Ida Wolden Bache: Research-Based Models in Monetary Policy Decision-Making

Why It Matters

Enhanced modelling improves Norges Bank’s ability to anticipate inflation and consumption responses, supporting more effective rate decisions in a volatile global environment. The insights also inform other central banks seeking to modernise their analytical toolkit.

Key Takeaways

• NEMO combines data‑driven forecasts with long‑term structural trends
• AI and machine learning are being piloted to lower forecast errors
• Micro‑data reveal debt‑laden households cut consumption after rate hikes
• Cash‑flow channel now explicit, complementing traditional substitution effects
• Model upgrades target supply‑side shocks and international trade links

Pulse Analysis

Norges Bank’s modeling ecosystem reflects a broader shift among central banks toward data‑rich, research‑driven policy analysis. While traditional autoregressive and VAR models remain core, the institution has layered them within a dynamic stochastic general‑equilibrium framework—NEMO—that captures Norway’s unique features, such as its petroleum sector and high household debt levels. By continuously weighting models based on out‑of‑sample performance, the bank can swiftly incorporate fresh indicators, from high‑frequency card‑transaction data to text‑based sentiment measures, ensuring that forecasts stay relevant during abrupt economic shifts like the pandemic.

A key frontier for Norges Bank is the integration of artificial intelligence and machine‑learning techniques. Early pilots suggest these tools can automate model selection, refine weighting schemes, and uncover nonlinear relationships that conventional econometric approaches may miss. This technological infusion promises to reduce forecast errors, especially in the short‑run horizon where policy decisions are most sensitive. Moreover, the bank’s commitment to micro‑data—spanning household finances, labour‑market dynamics, and five decades of price information—enables a granular view of how policy‑rate changes ripple through the economy via the cash‑flow channel, a mechanism that standard representative‑agent models often understate.

The practical implications of these advancements are significant. Empirical evidence now shows that a one‑percentage‑point rate increase can depress consumption by up to 1.5 percentage points for households with debt‑to‑income ratios three times the norm, while also dampening inflation over a multi‑year horizon. By embedding such heterogeneity into NEMO and complementary HANK models, Norges Bank can better assess distributional outcomes and trade‑off scenarios. As global shocks—from supply‑chain disruptions to geopolitical tensions—continue to test monetary policy, a more adaptable, AI‑enhanced modeling suite equips the bank to navigate uncertainty with greater precision and credibility.