Curated Measures

Key Business Risks

Data through Sep 30, 2026 Updated Oct 13, 2026 5 series

Compare topic-specific corporate risk discussion on a common sentence-count scale, implementing the decomposition in Hassan et al. (2025).

Based on peer-reviewed research Journal of Economic Perspectives, 2025

We’ll email you when the data is updated. No account required.

At a glance

Coverage
About 13,700 companies in 92 countries
Period
Jan 2016 – Sep 30, 2026
Updates
Quarterly
Files

CSV: time series, firm panel

Source

English earnings calls

Overview

Cardinal Risk Comparison implements the sentence-level topic risk measures from Figure 3 of Hassan, Hollander, Kalyani, van Lent, Schwedeler, and Tahoun (2025, Journal of Economic Perspectives). Because every topic series is constructed with the same sentence-level Risk rule—and can be divided by unconditional Overall Risk from the same calls—you can compare both the trajectory and the relative magnitude of different business risks over time.

Use it to

  • Compare the relative magnitude of different business risks (such as Inflation, Supply Chain, Trade, COVID-19, and Russia) on a common cardinal scale.
  • Compute each topic's share of all risk-related discussion by dividing call-level Topic Risk by unconditional Overall Risk (unconditional_risk_sentences_count).
  • Reproduce and extend the cardinal risk decomposition in Figure 3 of Hassan et al. (2025) to current quarters.
  • Rank which risks dominate executive and analyst discussion across sectors, countries, and macroeconomic shocks.

What it does not measure

  • Downloadable firm-panel CSVs contain raw call-level sentence counts (risk), not pre-divided shares of Overall Risk; compute the JEP decomposition ratio downstream using the Overall risk and sentiment export (unconditional.csv).
  • Topic shares do not sum to 100%: the topics do not exhaust all corporate risks, and a single sentence can match more than one topic query.
  • Measures the frequency of risk-conditioned discussion, not the probability of an event or expected financial losses.
  • Current values reflect the live earnings-call corpus and Updated search engine; exact replication of the published JEP figure requires the paper's replication package and sample restrictions.

Explore the data

What to look for

In the featured Supply Chain Risk series, discussion remains quiet between 0.01 and 0.02 from Q1 2016 through Q4 2019, triples to 0.06 by Q4 2020 during the initial COVID-19 lockdowns, and surges to an all-time peak of 0.14 in Q3 2022—more than ten times its pre-pandemic level—amid global shipping, semiconductor, and energy bottlenecks. It then unwinds to 0.03 by Q4 2024 before ticking up to 0.05 in H2 2025–Q3 2026, remaining roughly triple its pre-2020 baseline. Open the other topic series below to compare when COVID-19, Inflation, Russia, and Trade risk peaked on the same scale.

Click the chart to open the series, where you can filter it by headquarters country.

Risk decomposition: Supply Chain Risk over time

Date range: January 1, 2016 to September 30, 2026.

Time-series chart for Risk decomposition: Supply Chain Risk over time.
Loading figure...

Mean risk-related matched sentences per earnings call, measured in Matched sentences, shown quarterly using a 4-quarter moving average. Q1 2016 to Q3 2026, 43 data points. Higher values mean more risk discussion.

Explore series

Series

5 total Data through Sep 30, 2026

Browse the related series and open any series for its full query, filters, and methodology notes.

No series match your filter.

Methodology and data

1. Define the topic query

Each series uses the topic query documented in Hassan et al. (2025) (visible in full on each underlying series page). For sentence $s$ and topic $k$, Topic match(s, k) = 1 when sentence $s$ contains at least one match for topic query $k$, and 0 otherwise.

2. Identify risk-related sentences

The Risk dictionary is the deduplicated union of Oxford Thesaurus synonyms for risk, risky, uncertain, and uncertainty, excluding question, questions, and venture. For sentence $s$, Risk match(s) = 1 when sentence $s$ contains at least one Risk-dictionary term, and 0 otherwise.

3. Calculate call-level Topic Risk and Overall Risk

For earnings call $c$ and topic $k$, each sentence is counted at most once:

  • Topic Risk (risk): Σ Topic match(s, k) × Risk match(s) across all sentences $s$ in call $c$.
  • Overall Risk (unconditional_risk_sentences_count): Σ Risk match(s) across all sentences $s$ in call $c$, regardless of topic (delivered in the Overall risk and sentiment CSV).

4. Compute the JEP cardinal risk decomposition

For a selected firm, sector, or period, aggregate the numerator and denominator over the same calls and calculate:

Topic share of risk (%) = 100 × Topic Risk ÷ Overall Risk

For example, if 20 sentences match Supply Chain Risk and 200 sentences contain Risk-dictionary language overall in that sample, Supply Chain accounts for 10% of overall risk discussion. For exact replication of Figure 3 as published, use the AEA replication package.

Changes to this dataset

Updated 13 October 2026. We recalculated this dataset's series with our new search engine. They now run through 30 September 2026. Values for earlier quarters can differ slightly from the previous release, which is available on request.

References

Hassan, T. A. & Hollander, S. & Kalyani, A. & van Lent, L. & Schwedeler, M. & Tahoun, A., "Text as Data in Economic Analysis," Journal of Economic Perspectives, 39(3):193-220 (2025)

NL Analytics. (2026). Key Business Risks [Data set]. NL Analytics. https://apps.nlanalytics.tech/curated-measures/cardinal-risk-comparison/

Browse Curated Measures