Please enable JavaScript.
Coggle requires JavaScript to display documents.
Essay 1 - Coggle Diagram
Essay 1
Bridging the gap?
Lucas's framework is superior for policy evaluation, where one needs to model expectational responses and regime changes.
Friedman’s approach remains compelling for forecasting and empirical testing, where predictive accuracy is paramount.
In practice, modern macroeconomics reflects a tension between these views. Policymakers may use DSGE models for scenario analysis and SAM for short-term forecasting. As macroeconomists develop hybrid approaches—e.g., empirical DSGEs, estimated via Bayesian methods
In sum, Lucas and Friedman define complementary, not contradictory, criteria. The former ensures theoretical consistency under policy shifts; the latter prioritizes empirical validation. A truly “good” model today is likely one that balances both: structurally interpretable and empirically grounded.
What makes a good model?
- Two of the most influential and contrasting positions were articulated by Milton Friedman (1953) and Robert Lucas (1976).
-While Friedman emphasizes predictive accuracy,
-Lucas demands structural invariance under policy change.
Friedman (1953): Predictive Power Over Realism:
- Friedman outlines a view grounded in instrumentalism: the value of a model lies not in the realism of its assumptions, but in its ability to predict observable phenomena.
- “truly important and significant hypotheses will be found to have ‘assumptions’ that are wildly inaccurate descriptive representations of reality.”
- A good model, by his standard, works like a useful map—even if simplified, it helps us navigate reality effectively.
- Friedman’s philosophy allows economists to abstract from messy details of reality to focus on empirically relevant mechanisms.
-
"A model doesn’t have to be realistic. It’s like a map—it’s useful even if it’s not totally accurate."
Can the model predict the real world?
He doesn’t care if the assumptions (like “everyone is rational” or “markets are perfect”) are weird or wrong—as long as the model gets results right.
“Don’t tell me how the clouds work. Just tell me if it’ll rain.”
Strenght: Simple, often accurate short-run predictions. Weakness: Breaks down if policy changes or if people start acting differently.
Lucas (1976): The Lucas Critique and Structural Invariance: Lucas points out that empirical relationships found in macroeconomic data are not policy-invariant. - Lucas proposes that models should be built on deep, structural parameters—not correlations—that remain stable under different policy regimes. These include preferences, technology, and budget constraints. The model must reflect optimizing behavior and rational expectations so that it can be used to evaluate the effect of counterfactual policies.
NEMO, the core DSGE (Dynamic Stochastic General Equilibrium) model used by Norges Bank
Lucas (1976) disagrees.
He says models have to be based on real behavior, like how people make choices and respond to government policies.
Why? Because:
- If the government changes a rule (like taxes), people change their behavior.
- So a model that only uses past data might fail if the rules change.
He says a good model needs "deep parameters" — things like preferences and technology — that don’t change when policy changes.
“If you don’t understand how clouds form, your forecast will be wrong if the climate changes.”
Strenght: You can analyze policy changes because people in the model respond to those changes.
Weakness:
- Can be complex and unrealistic; hard to match real-world data.
- Often bad at forecasting short-run fluctuations.
NEMO:
- NEMO, the core DSGE model used by Norges Bank, is explicitly constructed to address Lucas’s concerns.
- It is structurally microfounded and designed to simulate how the economy responds to monetary policy changes.
- Therefore, NEMO satisfies Lucas’s criterion of a good model.
Friedman’s perspective: NEMO’s track record is more mixed. DSGE models often struggle with forecasting accuracy, especially out-of-sample. They are heavily calibrated, and their ability to match real-time data (e.g., inflation persistence or labor market dynamics) may be limited. So while theoretically sound, their empirical performance may fall short of Friedman’s standard.
NEMO: A Lucas-style framework.
- Used by Norway’s central bank.
- Has deep structure: people optimize, firms set prices, etc.
- Good for policy simulation, but not perfect for forecasting.
SAM framework (System for Averaging Models): SAM pools forecasts from multiple models—structural and reduced-form—prioritizing forecast accuracy over theoretical purity. It has no commitment to microfoundations or deep parameters but aims to capture the “wisdom of the crowd.”
SAM: A Friedman-style framework.
- Blends lots of models together to improve predictions.
- Doesn’t care as much about microfoundations.
- Great for forecasting, but not for analyzing deep policy changes.