How Money Is Actually Made in Real Estate: Cap Rates, Depreciation, and the LP/GP Bargain
Real estate is the only major asset class where the asset depreciates on the tax return while it appreciates on the balance sheet. That single contradiction is the engine that has compounded private wealth in this industry for a hundred years, and it is the reason real estate's professional class spends more time on capital structure and tax engineering than on bricks and mortar.
The professional question to ask is not whether real estate makes money. It is which engine made it. There are three engines, and they operate on different timelines, with different counterparties, and through different sections of the tax code. The investor who understands one engine in isolation underwrites a small part of the trade. The investor who runs all three together is operating on a different math entirely.
This article is the operating model. Cap rates as the pricing engine. Depreciation as the tax engine. The LP/GP waterfall as the allocation engine. And the moments where the three are pointed at each other rather than in the same direction.
1. The Cap Rate Is the Bond Equivalent of a Building
A cap rate is not a return. It is a price expressed as a yield.
The math is trivial. Net Operating Income divided by Value equals the cap rate. A property generating $1 million of NOI trading at a 5 percent cap rate is worth $20 million. The same property at a 6 percent cap rate is worth $16.7 million. The cap rate moved 100 basis points and the value dropped 17 percent. NOI did not change. The asset did not change. The pricing changed.
That sensitivity is the first thing every real estate professional internalizes, and it is the last thing most outside investors understand. A 5 cap to 5.5 cap move on a stabilized asset is a 9 percent loss of equity value. A 5 cap to 6 cap move is a 17 percent loss. A 5 cap to 7 cap move is a 29 percent loss. None of these moves require a single tenant to vacate, a single lease to roll, or a single dollar of NOI to disappear. They are pure pricing events driven by the cost of capital across the rest of the economy.
The reason cap rates move is that they are not set in real estate. They are set against the risk-free rate plus a real estate risk premium, and the risk-free rate is set by the Treasury curve. When the 10-year Treasury moves 200 basis points, cap rates eventually move with it, with a lag determined by the depth of the bid-ask spread between sellers anchored to old marks and buyers underwriting new debt.
A cap rate decomposes into a yield stack of four components: the real risk-free rate (the TIPS yield equivalent), inflation expectations (embedded in nominal Treasury yields), an illiquidity premium for private market exposure, and a real estate risk premium that varies by asset, market, and sponsor. The sum of those four layers is the cap rate the asset trades at.
The risk premium component is what allows real estate to outperform fixed income on a long horizon. It is also what gets re-rated when capital flows shift. When credit spreads widen, the real estate risk premium widens with them. When pension allocations to private real estate increase, the risk premium compresses. The cap rate is the visible artifact of those flows, and the asset value moves with the artifact whether the building has changed or not.
2. Cap Rate Compression Is the Largest P&L Line in the Industry
The single largest source of returns in private real estate over the past forty years has not been NOI growth. It has been cap rate compression. From the early 1980s, when cap rates on stabilized commercial assets routinely traded at 9 to 11 percent, through the 2021 cycle low when prime industrial traded inside 4 percent, the entire industry rode a 500 to 700 basis point compression that mathematically multiplied asset values by 2x to 3x before any operational improvement.
A simple decomposition. A property bought in 1990 at a 9 cap with $1 million of NOI was worth $11.1 million. The same property in 2021 with $2 million of NOI at a 5 cap was worth $40 million. NOI doubled, contributing roughly half the value gain. Cap rate compression contributed the other half. The operator who underwrote 2021 cap rates as the new normal was implicitly underwriting that the 40-year tailwind would continue. It did not.
The 2022 to 2024 cap rate expansion was the reverse of that trade. Properties that traded at 4 cap industrial moved to 5.5. Multifamily that traded at 4.25 moved to 5.5. Office that traded at 6 moved to 9 or did not trade at all. Sponsors who refinanced into the 2021 marks watched LTVs rise above 90 percent as values reset. The industry's largest open question in mid-2026 is how much further the expansion has to run before cap rates stabilize against the new rate regime.
The honest read is that NOI growth has been doing the work that cap rate compression used to do, and NOI growth is harder, slower, and more operationally taxing than waiting for the next 100 bps of compression. The operators outperforming in this cycle are the ones who built operating capability rather than financial engineering capability, because the financial engineering tailwind has reversed.
3. Why Cap Rates Matter for Development
Development underwriting hinges on a single number: the spread between yield-on-cost and the projected exit cap rate.
Yield-on-cost is stabilized NOI divided by total project cost. Exit cap is the cap rate at which the stabilized asset is sold or refinanced. The spread between the two is the development premium, and it is the entire reason developers exist as a category.
A multifamily project that costs $100 million all-in to build and stabilizes at $6 million of NOI has a yield-on-cost of 6 percent. If the exit cap rate is 5 percent, the asset is worth $120 million on stabilization. The $20 million spread is the development profit. Subtract sponsor promote, GP fees, and LP returns, and the residual is what made the project worth doing.
If the exit cap rate moves from 5 to 6 percent before stabilization, the asset is worth $100 million. The development profit goes to zero. The sponsor delivered the building, hit the underwriting NOI, and made nothing. If the exit cap rate moves to 6.5 percent, the asset is worth $92 million, the project has an $8 million loss, and the equity is impaired before the lease-up curve has even completed.
This is the structural risk of ground-up development that most LPs underestimate. The development premium is a function of cap rate spread, and cap rate spread is set in the capital markets, not in the construction trailer. A perfectly executed development project in a wrong-direction cap rate move loses money. An imperfectly executed project in a favorable cap rate move makes money. Across the cycle, sponsors who treat cap rate forecasting as a residual exercise discover that it is the entire exercise.
The disciplined development underwriting move is to demand a minimum yield-on-cost spread over current cap rates, typically 100 to 200 basis points depending on product type and market. A development that pencils only if exit caps compress further than entry caps is not a development. It is a leveraged bet on the capital markets dressed up as a construction project. Most institutional capital learned this distinction the hard way between 2022 and 2025.
4. Why Cap Rates Matter for Exits
The exit cap rate is the single most consequential assumption in any real estate pro forma. A 25 basis point error in the exit cap on a 10-year hold produces roughly a 4 to 5 percent error in projected IRR. A 100 basis point error produces 15 to 20 percent. The exit cap dominates every other underwriting input by a wide margin, and it is also the input most often handled with the least rigor.
The discipline gap is structural. Sponsors marketing a deal have an incentive to use the lowest defensible exit cap, because lower exit caps produce higher projected returns. LPs evaluating the deal are typically not in a position to challenge the assumption with stronger market data than the sponsor has. The result is a systematic bias toward exit cap optimism that compounds across the industry's underwriting in real time.
The corrective is to stress-test the exit cap, not negotiate it. A serious LP runs the deal at the sponsor's exit cap, at 100 bps wider than the sponsor's exit cap, and at the cap rate spread implied by current 10-year Treasury yields plus a long-run risk premium. If the deal still pencils across all three scenarios, the entry basis is conservative. If the deal only pencils at the sponsor's exit cap, the LP is implicitly underwriting the sponsor's cap rate forecast, which is rarely better than coin-flip accurate over a 7-year hold.
The non-obvious point is that exit cap risk and entry cap discipline are the same trade. A sponsor who buys at a generous cap rate has built in a cushion against unfavorable exit caps. A sponsor who buys at a compressed cap rate has eliminated that cushion and is dependent on the cap rate forecast holding. The cycle's worst-performing deals are not the ones with operational problems. They are the ones with no entry-cap cushion in a cycle where exit caps moved against the underwriting.
5. The Tax Engine: How Real Estate Depreciates on Paper While Appreciating in Value
The second engine of real estate returns is depreciation, and it is the most under-appreciated tax shield in American investing.
The mechanics are unusual. The IRS treats real property as a wasting asset for tax purposes. Residential rental real estate depreciates over 27.5 years on a straight-line basis. Commercial real estate depreciates over 39 years on a straight-line basis. Land is not depreciable, so the depreciable basis is the building component of the purchase price, typically 70 to 85 percent of the total.
The mathematical effect is that an investor who buys a $10 million apartment building, allocating $2 million to land and $8 million to building, generates roughly $290,000 of depreciation expense per year on a straight-line basis. That depreciation is a non-cash deduction that reduces taxable income from the property, often to zero or below, even when the property is generating positive cash flow. The investor receives cash. The IRS sees a loss.
The asset is not, in any economic sense, depreciating. Real property in most markets appreciates over time, often at rates above inflation. The depreciation is a tax fiction that allows the investor to defer income recognition for two to four decades while the underlying asset value compounds. That single mismatch is the structural reason private real estate has produced more long-duration tax-advantaged wealth than any other asset class in the United States.
A stylized 10-year hold makes the divergence concrete. Take a $10 million residential acquisition with $8 million allocated to the building. Straight-line depreciation runs the tax basis down by roughly 29 percent over the hold (a $2.3 million cumulative non-cash deduction). Over the same window, the asset's fair market value typically appreciates 30 to 70 percent depending on the market. The tax basis goes down. The market value goes up. The gap widens monotonically.
The tax basis declines linearly while market value compounds. The gap between the two at sale is the gain. Without further structuring, that gain is taxed at long-term capital gains rates plus a special 25 percent rate on the recaptured depreciation portion. With further structuring, it can be deferred indefinitely.
6. Cost Segregation: Pulling the Depreciation Forward
The straight-line schedule is the default. The professional schedule is cost segregation.
Cost segregation is an engineering-driven study that reclassifies components of a real estate asset from the 27.5 or 39-year schedule into shorter-lived categories: 5-year personal property (carpet, appliances, fixtures), 7-year personal property (specialized equipment), and 15-year land improvements (paving, landscaping, site utilities). On a typical multifamily acquisition, a cost segregation study reclassifies 20 to 35 percent of the depreciable basis into these accelerated categories. On a hospitality or retail acquisition with more specialized improvements, the percentage can run 30 to 45 percent.
The effect is that the bulk of depreciation deductions move from years 28 through 40 forward into years 1 through 15. On a present-value basis, accelerated depreciation is worth dramatically more than straight-line, particularly when paired with bonus depreciation provisions that allow same-year deduction of accelerated components. Bonus depreciation, introduced in various forms since the early 2000s and most expansively under the 2017 Tax Cuts and Jobs Act, has been the single most powerful lever in the modern depreciation toolkit. The TCJA's 100 percent bonus through 2022, followed by the legislated phase-down, drove a generation of high-net-worth and family office allocation into private real estate specifically for the accelerated tax benefit.
The cost segregation trade is not free. The studies cost $5,000 to $25,000 per property. The accelerated deductions reduce the asset's tax basis faster, which means more depreciation recapture at sale. And the recapture on personal property under Section 1245 is taxed at ordinary income rates, which can be higher than the 25 percent unrecaptured Section 1250 rate that applies to building depreciation. A sponsor who runs cost segregation aggressively and exits inside three to five years often pays more total tax than a sponsor who used straight-line, because the timing benefit reverses faster than the tax-rate arbitrage compounds.
The cost segregation play is therefore an LP-level decision, not a deal-level decision. An LP with a long hold horizon and a meaningful tax bill in the early years of ownership wins from cost seg. An LP with a short hold horizon or a low marginal tax rate (a pension fund, a sovereign wealth fund, a tax-exempt foundation) often does not benefit at all. GPs who default to cost segregation on every deal are sometimes optimizing for a different LP than the one who actually owns the equity.
7. The 1031 Exchange: Compounding Without Recognition
The single most consequential provision in the real estate tax code is Section 1031 of the Internal Revenue Code, which allows a real estate investor to exchange one investment property for another of like-kind without recognizing gain on the disposition. The deferred gain rolls into the basis of the replacement property, and the cycle can repeat indefinitely. Properly structured, an investor can compound real estate wealth across forty years and three or four asset cycles without paying federal capital gains tax until final disposition or death.
Death is the closing trick of the structure. Under current law, an heir receives a stepped-up basis on inherited property, which means the deferred gain across the predecessor's lifetime is permanently extinguished. The combination of 1031 deferral during life and stepped-up basis at death is the closest thing to a legal infinite-deferral structure in the US tax code. It is the reason multi-generational real estate wealth in this country is so concentrated and so durable.
The mechanical requirements of a 1031 exchange are precise. The replacement property must be identified within 45 days of the disposition and acquired within 180 days. A qualified intermediary must hold the proceeds in the interim. The replacement property must be of equal or greater value, and any cash received (boot) is taxable in the year of exchange. The like-kind requirement, post-2017, applies only to real property, not personal property, which has implications for cost-segregated assets where Section 1245 components do not roll cleanly.
For LPs allocating to real estate, the 1031 question is structural. A fund vehicle that distributes proceeds to LPs at sale prevents the LP from running their own 1031, because the LP receives cash, not a property interest. A vehicle that allows an opt-out 1031 at the LP level (through a TIC structure, a DST, or a Section 721 UPREIT contribution) preserves the deferral. The presence or absence of LP-level 1031 optionality often determines whether sophisticated taxable LPs participate in a fund at all.
8. The LP Economic Model
LPs are not a monolith. The economic model differs sharply across the LP types that allocate to private real estate, and the reason a deal closes or does not close is often a function of which LP type is at the table.
| LP Type | Primary objective | Tax sensitivity | Liquidity needs | Hold preference |
|---|---|---|---|---|
| Pension fund | Yield + capital appreciation | Tax-exempt, depreciation irrelevant | Long-duration | Long (10+ years) |
| Endowment / foundation | Total return | Tax-exempt | Long-duration | Long (10+ years) |
| Sovereign wealth fund | Capital appreciation, geographic diversification | Tax-exempt | Long-duration | Very long (20+ years) |
| Family office (taxable) | Tax-advantaged compounding | Highly sensitive, depreciation critical | Variable | Long, with 1031 optionality |
| Insurance company | Liability matching, predictable yield | Sensitive | Liability-driven | Medium (7-10 years) |
| Fund of funds | IRR on a fund vintage | Pass-through | Medium | Medium (5-7 years) |
| Individual accredited | IRR plus tax shield | Highly sensitive | Often shorter | Variable |
The tax-exempt LPs (pensions, endowments, sovereign wealth) are indifferent to depreciation, indifferent to 1031, and primarily focused on total return and risk-adjusted yield. They want long holds, predictable distributions, and operational quality. They are the LP base for core and core-plus strategies.
The taxable LPs (family offices, individual accredited, certain insurance vehicles) care intensely about depreciation, intensely about 1031, and operate with after-tax IRRs that can run 200 to 400 basis points above pre-tax IRRs depending on the structure. They are the LP base for value-add and opportunistic strategies, and they are the LPs most willing to accept variable-distribution structures in exchange for tax efficiency.
The misalignment between LP types is a frequent source of fund-formation friction. A GP who builds a fund optimized for tax-exempt LPs and then admits a few taxable LPs late in the raise often discovers that the fund's distribution policy, depreciation handling, and disposition strategy are wrong for the taxable LPs. The fund document language matters less than whether the GP has actually optimized the structure for the LP base it raised.
9. The GP Economic Model
The GP makes money through a layered fee and promote structure that is more asymmetric than most LPs fully appreciate.
The standard institutional GP economic stack:
- Acquisition fee. 1.0 to 2.0 percent of total project cost, paid at closing. On a $100 million deal, this is $1 million to $2 million of cash to the GP, recovered before any operational performance.
- Asset management fee. 1.0 to 1.5 percent of equity invested or total assets under management, paid annually. On a $30 million equity check, this is $300,000 to $450,000 per year, recurring through the life of the investment regardless of performance.
- Construction management fee. 2.0 to 4.0 percent of hard costs on development projects, paid through the construction period. On a $60 million construction budget, this is $1.2 million to $2.4 million.
- Disposition fee. 0.5 to 1.0 percent of sale price at exit, paid at closing of the sale. On a $130 million exit, this is $650,000 to $1.3 million.
- Refinancing fee. Often 0.5 to 1.0 percent of new loan proceeds at any refinance event.
- Promote / carried interest. A percentage share of equity profits above a defined return hurdle, structured through a waterfall.
The promote is the asymmetric upside. The cash fees compensate the GP for running the deal. The promote compensates the GP for outperforming.
A typical deal-level waterfall on a value-add multifamily transaction:
| Tier | Hurdle | LP share | GP share |
|---|---|---|---|
| Tier 1 | Return of capital + 8% pref | 100% | 0% |
| Tier 2 | Catch-up to 80/20 | 50% | 50% |
| Tier 3 | 8% to 15% IRR | 80% | 20% |
| Tier 4 | 15% to 20% IRR | 70% | 30% |
| Tier 5 | Above 20% IRR | 60% | 40% |
The structure is asymmetric in three ways. First, the GP receives cash fees regardless of whether the promote ever vests. Second, the promote is uncapped on the upside, with the GP's share of incremental dollars increasing as performance improves. Third, the GP typically contributes 1 to 10 percent of the equity (the GP co-invest), so the GP's economic exposure to downside is a small fraction of LP exposure while the GP's exposure to upside through the promote is multiples of the co-invest.
A fully promoted value-add deal in the cycle's good years can produce a GP take that runs 30 to 50 percent of total equity profits across all sources, on a 5 to 8 percent equity contribution. That asymmetry is the entire reason the institutional GP business exists as a stand-alone profession.
10. Where the LP and GP Are Pointed in Different Directions
The fee and promote structure creates moments where LP and GP economic interests diverge, and the disciplined LP knows where they are.
Hold extension. The asset management fee accrues for as long as the GP holds the asset. The disposition fee and the promote crystallize only on sale. A GP whose promote is unlikely to vest (an underwater deal) has an economic incentive to extend the hold and continue collecting management fees. The LP, locked into a specific fund duration, often wants the asset sold to redeploy capital. Funds with weak language around mandatory disposition windows produce some of the longest-running disputes in private real estate.
Refinance versus sale. A GP can crystallize promote on a sale. A GP can also crystallize promote on a refinance distribution if the waterfall allows it (interim promote). A GP whose promote vests on refinance has an incentive to maximize refinance proceeds, which can mean over-leveraging the asset and stripping out the equity cushion that protects the LP through the cycle. The GP collects the promote and the LP carries the leverage. Funds with strong LP-protective language require promote crystallization at final sale only, with refinance distributions returning capital first.
Fee creep. Acquisition fees are paid at closing. The structural incentive is to close more deals, not better deals, because every closing generates a fee independent of performance. Disciplined LPs negotiate caps on aggregate fees as a percentage of equity, and they monitor the GP's fee revenue against deployed capital across the fund. A GP whose fee revenue dominates promote revenue is running a deal-flow business, not an investment business.
Cost segregation timing. A taxable LP wants accelerated depreciation in the early years of ownership to offset other taxable income. A tax-exempt LP is indifferent. A GP running a mixed LP base often defaults to aggressive cost seg because it favors the marginal LP who cares most, which can produce recapture exposure for LPs who never benefited. Fund-level cost segregation policy is rarely negotiated explicitly, and it rarely should be a default.
Distribution preference versus reinvestment. A taxable LP often wants distributions they can use for 1031 exchanges or external deployment. A GP often wants to reinvest proceeds into the next deal in the fund, which generates additional acquisition fees and extends the asset management fee stream. The fund's distribution policy is a structural answer to this tension, and it is one of the most negotiated points in any LPA that closes well.
11. The Honest Math of Real Estate Returns
A useful exercise. Take a realistic value-add multifamily deal with a $30 million LP equity check, a 5-year hold, a 17 percent gross IRR at the property level, and a standard 8 percent pref / 80-20 promote structure with the fee stack above. Decompose where the dollars go.
| Source | Approximate $ to GP | Approximate $ to LP |
|---|---|---|
| Acquisition fee | $1.5M | - |
| Asset management fees (5 yr) | $1.8M | - |
| Construction / project mgmt | $1.5M | - |
| Disposition fee | $0.8M | - |
| LP return of capital + 8% pref | - | $44.0M |
| Promote (above 8% pref) | $5.5M | $22.0M |
| GP co-invest return | $1.5M | - |
| Total | $12.6M | $66.0M |
The LP put up $30 million and ended with $66 million on a 17 percent IRR. The GP put up $1.5 million in co-invest and ended with $12.6 million in fees and promote, on roughly the same IRR but with fee leverage that produced a multiple-of-money outcome closer to 8x on the GP's at-risk capital, against 2.2x on the LP's.
The asymmetry is not a scandal. It is the structural compensation for sponsoring deals, taking signature risk on debt, running operations, and producing the deal flow the LP could not produce without the GP. It is the price of access to the strategy, and at a 17 percent gross IRR, it is a price most institutional LPs are willing to pay.
It is also the math that explains why GP franchises trade at the multiples they do. A real estate sponsor producing $50 million of fee revenue and $30 million of average annual promote across a steady-state $5 billion AUM franchise is running an enterprise that rivals private equity general partners on after-tax economics, with lower regulatory burden and longer asset duration. The compounding of GP wealth in real estate happens over twenty to thirty years, across multiple funds, and it produces the family office capital that returns to the asset class as the next cycle's LP base.
12. Closing
Cap rates set the price. Depreciation defers the tax. The waterfall splits the proceeds. None of the three engines, in isolation, explains how money is actually made in real estate. All three together produce a return profile no other major asset class replicates.
The professionals run the three engines as a single integrated machine. Entry cap rates are negotiated against the LP base's tax horizon. Cost segregation is calibrated to the LP base's marginal rate. Promote structure is matched to the strategy's expected return distribution. Hold periods are set to optimize the depreciation schedule and the 1031 calendar simultaneously. The deal LPs see in the marketing materials is the visible artifact of a structure that was engineered around the engines, not around the building.
The amateurs run the three engines as three separate trades. They negotiate cap rates without thinking about exit. They claim depreciation without understanding recapture. They sign waterfalls without modeling the hold-extension scenario. The amateurs do not lose money on every deal. They lose money on the deals where the cycle moves against them and the structure had no integrated buffer.
The honest test of any real estate sponsor or any LP allocator is whether they can articulate, in a single sentence, how the cap rate forecast, the tax structure, and the waterfall fit together for the specific deal in front of them. If they cannot, they are running one engine. The professionals are running three.