Microfoundations Decomposition
Explain large economic outcomes by tracing the people, tasks, and incentives underneath
- Difficulty
- Moderate
- Time to result
- ~weeks to results
- Steps
- 5
- Confidence
- 92%
Microfoundations Decomposition starts with a broad relationship—such as technology and inequality or unemployment and inflation—and moves downward to the actors and tasks that generate it. Instead of treating the aggregate correlation as the explanation, the analyst asks what machines do, what workers seek, how jobs are designed, and how people are matched to opportunities. Those mechanisms are then connected back to the economy-wide result. The method prevents vague macro narratives from hiding contradictory local effects. It also supports better decisions because interventions can target a specific incentive, task, constraint, or matching failure rather than an abstract outcome. Its strength comes from moving deliberately between the micro and macro levels until the causal story is both detailed and systemically coherent.
Origin
Extracted from Masters in Business, where Daron Acemoglu explained the micro-level approach behind his doctoral work on the foundations of macroeconomics.
Core principles
- 01Break aggregate relationships into observable components.
- 02Trace outcomes to individual incentives, choices, and constraints.
- 03Study tasks and matching processes rather than relying on broad correlations.
- 04Reconnect micro-level mechanisms to the macro outcome.
How to run it
- 1
State the macro relationship
Define the large-scale outcome or correlation that needs explaining, such as automation reducing employment or institutions affecting growth.
Pro tip Phrase it as a causal question rather than a descriptive trend.
Watch out Do not treat the initial correlation as the mechanism.
- 2
Identify actors and tasks
List the workers, firms, machines, regulators, jobs, and transactions directly involved in producing the outcome.
Pro tip Use tasks as the unit of analysis when job titles hide important differences.
Watch out Avoid categories so broad that individual behavior disappears.
- 3
Map incentives and constraints
Determine what each actor is trying to achieve and what limits their choices, including skills, prices, rules, and available opportunities.
Pro tip Separate what actors want from what the system permits them to do.
Watch out Do not assume every actor responds to the same incentive.
- 4
Trace interactions
Explain how choices, task substitution, matching, and institutional rules interact to create local effects.
Pro tip Look for feedback loops and adjustment channels.
Watch out Ignoring adaptation can exaggerate immediate effects.
- 5
Reconstruct the macro outcome
Aggregate the micro mechanisms and assess whether they explain the original economy-wide pattern and relevant contrasting cases.
Pro tip Compare places or periods where one mechanism differs.
Watch out Do not generalize beyond the cases the mechanism can explain.
In the wild
An analyst studying falling local employment does not stop at the number of robots installed. They identify which production tasks robots replace, which workers previously performed them, whether firms expand output, and whether displaced workers can enter new tasks or sectors. They then aggregate those channels to estimate the local employment effect.
→ The analysis distinguishes direct displacement from offsetting demand, productivity, and reallocation effects.
Common mistakes
Treating correlation as explanation
A relationship between two aggregate variables does not reveal the actors or processes producing it. The framework requires a mechanism beneath the pattern.
Ignoring adjustment behavior
Workers and employers may retrain, reorganize work, invest in capital, or enter other sectors. Leaving out these responses produces a static and potentially misleading analysis.
Is it for you?
Best for
It is best for researchers, strategists, and policymakers investigating employment, inflation, inequality, or technological change.
Not ideal for
It is not ideal when only a rapid descriptive summary is required and causal detail cannot be observed.
From the transcript
“macro is so complicated we all struggle with it”
“What is it that new machines are doing,”
“So it's both the micro and the macro.”
From the episode
Interview With Daron Acemoglu: Masters in Business (Audio)
Daron Acemoglu