Generative AI, Managerial Expectations, and Economic Activity
Listen
Radio episode about this paper
Transcript
Introduction to the show: ident: Quantitative Finance. The day in quantitative finance: models of markets, risk and prices.
Ruth: I'm Ruth.
Warren: And I'm Warren, here with Iris.
Ruth: We're looking at the paper: "Generative AI, Managerial Expectations, and Economic Activity".
Warren: Here's the gist of the paper. The research utilizes generative artificial intelligence to extract managerial expectations regarding their economic outlook from over 120,000 corporate conference call transcripts.
Ruth: This process yields an "AI Economy Score" which is designed to predict various economic indicators—including GDP growth, production, and employment—up to ten quarters into the future, surpassing existing forecasting methods like traditional survey forecasts.
Warren: Furthermore, the study develops a composite measure that integrates managerial expectations concerning firm-level activities, industry conditions, and macroeconomic environments.
Ruth: This integrated measure significantly enhances the predictive power and extends the forecasting horizon for both national and sectoral growth projections.
Paper summary: Ruth: To recap, this paper "Generative AI, Managerial Expectations, and Economic Activity" uses generative artificial intelligence to extract managerial expectations about the economic outlook from over one hundred twenty thousand corporate conference call transcripts <ref:2410.03897#pg1,Generative AI, Managerial Expectations, and Economic Activity>. The thesis is that this process yields an "AI Economy Score" which predicts GDP growth, production, and employment up to ten quarters ahead beyond existing measures like survey forecasts.
Warren: What they claim is that this AI score, plus a composite measure integrating expectations about firm activities, industry conditions, and the macroeconomic environment, significantly improves the forecasting power and extends the predictive horizon for both national and sectoral growth projections.
Iris: So it’s leveraging large language models to get these expectations from transcripts to build a predictive tool that goes further out than traditional methods.
Ruth: And they found that managerial expectations offer unique insights into economic activity, with implications for both macroeconomic and microeconomic decision-making, which is the big picture here.
Warren: The results show the AI Economy Score is strongly predictive of next quarter’s real GDP, increasing the share of variation explained by four percentage points in a model that already incorporates things like the Term Spread, Real FFR, GZ Spread, and lag of GDP.
Iris: That predictive power is also moderately persistent for four quarters and it's associated with other future aggregate economic variables like industrial production, employment levels, and wages for up to ten quarters.
Ruth: A key finding they highlight is that shocks to the AI Economy Score are orthogonal to current economic activities; these shocks significantly influence future activities in a distinct way.
Warren: A positive shock predicts higher consumption, investment, and output growth for eight or more quarters, and it’s also positively associated with increased inflation, excess market returns, and long-term government bond yields.
Iris: That induced economic activity triggers tighter monetary policy responses because of these shifts in expectations.
Ruth: And they show that even after controlling for established GDP predictors, the AI Economy Score masked still remains a significant predictor of real GDP growth in the subsequent quarter.
Conclusion: Warren: So looking at this paper "Generative AI, Managerial Expectations, and Economic Activity" from authors Jha, Qian, Weber, and Yang November two thousand twenty-five <ref:2410.03897#pg1,Generative AI, Managerial Expectations, and Economic Activity>. The main implication is that we have a way to use public data through generative AI to get forward-looking information for researchers and investors at granular levels.
Ruth: It’s about using what managers believe right now—their expectations—to predict how the economy will behave far into the future, both nationally and at the firm level.
Iris: The paper suggests that this isn't just another data point; it's a tool that lets us see how individual firm beliefs translate into actual economic impacts, which is important for understanding both macro and micro decision-making.
Warren: Ultimately, the finding is that managerial expectations are powerful because managers can act upon their beliefs, leading to tangible economic impacts on the economy.
Manish Jha, Jialin Qian, Michael Weber, Baozhong Yang
Georgia State University · Rochester Institute of Technology · Purdue University Daniels School of Business and NBER
q-fin.CP, cs.LG, econ.GN, q-fin.EC
Submitted: 2024-10-04
Updated: 2026-11-09
Comments: 27 Pages, 5 Figures, 17 Tables
License: http://creativecommons.org/licenses/by-nc-nd/4.0/
Importance score: 92/100
The gist: The research utilizes generative artificial intelligence to extract managerial expectations regarding their economic outlook from over 120,000 corporate conference call transcripts.
Key concepts
- AI Economy Score
- A composite measure derived from generative AI analyzing over 120,000 corporate calls to capture managerial expectations about the economic outlook. It is designed to predict various indicators like GDP growth and production up to ten quarters into the future.
- Orthogonal Relationship
- The study found that shocks to the AI Economy Score have an orthogonal relationship with current economic activities, meaning changes in the expectation score significantly influence future activities. A positive shock forecasts higher consumption and investment for eight or more quarters.
- Managerial Expectations
- These are the beliefs held by managers about their economic outlook, which can be acted upon to create tangible economic impacts. The study found that pessimistic managers' outlooks are particularly informative, especially during economic crises.
- AI Weighted Score
- A novel score constructed using weights derived from a regression model linking scores to future firm sales. It improves predictions of national and industry GDP growth by incorporating factors like revenue, production, and capital expenditures.
Terminology
Summary
The research utilizes generative artificial intelligence to extract managerial expectations regarding their economic outlook from over 120,000 corporate conference call transcripts. p. 0 Generative AI, Managerial Expectations, and
Economic Activity*
Manish Jha†
Georgia State
Jialin Qian‡
Rochester Inst. Michael Weber§
Purdue and NBER
Baozhong Yang¶
Georgia State
November 2025
Abstract
We use generative AI to extract managerial expectations about their economic outlook from
120,000+ corporate conference call transcripts. Moreover, industry and firm-level measures provide valuable information
about sector-specific and individual firm activities. A composite measure that integrates
managerial expectations about firm, industry, and macroeconomic conditions further significantly improves the forecasting power and predictive horizon of national and sectoral
growth. Moreover, industry and firm-level measures provide valuable information
about sector-specific and individual firm activities. A composite measure that integrates
managerial expectations about firm, industry, and macroeconomic conditions further significantly improves the forecasting power and predictive horizon of national and sectoral
growth. A composite measure that integrates
managerial expectations about firm, industry, and macroeconomic conditions further significantly improves the forecasting power and predictive horizon of national and sectoral
growth. Our findings show managerial expectations offer unique insights into economic
activity, with implications for both macroeconomic and microeconomic decision-making.Where the paper says it
AI Economy Score
which is designed to predict various economic indicators—including GDP growth, production, and employment—up to ten quarters into the future, surpassing existing forecasting methods like traditional survey forecasts. Furthermore, the study develops a composite measure that integrates managerial expectations concerning firm-level activities, industry conditions, and macroeconomic environments. p. 0Where the paper says it
Where the paper says it
Where the paper says it
The AI Economy Score demonstrates strong predictive capability for next quarter’s real GDP, increasing the share of variation explained by four percentage points in a model that already incorporates common predictors such as the Term Spread, Real FFR, GZ Spread (Gilchrist and Zakrajšek, 2012), and lagged GDP. p. 2 We first focus on the main measure, the AI Economy Score, which captures the average
managerial expectation for the US economy in the next quarter. The AI Economy Score is
strongly predictive of next quarter’s real GDP, increasing the share of the variation explained by 4
percentage points in a model with common predictors for aggregate growth including the Term
Spread, Real FFR, GZ Spread (Gilchrist and Zakrajšek, 2012), and lag of GDP. The AI Economy Score is
strongly predictive of next quarter’s real GDP, increasing the share of the variation explained by 4
percentage points in a model with common predictors for aggregate growth including the Term
Spread, Real FFR, GZ Spread (Gilchrist and Zakrajšek, 2012), and lag of GDP. This predictive power
is also moderately persistent, lasting for 4 quarters. This predictive power
is also moderately persistent, lasting for 4 quarters. The AI Economy Score is also significantly
associated with other future aggregate economic variables, including industrial production,
employment, and wages, for up to 10 quarters in the future.Where the paper says it
Where the paper says it
Where the paper says it
A critical finding involves the orthogonal relationship between shocks to the AI Economy Score and current economic activities; these shocks significantly influence future activities. p. 2 To filter out the influence of other current economic factors, we employ a vector autoregression (VAR) framework to study the impulse responses to changes in the managerial expectation
score. We find that shocks to the AI Economy Score orthogonal to current economic activities
significantly influence future activities. We find that shocks to the AI Economy Score orthogonal to current economic activities
significantly influence future activities. A positive shock to the expectation score predicts higher
consumption, investment, and output growth for 8 or more quarters. It is also positively associated with higher inflation, excess market returns, and long-term government bond yields. The
increased economic activity induced by the shock also trigger tighter monetary policy. It is also positively associated with higher inflation, excess market returns, and long-term government bond yields. The
increased economic activity induced by the shock also trigger tighter monetary policy.Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
To address potential look-ahead bias concerns, the study demonstrates that the AI Economy Score masked remains a significant predictor of real GDP growth in the subsequent quarter even after controlling for established GDP predictors. p. 4 First, in a “horse race” against
the Survey of Professional Forecasters (SPF), the AI Economy Score retains its predictive power
and proves to be a slightly stronger predictor of real GDP when both measures are included in
the same model. Second, to address potential “look-ahead bias,” we anonymize transcripts, i. We
employ several methods to validate the responses generated by ChatGPT: (i) Manual review
and frequency analysis: We manually read through a random sample of around one thousand
conference calls and ChatGPT responses to verify the accuracy and relevance of the generated
content. We also examine the most commonly used words associated with both low and high AI
Economy Scores to gauge the consistency and appropriateness of ChatGPT’s outputs; (ii) Trend
analysis: We analyze trends over time and across various industries to assess the coherence and
relevance of ChatGPT’s responses within different contexts. 5 training data cutoff, and confirm that firm-level scores still robustly predict
future sales and value-added. Finally, we replicate the analysis using other large-language
models, including the recent GPT-5 mini model and another open-source model, showing that
our results are not dependent on a specific generative AI model. In all levels of analyses, we focus on forecasting the following common economic indicators:
GDP, production, employment, and wages. For firm-level predictions, we use Value-Add, or the
difference between sales and cost of goods sold, as a proxy for the overall value created by firms
and Sales as a proxy for production.Where the paper says it
Where the paper says it
Where the paper says it
AI Weighted Score,
it can forecast national GDP growth for up to 14 quarters. At the firm level, the AI Firm Score shows a strong positive correlation with metrics such as value-added, sales, employment, and wage levels over four years when controlling for other known macroeconomic indicators and firm covariates. p. 6Where the paper says it
The paper emphasizes that managerial expectations are powerful because managers can act upon their beliefs, leading to tangible economic impacts. p. 3 Why are managerial expectations so powerful in forecasting future economic activities? One
important reason is that managers can act upon their beliefs and their actions carry significant
economic impacts. realized economic outcomes. This analysis reveals that the outlooks of pessimistic managers are
particularly informative, demonstrating superior predictive accuracy, especially during economic
crises. We then examine how a firm’s operational structure shapes its manager’s perspective,
testing whether diversification provides a clearer or more muddled economic view. Our findings
indicate that forecasts from managers at more focused firms, or those with lower business and
geographic diversification, are significantly more powerful, suggesting that deep, specialized
knowledge provides a stronger economic signal. Therefore, aggregating such information
and insights from managers would be particularly fruitful. Overall, our study shows that the
new AI tools allow easy extraction of managerial expectations on a whole range of economic
indicators, providing a new set of forward-looking information to researchers, investors, and
regulators.Where the paper says it
Where the paper says it
Where the paper says it
AI Weighted Score,
constructed using weights derived from a firm-quarter-level regression model relating scores to future firm sales, further improves predictions of GDP growth at both national and industry levels, with influential components including revenue, production, wages, employment, industry prospects, and capital expenditures. Overall, the contribution lies in providing a readily constructible measure from public data that offers forward-looking information for researchers and investors at granular levels. p. 3Where the paper says it
Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.
Improvements for AI systems
-
Better macroeconomic forecasting capability: The AI Economy Score can predict GDP growth, production, and employment up to 10 quarters ahead, which is beyond existing measures like survey forecasts.
-
Enhanced microeconomic insights: The model allows for the construction of more micro-level expectation variables, such as an AI-based industry-level economy score, which exhibit substantial cross-sectional heterogeneity with both pessimistic and optimistic outlooks for different industries at the same point in time.
-
Improved long-term firm performance prediction: Firm-level AI scores forecast future firm sales and value added with predictive power for up to four years, suggesting that
firm-level expectations carrying predictive power for up to 4 years.
-
Enhanced multi-dimensional forecasting: By constructing the AI Weighted Score, which is a linear combination of 14 scores, the system can predict real GDP growth at both national and industry levels with a predictive horizon extending to up to 16 quarters.
-
Bias mitigation in analysis: The system's results are robust to look-ahead bias when analyzed using anonymized transcripts, where
the AI Economy Score masked remains positively and significantly associated with real GDP growth in the subsequent quarter, even after controlling for established predictors of GDP growth.
-
Behavioral insight integration: The system can classify managers into Optimist, Pessimist, and Realist scores based on their historical forecasting patterns relative to realized economic outcomes, demonstrating that
pessimistic managers are particularly informative, demonstrating superior predictive accuracy.
-
Diversification effect quantification: The system can quantify the impact of firm structure by showing that
the predictive power of economic forecasts is stronger for more focused firms compared to more diversified firms,
suggesting deep, specialized knowledge provides a stronger economic signal.
Abstract
Using generative AI, we extract managers' economic expectations from over 120,000 corporate conference call transcripts. The resulting AI Economy Score predicts GDP growth, production, employment, and wages five or more quarters ahead, beyond survey forecasts and other existing measures. Industry- and firm-level scores capture sector- and firm-specific activity, and a composite of firm, industry, and macroeconomic expectations further improves forecasting power and horizon. Predictability concentrates among historically more accurate, less biased managers, consistent with an information rather than a noise channel. Managerial expectations thus offer incremental insights into economic activity, with implications for macro- and microeconomic decision-making.
Sources
Related papers
- AFNO: Universal Certifiably-Arbitrage-Free Operator Learning Framework for European Call-Price Surfaces
- Diffusion models for dynamic volatility surface generation and data-driven hedging
- Rethinking Synthetic Scenario Realism: Compatibility, Not Fidelity, Drives Hedging Performance
- CosmosAlpha: Multi-Relational Graph Search with Pareto-Guided Exploration for LLM-Driven Alpha Mining
- Diversification Across Predictors: Heterogeneous Ensemble Learning for Cross-Sectional Return Prediction
- DSINC beyond Heston: the 3/2 and other non-affine stochastic-volatility models