Assessing Credit Quality and Duration Across S&P U.S. Corporate Bond Indices

While equity markets have rebounded, the International Monetary Fund, at its recent Spring Meetings, revised U.S. growth projections slightly upward while warning of fiscal imbalances. This fragile context may widen spreads, particularly for lower-tier credits. From a credit perspective, the S&P U.S. Investment Grade Corporate Bond Index’s option-adjusted spread (OAS) peaked at 366 bps on March 23, 2020, as noted below. Its OAS compressed to 74 bps by early 2026, down from 150 bps early 2019. The reduction reflects the Federal Reserve’s easing and the continued low-rate environment in 2021. Conversely, the S&P U.S. High Yield Corporate Bond Index’s OAS had significant volatility climbing to 1,020 bps on March 23, 2020. On the same chart, its OAS moved from 385 bps in 2019 to over 700 bps mid-2020, before pulling back below 300 bps early 2026. 

Both the iBoxx $ Liquid Investment Grade Index and the iBoxx $ Liquid High Yield Index have exhibited a comparable trend, with their respective OAS contracting YTD through June 1. 

Historical data on the S&P U.S. Investment Grade Corporate Bond Index shows three regimes based on interest rate sensitivity and credit risk premiums: Cluster 0 marks the pre- and early Covid-19 pandemic with the Federal Reserve’s 2019 - 2021 easing period driving refinancings. OAS pulled back to a normalized range after the pandemic spike as average duration to worst (DTW) nudged higher. Cluster 2 covers the tightening path in 2022 - 2023 when investment grade issuers turned cautious. DTW retraced as borrowing costs increased and supply stabilized. Cluster 1 highlights ongoing spread tightening and drifting DTW. S&P Global Market Intelligence’s Capital IQ Pro platform offers AI-driven features to explore corporate bonds across sectors.

The S&P U.S. High Yield Corporate Bond Index shows three distinct groups depicting declining duration and spreads: Cluster 0 with 2019 - 2021 low-rate setup, issuers locked in 4-year range DTW with average OAS 300–450 bps excluding a transitory pandemic increase. Cluster 2 shows 2022 - 2023 rate hikes where OAS widened, high yield bond prices mostly fell, and supply declined to avoid high coupons. Cluster 1 shows DTW drifting lower to a record low of 3.2 years by early April 2026--1.4 years lower than the mid-2022 peak.

Continuing with the credit quality stack, yield to worst (YTW) across all indices rose following the Fed’s hawkish guidance leading to 2022 - 2023 rate hikes and showed diverging trends. After bottoming in January 2021, YTW for the S&P U.S. Investment Grade Corporate Bond BBB Index remains elevated in the low 5% range, unwinding last year’s compression. Leverage here is under close scrutiny, as AI-led capex remains the primary driver of supply. Turning to S&P U.S. High Yield Corporate Bond CCC Index’s YTW, it reverted from the Liberation Day spike in April 2025. While it sits below its 11.3% long-term average, refinancing risk remains high. Lastly, the S&P U.S. High Yield Corporate Distressed Bond Index’s YTW has been rising since September, with market attention increasingly focused on upcoming refinancing requirements.

To further analyze these trends, the Kensho LLM-Ready API provides an expanding array of datasets, including Machine Readable Transcripts that offer corporate call insights for both IG and HY debt issuers. This data is delivered in a format optimized for Natural Language Processing (NLP) applications with metadata tagging. S&P Global Market Intelligence data can be ingested directly into large language models (LLMs) such as ChatGPT by OpenAI and Claude by Anthropic. MCP server access makes it easy to connect the Kensho LLM-Ready API via Claude’s and ChatGPT’s web, desktop, and mobile apps.

The chart above shows frequency scores for textual data using ProntoNLP, which is S&P Global's fine-tuned LLM layered on top of the Machine Readable Transcripts dataset. Based on a sample of mid and small cap equity size companies, their corporate calls and other materials dated January through May 2026 were evaluated. ProntoNLP uses neural sparse search to identify sentences across such transcripts that are semantically relevant to a query - in this case, "high yield bond issuance, debt refinancing, leverage." The natural language processing tool goes beyond simple keyword-match and understands contextual meaning to enhance analytics and also produce a word cloud. ProntoNLP does not restrict the sample only to confirmed high yield issuers -rather, it surfaces companies whose transcript language is most semantically aligned, in this example, with high yield bonds.