Inflation Forecasting

The Evolution of Inflation Challenges

• Historical Shift in Central Bank Policy
  • Pre-Global Financial Crisis: Struggled to control inflation (fear of wage hikes, fiscal expansion).
  • Post-Pandemic (2022-2025): Struggling to push inflation down amid volatile global trends.
• Factors Influencing Inflation Forecasting:
  • Globalization & Labor Markets: Entry of low-wage workers complicates wage-inflation dynamics.
  • Macroeconomic Uncertainty: Geopolitical issues (e.g., trade wars, oil price fluctuations) impact inflation.

RBI’s Approach to Inflation Forecasting

• India’s Inflation Landscape:
  • CPI inflation highly impacted by food prices, global commodity trends, and gold purchases by central banks.
  • Emerging monetary policy challenges require advanced forecasting techniques.
• AI’s Role in Enhancing Economic Research:
  • RBI Governor has emphasized the need for advanced tools in forecasting inflation trajectories.

How Large Language Model (LLM) Can Revolutionize Inflation Forecasting

A. Nowcasting & Real-Time Data Analysis

• Traditional forecasting relies on numerical data, whereas AI-driven LLMs analyze:
  • News articles, reports, social media trends to identify inflationary trends.
  • “Nowcasting” short-term inflation predictions more accurately.

B. AI-Driven Forecasting Accuracy

• Studies like Faria-e-Castro & Leibovici (2024) at the Fed show that LLMs:
  • Provide lower mean-squared errors in inflation forecasts.
  • Outperform traditional Survey of Professional Forecasters methods.
• Bybee (2023) used GPT-3.5 to simulate economic expectations successfully.

C. AI Agents in Inflation Expectation Surveys

• AI-driven survey simulations can provide insights into household inflation expectations:
  • Households’ expectations are crucial for monetary policy but are costly to survey.
  • AI agents can simulate consumer responses, providing cost-effective alternatives.

AI’s Expanding Role in Economic Decision-Making

• Rise of Generative AI (GAI):
  • 40% of US adults (by Aug 2024) use AI, with 28% using it at work (Bick et al., 2024).
  • Half of US households now use AI tools (Aldasoro et al., 2024).
• AI’s Influence on Policy:
  • As AI-assisted decision-making increases, central banks must adapt AI-driven forecasting models.

Challenges & Limitations of LLMs in Forecasting

• Data Control Issues:
  • LLMs are trained on external datasets; central banks lack control over training data.
  • Training data is not timestamped, making real-time retraining difficult.
• Model Replicability Issues:
  • Publicly available LLMs are retrained periodically, posing challenges in maintaining consistent results.
• Despite these caveats, LLMs offer a new frontier in economic forecasting.

AI Success Story: Predicting Indian Elections (2024)

• AI-based election forecasting outperformed traditional exit polls:
  • Kcore Analytics used AI-driven analysis of social media trends to predict results.
  • Incorporated economic sentiment, including inflation, into its predictive model.
• Lesson: AI can enhance economic and political forecasting, influencing decision-making.

The Future of AI in Inflation Forecasting

• AI-driven inflation forecasting is a game-changer, offering:
  • Faster, more accurate predictions using real-time data.
  • Cost-effective alternatives to traditional surveys.
  • Enhanced economic policymaking tools for central banks.
• Despite limitations, integrating LLMs into monetary policy is a logical next step.