A new apartment building has to charge enough rent to pay its lender, pay its investors and pay its bills. The building used throughout this article is Reference Project B, the podium project, one of the two illustrative buildings used across this site, with every assumption listed at the end: a 120-unit mid-rise over a structured garage, costing $30,000,000, which needs $2,415 per unit per month: $1,081 for debt service, $614 for the investors' return and $719 for operating expenses. Roughly $1,695 of that, the first two pieces, is the price of capital.
That raises an obvious question. If the cost of capital is the largest piece of rent, and governments borrow more cheaply than developers, could a public body finance the same building and charge less? The short answer is yes. The longer answer is that the saving comes from somewhere specific, and where it comes from determines who pays.
This article runs six ways public capital can enter a project through the same rent-math used across this site, and shows what each does to required rent, who bears the cost and what the risks are. It does not recommend any of them. All figures are illustrative assumptions, listed in full below.
Why capital has a price
Capital is not free for anyone, public or private, because supplying it means two things: giving up whatever else the money could have done, and accepting the risk that it does not come back. A lender who advances $19,500,000 for thirty years could have bought government bonds instead; the interest rate on the loan has to beat that alternative by enough to cover the chance of default. An investor who puts in $10,500,000 of equity stands last in line if the building fails, and asks for a higher return in exchange.
In the base project the lender is assumed to charge 6.5% and the equity investors to require an 8% cash-on-cash yield, producing $1,479,039 of annual debt service and $840,000 of annual equity return, or $1,695 of the $2,415 monthly rent. Any model that lowers rent through financing has to lower one of those two numbers, or reduce the amount of capital they apply to.
Where public capital's advantage comes from
When a public body finances housing more cheaply, the advantage comes from one or more of three sources. They are often bundled together in practice, but they are different things with different payers, and it helps to keep them apart.
- Tax exemption. Interest on certain bonds issued by public entities is exempt from federal income tax, so investors accept a lower rate. The building saves interest; the federal treasury forgoes the tax it would have collected. Other taxpayers cover the difference.
- Public credit. A city, state or agency with the power to tax can borrow more cheaply than a developer because lenders expect to be repaid even if a single project fails. The risk has not gone away. It has moved from the project to the public balance sheet.
- Patient return. A public fund or agency may accept a lower or slower return than a private investor would. The gap between what it accepts and what the market would require is a subsidy, paid by whoever owns the fund: taxpayers, pensioners or the beneficiaries of whatever else the money could have done.
None of these is illegitimate. But the clearest way to compare models is to ask, for each one, which of the three is doing the work, and who is on the other side of it.
Six models and what each does to rent
A. Tax-exempt bond financing
Public agencies can issue tax-exempt private-activity bonds and lend the proceeds to a housing project; the Low-Income Housing Tax Credit is commonly paired with them. We assume the project replaces a taxable loan at 6.5% with a bond at 4.5%, a 200-basis-point spread, on the same $19,500,000 of debt. The mortgage constant over a 30-year amortization falls from 7.585% to 6.080%, and annual debt service falls from $1,479,039 to $1,185,644, a saving of $293,396 per year.
200-basis-point lower interest rate on $19,500,000 of debt
−$293,396/year
−$214/month
Mortgage constant at 6.5%, 30 years = 7.585%; at 4.5%, 30 years = 6.080%
$19,500,000 × 7.585% = $1,479,039 per year; $19,500,000 × 6.080% = $1,185,644 per year
Difference $293,396 ÷ (1 − 0.05 vacancy) = $308,838 of required revenue
$308,838 ÷ 120 units ÷ 12 months = $214 per unit per monthIllustrative; the taxable and tax-exempt rates are assumptions, not market quotes.
Bonds are not free to issue. Underwriting, bond counsel, a trustee and a rating all cost money, and we assume those issuance costs at 2.5% of the bond, or $487,500, added to project cost and financed alongside it. That adds back about $24 per month. The net effect is a required rent of $2,224, a reduction of $190 per month.
Required rent: taxable loan versus tax-exempt bond
Difference−$190/month
The gross interest saving is −$214/month; issuance costs of $487,500 add back about $24/month. Rent or income restrictions and ongoing compliance costs, which usually accompany bond financing, are not modeled.
Three caveats. Bond financing usually comes with rent or income restrictions on some units and an ongoing compliance obligation, neither of which is in this arithmetic. The interest saving is not free: bondholders pay no federal tax on the interest, so the federal treasury forgoes revenue that other taxpayers make up. And whether the bond rate really sits 200 basis points below the taxable rate depends on the issuer's credit and the market at the time; the spread can be narrower.
B. Public land
A land bank or public agency may already own a site and contribute it to the project. Removing the $3,600,000 of land from the capital budget removes $2,340,000 of debt and $1,260,000 of equity, which together required $278,285 a year of debt service and equity return. Grossed up for vacancy and spread across the units, that is a reduction of $203 per month.
The cost is the land's opportunity value: what the public body could have sold, leased or used the site for. Because the land is already owned, this cost appears in no budget line, which makes it easy to overlook and hard to compare with a cash subsidy of the same size.
C. Patient public equity
A public developer, a state fund or a public pension vehicle could supply the $10,500,000 of equity itself. If it accepts a 4% cash yield instead of the 8% a private investor requires, the required return falls by $10,500,000 × 4% = $420,000 a year, which is $307 per month of rent, or 12.7% of the base figure. This is the largest single-model effect in the set.
It is also the clearest case of the third source of advantage. The four points of forgone yield are a subsidy, paid by the fund's owners, whether taxpayers or pensioners. And the public entity now carries the construction and lease-up risk that private equity would have absorbed: if costs run over or units lease slowly, the loss lands on the public balance sheet. Some practitioners regard this as a reasonable use of public capacity; others argue that governments underprice the risk they take. The rent-math cannot settle that; it can only show what the yield gap is worth.
D. Revolving construction loan fund
The base project carries $1,800,000 of financing costs: construction-period interest and loan fees on the construction loan. A public revolving fund that lends during construction at below-market rates could reduce that line. We assume it halves it, to $900,000, which lowers required rent by $51 per month.
A revolving fund is only revolving if its capital comes back. Its cost to the public is the below-market interest it accepts plus the losses on projects that fail to repay. If it is not repaid, the fund shrinks and finances fewer projects in the next cycle.
E. Direct capital grant
The simplest model is a grant that reduces the amount the project has to finance. We assume $50,000 per unit, or $6,000,000 in total, which brings required rent to $2,075, a reduction of $339 per month. The cost to the public is exactly the grant, paid up front, with no repayment. Grants are transparent and easy to compare, and they raise the sharpest version of the allocation question, which gets its own section below.
F. Property tax abatement
The last model works on the operating side rather than the capital side. We assume property taxes of $2,500 per unit per year, or $300,000, which is 30% of the $984,000 annual operating budget. A full abatement removes that $300,000 of operating expenses, and after the vacancy gross-up that is $219 per month of rent. Note that this model is the one most sensitive to the choice of building: the tax line is an assumed 1.0% of development cost, so a costlier building has a larger abatement to give away.
The municipality forgoes $300,000 a year for as long as the abatement runs. Whether that is a real cost is the "but-for" question: if the building would not have been built without the abatement, the city gave up revenue it would never have collected; if it would have been built anyway, the city gave away $300,000 a year. That question is contested in almost every abatement debate and rarely provable either way.
| Model | What it changes | Illustrative rent effect | Who bears the cost | Main risks |
|---|---|---|---|---|
| A. Tax-exempt bonds | Interest rate on $19,500,000 of debt, 6.5% to 4.5% | −$190/month net | Federal treasury (forgone tax); other taxpayers | Rent restrictions; compliance cost; spread may be narrower |
| B. Public land | Removes $3,600,000 of land from cost | −$203/month | Public body (opportunity value of the site) | Hidden cost; site may not fit the project |
| C. Patient equity | Equity yield 8% to 4% on $10,500,000 | −$307/month | Fund owners: taxpayers or pensioners | Construction and lease-up risk on public books |
| D. Revolving loan fund | Halves $1,800,000 of construction financing cost | −$51/month | Fund (below-market interest and losses) | Non-repayment shrinks the fund |
| E. Capital grant | $6,000,000 removed from financed cost | −$339/month | Grant-making body, up front | Depth versus breadth; no recycling |
| F. Tax abatement | Removes $300,000/year of property tax | −$219/month | Municipality, every year of the abatement | But-for question; ongoing revenue loss |
The tools compound. Bond financing, public land and a full abatement together bring the required rent to roughly $1,827 per month on this project. That figure rebuilds the pro forma once rather than adding the three savings together: total development cost becomes $26,887,500 ($30,000,000 less the $3,600,000 site, plus the $487,500 of issuance costs), all of it financed in the same 65/35 proportions with the debt at the 4.5% bond rate, and operating expenses fall to $684,000 once the $300,000 tax line is abated. But each stacked layer is a further public cost on a different ledger: the federal treasury, the land-owning agency and the municipal budget are all contributing to that one number.
Depth or breadth
A grant of $50,000 per unit lowers rent by $339. Suppose the goal is instead a rent of $1,200, a reduction of $1,215 from the base. At the same rate, that takes a grant of about $179,109 per unit, or $21,493,029 for the whole building, which is 72% of its total development cost. The public would be paying for nearly three quarters of a building it does not own.
Under the modeled assumptions; the choice between depth and breadth is a policy decision, not an arithmetic one
Now hold the budget fixed instead of the rent. With $6,000,000 to spend, the money can produce 120 units at $2,075, or about 33 units at $1,200, or any mix in between. Neither answer is right. Deeper subsidy reaches households with lower incomes; broader subsidy reaches more households. Every subsidy program makes this choice, explicitly or by default; it is the central allocation decision in subsidy design. The rent-math can price the options; it cannot rank them.
What public capital can do that private capital cannot
Public capital also brings things a private pro forma cannot.
- Lower cost of capital. Through the three sources above, a public body can genuinely reduce the largest component of rent, and the reduction is real for the tenant even though someone else is paying.
- Patience through cycles. Private capital tends to arrive when returns are high and leave when they are not. A fund that does not need an IRR within a fixed window can keep building when private investors have stopped, and can hold assets rather than sell them at the exit cap rate.
- Alignment with public goals. A public owner can keep rents restricted indefinitely, house particular populations or locate near public investments, without the tension a private developer faces between mission and return.
- Land already owned. A public body holding a site can accept a lower return on it than a private seller would, and can decide that housing is the highest use even when a sale would fetch more.
What can go wrong
The risks below are not arguments against public capital. They are the places where the arithmetic depends on things going as planned.
Execution capacity
Development is a specialized business: assembling sites, managing entitlement, supervising a general contractor, leasing up and operating. Whether a public agency has that staff in-house, at the scale required, comes before the financing question, because cheap capital does not lower rent on a building that is late, over budget or badly run.
Political cycles and multi-year projects
A building takes years from concept to occupancy; budgets and administrations turn over on shorter cycles. Funding appropriated one year can be cut the next, and a project stalled mid-stream is the most expensive outcome of all. The site's analysis of what a year of delay costs applies to public timelines: a single year of delay adds $71 per month to required rent on this project, which offsets a large share of the advantage from any one model above.
Balance-sheet exposure
When public credit stands behind a project, a cost overrun or a failed lease-up becomes a public liability. On models C and D this exposure is direct; on model A it can be contingent, appearing only when something goes wrong, which makes it easy to under-count when the program is designed.
Crowding out
Public capital can displace private capital that would have built anyway, in which case the public cost buys no additional housing. Public projects also compete for the same contractors, materials and sites as private ones, and in a tight market that competition can raise costs for everyone, including the public project itself. Economists disagree about how large this effect is and when it appears.
Depth, breadth, restrictions and time
- Subsidy depth per unit versus number of units. Every dollar spent making one unit deeper is a dollar not spent on another unit; the arithmetic above does not resolve which matters more.
- Compliance and restriction costs. Rent and income limits reduce revenue; monitoring them adds cost. Neither is in the figures above.
- Slower timelines. Public procurement, approvals and oversight tend to take longer than private equivalents, and the rent-math prices delay at $71 per month per year on this building.
Reference Project B (the podium project): an illustrative 120-unit mid-rise over a structured garage of 120 spaces; total development cost $30,000,000 ($250,000 per unit); operating expenses $8,200 per unit per year; required rent $2,415 per unit per month. Why this project: it is capital-intensive and carries an ordinary private capital stack, which is what each public model is being compared against. Both reference projects are defined side by side on our Methodology page.
Base budget: land $3,600,000; hard costs $19,500,000 (including 120 structured parking spaces at $30,000 each, $3,600,000); soft costs $3,300,000 (of which permit and impact fees $1,440,000, $12,000 per unit); developer fee $1,200,000 (4% of total development cost); financing costs (construction-period interest and loan fees) $1,800,000; lease-up and operating reserves $600,000.
Base financing: construction/permanent debt at 65% of cost ($19,500,000) at 6.5% interest, 30-year amortization (mortgage constant 7.585%), annual debt service $1,479,039; equity 35% ($10,500,000) at an 8% target cash-on-cash yield, $840,000 per year. Required NOI $2,319,039.
Base operations: operating expenses $8,200 per unit per year ($984,000): property taxes $2,500 per unit ($300,000, an assumed 1.0% of development cost), insurance $1,000, operations $3,500, owner-paid utilities $800 and capital reserves $400. Vacancy 5%; required gross revenue $3,476,883; base required rent $2,415 per unit per month ($1,081 debt service + $614 equity return + $719 operating expenses).
Model A: we assume a taxable loan at 6.5% is replaced by a tax-exempt bond at 4.5% on the same $19,500,000 of debt (mortgage constant 6.080%, annual debt service $1,185,644, saving $293,396 per year, −$214 per month). We assume bond issuance costs of 2.5% of the bond ($487,500), added to project cost and financed, adding about $24 per month. Net required rent $2,224 (−$190 per month).
Model B: we assume a public agency or land bank contributes the site, removing $3,600,000 of land from the capital budget: $2,340,000 less debt and $1,260,000 less equity, $278,285 less annual debt service and equity return, −$203 per month.
Model C: we assume a public entity supplies the full $10,500,000 of equity and accepts a 4% cash yield instead of 8%, reducing the required equity return by $420,000 per year, −$307 per month (12.7% of base rent).
Model D: we assume a public revolving loan fund halves the $1,800,000 financing-cost line to $900,000, −$51 per month.
Model E: we assume a capital grant of $50,000 per unit ($6,000,000), bringing required rent to $2,075 (−$339 per month). At that rate, a rent of $1,200 (−$1,215) would require about $179,109 per unit, $21,493,029 in total, 72% of total development cost; a fixed $6,000,000 produces 120 units at $2,075 or about 33 units at $1,200 ($6,000,000 ÷ $179,109 = 33.5, stated as the 33 units the money fully covers).
Model F: we assume property taxes of $2,500 per unit per year ($300,000, or 1.0% of development cost and 30% of the operating budget) and a full abatement, −$219 per month.
Stacking illustration: we assume bond financing, public land and a full abatement together, computed by rebuilding the pro forma once rather than by adding the three individual savings. Total development cost $26,887,500 ($30,000,000 − $3,600,000 of land + $487,500 of issuance costs); debt $17,476,875 at the 4.5% bond rate (mortgage constant 6.080%); equity $9,410,625 at 8%; operating expenses $684,000 after the abatement; required rent roughly $1,827 per month. Because the whole capital stack is repriced at the bond rate, this figure is not the sum of the −$190, −$203 and −$219 shown for models A, B and F alone.
Delay reference: one year of delay adds $71 per month to required rent on this project, from the site's delay analysis.
Every rent figure uses the site's shared rent-math: annual debt service is the mortgage payment on (capital cost × loan-to-cost) at the interest rate over the amortization period; annual equity return is capital cost × (1 − loan-to-cost) × target equity yield; required NOI is the sum of the two; required revenue is (required NOI + operating expenses) ÷ (1 − vacancy); and required rent per unit per month is required revenue ÷ units ÷ 12. Intermediate values are not rounded; displayed figures are rounded to the nearest dollar.
Each of the six models is computed alone against the base project. The stacking figure is the only combination shown. Rent and income restrictions, compliance costs, and the federal revenue cost of the tax exemption are described but not quantified. Public execution risk, timeline differences and crowding-out effects are described but not quantified.
The +$24 per month issuance-cost effect treats the $487,500 as added to total development cost and financed in the same debt-and-equity proportions as the rest of the project, with the debt at the bond rate. The same treatment carries into the stacking figure, which is computed by rebuilding the pro forma once at the reduced cost, the bond rate and the abated operating budget, so it is smaller than the sum of the three models taken separately.
The rates, spreads, fees and tax levels used above are inputs we chose to make the six comparisons legible on a single project. They are not terms quoted by any agency, lender or municipality, and no external data set was consulted to set them. Public financing programs differ substantially between states and even between cities within a state, and the size of every saving shown here depends on terms that are published program by program. A reader evaluating a specific program should work from that program's own published terms and fee schedule and rerun the arithmetic.
A narrower bond spread. If the tax-exempt rate sits 100 rather than 200 basis points below the taxable rate, the interest saving in model A roughly halves, and issuance costs take a larger share of it.
Higher issuance costs or a smaller bond. Issuance costs are partly fixed, so on a smaller project they consume a larger share of the interest saving.
Rent restrictions that cut revenue. If bond or grant terms require some units to rent below the modeled figure, the remaining units must carry more, or the subsidy must deepen.
Public construction timelines. A year of public-process delay adds $71 per month on this project, per the site's delay analysis, which can offset much of a financing advantage.
Cost overruns landing on the public balance sheet. Under models C and D, an overrun raises the capital the public entity must supply and lowers the return it actually earns; the modeled savings assume the budget holds.
Whether the project would have been built anyway. If private capital would have financed the same building, the public cost buys no additional housing, and the rent reductions above are transfers rather than new supply.
What this means
Can government finance housing more cheaply? Yes. On the modeled building, each of the six models lowers required rent, by between $51 and $339 per month alone and by more when combined. The reason is not that public bodies have found a cheaper way to build. It is that they are paying part of the cost somewhere else, accepting a risk that private capital would have charged for, or accepting less than the market would demand. Each is a legitimate choice. None is free.
The useful questions, then, are not whether public capital is cheaper but which payer is on the other side of the saving, how much it costs per unit, how many units that buys, and whether the public body can deliver the building on time and on budget. Two programs with the same headline rent can have very different answers to all four.
Readers who want the other half of this picture, what private capital charges and why, can turn to our analysis of developer profit, and to what happens to rent when interest rates rise, which is the same arithmetic run in the opposite direction on Reference Project A, the site's cheaper garden building, whose base rent is $1,996 rather than $2,415.