Do Brazilian REITs work as a retirement asset?
Over the last five years, IFIX — the B3 exchange index that tracks Brazil's listed real estate funds, known locally as FIIs — returned 0.6% a year above inflation. Over seven years it returned 0.3% a year below it. Cash, over those same windows, delivered 6.4% and 3.9% in real terms. If you hold FIIs for the monthly income and feel like you have gone nowhere, you are reading the data correctly.
What is being measured here. Total return — price plus reinvested distributions — for IFIX, the Ibovespa (Brazil's main equity benchmark) and IDIV (B3's dividend index). All three are total-return indices. They are compared against the CDI, the interbank rate that serves as Brazil's cash benchmark, and against IPCA, the country's official inflation index. Every window ends in July 2026, and "real" always means net of the inflation of that same period.
Five windows, one ending month
A single number proves nothing. The same asset class looks brilliant or disastrous depending on where you place the ruler — which is why the table below cuts it five different ways, all ending in the same month.
| Window | IFIX | CDI (cash) | IDIV | Ibovespa | Inflation |
|---|---|---|---|---|---|
| 3 years | +1.3% | +7.8% | +11.1% | +8.3% | 4.7% |
| 5 years | +0.6% | +6.4% | +7.2% | +2.1% | 5.6% |
| 7 years | −0.3% | +3.9% | +6.0% | +2.5% | 5.6% |
| 10 years | +3.1% | +4.2% | +10.5% | +6.7% | 5.0% |
| Since 2010 | +3.0% | +3.9% | +4.7% | +0.5% | 5.8% |
Annualised real return, net of inflation. Last column: the inflation of that same period, in % per year.
The recent frustration is real, then. But the uncomfortable part is the bottom row: this did not start in 2021. Measured from the index's inception in December 2010 — fifteen years and seven months of complete history, every rate cycle included — IFIX delivered 3.0% a year above inflation while cash delivered 3.9%. An investor who bought the whole real estate index and reinvested every distribution ended up almost a full percentage point a year behind someone who simply held cash, and took considerably more risk to get there.
"You picked a bad starting point"
That is the obvious objection, and it deserves a real test rather than a rebuttal built on another convenient window. So we measured all of them: each of the 128 rolling five-year windows that fit between December 2010 and July 2026, starting in every calendar month.
Annualised real return across 128 rolling five-year windows, Dec 2010 to Jul 2026.
The conclusion survives the change of window. Cash beat IFIX in 77 of the 128 five-year windows (60%). IDIV won 78 of them (61%) and so did the Ibovespa (61%). And IFIX finished below inflation in 29 of those windows (23%) — roughly one in four consecutive five-year stretches destroyed purchasing power.
For anyone planning retirement, though, one number matters more than the median. Cash's worst five-year stretch was −0.13% a year in real terms, essentially a tie with inflation. IFIX's worst was −6.09% a year, which compounds into roughly 27% of purchasing power gone over five years. That is the asymmetry in one line: equity-like downside paired with an upside ceiling (best window: +12.79%) that fixed income does not have, but also does not need.
Stretch the horizon and the verdict flips
Stopping here would produce a clean conclusion and a wrong one. We ran the same exercise over ten-year windows, and the picture changes character.
| Ten-year windows (68 in total) | Median real | Worst window | Below inflation |
|---|---|---|---|
| IFIX | +3.10% | −0.29% | 3 of 68 (4%) |
| Cash (CDI) | +3.14% | +1.94% | 0 of 68 (0%) |
| IDIV | +4.02% | +0.79% | 0 of 68 (0%) |
| Ibovespa | +2.65% | −0.58% | 10 of 68 (15%) |
Over ten years IFIX ties with cash — 3.10% against 3.14% median real — and the ugly tail almost disappears: the worst decade lost 0.29% a year to inflation, against 6.09% in the five-year cut. More striking still, over that horizon IFIX failed to beat inflation in only 4% of windows while the Ibovespa failed in 15%. For a long-horizon investor, Brazilian real estate funds were historically more dependable than Brazilian equities at the single job of not losing purchasing power.
Reading both tables together: in Brazil, listed real estate has not been an asset that pays a high premium. It has been an asset that behaves roughly like fixed income over long horizons while swinging like equity along the way. A five-year investor carried risk without the premium. A fifteen-year investor matched cash while collecting tax-exempt monthly income — a different proposition, not necessarily a worse one.
So are equities the better answer?
In the data, IDIV — the B3 index of Brazil's highest-paying dividend stocks — wins almost every cut: 5.08% real a year at the five-year median, 4.02% at ten years, and it beat IFIX in 81% of ten-year windows. The Ibovespa does not: since 2010 it delivered 0.5% real a year, less than IFIX and far less than cash.
The price of IDIV's return shows up in the column nobody quotes. Its worst five-year window was −12.31% a year, twice as deep as IFIX's worst, and it ended below inflation in 28% of five-year windows against IFIX's 23%. IDIV pays more and hurts more. These are different shapes of pain, not a straightforward ranking.
There is also a structural difference the table cannot show. Distributions from Brazilian real estate funds are exempt from income tax for individual investors under current rules, while stock dividends are not. In a portfolio built to live off income, that gap lands in the bank account every month.
What if rates fall? The question every holder asks
The reasoning is familiar: with rates high, investors abandon real estate funds for government bonds; when rates fall, the flow returns and prices rise. It has logic — and a problem when it is used as a long-term thesis.
First, because the same move lifts the alternatives. A ladder of inflation-linked government bonds (Tesouro IPCA+, Brazil's retail inflation-linked treasury) benefits from exactly the same fall in real rates, and it does so under contract: the real coupon and the principal are fixed at purchase. If long real rates fall, the bond you already own marks up just like the fund's price does, with the difference that holding to maturity delivers the contracted real return regardless of whether a manager allocates well.
Second, because the historical series does not support an automatic link between the policy rate and the rate that actually prices property. Across 196 months of history, the pass-through from Selic — Brazil's overnight policy rate — to the long inflation-linked yield that discounts a real estate fund's cash flow shows a beta of 0.12 and a correlation of 0.17. The long end responds to fiscal risk and term premium far more than to the overnight rate. "Rates will fall, so FIIs will rise" describes something that sometimes happens; it is not a mechanism you can contract.
Three funds that worked — and what they share
An index is an average, and averages hide everything. We measured total return fund by fund, reinvesting each distribution at that month's price and correcting the history for share splits, always against the IFIX and the cash rate of that fund's own window. These are cases with nearly a decade of clean history:
| Fund | Window | Price | Total p.a. | Real p.a. | IFIX same period |
|---|---|---|---|---|---|
| BTLG11 · logistics | Jan 2017 – Jun 2026 | +50% | +12.2% | +6.7% | +7.5% |
| MXRF11 · credit and hybrid | Apr 2017 – Jul 2026 | −0.3% | +11.3% | +5.9% | +7.0% |
| HGLG11 · logistics | Mar 2017 – Jul 2026 | +26% | +11.0% | +5.7% | +7.0% |
| HGRU11 · urban income | Jul 2018 – Jul 2026 | +17% | +10.8% | +5.2% | +7.4% |
| KNRI11 · offices and logistics | Dec 2016 – Jul 2026 | +10% | +8.2% | +3.0% | +7.8% |
The top three delivered between 5.7% and 6.7% real a year for close to a decade — above cash for the same period and well above the index. MXRF11 is the most instructive entry on the list: its price is essentially where it was in 2017 (−0.3%), and it still compounded at 11.3% a year. The entire return came from reinvested distributions. Anyone reading only the price chart concluded the fund went nowhere; anyone adding the income collected one of the best results in the sample.
What these cases share is not a segment — logistics, credit and urban income all appear. It is a large manager, a diversified portfolio, and follow-on offerings the market absorbed without punishing the price. And the counterexample sits inside the same list: KNRI11, from an equally large manager with an equally long record, returned 3.0% real a year and finished below the IFIX of its own window. A strong name guarantees nothing.
And the ones that collapsed
At the other end, the destruction is on a different scale. Every figure below is total return, with all distributions already added back:
| Fund | Window | Price | Total p.a. | Real p.a. |
|---|---|---|---|---|
| XPCM11 · single-tenant office | Dec 2016 – Apr 2026 | −91% | −15.3% | −19.4% |
| EDGA11 · corporate office | Nov 2016 – Jun 2026 | −77% | −9.6% | −13.9% |
| VVCR11 · active credit | Jun 2018 – Jun 2026 | −82% | −9.2% | −13.8% |
| HCTR11 · high yield credit | Jul 2019 – Jul 2026 | −85% | −6.5% | −11.5% |
| BRCR11 · corporate offices | Nov 2016 – Jun 2026 | −55% | −0.3% | −5.1% |
| URPR11 · high yield credit | Jan 2024 – Jul 2026 | −78% | −37.4% | −40.3% |
Note that the generous distributions rescued none of them. HCTR11 paid a high yield for much of the period and still returned −11.5% real a year: an investor collecting 1.5% a month was collecting his own capital back while the price melted. That is the difference between income and return of principal, and it only becomes visible when you measure total return instead of dividend yield.
Of the 127 funds in our database with a clean series and at least three years of history, 33 (26%) saw the price drop by 30% or more, 13 fell by half or more and 4 lost over 70%. On a total-return basis, 14 of them (11%) are nominally negative — meaning that even after adding every distribution received, the investor holds less money than he put in.
The number this study cannot show you
One limitation has to be stated plainly, because it pushes every result above in the same direction.
Of 166 funds with enough history, 39 were discarded because the price series contains a step that the declared split does not explain, or because the implied distribution in some month is too large to be income — the signature of principal amortisation booked as a dividend. Measuring those without handling each case individually would produce wrong numbers, and a wrong number is worse than a missing one.
The trouble is that those 39 are not a random draw. Reverse splits and mass amortisations are precisely what happens to a fund that is shrinking or winding down. The filter therefore removes the worst cases preferentially. The 2.0% real annual median we measured among seven-year survivors overstates what the average holder actually received — and it does not even count the funds that stopped existing and so never entered the database at all.
IFIX itself carries no such bias: B3 calculates it from the theoretical portfolio of each moment. That is why the index figures in this article are the solid part of the exercise, while the individual-fund figures should be read as a portrait of those left standing.
What the whole thing says about retirement
The data supports three claims and undermines two.
Supported: (1) over a five-year horizon, Brazilian real estate funds historically delivered equity-like risk with below-cash returns, and that holds across 60% of measured windows rather than one unlucky slice; (2) over ten years or more, they match cash and are more dependable than the equity index at preserving purchasing power; (3) the gap between a good and a bad fund is wider than the gap between asset classes — 6.7% real a year in BTLG11 against −19.4% in XPCM11 is a chasm no allocation decision between the index and cash could ever produce.
Undermined: (1) the idea that a high dividend yield signals a high return — the biggest wipeouts were paying the biggest yields; (2) the idea that selection does not matter and the index is enough — index buyers trailed cash over fifteen years while several individual funds beat it comfortably for a decade.
On the suspicion that high yield credit funds and smaller managers are structurally doomed: among the funds that survived with a clean series, the credit category's median was 3.4% real a year, above the property category's. But that sample excludes precisely the funds that blew up — HCTR11, VVCR11, URPR11 and CACR11 all fall outside it, either for dirty data or short history. With what we have today, the claim can be neither confirmed nor dismissed by the numbers: the individual failures do cluster in high yield credit and in single-tenant offices, but the surviving sample does not show the whole category losing.
What to watch from here
- The long real yield, not the policy rate. It is what discounts these funds' cash flows, and it answers to fiscal risk rather than the overnight rate.
- Payout ratios persistently above 100%. Distributing more than the fund generates is returning principal; the effect shows up in the price before it shows up in the distribution.
- Tenant and borrower concentration. Every fund on the list above that lost more than 70% was concentrated — in one tenant, one borrower group or one segment.
- Amortisation disguised as income. Regulatory filings separate the two; brokerage apps frequently show them added together.
- Your actual horizon. The five-year and ten-year tables tell different stories about the same asset, and the one that applies is the one matching how long the money will stay invested.
