The Rational Borrower

The Trade That Looks Stupid

An institution buys $1,000,000 in tokenized US Treasuries. The yield is 4.5% per year. That same day, they walk into a lending protocol and borrow $666,667 in stablecoins against those Treasuries at 10% annual interest.

They earn 4.5%. They pay 10%. That's a negative carry of 5.5% on every borrowed dollar. Over a year, the interest bill is $66,667. The Treasury yield pays them $45,000. The gap is a $21,667 loss — before fees, before gas, before anything else.

Most people stop here. They look at the borrow rate, compare it to the collateral yield, and walk away. That's the wrong conclusion. The reasons why reveal something fundamental about how institutional capital actually works — both in traditional finance and increasingly onchain.

The Collateral Is Locked. Borrowing Is the Only Key.

Many of the highest-quality tokenized assets onchain can't be freely sold. Franklin Templeton's BENJI, WisdomTree's WTGXX, and similar institutional-grade tokens are restricted securities. No public secondary market. No DEX pair. No order book. You buy them through an authorized channel, and then you hold them.

The yield is attractive. The asset is creditworthy. The capital is frozen.

For these holders, borrowing isn't an alternative to selling. It's the only mechanism for extracting liquidity from the position. The relevant comparison isn't "10% cost vs. 4.5% yield." It's "10% cost vs. no liquidity at all."

That reframes the entire trade. The borrow rate isn't the cost of leverage. It's the price of access to capital that would otherwise be completely locked.

Even If You Could Sell, You Wouldn't

Suppose the asset isn't restricted. You could sell your Treasuries on the open market tomorrow. Would that be the better move?

Usually not, for two independent reasons.

The tax problem. Selling a profitable position triggers a taxable event. If the institution bought $1M in Treasuries at a cost basis of $850,000, selling generates $150,000 in capital gains — a tax bill of $30,000 to $60,000 depending on jurisdiction.

The annual cost of a $667K loan at 10% is $66,667. But that's a deductible expense. And critically, it's temporary — when the loan is repaid, the institution still owns the Treasuries. Selling destroys the position permanently. The tax is paid permanently. The borrow cost is the rental price of capital; the tax is the destruction of it.

This isn't an exotic DeFi insight. It's the same reason high-net-worth individuals borrow against stock portfolios instead of selling shares, and why real estate investors refinance instead of liquidating properties.

The exposure problem. Beyond tax, there's a simpler reason: the institution wants to own the asset. They're deliberately long Treasuries — because they believe in the yield, because their fund mandate requires the allocation, because they need it on their balance sheet, or because they expect capital appreciation as rates decline.

Selling exits the trade. Borrowing preserves it. The institution gets liquidity and keeps the position. In portfolio construction terms, borrowing against an asset is the difference between "I need cash" and "I no longer want to own this." Those are very different statements.

The Real Cost Is Not What You Think

Look at the arithmetic more carefully. The institution pays 10% on $666,667 — that's $66,667 per year. But the $1,000,000 in collateral doesn't stop earning just because it's been posted. BENJI continues accruing 4.5%, generating $45,000 per year.

The headline rate is 10%. The net cost, after the collateral's ongoing yield, is 2.2% on the full position. That's the actual price of $667K in liquid, deployable capital.

And that assumes the borrowed money sits idle. Nobody borrows $667K to let it sit.

What the Borrowed Capital Funds

The strategies below explain why paying that 2.2% net cost is actively profitable.

The institution deposits the borrowed USDC into a yield-bearing stablecoin vault — a diversified vehicle targeting 8% APY from a blend of Treasuries, institutional private credit, and DeFi yield strategies. From there, the playbook branches.

Direct yield capture. Deposit $667K into an 8% vault while paying 5–10% to borrow it. With the collateral's native yield offsetting most of the cost, total return on the position exceeds what holding alone would produce.

Recursive leverage ("looping"). The yield vault token is itself composable collateral. Deposit it on a secondary lending market, borrow more USDC at ~5%, buy more vault tokens. Each iteration earns the same spread on a larger capital base, compounding a modest per-dollar edge into a significant portfolio-level return.

Liquidity provision. The borrowed USDC becomes the pair side of an LP position — in a stablecoin pool on a DEX, for example — earning trading fees on capital that would otherwise have been locked.

Arbitrage and market-making. Market makers borrow against inventory to fund operations across venues — capturing DEX–CEX price dislocations, vault NAV mispricings, or cross-chain spreads. The borrow rate is operational cost; the arbitrage revenue more than covers it.

Basis trading and hedging. Hold the spot asset, borrow stablecoins to fund a short on the corresponding perpetual when rates are positive. The funding income alone can exceed the borrow cost.

NAV discount capture. Yield-bearing RWAs periodically trade below their net asset value on secondary markets — typically when rate spikes force leveraged positions to unwind, creating mechanical selling pressure. Borrowing against existing holdings funds the purchase of the discounted asset with leverage, amplifying the return when it reprices to NAV. A 0.4% discount at 7.5x leverage becomes a 3% gain in days.

The specific strategy matters less than the principle: borrowed capital is deployed, not parked. The borrow rate is the cost of putting idle capital to work, and the return on deployment is what justifies it.

The Numbers

The "Hold only" row is the baseline — what the institution earns by doing nothing. Every row below it shows the incremental impact of borrowing and deploying.

Scenario A: BENJI at 5% SAV Rate

$1M BENJI at 4.5% native yield. SAV borrow: 5%. Secondary borrow: ~5%. Deployment yield: 8%.

Scenario B: sUSDe at 7% SAV Rate

$1M sUSDe at 12% native yield. SAV borrow: 7%. Secondary borrow: ~5%. Deployment yield: 8%.

Each loop iteration adds roughly 2 percentage points of return. The collateral's native yield sets the floor; the loop determines how far above it you can climb. In both cases, the collateral has never been sold. The borrower retains full beneficial ownership throughout.

One note on execution: much of the emerging literature on RWA leverage focuses on the friction of looping when each cycle requires subscribing through a transfer agent and waiting for T+1 or T+3 settlement — a process that can take weeks to build a 5x position. That friction is real, but it's a primary market problem. When the collateral is already a token in the borrower's wallet and the looping happens entirely through onchain lending markets and DEX liquidity, the full position can be constructed in minutes. The settlement constraint disappears because the loop never leaves the chain.

Case Study: Buying a Dollar for 99.6 Cents

In March 2026, a trader on Kamino executed a trade that illustrates several of these dynamics in a single transaction — and surfaces one more reason institutions borrow against low-yielding assets.

PRIME is a yield-bearing RWA token with a net asset value of approximately $1.023. It earns a modest yield, and like most RWAs, it normally trades close to NAV. A few days before the trade, stablecoin borrow rates on Kamino spiked above PRIME's native yield, turning the loop spread negative.

This is the rate inversion scenario described in the risk section below. And the leveraged holders did exactly what rational operators do: they unwound. To unwind, they had to sell PRIME on secondary markets. That selling pressure pushed the market price down to roughly $1.018 — a 0.4% discount to NAV — even though nothing about the underlying asset had changed.

The trader saw the gap and acted. They bought $10,000 of PRIME at the discounted price and looped the position to 7.5x leverage on Kamino, borrowing USDS. Seven days later, the position was worth $10,150 — after approximately $100 in borrow interest. A 1.5% return in one week, annualizing to roughly 78% APY.

Rate spike → loop spread inverts → leveraged holders unwind → selling pressure → PRIME drops below NAV → discount trader enters with leverage → NAV converges → profit.

The meta-lesson: one borrower's rational exit creates another borrower's rational entry. The ecosystem is self-correcting — and at every stage, the borrowing is the mechanism that makes the trade possible.

This kind of NAV discount isn't a one-off anomaly. It's a structural feature of yield-bearing RWAs that trade on secondary markets. Whenever rates spike, leveraged positions unwind, and mechanical selling pushes price below intrinsic value. The discount persists until someone borrows capital, buys the asset at the lower price, and either holds for convergence or loops the position to amplify the return.

For lenders, this dynamic matters. It means demand doesn't evaporate during rate spikes — it shifts. The loopers who were borrowing at 5% leave, and the discount traders who borrow at the higher rate enter. The demand is countercyclical: it increases precisely when rates are elevated, because the arbitrage opportunity widens as prices dislocate further from NAV.

The Deeper Reason: Yield-Bearing Working Capital

Everything above describes a borrower with a specific strategy: deploy capital, capture a spread, repeat. But there's a category of institutional participant for whom the borrowed capital isn't a yield play. It's operational infrastructure.

Consider a DeFi market maker. At any given moment, they have capital parked across a dozen venues: posted as margin on a perpetual exchange, sitting in a DEX LP position waiting for flow, held in reserve on a lending protocol to backstop liquidations, queued for an arbitrage that hasn't materialized yet. Most of that capital is doing nothing most of the time. It's waiting. And while it waits, it earns zero.

This is the dead capital problem. A market maker with $5M in working capital might have $3M of it idle at any given moment — not because they're bad at their job, but because operational readiness requires capital to be available, not deployed. Posted margin isn't earning. Reserve capital isn't earning. The USDC sitting in a wallet waiting for the next arbitrage opportunity isn't earning.

Now replace that idle USDC with a yield-bearing stablecoin that earns 8% while sitting in a wallet, while posted as margin, while providing LP liquidity, while waiting for the next trade. The capital never stops earning, even when it's not actively being used.

That changes the entire cost structure. On $3M of working capital, the difference between 0% and 8% is $240,000 per year. That's not a yield strategy. It's an operational advantage — a structural reduction in the cost of running a DeFi business.

For these participants, the borrow rate isn't compared against the collateral's yield or against any specific deployment return. It's compared against the cost of capital sitting dead. If your alternative is 0% on idle USDC, then paying 5–10% to borrow against an asset you want to hold — and immediately converting that capital into a yield-bearing form that earns while it works — isn't a loss. It's the elimination of a loss you were already absorbing silently.

A yield-bearing stablecoin that's accepted as collateral, margin, and LP base pair across DeFi venues isn't a yield product. It's the base currency for capital-efficient operations. The bigger the operation, the more idle capital you have, and the more the passive yield matters. A market maker running $20M across ten venues with 60% average idle time is leaving $960,000 per year on the table by holding plain USDC.

This thesis is distinct from looping. Looping is a deliberate yield strategy with a specific risk profile. The yield-bearing working capital thesis is a passive, structural improvement to the economics of every DeFi participant who uses stablecoins — from market makers to LPs to DAOs to treasury managers. It requires no active strategy. It just requires that the default form of your USDC earns yield instead of zero.

Two More Tailwinds

Token incentives reduce the effective rate. During bootstrapping phases, lending protocols offer token-based incentives to early participants. Borrowers earn governance tokens through programs like Strands, which convert to transferable tokens at the end of each rewards season. These can offset 100–300 basis points of the effective borrow rate. A 10% sticker rate becomes 7–8% after the token allocation. A 5% rate becomes 2–3%. These subsidies are time-limited — typically 12 months — but during the bootstrapping window, they make an already rational trade even more compelling.

Fixed rates eliminate volatility risk. In variable-rate lending markets, borrow rates can spike to 15%+ during periods of high utilization. A borrower who enters at 5% can find themselves paying triple in hours. Fixed-rate, fixed-term lending eliminates this risk entirely. The borrower locks in their cost of capital for the duration. For the looping strategy specifically, rate certainty on the primary leg is critical: it isolates risk to only the secondary market's variable rate, rather than having both legs move against the borrower simultaneously.

The Trade Nobody Questions in TradFi

The repo market — the largest short-term funding market in the world at over $4 trillion in daily volume — operates on exactly this principle. A hedge fund posts US Treasuries yielding 4% as collateral, borrows cash at the repo rate, and deploys that cash into higher-yielding instruments. The borrow rate routinely exceeds the collateral yield. Nobody calls this irrational.

Onchain lending against tokenized RWAs is the same trade, on different infrastructure. The collateral is the same (Treasuries, money market funds). The economic logic is the same (borrow against low-risk assets to deploy into higher-returning strategies). The difference is that onchain infrastructure makes it composable, transparent, and accessible 24/7 — with collateral that's verifiable in real time rather than buried in a counterparty's balance sheet.

What This Means for Lenders

Borrowing demand is structural. Looping, LP provision, market-making, and the yield-bearing working capital thesis all create persistent demand for stablecoin borrowing. This isn't speculative leverage that disappears when sentiment shifts. It's a function of quantifiable economics.

The borrower's incentive to repay is strong. Default forfeits institutional-grade collateral the borrower needs and can't easily replace. The strategies funded by the loan generate returns that comfortably cover the borrow cost. Incentives are aligned.

Leverage is downstream. The lender's capital is secured at 150% by high-quality collateral. The borrower's looping or operational activity happens on secondary markets. It doesn't affect the SAV-level collateral cushion.

A high rate is a signal, not a warning. When a borrower willingly pays 7–10%, it means the capital has a productive destination that exceeds the cost. That productivity is what ensures the lender gets paid.

What Can Go Wrong

Leverage amplifies returns and exposure. The risks are real.

Rate inversion. If secondary borrow rates rise above the deployment yield, the loop spread turns negative. Sophisticated operators monitor this and unwind before it becomes material. As the PRIME case study shows, this unwinding can itself create secondary opportunities — but looping returns are not permanent and require active management.

NAV decline. A credit event or depeg in the yield-bearing stablecoin can trigger secondary liquidation. At 5x leverage, a 2–3% NAV decline breaches the threshold. At 3x, the buffer is substantially wider. Conservative operators cap leverage at 3x.

Liquidity constraints. Unwinding a looped position requires selling or redeeming the yield-bearing stablecoin at par. Deep DEX liquidity and protocol-level redemption mechanisms are prerequisites.

Smart contract risk. An exploit in any contract in the chain compromises the position. Multiple independent audits, formal verification, and phased TVL caps are baseline requirements.

Every Dollar Loses. The Portfolio Wins.

The trade that looked stupid is a structured strategy institutions have run for decades, adapted for onchain infrastructure.

The borrow rate is high because the collateral is restricted and borrowing is the only path to liquidity. It's tolerable because selling would cost more in taxes. It's less than it appears because the collateral keeps earning, reducing the net cost to a fraction of the headline rate. It's justified because the borrowed capital is deployed into strategies — from yield vaults to loops to NAV arbitrage — that more than cover the cost.

When rate spikes force one class of borrower to exit, the resulting price dislocations create opportunities for another class to enter. The ecosystem is self-correcting, and borrowing is the mechanism at every stage. For market makers, LPs, and operational participants, there's an even simpler case: the borrowed capital, converted into a yield-bearing form, earns a return while doing its job. The dead capital problem disappears.

On each individual borrowed dollar, the institution may be losing money. On the full position, they're earning 10–20% on assets that were yielding 4–12% passively.