Credit factor models in a new macro world

Can factor models capture recent geopolitical developments?

March 2025 marks a turning point in the current geopolitical environment. Europe and the US are changing course, with far-reaching consequences for sectors and individual companies around the world.

In Germany, decades of conservative fiscal policy have come to an end. The federal government has loosened the debt brake enshrined in the constitution and adopted a defence and infrastructure package worth hundreds of billions of euros – financed by new debt. This has had an impact on European interest rates and, in the medium term, on inflation rates. The longer-term political consequences, such as the unenforceability of the Maastricht criteria at the European level following this precedent, are not yet foreseeable.

In the US, President Trump is continuing his unconventional policies and is causing an adjustment recession by dismantling the bloated state apparatus, dissolving government agencies and dismissing their employees. While these measures will strengthen growth in the longer term and attempt to bring out-of-control government finances under control, they represent a short-term economic headwind. The threat of tariffs to achieve better deals for the US economy is causing additional uncertainty.

From the perspective of a systematic credit investor, the question arises: Are factor models able to absorb these structural breaks? Any model adjustment always takes a few days and weeks, but can these factors capture the new economic environment at all? Are there variables and mechanisms in such models that ensure that information about the German fiscal measures or the current US economic policy and its impact on companies and sectors is incorporated into the forecasts?

In particular, two systematic factors in factor models should – if properly formulated – capture this information.

1. Stock momentum:
   Stock momentum is an important factor in most systematic credit models. The model analyses which stocks are performing particularly well (poorly) and uses this information as a signal to buy (sell) the bonds of the same company. This is based on the assumption that a company’s stocks absorb information about the company faster and more completely, and thus provide an indication of the future relative performance of the company’s bonds. There is academic evidence to support this assumption.
   
   The stock market’s reaction to political developments is thus directly incorporated into the model. If market developments cause defence stocks to outperform the rest of the market, this immediately leads to a positive momentum signal and the corporate bond forecast improves. Similarly, the expectation of rising interest rates leads to an underperformance of stocks in interest-sensitive sectors. This causes a deterioration in the corporate bond forecast from these sectors.
   
   
2. Value:
   Value is another factor found in almost every systematic credit model. Value measures whether a bond is overvalued or undervalued relative to its fair value. This is done using a fair value model that incorporates various variables that measure the risk of the company. The market spread of the bonds is then compared with the fair theoretical spread from the model. If the variables in the fair spread model react quickly to changes in the market environment, the company’s value forecast will also change quickly.
   
   Such variables may include the following aspects (other variables are possible):
   • Stock volatility: Academic research has shown a strong correlation between the bond spread and the volatility of a company’s stock. Therefore, an increase in stock volatility leads to an increase in the fair spread required by the model. If the company’s spread does not follow suit to the same extent, the attractiveness of the bond will decrease.
   • Analyst estimates: Analyst’s estimates of the attractiveness of companies and their earnings performance are a possible variable in a value model. Analysts generally react very quickly to developments that have a lasting impact on a company’s business performance, often within hours or a few days. The change in these estimates influences the fair spread of a company and thus its attractiveness.
   • Stock valuation: When the medium-term outlook for a company changes, this is often reflected in the company’s valuation. A change in the price/earnings ratio often reflects market expectations of significant changes in the company’s earnings momentum. If this variable is incorporated into a value model, the change will also affect the valuation of the company’s bonds in the medium term.

Two recent examples illustrate the impact of systematic factors on credit forecasts.

The French defence company Thales is one of the big winners from the European defence programme. The company’s stock price rose from EUR 138.65 on 31 December 2024 to EUR 252.40 on 18 March 2025, an increase of 82 % within two and a half months, while the bond’s credit spread tightened from 120 to 77 basis points over the same period. This spread tightening accelerated significantly in mid-March.

The chart above shows how the company’s forecast has improved, and which components are largely responsible for this. After being around 50 % at the turn of the year and thus in neutral territory, the forecast shot up to over 90 % in early March, suggesting a very attractive bond. While the value factor and the spread in particular improved versus the peer group at the beginning of January, the rising equity momentum in February led to a further improvement in the forecast.

At the current margin, the forecast is slightly lower due to the recent correction in the stock price (momentum) and the simultaneous decline in the spread (value). Nevertheless, the company’s bond remains attractive.

The real estate sector in Germany is seen by the market as one of the losers in the new market environment. Real estate companies such as Vonovia are more sensitive to interest rates than many other sectors, and the massive rise in interest rates could be a headwind for the company’s future business development. The company’s stock price fell by almost 14 % from 4 March to 19 March 2025, while credit spreads widened from 151 to 165 basis points from 5 March.

While the Vonovia bond was still neutrally valued in early March, with an overall signal of just over 50 %, the value and momentum factors deteriorated significantly as the market reacted to the political change in Germany. The underperformance of the stock led to a significant deterioration in the momentum factor. The value factor also requires a higher spread premium than the market has priced in with a spread widening of 14 basis points. Both factors therefore lead to a deterioration of the overall forecast into negative territory, around the 20% percentile.

Conclusion

Both examples show how a systematic factor model can capture market information and incorporate it promptly into a factor forecast. Equity momentum transfers the movement of the equity market into the bond forecast, while a fair value model can process many other market variables and thus influence the bond forecast. Well-formulated factor models, in which the systematic factors react promptly to market variables, are therefore ideally suited to adjusting forecasts to a changing economic environment in a short period of time.