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.
Source: BS
