Geopolitical risk is the threat, realization, and escalation of adverse events tied to wars, terrorism, and tensions among states. In "Measuring Geopolitical Risk," Caldara and Iacoviello (2022) built the GPR index from about 25 million newspaper articles. Across 26 countries from 1900 through 2019, a one standard deviation rise in global GPR raised disaster probability by 18 percentage points.
What the Study Found
After controlling for global factors, a one standard deviation rise in country-specific GPR raised disaster probability by 9 percentage points. The same shock lifted the probability of disaster onset from about 2.2% to 9%, an increase of 6.8 percentage points. In a US quarterly VAR from 1986 to 2019, a two standard deviation GPR shock cut fixed investment by 1.5% after about one year. Hours fell 0.6% one year after the shock, and GDP dropped 0.3% over the first year. After a two standard deviation GPR shock, highly exposed firms cut first-year investment about 1 percentage point more than nonexposed firms.
"Higher geopolitical risk foreshadows lower investment and is associated with higher disaster probability and larger downside risks to GDP growth."
Caldara and Iacoviello (2022), Measuring Geopolitical Risk, p. 1223.
Methodology
The recent GPR index searches 10 newspapers from 1985, and the historical index searches three US newspapers from 1900. Each month's index is the number of articles matching eight search categories of geopolitical threats and acts, divided by total articles. The macroeconomic VAR runs from 1986:I through 2019:IV, the country panel from 1900 to 2019, and earnings-call data from 2005:I through 2019:IV. Controls include oil prices, the S&P 500, the VIX, country fixed effects, lagged GDP growth, Tobin's Q, and firm cash flows.
Key Statistics
| Metric | Finding | Context |
|---|---|---|
| Correlation with narrative index | 0.86 | Manual scoring of 44,000 New York Times front pages, 1900–2019 |
| Correlation with human-coded index | 0.93 | Annual frequency; more than 7,000 articles read and coded by hand |
| Correlation, non-US vs US newspapers | 0.88 | Cronbach alpha across the ten newspapers is 0.96 |
| Correlation with worldwide war deaths | 0.82 | Historical GPR index, annual data |
| GDP growth left tail | Decline four times larger than the OLS effect | Tenth quantile of one-year-ahead GDP growth, country-specific GPR |
| Firm-level GPR from earnings calls | Investment falls more than 1% at the trough | Two standard deviation rise; stays below baseline for up to one year |
| Disaster probability model (Equation 1) | D(i,t) = α(i) + β·GPR(t) + γ·GPRC(i,t) + δ·ΔGDP(i,t−1) + controls + u(i,t) | Linear probability model; D is a disaster dummy, GPRC is country-specific GPR |
Geopolitical Acts vs Geopolitical Threats
| Measure | Geopolitical acts (GPA) | Geopolitical threats (GPT) |
|---|---|---|
| Search categories in Table 1 | 3 (categories 6–8: beginning of war, escalation of war, terrorist acts) | 5 (categories 1–5: war, peace, military buildup, nuclear, and terrorist threats) |
| Correlation with worldwide war deaths | 0.83 | 0.46 |
| Effect of its shock on the other index | Sharp and significant increase in threats | Small and short-lived increase in acts |
| Investment and hours response | Similar decline, more persistent | Similar decline |
Why This Matters
The index separates news about wars that might happen from wars that do, which matters for anyone pricing event risk. Threats alone produced contractionary effects with acts held fixed, so tension that never becomes conflict can still weigh on activity. Portfolio managers can treat a rising GPR reading as a tail-risk signal for growth and investment. The authors caution that future geopolitical risks could take different forms than past ones.