Policy Analysis Framework
Cost-benefit, stakeholder mapping, implementation science — how to evaluate a policy proposal.
Frameworks
Bardach's 8-fold path
Eugene Bardach's 'A Practical Guide for Policy Analysis' is the canonical analytic template taught at Berkeley, the Kennedy School, and most MPA programs.
Key Points
- 1. Define the problem.
- 2. Assemble evidence.
- 3. Construct alternatives.
- 4. Select criteria.
- 5. Project outcomes.
- 6. Confront trade-offs.
- 7. Decide.
- 8. Tell your story.
Cost-benefit analysis
CBA monetizes costs and benefits in common units to enable comparison. Not always appropriate (civil rights? human dignity?) but essential for infrastructure, environmental, and health policy.
Direct costs
Out-of-pocket expenditure by government.
Indirect costs
Compliance, opportunity costs, administrative burden.
Direct benefits
Measurable gains (reduced mortality × VSL, jobs created, emissions avoided × social cost of carbon).
Distributional analysis
Who pays and who benefits? Aggregate NPV can mask serious equity problems.
Stakeholders
Stakeholder mapping
Every policy affects different groups differently. Map them explicitly before making a recommendation.
Key Points
- Who are the direct winners and losers?
- Who has power to block or enable implementation?
- What is each stakeholder's existing public position?
- Are there latent stakeholders who only mobilize when the policy launches?
Power / interest matrix
A 2×2 grid: high/low power × high/low interest. Yields four management approaches.
High power + high interest
Manage closely — cabinet peers, committee chairs, major unions.
High power + low interest
Keep satisfied — treasury, regulators.
Low power + high interest
Keep informed — advocacy groups, affected communities.
Low power + low interest
Monitor — general public, distant stakeholders.
Implementation
Pressman & Wildavsky's complexity problem
Their 1973 book 'Implementation' studied why a federal jobs program in Oakland kept failing. Insight: success requires every step in a chain of decisions to go right; each 90% probability multiplies quickly to failure.
Key Points
- 10 decision points × 90% success each = 35% overall success.
- Reduce decision points. Align incentives. Build in feedback loops.
- Design for the 'street-level bureaucrat' (Lipsky 1980) actually implementing.
Measuring what matters
Key Points
- Output metrics: immediate deliverables (forms processed, vaccines administered).
- Outcome metrics: the change you actually want (health outcomes, employment, test scores).
- Pre-register the outcome metric before the policy ships — prevents post-hoc goalpost-moving.
- Randomized evaluations where possible (J-PAL's 1000+ RCTs on development policies).
FAQ
What counts as strong policy evidence?
A hierarchy: RCTs > quasi-experimental (difference-in-differences, regression discontinuity) > observational studies > expert opinion. Context matters — external validity from a Bangladesh RCT may not transfer to Chicago.
Policy vs politics — where does analysis end?
Analysts typically stay one layer above the politics. Identify the politically feasible set; let the principal decide within it. But never sanitize a policy's distributional consequences — that's where analysts earn their salary.
Continue learning
Explore related MUN guides to deepen your skills.