Notes on Decision Making in Public Policy: Rational Choice, Bounded Rationality, and Decision-Making Models
Group project logistics
Next week in tutorials: start forming groups; self-allocation this year (groups of 3–4 for a 4,000-word paper).
Finalizing group membership and starting the group writing process at the end of week seven.
Additional guidance to be provided in tutorials (by instructor and Nick).
Apology for sniffly/nasal recording; this is a note about the lecture context, not content.
Context and scope of the lecture
Topic: decision making within the policy process.
Context: policy cycle; decisions occur after consultation and coordination; authoritative decisions select a course of action from identified alternatives.
Key idea: decision making is the stage of the policy cycle where government resources and authority are formally committed to a chosen action.
Important caveat: all processes occur within highly contextual environments with constraints.
Types of policy decisions:
Positive decisions: alter the status quo.
Negative decisions: consciously not changing the status quo.
Non-decisions: decisions where alternatives are not allowed to be discussed.
Emphasis on models to understand how decisions happen (and their limitations).
Core theoretical frameworks for decision making in public policy
The main models discussed: rational choice, bounded rationality, incrementalism, the garbage can model, mixed scanning, and decision accretion.
Distinction between positive and normative theories:
Positive theory: how decisions are made.
Normative theory: how decisions should be made.
Each framework seeks to explain or predict actor behavior and policy outcomes, and each carries strengths and limits when applied to real-world policy environments.
Rational Choice Theory (RCT)
Public policy version: a framework for understanding and modeling decision making where policymakers are viewed as rational actors maximizing utility given information and available alternatives.
Core assumptions:
Actors maximize utility (utility = 2) based on evaluation of costs and benefits of options.
Macro outcomes are the result of individual decisions (micro-foundations).
The emphasis is often on predicting or explaining observed choices as optimal given constraints.
Key sub-frames within RCT:
Game theory: strategic interaction among rational actors; outcomes depend on others’ actions; concepts of interdependence and strategic choice.
Social choice theory: aggregates individual preferences into a collective decision; aims to explain how to produce a coherent group decision from diverse preferences.
Public choice theory: economics applied to political processes; self-interest drives decisions of politicians, bureaucrats, and voters; draws parallels between market incentives and political incentives.
Nash equilibrium (example of game theory):
Definition: a strategy profile is a Nash equilibrium if no participant can benefit by unilaterally changing their strategy while others keep theirs unchanged.
Formal intuition (informal): each actor’s chosen strategy is optimal given the strategies of others.
Notation (illustrative): for a strategy profile s = (s1, s2, …, sn), no player i has an incentive to deviate: ui(si, s{-i}) \c \u2272 ui(s'i, s{-i}) for all feasible s'_i.
Arrow’s Impossibility Theorem (social choice theory):
Claim: No social welfare function can simultaneously satisfy a set of seemingly reasonable conditions (non-dictatorship, unrestricted domain, independence of irrelevant alternatives, Pareto efficiency) without leading to logical inconsistencies or dictatorship.
Implication for policy: aggregating diverse individual preferences into a single social choice is fraught with fundamental trade-offs and potential unfairness or instability.
Public choice theory (politics and bureaucracy):
Core claim: actors in government act in self-interest, not purely for public good; bureaucrats may seek larger budgets, greater power, influence, or job security.
Policy implications: outcomes often reflect bargaining, coalitions, and the pursuit of individual incentives rather than purely public-spirited motives.
Noted scholars/figures: reference to economists/political scientists who emphasize budgets and incentives (lecture cites Nick Sondland in a humorous mispronunciation; historically linked to the idea of budget-maximizing bureaucrats).
Uses and critiques in public policy:
RCT can generate predictive models to explain and anticipate policymaker behavior.
It is instrumental for analyzing how actions maximize utility, given constraints.
Critiques of rational choice in policy:
Unrealistic assumptions: humans are perfectly rational with complete information; decisions occur under uncertainty and cognitive biases.
Empirical limitations: predictions often do not hold in real-world settings; post hoc explanations rather than testable predictions.
Reductionism: oversimplifies social, cultural, and institutional contexts; neglects altruism, social norms, and collective identities.
Cultural/historical specificity: rooted in Western, particularly American, norms; may not generalize.
Methodological concerns: overreliance on deductive, mathematical modeling at expense of empirical inquiry.
Overall takeaway: RCT offers analytical structure, emphasizing instrumental rationality, but faces significant challenges when applied naively to real-world policymaking.
Bounded Rationality (BR)
Origin: Herbert Simon as a critique/alternative to full rationality.
Core idea: decision makers have cognitive limitations that prevent perfect rationality due to complexity, limited information, time constraints, and human cognitive limits.
Key concepts:
Satisficing: actors search for a solution that is good enough, not optimal, given constraints.
Sequential decision making: due to limits on information processing, searches proceed step-by-step and stop when a satisfactory option is found.
Heuristics: cognitive shortcuts used to simplify decision making under constraints.
Limited information processing: actors cannot access or process all information; decisions occur within a bounded environment.
Implications for public policy:
BR recognizes realism: policymakers operate under constraints and typically do not achieve perfect optimization.
Decision processes become more about finding satisfactory options rather than the optimal one.
Heuristics in BR (and how they help, but also bias):
Availability: rely on information that is readily available, recent, or memorable; quick judgments; can bias toward salient events.
Representativeness: judge likelihood by similarity to a prototype or stereotype; can lead to base-rate neglect or pattern-based errors.
Anchoring: initial information sets a reference point; subsequent judgments are biased toward the anchor.
Framing, priming, and loss aversion (within BR context):
Framing: how information is presented changes decision outcomes (gains vs losses, etc.).
Priming: exposing relevant information before a decision prepares the mind to respond in a certain way.
Loss aversion: losses loom larger than gains; people prefer avoiding losses to achieving equivalent gains.
Nudging/ethical framing: using framing to influence decisions toward beneficial policies; must consider ethical implications.
Framing and heuristics examples from the lecture:
Anchoring example: starting budget forecasts influence subsequent adjustments.
Availability framing: highlighting a recent disaster to justify increased security spending.
Representativeness framing: presenting a new health policy as similar to a well-regarded vaccination program to gain support.
Gains vs losses framing: renewable energy framed as protecting children’s future (gain framing) vs catastrophic fossil-fuel outcomes (loss framing).
Priming example: preceding crime data to prime support for tougher measures.
Loss aversion demonstration: a hypothetical flu outbreak with different program choices shows people often prefer options with a guaranteed but smaller benefit over probabilistic but larger potential losses.
Ethical considerations: acknowledge that exploiting heuristics and framing carries ethical risks; nudging should promote genuine public good, not merely manipulate decisions.
Practical takeaway: framing and heuristics help explain and influence policy decisions, but require careful ethical consideration and awareness of cognitive biases.
Mixed Scanning Model (MSM)
Purpose: to combine strengths of rational choice and incrementalism, addressing their weaknesses.
Two-stage process:
1) Pre-decisional (representative) phase: a cursory scan of the policy environment to broadly identify concerns and opportunities without heavy resource use; broad review of issues and alternatives.
2) Detailed probing: after identifying promising options, conduct thorough, systematic analysis of these options using rigorous criteria and tools, gathering expertise.Key features:
Conserves resources by avoiding exhaustive analysis of all options in stage 1.
Maintains flexibility to adjust to new information and changing contexts.
Stage 2 resembles rational analysis (costs, benefits, risks, implications) but is focused on a subset of options.
Strengths: pragmatic, prescriptive and descriptive; reduces risk of overlooking factors while remaining efficient.
Limitations: still a model, and real-world decisions can be complex; requires careful selection of options to move to stage 2.
Practical takeaway: MSM attempts to balance broad exploration with focused, rational evaluation.
Garbage Can Model (GCM)
Radical shift from orderly rational/incremental models to chaos and disorder in decision making.
Core assumptions:
Decision making is inherently ambiguous and unpredictable.
Problems, solutions, and participants flow into decision opportunities like garbage in a can.
Problems are not well-defined; solutions exist looking for problems; participants’ involvement is inconsistent; decisions arise from the interaction of these elements rather than deliberate planning.
Metaphor/illustration:
Garbage can with problems, solutions, participants, and opportunities all inside; when ideas touch, a decision may emerge by chance rather than through planned alignment.
The model highlights timing and luck as crucial determinants of outcomes.
Applications and critique:
Useful for understanding decision making in highly turbulent, complex, or uncertain environments (e.g., universities, large bureaucracies).
Criticized as overly pessimistic and chaotic; may overstate disorder in many settings.
Historical origin: Cohen, March, and Olsen; often applied to organizational contexts (including universities).
Decision Accretion (DA) Model
Concept: decisions accumulate over time through small, incremental decisions made by many actors across an organization, rather than through a single deliberate act.
Key features:
Small incremental decisions contribute to a final outcome without a conscious, centralized plan.
Involves multiple actors across levels; no single actor fully aware that a final decision is being made.
Outcomes emerge as layers accumulate, like pearls forming in an oyster.
Heavily shaped by organizational context (structure, culture, SOPs).
How it differs from incrementalism:
Incrementalism: deliberate, small, intentional changes to reduce risk; centralized/aware decision-making; controlled.
DA: organic, unstructured accumulation; decisions emerge from uncoordinated actions; no single intention guiding the final outcome.
Implications for policy: final policies may be unintended or unforeseen consequences of many small actions across the organization.
Summary contrast:
Incrementalism: deliberate, small changes with some strategic intent; controlled.
DA: emergent, cumulative, often unplanned; decentralized and uncoordinated.
Quick comparison guide: when to apply which model
Howlett and Ramesh framework (low vs high constraints; few vs many actors):
Rational decision making: low constraints, few actors, clear problem definition.
Incrementalism: low constraints, few actors, but with existing policy framework and risk aversion encouraging small tweaks; controlled changes.
Decision Accretion (DA): high number of actors, clear problem, but complexity drives emergent accumulation in a less coordinated way.
Garbage Can Model: high constraints, many actors, high uncertainty; chaotic, opportunistic alignment of problems/solutions/participants.
Summary mapping from the lecture's chart:
Low level of actors + clear problem -> Rational decision making.
High number of actors + clear problem -> DA.
High severity of constraints + low actors -> Incremental decision making.
High constraints + high actors -> Garbage Can Model.
Real-world relevance and context
The professor ties these models to real-world environments, including universities (as an example of a chaotic environment where GCM may apply).
The models illustrate that policy decisions often emerge from a mix of deliberate planning, incremental adjustments, and uncoordinated institutional processes.
The practical takeaway for students: understand which model best captures the decision context you study or work within, and recognize that multiple models may apply at once depending on constraints, actors, and problem definition.
Framing, priming, nudges, and ethical considerations (in BR context)
Framing: presenting information to influence how decisions are perceived (gains vs losses, etc.).
Priming: exposing related information before a decision to prepare the mind to respond in a certain way.
Loss aversion: people fear losses more than they value gains; can drive stronger policy support when losses are emphasized.
Nudging: guiding choices through framing and presentation while preserving freedom of choice; ethical concerns about manipulation vs genuine public good.
Policy framing examples (from lecture):
Gain framing: “protecting our children’s future” for renewable energy.
Loss framing: focusing on avoiding the loss of biodiversity or natural resources.
Anchoring: initial forecasts guide later adjustments; risk of insufficient adjustment from anchor.
Availability: highlighting recent events to influence risk assessments and resource allocation.
Representativeness: framing new programs as similar to well-known successes to gain support.
Illustrative example: framing and risk preferences in public policy
Melbourne flu outbreak thought experiment (illustrative study):
Scenario: outbreak predicted to kill 600 people; two program sets with different risk profiles.
A vs B:
A: 200 people saved (certain gain).
B: one-third chance 600 saved, two-thirds chance 0 saved.
Most participants chose A (certain gain).
C vs D:
C: 400 people die (certain loss).
D: one-third chance 0 die, two-thirds chance 600 die (loss scenario).
Most participants chose D (riskier option with potential large loss, i.e., evaluating the loss domain).
Explanation: risk aversion in the gain domain (A preferred) and risk seeking in the loss domain (D preferred); the study reported 67% choosing A and 67% choosing D, illustrating framing effects and loss aversion.
Takeaway: risk framing interacts with loss aversion; decisions can be steered by how probabilities and outcomes are framed, which is essential in policy communications and design.
Practical implications for exam preparation
Remember key concepts and definitions:
Rational Choice Theory: utility maximization by rational actors; macro outcomes from micro decisions.
Bounded Rationality: cognitive limits lead to satisficing and heuristic-guided decisions; decision environment is bounded.
Heuristics: availability, representativeness, anchoring; each can bias judgment and decision outcomes; framing and nudging can influence choices.
Mixed Scanning: two-stage process combining breadth (pre-decisional scan) and depth (detailed probing).
Garbage Can Model: decision making as chaotic with problems, solutions, participants, and opportunities moving independently; timing and chance matter.
Decision Accretion: decisions accumulate through small, uncoordinated actions across actors; emergent and often unintended outcomes; contrast with Incrementalism.
Incrementalism: deliberate, small changes; centralized decision making; risk reduction.
Be able to identify which model best fits a given policy context based on constraints and number of actors.
Be able to discuss the ethical implications of framing/nudging in public policy.
Be able to articulate core criticisms of each model and understand that real-world decision making often reflects a mix of these models rather than a single framework.
Know the main equations and formal ideas surfaced:
Nash equilibrium concept: concept of strategic stability in game-theoretic contexts. Definition in practice.
Arrow’s impossibility theorem: fundamental incompatibility among reasonable fairness criteria when aggregating individual preferences into a group decision.
Utility maximization basics (conceptual, not a full calculus): for a given action a, choose to maximize utility: ext{maximize } u(a) ext{ over } a ext{ in } A.
Final reminder on exam readiness: understand definitions, apply the right model to the scenario, and critique the model’s applicability and limitations; be prepared to discuss ethical implications of policy framing and nudging.
Quick references and closing notes
The instructor’s closing reflections emphasize the variety of models and the need to choose context-appropriate frameworks.
The discussion also references foundational scholars and frameworks (Howlett & Ramesh; Lindblom; Cohen, March, and Olsen) in shaping decision-making theory in public policy.
Upcoming tutorials: last week before break; engagement with group work and further exploration of these models.
Acknowledgements: light tone and visuals (graphics for the models) used to illustrate concepts; humor about the lecturer’s own design efforts and classroom dynamics.
ext{Nash equilibrium: } orall i, ui(si^,s{-i}^) \c ui(s'i,s{-i}^*) ext{ for all } s'_i ext{ feasible}.
ext{Arrow's impossibility: No social welfare function } W ext{ can satisfy simultaneously:}
\text{(i) unrestricted domain, (ii) non-dictatorship, (iii) Pareto efficiency, (iv) independence of irrelevant alternatives.}
ext{Utility maximization (generic): } \max_{a\in A} u(a).
This collection of notes should serve as a comprehensive study aid, consolidating the key ideas, models, critiques, and real-world implications discussed in the lecture. You can use it to compare models, recall definitions, and apply them to exam-style prompts.