Arrow’s Impossibility Theorem

In one sentence

Arrow’s impossibility theorem says that with three or more alternatives, no social choice rule can turn individual preference rankings into a complete and transitive social ranking while simultaneously satisfying a set of appealing fairness axioms.

The setting (what Arrow studied)

Individuals have rankings over alternatives (ordinal preferences).
A “social welfare function” takes those individual rankings and outputs a social ranking.

Arrow asked: can we design such a rule that is fair and coherent for all possible preference profiles?

The key axioms (informal)

Common statements of the theorem use some version of:

Unrestricted domain (universality): the rule works for any pattern of individual rankings.
Pareto (unanimity): if everyone prefers $A$ to $B$, society ranks $A$ above $B$.
Independence of irrelevant alternatives (IIA): society’s ranking of $A$ vs $B$ depends only on individuals’ rankings of $A$ vs $B$, not on other options.
Non-dictatorship: no single voter always determines the social ranking.
Transitivity (coherence): if society prefers $A$ over $B$ and $B$ over $C$, it must prefer $A$ over $C$.

The conclusion (what’s “impossible”)

When there are at least three alternatives, you can’t satisfy all of the above at once. Something has to give:

weaken fairness (e.g., accept a dictator),
restrict preferences (e.g., single-peaked),
relax coherence (allow cycles),
or change the informational basis (use cardinal utilities / interpersonal comparisons).

A concrete intuition: majority cycles

Even simple majority voting can generate cycles.

Example with three voters and three options:

Voter 1: $A \succ B \succ C$
Voter 2: $B \succ C \succ A$
Voter 3: $C \succ A \succ B$

Then a majority prefers $A$ to $B$, $B$ to $C$, and $C$ to $A$.

    flowchart LR
	  A["A"] -->|majority| B["B"]
	  B -->|majority| C["C"]
	  C -->|majority| A

Why it matters in economics

Arrow’s theorem is foundational in:

welfare economics (limits of defining a “social preference” from individuals),
political economy (agenda manipulation, cycling, voting paradoxes),
mechanism design (designing rules under informational constraints).

Social Welfare Function: A rule mapping individual preference orderings into a social ordering.
Condorcet Paradox: Majority voting can yield cycles even when individual preferences are transitive.
IIA (Independence of Irrelevant Alternatives): The social ranking of $A$ vs $B$ should not depend on irrelevant options.
Single-Peaked Preferences: A restricted domain where majority voting can be transitive.
Gibbard–Satterthwaite Theorem: Another impossibility result about strategy-proof voting rules.

Quiz

### Arrow’s impossibility theorem requires at least how many alternatives? - [ ] 2 - [x] 3 - [ ] 4 - [ ] Any number > **Explanation:** The impossibility kicks in with three or more alternatives. ### IIA (Independence of Irrelevant Alternatives) means: - [x] Society’s ranking of $A$ vs $B$ depends only on individuals’ rankings of $A$ vs $B$ - [ ] Everyone must agree on the top choice - [ ] The voting rule must be majority rule - [ ] The social ranking can be intransitive > **Explanation:** Adding/removing a third option shouldn’t change the social $A$ vs $B$ ranking if individual $A$ vs $B$ rankings don’t change. ### Which outcome is typical under majority voting when preferences form a Condorcet cycle? - [x] No transitive social ranking exists that matches pairwise majorities - [ ] A dictator emerges automatically - [ ] Pareto efficiency fails for all profiles - [ ] Every alternative ties > **Explanation:** Cycles imply intransitivity in pairwise majorities. ### True or False: Arrow’s theorem says majority rule is always irrational. - [ ] True - [x] False > **Explanation:** The theorem is about *all* rules and all profiles; majority rule is one example that can cycle. ### One common “escape hatch” that can restore transitive majority outcomes is: - [x] Restricting preferences to be single-peaked - [ ] Forbidding people from voting - [ ] Removing the Pareto condition - [ ] Pricing goods in a market > **Explanation:** Domain restrictions can eliminate cycling. ### The “non-dictatorship” axiom rules out: - [x] Any rule where one voter’s ranking always becomes society’s ranking - [ ] Any rule that uses rankings - [ ] Any rule with more than three options - [ ] Any rule that is transitive > **Explanation:** A dictator would determine social preferences regardless of others’ rankings. ### Arrow’s theorem is a result about: - [x] Aggregating individual rankings into a social ranking for all preference profiles - [ ] Solving for competitive equilibrium prices - [ ] Estimating regression coefficients - [ ] Forecasting inflation > **Explanation:** It is a social choice (collective decision-making) impossibility theorem. ### If there are only two alternatives (A vs B), many “reasonable” voting rules can avoid Arrow-type contradictions because: - [x] Transitivity issues (cycles) do not arise with only two options - [ ] IIA becomes impossible by definition - [ ] Pareto can never hold - [ ] Non-dictatorship cannot be defined > **Explanation:** Cycles require at least three alternatives. ### Which condition is most controversial in practical voting contexts because it ignores “context effects” from other options? - [x] Independence of irrelevant alternatives (IIA) - [ ] Pareto (unanimity) - [ ] Transitivity - [ ] Unrestricted domain > **Explanation:** IIA says the social A vs B comparison should not depend on other choices. ### A standard way to “escape” Arrow’s impossibility while keeping coherent choices is to: - [x] Restrict the domain (e.g., single-peaked preferences) or add more structure to preferences - [ ] Require that everyone has identical preferences always - [ ] Remove all elections and choose randomly - [ ] Force ties in every election > **Explanation:** Additional structure can restore transitivity for majority voting and other rules.