Why Multifamily Wins: Five Decades of NCREIF Data on Returns, Risk, and the Arithmetic of Housing Scarcity
Every property sector has a story. Office had the corporate century. Retail had the suburban consumer. Industrial has the container ship and the shopping cart that lives on your phone. Hotels have the business cycle wearing a name tag.
Multifamily has something better than a story. It has a distribution.
Pull the quarterly return series from the NCREIF Property Index, the longest continuously maintained record of institutional real estate performance in the United States, running since the first quarter of 1978, and compute the statistics an allocator actually cares about. Multifamily does not win by having the single highest compound return in every window. It wins the way a great pension asset is supposed to win: high compounding, low dispersion, a thin left tail, and a demand base that regenerates itself every year regardless of what the Federal Reserve is doing. It is the sector where the mean is good and the distribution around the mean is better.
This paper is the full case. We will go through the dataset itself, the inventory that underlies it, the compound growth rates, the standard deviations, the ratio between the two, the extremes, the frequency of negative quarters, and the behavior of the sector inside every major drawdown since Jimmy Carter was president. Then we will do the more important work: explaining why the distribution looks the way it does, because a statistical edge you cannot explain is an edge you cannot trust. The explanation runs through lease microstructure, tenant granularity, government-sponsored debt, occupancy physics, and the demand-side engine of the next two decades, a single-family housing shortfall measured in the millions of units.
A note before we begin. The current index composition, sector returns, cap rates, and NOI figures in this paper are taken directly from NCREIF's published NPI Flash Report and press release for the first quarter of 2026 (released April 25, 2026), the most recent print available as of this writing. The long-run since-inception statistics are drawn from the NPI quarterly history and NCREIF's published sector research, rounded to the precision the appraisal-based data actually supports; anyone with NCREIF query access can reproduce them, and a methodology section at the end states the conventions. The point of this paper is not the second decimal. It is the shape of the evidence, which is robust to any reasonable set of conventions.
1. The Dataset: What NCREIF Is and Why It Is the Right Court for This Trial
The National Council of Real Estate Investment Fiduciaries collects property-level operating and valuation data from institutional investment managers (pension fund advisors, open-end core funds, insurance companies) and aggregates it into the NCREIF Property Index (NPI). Three features make it the right instrument for this question.
First, longevity. The index begins in 1978Q1. That gives us roughly 188 quarters, 47 years, spanning the Volcker rate shock, the 1986 tax reform, the S&L collapse, the 1990s expansion, the dot-com recession, the housing bubble, the Global Financial Crisis, the longest expansion in American history, a pandemic, the fastest hiking cycle since 1981, and the 2022 to 2024 valuation reset. Any sector statistic that survives that gauntlet is not a regime artifact.
Second, quality control. The NPI is unlevered, held at the property level, appraisal-based, and restricted to operating (stabilized) properties held in fiduciary environments. It strips out the two things that contaminate most real estate return debates: leverage, which is a capital-structure choice rather than an asset property, and development risk, which is a different business. What remains is the clean question of what the asset class earns, and how violently.
Third, breadth. As of the first quarter of 2026, the index carries 12,996 properties and $926.5 billion in market value across more than 100 metropolitan markets. The five historic property types, apartment (now reported as "residential"), industrial, office, retail, and hotel, have been joined in the expanded NPI by self-storage, seniors housing, and an "other" bucket. This is not a survey. It is close to a census of institutionally held American real estate.
One structural caveat belongs up front, because we will return to it in Section 9: the NPI is appraisal-based, and appraisals smooth. Reported standard deviations understate true economic volatility for every sector. This matters for absolute risk statements. It matters much less for cross-sector ranking, which is what this paper is about, because the smoothing mechanism is common to all property types while the ordering of volatility across them is consistent with transaction-based indices like Green Street's CPPI. Multifamily's edge is a relative claim, and relative claims survive smoothing.
2. Inventory: Depth of Sample Is Itself a Finding
Start with the least glamorous statistic, how many properties are actually in the index, because inventory is not just a data-quality footnote. It is economic information. Here is the exact composition of the NPI as of the first quarter of 2026, per NCREIF's published Flash Report:
| Property Type | Properties (1Q2026) | Market Value ($M) | Share of NPI Market Value |
|---|---|---|---|
| Industrial | 5,685 | $309,335 | 33.4 percent |
| Residential (apartment) | 2,886 | $274,060 | 29.6 percent |
| Office | 1,708 | $160,967 | 17.4 percent |
| Self-Storage | 1,235 | $24,973 | 2.7 percent |
| Retail | 1,085 | $116,147 | 12.5 percent |
| Seniors Housing | 219 | $13,996 | 1.5 percent |
| Other | 156 | $24,659 | 2.7 percent |
| Hotel | 22 | $2,410 | 0.3 percent |
| Total Index | 12,996 | $926,547 | 100.0 percent |
Four observations.
The index votes with its feet. NPI composition is not fixed; it reflects what fiduciaries actually choose to hold on behalf of pensioners. In 1987, office was roughly half the index by value and apartments were a rounding error near 10 percent. Today residential is 29.6 percent of institutional value, $274 billion across 2,886 properties, while office has fallen to 17.4 percent and continues to shed both properties and share. And look at what institutions have added around the core: self-storage and seniors housing, the two newest subindexes, are both housing-adjacent bets on the household, not the corporation. Sum the residential complex (residential plus seniors housing plus a large fraction of self-storage demand, which is downstream of household moves and downsizing) and roughly a third of all institutional real estate value is now a wager on where Americans live. Over four decades, the most conservative capital in the country has executed a slow-motion rotation out of the sectors with concentrated tenant risk and into the sectors with granular, demographic demand. Revealed preference is a dataset too.
Hotels have nearly left the index entirely. Twenty-two properties. Three-tenths of a percent of value. The sector with the worst volatility and the deepest COVID drawdown in NPI history has been sold down to a rounding error, a quiet institutional verdict rendered one disposition at a time. Any hotel statistic from here forward describes a vestige, and NCREIF itself now cautions that the subindex "may not reflect the broader hotel sector."
Apartment inventory almost never contracts. Chart the count of NPI apartment properties by quarter and you get one of the most boring lines in institutional finance: a staircase that goes up. Quarters in which the apartment inventory meaningfully shrank are rare and idiosyncratic (fund redemptions, portfolio sales), not thesis-driven exits. Office and retail, by contrast, have spent the last decade in managed contraction: properties leave the index because managers are selling the sector down, writing assets off, or handing keys back to lenders. When a property type's index count falls for years at a time, that is disinvestment showing up in the sample itself. Multifamily has never had such an era. Not in the S&L bust, not in the GFC, not in 2020, not in the 2022 to 2023 rate shock.
Depth stabilizes the statistics. With 2,886 properties spread across more than 100 metropolitan markets, the residential series is diversified at the index level in a way the 22-property hotel series simply cannot be. The sector statistics that follow are estimated from a deep, geographically dispersed sample. They are not the echo of a few large assets in a few large markets.
3. The Anatomy of a Return: Income Is a Floor, Appreciation Is a Vote
Before the summary table, decompose the return, because the decomposition is where multifamily's risk profile is manufactured.
Every NPI total return is the sum of two components. The income return is net operating income over value, the cash the building throws off. The appreciation return is the change in appraised value, the market's opinion. The two components have completely different statistical personalities:
- The income return is almost pathologically stable. For apartments it has run in a band of roughly 1.0 to 1.5 percent per quarter (call it 4 to 6 percent annualized) for five decades, and it has essentially never printed negative at the index level. Buildings full of rent-paying households produce cash in expansions, recessions, pandemics, and hiking cycles.
- The appreciation return carries nearly all of the variance and all of the negativity. Every negative total-return quarter in the sector's history is a quarter in which appraised values fell by more than the income cushion.
This is why the income return functions as a floor. In the most recent print, the first quarter of 2026, the residential subindex earned an income return of 1.07 percent against an appreciation return of negative 0.17 percent, netting a positive 0.90 percent total return in a quarter when appraisers were still marking values down. That is the floor working in real time. For a diversified apartment portfolio to post a negative total return, values must decline by more than roughly 1.1 percent in a single quarter; appraisers must mark the sector down faster than the buildings pay you. In 47 years, that has happened in only three clusters: the early-1990s S&L workout, the GFC, and the 2022 to 2023 rate reset. Everything else, from 1981 and 1982 with the fed funds rate at 19 percent to the dot-com recession, the taper tantrum, and COVID itself, was absorbed by the income floor.
Hold that mechanism in mind. The table you are about to read is mostly a story about which sectors have the thickest floors and the least manic appraisal lines above them.
4. The Core Table: CAGR, Volatility, and the Ratio That Actually Matters
Here are the headline statistics for the five historic NPI property types over the full available history since 1978 (hotel enters the index slightly later, and its modern sample of 22 properties as of 1Q2026 is thin enough that NCREIF itself flags the subindex as unrepresentative). Returns are unlevered total returns; CAGR is the annualized compound growth rate of the quarterly series; volatility is the annualized standard deviation of quarterly returns; extremes are single worst and best quarters; the final column is the share of all quarters with a negative total return.
| Sector | CAGR (since inception) | Ann. Std. Dev. | CAGR ÷ Std. Dev. | Worst Quarter | Best Quarter | Negative Quarters |
|---|---|---|---|---|---|---|
| Apartment | ~8.6 percent | ~7.4 percent | ~1.16 | ~ −7.6 percent (2008Q4) | ~ +6.9 percent (2021Q4) | ~11 percent of quarters |
| Industrial | ~9.3 percent | ~8.1 percent | ~1.15 | ~ −7.9 percent (2009Q1) | ~ +13.3 percent (2021Q4) | ~12 percent |
| Retail | ~8.8 percent | ~7.1 percent | ~1.24 | ~ −7.3 percent (2009Q1) | ~ +7.5 percent (early 1980s) | ~13 percent |
| Office | ~7.3 percent | ~9.3 percent | ~0.78 | ~ −9.0 percent (2008Q4) | ~ +6.6 percent (early 1980s) | ~19 percent |
| Hotel | ~7.5 percent | ~11.5 percent | ~0.65 | ~ −19 percent (2020Q2) | ~ +9 percent | ~23 percent |
Read this table the way an economist reads it, not the way a leaderboard reads it.
The CAGR column understates the gap. Multifamily's ~8.6 percent sits in the top cluster, behind industrial's ~9.3 percent, and industrial's edge is real and recent, earned mostly in a single spectacular decade (2015 to 2022, the e-commerce repricing). But compound returns are the output of a distribution, and two sectors with similar CAGRs and different variances are not similar investments. Volatility drag is arithmetic, not opinion: a series that compounds at 8.6 percent with 7.4 percent volatility preserves far more of its arithmetic mean than one compounding through 9 or 11 percent swings. In a levered vehicle (and virtually all private real estate is held levered) variance at the asset level gets multiplied at the equity level, and the low-variance asset can support more leverage at the same equity risk. Unlevered NPI statistics therefore understate multifamily's advantage in the form investors actually hold it.
The ratio column is the verdict. CAGR divided by standard deviation, a Sharpe ratio without the risk-free subtraction, which is appropriate for comparing sectors within one asset class, separates the field into two camps. Apartment, industrial, and retail cluster between roughly 1.15 and 1.25: more than a unit of compound return per unit of reported risk. Office and hotel sit at 0.78 and 0.65: structurally inferior compensation for structurally higher variance. Office is the finding people underweight. Over 47 years, including office's golden decades, the sector delivered the lowest return of the big four and the second-highest volatility. That is not a post-COVID story. It was true before anyone had heard of remote work, because office economics (long leases that reprice rarely, catastrophic capital costs to re-tenant, binary single-tenant risk) were always fragile beneath the prestige.
Retail's ratio carries an asterisk that flatters multifamily. On the full window, retail's risk-adjusted ratio actually edges out apartments, a fact an honest paper must print. But decompose it by era and the ratio is a museum piece: it was earned overwhelmingly in 1978 to 2007, when the enclosed mall was a monopoly on the American consumer. Since 2016, retail's rolling returns have run below its long-run mean with rising dispersion, its NPI property count is shrinking, and its terminal-value uncertainty (what is a B-mall worth in 2040?) is exactly the kind of risk a standard deviation cannot see. Multifamily's ratio, by contrast, is earned in every era, which brings us to the consistency evidence.
The extremes tell you about the tails. Multifamily's worst quarter in 47 years is roughly negative 7.6 percent: painful, and shallower than office's worst, a fraction of hotel's negative 19 percent COVID quarter, and recovered within two years. Its best quarter, positive 6.9 percent in 2021Q4, is modest next to industrial's positive 13.3 percent. That pairing of a shallow left tail and an unspectacular right tail is the signature of the distribution. Multifamily does not pay you in lottery tickets. It pays you in the absence of catastrophes, which for a compounding vehicle is worth more.
The negative-quarter column is the sleeper statistic. Roughly one apartment quarter in nine has been negative, versus roughly one in five for office and nearly one in four for hotels. And the composition is even more lopsided than the frequency: essentially all of multifamily's negative quarters are packed into three episodes (1990 to 1992, 2008 to 2009, 2022 to 2023). Office scatters negative quarters across seven distinct episodes. If you charge a portfolio manager with delivering steady appraised value to a pension board every quarter for thirty years, the sector that is negative 11 percent of the time in three predictable clusters is a different fiduciary product than the sector that is negative 19 percent of the time in seven.
5. Consistency: The Decade Test
Full-period statistics can hide regime dependence, so run the harder test: compound each sector by decade and ask who shows up in the top half every time.
| Sector | 1980s | 1990s | 2000s | 2010s | 2020–2024 |
|---|---|---|---|---|---|
| Apartment | ~9.5 percent | ~9.6 percent | ~7.0 percent | ~9.4 percent | ~4.5 percent |
| Industrial | ~9.0 percent | ~8.6 percent | ~7.5 percent | ~12.5 percent | ~9.0 percent |
| Office | ~8.5 percent | ~5.3 percent | ~6.7 percent | ~8.5 percent | ~ −2 percent |
| Retail | ~9.0 percent | ~5.6 percent | ~9.5 percent | ~8.6 percent | ~2.5 percent |
| Hotel | n/a (thin) | ~7.5 percent | ~7.0 percent | ~8.0 percent | ~2 percent |
Every other sector has a lost decade. Office lost the 1990s to the S&L hangover and is losing the 2020s to the work-from-home repricing. Retail lost the 1990s to overbuilding and the late 2010s onward to e-commerce. Industrial spent the 1990s and 2000s as a respectable but unremarkable 7 to 8 percent asset before its 2010s apotheosis. Hotels lose whichever decade contains the recession.
Apartments have never had one. The worst apartment decade on record, the current half-finished 2020s carrying the full weight of the 2022 to 2023 rate shock and the largest supply wave since 1973, still compounds positively. In every completed decade since the index began, multifamily lands between roughly 7 and 10 percent. That 300-basis-point band across five utterly different macro regimes is the narrowest of any property type, and it is the statistical fingerprint of a demand base that does not depend on the business model of the American corporation (office), the format of the American store (retail), the geometry of the American supply chain (industrial), or the discretionary travel budget (hotel). It depends on household formation. Households keep forming.
6. Drawdown Forensics: Five Episodes, One Pattern
Averages are for brochures. Allocators live in the drawdowns, and in the recoveries that follow them, so walk through all five.
The S&L era, 1990 to 1992. The one crisis where multifamily was near the center of the blast, because the 1981 to 1986 tax code had turned apartment construction into a tax shelter and the 1986 reform detonated it. Even so, NPI apartment peak-to-trough value declines ran shallower than office, which suffered its worst stretch of the century outside the GFC as see-through towers sat empty across Texas and New England. Both sectors taught the same lesson from opposite sides: supply kills, but supply absorbed by demographic demand (apartments, filled within three years) is a different disease than supply facing no marginal tenant (office, a decade to digest).
The Global Financial Crisis, 2008 to 2009. The control experiment. A housing-triggered crisis should have been maximally hostile to residential assets, and yet apartment fundamentals were the most resilient in the index. National apartment occupancy bottomed around 92 percent; office availability blew out for years. Apartment NOI at the index level fell mid-single digits peak-to-trough and recovered by 2011; office NOI declined longer and deeper on a lagged fuse as leases rolled into a dead market. Apartment values were first into the drawdown, first out: the sector posted the NPI's fastest return to prior-peak value, aided by the fact that the collapse of the for-sale housing market manufactured renters out of foreclosed homeowners in real time. And critically, the debt never left. Which is Section 7's subject.
COVID, 2020. Hotels: roughly negative 19 percent in a single quarter, the worst print in NPI sector history. Retail: rent collection crises and co-tenancy dominoes. Office: the beginning of a structural, still-unresolved repricing. Apartments: collections dipped to the mid-90s percent range, occupancy barely moved, and by late 2021 the sector was printing the best quarters in its recorded history (positive 6.9 percent in 2021Q4 alone). Total pandemic drawdown at the index level: barely two soft quarters. People can stop traveling, stop commuting, and stop shopping in person. They cannot stop living somewhere.
The rate shock, 2022 to 2023. The honest entry, because multifamily took real damage: peak-to-trough appraised value declines around 20 percent (transaction-based indices printed negative 25 to negative 30 percent), driven by cap-rate expansion from the low 4s to the mid 5s and a 500-basis-point hiking cycle. Two features distinguish it. First, it was a discount-rate event, not a cash-flow event. Apartment NOI kept growing through the entire drawdown, which is why the value decline stopped the moment the rate path stabilized. Compare office, which faced the same discount-rate shock plus a collapsing numerator. Second, the market cleared. Multifamily transaction volume recovered faster than any other sector because the buyer pool (GSE-financed, 1031-driven, institution-deep) never disbanded. Office bid-ask spreads stayed unbridgeable for years. The ability to find the bottom quickly is itself a risk statistic, one that never shows up in standard deviation.
The healing tape, 2024 to 2026. The recovery is now printing, and the most recent NCREIF release (1Q2026) is worth reading closely because it shows the cycle mid-turn. The total index returned 1.23 percent in the quarter, its fourth consecutive positive quarter, for a trailing-year total of 4.94 percent, composed of a 1.15 percent income return and a capital return that has just crossed back over zero (positive 0.08 percent, after 0.00 percent the prior quarter). Appraisal cap rates for unsold properties stand at 4.57 percent; properties that actually traded cleared at 5.27 percent, a 70-basis-point spread that tells you appraisals are still catching down to the transaction market even as index NOI growth turned positive (0.56 percent). By sector, the trailing year reads: seniors housing 12.8 percent, self-storage 7.8 percent, retail 6.9 percent, residential 4.9 percent, industrial 4.4 percent, office 3.9 percent, hotel 2.0 percent.
Two things in that tape matter for this paper's argument. First, the two best-performing sectors in the entire index, seniors housing and self-storage, are both housing-adjacent, both demographically driven, both downstream of the household rather than the corporation. The index's own leaderboard is currently a ranking of proximity to residential demand. Second, an honest reading: core residential itself printed the softest quarter of the major food groups in 1Q2026 (0.90 percent total, with appreciation at negative 0.17 percent as the 2021 to 2023 supply wave finishes digesting), even as its income return held at 1.07 percent and its trailing year matched the index at 4.9 percent. That is exactly the pattern Section 9 describes. Multifamily's downturns are supply-timing events with intact cash flows, not demand-destruction events. The last two times the sector printed this profile (1993, 2010) were, in retrospect, the best entry points of their respective decades.
Five episodes, one pattern: multifamily is never the epicenter twice, never the deepest drawdown, always among the first to reprice and re-trade. In the language of the table: the thin left tail is not luck. It is repeated behavior under maximally varied stress.
7. The Machinery: Why the Distribution Looks Like This
A statistical edge without a mechanism is a backtest. Here is the mechanism: five interlocking structural features, each one visible in the data.
One-year leases: the repricing advantage. An apartment building reprices its entire revenue stack every twelve months. An office building reprices perhaps a tenth of its rent roll in a good year, on leases written up to fifteen years ago. This single parameter, lease duration, explains an astonishing share of the sector return patterns. In inflations, multifamily passes through costs in real time (see 2021 and 2022, when apartment rent growth ran double digits while office rents sat contractually frozen beneath CPI); the NPI's inflation-hedging literature consistently ranks apartments at the top for exactly this reason. In recessions, the same short duration lets the sector find the market-clearing rent within quarters and refill the building. Occupancy self-heals because the price is allowed to move. Office's long leases smooth reported income on the way in and then deliver the entire accumulated repricing as a cliff at expiry, plus a seven-figure invoice for tenant improvements. Multifamily's revenue line is a thermostat; office's is a time bomb with a long, quiet fuse.
Tenant granularity: credit risk as a law of large numbers. A 300-unit apartment property is a portfolio of 300 independent household credits, each one a low-single-digit percentage of revenue. A single-tenant office building is one corporate credit at 100 percent. The apartment's vacancy process is binomial and mean-reverting; the office's is binary and catastrophic. This is why apartment income returns look like a bond coupon in the data: idiosyncratic tenant risk has been diversified away inside each asset, before the index diversifies across assets. No other major property type gets its diversification wholesale like this. It must be assembled tenant by tenant, lease by lease, and it still concentrates.
GSE debt: the countercyclical liquidity spine. Multifamily is the only commercial property type with a permanent, policy-mandated lender of first resort. Fannie Mae and Freddie Mac fund roughly 40 to 50 percent of all multifamily lending, and their share expands in crises. 2009 and 2020 both saw the agencies underwriting loans in quarters when the CMBS market and the balance-sheet banks had simply closed. The consequences compound: refinancing risk, the classic mechanism by which real estate downturns metastasize into forced-sale spirals, is structurally muted; cap rates carry a durable liquidity premium (the asset is always financeable, so it is always biddable); and drawdowns terminate early because a functioning debt market lets buyers underwrite the bottom. When people ask why apartment values found a floor in 2009 and 2024 while office kept falling, a large part of the answer is that one sector's mortgage market never stopped answering the phone. This subsidy is not going away; housing finance is one of the few genuinely bipartisan commitments in American policy.
The occupancy floor and the substitution ladder. National apartment occupancy has spent five decades oscillating in a band roughly between 92 and 97 percent. It does not visit the 70s. Office does. The reason is that housing demand does not disappear in recessions. It trades down the quality ladder within the same asset class. The Class A tenant moves to B; the B tenant moves to C; the household that delays buying a home stays put and renews. Every rung of the ladder remains an apartment, so the sector's aggregate occupancy is nearly recession-proof even as individual assets compete harder. Office demand, when it contracts, exits the asset class entirely: to the spare bedroom, to the hybrid schedule, to no lease at all. One sector's demand shocks are redistributive; the other's are absolute.
Capital expenditure economics. The dirty secret of office NOI is how little of it survives contact with the capital account: tenant improvements, leasing commissions, and lobby arms races routinely consume 20 to 40 percent of NOI across a cycle. Apartment turns cost paint, carpet, and a leasing agent's afternoon. NCREIF's own cash-flow research shows apartments converting more of their income return into distributable cash than office by a wide margin. Since the NPI income return is computed before most of this capital drag, the reported statistics, remarkably, understate multifamily's advantage in the currency that ultimately matters, which is cash in the partnership's account.
8. The Demand Engine: The Single-Family Shortfall Is Multifamily's Structural Edge
Everything to this point explains multifamily's past distribution. This section is about why the forward distribution may be better still, and it rests on the most consequential imbalance in the American economy: the country stopped building enough homes a generation ago and has never caught up.
The deficit is cumulative and large. From 1968 through 2008, the United States completed roughly 1.4 to 1.6 million housing units per year. From 2009 through 2021, it averaged barely 900,000 to 1.1 million against household formation that never stopped. Every credible estimate of the resulting cumulative shortfall lands in the millions. Freddie Mac put it at 3.8 million units as of 2021; Realtor.com and NAR-affiliated work has ranged toward 5 to 7 million; even conservative methodologies concede several million. The subsequent 2021 to 2024 construction surge slowed the bleeding without closing the gap, because it coincided with the largest household-formation wave since the baby boom hit the housing market. Whatever the exact figure, the direction is undisputed: America is structurally short of places to live, and shortages set the price of the marginal unit.
The shortfall lands hardest on the entry-level buyer, which converts owners-to-be into renters-who-stay. The missing units are disproportionately starter homes: the sub-1,800-square-foot single-family house fell from roughly 40 percent of new construction in the early 1980s to under 10 percent by the 2010s, killed by land costs, impact fees, minimum-lot zoning, and builder consolidation into higher-margin product. The buyer who cannot find an entry-level home does not vanish. She renews her apartment lease. The median age of the first-time homebuyer, 29 in 1981, reached 38 by 2024, an all-time record. That is nearly a full decade of additional rentership per household, multiplied across the largest generational cohorts in American history. In present-value terms, the multifamily industry's average customer lifetime has been extended by policy failure in an adjacent market.
The rate lock-in deepened the moat. After 2022, the math turned from difficult to prohibitive. With the vast majority of outstanding mortgages fixed below 5 percent and prevailing rates well above 6, existing owners cannot afford to sell (the lock-in effect starves resale inventory), while the monthly cost of owning the median home rose to a premium of $1,000 or more per month over renting the comparable unit, the widest own-versus-rent spread in the modern record. The homeownership margin, where apartment demand historically leaked away, is frozen at both ends: no starter-home supply, no affordable exit. Multifamily's biggest competitor has, in effect, withdrawn from the field for an extended period.
And the demand is demographic, not cyclical. Millennials and Gen Z together represent roughly 140 million Americans moving through peak household-formation age over the next fifteen years, layered over immigration flows that skew heavily toward rental housing in their first decade. Demand of this kind does not require GDP growth, corporate profits, consumer confidence, or trade volumes, the respective demand engines of office, retail, and industrial. It requires birthdays. Birthdays are the most reliable time series in economics, and the current NCREIF tape corroborates it from the other end of the age distribution, where seniors housing, the purest demographic bet in the index, is the best-performing sector in institutional real estate at 12.8 percent over the trailing year. The same actuarial table is driving both subindexes. Demography is doing the underwriting.
Connect this back to the NCREIF statistics and the logic closes. The shortfall raises the sector's mean (structural excess demand supports real rent growth above inflation over full cycles). The lock-in and the entry-level drought thin the left tail (the marginal renter has fewer exit options precisely when the economy weakens, making downturn occupancy stickier than in any prior regime). And the demographic engine stabilizes the whole distribution (the demand base regenerates independently of the macro cycle). The single-family shortfall is not a talking point appended to the data. It is a mechanism that acts on every column of the table in Section 4, in multifamily's favor, simultaneously.
9. The Honest Ledger: What Can Hurt You
A white paper that finds no risks has found no truth. Four entries on the other side of the ledger.
Appraisal smoothing understates absolute risk. Unsmoothed or transaction-based apartment volatility likely runs 1.5 to 2 times the reported NPI figure, and 2022 to 2023 demonstrated that private valuations can fall 20-plus percent when the discount rate moves 200 basis points. The relative ranking against other sectors survives (they smooth too, and their transaction-based indices fell further), but no one should read 7.4 percent volatility as literal. Lever the asset 60 percent and the equity experience in a rate shock is genuinely violent.
Supply is the sector's recurring self-inflicted wound. Multifamily's demand story is so legible that capital periodically overruns it: 1972 to 1974, 1984 to 1986, and 2021 to 2023, when units under construction peaked near 1.1 million, the most since 1973. The 2024 to 2026 delivery wave produced flat-to-negative rent growth across the Sun Belt's highest-supply metros, exactly as the history predicts, and it is visible in the current index print, where residential appreciation was still marginally negative (0.17 percent below zero in 1Q2026) while every other statistic in the sector's profile held. The regional subindexes tell the same story: the high-supply South still compounded at 6.0 percent over the trailing year on the strength of demand, while the West, where the rate shock hit richly priced coastal assets hardest, trailed at 3.7 percent. The saving grace, visible in every prior episode, is that demographic demand absorbs apartment oversupply in two to four years, whereas office oversupply can take a decade or more. Supply risk in multifamily is a timing risk, not a terminal-value risk. But levered vehicles die of timing.
The expense line has become the new battleground. Insurance premiums doubling in coastal markets, property-tax reassessment, payroll, and the general repricing of operational risk compressed apartment NOI margins meaningfully after 2022. The revenue line's inflation pass-through remains the best in real estate, but the expense line increasingly passes inflation through in the other direction. Underwriting that still assumes 2015-era expense ratios is fiction.
Regulatory risk scales with the sector's success. Rent control and rent stabilization regimes expanded in the 2019 to 2025 window, and the same housing shortage that powers multifamily's economics generates the political pressure to cap them. The risk is jurisdiction-specific and, at the national-index level, historically modest, but it is correlated with exactly the markets where scarcity rents are largest. The sector's tailwind and its political risk are the same phenomenon viewed from opposite sides of the ballot.
None of these reverse the verdict. All of them price it. The correct conclusion is not that multifamily is riskless. It is that multifamily is the sector whose risks are cyclical and insurable (supply timing, expense inflation, rate duration) rather than structural and terminal (obsolescence of the office contract, of the retail format, of the business-travel budget).
10. Implications for Portfolio Construction and Underwriting
For the allocator, the NCREIF record supports treating core multifamily not as one of four interchangeable property bets but as the anchor allocation: the sector whose return-to-risk ratio, negative-quarter frequency, and drawdown recovery speed most resemble what real estate is hired to do in an institutional portfolio, which is to deliver bond-plus returns with equity-plus tax efficiency and genuine inflation linkage. Sector deviations from a multifamily-heavy baseline should be justified as tactical (industrial's secular demand, selective retail basis plays), not treated as neutral diversification. Diversifying from apartments into office over the last half-century meant accepting more volatility for less return, the textbook definition of an inefficient trade, and the market-value shares in Section 2 show institutional capital has spent forty years arriving at the same conclusion.
For the operator and developer, the data disciplines the pro forma. The sector's long-run unlevered 8.6 percent decomposes into roughly 4.5 to 5 percent income and 3.5 to 4 percent appreciation, so any underwriting that requires double-digit unlevered returns from a stabilized asset is assuming either uncompensated leverage, un-repeatable cap-rate compression, or a rent-growth regime the 47-year record has never sustained. The durable excess returns in multifamily have come from the edges the data identifies: buying into the supply-wave troughs the sector reliably manufactures (1993, 2010, arguably 2024 to 2025), operating the expense line as aggressively as the revenue line, and positioning product one rung below the frozen entry-level ownership market, where the substitution ladder deposits its tenants.
For the economist, the conclusion generalizes: multifamily wins because it is the property sector most directly plugged into the household, the one economic unit that reprices annually, diversifies naturally, borrows with a federal guarantee, and multiplies demographically. Every other sector is a derivative of a business model that can be disrupted. Multifamily is a derivative of the population.
11. Methodology and Sources
Current-quarter statistics, meaning index composition (12,996 properties, $926.5 billion), property counts and market values by subindex, 1Q2026 and trailing-year returns by sector and region, income and appreciation splits, cap rates (4.57 percent appraisal, 5.27 percent transaction), and NOI growth (0.56 percent), are taken directly from NCREIF's NPI Flash Report, 1st Quarter 2026 and the accompanying NPI press release dated April 25, 2026, both published at ncreif.org. Long-run return statistics reference the NCREIF Property Index (NPI), quarterly, unlevered, appraisal-based, 1978Q1 forward, property-type subindices as published by NCREIF; hotel statistics reflect the subindex's later inception and now-vestigial sample. CAGR is the geometric annualization of linked quarterly total returns; standard deviation is the annualized standard deviation of quarterly total returns; negative-quarter frequency is the count of quarters with total return below zero over quarters available. Since-inception figures are rounded, deliberately, to the precision appraisal-based data supports, and holders of NCREIF query access should reproduce exact values for any decision-grade use; sub-period boundaries shift point estimates without altering rankings. Housing-shortfall estimates reference Freddie Mac's 2021 analysis (3.8 million units) and the broader published range; construction, household-formation, and first-time-buyer statistics reference U.S. Census Bureau completions data and NAR buyer-profile surveys. Transaction-based comparisons reference Green Street's CPPI. Where this paper says "approximately," it means it.
12. Conclusion
Forty-seven years of the most carefully assembled dataset in institutional real estate reduce to a short verdict. Multifamily compounds near the top of the field, with the narrowest band of outcomes across decades, the fewest negative quarters, the shallowest tails, and the fastest recoveries. Every one of those statistical properties traces to a durable mechanism: annual repricing, granular credit, guaranteed debt, an occupancy floor built from the substitution ladder, and a capital-expenditure profile that lets the income actually reach the investor. Layered on top is a demand engine no other sector possesses: a multimillion-unit national housing shortfall, an entry-level ownership market frozen by rate lock-in and land-use policy, and the largest renter cohorts in American history forming households on a schedule set by demography rather than by the business cycle.
Office needed the corporation. Retail needed the format. Industrial needed the supply chain. Hotels needed the recovery. Multifamily needs only what has never once failed to appear in the entire history of the index: next year's households, looking for somewhere to live.
That is why multifamily wins. Not as a slogan. As a distribution.