Mapping Market Turmoil with ‘Material’ News Attention

Since early April, U.S. reciprocal tariffs and subsequent revisions have sharply raised global business risks. Perhaps more importantly, the perceived risk of disruption and the level of economic uncertainty are likely to have increased for the foreseeable future, as many investors and companies have lost confidence in any predictability of trade policy. With this backdrop, investors are seeking metrics that quantify the level and spread of material risks and uncertainty on a high-frequency and timely basis.  

We describe how to generate weekly macro signals reflecting material company concerns and illustrate the approach with a “Tariff Risk Indicator” (TRI): the level of news-based attention to concrete tariff-related and company-relevant risks. 

When does news matter? 

Any global crisis is sure to generate lots of commentary and speculation. But when does it matter to an investor? In our framework, we can consider a news item as company-relevant once the LLM-powered workflow can establish the news summary indicates a direct risk to the company’s business from tariff-related actions.¹ Specifically, each news item must pass through three distinct gates:  

1. Risk to business: It should indicate specific concerns or difficulties for the company with respect to revenue, operations, supply chain, financial stability or reputation. General news coverage of stock price movements/volatility, earnings volatility, or past and mitigated risks is excluded.  
2. Geopolitical risk: It should indicate that the company is likely to be impacted by the actions of a country through foreign policy, trade relations, international conflicts or diplomatic actions.²
3. Trade restriction and economic nationalism (TREN): A geopolitical-risk news item must have a summary that indicates alignment with at least one of: 
   • Tariffs, trade wars, export bans and import restrictions between countries 
   • Forced divestitures or bans targeting foreign firms for nationalistic reasons 
   • Retaliatory trade measures that disrupt access to international markets 

Thousands of news items are collected and summarized on a weekly basis by MSCI risk systems, but for speed and efficiency, we do not need to process all of these in detail. Instead, we first apply a “semantic filter” on the news universe by screening on the high-level alignment between a story and the metric definition.³ We select the top 25% of news items from this filter. We evaluate the selected news items through the full workflow.  

On a weekly basis, the TRI is calculated as follows: 

TRI = (number of news items assessed to indicate TREN / number of items in news universe)⁴

The TRI effectively indicates the minimum percentage of news items with business risks due to TREN. Companies flagged by the framework workflow can be thought of as “canaries in the coal mine” or “poster children” for this category of disruption. 

Tracking the Tariff Risk Indicator in the US

We calculated the TRI for MSCI USA Index constituents and plotted it with a rolling four-week average using the available data from the last two years. At the end of the third week in April 2025, the TRI was 5.2% (vs. a four-week average of 4.2%) — unsurprisingly its highest level for the preceding 24 months. Moreover, the mid-April high was nearly 20 times higher than the average of approximately 0.3% during that period — and it more than doubled between the end of March and the middle of April.

Since then, the TRI has dropped but remains elevated. The recent modest drop indicates that, while tariff anxiety has eased somewhat, widespread risks persist, and corporate effects are now being more explicitly quantified through earnings calls and other communications.

Every week, on average, the news universe included 1,944 news items from 348 unique issuers in 107 GICS sub-industries. Over the analysis period, the news universe included items from 668 unique issuers, out of 674 MSCI USA Index constituents. On average, 559 news items from 102 issuers and 52 GICS sub-industries were processed after the semantic-filtering step.  

The TRI can also measure the spread of attention to tariff risks. The sharp increase in TRI over the last five observations soon spread to many companies and industries reflecting the scale of market-wide concerns and uncertainty. In the chart below, we can see that, for the week ending on May 2, news was flagged by 24 Global Industry Classification Standard (GICS®)⁵ sub-industries for material and elevated risks from TREN, much higher than prior to January 2025.

Each circle represents a week between March 3, 2023, and April 18, 2025. The yellow line traces the evolution of the TRI since Feb. 7, 2025. 

Lastly, the TRI can be calculated at the sector or industry level. At the end of March of this year, the four-week-average TRI was highest in the autos and components industry group and was zero for eight others. By the end of April, news attention to tariff risk was widespread and had spiked for all industry groups. Consumer-discretionary companies, such as Nike Inc., Lululemon Athletica Inc and Ford Motor Company, were leading contributors to the higher TRI because of their exposure to both U.S. tariffs and China’s retaliatory measures. Semiconductor companies were consistently represented, while aerospace companies, such as The Boeing Company and Howmet Aerospace Inc., were flagged when Chinese companies stopped taking deliveries. The perspective that the TRI offers is complementary to the factor and economic exposure lenses we recently used to analyze post-tariff-announcement performance.

The rolling four-week average of the TRI for the industry groups in the MSCI USA Index.

In our approach, we have leveraged the news-processing power of MSCI AI Portfolio Insights to create top-down market-level and sector-based indicators of stress that allow investors to quantify exposure to material and elevated tariff risks. More generally, we can use LLMs to build a systematic, rules-based model of market-wide geopolitical risks arising from foreign government actions motivated by concerns around security, technology, economy and regulations. Over time, investors can track these indicators as part of their decision toolkit and assessment of the market environment. Other applications of the approach would be to calculate a range of geopolitical and macro indicators that could also be used for portfolio analysis and the creation of hedge portfolios.