Geopolitics doesn’t matter, until it does

I always say that geopolitics doesn’t matter because, in nine out of ten cases, geopolitical events like wars or terrorist attacks only affect short-term sentiment, not market fundamentals. But what about the tenth event?

---

We’ve reset our strategy to grow shareholder value

We are targeting to grow free cash flow* at more than 20% (CAGR 2024-2027) and deliver ROACE of more than 16% in 2027.

*Adjusted operating cash flow

Learn more

---

If you look at the Geopolitical Risk Index (GPR) by Caldara and Iacoviello, which only measures instances of armed conflict (incl. terror attacks), you can immediately see that this isn’t a symmetric distribution. Most of the time, the changes in the index are small, but occasionally, you get a major spike.

Geopolitical Risk Index

Source: Caldara and Iacoviello (2021)

Hence, employing standard econometric models to analyse the link between the GPR and stock markets doesn't make sense. Standard econometric models are linear models that assume a regular (symmetric and without fat tails) distribution. But even if you ignore the big spikes, the GPR's distribution is far from normal.

Distribution of GPR around the median

Source: Panmure Liberum, Caldara and Iacoviello (2021)

This is why I like the approach of Davide Brignone, Luca Gambetti, and Martino Ricci. They built a nonlinear model to assess how spikes in the GPR influence stock markets and other variables. This approach allows them to compare the predictions of a standard linear model with their nonlinear model for extreme spikes of 2, 4, or 8 standard deviations (corresponding to monthly spikes in the GPR of 70, 140 and 280 points).

The charts below show that a standard linear model significantly underestimates a sizeable geopolitical shock's impact on both the S&P 500 and real consumption. Indeed, the nonlinearities stem from the change in consumer and investor behaviour for large geopolitical shocks. More minor shocks lead to a sentiment shock that dissipates quickly. In contrast, more significant shocks fundamentally change consumer behaviour and lead to recession-like behaviour that feeds back to a weaker earnings growth outlook for stocks, etc.

Market reaction to extreme GPR shocks

Source: Brignone et al. (2024). Note: Red dashed lines are forecasts from the linear model, and blue solid lines from the nonlinear model.

As investors, we can learn from this study to check the GPR regularly but only react if it spikes by a lot (more than 70 points or so) because everything below that will likely be inconsequential. And if you want to check the GPR, it is available for free here: https://www.matteoiacoviello.com/gpr.htm