Accessible Version
How U.S. Bank Stock Prices Respond to Geopolitical Risk, Accessible Data
Figure 1. The KBW Bank Index and S&P 500 Around Select Geopolitical Events
This figure is a panel of four charts. Each chart plots two lines – the S&P 500 and the BKX bank index – that are indexed to 100 one day before an event. The x-axis is daily, and there is a vertical line visualizing the event date for each panel, which are (1) the Paris terrorist attacks on November 13, 2015; (2) Iranian attacks on U.S. military bases in Iraq on January 7, 2020; (3) the Russian invasion of Ukraine on February 24, 2022; and (4) the U.S. tariff announcements on April 2, 2025.
In chart (1) the Paris terrorist attacks on November 13, 2015, the two indexes move in tandem, dipping to around 90 on the event date, increasing back to 100 three days after the event, and spiking to 102 after five days, where it stays until seven days post-event.
In chart (2) Iranian Missiles Attacking U.S. Bases in Iraq on January 7, 2020, the indexes move in tandem, but at different levels. Both indexes dip at the event date, increase to above 100 after two days post-event, dip again 3 days post-event, then continue by increasing steadily. The S&P 500 index is consistently higher than the BKX index, with the gap widening after three days post-event.
In chart (3) the Russian invasion of Ukraine on February 24, 2022, the S&P 500 index increases on the event date while the BKX decreases. Both move in tandem following the event date, increasing to above 100 after one day, staying level for a few days, dipping five days post-event, increasing 6 days post-event, and finally ending on a slight decline. The S&P 500 index remains higher in level than the BKX index for the duration of the time period.
In chart (4) Liberation Day, U.S. Tarriff Announcements on April 2, 2025, the two indexes move in tandem for the duration of the time period, very slightly increasing on the event date, then immediately decreasing to between 85 and 90 after two days post-event. The indexes remain at these levels until seven days post-event, when they spike to 95, but still remain below the initial 100.
Notes: This figure plots log changes in the KBW Bank Index (BKX) and the S&P 500 around major geopolitical events. In each panel, the vertical line marks the date of the event. The indexes are normalized to 100 on the day before the event to facilitate comparison of their movements over time. Log changes can be interpreted approximately as percentage changes.
Sources: Bloomberg, Caldara and Iacoviello (2022), and authors’ calculations.
Figure 2. Distribution of Bank-specific GPR Betas
This figure plots a distribution of bank-specific betas as a histogram, with the x-axis plotting betas with a range of -0.35 to -0.05 and the y-axis giving the fraction of banks within each beta range, from 0 to 0.25. From left to right, the chart has seven bins of bank beta ranges that are slightly left-skewed but centered around -0.20. The bins each roughly range 0.04 in size, and the fraction of banks in each bin, from left to right, are 0.05, 0.075, 0.24, 0.21, 0.21, 0.16, and 0.05.
Notes: The figure shows the distribution of the estimated GPR beta coefficients across 38 U.S. banks.
Sources: Caldara and Iacoviello (2022), CRSP, and authors’ calculations.
Figure 3. Bank Characteristics and GPR Stock Price Betas
This figure is a panel of six charts that each plot the relationship between bank-specific betas (y-axis) and a different bank characteristic (x-axis). Each chart is a scatter plot, including points for each bank and a line of best fit. The first four charts have a y-axis range of -0.3 to -0.1. Chart (a) plots a strong, positive relationship between a bank’s returns on average assets (ROAA) and the bank beta, chart (b) plots a strong, positive relationship between a bank’s liquidity ratio and the bank beta, chart (c) plots a loose, negative relationship between a bank’s non-performing loan (NPL) ratio and the bank beta, and chart (d) plots a right-skewed, negative relationship between a bank’s trading asset ratio and the bank beta. The final two charts are bin scatters with lines of best fit, each having a y-axis range of -0.24 to -0.16 and four bins. Chart (e) plots a strong, negative relationship between a bank’s share of foreign claims and the bank beta, and chart (f) plots a looser negative relationship between bank-specific geopolitical risk (BGPR) and the bank beta.
Notes: The various panels in the figure plot the relationship between different bank characteristics and estimated GPR betas, showing bin scatter plots in panels (e) and (f) to preserve the confidentiality of the data. Each point represents the (average) estimated GPR beta and the corresponding bank characteristic. BGPR is computed following Niepmann and Shen (2025) by weighting country-level GPR indexes by each bank's geographic exposure shares and is summed over the sample period of stock returns used to estimate the GPR beta of each bank. The line shows the fitted relationship from a linear regression.
Sources: FR Y-9C regulatory filings, FFIEC 009, Caldara and Iacoviello (2022), CRSP, and authors' calculations.
Figure 4. Explaining Heterogeneity in Stock Price Sensitivity across Banks
This figure plots the differential stock price effects, in basis points, for banks at the 90th and 10th percentile of the distribution of a given bank characteristic. There are seven characteristics: liquid asset ratio, return on average assets (ROAA), Tier 1 capital ratio, non-performing loan (NPL) ratio, trading assets ratio, foreign asset ratio, and bank-specific geopolitical risk (BGPR). The differential effects of each characteristic, in this order, range from roughly 1 to 5, 0.75 to 5.25, -3 to 3, -4 to 4, -4.75 to 0, and -5.25 to -0.25.
Notes: Each bar shows the difference in the stock return response to a one-standard-deviation increase in the GPR index between a bank at the 90th percentile and a bank at the 10th percentile of the indicated characteristic. Units are basis points. Bank characteristics other than BGPR are measured with a 60-trading-day lag. Estimates are based on separate regressions of daily log equity returns on the GPR index interacted with each characteristic for large banks (total assets ≥ $250 million), controlling for day fixed effects. Whiskers denote 95 percent confidence intervals. Standard errors are clustered by bank for balance sheet variables and by day for BGPR. See the appendix for more details.
Sources: FR Y-9C regulatory filings, FFIEC 009 Caldara and Iacoviello (2022), CRSP, and authors' calculations.
Figure A1. Response of the KBW Bank Index to Geopolitical Risk
This figure plots the three estimated coefficients from regressions of the log change in the KBW Bank Index (BKX) on the log changes of the GGPR index unconditionally (column 1), unconditionally during crises times (column 2), and conditioned on the return of the S&P 500 index (column 3). The “Unconditional” estimate (column 1) is -0.217, with 95% confidence interval bars extending from roughly -0.05 to -0.38. The “Unconditional, Crisis Times” estimate is -0.501 with confidence intervals extending from -0.10 to -1.05. Finally, the “With S&P 500 control” point estimate is -0.0121, with a confidence interval bar ranging from 0.10 to -0.10.
Notes: This figure plots the estimated coefficients from a regression of the log change in the KBW Bank Index (BKX) on the log changes of the GGPR index unconditionally (columns 1 and 2) and conditioned on the return of the S&P 500 index (column 3).
Sources: Bloomberg, Caldara and Iacoviello (2022), and authors’ calculations.
Figure A2. Local Projections of Monthly Bank Stock Prices on BGPR
This figure plots the cumulative response, as log points in percents, of U.S. bank stock prices (y-axis) to a shock in bank-specific geopolitical risk (BGPR) over 6 months (x-axis). The y-axis ranges from -1.5 to 1, and the chart includes points estimates and 90% and 95% confidence interval bands. Starting from month zero, the point estimates increase from -0.75 to 0.5 at month two, hover around 0 to 0.25 until month five, and then decrease to -0.5 at month 6. The two confidence bands are roughly similar, with the 90% and 95% intervals remaining very tight together. In general, these bands range from 0.5 below to 0.5 above the point estimate, being widest at month zero and tightest at month four.
Notes: The figure shows the cumulative response over six months of U.S. bank stock prices to a shock to BGPR. The local projections include one lag and are estimated over a six-month horizon. The regression equation includes bank-fixed effects and four bank-level controls: a bank’s Tier1 capital ratio, its non-performing loan ration, its return on average assets, and its liquid-asset ratio. Standard errors are clustered at the bank level. The sample includes 30 banks.
Sources: FR Y-9C regulatory filings, Caldara and Iacoviello (2022), CRSP, and authors' calculations.