement stretch, and never recovered.

Run the two engines forward and there's a specific year where they cross: annual deaths nationally exceed annual births, and natural increase (births minus deaths) goes negative for the country as a whole. Two independent estimates bracket it:

This is already true locally in parts of the country. As early as 2017, the Census Bureau found roughly two states and a third of all U.S. counties already had more deaths than births in a given year — almost entirely rural counties with older, longer-settled populations. The national crossover is the point at which that pattern, currently confined to the oldest and most rural geography, becomes the country's default.

Section 04 established that immigration becomes the only source of U.S. population growth once natural increase turns negative. That makes current immigration policy directly relevant to this brief's thesis — and as of 2026, that policy is actively shrinking net migration, not sustaining it.

Enforcement in 2025 was consistent with the administration's stated policy of enforcing immigration law across the board, not limiting removals to those with additional criminal convictions: more than a third of removals from detention involved no criminal record beyond the immigration violation itself, and most of the remainder's records were misdemeanors rather than felonies.

None of this changes the Section 04 crossover mechanics — it sharpens them. A national population and labor force that was already going to depend on immigration as its sole growth lever after the 2030s is now drawing down that lever years earlier than the baseline projections assumed.

The pattern shows up market by market, not just in the national aggregate — and it lines up with where enforcement has been heaviest. The White House's Jan. 2026 comparison of the 20 U.S. metro areas with the largest illegal-immigrant populations found year-over-year list prices had fallen in 14 of the 20 by December 2025 — and the three that instead posted increases were all sanctuary-city metros that limit local cooperation with federal enforcement.

Rents show the same pattern. Reported year-over-year declines by February 2026 included Austin (-6.6%), Denver (-5%), Phoenix (-4%), Jacksonville (-4.2%), and Houston (-3%) — a run of declines Treasury Secretary Scott Bessent summarized directly in August 2026: "People don't want to admit it: where ICE goes, rents go down."

Other factors are in the mix too — new apartment supply hit several of these same metros at the same time, and deportations also pull from the construction workforce, which cuts against new-home supply. So enforcement isn't the only variable. But the direction is consistent in metro after metro with heavy enforcement and a large illegal-immigrant population, and it reverses in the sanctuary-city holdouts — which is itself evidence for the Section 04 thesis: this population is a real, measurable share of local housing demand, and removing it measurably cools the market.

A different, and easily confused, statistic: how often sellers are cutting the asking price. Everything above is the level of list prices, year-over-year. A separate metric — the share of active listings that have had their asking price reduced at least once — is tracked monthly by Redfin and Realtor.com and is the one usually shown as a state-colored U.S. map. It's a leading indicator of a cooling market (sellers admitting the market moved before the price did), and it runs hottest in the same Sun Belt states named above.

This is the mechanism behind the cheap-European-property articles that started this thread. The U.S. version of the same story is already underway, though — importantly — slower and patchier than early forecasts assumed.

Europe is further along the same curve, and the houses are already showing up for sale — at prices that read like a typo. With fertility below replacement for decades longer than the U.S. (Section 03) and rural depopulation well underway, several European countries are already living through the "who's left to buy this house" problem the U.S. tiles above only project.

None of the above says the economy shrinks on a fixed schedule — immigration, productivity, and automation are all live variables the projections above hold roughly constant. But a few structural pulls look durable regardless of how those variables move:

Demand shift A larger, longer-lived retired population is a permanent tailwind for healthcare, caregiving, home services, and anything adjacent to aging in place — sectors that are already the fastest-growing part of the U.S. job market and are structurally protected from the labor shortage described below, because the demand itself is generated by the same demographic.

Labor scarcity A shrinking pool of new entrants doesn't require total population decline to bite — entry-level trades, healthcare support roles, and other jobs that depend on a steady inflow of 18–30 year-olds face a tightening supply well before the national population peaks, simply because that age band is the one directly squeezed by the "dip" in the cascade chart above.

Geography, not just totals The national numbers are an average. Places that keep or attract young working-age people will feel none of this; places that don't will feel all of it, regardless of what the national total does. Montana's voter data already shows a corridor where the young are thin on the ground relative to the population baseline — the open question for a place like North Idaho isn't "will there be a demographic crunch" but "does this specific region attract or lose the cohort that determines its own labor and housing future."

Immigration as policy lever Once natural increase goes negative — nationally by the early-to-late 2030s on current estimates, possibly sooner given the immigration slowdown in Section 05 — immigration policy stops being one input to local labor supply among several and becomes close to the only lever left at the national level. Regions and employers will increasingly compete on their ability to attract and retain people from elsewhere, whether that's other U.S. regions or other countries.

Every conclusion in Section 07 treats automation the way it has actually behaved for the last several decades: it eats routine, structured, primarily cognitive or repetitive-motion work (data entry, assembly-line stations, routine driving) and leaves alone anything that requires improvising in an unstructured physical environment — a plumber diagnosing a leak in a wall no blueprint matches, a CNA adjusting to a specific patient's mobility that day. That asymmetry is the entire basis for favoring trades and hands-on care as automation-resistant.

The mechanism that could break it Humanoid and mobile-manipulation robotics companies (Tesla Optimus, Figure, Boston Dynamics/Hyundai, 1X, Physical Intelligence, among others) are not trying to hand-program every task the way industrial robots historically were. They're building toward large, general manipulation models trained across simulation and fleets of physical units, where a skill learned or demonstrated once — fold this shirt, navigate this kind of clutter, use this hand tool — is distributed to every deployed unit as a software update. That is a fundamentally different diffusion curve than a human trade: training one apprentice takes years and doesn't transfer to the next apprentice; training one robot, if the approach works, effectively trains all of them simultaneously.

Why it hasn't happened yet As of 2026, no fleet has demonstrated this at the level of a licensed electrician or an experienced home health aide — current humanoid deployments are still narrow (structured warehouse and factory tasks, scripted demos) and the "long tail" of physical judgment calls in an unpredictable home, job site, or patient's body remains the hard, unsolved part. The gap between "folds laundry in a lab" and "rewires a 1974 farmhouse to code" or "recognizes a diabetic patient is going into shock" is still large. This is a real trajectory, not a settled outcome — the honest position is uncertainty about the timeline, not confidence that it will or won't arrive within a homeschooler's working career.

What would actually change the conclusion The Section 07 labor-shortage case degrades gradually, not on a single date — the earliest and clearest hit would land on the more structured, repeatable end of "the trades" (routine installation, standardized inspection tasks) well before it reached the improvisational end (custom diagnosis, patient-specific care, anything requiring trust and physical presence with another person). Caregiving in particular carries a demand component — humans generally want a human present for hands-on care of the elderly and disabled — that doesn't fully disappear even if the physical task becomes automatable, which is a real ballast the purely mechanical trades don't have to the same degree.