Optimizing the Capital Stack: How Leverage Amplifies (and Destroys) Equity Returns in Commercial Real Estate

Key Takeaways

How Leverage Amplifies Equity Returns

Leverage in commercial real estate works the same way it works in every other asset class: you borrow capital at a fixed cost and invest it in an asset that (you expect) earns more than the borrowing cost. The difference between the asset's return and the borrowing cost accrues entirely to the equity. When the asset return exceeds the borrowing cost, every dollar of debt magnifies the equity return. When the asset return falls below the borrowing cost, every dollar of debt erodes it.

The quantitative relationships that govern leverage are the foundation for everything in the capital stack. Start with a $50M stabilized industrial property. In-place NOI: $3.25M. Going-in cap rate: 6.50%. If you buy it with all equity (no debt), the unlevered return equals the property's yield. Your Year 1 cash-on-cash return is the cap rate: 6.50%. Your unlevered IRR over a 5-year hold with 3% annual NOI growth and a 7.00% exit cap depends on the exit price, but call it approximately 8.4%.

Now introduce debt. Borrow $30M at 60% LTV with a fixed rate of 5.75%, 30-year amortization. Your annual debt service is approximately $2.10M. Your equity investment is now $20M instead of $50M. Cash flow after debt service in Year 1: $3.25M minus $2.10M = $1.15M. Cash-on-cash return to equity: $1.15M / $20M = 5.75%. The cash yield to equity appears lower than the unlevered cap rate, but that is because the debt service includes principal amortization. The equity's total return, which includes the principal paydown benefit and the leveraged exit, is substantially higher.

Over a 5-year hold with the same 3% NOI growth and 7.00% exit cap, the levered equity IRR is approximately 14.2%, compared to 8.4% unlevered. The equity multiple is 1.82x, compared to 1.45x. Leverage produced 580 bps of incremental IRR and 0.37x of incremental equity multiple, using $30M of debt that costs 5.75%. The equity investor earned a 14.2% IRR on $20M of capital while the debt investor earned a 5.75% return on $30M. The spread between the property return and the debt cost was captured entirely by the equity.

The Leverage Amplification Formula

The relationship between levered and unlevered returns can be expressed precisely. The levered return on equity is:

When R_asset exceeds R_debt, the amplification term is positive. More debt (higher D/E ratio) means a higher equity return. When R_asset is below R_debt, the amplification term is negative. More debt means a lower equity return. When R_asset equals R_debt, leverage has no effect on equity returns. You are borrowing at exactly the rate the property earns. The formula explains why leverage works in real estate: cap rates have historically exceeded mortgage rates for stabilized properties, so the amplification term has been positive for most deals in most environments. But it also explains why leverage becomes dangerous when rates rise or property yields compress.

Note that this formula is a single-period simplification. In practice, the relationship is dynamic: R_asset changes as NOI grows or declines, R_debt changes for floating-rate loans, and the D/E ratio changes as the loan amortizes. A full discounted cash flow (DCF) model captures these dynamics. The formula is useful for intuition and quick screening, not for final underwriting.

Positive vs Negative Leverage

The terms "positive leverage" and "negative leverage" describe which side of the amplification equation a deal falls on. They are the single most important concepts in capital stack optimization, and they are frequently misunderstood.

Positive Leverage Defined

Positive leverage exists when the property's yield exceeds the cost of debt. In a single-period analysis, the test is whether the cap rate exceeds the mortgage constant (annual debt service divided by loan amount). In a multi-period DCF, the test is whether the unlevered IRR exceeds the effective cost of borrowing (including origination fees, prepayment penalties, and interest rate on the debt).

When positive leverage is present, every incremental dollar of debt (up to the point where lender constraints bind) increases the equity IRR. The equity investor is borrowing cheap capital and investing it in a higher-returning asset. The spread between the two rates is pure amplification that flows entirely to the equity.

Example: a stabilized multifamily property with a 5.75% cap rate and a 5.25% mortgage constant (after accounting for amortization). The spread is 50 bps. At 60% LTV (D/E ratio of 1.5x), the static leverage amplification is 50 bps times 1.5 = 75 bps of incremental equity return. At 70% LTV (D/E ratio of 2.33x), it is 50 bps times 2.33 = 117 bps. Each 10 points of additional LTV increases the equity return by more, not less, because the D/E ratio rises geometrically. This is the mathematical reason why leverage is seductive: the marginal return to each additional dollar of debt increases as leverage increases.

Negative Leverage Defined

Negative leverage exists when the cost of debt exceeds the property's yield. In this environment, every additional dollar of debt reduces the equity return. The equity investor is borrowing expensive capital and investing it in a lower-returning asset. The negative spread drains value from the equity.

Example: the same multifamily property at a 5.75% cap rate, but with rates that have risen so the mortgage constant is now 6.25%. The spread is negative 50 bps. At 60% LTV, the negative amplification is 50 bps times 1.5 = 75 bps of equity return destruction. At 70% LTV, it is 50 bps times 2.33 = 117 bps. The more you borrow, the worse your equity return gets. And just as positive leverage accelerates geometrically, negative leverage decelerates geometrically. The damage compounds as LTV rises.

Negative leverage does not necessarily mean the deal is bad. It means the deal does not benefit from debt. If the unlevered return is acceptable on its own (for example, 9% unlevered IRR on a core-plus multifamily deal), the investment may still make sense. But the capital stack should be structured with less debt, not more. The optimal leverage in a negative-leverage environment is zero, or as close to zero as the investor's capital availability allows.

The Breakeven Test

The breakeven point between positive and negative leverage occurs when the property yield equals the cost of debt. In practice, practitioners test this by comparing two numbers:

The distinction between the interest rate and the mortgage constant matters. A loan at 5.75% interest with 30-year amortization has a mortgage constant of approximately 7.00% (because the annual payment includes both interest and principal). A cap rate of 6.50% exceeds the interest rate (positive by 75 bps) but is below the mortgage constant (negative by 50 bps). Whether leverage is "positive" depends on which comparison you use. The correct comparison for single-period cash flow analysis is cap rate vs mortgage constant. The correct comparison for total-return analysis (including principal paydown benefit) is more nuanced and is best resolved through a full DCF.

As a practical matter, most institutional investors use the cash-on-cash comparison (cap rate vs mortgage constant) as a screening tool and the full DCF as the underwriting tool. If a deal shows negative leverage on a cash-on-cash basis, the analyst flags it and examines whether the total return from NOI growth, principal paydown, and exit value overcomes the negative current yield. In many value-add strategies, the answer is yes: the going-in cap rate is below the mortgage constant, but the projected stabilized yield (after renovations and lease-up) exceeds it. The capital stack is built on forward expectations, not current yields.

Five-Level Leverage Sensitivity Table

The most useful analytical tool for capital stack optimization is a sensitivity table that shows the same deal at multiple leverage levels. Below is the $50M stabilized industrial property analyzed at five LTV levels: 0%, 50%, 60%, 70%, and 80%.

Base assumptions. Purchase price: $50M. In-place NOI: $3.25M (6.50% cap rate). NOI growth: 3% per year. Hold period: 5 years. Exit cap rate: 7.00%. Debt terms: 5.75% fixed rate, 30-year amortization, 10-year term. No prepayment penalty. No mezz or preferred equity. Closing costs and transaction fees excluded for clarity.

Several patterns emerge from the table. First, equity IRR increases monotonically with leverage. Going from 0% to 80% LTV more than doubles the equity IRR, from 8.4% to 18.9%. This is the leverage amplification effect at work: the cap rate (6.50%) exceeds the effective cost of debt, so every additional dollar of debt increases the equity return. Second, the cash-on-cash yield declines with leverage. The Year 1 cash-on-cash drops from 6.50% at 0% LTV to 4.50% at 80% LTV. This is because the amortizing debt service consumes an increasing share of current NOI. Third, DSCR tightens as leverage rises. At 70% LTV, the DSCR is 1.33x, which meets most institutional lender minimums. At 80% LTV, the DSCR drops to 1.16x, which is below most lender minimums (typically 1.20x to 1.25x for stabilized industrial). Fourth, the equity multiple follows the same pattern as IRR. It rises from 1.45x to 2.30x as leverage increases, reflecting both the leveraged exit and the smaller equity denominator.

The table shows that leverage works in this scenario. But the table alone does not show where leverage stops being productive and starts being reckless. For that, you need to look at the marginal return to each incremental unit of leverage and weigh it against the incremental risk.

Marginal IRR per Unit of Leverage

Look at the IRR gains between each leverage level:

• 0% to 50% LTV: +440 bps of IRR. This is the largest absolute gain. The first 50 cents of debt per dollar of asset value produces the most return per unit of leverage.
• 50% to 60% LTV: +140 bps. Still productive, but the marginal return is declining.
• 60% to 70% LTV: +290 bps. A jump, because the D/E ratio accelerates from 1.50x to 2.33x. Each incremental LTV point carries more leverage amplification as the equity base shrinks.
• 70% to 80% LTV: +180 bps. The marginal IRR gain is lower than the 60% to 70% step, even though the D/E ratio jumps from 2.33x to 4.00x. The diminishing return is starting to show.

The marginal gain is not the whole story. The marginal risk is equally important. At 80% LTV, the DSCR is 1.16x. A 5% decline in NOI (a single tenant vacancy, a maintenance surprise) pushes the DSCR below 1.00x, meaning the property cannot service its debt from operating income. The equity investor must fund the shortfall from reserves or additional capital. At 70% LTV, the same 5% NOI decline brings the DSCR to 1.26x, still above the threshold. The risk difference between 70% and 80% LTV is not 10 points of leverage. It is the difference between surviving a moderate stress scenario and needing to write a check.

Leverage Sensitivity Chart

Interest Rate Impact on Optimal Leverage

The leverage sensitivity table above assumes a fixed 5.75% interest rate. Change the rate and the entire analysis shifts. The cap rate vs cost of debt relationship is the engine of leverage amplification, and interest rates are half of that equation.

Consider three interest rate scenarios on the same $50M industrial deal at 70% LTV:

At a 4.75% interest rate, the mortgage constant (6.26%) is close to the cap rate (6.50%). The leverage spread is a thin +24 bps on a cash-flow basis, but the total-return leverage (which includes principal paydown and leveraged exit) is strongly positive. The equity IRR at 70% LTV is 19.4%.

At 5.75%, the mortgage constant (7.00%) exceeds the cap rate (6.50%) by 50 bps. On a cash-flow basis, this is technically negative leverage: Year 1 cash-on-cash (5.33%) is below the unlevered yield (6.50%). But the total-return leverage is still positive because principal paydown and the leveraged exit more than offset the negative current yield. The equity IRR is 17.1%, still well above the 8.4% unlevered return.

At 6.75%, the mortgage constant (7.78%) exceeds the cap rate by 128 bps. The negative cash flow leverage is now severe: Year 1 CoC is only 4.00%, and the DSCR has dropped to 1.19x. The total-return IRR is still above the unlevered return (14.8% vs 8.4%), so leverage is creating value in total-return terms. But the margin of safety is thin. A small NOI decline pushes the DSCR below 1.0x. A larger NOI decline triggers covenant violations and potential default.

The lesson is that the optimal LTV depends on the interest rate environment. At 4.75% rates, 70% LTV is comfortably within the positive leverage zone and 80% LTV is defensible. At 5.75% rates, 70% LTV is the practical ceiling for most stabilized deals. At 6.75% rates, 60% LTV may be the prudent maximum, with 70% LTV possible only for high-conviction value-add deals where NOI growth is expected to improve the coverage ratio quickly.

Floating-Rate Risk and Leverage

The analysis above assumes fixed-rate debt. Floating-rate debt introduces an additional risk dimension: the leverage spread can change during the hold period. A deal that starts with positive leverage at a 5.00% SOFR spread plus 200 bps (7.00% all-in) can shift to negative leverage if SOFR rises 150 bps during the hold.

Floating-rate borrowers often purchase interest rate caps to limit upside rate exposure. An interest rate cap puts a ceiling on the reference rate (for example, capping SOFR at 4.00%), which limits the maximum all-in cost of debt. The cost of the cap is part of the effective cost of borrowing and should be included in the leverage analysis.

The Mortgage Bankers Association's commercial/multifamily research tracks quarterly origination volumes by loan type (fixed vs floating, bank vs nonbank). Their data shows a continued shift back toward fixed-rate originations in 2026 as borrowers lock in rates and lenders price floating-rate loans with wider spreads. For capital stack optimization purposes, the choice between fixed and floating is itself a leverage decision: floating rate gives you a lower starting cost of debt (and therefore a wider positive-leverage spread) at the cost of rate risk that can eliminate the spread entirely.

Stress-Testing the Rate Environment

Any capital stack analysis that does not include interest rate stress testing is incomplete. At minimum, underwrite the deal at three rate levels:

1. Base case: Current market rate (or the locked rate if you have a commitment). This is the scenario you expect.
2. Upside stress (+100 to +150 bps): Rates rise by 100 to 150 bps from current levels. Test whether the deal still covers debt service and meets return thresholds at this rate. If you are using floating-rate debt, this is the rate that hits after your cap expires.
3. Downside scenario (−75 to −100 bps): Rates decline. Test whether the return improvement is large enough to warrant a refinancing event. If the deal is fixed-rate with a prepayment penalty, a decline in rates is an opportunity cost, not a windfall.

Stress testing is not a formality. It is the mechanism that prevents you from building a capital stack that works in exactly one interest rate scenario. The CBRE Cap Rate Survey for H1 2026 provides a reference frame for where cap rates sit relative to borrowing costs across property types. When the spread is tight (as it is for core office and some gateway multifamily), the margin for error is thin and the optimal leverage point is lower.

Diminishing Returns and Risk-Adjusted Thinking

The sensitivity table shows that equity IRR increases with leverage at every LTV level. So why not maximize leverage? Because the analysis so far has measured only the numerator (return) without adjusting for the denominator (risk). Risk-adjusted thinking asks a different question: how much incremental risk am I taking for each incremental unit of return?

The Diminishing Returns Curve

Plot the equity IRR against LTV and the curve is concave. It rises steeply at lower leverage levels and flattens at higher levels. The first 50 points of LTV produce 440 bps of IRR gain (8.8 bps per LTV point). The last 10 points (70% to 80%) produce only 180 bps (18 bps per LTV point). On an absolute basis, the marginal return per LTV point at high leverage (18 bps) is higher than at low leverage (8.8 bps). But the marginal risk per LTV point at high leverage is dramatically higher.

Risk in this context means several things simultaneously:

• DSCR erosion. At 80% LTV, the DSCR is 1.16x. A 10% NOI decline drops the DSCR to 1.04x. At 70% LTV, the same NOI decline drops the DSCR from 1.33x to 1.20x. The 80% LTV structure survives by a thread. The 70% LTV structure survives comfortably.
• Refinancing risk. At 80% LTV, the borrower must refinance or sell at a valuation that supports the outstanding loan balance. If cap rates rise 50 bps (property value declines approximately 7%), the LTV at maturity exceeds 80% and the borrower must bring cash to the refinancing table. At 70% LTV, the same cap rate expansion leaves the LTV at approximately 75%, still refinanceable.
• Equity cushion. At 80% LTV, the equity cushion is $10M on a $50M property. A 20% decline in property value eliminates the entire equity position. At 70% LTV, the equity cushion is $15M, and a 20% decline leaves $5M of residual equity. The 80% LTV investor is one stress scenario away from total loss. The 70% LTV investor has a buffer.
• Operational constraint. Higher leverage means less free cash flow for capital expenditures, leasing costs, and unforeseen repairs. A building that requires $500K of roof work in Year 3 is a manageable event at 60% LTV (the equity investor uses excess cash flow) and a capital call event at 80% LTV (there is no excess cash flow).

The Risk-Adjusted Return Framework

A simple risk-adjusted framework assigns a volatility or risk score to each leverage level and computes a risk-adjusted return. While institutional investors use different methodologies (Sharpe-like ratios, value-at-risk, stress-scenario analysis), the core logic is the same: divide the expected return by a measure of the downside.

One practical approach is to compute the "breakeven NOI decline" at each leverage level. This is the percentage decline in NOI that brings the DSCR to 1.00x (the point where the property cannot cover debt service from operations). A larger breakeven number means more cushion.

The "IRR per 1% of breakeven cushion" column reveals the trade-off. At 50% LTV, you earn 0.28% of IRR for each percentage point of NOI-decline cushion. At 80% LTV, you earn 1.37% per point. That looks like a better ratio, but the cushion itself is shrinking. The 80% LTV investor is earning a high return per unit of safety margin precisely because the safety margin is dangerously thin. A 14% NOI decline (one anchor tenant departure in a multi-tenant industrial portfolio, or a 6-month vacancy during re-tenanting) wipes out the DSCR.

The optimal leverage point for risk-adjusted returns depends on the investor's risk tolerance. Conservative institutional investors (insurance companies, pension funds) typically target 55% to 65% LTV, accepting lower IRR in exchange for a wide breakeven cushion. Opportunistic investors and private equity funds target 70% to 75% LTV, accepting thinner cushions for higher IRR. Anything above 75% LTV on a stabilized deal typically requires mezzanine debt or preferred equity to fill the gap between the senior lender's maximum and the desired total leverage, which introduces intercreditor complexity and higher blended cost of capital.

DSCR and LTV Guardrails

Lender underwriting standards impose hard constraints on leverage that override the investor's return optimization. The two primary guardrails are the loan-to-value ratio (LTV) and the debt service coverage ratio (DSCR). Both must be satisfied simultaneously, and the binding constraint (whichever is more restrictive) determines the maximum loan amount.

LTV Constraints

LTV is the ratio of the loan amount to the appraised property value. Lender maximums vary by property type, loan type, and market conditions:

• Agency (Fannie Mae, Freddie Mac): Up to 80% LTV for stabilized multifamily, 75% for most other property types.
• CMBS conduit: Typically 65% to 75% LTV, depending on property type and DSCR.
• Life company: 55% to 65% LTV for stabilized assets. Conservative by design.
• Bank: Varies widely. 60% to 75% LTV for stabilized, lower for transitional or construction. The Federal Reserve's SLOOS data tracks how bank lending standards change over time. As of Q3 2026, the SLOOS reports continued easing at large banks and mixed signals from regional and community banks.
• Bridge/debt fund: Up to 75% to 80% of as-stabilized value for transitional deals. Higher LTV but also higher rates and shorter terms.

The appraised value is itself a judgment call. Lenders may use the lesser of the purchase price and the appraised value (the "lesser of" rule), which limits LTV on acquisitions where the buyer pays a premium to market. For refinancings, the appraised value is the only reference point, and disagreements between the lender's appraiser and the borrower's expectations are common.

DSCR Constraints

DSCR is the ratio of net operating income to debt service. It measures the property's ability to cover its debt payments from current operations. Lender minimums vary by property type and loan type:

• Stabilized multifamily: 1.20x to 1.25x (agency), 1.25x to 1.30x (CMBS, bank).
• Stabilized industrial/retail: 1.25x to 1.35x.
• Office: 1.30x to 1.50x (reflecting higher perceived risk in the current environment).
• Hospitality: 1.40x to 1.60x.

The DSCR constraint often binds before the LTV constraint. On the $50M industrial deal at a 6.50% cap rate and 5.75% interest rate, the LTV constraint might allow 75% leverage ($37.5M loan), but the DSCR constraint at 1.25x minimum allows only $3.25M / 1.25 = $2.60M of annual debt service, which translates to a maximum loan of approximately $37.1M (74.2% LTV). The DSCR is the binding constraint by a small margin. At higher interest rates, the DSCR binds more aggressively because each dollar of loan balance requires more annual debt service.

According to LoopNet's leverage ratio explainer, the leverage ratio (another name for the LTV constraint) is the single most common screening criterion that CRE lenders apply at the term sheet stage. Deals that exceed the lender's maximum LTV are declined before the DSCR, debt yield, or any other credit metric is evaluated. This makes the LTV constraint a gate, not a sliding scale: you either pass or you do not.

Debt Yield

A third constraint that has gained prominence since 2020 is the debt yield: NOI divided by the loan amount. Debt yield is independent of the interest rate and amortization schedule, which makes it a more stable measure of loan-level risk than DSCR (which changes with rates and amortization). Typical lender minimums are 8% to 10%. On the $50M industrial deal, the debt yield at 70% LTV is $3.25M / $35M = 9.3%, above the typical minimum. At 80% LTV, the debt yield drops to $3.25M / $40M = 8.1%, barely above the minimum and vulnerable to any NOI decline.

Common Leverage Mistakes

The mechanics of leverage are well understood. The mistakes practitioners make are not about arithmetic. They are about framing, assumptions, and the failure to stress-test.

1. Optimizing leverage at origination without stress-testing the exit. The capital stack that maximizes IRR at a 6.50% cap rate and 5.75% fixed rate may not survive a market where cap rates have expanded 50 bps at exit. The exit assumption is half the return. If the exit cap rate widens by even 25 bps, the levered equity return drops disproportionately because the debt balance does not decline as fast as the property value. Practitioners who optimize the capital stack for origination-day conditions without modeling exit-cap sensitivity are building a single-scenario deal.

2. Confusing the interest rate with the cost of debt. The interest rate on the loan (5.75%) is not the cost of debt. The cost of debt includes origination fees (1% to 2% of the loan amount), legal costs, rate lock fees, and any prepayment penalty that will be incurred at exit. A 5.75% loan with 1.5% of origination fees and a 1% prepayment penalty at Year 5 has an effective cost of debt closer to 6.15%. The leverage spread (cap rate minus cost of debt) is thinner than it appears.

3. Treating DSCR as a constraint rather than a signal. Lender minimum DSCRs (1.20x, 1.25x) are floors, not targets. A deal that underwrites to exactly the lender's minimum has no cushion for underperformance. Institutional investors typically target 10 to 15 points of DSCR above the lender minimum (for example, a 1.35x target when the lender minimum is 1.25x). The gap between the lender minimum and the investor target is the margin of safety for operational surprises, tenant credit events, and capital expenditure needs.

4. Ignoring the interaction between leverage and promote economics. In LP/GP structures with promote tiers, higher leverage amplifies the GP's promote by compressing the equity check and increasing the equity IRR. A GP who pushes for higher leverage is not just optimizing the deal return. They are also optimizing their own promote economics, sometimes at the expense of LP risk-adjusted returns. LPs should evaluate leverage recommendations in the context of how they affect the GP's promote waterfall, not just the blended equity return.

5. Using the cap rate as the sole leverage test without considering NOI growth. A deal with a 5.50% going-in cap rate and a 6.00% mortgage constant appears to have negative leverage. But if the business plan projects 20% NOI growth over three years (through renovation, lease-up, or mark-to-market rent increases), the stabilized yield on cost may be 7.00% or higher. The cap rate vs mortgage constant test is a snapshot. For value-add and development deals, the leverage test should use the projected stabilized yield, not the going-in cap rate. The risk is that the projected growth does not materialize, leaving the deal permanently in negative leverage territory.

6. Assuming that higher leverage always means higher risk. In some structures, higher leverage reduces the equity investor's dollar exposure to the deal. If the investor has a $50M allocation and is choosing between one all-equity deal at $50M and three 70% levered deals at $50M each (with $15M of equity per deal), the levered portfolio has more diversification. The risk at the deal level is higher (higher leverage per deal), but the risk at the portfolio level may be lower (diversification across three assets instead of one). Capital stack optimization at the portfolio level is a different exercise from deal-level leverage optimization.

7. Failing to model the refinancing event. Most institutional CRE loans have 5 to 10 year terms, even with 25 or 30 year amortization schedules. The loan matures before the amortization period ends, and the borrower must refinance or sell. If interest rates have risen by 100 bps during the hold, the refinancing rate is higher, the DSCR at the refinancing is tighter, and the maximum loan amount may be lower than the outstanding balance. The borrower must bring cash to close the refinancing gap. This "maturity wall" risk is a form of leverage risk that does not appear in a simple IRR analysis that assumes a clean exit or sale.

Optimize It in Apers

Related Articles

• LP/GP Structures, Promote, Catch-Up, and Clawback. How the equity waterfall above the debt interacts with leverage. Higher leverage amplifies the GP promote by compressing the equity check and accelerating the return past promote hurdle rates.
• Bank Debt: Recourse vs Nonrecourse. The senior debt layer that leverage analysis sits on top of. How recourse provisions, covenants, and carveout guaranties constrain the leverage decision.
• Mezzanine Debt and Intercreditor Agreements. When the optimal total leverage exceeds the senior lender's maximum LTV, mezzanine debt fills the gap. How mezz pricing and the intercreditor apparatus affect blended cost of capital and the leverage spread.
• Preferred Equity: Priority of Payments and Redemption. The equity-side alternative to mezzanine debt for filling the leverage gap. How preferred equity structures differ from mezz in their effect on the leverage analysis.
• Bridge Loans, Floating Rate Risk, and Exit Assumptions. Floating-rate bridge debt introduces rate risk into the leverage analysis. How to model the interaction between floating rates and leverage optimization.