The Journal of Fixed Income

Can AI Generate Value in Global Fixed Income Portfolios?

Luis Ceballos — 2026-06-15

We evaluate the economic value of large language models (LLMs) in global fixed-income portfolio management. Using a sample of more than 500 actively managed global bond funds, we construct AI-driven portfolios by leveraging ChatGPT to generate country-allocation weights that mimic each fund’s mandate, allowing us to evaluate their performance relative to the actual funds. We document that, although AI-driven portfolios are more diversified, they deliver lower returns, higher volatility, and lower portfolio turnover relative to actual funds. On a risk-adjusted basis, the performance spread between AI-driven and actual portfolios is largely negative, with fewer than 5% of funds showing significantly positive values. The underperformance persists across credit quality, duration, and currency segments and remains robust to the exclusion of major risk-off periods. These results suggest that LLMs remain limited in replicating the forward-looking judgment of portfolio managers.

Issuer-Paid versus Subscriber-Paid Credit Measures: Evidence from U.S. Corporate Bonds

Yoon Shin — 2026-06-15

Using 4,230 new corporate bonds issued by U.S. industrial firms from 2001 to 2020, we compare the performance of issuer-paid credit ratings from Standard & Poor’s with two subscriber-paid credit risk measures: Egan-Jones Ratings and Bloomberg Default Grades. We examine their ability to explain offering yields, yield spreads, credit default swap (CDS) spreads, and actual default events.

We find that Bloomberg’s market-based default grades consistently outperform traditional credit ratings in explaining yield and CDS spreads, reflecting their ability to incorporate timely market information. While S&P ratings significantly influence bond pricing, they exhibit lower explanatory power than Bloomberg, and Egan-Jones ratings perform weakest across spread specifications. In contrast, S&P outperforms Egan-Jones in predicting realized default events. For investment-grade bonds, all measures lose predictive power for defaults due to their extreme rarity. Overall, market-based credit risk measures dominate agency ratings in pricing credit risk.

The Comparative Advantages of Factor Diversification versus Holding Diversification in Actively Managed High-Yield Bond Portfolios

Muting Ren — 2026-06-15

This paper examines the comparative impacts of factor diversification and holding diversification on high-yield bond portfolio performance. Traditional portfolio construction methods typically focus on reducing idiosyncratic risk by increasing the number of individual holdings. The rise of multi-factor investing in corporate bonds introduces new dimensions to risk management. Using empirical analysis, this study assesses how allocation across multiple active return sources—carry, value, momentum, quality, low volatility, and size—interacts with conventional holding-diversification strategies. The research also considers these diversification methods beyond a friction-free environment and incorporates practical constraints such as transaction costs and limited portfolio turnover. Results indicate that factor diversification offers significant risk mitigation advantages, equal to or greater than those provided by expanding the number of holdings. Additionally, for high-yield active managers aiming to outperform their benchmark, factor diversification maintains exposure to various active return sources and delivers attractive risk-adjusted returns. This study provides practical guidance for portfolio managers on diversifying high-yield bond portfolios utilizing both the range of holdings and factor exposures.

LONG-TERM BREAKEVEN INFLATION RATES AND FEDERAL RESERVE TREASURY MARKET ACTIONS: 2018-2024

Kyle Allen — 2026-06-15

We examine how Federal Reserve actions in Treasury markets shaped market-based measures of long-term inflation expectations from 2018 through 2024. Using daily breakeven inflation (BEI) rates at 5- and 10-year horizons, we compare QE4 with the periods immediately before and after. During QE4, BEI exhibited heightened volatility and a breakdown in the usual correlation between nominal and real yields. Fed purchases of nominal Treasuries significantly reduced BEI, while TIPS purchases raised it. These effects are robust across data frequencies, indicating that BEI during QE4 reflected Federal Reserve balance sheet composition rather than solely market-driven expectations.

Measuring the Liquidity Risk Premium in Credit Markets

Alberto Pellicioli — 2026-06-15

This article presents a comprehensive analysis of the Liquidity Risk Premium (LRP) in corporate bond markets. We place the Exposure‑Matched Approach (EMAP) at the center of the methodological framework, positioning it as the primary tool for isolating and quantifying liquidity-driven spread differentials. EMAP’s portfolio‑based, exposure‑controlled construction allows the estimation of liquidity premia while holding constant all other systematic risk sources.

We complement this with a regression‑based robustness check, ensuring statistical consistency and reinforcing our main findings. Using both transaction‑cost‑based liquidity measures (LCS) and an alternative structural measure combining age and outstanding amount, we examine US and European IG and HY markets over the 2011–2026 period.

Our results reveal a secular decline in liquidity premia - more pronounced in the US than Europe—interrupted by sharp, crisis‑driven spikes. The high correlation between EMAP and regression-based LRP highlights the robustness of the findings and supports the use of EMAP as a practical and economically grounded measure for practitioners.

Fixed-Income Diversification in Multi-Asset Portfolios: A Bayesian Copula Approach to Tail Risk and Hedging

A. Bouteska — 2026-06-15

In the current landscape of high inflation and shifting monetary policy, the diversification potential of core assets is a primary concern for institutional bond managers. This paper introduces a Bayesian modeling framework utilizing a mixture copula to analyze the joint return dynamics of U.S. Treasury Bond futures, global equities (S&P 500), and real estate (REITs). By combining Clayton (and its rotations), Frank, and Gaussian copulas, our model captures complex, non-linear dependencies and tail risks often missed by traditional correlation metrics. We evaluate these interactions through ten distinct risk measures, including Expected Shortfall and Spectral Risk, providing evidence that fixed-income assets offer a 33% hedging effectiveness for real estate even during market stress. Our findings identify an optimal fixed-income allocation range of 23% to 32% for risk-aware portfolios, offering institutional investors a quantitative toolkit for enhancing "all-weather" portfolio resilience.

"Street-Smart" Optimization: A Pragmatic Approach to Corporate Bond Position Weightings

Mark Vandermyde — 2026-06-15

Despite advances in quantitative optimization techniques, their application to corporate bonds continues to be hindered by intractable liquidity issues, particularly for the high yield sector. As a result, security selection and position weightings often reflect a more qualitative, and therefore idiosyncratic, process which may have implications for portfolio diversification. We develop a conceptual framework to evaluate the risk/return trade-off inherent with such “ad hoc overweights”, as a function of the number of additional positions required to replicate the volatility of an equally-weighted reference portfolio. We then apply various model-derived weightings patterns to historical returns to assess under what conditions the sizing dispersion observed in typical bond portfolios might be expected to enhance performance over a simpler 1/N approach.