Floating-Rate Mechanics, SOFR & the Index Transition

A floating-rate loan (also called an adjustable-rate or variable-rate loan) does not carry a single interest rate fixed for the life of the loan. Instead, its rate resets periodically to track a published market benchmark, so the borrower's interest cost rises and falls as that benchmark moves. Floating-rate structures dominate bridge loans, construction loans, and much of the CLO- and warehouse-financed lending market, because the lender's own cost of funds is typically floating as well -- matching a floating asset (the loan) to floating liabilities (the lender's funding) removes a major source of interest rate mismatch risk for the lender, and shifts that rate risk onto the borrower unless the borrower separately hedges it (the subject of the next topic in this block).

Understanding floating-rate mechanics requires precision about three things: which index the loan references, how the spread over that index is set, and how often -- and on what basis -- the rate actually resets. Get any one of these wrong in a proforma or a payment calculation, and the error compounds every period.

Every floating-rate loan's periodic interest rate is built from two components: the index (also called the reference rate or benchmark) and the spread (also called the margin). The formula is:

All-in Rate = Index Rate + Spread

The spread is negotiated at origination and is generally fixed for the life of the loan (absent a step-up or step-down feature written into the loan documents); it compensates the lender for credit risk, liquidity, and profit margin on that specific borrower and asset. The index is a published, third-party market rate that neither party controls, and it is this component that resets -- typically monthly, though quarterly resets also appear in the market -- on dates specified in the loan agreement, known as reset dates.

Interest on floating-rate CRE loans is almost always computed on an Actual/360 day-count basis: the stated annual rate is applied to the actual number of days in the interest period, divided by a 360-day (not 365-day) year. This convention, inherited from money-market lending practice, means that for an identical nominal annual rate, Actual/360 produces slightly more dollars of interest than an Actual/365 calculation would -- a distinction worked through numerically later in this topic.

For decades, the dominant floating-rate index in CRE lending was LIBOR (the London Interbank Offered Rate), a rate intended to represent what major banks would charge each other for unsecured short-term borrowing. LIBOR was set each day by a panel of banks submitting estimates of their borrowing costs, which were then averaged -- it was not, for most tenors and currencies, calculated from a deep pool of actual observed transactions. That design had two structural weaknesses. First, as unsecured interbank lending declined sharply after the 2008 financial crisis, many submitting banks had few or no real transactions to base their estimates on, so submissions increasingly reflected judgment rather than observable trades. Second, because submissions were self-reported and tied directly to instruments banks and their trading desks held, LIBOR proved vulnerable to manipulation -- a series of enforcement actions beginning around 2012 found that traders at multiple panel banks had coordinated to skew submissions to benefit their own derivatives positions.

Global regulators concluded LIBOR could not be sustained as a benchmark underlying hundreds of trillions of dollars in contracts. In the United States, the Federal Reserve convened the Alternative Reference Rates Committee (ARRC), which selected the Secured Overnight Financing Rate (SOFR) as LIBOR's successor for U.S. dollar contracts. SOFR is calculated by the Federal Reserve Bank of New York from a large volume of actual overnight Treasury repurchase ("repo") transactions -- borrowing and lending collateralized by U.S. Treasury securities -- making it transaction-based, deeply liquid, and effectively impossible to manipulate through self-reporting. Most USD LIBOR tenors stopped being published after mid-2023, and virtually all new CRE floating-rate loans now reference SOFR. Loan documents originated on legacy LIBOR before the transition were generally amended to reference SOFR through fallback provisions, commonly including a fixed credit spread adjustment (CSA) layered on top of the new SOFR-based spread -- intended to approximate the historical gap between LIBOR and SOFR so outstanding contracts would not see an abrupt, unintended change in economics purely from the index swap. The exact cessation dates, fallback mechanics, and any residual legacy-LIBOR exposure are administrative details that can vary by contract vintage; confirm them against the specific loan documents and current ARRC or regulatory guidance rather than assuming a fixed date applies universally.

SOFR is published as an overnight rate, but loans need a rate that applies over an entire interest period (a month, a quarter), so the market converts overnight SOFR into period rates in two different ways.

Term SOFR is a forward-looking rate, published by CME Group for standard tenors (commonly one, three, and six months), derived from SOFR futures market pricing. Like legacy LIBOR, Term SOFR is set in advance, at the start of an interest period, so the borrower and lender both know the exact all-in rate -- and therefore the exact payment -- before the period even begins. This is the index most commonly used in cash CRE loan products because of the payment certainty it provides.

Daily Simple SOFR (and related compounded-in-arrears SOFR conventions) instead averages the actual daily overnight SOFR fixings over the interest period itself, meaning the rate is not fully known until the period is ending or has just ended. To manage the practical difficulty of billing a rate that isn't final until the last day, agreements using daily simple SOFR typically apply a lookback (using the SOFR fixing from a few business days earlier than the actual day) or a lockout (freezing the rate for the last few days of the period at its most recent value) so a payment can be calculated and delivered on time. Daily Simple SOFR is more common in syndicated corporate loans and derivatives markets than in conduit or bridge CRE lending, precisely because CRE borrowers and their lenders generally prefer the payment certainty Term SOFR provides.

Consider a $20,000,000 interest-only floating-rate bridge loan priced at 1-month Term SOFR (reset each quarter for this example) plus a spread of 250 basis points (2.50%), with interest computed Actual/360.

Period 1 (90 days). At the start of the period, Term SOFR fixes at 5.10%. The all-in rate for the period is: All-in Rate1 = 5.10% + 2.50% = 7.60% Interest1 = $20,000,000 x 7.60% x (90 / 360) = $20,000,000 x 0.0760 x 0.25 = $380,000.00

Period 2 (92 days). SOFR resets down to 4.85% (reflecting, say, a Federal Reserve rate cut between periods). The all-in rate becomes: All-in Rate2 = 4.85% + 2.50% = 7.35% Interest2 = $20,000,000 x 7.35% x (92 / 360) = $20,000,000 x 0.0735 x 0.255556 = $375,666.67

Total interest across the two periods (182 days) = $380,000.00 + $375,666.67 = $755,666.67. Note that the spread (2.50%) never changed -- only the index component moved, and every dollar of that 25-basis-point drop between periods flowed directly through to the borrower's payment. This period-by-period recalculation, not a single blended annual rate, is how floating-rate debt service is actually billed and paid.

The day-count convention is not a rounding footnote -- it changes the dollar amount due even when the stated annual rate is identical. Take Period 2 above: a 92-day period at an all-in rate of 7.35% on a $20,000,000 balance.

Under Actual/360 (the market-standard convention for floating-rate CRE loans): Interest = $20,000,000 x 7.35% x (92 / 360) = $375,666.67

Under Actual/365, holding the same nominal rate and the same 92 days: Interest = $20,000,000 x 7.35% x (92 / 365) = $370,520.55

The difference -- $5,146.12 on this single period alone -- arises purely from dividing by 360 instead of 365 days. Because 365 / 360 is approximately 1.0139, Actual/360 effectively costs the borrower about 1.39% more in interest than the same stated nominal rate would under Actual/365, compounding period after period over the loan's term. Borrowers and analysts building a proforma off a lender's quoted 'rate' without confirming the day-count convention in the loan documents will consistently understate the true cost of debt service.