When Tax Equity Runs Out of Basis: Four Gates, Two Ledgers, and Losses That Die on Exit

September 17, 2026 · Dezzmond Team

Financial Modeling Data Analysis Excel

A tax equity model allocates 99% of losses to the investor and reports a return. Whether the investor can actually deduct those losses is a different question, governed by four separate limitations, any one of which can suspend them.

And a suspended loss does not wait indefinitely for the investor. When the investor exits, losses still suspended are extinguished — they do not transfer to the buyer and they do not offset the gain on sale. Benefit that was priced into the deal simply disappears.

This is the third post in the tax structuring series and the last on tax equity structures. You will get the difference between the two ledgers that track an investor's position, the four gates a loss has to clear, why the ITC basis reduction makes the problem worse before depreciation even starts, what a deficit restoration obligation actually buys, and why the depreciation election changes the DRO by a factor of seven.

ℹ️ Note: This describes how these rules work in practice. It is not tax advice — partnership loss limitations are technical, interact with one another, and depend on each partner's own circumstances.

Why Does an Investor Run Out of Room?

Because the benefits are front-loaded and the capital is not. An investor contributes a fixed amount and is then allocated credits and deductions substantially larger than it, in the first few years.

That is the deal working as intended. Accelerated depreciation on a capital-intensive asset generates deductions well in excess of the contribution that funded it, and the whole purpose of the structure is to move those deductions to someone who can use them.

The constraint is that partnership tax does not allow losses to be deducted without something to absorb them. The investor's ability to take the allocation is capped, and when the cap binds the loss does not disappear — it suspends, and waits.

Two Ledgers, Not One

The single most common modelling error in this area is conflating the capital account with outside basis. They are different measures, they move differently, and they do different jobs.

	704(b) capital account	Outside basis
What it tracks	The partner's book equity in the partnership	The partner's tax basis in its partnership interest
Can it go negative?	Yes, with a DRO	Never
What it governs	Whether an allocation has economic effect; liquidating distributions	Whether a loss can be deducted
Affected by 163(j)	No	Yes
When exhausted	Requires a DRO to keep allocating	Losses suspend under 704(d)

Norton Rose Fulbright illustrates the separation with a concrete example: the section 163(j) interest limitation "only effect[s] outside basis and not capital accounts," which is only possible if the two are genuinely distinct ledgers.

The practical consequence is that a model tracking one number is tracking the wrong thing half the time. Norton Rose notes that tax equity models commonly carry six capital accounts — two for the investor, two for the sponsor, and two at partnership level — precisely because a single balance cannot answer both questions.

The Four Gates

A loss allocated to a partner must pass four separate limitations before it is deducted. They apply in order, and clearing one says nothing about the next.

Gate	Test	If it fails
§704(d)	Loss cannot exceed outside basis	Suspended, carried forward until basis exists
§465	Loss cannot exceed the amount at risk	Disallowed, carried forward indefinitely
§469	Passive losses offset only passive income	Suspended until passive income or disposition
§461(l)	Excess business loss limitation	Deferred

The at-risk gate is the one most likely to surprise. Because a partnership is a pass-through, the at-risk amount is measured at partner level — and a partner with no personal liability for partnership debt is not at risk for that debt. A limited partner with a $500,000 contribution alongside $5m of recourse partnership debt is at risk for $500,000.

For a tax equity investor the practical reading is that the four gates are not alternatives. A loss has to survive all of them, and a model that tests only outside basis will report deductions that the at-risk or passive activity rules independently disallow.

How Does the ITC Basis Reduction Make It Worse?

By removing half the credit from basis before depreciation is even calculated. Norton Rose sets out the mechanic: "section 50(c)(3)(A) provides for the basis reduction of half of the ITC," and importantly the ordering — "the partnership's basis is reduced by the ITC basis adjustment and the balance of the basis is then subject to the depreciation rules."

So the sequence is: claim the credit, reduce basis by half of it, then depreciate what is left.

That compresses the room available twice over. The credit is claimed immediately, reducing the asset base. The reduced base then generates less depreciation in absolute terms, but the deductions still arrive faster than the contribution can absorb them.

Norton Rose gives the worked illustration, and it is worth reading slowly:

"a partner contributes $100 and is allocated a $200 credit with $100 basis reduction and $150 loss, resulting in a negative $150 capital account"

One hundred of capital, and a capital account at minus one hundred and fifty within the period. That is not a stressed case — it is the ordinary arithmetic of a credit-driven structure, and it is the reason the deficit restoration obligation exists at all.

What Does the DRO Actually Buy?

Permission for the capital account to go negative, backed by a real obligation. As the flip post covered, a DRO is a promise to contribute on liquidation to restore a deficit. Norton Rose describes the consequence on transfer: a buyer of the partnership interest "steps into the seller's negative capital account and the associated DRO."

The size varies enormously, and Norton Rose ties it directly to the depreciation election:

"tax equity deals in which 12-year straight-line depreciation is elected and the tax equity investor's DRO cap is as little as 15% of its capital contributions"
"bonus depreciation deals" where the "DRO cap is more than 100% of the tax equity investor's capital contributions"

A factor of roughly seven between the two, driven by a depreciation election.

That is a genuinely useful structuring lever and it is under-used. An investor unwilling to accept a large DRO is not necessarily unwilling to do the deal — it may be willing on a slower depreciation election. Electing 12-year straight-line slows the deductions, flattens the capital account trajectory, and can bring the required DRO inside what the investor will accept. The cost is that the benefits arrive later, which reduces their present value and therefore the price.

The trade is explicit: depreciation speed versus DRO size, and it should be run as a two-way sensitivity rather than settled by assumption.

Why Do Distributions Make It Worse?

Because cash paid out reduces outside basis, and outside basis is what absorbs losses. A project that distributes well is consuming the same capacity its deductions need.

That is counterintuitive and it matters. Distributions are the good news in a project — the asset is performing, cash is flowing, the investor is receiving its return. Each distribution also reduces the investor's basis in its partnership interest, which brings the 704(d) ceiling down and accelerates the point at which further losses suspend.

The effect compounds with the front-loading problem. In the early years the investor is simultaneously receiving cash distributions (reducing basis), being allocated large depreciation deductions (reducing basis), and taking a credit that carried a basis reduction with it. Three claims on the same limited pool, all arriving at once.

Two practical consequences.

Distribution timing is a tax variable, not only a cash one. Deferring distributions in the early years preserves basis for absorbing deductions, which can be worth more than receiving the cash sooner. That trade should be run explicitly rather than settled by the cash waterfall.

A high-performing project can suspend more losses than a mediocre one. Strong early cash generation draws down basis faster. It is an uncomfortable result and it is the arithmetic.

How Does Partnership Debt Change the Answer?

It relieves the first gate and does nothing for the second, which moves where the constraint binds.

A partner's outside basis includes its share of partnership liabilities. So project-level debt increases the investor's outside basis and, on its own, creates more room before losses suspend under 704(d).

The at-risk rules do not work the same way. As the earlier search of the limitation hierarchy sets out, a partner is at risk only for amounts it stands to lose — and a partner with no personal liability for partnership debt is not at risk for that debt. The illustration is stark: a limited partner with a $500,000 contribution alongside $5m of recourse partnership debt is at risk for $500,000.

Put those together and the interaction is precise. Partnership debt raises the 704(d) ceiling and leaves the 465 ceiling where it was. The binding constraint simply moves from the first gate to the second, and a model testing only outside basis will conclude the losses are deductible when the at-risk rules independently disallow them.

This connects back to the structuring point in the flip post. Tax equity resists project-level debt because it ranks ahead of the distributions the investor needs, which is why back-leverage sits at the sponsor holdco instead. That structural choice also removes the basis the debt would have contributed — so back-leverage protects the investor's cash flow priority and tightens its loss absorption capacity at the same time. Both effects are real and they point in opposite directions.

What Is a Book-Up?

A revaluation of partnership assets that resets capital accounts, and Norton Rose notes it can "eliminate or reduce a DRO."

That sounds like a clean solution and is only a partial one, for a reason that leads directly to the next section. Norton Rose is explicit that the apparent benefit is offset: upon investor exit, "the investor's suspended losses are eliminated and do not get transferred to the buyer," which cancels out what looks like a windfall from DRO avoidance.

What Happens to Suspended Losses on Exit?

They die. This is the most consequential sentence in this post.

Norton Rose states it twice, in two different pieces. Suspended losses "neither offset the gain nor transfer to the buyer" on a sale of a partnership interest. And: "the investor's suspended losses are eliminated and do not get transferred to the buyer."

Three consequences follow, and they are not symmetrical.

The benefit is lost, not deferred. A suspended loss is not a timing difference that resolves on exit. If the investor sells while losses remain suspended, that value is gone — for everyone.

It does not even shelter the exit. The intuition that suspended losses would at least offset gain on sale is wrong. They do not.

The buyer gets nothing. A purchaser of the interest inherits the negative capital account and the DRO, but not the suspended losses attached to it. It takes the liability and not the asset.

There is one piece of good news, and it applies only if the investor stays. Norton Rose notes that "there is not limit on the ability to use losses that are suspended under section 704(d) in future years once the losses are released" — unlike net operating loss carryforwards, which carry their own restrictions. Suspended 704(d) losses are patient. They just cannot be sold.

What Is a Stop Loss Provision?

A cap in the partnership agreement that prevents allocations from driving the capital account past the point the DRO will support.

The logic is straightforward once the DRO is understood. A deficit restoration obligation is capped — at 15% of contributions in a slow-depreciation deal, at more than 100% in a bonus depreciation one. An allocation that would push the capital account below that cap has no economic effect to support it, because there is no obligation standing behind the deficit it would create.

A stop loss provision handles that by halting the allocation at the cap and redirecting the excess to the other partner, typically the sponsor. The sponsor is then allocated deductions it may be no better placed to use — which is the whole reason the structure exists — but the allocation is at least respected.

Two features are worth carrying into a model.

It is a different failure from suspension. A suspended loss under 704(d) is still the investor's; it waits for basis. A loss redirected by a stop loss provision was never allocated to the investor at all. The first is a timing problem, the second is a permanent transfer of benefit to the party least able to use it.

It changes what the investor will pay. Pricing is built on the benefits the investor expects to receive. If the stop loss provision is expected to redirect a meaningful share of deductions, the investor discounts them — which is another reason the DRO cap and the depreciation election have to be settled together rather than sequentially.

How Do You Model This in Excel?

With two ledgers and four tests, not one balance and a loss allocation.

The inputs

Assumptions, labelled as such:

Tax equity contribution               $75,000,000
ITC claimed                           $75,000,000
ITC basis reduction (50%)             $37,500,000
DRO cap accepted                      50% of contribution  = $37,500,000
Depreciation election                 bonus vs 12-year straight line

The two ledgers, tracked separately

Capital_Account(t) = Capital_Account(t−1)
                   + income allocated
                   − loss allocated
                   − distributions
                   − ITC basis reduction effect
                   → may go negative, floored at −DRO_Cap

Outside_Basis(t)   = Outside_Basis(t−1)
                   + income allocated
                   − loss allowed
                   − distributions
                   → floored at ZERO, never negative

Build these as two rows, not one. The floor on the second row is what generates suspension.

The suspension test

Loss_Allocated      = 99% × partnership loss
Loss_Allowed        = MIN(Loss_Allocated, Outside_Basis)
Loss_Suspended(t)   = Loss_Allocated − Loss_Allowed
Cumulative_Suspended = running total

The DRO constraint

DRO_Required        = MAX(0, −Capital_Account)
DRO_Headroom        = DRO_Cap − DRO_Required
Allocations_Blocked = DRO_Required > DRO_Cap

Where allocations are blocked, the deduction cannot be allocated at all — a different failure from suspension, and one that reduces what the investor will pay rather than merely delaying it.

The exit test — the number most models omit

Suspended_At_Exit   = Cumulative_Suspended at flip / sale date
Value_Destroyed     = Suspended_At_Exit × Marginal_Tax_Rate

On a deal carrying $20m of suspended losses at exit and a 21% rate, that is $4.2m of benefit extinguished — value the pricing assumed and nobody receives. It should appear as a line in the base case, not as a footnote.

The sensitivity worth running

Depreciation election  ×  DRO cap  →  cumulative suspended losses at exit

Bonus depreciation, DRO 50%    → high suspension
12-yr straight line, DRO 15%   → low suspension, later benefits

The two levers move in opposite directions and are usually chosen by different people — the election by the tax team, the DRO by the investor's credit committee. This grid is where those two decisions meet.

ℹ️ Note: Model all four gates, not just 704(d). At-risk and passive activity limitations operate at the investor's own level and on its own circumstances, which means the project model cannot answer them alone — but it can flag where the allocations assume they are satisfied.

To run the full version — both ledgers across the deal life, all four limitations tested, and suspended losses valued at exit — prompt Dezzmond with your allocation schedule and depreciation election.

What Do Investors and Sponsors Actually Check?

Are capital account and outside basis modelled separately? They diverge, and only one of them floors at zero.
What is cumulative suspended loss at the expected exit? That value is destroyed, not deferred.
Is the DRO cap sufficient for the depreciation election chosen? Bonus depreciation can require a DRO above 100% of contribution.
Has a slower depreciation election been priced? It can cut the required DRO to as little as 15%.
Are the at-risk and passive activity limitations addressed? They apply at investor level and are frequently assumed away.
Was the ITC basis reduction applied before depreciation? Half the credit, then depreciate the remainder.
What does the buyer of the interest actually inherit? The negative capital account and DRO, but not the suspended losses.

Frequently Asked Questions

What is the difference between a capital account and outside basis?

The 704(b) capital account tracks book equity and can go negative where a deficit restoration obligation exists. Outside basis tracks tax basis in the partnership interest, can never be negative, and is what governs whether a loss is deductible.

What happens when outside basis reaches zero?

Further losses are suspended under section 704(d) and carried forward until the partner has basis available again. There is no limit on using them in future years once released.

Do suspended losses transfer when the investor sells?

No. They neither offset the gain on sale nor pass to the buyer — they are eliminated. The buyer inherits the negative capital account and the DRO without them.

Why does the depreciation election change the DRO?

Because it changes how fast the capital account goes negative. Deals electing 12-year straight-line depreciation have been done with DRO caps as low as 15% of contributions, while bonus depreciation deals can require more than 100%.

Does the ITC reduce basis before or after depreciation?

Before. Basis is reduced by half the credit under section 50(c)(3)(A), and the remaining basis is then subject to the depreciation rules.

Closing: The Benefit Has to Fit Through Four Doors

A tax equity structure is usually described in terms of what is allocated. The harder question is what is absorbed — and the answer is governed by a set of limitations that live at the investor's level, in its own tax circumstances, on ledgers the project model does not naturally carry.

Three things are worth taking from it. The capital account and outside basis are different ledgers and only one of them can go negative. The DRO and the depreciation election are the same decision approached from two sides, and a factor of seven separates the outcomes. And suspended losses are extinguished on exit — which turns a timing problem into a permanent one at exactly the moment the deal is being celebrated.

That closes the tax equity structures sub-series. The next three posts take monetisation after the IRA — transferability, direct pay, and the hybrid structures that combine a transfer with tax equity — where the question stops being how to bring a taxpayer into the project and becomes whether you need one at all.

Sources: Norton Rose Fulbright — Demystifying Capital Accounts in Tax Equity Transactions · Norton Rose Fulbright — Capital Account Implications for Renewable Energy Tax Credits · IRS — Partner's Outside Basis (LB&I Process Unit)