Applied Microeconomics

Using microeconomic theory and data to answer real-world questions about households, firms, and policy with credible causal inference.
On this page

In one sentence

Applied microeconomics uses microeconomic models plus data to estimate causal effects and evaluate policies, institutions, and market designs.

What applied microeconomists do

Applied micro typically starts with a concrete question (about incentives, constraints, and behavior) and then asks what evidence can identify a causal answer.

Common objects of study include:

  • households (consumption, labor supply, education),
  • firms (pricing, productivity, entry/exit, innovation),
  • markets (competition, matching, bargaining),
  • policy (taxes, transfers, regulation, procurement, nudges).

Core empirical challenge: identification

Correlation is usually not enough. If a policy is adopted where conditions are already improving, outcomes may rise even if the policy did nothing. Applied micro relies on research designs that can isolate causality:

Common designs (toolkit)

  • Randomized controlled trials (RCTs): treatment assigned randomly.
  • Difference-in-differences (DiD): compare changes over time between treated and comparison groups.
  • Regression discontinuity (RD): exploit cutoff-based assignment rules.
  • Instrumental variables (IV): use a variable \(Z\) that shifts treatment \(D\) but affects outcomes \(Y\) only through \(D\).
  • Structural estimation: estimate primitives (preferences/technology) to simulate counterfactuals.

One compact way to see IV is:

\[ \text{IV estimand} = \frac{\text{Cov}(Z,Y)}{\text{Cov}(Z,D)} \]

    flowchart TD
	  Q["Question<br/>(what changes behavior?)"] --> D["Data<br/>(who, where, when)"]
	  D --> I["Identification strategy<br/>(RCT, DiD, RD, IV)"]
	  I --> E["Estimate causal effect"]
	  E --> C["Counterfactuals<br/>(what if policy changed?)"]
	  C --> P["Policy or business decision"]

Examples of applied micro questions

  • Labor: Do minimum wages reduce employment? How large are firm wage premia?
  • Public finance: How do taxes affect labor supply and evasion? What is the incidence of a sugar tax?
  • IO/competition: Do mergers raise prices? How do entry barriers affect market power?
  • Health: How do copayments change utilization and health outcomes?
  • Education/development: Do tutoring, cash transfers, or information interventions improve learning and earnings?
  • Identification: The conditions under which a causal parameter can be learned from the observed data.
  • Causal Inference: Methods for estimating cause-and-effect relationships rather than associations.
  • Randomized Controlled Trial (RCT): An experiment where treatment is randomly assigned.
  • Difference-in-Differences (DiD): A design comparing changes over time across treated and control groups.
  • Instrumental Variables (IV): A method using an instrument to identify causal effects when treatment is endogenous.
  • Structural Model: A model that estimates primitives (preferences/technology) to simulate counterfactuals.

Quiz

### Applied microeconomics is best described as: - [x] Using theory and data to estimate causal effects in real settings - [ ] Studying only abstract models with no data - [ ] Forecasting GDP using only time series - [ ] Accounting and bookkeeping > **Explanation:** Applied micro is about credible empirical answers to microeconomic questions. ### A central goal of “identification” is to: - [x] Separate causation from correlation - [ ] Maximize the number of regressors - [ ] Avoid collecting any data - [ ] Ensure prices always clear markets > **Explanation:** Identification is about what assumptions/design make a causal interpretation valid. ### Which research design uses a cutoff rule for assignment? - [ ] Difference-in-differences - [x] Regression discontinuity - [ ] Time-series smoothing - [ ] National accounting > **Explanation:** RD exploits discontinuities at thresholds. ### In difference-in-differences, you compare: - [x] Changes over time in treated vs control groups - [ ] Levels at one point in time only - [ ] Two unrelated countries with no common trends - [ ] Only the treated group before and after > **Explanation:** DiD uses a comparison group to net out common shocks. ### Instrumental variables are typically used when: - [x] Treatment choice is endogenous (correlated with unobservables) - [ ] Randomization is perfect and costless - [ ] There is no measurement error - [ ] Prices are fixed by law > **Explanation:** IV can address endogeneity if the instrument is valid. ### True or False: Applied micro often combines a simple economic model with an empirical design. - [x] True - [ ] False > **Explanation:** Theory helps define the mechanism and parameter; the design helps identify it. ### A major threat to causal interpretation in observational data is: - [x] Selection bias / omitted variables correlated with treatment and outcomes - [ ] The existence of markets - [ ] The definition of GDP - [ ] The law of demand > **Explanation:** Endogeneity can create correlations that are not causal. ### In DiD, a key identifying assumption is: - [x] Parallel trends (treated and control would have evolved similarly absent treatment) - [ ] Zero inflation always - [ ] Perfect competition in all markets - [ ] Random assignment by definition > **Explanation:** Without parallel trends, changes may reflect different underlying trajectories. ### In RCTs, “noncompliance” means: - [x] Some units do not take the assigned treatment status - [ ] The treatment has zero effect - [ ] The sample size is infinite - [ ] The outcome is measured without error > **Explanation:** Noncompliance creates a gap between assignment and actual treatment received. ### When treatment varies at a group level (e.g., by school or village), a standard practice is to: - [x] Cluster standard errors at the assignment level - [ ] Always use no standard errors - [ ] Treat all observations as independent even within groups - [ ] Replace the design with a supply curve > **Explanation:** Group-level shocks create correlation within clusters and affect inference.
Apprenticeship
Antitrust