Financial Conditions Credit Spread Index (FXI)

Updated 10 July 2026

FXI is a transaction-based benchmark that measures the marginal cost of unsecured wholesale funding for U.S. institutions, serving as a broad macroeconomic gauge distinct from bank-specific indices.
It is constructed from actual wholesale funding transactions filtered by size, tenor, and data quality, with spreads aggregated and smoothed using a 21-business day moving average.
FXI functions as both a fallback benchmark in credit contracts and a diagnostic tool to distinguish system-wide funding stress from bank-specific liquidity issues.

The Financial Conditions Credit Spread Index (FXI) is a transaction-based benchmark that measures the marginal cost of unsecured wholesale debt funding faced by U.S. institutions. It is designed as a broader market companion to the Across-the-Curve Credit Spread Index (AXI) and as a fallback benchmark or reference for AXI in loan and derivative contracts. Because FXI includes all wholesale funding transactions that meet the same size, tenor, and data-quality filters used for AXI, but is not limited to bank issuers, it functions as a broader, macroeconomic gauge of funding conditions rather than a purely bank-specific funding measure (Tsyrennikov, 3 Sep 2025).

1. Definition, scope, and benchmark role

FXI is defined by issuer coverage and use case. Relative to AXI, which is bank-specific, FXI measures unsecured wholesale funding conditions across the broader set of U.S. institutions. The underlying rationale is that a post-LIBOR benchmark ecosystem can distinguish between a bank-funding-sensitive index and a broader financial-conditions companion without abandoning transaction-based benchmark design. In this framework, FXI is both a macro barometer and a contractual fallback (Tsyrennikov, 3 Sep 2025).

The benchmark role is operational as well as conceptual. FXI shares AXI’s computation methodology and is explicitly positioned for use in legal contracts and derivatives as a continuity mechanism if AXI’s transaction base declines or becomes less representative. The same source states that FXI is aligned with IOSCO Principles for Benchmark determination and is operationally compatible with SOFR-based infrastructure, which places it within the benchmark reform architecture rather than treating it as a purely academic index (Tsyrennikov, 3 Sep 2025).

A recurring misconception is that FXI is simply AXI with a different label. The distinction in the data is narrower and more precise: AXI is a transparent, transaction-based measure of wholesale bank funding costs, whereas FXI is the broader market companion that measures the marginal cost of unsecured wholesale debt funding faced by U.S. institutions generally. The difference is therefore one of coverage and interpretation, not of basic construction logic (Tsyrennikov, 3 Sep 2025).

2. Construction mechanics

FXI is constructed from actual wholesale funding transactions, not hypothetical quotes. It includes all eligible transactions—whether the issuer is a bank, non-bank financial, or corporate—that pass the same size, tenor, and data-quality filters used for AXI. The common methodology aggregates spreads across short- and long-term maturities, then smooths the resulting daily series with a 21-business day moving average (Tsyrennikov, 3 Sep 2025).

The aggregation proceeds in stages. For short-term maturities, defined as $0$-$1$ year, the construction uses the dollar-volume-weighted median spread. For long-term maturities, defined as $1$-$5$ years, it takes the volume-weighted average of median spreads for the $1$-$2$ year, $2$-$3$ year, $3$-$4$ year, and $1$0-$1$1 year sub-buckets. Maturity weights are then computed as

$1$2

and

$1$3

The daily unsmoothed spread is then

$1$4

and the published index is the 21-business day moving average

$1$5

This construction is intended to combine transparency, transaction depth, and robustness across the maturity spectrum (Tsyrennikov, 3 Sep 2025).

The use of a moving average is not incidental. In the source description, smoothing is part of the benchmark methodology itself, rather than a downstream analytical choice. This indicates that FXI is designed to be operationally stable enough for benchmark use while still preserving sensitivity to changes in underlying wholesale funding spreads (Tsyrennikov, 3 Sep 2025).

3. Statistical profile and behavior under stress

FXI is supported by a substantially larger transaction pool than AXI. Its average daily supporting volume is reported as $1$61 trillion. The source characterizes this as making FXI statistically more robust and less vulnerable to idiosyncratic or technical market events, which is central to its role as both macro indicator and fallback benchmark (Tsyrennikov, 3 Sep 2025).

Metric	FXI value	Interpretation
Average daily supporting volume	$1,558 billion	Large transaction base
Minimum supporting volume	Always above $1 trillion	Robustness
Mean	0.6905	Average index level
Std. Dev.	0.3200	Dispersion
Coef. Variation	0.4635	Relative stability
LT weight (Mean)	0.805	Long-term dominance
FXI LT Spread	0.8280	Long-term component
FXI ST Spread	0.1203	Short-term component

The reported statistics imply that FXI is more heavily influenced by longer-dated wholesale funding conditions than AXI, with a mean long-term weight of $1$7. The details further note that FXI’s average value is higher than AXI’s and that its coefficient of variation is lower, which is interpreted as greater stability attributable to the broader pool and the heavier long-term weight (Tsyrennikov, 3 Sep 2025).

Co-movement with AXI is strong in both normal and stressed conditions. The overall correlation of daily changes is reported as

$1$8

During the onset of COVID-19, from March 1 to June 30, 2020, the correlation of daily changes rises to 93.6%. During the Silicon Valley Bank episode, from March 1 to June 30, 2023, it remains 80.0%. The interpretation given is that FXI and AXI track each other closely in normal times, become even more tightly linked under economy-wide shocks, and still retain strong co-movement in bank-specific stress (Tsyrennikov, 3 Sep 2025).

A further empirical regularity is that FXI generally exceeds AXI except in bank-specific stress episodes. The significance of divergence is explicitly interpretive: when the two indices diverge, the pattern helps distinguish whether turmoil is concentrated in banks or is spilling over to the broader economy. In that sense, FXI is not only a level measure but also part of a relative-diagnostics pair with AXI (Tsyrennikov, 3 Sep 2025).

4. Financial-conditions interpretation and contractual use

FXI serves as a macroeconomic gauge of funding conditions because it spans all eligible institutions rather than only banks. This broader coverage makes it suitable for monitoring general stress or looseness in U.S. credit markets and for contract settings in which a broad-market credit spread is more appropriate than a bank-funding-specific measure (Tsyrennikov, 3 Sep 2025).

Its benchmark role is especially clear in fallback design. Because FXI uses the same methodology as AXI while drawing on a larger and more diversified transaction base, it is explicitly designed to preserve continuity in loan and derivative contracts. The source emphasizes that this matters when a reference rate must remain robust even if the narrower benchmark becomes less representative. The contractual significance of FXI is therefore tied to benchmark resilience rather than merely to statistical convenience (Tsyrennikov, 3 Sep 2025).

The same research situates FXI within a loan-pricing ecosystem centered on the restoration of credit sensitivity lost in the USD LIBOR transition. The quantified loan-pricing results in that study are reported for SOFR+AXI, not for FXI directly: SOFR+AXI can support spread discounts of up to 65 basis points without lowering risk-adjusted returns, while banks relying on SOFR-only pricing can fail to recover as much as 15 basis points on revolving credit lines over as little as three months. Within that architecture, FXI functions as the broader market companion and fallback, rather than as the primary object of the reported pricing experiment (Tsyrennikov, 3 Sep 2025).

This division of roles is important for interpretation. FXI is not presented as a replacement for every credit-sensitive funding benchmark. It is instead positioned as the broader market measure that complements a bank-specific index, with the two used jointly to separate system-wide financial conditions from bank-centric funding stress (Tsyrennikov, 3 Sep 2025).

5. Research directions for richer FXI design

Several recent studies propose extensions that are directly relevant to future FXI construction. One study builds a 167-indicator comprehensive credit risk indicator set combining 81 macroeconomic indicators, 47 corporate financial indicators, 9 bond-specific indicators, and 30 corporate non-financial indicators, and states that future FXIs should integrate both financial and non-financial corporate indicators. In that study, adding non-financial indicators increases average out-of-sample predictive $1$9 from 0.117 to 0.393 and makes 7 of the top 10 most important predictors non-financial; the same summary also states that dynamic weights and SHAP-based mechanism analysis would enhance real-time monitoring and risk management utilities of the FXI (Wu et al., 23 Sep 2025).

Textual information is another proposed augmentation. Research on earnings-call transcripts develops a text-based credit score that forecasts future CDS spread changes and reports that the information extracted from earnings calls is not spanned by fundamental or market variables. The same work states that a text-implied “fair value” CDS spread or credit score can enhance broad indices by providing a management-guided, forward-looking complement to purely market-based measures, particularly when CDS indices are confounded by illiquidity, sentiment, or supply/demand imbalances (Mamaysky et al., 2022).

A related sentiment-based line of work proposes Multi-Level Sentiment Analysis combining firm-specific micro-level sentiment, industry-specific meso-level sentiment, and duration-aware smoothing. Applied to a Chinese bond market corpus, it produces a daily composite sentiment index and reports statistically significant forecasting improvements when sentiment is added, including a 3.25% MAE reduction and 10.96% MAPE reduction for the $1$0 horizon. The same summary states that the methodology can be applied to construct an FXI-like index by aggregating multi-level sentiment at sector, region, or aggregate market levels (Liu et al., 3 Apr 2025).

High-frequency market-based credit estimates also provide a possible extension. A Bayesian yield-curve approach estimates corporate default spreads from daily bond price data, with the stated implication that high-frequency, firm-level default spreads allow more prompt detection of deteriorating credit conditions and can improve the granularity and timeliness of financial conditions indices such as a credit spread index or FXI. This suggests a path toward combining transaction-based benchmark construction with higher-frequency issuer-level credit information (Papenkov et al., 4 Mar 2025).

FXI is a broad-market spread index, but research on other credit-spread settings shows that broad aggregation can obscure structurally different transmission channels. For emerging markets and developing economies, a dynamic panel model for sovereign spreads uses JP Morgan EMBI Global Sovereign Spread ("EMBIG") as the spread measure and shows that commodity terms of trade and exporter/importer status alter both the direct and indirect impact of global financial conditions on spreads. The methodological lesson drawn in that work is explicit: any credit spread index or FXI centered on spreads in EMDEs should account for commodity terms of trade, trade structure, and the distinction between direct and indirect transmission from global conditions (Carrera et al., 2021).

Segmented markets create a different interpretive problem. A three-currency HJM framework for Brazilian debenture markets shows that standard credit spread indices can be distorted if they aggregate across segments without accounting for segmentation, because they may conflate genuine credit risk with regulatory, clientele, and liquidity frictions. In that setting, the within-issuer triangle residual at the 3-year tenor averages 640 basis points and remains stable through tightening and easing cycles; the same study concludes that practitioners and policymakers should allow for multiple, segment-specific models and indices when persistent segmentation is present (Coelho, 28 May 2026).

Cross-currency credit spread aggregation raises additional issues. A reduced-form model for multi-currency CDS introduces default-driven FX devaluation jumps and states that basis spreads between domestic- and foreign-currency CDS are not merely noise or liquidity effects but can reflect market-perceived default-triggered devaluation risk. The paper’s key approximation,

$1$1

implies that a cross-currency credit spread index that ignores devaluation jumps can understate risk when the market prices material FX-default linkage (Brigo et al., 2015).

Finally, dependence structure matters for aggregation. Vine copula models with tail dependence are reported to provide more accurate cross prediction results than linear regressions for Chinese corporate bond spread data and to improve conditional inferences relevant for sectoral risk transfer, while a copula-based Markov reward approach for EU sovereign spreads models total credit spread, dynamic Theil inequality, and contagion through covariance across countries. A plausible implication is that system-level FXI design benefits from dependence-aware aggregation when sectoral or cross-country contagion is a first-order concern (Pan et al., 2020, D'Amico et al., 2019).