Skip to contentScholarGate
LibraryBookshelfDeskReview StudioAssistant
Sign in
On this page
IntuitionHow it worksWhen to use itStrengths & limitationsCommon pitfallsApplicationsFrequently asked🔒 Read the full methodSourcesRelated methods
Cite this pageSpotted an issue on this page? Report or suggest a fix →
Home›Causal inference›Policy Evaluation with Instrumental Variables
Regression modelQuasi-experimental / causal inference

Policy Evaluation with Instrumental Variables

Instrumental Variables Estimation for Policy Evaluation · Also known as: IV policy evaluation, 2SLS policy analysis, natural-experiment IV, policy IV estimation

Instrumental Variables (IV) estimation for policy evaluation is a quasi-experimental technique that uses an exogenous instrument — a variable that shifts exposure to a policy but is otherwise unrelated to the outcome — to recover the causal effect of a program or intervention from non-experimental data. Popularised in policy research by Angrist, Imbens, and Rubin (1996), it identifies the Local Average Treatment Effect (LATE) among units whose treatment status is changed by the instrument.

ScholarGate
  1. Regression model
  2. v1
  3. 2 Sources
  4. PUBLISHED
Cite this page →
Tools & resources
Download slides
Learn & explore

Read the full method

Members only

Sign in with a free account to read this section.

Sign in

Method map

The neighbourhood of related methods — select a node to explore.

Policy Evaluation Instrumental Variables
2SLS RegressionDifference-in-DifferencesInstrumental Variables i…Local Average Treatment…Propensity Score Matching

When to use it

Use IV policy evaluation when treatment assignment is not random but a plausible instrument exists — for example, a lottery, a geographic boundary, a policy-rollout cutoff, or a natural experiment that creates exogenous variation in who receives the intervention. It is appropriate for cross-sectional or panel data with continuous or binary outcomes, and where OLS or DiD are suspect due to selection bias or simultaneous causality. Do not use IV if no credible instrument is available, if the first-stage F-statistic is below 10 (weak instrument), if the instrument appears to directly affect the outcome, or if the complier sub-population is too narrow to be policy-relevant.

Strengths & limitations

Strengths
  • Recovers a causal policy effect from observational data when randomisation is infeasible, by exploiting exogenous variation in treatment exposure.
  • Accounts for both self-selection into treatment and simultaneous causality, biases that OLS cannot correct.
  • The LATE estimate has a clear, policy-relevant interpretation: the effect for the units whose behaviour the policy instrument actually shifts.
  • Natural-experiment instruments (lotteries, eligibility thresholds, geographic variation) are often highly credible and widely accepted in academic and policy audiences.
  • Easily extended to panel data and heterogeneous-treatment-effect frameworks, including the LATE with heterogeneous treatment effects.
Limitations
  • A valid instrument is genuinely difficult to find; most observational datasets do not contain an obvious and defensible IV.
  • The estimate is LATE — causal only for compliers — and may not generalise to the broader population or to other policy contexts.
  • Weak instruments (low first-stage F) produce large standard errors and finite-sample bias toward OLS, undermining the advantage of IV.
  • The exclusion restriction is untestable; its plausibility depends entirely on the researcher's theoretical argument, leaving room for disagreement.
  • With multiple instruments or heterogeneous effects, interpretation becomes complex and requires careful consideration of which complier group is identified.

Frequently asked

What makes an instrument valid?

Two conditions must hold. Relevance: the instrument must be correlated with the treatment variable, verified by a first-stage F-statistic above 10. Exclusion: the instrument must affect the outcome only through the treatment, not through any independent channel. Relevance is testable; exclusion is a theoretical claim that must be argued on substantive grounds.

What is the Local Average Treatment Effect (LATE)?

LATE is the average causal effect of treatment for compliers — units whose treatment status changes because of the instrument. It is not the effect for the entire population; never-takers and always-takers, who do not respond to the instrument, are not identified. Whether LATE is the policy-relevant parameter depends on whether compliers are the group the policymaker cares about.

How is 2SLS related to IV estimation?

Two-Stage Least Squares (2SLS) is the standard estimation procedure that implements IV with one or more instruments and covariates. In the first stage, treatment is regressed on the instrument(s) and covariates. In the second stage, the outcome is regressed on predicted treatment values. With a single instrument and no covariates, 2SLS is algebraically identical to the simple IV ratio estimator.

What if my first-stage F-statistic is below 10?

A weak instrument produces IV estimates with large standard errors and finite-sample bias that can be as bad as OLS. You should report the weakness, consider weak-instrument-robust confidence intervals (e.g. Anderson-Rubin), search for stronger instruments, or be transparent that the IV approach is not credible with the available data.

Can I use IV with panel data?

Yes. IV naturally extends to panel settings, where individual fixed effects are included in both stages to control for time-invariant unobservables. Panel IV is common in policy evaluation when treatment varies over time within units and a time-varying instrument can be constructed.

Sources

  1. Angrist, J. D., Imbens, G. W., & Rubin, D. B. (1996). Identification of Causal Effects Using Instrumental Variables. Journal of the American Statistical Association, 91(434), 444-455. DOI: 10.1080/01621459.1996.10476902 ↗
  2. Angrist, J. D., & Pischke, J.-S. (2009). Mostly Harmless Econometrics: An Empiricist's Companion. Princeton University Press. ISBN: 978-0691120355

How to cite this page

ScholarGate. (2026, June 3). Instrumental Variables Estimation for Policy Evaluation. ScholarGate. https://scholargate.app/en/causal-inference/policy-evaluation-instrumental-variables

Related methods

2SLS RegressionDifference-in-DifferencesInstrumental Variables in Health ResearchLocal Average Treatment EffectPropensity Score Matching

Which method?

Set this method beside its closest kin and read them side by side — the library lays the books on the table; the choice is yours.

  • 2SLS RegressionEconometrics↔ compare
  • Difference-in-DifferencesEconometrics↔ compare
  • Instrumental Variables in Health ResearchHealth Economics↔ compare
  • Local Average Treatment EffectCausal inference↔ compare
  • Propensity Score MatchingResearch Statistics↔ compare
Compare side by side →

Similar methods

Instrumental Variables in Education ResearchInstrumental Variables in Health ResearchHeterogeneous treatment effect Instrumental variablesTwo-Stage Least Squares (2SLS)2SLS RegressionPanel Data Instrumental VariablesLocal Average Treatment EffectSpatial Instrumental Variables

Related reference concepts

Instrumental Variables (IV) EstimationInstrumental Variables (IV) EstimationSingle Equation Models • Single VariablesNatural ExperimentQuasi-Experimental and Natural Experiment DesignCausal Identification

Spotted an issue on this page? Report or suggest a fix →

ScholarGate — Policy Evaluation Instrumental Variables (Instrumental Variables Estimation for Policy Evaluation). Retrieved 2026-07-21 from https://scholargate.app/en/causal-inference/policy-evaluation-instrumental-variables · Dataset: https://doi.org/10.5281/zenodo.20539026
Quick facts
Originator
Angrist, Imbens & Rubin (canonical 1996 JASA framework); foundational IV roots in Wright (1928) and Theil (1953)
Year
1996 (modern policy-evaluation framing); IV roots 1920s
Type
Quasi-experimental causal inference / IV regression
DataType
Observational or natural-experiment data with a valid instrument; cross-sectional or panel
Subfamily
Quasi-experimental / causal inference
Related methods
2SLS RegressionDifference-in-DifferencesInstrumental Variables in Health ResearchLocal Average Treatment EffectPropensity Score Matching
ScholarGate

A content-first reference library for research methods — what each one is, how it works, and where it comes from.

Open data (CC-BY)

Explore

  • Library
  • Search the library…
  • Browse by field
  • Fields
  • Journey
  • Compare
  • Which method?

Reference

  • Subjects
  • Atlas
  • Glossary
  • Methodology
  • Philosophy

Your tools

  • Bookshelf
  • Desk
  • Chat

Company

  • About
  • Pricing
  • Contact
  • Suggest a method

Entries are compiled from published sources for reference. Verifying the accuracy and suitability of any information for your own use remains your responsibility.

© 2026 ScholarGate · A research-method reference library
  • Privacy
  • Cookies
  • Terms
  • Delete account