AXI – Frequently Asked Questions

These Frequently Asked Questions were prepared by SOFR Academy for use by market participants. This page will evolve as new developments and questions arise. If you have a question to which you are seeking an answer, inquiries can be directed to [email protected] Thank you.

The Across-the-Curve Credit Spread Index (AXI) is a weighted average of the credit spreads of unsecured bank funding transactions with maturities ranging from overnight to five years, with weights that reflect both transactions volumes and issuance volumes. AXI can be added to Term SOFR (or other SOFR variants) to form a credit-sensitive interest rate benchmark for loans, derivatives and other products. Key features of AXI are summarized in our user friendly infographic.

The Financial Conditions Credit Spread Index (FXI)TM is an extension of AXI that incorporates data based on transactions of both financial and non-financial corporate debt instruments and is approximately 500% more robust (Berndt, Duffie, Zhu, 2020).

In addition, ‘all-in’ benchmark rates are being calculated and published where across-the-curve credit spreads are combined with variations of SOFR that are published and administered by the NY Fed. A benchmark comprised of a variation of SOFR plus AXI is referred to as ‘SOFRx’. A benchmark comprised of a variation of SOFR plus FXI is referred to as ‘SOFRy’.

To the extent practicable, AXI will seek alignment with the key principles set out by the Alternative Reference Rates Committee (ARRC).

No. AXI was not developed to emulate LIBOR. Criteria taken into account when developing AXI included:

First – Hedging effectiveness: Highly correlated with U.S. bank cost of funds, as determined by recent market credit spreads for wholesale unsecured issues of U.S. banks and bank holding companies.

Second – Robustness: Computed from a large enough pool of market transactions that the index can underlie actively traded derivatives instruments used by banks and their borrowing customers to hedge their floating-rate exposures, without significant risk of statistical corruption or manipulation.

Third – Adaptable to changes in issuance patterns: The index should, within reason, maintain the first two properties even as banks change the maturity and instrument composition of their issuances in response to changes in regulation and market conditions.

A future change in bank regulation or evolving market trends may result in banks adjusting their debt maturity structure. AXI was designed to automatically adapt to these changes. 

When bank funding turns out to be short term, AXI gives greater weight to short-term spreads. When bank funding turns out to be longer term, AXI weighs to the long term.

Since there are almost no transactions with which to fix LIBOR, a representative credit spread would have to stay away from LIBOR and instead focus on the actual funding banks are getting. 

AXI is a dynamic credit spread that moves to where the actual bank funding transactions occur. For example, if banks began funding themselves at the 3 year point of the yield curve, AXI would weight transactions at that tenor.

AXI is fundamentally different to the other CSR proposals in a few important ways: (1) it takes into account funding transactions that occur out to 5-years so it’s not limited to the short-term markets that once underpinned LIBOR, (2) it is a credit spread add-on to SOFR, not a standalone rate and therefore promotes the adoption of SOFR (3) AXI automatically adapts to changes in bank funding composition, so its representativeness and robustness are sustained over time.

SOFR Academy is grateful to the team of leading academics that authored the paper in which the Across-the-Curve Credit Spread Index (AXI)TM and the Financial Conditions Credit Spread Index (FXI)TM were conceived. Antje Berndt is a Professor of Finance at the College of Business and Economics, Australian National University. Darrell Duffie is the Adams Distinguished Professor of Management and a Professor of Finance at the Graduate School of Business, Stanford University, as well as a research fellow of the National Bureau of Economic Research. Yichao Zhu is a Senior Lecturer in finance at the College of Business and Economics, Australian National University. Professor Duffie chaired a Market Participants Group on Reforming Interest Rate Benchmarks (or the Market Participants Group on Reference Rate Reform). The group was established by the Financial Stability Board (Market Participants Group (2014)).

AXI was discussed at the Credit Sensitivity Group Workshops hosted by the Federal Reserve Bank of New York. Four workshops were held between June and August 2020 and aimed to build a shared understanding of the challenges that banks of all sizes and their borrowers may face in transitioning loan products away from LIBOR. They also explored methodologies to develop a robust lending framework that considers a credit sensitive rate/spread that could be added to SOFR.

No. Whenever a loan is indexed to a AXI, it would also reference SOFR. So, use of a credit spread index does not imply less frequent reference to SOFR. The construction of floating-rate loan terms implies that the credit spread index would be used in different ways for different applications, including loans with different coupon periodicity and derivatives of different types. 

With the emergence of an active loan market based on a credit spread index, there would be many potentially important useful applications for hedging and speculation based on that index, some of which we have outlined here. It would tempt fate to choose a credit spread index that is not based on a large pool of transactions. Regulators and market participants can avoid painting themselves into that corner.

AXI interest rate swaps are priced the same way that a LIBOR swap (and most other swaps) are priced — in effect, supply and demand cross at some fixed-side rate. Like any new market, such as SOFR, at the beginning, there is relatively little closely related rate information. Upon AXI launch, dealers will offer quotes (bids and offers). At first, they will rely on the historical relationship between AXI, Libor, OIS and quote with wide bid-offer spreads. Later, as trades are executed and price discovery occurs, bid offers will narrow as trade activity picks up.

Reliance on an index that depends only on a much smaller available daily sample, however, would imply a much noisier index, having greater susceptibility to manipulation. Moreover, the intent is an index that, when added to a risk-free rate, is correlated with a bank’s recent cost of funds.

A bank’s recent cost of funds is determined by the yields of its recently issued stack of liabilities, most of which are issued on a range of past dates. This explains the once-prevalent use of long-lag COFI reference rates, which move very slowly with changes in cost of funds. A credit spread index of the sort that we have illustrated is a compromise between using a very stale index, like 11th District COFI, and a current-market credit-sensitive rate such as LIBOR (which is no longer feasible). Finally, it bears mentioning that linking a loan interest payment to a lagging market spread index implies no sort of “arbitrage.” The impact of any recent changes in market spreads can be embedded into the fixed spread (including the borrower-specific component) that is added to the variable rate components when setting the loan terms. Lenders would undoubtedly have in mind recent changes in market conditions when offering loan terms. Borrowers that are negotiating loan terms would benefit from staying up to date on changes in market credit spreads.

One wants an index that is based on a large pool of transactions. Studies can determine whether the funding costs of smaller banks are sufficiently correlated with the funding costs of larger banks that smaller banks would want to use this sort of credit spread index. Smaller and mid-size banks have been content to reference LIBOR, which is based exclusively on the funding costs of the largest banks, including non-U.S. global banks. Ameribor may be a useful alternative for some small banks.

When contracting 3-month floating-rate interest payments, it makes sense for AXI to be scaled down. For example, a bank loan linked to AXI would have a floating interest payment R(t) on date t of the contractual form R(t) = SOFR(t) + B*AXI(t) + borrower fixed spread, where B is a constant scaling factor specific to the 3-month length of the coupon period and SOFR(t) is SOFR for the 3-month coupon period ending on date t, obtained from daily SOFR compounded in arrears over the coupon period.

With this construction of loan terms, floating-rate risk can be managed with combinations of derivatives linked to SOFR and derivatives linked to AXI. For example, a loan of principal P paying a fixed spread over SOFR + B*AXI can be swapped to a fixed rate by entering a SOFR payer swap with notional P and an AXI payer swap with notional B*P.

Yes. AXI is scaled down to create a term structure in standardized tenors including Overnight, 1 Month, 3 Months, 6 Months and 12 Months.

No. SOFR Academy is supportive of the work of the Alternative Reference Rates Committee (ARRC). To publish AXI as a standalone benchmark index would create a path for banks to skip over SOFR which is counter to the recommendations of the Financial Stability Board (FSB). The FSB is an international body that monitors and makes recommendations about the global financial system and was established by the G20 in 2009. 

AXI is a spread which is added to SOFR. For example, it could be used on top of CME Term SOFR, simple daily SOFR, or other SOFR variants to form a credit-sensitive interest rate benchmark for loans. Importantly, input data for AXI is not confined to the thin, short-term bank funding markets that once underpinned LIBOR. AXI plus SOFR is referred to as SOFRx.

No. AXI is a robustly determined and representative dynamic credit spread that is added to SOFR. The ARRC identified SOFR as the rate that represents best practice for use in certain new USD derivatives and other financial contracts. 

During times of market stress, such as March 2020, LIBOR and LIBOR-like rates tend to rise rapidly, at the same time, the number of interbank bank funding transactions drops in the short-end of the yield curve, to zero in some cases (Schooling Latter, 2021).

AXI will move higher during market stress events but only in correlation with changes in bank funding costs, it will not move as high or with the same level of volatility as LIBOR during these period. 

The Financial Conditions Credit Spread Index (FXI) is an extension of AXI that incorporates data based on transactions of both financial and non-financial corporate debt instruments. An advantage of including non-financial issues is the much larger volume of transactions that can be exploited for robustness. It has been estimated that this scales up the dollar volume of covered transactions over the period 2014–2019 by factor of nearly 500%. AXI and FXI are highly correlated, especially over the past few years.

FXI vs AXI chart | SOFR Academy


Image Source: NY Federal Reserve – Forum on Ongoing Innovation in Reference Rates for Commercial Lending

Please email questions to [email protected] and we will arrange to have them answered. Selected questions and answers may be added to this publicly available list of frequently asked questions.


The answers to certain questions were sourced from the following reference:

Berndt, A., D. Duffie, and Y. Zhu (2020), Across-the-Curve Credit Spread. Stanford University Graduate School of Business Research Paper No. 3884, Available at SRN:

Please contact [email protected] to obtain AXI licensing information.

No. AXI is currently not a benchmark available for use in the United Kingdom.

How can we help you?

Get in touch with us and register your interest.