Font size:
Print
Role of AI in Inflation Forecasting
Context:
The year 2022 marked a turning point in global economic trends as inflation surged unexpectedly, reversing previous patterns of low inflation.
More on News
- This shift underscored the complexities of monetary policymaking in the post-pandemic era, particularly in India, where the Reserve Bank of India (RBI) now faces similar challenges.
- The interconnected nature of global markets has significantly influenced inflationary cycles, further complicated by the integration of low-wage workers into the global workforce.
Response of Central Banks
- Modernised Methods: In response to these challenges, central banks, including the RBI, must modernise their economic forecasting methods.
- The new RBI governor has already emphasised the need for technological advancements in this domain.
- LLMs: A key innovation in economic forecasting is the integration of Large Language Models (LLMs).
- These AI-driven models offer a transformative approach to understanding and predicting inflation.
- Given that central banks invest substantial resources in surveying consumer inflation expectations, LLMs could serve as a viable alternative by replicating human survey patterns and processing economic information in real-time.
Leveraging LLMs for Inflation Estimation
LLMs can improve inflation forecasting in two primary ways:
- Analysing Large Textual Data Sources: LLMs can process vast amounts of information from news articles, economic reports, and social media to detect inflation trends and sentiments.
- This approach provides a more nuanced and timely assessment compared to traditional methods, which rely solely on numerical data.
- By enhancing “nowcasting” techniques—short-term inflation forecasting—AI and machine learning models are becoming integral to economic analysis.
- Deploying AI Agents for Inflation Surveys: AI-driven survey agents can simulate consumer responses to structured inflation expectation surveys, reducing the reliance on costly and time-consuming household surveys.
- These agents incorporate external knowledge, helping policymakers understand how economic information shapes public expectations.
- Given that household inflation expectations influence consumer behaviour and central bank policies, this approach could significantly enhance economic decision-making.
- Effectiveness: Recent studies underscore the effectiveness of AI in inflation forecasting.
- Research by Miguel Faria-e-Castro and Fernando Leibovici (2024) at the Federal Reserve Bank of St. Louis highlights how LLMs like Google’s PaLM generate inflation forecasts with lower errors compared to traditional survey-based models.
- Similarly, Bybee (2023) demonstrated the potential of GPT-3.5 in simulating economic expectations, reinforcing the credibility of AI-driven forecasting techniques.
Challenges and Future Prospects
- Despite its promise, the use of LLMs for inflation forecasting comes with limitations.
- These models are trained on specific datasets selected by developers, restricting user control over the training process.
- Additionally, the absence of time-stamped data prevents true out-of-sample forecasts.
- Furthermore, publicly available LLMs undergo regular retraining, making replicability a challenge for researchers.
AI’s Growing Influence on Economic Forecasting
- Studies indicate that nearly 40% of U.S. adults had used GAI by August 2024, with 28% incorporating it into their work.
- As AI-driven decision-making becomes more prevalent, understanding its implications for economic forecasting is essential.
- A compelling example of AI’s predictive capabilities emerged during India’s 2024 general elections.
- While traditional exit polls failed to accurately predict results, Kcore Analytics, an AI-driven research firm, successfully forecasted voter preferences.
- By analysing social media interactions—what people read, wrote, and engaged with—alongside key economic indicators like inflation, the firm delivered an accurate prediction. This demonstrates AI’s potential to process real-world economic and political data effectively.
In the words of Milton Friedman, albeit with a modern twist: “Inflation is everywhere a monetary (and a political) phenomenon.” As AI continues to reshape economic research, central banks must embrace these technologies to enhance forecasting accuracy and support informed policymaking.
