Aligning US and European credit market models: A comparative approach to forecasting credit spreads and bond fund returns
In financial forecasting, a good model is one that can generate realistic and robust scenarios, allowing for accurate predictions of future market behaviors. A key aspect of creating such models is the incorporation of so-called stylised facts – widely observed empirical patterns in financial markets that are assumed to be consistent over time. These stylised facts serve as foundational assumptions, helping to ensure that the model accurately reflects typical market behavior and improves its predictive and explanatory power.

In America, a model developed by the National Association of Insurance Commissioners (NAIC) focusing on corporate credit spreads and bond index fund excess returns has been presented (Kehrberg et al. (2022)). It is regulatory-compliant, peer-reviewed, reflects stylised facts and hits the proposed acceptance criteria. The NAIC have proposed eight stylised facts, and in a joint paper, Kidbrooke and Master’s students from Kungliga Tekniska Högskolan (KTH) have investigated these US and European credit market models, as well as implemented and simulated credit spreads and bond index fund excess returns.

However, no such model has publicly been implemented on the European market. To motivate the use of such a model, we must make sure the credit markets behave similarly. This is done by examining each stylised fact quantitatively and qualitatively. While there are minor deviations that need consideration, our paper finds that European credit spreads exhibit similar behaviors to their American counterparts – hence using a similar simulation framework is well-motivated. 

Additionally, the research aims to enhance the alignment between US and European models by exploring potential refinements to the Kidbrooke Economic Scenario Generator. The focus is on further improving its ability to accurately predict bond index fund returns and corporate credit spreads across a variety of market conditions. 

Using parameter estimation techniques and historical data, we calibrate the model and implement it into the European market. Simulating over thirty years, we expectedly find that while High Yield bond funds on average have a higher excess return than its Investment Grade counterpart, they are also outperformed by government bonds in some scenarios – underlining their high risk and high reward nature.

Stylised facts for European credit spreads
Historically, stylised facts in credit markets have not been as extensively studied as they deserve to be. They provide useful information which guides any modeling and design choice. To achieve the most accurate credit model possible, it is essential to research the market of interest thoroughly to see which properties are most important for what one tries to simulate.  

One challenge in modeling European credit spreads is the difference in market dynamics compared to the US. The European market contains a diverse range of countries, each with its economic conditions and market behaviors, which warrants a proper investigation on how its market differs from the American counterpart. This diversity can lead to varying credit spread behaviors, which must be accurately captured in any financial model. Hence, one must tread with caution. 

When examining the behavior of credit spreads in Europe, a logical starting point is to see if the stylised facts are the same as the ones proposed in the United States. For instance, one stylised fact presented in the US is:  

Credit spreads tend to be higher in equity bear markets

To assess this, we first need to define bear markets for the European stock market. While there is no set-in-stone definition, some analysts use a 20% decrease in a major stock index from a yearly high is an indicator of a bear market’s beginning, and conversely a 20% increase from a yearly low the end of that bear market (“Bull Market vs. Bear Market”). Using the major index STOXX Europe 600, we proceed using this definition. 

The reader can examine whether the credit spreads do indeed increase during bear markets and also on the European market both visually and numerically in the following graph and table.

It is apparent that there is a widening in spreads during this type of market stress, suggesting the higher perceived risk in corporate bonds during financially turbulent times (the worse the economy, the more likely corporations are to default on their borrowing).  

To confirm the visual intuition, one can see in the above table that the average spread rises significantly when filtering for bear markets. This quantitative evidence supports the idea that credit spreads can serve as a barometer for market sentiment and perceived credit risk. 

Given the similarities between the US and European financial markets, it is well motivated to use a similar model with the same characteristics but different parameters, adjusted for the European market.  

Therefore, the credit spread model should be a stochastic model that exhibits mean reversion and remains positive. A lognormal Ornstein-Uhlenbeck-like process has these properties and is proposed by the NAIC.

Simulating credit spreads and bond index fund excess returns
By implementing the model and calibrating the parameters numerically using maximum likelihood and least squares estimation to European credit data, we obtain results.

For a reasonability check, one can compare historical levels for Investment Grade spreads with simulated ones and see similar behavior qualitatively, indicating that the spread model produces realistic results.

The excess return profiles for different indices show that the Investment Grade bond index fund for securities with maturity between one and five years have a moderately high return, on average slightly outperforming the risk-free rate for similar maturities while never underperforming.

In comparison, bond funds containing High Yield corporations in the median scenario have an annualised yearly excess return of north of two percent, while a non-negligible number of scenarios (3.5% av the 10 000 simulations run) perform worse than comparable government bonds. 

Conclusion
In the broader context of the financial industry, adopting empirically supported, regulated models are crucial because they ensure the accuracy and reliability of risk assessments.  

In this research, we have focused on bridging the gap between U.S. and European financial modeling, particularly in the context of credit spreads and bond index fund returns. By examining and validating the stylised facts of European credit markets, we demonstrated that, despite minor differences, the behaviors in Europe are largely similar to those in the U.S. This justifies the application of a similar simulation framework, adapted with specific parameters for the European context. Ensuring that the models used in a particular market accurately reflect the stylised facts of that market will help to generate robust future scenarios.

The findings confirm that while high-yield bond funds often provide higher excess returns than investment-grade counterparts, they also come with increased risk, occasionally underperforming against government bonds in certain scenarios.

Ultimately, this research underscores the importance of conducting thorough market analysis before selecting a model for forecasting credit spreads. By doing so, we can provide more accurate and reliable predictions, refining the decision-making process for stakeholders in the European financial markets.

Read the original article here.