My prior report on this market established that four years of NGCB data support genuine, if modest, expansion of the Las Vegas locals market since Durango opened — and that Red Rock Resorts captured real share on top of that expansion. One variable was conspicuously absent from that analysis: the demand driver everyone in this market, RRR included, points to as the underlying reason any of it is happening at all. Clark County population growth. This report tests that claim directly, using the same primary-source discipline as the prior one — the state's own certified population data, a live regression rather than an assumed correlation, and a side-by-side check of what RRR's and Boyd's own investor materials claim against what the independent data shows.
- Population growth and locals-market revenue growth show no detectable headcount relationship, 2022–2025, once the shared upward trend is removed — consistent across annual, monthly, and 65+-specific tests, though the underlying sample is small. Age composition is a separate, smaller, better-evidenced story (see below): headcount plus mix-shift together support roughly 0.9%/year, not zero.
- The single best-evidenced explanation for the gap between population and revenue growth is new supply, not demographics: 2024's 5.8% three-area GGR growth against 1.3% county-wide population growth is a 4.5-point gap traced directly to Durango's December 2023 opening — a gap that widens to 9.4 points when narrowed to the Balance of County sub-area where Durango actually sits.
- A sharper sub-area pattern: Henderson has the fastest population growth and the slowest gaming-revenue growth of the three NGCB sub-areas — population proximity alone doesn't predict revenue growth.
- Age composition, not headcount, is the better-supported story — the 65+ population has grown 67% since 2010 versus 25% for the total population, and spends roughly 2x the average share of income on entertainment.
- A cohort-survival decomposition shows that 65+ growth is not driven by retirees moving to Las Vegas — Clark County has net out-migration past 65. The inflow happens at ages 50–64 (roughly +4,100/year), which then ages into the high-spend cohort a decade later.
- A demographic decomposition suggests population and age-mix alone support RRR's own low-end (3%/year) locals-market growth scenario, but reaching the mid or high case requires an additional, currently unverified assumption about per-capita spend growth.
- RRR's own investor-materials population claims largely check out against independent state data; where they diverge, it traces to a different, legitimate data vendor, not an error.
- Boyd's public materials treat this topic with far less rigor — one boilerplate 10-K sentence and a single project anecdote, versus RRR's multi-vendor demographic apparatus.
- Open question that could undercut the whole forward-looking thesis: the 2.0x entertainment-spend premium this report attributes to the 65+ cohort may be an age effect (durable, applies to whoever turns 65) or a Baby Boomer cohort effect (specific to this generation, and not necessarily true of Gen X or Millennials as they age into 65+). Independent research doesn't resolve this, and RRR's own bear case already names "younger generations' shifting entertainment preferences" as a risk. Dating the risk: Baby Boomers finish aging into 65+ by 2029, so this report's near-term 2026–2031 figures are unaffected — but exposure grows sharply for the 2036 and 2044 projections, and the migration-inflow finding above is itself a Boomer-specific data point, not yet tested for Gen X.
- The historical dataset was deliberately not extended further back, because the years with real population-growth variance (the 2008 recession, the 2020 pandemic) are also the years dominated by confounds that would muddy the test rather than clarify it. That choice has an honest cost, stated plainly: the clean window is also a short one, which is exactly why the regression can report a null result but cannot statistically rule a relationship out — the two limitations are the same tradeoff, and the planned follow-up (extended data with explicit macro controls) is the way to resolve both at once.
The findings are summarized above; what follows is the full analysis, source by source and test by test, for readers who want to see the work behind the conclusions.
The Data: Clark County Population, Certified and Projected
Population estimates for Nevada counties come from two related but distinct official pipelines, and conflating them is an easy way to misstate this market's growth. The Governor's Certified Population Estimates, issued annually by the Nevada Department of Taxation, are the official as-of-July-1 population figures used for state revenue distribution — the actuals. Separately, the same office publishes near-term five-year projections each March and 20-year projections (developed with UNLV's Center for Business and Economic Research) each October. These are related but not identical series, built at different times on different baseline assumptions, and they can diverge from each other for the same forecast year.
Two things are worth flagging up front. First, a separate Census Bureau/FRED intercensal series (NVCLAR3POP) puts 2025 Clark County population at roughly 2.41 million — about 40,000 below the state's certified figure — a known, methodology-driven gap between the two data pipelines rather than an error in either one. Second, the state's longer-run 20-year model (developed with UNLV CBER, most recently updated in 2025) projects Clark County reaching approximately 2.92 million by 2040 and 3.23 million by 2060 — figures that provide longer-run context for the same demographic trajectory RRR cites in its investor materials, but sit on a different vintage than the near-term table above.
The Housing Pipeline
Population growth doesn't happen without somewhere for people to live, and the residential pipeline data — independently sourced from SalesTraq, Clark County Comprehensive Planning, and Home Builders Research (a Zonda company), as cited in RRR's own materials — provides useful, verifiable structural context distinct from the population-count debate above.
Actively-selling subdivisions across the valley rose 7.4% year over year to 174 as of March 2026 — a genuine, independently trackable leading indicator of near-term population inflow that moves faster than annual population estimates, which can lag by up to a year. The housing market's risk profile also looks structurally different than it did heading into the last downturn: adjustable-rate mortgages, a major amplifier of the 2008 crisis, made up roughly 25% of the Nevada market in January 2007 versus under 2% in mid-2025. None of that guarantees the current growth cycle continues uninterrupted, but it is a fair, verifiable reason to treat this cycle's housing foundation as meaningfully more stable than the last one that mattered for this market.
Does Population Growth Explain Gaming Revenue Growth?
Here is the test the prior report couldn't run without this data: does population growth in Clark County actually track gaming revenue growth in the locals market? The honest answer is more interesting than a simple yes or no.
On raw levels, the two series look tightly related — and that is the wrong conclusion to draw. Regressing combined North Las Vegas + Boulder Area + Balance of County gaming win against Clark County population, year by year from 2022 to 2025, produces an R² of 0.92 (p = 0.04). That looks like a strong, statistically meaningful relationship. It is not, and it's worth being precise about why, since both series genuinely are increasing — that's exactly the trap. Any two things that are both generally rising over the same few years will look statistically related even with zero real connection between them: plot Clark County population against, say, national Starbucks store counts over the same four years, and you'd get an equally tight-looking fit, for reasons that obviously have nothing to do with each other. This is a formally documented statistical trap — Granger and Newbold demonstrated it rigorously in 1974 using series built by construction to have no real relationship, which still produced significant-looking regressions purely because both trended. A regression on raw levels is really asking "does a higher X tend to come with a higher Y" — and with only a handful of years, both climbing almost every year, the answer is yes almost by construction, regardless of whether any real link exists.
The honest test is on growth rates, not levels — and there, the relationship disappears. The sharper, more useful question was never "is population higher in years when revenue is higher" — with a growing county and a growing market, that was close to guaranteed regardless of any real link. The real question is whether the specific years when population grew faster or slower than its own normal pace also saw gaming revenue move faster or slower than its own normal pace, in the same direction. Testing growth rates instead of levels is exactly how that gets isolated — it strips out the part both series would have done anyway and checks whether what's left moves together. Here, it doesn't: regressing population growth against gaming-revenue growth, year over year, produces an R² of 0.01 (p = 0.92) — no detectable relationship. Running the same test on a larger, monthly-interpolated version of the same data is consistent with the annual result: the underlying series spans 52 months (population linearly interpolated between the certified annual anchors), but because the regression is run on year-over-year growth rates, the first 12 months serve only as the lookback base for computing that first growth figure — leaving 40 usable observations for the regression itself. On that basis: R² of 0.0001, p = 0.95. One honest qualification on that monthly version: because the population side is interpolated, its month-to-month growth path is smooth by construction and carries no genuine monthly information — nearly all the monthly variation sits on the revenue side — so an R² near zero is close to guaranteed by the interpolation itself. The monthly test is best read as a consistency check on the annual finding, not as independent confirmation of it. Concretely: in 2024, population grew slower than normal (1.3%) while gaming revenue spiked (5.8%) on Durango's opening — the opposite of what a real relationship would predict. In 2025, population accelerated (2.3%) while revenue growth decelerated to match it exactly (2.3%) — the two lines up, but only because Durango's supply effect had already worked through the year-over-year comparison base by then, not because population was driving the number in any stable, repeatable way.
What this actually means: 2024 and 2025 tell two different stories, and the difference is instructive. In 2024, Clark County population grew a modest 1.3% while three-area gaming win jumped 5.8% — a gap almost entirely explained by Durango's December 2023 opening, a supply-side event, not a demand-side one, as the prior report established in detail. In 2025, population growth accelerated to 2.3% while gaming-revenue growth decelerated to 2.3% — the two rates converged almost exactly, and per-capita gaming revenue was flat year over year. Read together, the honest interpretation is that 2024's above-trend growth was a Durango story, not a population story, while 2025 looks like a year where the market's growth simply tracked population growth roughly one-for-one.
There is a subtler pattern in the same table worth making explicit, because it anticipates a fair criticism of the regression: with only three growth-rate observations and one of them (2024) dominated by a known supply shock, a skeptic could argue the null result is just one outlier swamping a tiny sample. Look at the two years without an active supply shock, though, and the picture is not random — it is exactly what this report's own demographic framework predicts. In 2023, revenue grew 1.6% against 1.0% population growth; in 2025, 2.3% against 2.3%. In both clean years, revenue growth ran at or modestly above population growth — consistent with the roughly 0.9%/year that headcount plus age-mix composition together support (Scenario A, below), plus ordinary year-to-year noise. In other words, the data is consistent with a market whose baseline growth is demographic and whose deviations from baseline are supply-driven — which is a more precise statement of this report's thesis than "population doesn't matter," and it is the version the rest of this report defends.
| Year | Population | Pop. YoY% | 3-Area GGR | GGR YoY% | GGR / Capita |
|---|---|---|---|---|---|
| 2022 | 2,338,127 | +0.8% | $2,935.0M | +2.5% | $1,255 |
| 2023 | 2,361,285 | +1.0% | $2,981.1M | +1.6% | $1,262 |
| 2024 | 2,392,490 | +1.3% | $3,152.8M | +5.8% | $1,318 |
| 2025 | 2,448,576 | +2.3% | $3,225.9M | +2.3% | $1,317 |
Population and GGR YoY% figures for 2022 are calculated against a 2021 baseline: Clark County population of 2,320,551 (Nevada Dept. of Taxation certified estimate) and combined North Las Vegas + Boulder Area + Balance of County twelve-month gaming win of $2,864.7M (Nevada Gaming Control Board, December 2021 Monthly Revenue Report, "Twelve Months" column for each sub-area).
One honest limitation applies to every version of this test: n=4 annual observations (n=3 for the growth-rate comparison) is too small a sample to prove the absence of a relationship with statistical confidence. To put a concrete number on how small: with three growth-rate observations, only a near-perfect correlation (an R² above roughly 0.99) would register as statistically significant at conventional thresholds — so this test was, by construction, capable of detecting only an essentially perfect relationship, and its null result should be read as "not detected in this sample," not "shown to be absent." That low power is not an oversight; it is the direct, deliberate cost of restricting the analysis to a confound-light window, as discussed at the end of this report — the same short span that keeps the recession and pandemic from contaminating the test is also what limits how much the test can statistically rule out. The two limitations are one tradeoff, and the planned multivariate follow-up with an extended dataset and explicit macro controls is designed to resolve both at once.
A second limitation, equally important: this is an incomplete regression by deliberate design. In econometric terms, an incomplete regression is one in which relevant explanatory variables are omitted — and this report's bivariate model omits employment, housing wealth, new supply, regional income, and every other driver discussed in "So What Does Explain Gaming Growth?" below. That incompleteness means the slope coefficient on population growth cannot be reliably interpreted as population's true marginal effect on revenue; it only establishes whether the two variables move together in this sample once the trend is removed. The null result — they don't — is the correct and intended finding, but it is the finding of an incomplete model. The true marginal effect of population growth, controlling for supply shocks and macroeconomic factors, is precisely what the multivariate follow-up is designed to estimate. What can be said now is that the finding is consistent across every specification tested — annual, monthly-interpolated, and a specific test using only the population 65 and over, below — and a small, thin sample telling the same story multiple ways, alongside the event-based evidence (Durango) and the sub-area cross-section (Henderson), is meaningfully more trustworthy than the regression alone would be, even if it falls short of proof.
A third limitation worth stating formally: 2024 is almost certainly an influential observation in the growth-rate regression. With only one degree of freedom in the growth-rate comparison, any single year with an unusual Y value relative to its X value carries enormous leverage over the estimated slope. Running the regression without 2024 — the year Durango dominated the GGR result — and comparing it against the full-sample result is the standard diagnostic for this, and the informal version of that test is already in the section below ("In 2023, revenue grew 1.6% against 1.0% population growth; in 2025, 2.3% against 2.3%"). The clean-year comparison effectively shows that excluding the Durango year, the remaining observations behave exactly as Scenario A predicts — which confirms 2024 is a high-leverage outlier driven by a documented, dollar-quantified supply event rather than an anomaly in the population-revenue relationship itself. The formal influential-observation sensitivity table (regression with and without 2024, with Cook's distance or simple slope comparison) will be reported explicitly in Part 2 alongside the extended dataset.
RRR's investor materials lead with population growth prominently — Nevada's population growth ranks 4th nationally, Clark County is projected to add 402,000 residents by 2040, and a slide titled "Locations Strategically Positioned Across Valley" states that more than 70% of Clark County's future population growth is located within three miles of a Red Rock-owned property or development site, sourced to Claritas. A reader who has seen that deck could reasonably ask why this report finds no relationship when the company itself is pointing at the same data.
The answer is that RRR's claim and this report's test are not actually measuring the same thing, and the difference matters. RRR's 70%-within-3-miles statistic is a catchment-area claim about where the company chose to buy land relative to where growth is expected to concentrate — it says nothing about whether county-wide revenue moves in step with county-wide population, year to year. This report's regression tests that second, broader, more falsifiable question, because that is the question an investor actually needs answered to know whether demographic growth is doing the work in the P&L today, versus a story about land banked for the future. Those are different claims, tested at different levels of aggregation, and a null result on one does not contradict a true statement about the other. A real estate map showing where future residents will live relative to existing casinos is not the same evidence as a time series showing that revenue accelerated in the years population accelerated — and this report's null result speaks only to the latter.
It's also worth being fair to what RRR's deck is actually for: an investor presentation's population slides are underwriting context for a land-bank and capital-allocation thesis, not a peer-reviewed causal claim about historical revenue drivers. Framing population growth as a tailwind for a six-site Las Vegas Valley development pipeline is a reasonable, standard use of that data. Where this report's findings and RRR's own numbers line up directly, rather than talk past each other, is composition: RRR's own deck shows the 65-and-over cohort growing roughly 3.7x faster than the total population historically (92.6% vs. 25.3%, 2010–2026, per Claritas), projects it to grow more than 3.2x faster over the next five years (16.8% vs. 5.3%, 2026–2031), and shows older consumers consistently allocating a larger share of discretionary income to entertainment (23.5% for the 75+ bracket against an 8.8% average, per BLS). That is the same mechanism this report's shift-share analysis found — a real, if modest, composition effect — presented from the company's side rather than derived independently. The two analyses agree on the part of the story that's actually about spending behavior; they diverge only on the much larger claim that raw population growth alone explains locals-market revenue growth — a claim this report tested directly, and one RRR's deck itself never actually makes.
A Sub-Area Puzzle: Population and Revenue Growth Move in Opposite Directions
The county-level test masks something sharper at the sub-area level. Two of the three NGCB gaming sub-areas map cleanly onto municipal population data: North Las Vegas (the gaming sub-area and the city share the same boundary) and Boulder Area (defined, per NGCB's own classification, as casinos within Henderson city limits — so Henderson's certified population is a direct, high-confidence proxy). The third, Balance of County, does not map to any single incorporated boundary; it is built here as a residual — Clark County total minus Henderson, North Las Vegas, the City of Las Vegas, Boulder City, and Mesquite, minus Laughlin (which NGCB reports as its own separate gaming market, the same reason Mesquite is excluded) and six additional remote, low-density unincorporated communities (Bunkerville, Indian Springs, Moapa, Moapa Valley, Mt. Charleston, and Searchlight) with no meaningful connection to Balance of County casino revenue. What remains is the population of Enterprise, Paradise, Spring Valley, Summerlin, Sunrise Manor, Whitney, and Winchester — the contiguous, urbanized unincorporated communities that actually make up the Las Vegas Valley outside the incorporated cities — plus a small unallocated remainder not broken out by name in the certified series. This is a closer match to the area NGCB's Balance of County revenue figure covers than a simple county-minus-cities residual, though it has not been independently confirmed against Clark County Comprehensive Planning's own small-area geography.
Henderson is the puzzle. It has been the fastest-growing sub-area by population every single year in this dataset, accelerating to 3.7% growth in 2025 — and it has the slowest-growing gaming revenue of the three. If population growth mechanically drove gaming revenue, Boulder Area should be the standout, not the laggard. It is not. Whatever is driving gaming-revenue growth in this market, sub-area population growth alone does not appear to be doing much of the work — supply, competitive dynamics (Green Valley Ranch has been mid-renovation through this period, a real disruption unrelated to demographics), and the composition of who is moving in likely matter more than the headcount. North Las Vegas, by contrast, shows the pattern you would naively expect: modest, steady population growth alongside modest, steady gaming-revenue growth.
It's Not Headcount — It's Composition
The county-level and sub-area findings both point the same direction: simple population count is not doing much explanatory work. RRR's own investor materials — and this author's prior initiation report, in the "equity cash-out migrant" section — have consistently argued something more specific: it's not how many people are moving to Clark County, it's who.
The Nevada State Demographer's Age, Sex, Race, and Hispanic Origin (ASRHO) estimates and projections — an independent series from the same office that produces the certified population and near-term forecast figures used throughout this report — break Clark County's population into five-year age cohorts, with actual estimates back to 2000 and projections through 2044. The chart below draws two specific years from that longer series — 2010 and 2026 — to compare how much each cohort grew over that 16-year stretch; the full 2026–2044 projection from the same series is used later in this section to look forward instead of back.
That is a clean, monotonic gradient: growth intensifies the further up the age ladder you go. It is exactly consistent with an aging-in-place-plus-retiree-inflow story, and it directly supports the mechanism this author's prior report described in detail — the "equity cash-out migrant," typically 50+, arriving from coastal California with substantial home-equity gains and a documented tendency to allocate an outsized share of discretionary income to entertainment. Running the same regression discipline against this narrower cohort doesn't rescue the naive test, for what it's worth: a level regression of 65+ population against gaming revenue still shows the same trend-inflation problem (R² = 0.94), and the differenced version, while directionally a bit stronger than the total-population version, remains statistically inconclusive at this sample size (R² = 0.11, n=3). The value of the age-cohort data here is not a cleaner regression — it's the descriptive case for why the regression comes up empty: the thing actually driving above-trend revenue growth, per RRR's own framing, is a spend-per-person mechanism layered on modest headcount growth, and a headcount-only test was never going to see it.
Why Age-Specific Growth Rates Are the Better Signal
The age gradient above is a data pattern; it's worth being explicit about why it's plausible as more than a correlation worth noting. Three things distinguish an aging population from a merely growing one, as far as locals gaming demand specifically is concerned — plus one real limitation that keeps the case from being fully closed.
Spend share. The most directly quantified mechanism: older residents allocate a much larger share of discretionary income to entertainment. The BLS data behind RRR's own investor materials — used later in this report to build a quantitative estimate — shows 65-74 year-olds spending 13.4% of discretionary income on entertainment and 75+ spending 23.5%, against an 8.8% all-ages average, roughly double the average rate. A resident moving from the 45-54 bracket into the 65+ bracket doesn't just add one more person to the county — they represent a meaningfully larger share of entertainment spending than they did the year before, independent of any change in headcount.
Visit frequency. The locals gaming model, as this author's prior report on this market established, is built on frequency rather than destination visits — RRR's own materials cite 76% of carded slot revenue coming from guests who visit four or more times a month. A working-age resident's gaming visits compete with a 40-hour work week; a retired resident's don't. That is a plausible, if not independently quantified in this report, structural reason the 65+ cohort specifically — not "older" in some vaguer sense — matters more to a frequency-driven business model than it would to a destination-tourism one.
Wealth arbitrage. A meaningful share of Nevada's older in-migrants arrive from higher-cost coastal markets, Southern California in particular, often having sold a home at a substantial gain and relocated to a state with no income tax and a materially lower cost of living. That combination — a lump-sum wealth gain plus lower fixed costs — plausibly supports higher discretionary spending than the household's income alone would suggest. This report carries that mechanism over from the prior report's analysis rather than re-verifying the underlying migration and home-equity data itself.
Is it migration, or aging in place? The mechanisms above all assume something specific: that the growing 65+ cohort represents new spending power entering Clark County. But an alternative explanation would produce the identical age-cohort data — existing residents simply aging past 65 in place, already captured as customers, merely crossing a birthday. The distinction matters enormously for an investor: genuine in-migration is unambiguously incremental demand, while aging in place is a demographic bulge working its way through a population the market already serves. The ASRHO series used throughout this report reports population stocks at each age band, not the sources of change, so it does not answer this question on its face. It can, however, be made to answer it.
Decomposing the Cohort: Where the Older Population Actually Comes From
The standard demographic technique here is a cohort-survival residual: take a five-year age band in a base year, apply mortality rates to project how many should still be alive five years later, and compare that expectation against the actual population of the next-older band. The gap is implied net migration. Applied to Clark County's older cohorts, using a clean pre-pandemic window (2013–2018) and a sensitivity band of survival rates rather than a single point estimate:
| Cohort Transition | 2013 Population | × Survival Rate | = Expected Survivors | 2018 Actual Population | Implied Net Migration |
|---|---|---|---|---|---|
| 50-54 → 55-59 | 130,101 | 0.9724 | 126,505 | 133,098 | +6,593 |
| 55-59 → 60-64 | 122,682 | 0.9584 | 117,579 | 126,831 | +9,252 |
| 60-64 → 65-69 | 103,277 | 0.9394 | 97,016 | 101,568 | +4,552 |
| 65-69 → 70-74 | 90,992 | 0.9150 | 83,261 | 82,376 | −885 |
| 70-74 → 75-79 | 69,167 | 0.8740 | 60,455 | 56,598 | −3,857 |
Source: The "2013 Population" and "2018 Actual Population" columns are both pulled directly from the same underlying series — the Nevada State Demographer's ASRHO (Age, Sex, Race, and Hispanic Origin) estimates, which report Clark County's population by five-year age band. The "Survival Rate" column comes from the Social Security Administration's 2023 Period Life Table (national, both sexes averaged — see "Two methodological cautions on this decomposition," below, for how each rate was derived and sensitivity-tested). "Expected Survivors" and "Implied Net Migration" are both the author's own calculations, not reported figures from any source: the former is the direct product of the two preceding columns, 2013 Population × Survival Rate, rounded to the nearest person; the latter is 2018 Actual Population minus Expected Survivors.
Cohort Transition identifies a five-year age band as recorded in the Nevada State Demographer's 2013 ASRHO estimate (e.g., "50-54") and the corresponding, five-years-older band in the 2018 estimate ("55-59") — the same group of people, five years on.
Expected Survivors is not a reported figure from any data source — it's the author's calculation, shown directly in the table as the product of the two preceding columns (e.g., 130,101 × 0.9724 = 126,505 for the 50-54→55-59 transition). It projects how many of the people counted in the starting age band in 2013 should still be alive and in the next age band by 2018, assuming zero net migration in either direction.
2018 Actual Population is simply the Nevada State Demographer's actual, reported 2018 ASRHO population figure for that next-older age band (e.g., the actual number of 55-59 year-olds counted in Clark County in 2018) — no calculation involved, a direct data point. The gap between this figure and Expected Survivors is what the table attributes to migration: more actual people than expected survivors implies net in-migration; fewer implies net out-migration. See "Two methodological cautions on this decomposition," below, for the specific survival-rate sourcing and sensitivity testing behind the Expected Survivors figures.
Note: The subtotal figures are computed from unrounded intermediate values; the rounded per-transition rows displayed above sum to within ±1 of each subtotal.
The answer is: both, but not where you'd expect. Clark County has substantial, sustained net in-migration in the 50-64 band — and net out-migration among residents already past 65. In other words, the 65+ population is not growing because retirees are moving to Las Vegas. It is growing because people move to Las Vegas in their fifties and early sixties, and then age into the 65+ cohort five to fifteen years later, alongside longtime residents doing the same. The inflow and the spending cohort are separated by roughly a decade.
That's consistent with, without independently confirming, the "equity cash-out migrant" mechanism described in this author's prior report — this report has not tested age-specific data on home-sale-triggered relocation, so the connection is suggestive rather than established. But it changes the shape of the forecast regardless of that mechanism question. The 65+ spending cohort is being fed by a pipeline of people already living in Clark County, which makes the next decade of 65+ growth more predictable than an ongoing-inflow assumption would suggest — and also means it is not self-sustaining indefinitely, since it depends on continued arrivals into the 50-64 band today to supply the 65+ cohort of the late 2030s.
First, this analysis deliberately uses a pre-pandemic window — here is exactly why, with the numbers that make the case. The table above uses 2013–2018, and its five rows sum to a total 50+ residual of +15,655 (rounded to +15,700 in the discussion below). Running the identical cohort-survival method — same technique, same five-year age bands, same SSA-derived survival rates — just shifted five years later to 2019–2024, produces a total 50+ residual of roughly −1,600 instead.
| Window | Total 50+ Residual | Reads As |
|---|---|---|
| 2013–2018 (pre-pandemic) | +15,700 | Substantial net in-migration |
| 2019–2024 (spans pandemic) | −1,600 | Roughly flat, mildly negative |
| Swing between windows | +15,700 − (−1,600) = ≈17,300 | Same method, sharply reduced inflow |
What the "~17,300 swing" actually means: it is not evidence that 17,300 people specifically emigrated, or that Clark County's older population suddenly reversed course between the two windows. It means that the same calculation — comparing how many people should still be alive in each age band against how many actually were, using the SSA-derived survival rates described in "Second," below — collapses from a substantial net inflow to a roughly flat, mildly negative reading once the window is shifted five years later to include 2020–2021. That collapse is the tell. The 2019–2024 window overlaps the pandemic, and COVID-19 caused a real, well-documented spike in excess mortality concentrated among older residents — meaning more people in the older 50+ bands failed to survive to the next five-year mark than the survival-rate assumption expects, for a reason that has nothing to do with migration. Because this method has no way to separate "didn't survive due to a pandemic" from "moved away," it silently books excess pandemic deaths as if they were out-migration, and a swing this large (roughly the same order of magnitude as documented COVID-19 excess mortality among Clark County's older residents) is exactly the signature that would produce. That is precisely why this report uses the 2013–2018 window instead of the more recent one: not because the more recent window is unavailable, but because feeding it into this specific method would produce a confidently-wrong answer rather than a merely imprecise one.
Second, the survival rates applied above are derived directly from the Social Security Administration's published 2023 Period Life Table (national, both sexes averaged) rather than an unsourced approximation — for example, the 60-64→65-69 transition uses a 0.939 survival rate, the average of the SSA's male (0.926) and female (0.953) five-year survival probabilities for that age range. Testing the actual male-only and female-only rates as sensitivity bounds, rather than an arbitrary round-number band: the 50-64 net in-migration holds positive at both extremes (+16,856 at the higher-mortality male rate to +23,937 at the higher-survival female rate, versus +20,397 at the sex-combined average), and the 65+ net out-migration holds negative at both extremes too (−1,602 to −7,880, versus −4,742 at the average) — both headline conclusions are robust to which sex-specific rate is used. The one line item that is not fully robust is the individual 65-69→70-74 transition, which is positive at the male rate and negative at the female rate; it is the 65+ out-migration total, not this single row, that holds up across the sensitivity range. This SSA benchmark is still national, not Nevada-specific, so a fully independent confirmation would use IRS Statistics of Income age-bracketed migration data, which tracks actual year-over-year address changes on tax returns rather than inferring migration as a residual — that cross-check has not been performed here and remains the right next step, particularly since Nevada's own ASRHO model also derives migration residually, leaving some circularity in the above.
▸ How the 0.939 survival rate is calculated from the SSA table
Source: SSA 2023 Period Life Table, "Number of lives" column — survivors from a hypothetical cohort of 100,000 born alive, at each exact age. ssa.gov/oact/STATS/table4c6.html
Step 1 — Average male and female lives at each age (both sexes combined):
| Age | Male lives | Female lives | Both-sexes avg |
|---|---|---|---|
| 60 | 84,544 | 91,080 | 87,812.0 |
| 61 | 83,585 | 90,450 | 87,017.5 |
| 62 | 82,563 | 89,767 | 86,165.0 |
| 63 | 81,473 | 89,029 | 85,251.0 |
| 64 | 80,314 | 88,238 | 84,276.0 |
| Sum 60–64 | 430,521.5 | ||
| 65 | 79,084 | 87,399 | 83,241.5 |
| 66 | 77,783 | 86,508 | 82,145.5 |
| 67 | 76,416 | 85,567 | 80,991.5 |
| 68 | 74,984 | 84,569 | 79,776.5 |
| 69 | 73,486 | 83,509 | 78,497.5 |
| Sum 65–69 | 404,652.5 |
Step 2 — Band-to-band survival rate:
404,652.5 ÷ 430,521.5 = 0.9399 ≈ 0.939
The 0.939 is the ratio of total person-years lived in the 65–69 band to total person-years in the preceding 60–64 band — the correct method for deriving a 5-year band-to-band transition rate from a single-year period life table. Computed independently from primary source data; matches the report's stated figure to four decimal places.
Source: Social Security Administration, Period Life Table, 2023 (as used in the 2026 Trustees Report), ssa.gov/oact/STATS/table4c6.html.
How Far Does This Run? Age-Cohort Projections Through 2044
The historical gradient answers "has this happened." The more useful question for an investor is "how far does this keep going" — and the same ASRHO series that produced the 2010–2026 figures also projects Clark County's age structure forward through 2044.
| Year | 65+ Population | 65+ Cum. Growth from 2026 | 65+ Share of Total | 50+ Population | 50+ Cum. Growth from 2026 | 50+ Share of Total |
|---|---|---|---|---|---|---|
| 2026 | 377,459 | — | 15.4% | 832,713 | — | 33.9% |
| 2031 | 425,031 | +12.6% | 16.6% | 909,806 | +9.3% | 35.6% |
| 2036 | 465,991 | +23.5% | 17.7% | 952,316 | +14.4% | 36.2% |
| 2044 | 515,481 | +36.6% | 18.9% | 1,029,513 | +23.6% | 37.8% |
†2026, 2031, 2036, and 2044 shares-of-total all use the ASRHO series' own directly-reported total-population figures for that year (Nevada State Demographer, 2025 ASRHO Estimates and Projections, 2000–2044 vintage): 2,456,291 / 2,556,081 / 2,628,250 / 2,726,185, respectively. Across the full 2026–2044 window, the 65+ share of Clark County's population rises from 15.4% to 18.9% (passing through 17.7% at 2036), and the 50+ share rises from 33.9% to 37.8% (36.2% at 2036) — both cohorts growing meaningfully faster than the county as a whole, with steady, uninterrupted growth at every checkpoint.
Read against the county-wide forecast established earlier in this report — total population growth decelerating from 1.5% to 1.2% annually through 2030 — the age-cohort trajectory tells a different story: the 65+ share of Clark County's population is projected to rise from roughly 15.4% to 16.6% in just five years, and the cohort keeps compounding at a meaningfully faster rate than the county as a whole through at least 2044. Whatever locals-market growth premium composition contributes, per the quantitative estimate later in this report, it is not a short-lived effect confined to the current decade — it is a structural feature of Clark County's age profile for the foreseeable future.
Putting a Number On It: How Much of RRR's Growth Scenarios Can Demographics Actually Support?
The composition finding above is qualitative — older, higher-spend residents are the fastest-growing group. The more useful question for an investor evaluating RRR's own 3%/5%/7% annual growth scenarios for the total Las Vegas locals gaming market — gross gaming revenue across all operators, not RRR's own company revenue — through 2036 is quantitative: how much of that range can population and age-composition alone actually explain? A simple shift-share decomposition, built entirely from data already sourced in this report, gives a defensible answer.
Start with a spend-intensity weight for the 65+ cohort, built directly from RRR's own BLS-sourced entertainment-spending chart: 65–74 year-olds spend 13.4% of discretionary income on entertainment, 75+ spend 23.5%, against an 8.8% all-ages average. Population-weighting those two brackets against Clark County's actual 65-74/75+ split (from the ASRHO data) puts the 65+ cohort's blended entertainment-spend share at roughly 17.5% — about 2.0x the population average. That multiplier, applied to each cohort's population and held constant, isolates the pure "mix-shift" effect of the population aging, independent of any assumption about incomes rising.
| Basis | Annual % | Cumulative 2026–2036 |
|---|---|---|
| Pure headcount growth (population only) | 0.68% | ~7.0% |
| + Age-composition mix-shift (Scenario A) | 0.88% | ~9.2% |
| + Per-capita income growth within cohorts (Scenario B)† | 4.21% | ~51.0% |
| RRR Low scenario, total locals market (3%/yr)‡ | 3.00% | ~34.4% |
| RRR Mid scenario, total locals market (5%/yr)‡ | 5.00% | ~62.9% |
| RRR High scenario, total locals market (7%/yr)‡ | 7.00% | ~96.7% |
Show full calculation and source trace for each row
1 · Pure headcount growth (population only)
Source: total Clark County population, Nevada State Demographer 2025 ASRHO Estimates and Projections, Clark County "W GQ" table, pages 25–26 of 114.
| Line | Value | Formula / source |
|---|---|---|
| Total population, 2026 | 2,456,291 | ASRHO p.25, "Total" row |
| Total population, 2036 | 2,628,250 | ASRHO p.25, "Total" row |
| Cumulative growth, 2026–2036 | 7.0% | (2,628,250 ÷ 2,456,291) − 1 |
| Annualized (CAGR) | 0.68%/yr | (2,628,250 ÷ 2,456,291)1/10 − 1 |
2 · Age-composition mix-shift (Scenario A)
Method: weight the 65+ cohort at a 2.0× spend intensity and hold non-65+ at 1.0×, then measure how the spend-weighted index grows as the 65+ share rises. Spend multiplier from RRR Q1 2026 deck, p.11 (BLS-sourced: 65–74 spend 13.4% and 75+ spend 23.5% of discretionary income on entertainment vs. an 8.8% all-ages average, population-weighted across ASRHO's 65–74/75+ split). Population figures: ASRHO p.25.
| Line | 2026 | 2036 | Formula / source |
|---|---|---|---|
| 65+ population | 377,459 | 465,991 | ASRHO p.25, "65 Years and Over" row |
| Non-65+ population | 2,078,832 | 2,162,259 | Author: Total − 65+ |
| Spend-weighted index = (65+ × 2.0) + non-65+ | 2,833,750 | 3,094,241 | 2026: (377,459×2.0)+2,078,832 2036: (465,991×2.0)+2,162,259 |
| Cumulative growth, 2026–2036 | 9.2% | (3,094,241 ÷ 2,833,750) − 1 | |
| Annualized (CAGR) | 0.88%/yr | (3,094,241 ÷ 2,833,750)1/10 − 1 | |
3 · Per-capita income growth within cohorts (Scenario B)
Method: to the 2026–2031 mix-shift rate, add annualized within-cohort income growth, then extrapolate the combined rate flat through 2036. Population figures: ASRHO p.25. Income growth: Claritas-sourced 16.8% for the 65+ cohort, 2026–2031 (RRR Q1 2026 deck, p.10) — the single externally-sourced input, subject to the circularity caveat below.
| Step | Result | Formula / source |
|---|---|---|
| Index2026 | 2,833,750 | (377,459 × 2.0) + 2,078,832 · ASRHO p.25 |
| Index2031 | 2,981,112 | (425,031 × 2.0) + 2,131,050 · ASRHO p.25 |
| Step 1 — mix-shift rate (2026–2031) | 1.02%/yr | (2,981,112 ÷ 2,833,750)1/5 − 1 |
| Step 2 — income growth, annualized | 3.15%/yr | (1 + 0.168)1/5 − 1 · Claritas, RRR deck p.10 |
| Step 3 — combine (multiplicative) | 4.21%/yr | (1.0102 × 1.0315) − 1 |
| Step 4 — extrapolate flat, 10 yrs | ~51.0% cumulative | (1.0421)10 − 1 |
▸ Why the Scenario B combination works the way it does
The step-by-step figures and their sources are in the Scenario B table above (all population inputs from ASRHO page 25 of 114; income growth from RRR Q1 2026 deck, page 10). The notes below explain the two methodological choices in that calculation that aren't self-evident from the numbers alone.
What (1.0102 × 1.0315) − 1 means in plain English: Each factor converts a growth rate into a "for every $1.00, you end up with $X" multiplier. So 1.0102 means "the age-mix shift alone turns $1.00 into $1.0102," and 1.0315 means "income growth alone turns $1.00 into $1.0315." Multiplying them together — $1.0102 growing by a further 3.15% — gives $1.0421, meaning both effects happening simultaneously on the same dollar of economic base produce $1.0421. Subtracting the original $1.00 gives the net combined growth rate of 4.21%. The reason to multiply rather than simply add (1.02% + 3.15% = 4.17%) is that the income growth applies to a cohort that is already slightly larger due to the mix-shift — so the income effect runs on a bigger base than if calculated in isolation. The difference is small (0.04pp), but multiplication is the arithmetically correct form.
Why multiplicative in Step 3: Combining two growth rates as (1+r₁)×(1+r₂)−1 correctly captures the cross-term (r₁×r₂ = 0.032%), which additive combination (r₁+r₂ = 4.17%) misses. The difference is small but the multiplicative form is arithmetically correct.
Why extrapolate flat in Step 4: Scenario B is built from 2026–2031 inputs because Claritas's income-growth figure is only defined for that window. No independent 2031–2036 income forecast exists; extending the 2026–2031 combined rate flat through 2036 is the disclosed assumption. This means Scenario B's ~51.0% is best read as "the 2026–2031 combined rate, compounded for 10 years" rather than a true 10-year independent estimate.
†Headcount and Scenario A reflect the full 2026–2036 ASRHO trajectory; Scenario B is constructed differently, using only 2026–2031 inputs before extrapolating flat through 2036 — see the calculation detail above for the full construction and the caveat below for reliability concerns.
‡"RRR Low/Mid/High" are RRR's own disclosed scenarios for growth of the total Las Vegas locals gaming market — gross gaming revenue across all operators, not RRR's own company revenue — per the "Locals Gross Gaming Revenue Potential" chart in RRR's Q1 2026 investor presentation (page 29), where RRR frames the $3.2B 2026(a) starting point as roughly 1.9% of Southern Nevada personal income ($3.2B of gaming revenue as a share of $167B of personal income). This report's headcount, mix-shift, and income-growth rows above are measured on that same market-wide basis for direct comparability.
Each bar above is a complete, standalone cumulative estimate through 2036 under its own full set of assumptions — not an incremental slice meant to be added to the bar before it. The second bar already contains all of the first bar's headcount effect plus the age-mix shift on top of it; the third bar already contains both of those plus the income-growth assumption. Reading the chart left to right shows how the estimate builds as each additional assumption is layered in, but the bars themselves should never be summed. The read is nuanced in a useful way. On the conservative basis — population growth plus the age-mix shift toward higher-spend residents, with spend intensity per cohort held flat — demographics explain roughly 9% cumulative growth through 2036, about a quarter of even RRR's own low-end 3%/year scenario (34% cumulative). That is the high-confidence number: it rests only on population data and a spend-intensity ratio built from RRR's own published BLS chart, nothing more speculative than that. It means the bulk of even RRR's most conservative growth case has to come from something other than headcount and age-mix alone — share gains, the company's land-bank development pipeline, price or hold-rate increases, or per-capita spend growth beyond simple demographic composition, which this report's own Scenario B attempts to quantify below, though with the independence caveat noted there.
That borrowed figure is what deserves scrutiny. RRR's investor deck separately discloses 65+ population growth of 16.8% for the identical window — a different, narrower geography than Clark County's ASRHO figure, and an exact match to the deck's own income-growth number. That raises an unresolved question: is Claritas modeling 65+ income growth independently, or mechanically tying it to its own population projection for that cohort? If the latter, the income figure adds nothing beyond Scenario A — it would restate the population figure rather than show genuine spending growth. Until RRR or Claritas confirms independence, this report treats the resulting 4.21% figure as unverified and does not rely on it as a standalone conclusion.
One further note: 4.21% is built only from 2026–2031 inputs, then held flat through 2036, since Claritas has not published a 2031–2036 projection to blend against.
Scenario A holds the 65+ cohort's 2.0x entertainment-spend multiplier constant through 2036 — flagged elsewhere in this report as the single biggest open question in the age-vs-cohort-effect discussion (see "Where This Report Could Be Wrong"). That risk is real, but it has not, until now, been quantified. Because the mix-shift effect is linear in the multiplier by construction — population shares are fixed, and the multiplier only rescales the 65+ bracket's weight — the sensitivity can be solved exactly from two figures already published in this report: pure headcount growth (0.68%/yr, ~7.0% cumulative, which is what the model returns at a multiplier of 1.0x, i.e., zero spend premium) and Scenario A itself (0.88%/yr, ~9.2% cumulative, at the current 2.0x multiplier).
| 65+ Spend Multiplier | Interpretation | Annual % | Cumulative |
|---|---|---|---|
| 2.0x | Full age effect persists (Scenario A, as published) | 0.88% | ~9.2% |
| 1.5x | Partial cohort erosion — premium roughly halves | 0.79% | ~8.2% |
| 1.0x | Full cohort effect — premium fully erodes to average | 0.68% | ~7.0% |
Two things follow from this, and they cut in different directions. First, the critique that this report's forward-looking numbers rest on an unstress-tested assumption is fair, and this table is the fix. Second, the range this produces is narrower than "predictive validity drops to near zero" would suggest: because the mix-shift effect is a small overlay (0.20 points/year at its maximum) on top of pure headcount growth, even the harshest cohort-effect scenario — the 65+ premium vanishing entirely — still leaves 0.68%/yr and ~7.0% cumulative on the table, not zero. What the age-vs-cohort risk actually threatens is the marginal ~2.2 points of cumulative growth between the 7.0% floor and the 9.2% headline figure, not the demographic case in its entirety. This is a genuine sensitivity, worth taking seriously, and it is also a smaller one than the qualitative framing implied.
One terminology note, for precision: describing the flat-multiplier assumption as introducing "omitted variable bias" overstates the econometric claim. Omitted variable bias is a specific concept — it describes bias in an estimated regression coefficient caused by a correlated variable left out of that regression. This report's shift-share decomposition is an accounting identity built on an assumed, not estimated, weight; the more precise framing is parameter instability or structural-break risk in the multiplier itself, which is what the sensitivity table above tests directly.
The practical takeaway for an investor: even RRR's low-end growth scenario requires the least help from non-demographic drivers of the three cases, but "least" is still most of it — demographics alone explain barely a quarter of the low-end case's cumulative growth. Reaching any of RRR's scenarios, low included, requires either accepting Scenario B's income-growth assumption at face value — which, per the caveat above, deserves a direct question to the company before being relied upon — or crediting a meaningfully larger share of the growth to non-demographic drivers: RRR's land-bank development pipeline, gaming market share gains against Boyd and smaller operators, or price and hold-rate increases. That is a sharper, more falsifiable framing than "the demographic tailwind supports the growth thesis" — it identifies specifically which lever needs to work, and how hard, to hit each scenario.
What RRR Tells Investors — and What Independent Data Shows
RRR's Q1 2026 investor presentation devotes several slides to exactly this demographic case, and it is worth testing those specific claims against the independent data assembled above — not because there is reason to expect they don't hold up, but because that is the standard this report applies to every claim examined, company materials included.
Three Claims, Checked Against Independent Data
RRR's presentation shows Clark County population growing by roughly 402,000 residents from the mid-2020s to 2040, sourced to the U.S. Census Bureau and UNLV CBER, with a footnote specifying the CBER estimates "have been updated to reflect current Clark County population estimate." That disclosure matters — it means the figure is a transparently rebased version of CBER's published trajectory, not a stale or cherry-picked number.
On the 65-and-over growth chart, RRR cites Claritas — a proprietary demographic vendor with its own cohort methodology, different from the Nevada State Demographer's ASRHO model used elsewhere in this report. The independent ASRHO data shows 67.0% growth for the 65+ cohort versus RRR's cited 92.6%, while the total-population figure (25.4% independent vs. 25.3% RRR) is an almost exact match. That is best read as two different, both-legitimate demographic vendors producing different cohort-growth estimates — a reminder that "population data" is not one number from one source, not evidence that either figure is wrong.
RRR's materials actually make two distinct proximity claims, and they're worth separating rather than treating as one "radius strategy," since they serve different arguments and rest on different data.
Claim 1 — current density. RRR's own SEC filings state the figure directly: the company's FY2025 10-K describes its properties as "conveniently located throughout the Las Vegas valley," with "over 90% of the Las Vegas population... located within five miles of one of our gaming facilities" — language that has appeared consistently in RRR's 10-K filings for years. This is a claim about today's existing customer base, not future growth — it functions as a moat argument, and it is the geographic mechanism behind the 76%-of-carded-slot-revenue-from-4-plus-monthly-visitors statistic: convenience density drives visit frequency, and visit frequency drives the recurring-revenue profile RRR describes as closer to a subscription business than to destination gaming.
Claim 2 — future growth capture. Separately, RRR states that more than 70% of Clark County's future population growth is expected to occur within three miles of an RRR-owned property or development site — a claim sourced to Claritas in RRR's own investor presentation, and a parcel-level claim this report has no independent way to verify directly, but a materially more useful one for this report's purposes than Claim 1, because it is testable, site by site, as RRR's land bank actually develops. Durango is the completed test case; the remaining six entitled parcels (Cactus/Las Vegas Boulevard, Wild Wild West, Losee, Inspirada, Town Center, and Skye Canyon) are future ones. If the population-versus-revenue disconnect this report documents at the county and sub-area level — most visibly, Henderson's fastest-population-growth-but-slowest-revenue-growth pattern — persists at the individual-parcel level as these sites open, that would be a real-world confirmation of this report's central finding: that proximity to population growth alone doesn't reliably predict gaming-revenue growth. If specific sites instead outperform that pattern, that would be informative in the other direction, and worth revisiting as Durango's post-stabilization data and the next site's opening both become available.
The overall read on RRR's demographic materials: well-sourced, and where this report can independently check specific figures, they either match closely or are transparently reconciled to a named, different data vendor. Nothing examined here suggests the company's population claims are inflated or unreliable.
Boyd's Different Story
Boyd Gaming's investor materials do not include a comparable demographic slide deck — this author checked Boyd's investor relations site directly and found no dedicated population-growth analysis of the kind RRR publishes each quarter. What Boyd does offer is narrower and more site-specific, and it grew slightly more detailed across two consecutive earnings calls. On the Q4 2025 call (February 2026), CEO Keith Smith noted that Cadence — the master-planned community adjacent to Boyd's new Cadence Crossing casino in Henderson — ranked third in the nation in 2025 for home sales. On the following Q1 2026 call, after Cadence Crossing opened in March, Smith added that Cadence is planned for more than 12,000 homes at full build-out. Both are real, specific, checkable claims — but both are about a single project, not a market-wide demographic case.
Boyd's most formal disclosure document shows an even thinner treatment. The company's 10-K business description of its Las Vegas Locals segment has used nearly identical boilerplate language across at least three consecutive fiscal years — the FY2022, FY2024, and FY2025 filings all describe Las Vegas as having "strong demographics that include a large population of retirees and other active gaming customers," a one-sentence characterization with no population figures, no growth rate, and no citation to Census, CBER, or any demographic vendor. One change across those filings is worth noting: the FY2025 10-K (filed February 2026) swaps the prior two years' reference to positive trends in "visitation" for positive trends in "average weekly wage growth" — a small wording change, but one consistent with the destination-travel softness at The Orleans and Suncoast disclosed on Boyd's own recent earnings calls, and with a broader shift toward wage-and-employment framing, rather than population or visitation framing, that shows up across both companies' more recent commentary on the locals market's health.
That is a real, useful contrast for understanding how the two companies communicate the same underlying opportunity. RRR builds a multi-slide, multi-vendor demographic apparatus (Census Bureau, UNLV CBER, Claritas) as a standing feature of every quarterly deck. Boyd's version of the same argument is a single project-level data point, repeated and lightly updated quarter to quarter, sitting alongside one boilerplate sentence in its annual SEC filing. Neither approach is more or less valid — they reflect different investor-communication styles and different portfolio structures — but a reader comparing "what RRR says about population" to "what Boyd says about population" is comparing a comprehensive analytical framework to a single anecdote, not two versions of the same disclosure. Notably, Cadence Crossing sits in the sub-area (Henderson/Boulder Area) with the fastest population growth and the slowest gaming-revenue growth of the three NGCB sub-areas in this dataset — it is too early to know whether the property will buck that pattern or extend it.
What This Analysis Can't Tell You
Two limitations here are worth stating directly rather than leaving implicit, because they're different from the small-sample caveat already noted throughout this report, and arguably more important.
Why the historical series wasn't extended
The two red bands are exactly the periods with the most real year-to-year variance in population growth — which is also exactly why they're excluded here rather than included. Each band has its own dominant, non-demographic shock (a national recession and a bank-driven regional bust; a government-ordered shutdown) running through both series at once, which is what would confound a simple bivariate test rather than clarify it. This report's window (green) is comparatively uneventful by design — low variance, but also low confound risk — with the multivariate model in "Coming Next" built specifically to bring the red bands back in with explicit controls instead of leaving them out entirely.
The obvious response to "n=4 is too small" is "get more data." That response doesn't actually work here, and it's worth explaining why precisely, since the reasoning is sharper than "the sample is too small." Population is a slow-moving variable — it only shows real year-to-year variance during major macro disruptions. In Nevada's recent history, that means essentially two windows: the 2008–2011 recession and the 2020–2021 pandemic. Both windows have a third factor running through them that would confound a simple population-versus-revenue test, and the risk isn't just added noise — it's a specific chance of a misleading result in either direction. In 2008–2011, the recession itself plausibly suppressed both population growth (people generally don't relocate to a state whose housing market and economy are collapsing) and gambling spending (mass unemployment, wealth destruction, plus Station Casinos' own Chapter 11 restructuring) at the same time. Adding that period to a simple regression could easily manufacture an apparent positive relationship between population and revenue — both series moving down together — that isn't causal at all; it would really be two things separately responding to the recession, not one driving the other. That's a more insidious outcome than a null result, because it would look like confirming evidence while actually being a comment on the recession. 2020 has the opposite shape but the same underlying problem: population barely moved while gaming revenue fell to near-zero for weeks under a government closure order, an extreme, one-sided data point that would distort a small-sample regression without testing the population-revenue link at all. Properly using either period would require a model that explicitly controls for the recession or the closure as its own variable, not simply more (population, revenue) pairs fed into the same bivariate test — which is exactly the multivariate project already scoped as a future piece, not an extension of this one. A bigger sample built from those specific years, without that control, would look more rigorous — more data points, a chart with more history — while being less trustworthy than the honest, small-sample finding this report settles for instead.
What a clean regression still wouldn't tell you about the future
It's also worth being explicit that even a larger, cleaner, statistically airtight version of this regression would have real limits as a forecasting tool — limits that have nothing to do with sample size.
First, a regression describes the past under a specific set of conditions, not a durable law. "No detectable relationship, 2022–2025" is a finding about a window dominated by Durango's opening — a supply-side event. It says nothing about whether population would matter in a different four years without a comparable supply shock sitting on top of it. Second, regression can't see structural breaks it hasn't been trained on. Durango, Cadence Crossing, whatever RRR eventually builds on its land bank — none of these appear as a "population" variable at all; they're supply shocks, and this market's history (as the prior report on this market established) has been driven more by them than by demographics. A population regression, however well-fit to history, is blind to the next one of these until after it happens. Third, correlation — even a strong, statistically solid one — still isn't a mechanism. This report's own central finding illustrates the point: total population, the "obvious" variable, turned out to be the wrong one, and it took an entirely different data source (age cohorts) to find a more plausible mechanism. A cleaner regression on the wrong variable is still a regression on the wrong variable. Fourth, the variable itself is drifting, not stable — Henderson's population growth alone accelerated from 2.2% to 2.6% to 3.7% over just three years in this dataset. Any historical fit implicitly assumes the future resembles the estimation period, and a population growing at an unprecedented pace for this specific market is exactly the situation where that assumption is riskiest.
None of this is an argument against the analysis in this report — the tests run here are honest, correctly specified, and consistent across every version tried. It's an argument for being precise about what a regression, done well, actually buys you: a defensible statement about what happened in a specific window, not a predictive model of what RRR or Boyd do next. The right tool for that is the multivariate, macro-factor analysis (employment, housing prices, and regional economic indicators) flagged as a separate future piece — and even that, done as well as it can be done, will face the same four limits described above.
So What Does Explain Gaming Growth?
If population and age composition together explain only a modest fraction of this market's growth, it's worth stating plainly what's actually carrying the rest — ranked by how well-evidenced each candidate is across this report and the prior one, not by how often it gets cited in investor materials.
| Driver | Evidence Strength | Quantified Effect | Status |
|---|---|---|---|
| New supply (Durango) | Strong | ~4.5pp gap in 2024 alone | Confirmed, prior report |
| Age composition (mix-shift) | Modest, quantified | ~0.2pp/yr on top of 0.68% headcount (~0.88%/yr combined, ~9.2% cumulative by 2036) | Confirmed, small |
| Per-capita spend growth within cohorts | Plausible | ~4.21%/yr if valid (Scenario B) | Unresolved — see verification caveat |
| Price / hold-rate changes | Plausible | Not quantified | Untested in this report |
| Capital reinvestment in existing properties | Plausible, cited by RRR | Not isolated | Untested in this report |
| Loyalty / engagement of existing customers | Plausible, cited by RRR | Not isolated | Untested in this report |
| Macro factors (wages, employment, housing wealth) | Plausible | Not tested | Deferred to future report |
New supply is the best-evidenced driver in this entire body of work. 2024's 5.8% GGR growth against only 1.3% population growth is a gap of roughly 4.5 points that population and composition together can't explain (composition alone is worth about 0.2 points a year). The disconnect is sharper still when narrowed to the specific sub-area where Durango actually opened: Balance of County population grew just 1.1% in 2024 (Nevada Dept. of Taxation certified estimates, author's residual calculation — Clark County total less Henderson, North Las Vegas, the City of Las Vegas, Boulder City, Mesquite, Laughlin, and six remote unincorporated communities), while Balance of County GGR alone grew 10.5% (NGCB monthly extraction, author's calculations) — a gap of roughly 9.4 points, more than double the county-wide comparison, in the exact geography Durango's new supply landed in. The prior report on this market traced that gap directly to Durango's December 2023 opening — a new $800 million property in an underserved corridor. That's not "more people"; it's more capacity and a new amenity drawing spend that either didn't exist before or was previously going to competitors or the Strip.
Age composition is real but small. The shift-share section above put the combined demographic effect — headcount growth plus the mix-shift toward higher-spend residents — at roughly 0.9% a year, of which the mix-shift itself contributes about 0.2 points. Meaningfully positive, and better-evidenced than population alone, but nowhere near enough to explain most individual years on its own.
Per-capita spend growth is the largest plausible lever this report identified, and the least resolved. Scenario B — if 65+ spending genuinely grows in line with RRR's disclosed income projections — implies roughly 4.21% annual growth, close to RRR's mid-case total-market scenario. But RRR's own deck discloses identical 16.8% figures for 65+ population growth and 65+ income growth over that same 2026–2031 window — and it's unresolved whether that's genuine, independently modeled income growth or the income figure mechanically tracking population. This is the report's single most consequential open question: the best candidate for closing the gap between RRR's low and mid cases, but unverified rather than untested — worth a direct question to RRR investor relations or Claritas before relying on it.
The rest — price, capital reinvestment, loyalty-driven engagement, and broader macro conditions — are all plausible and none of them have been tested here. RRR's own materials cite over $6B invested across its portfolio with no deferred maintenance capital, and an extensive loyalty database converting proximity into the 76%-from-4-plus-monthly-visits statistic discussed earlier in this report — both are real, cited growth mechanisms that this report has not isolated or quantified. Wages, employment, and housing wealth are addressed nowhere in this report by design, and are the subject of the future macro-factor piece already flagged in the previous section.
One nuance worth stating plainly: this isn't a case of population explaining nothing in every year. 2023 (1.6% GGR growth against 1.0% population growth) and 2025 (2.3% against 2.3%) both track reasonably close to population growth, with only a modest unexplained residual even after accounting for composition. It's specifically 2024 that breaks the pattern, and it breaks it by enough that a four-year regression can't distinguish a real, small population effect from noise. The honest summary isn't "population explains nothing" — it's that population and composition together explain a little, and neither comes close to explaining 2024, which needed a new casino to close the gap.
What This Means
Population growth is real, it is measurable, and it is decelerating on a percentage basis even as the state's own certified data shows it running slightly ahead of official near-term forecasts. It is not, on the evidence assembled here, a simple, mechanical explanation for locals-market gaming-revenue growth — the relationship disappears entirely once the shared upward trend is stripped out of both series, a finding that holds up across every specification this report tested, including a larger monthly sample and a narrower, more theoretically appropriate age-cohort test.
What the data does support is a more specific, more interesting story than "more residents, more revenue": the population growing fastest is the one with the most discretionary income to spend on entertainment — fed not by retirees moving to Las Vegas, but by households arriving in their fifties and early sixties, partly via a home-equity arbitrage from coastal California, who then age into the high-spend 65+ cohort a decade later. Layered on top of that is a submarket-by-submarket pattern that does not track headcount growth in any simple way — Henderson's population is growing faster than anywhere else in the county and its gaming revenue is growing the slowest. That is the finding worth building an investment thesis around, and it is consistent with, not contrary to, what RRR itself is telling investors.
The honest caveat running through this entire report is sample size. Four years of comparable NGCB sub-area data is enough to establish a Durango baseline, as the prior report did — it is not enough to prove or disprove a population-revenue relationship with real statistical confidence. See the previous section for why that history wasn't simply extended, and for the deeper limits of what a regression like this one can and can't tell you going forward.
Where This Report Could Be Wrong
Every claim in this report has been tested against primary sources — but the report's own methods and conclusions haven't yet been tested against outside research that might contradict them. That's a different, and arguably more important, kind of check. Three findings from independent research complicate parts of this report; a fourth is a methodological gap worth naming directly rather than leaving implicit.
1. Las Vegas leads the nation in gross retiree inflow — which complicates, but doesn't overturn, the net out-migration finding
A February 2026 study by HireAHelper, based on relocation data from PGM Solutions, found the city of Las Vegas ranked first in the country for inbound movers aged 65 and over in 2025 — 7,854 retirees, ahead of Tucson (7,627) and Houston (7,287). Read on its own, that looks like a direct contradiction of this report's cohort-survival finding of net out-migration among Clark County residents past 65.
It isn't, but the reconciliation matters. The same study found Florida led the nation in gross inbound retirees (45,696) — and also led in gross outbound (44,881), netting just 815 new residents. On net state-level retiree gains specifically, the study's own reporting notes Nevada "ranked relatively low," well outside the leaders (South Carolina, Texas, and North Carolina topped that list). A place can lead the country in gross retiree inflow and still show a small or negative net gain if outflow is comparably large. That is directionally consistent with what this report's cohort-survival method found — high churn, not low retiree interest — but it is not a precise match, and this report cannot reconcile the two data sources exactly without a Nevada-specific gross-versus-net breakdown, which the HireAHelper study does not provide at the state level. Read together: Las Vegas is genuinely a major retiree draw, and it also loses a large number of retirees to outflow — the net figure alone obscures that churn.
2. A more serious question: is the 65+ spending premium an age effect, or a Baby Boomer cohort effect?
This is the finding most worth taking seriously, because it doesn't just complicate a number — it questions an assumption baked into the report's entire forward-looking section. The shift-share decomposition and the 2044 population-cohort forecast both treat the 2.0x entertainment-spend multiplier for the 65+ cohort as a fixed structural parameter — something that will apply to whoever is 65+ in 2030, 2036, or 2044, not just to the people who are 65+ today.
Independent research raises a real doubt about that assumption. A widely cited casino-visitation trend survey found 79% of Baby Boomers visited a land-based casino in the past 12 months, versus 69% of Gen X and 54% of Millennials — consistent with this report's premise. But TransUnion's Q1 2025 US Betting Report found Baby Boomer and Gen X gambling engagement rising in late 2024 while Millennial engagement fell, and multiple industry surveys describe Millennials and Gen Z as favoring mobile and online formats over physical casino visits. That is the classic demographic identification problem, applied directly to this report's thesis: is high land-based casino spend among the 65+ cohort an age effect — something that happens to most people as they age into that bracket, regardless of generation, which would make the multiplier durable as new cohorts turn 65 — or a cohort effect specific to how Baby Boomers relate to physical casinos, which would not carry forward as Gen X and eventually Millennials age into 65+ over this report's own forecast horizon?
This is not a hypothetical concern this report is raising in isolation. A bear-case summary of Red Rock Resorts from October 2025 names "younger generations' shifting entertainment preferences" directly as a demographic headwind risk to the company's thesis. If the age effect this report assumes is really a cohort effect, the 2.0x multiplier — and every downstream figure built on it, including the 2036 and 2044 cohort forecasts and both shift-share scenarios — would shrink as today's 65+ Boomers are gradually replaced by Gen X and Millennial retirees with different, less land-based-casino-oriented habits. That would make the demographic tailwind this report identifies smaller and more time-limited than the report currently implies, not just smaller than RRR's own market growth scenarios.
Putting a date on the risk: when does Gen X actually start entering the 65+ bracket?
Standard generational birth-year bounds (Pew Research) put Baby Boomers at 1946–1964 and Generation X at 1965–1980. That produces a hard, calculable pivot: 2029 is the last possible year any Baby Boomer turns 65, and 2030 is the first possible year any Gen Xer does. Mapped against this report's own forecast checkpoints:
| Forecast Year | Birth Year of Newest 65th Birthday | Generation |
|---|---|---|
| 2026 | 1961 | Baby Boomer |
| 2031 | 1966 | Gen X — but only 2 of Gen X's 16 birth-years have crossed in |
| 2036 | 1971 | Gen X — 7 of 16 birth-years in |
| 2044 | 1979 | Gen X — 15 of 16 birth-years in |
That timeline cuts two ways for this report. It's favorable for the near-term figures: Scenario A and Scenario B are both built on the 2026–2031 window, which — per the table above — is a population that is still almost entirely Baby Boomer. The age-versus-cohort-effect risk described above is real, but it is a 2036-and-beyond exposure, not a threat to this report's 5-year numbers.
It cuts the other way for the migration-decomposition finding, though. The five cohort-transition rows in "Decomposing the Cohort" above — ages 50-54 through 70-74, as measured in 2013 — correspond to birth years 1939 through 1963: entirely Baby Boomers and the Silent Generation, with zero Gen X representation. The +20,397 net in-migration this report measured in the 50-64 band is, by construction, a Baby Boomer-specific data point. Whether Gen X in-migrants of the same age arrive in Clark County at a similar rate is a distinct, untested assumption layered on top of the age-versus-cohort-effect question above, not a resolution of it — and it is the same open question the next report in this series, using an extended dataset and IRS Statistics of Income data, would be positioned to actually test.
3. A methodological note: differencing is a reasonable simplification, not the state of the art
This report tests population against gaming revenue by comparing year-over-year growth rates rather than raw levels, which correctly avoids the spurious-trend trap described earlier. The more rigorous econometric tool for testing whether two trending series share a genuine long-run relationship is cointegration testing (the Engle-Granger or Johansen methods), which can detect a real equilibrium relationship between two non-stationary series even when their short-run differences appear uncorrelated — something simple differencing can miss entirely. This report does not use cointegration testing because it requires substantially more observations than are available here; with n=4, it is not a matter of imprecision but of infeasibility. That is a fair characterization for a reader with econometrics training to make, and it is noted here directly rather than left for a critic to point out.
Coming Next
This report ruled out simple headcount as the explanation for this market's growth and found composition to be a real but modest contributor. That leaves the larger share of the growth story open — and the ranked table above points directly at what a follow-on report needs to cover.
The next piece in this series will build a proper multivariate model of locals-market GGR growth, rather than the single-variable regressions used throughout this report. That means extending the historical NGCB dataset back through the 2008–2011 recession and the 2020–2021 pandemic — deliberately not done here, for the confounding reasons explained above — but this time with explicit control variables for the recession and the government-mandated closures, so those periods can finally contribute real evidence instead of noise or false signal. Alongside that extended dataset, the model will incorporate Las Vegas MSA employment and unemployment data, regional housing-price indices (FHFA/Case-Shiller), consumer confidence measures, and at least one regional leading indicator — Port of Los Angeles and Long Beach inbound container volume, tested explicitly for a lag structure against Southern California migration and visitation into Clark County, rather than assumed as a same-period correlation. One methodological commitment worth stating plainly in advance: simply lengthening the dataset without those controls would not be an improvement on this report — it would risk manufacturing an apparent population-revenue relationship out of the recession and pandemic years that is really just two series responding separately to the same shock, which is a worse and more misleading result than the honest null finding presented here.
That report will also carry over five specific open items from this one: resolving whether RRR's disclosed 65+ income-growth projection is modeled independently of its population projection, which determines whether this report's Scenario B (~4.21%/year) is real evidence or a restated population figure; confirming this report's cohort-survival migration decomposition — the finding that Clark County's inflow arrives at ages 50–64 rather than 65+ — against IRS Statistics of Income age-bracketed migration data, which measures actual address changes rather than inferring migration as a residual; quantifying the price, capital-reinvestment, and loyalty-engagement effects this report named but did not test; independently verifying the Balance of County population approximation against Clark County Comprehensive Planning's own small-area geography; and reporting a formal influential-observation sensitivity table (regression with and without 2024 — the Durango year — with Cook's distance and slope comparison) to formally establish what the clean-years analysis in this report demonstrates informally: that 2024 is a high-leverage supply-shock observation, not an anomaly in the population-revenue relationship itself.
