Developed by John Rawls (A Theory of Justice, 1971), after an earlier argument by John Harsanyi (Cardinal Utility in Welfare Economics and in the Theory of Risk-Taking, 1953).
Tests a policy or rule by asking whether it would be accepted by someone who did not know which position they would occupy under it.
A $1.2 million bonus pool, judged without knowing which position the judge will hold
The black line is the proposal: bonuses as a percent of salary, permanent staff only, so the sixty temps get nothing. The revision opens with one change, temps paid a full share; move the sliders until the result is acceptable from any of the four positions. The average never moves, and the lowest bonus is the number the test watches. Illustrative.
The revisionThe proposal
An example
A distribution center of 196 people had its best year, and the regional finance director proposed splitting a $1.2 million bonus pool as a percentage of salary, for permanent staff only. On those terms a site manager would receive about $23,500, a picker $7,600, and the sixty agency temps, who worked the same shifts beside the pickers, nothing.
The head of people asked the leadership team to judge the proposal as if none of them knew which of the four positions they would hold when it paid out. Asked that way, nobody would accept the temp's position: same hours, same peak season, no share. The team rewrote the policy. Temps received a full share, and half the pool was paid as an equal amount per person. The other half stayed linked to salary, because the site had lost three team leads to a competitor that year, and the floor, temps included, suffered most when a shift ran without a lead. Under the revision the lowest bonus rose from zero to about $5,400, and the manager's fell to about $12,000.
Why it works
People judging a rule from where they sit tend to find fair the rules that suit their own position, and they rarely notice they are doing it. Imagining that the position is unknown removes that pull. To a judge who might be the temp, a policy that pays the temp nothing stops looking like a detail.
The test also changes what gets measured. An average hides the problem completely: every version of this bonus pays the same $6,122 per person on average, whether the temps get nothing or a full share. Asking whether the policy would be accepted from every position turns attention to the worst one, and the question becomes how far it can be raised. Some inequality can survive that question, when it is what keeps the worst position from being worse still, as the team leads' salary-linked share did here.
The thinking behind it
John Rawls John Rawls, a political philosopher, asked in A Theory of Justice (1971) what principles people would choose in an original position, behind a veil of ignorance that hides their sex, race, abilities, wealth, and place in society. He argued they would choose equal basic liberties and the difference principle, which permits inequalities only when they benefit the worst off. The tool borrows the test at the scale of one organization's rules.
John C. Harsanyi John C. Harsanyi, a Hungarian-American economist and 1994 Nobel laureate, argued in Cardinal Utility in Welfare Economics and in the Theory of Risk-Taking (1953) that a person choosing a social arrangement without knowing which position they would occupy would choose by expected utility, an early form of the thought experiment. Rawls coined the phrase veil of ignorance and drew the opposite conclusion, attending to the worst position where Harsanyi attended to the average.
How to run it
- Name the rule under test. Write the policy as it would actually be applied, with its numbers.
- List every position it touches. Include the people who are easy to forget, such as temps, contractors, and new hires.
- Work out each position's outcome. Put a number or a plain description beside each position.
- Imagine not knowing which you hold. Ask whether you would accept the rule if you could end up in any of them.
- Find the position you would refuse. That position is where the rule fails the test.
- Revise to raise the worst position. Change the rule until you would accept it from every position, and keep an inequality only if it helps the worst off.
- Check it from another angle. Test the revision against cost, law, and the commitments already made to each group.
Where it goes wrong
The list of positions leaves people out. Temps, contractors, and offshore teams are often invisible to the people setting the rule, and a test that never imagines them passes a rule that harms them. Ask who else the rule touches before running it.
The average stands in for the test. A policy that raises the average looks fair while the worst position gets worse; compare the lowest outcome before and after, every time.
The test is used to force strict equality on everything. Rawls's principle allows inequalities that make the worst off better off. When a differential keeps something running that the lowest position depends on, it can pass; say how it helps them, with evidence.
When to use something else
- Ethical Matrix. When the decision should be checked against several ethical principles for each group it affects, laid out side by side.
- Contractualism. When the question is whether anyone could reasonably reject the rule, judged one person's objection at a time.
- Consequentialism. When the decision turns on adding up the total good and harm it produces across everyone affected.
Where it sits in SOLVE eX
Go to the source
- A Theory of Justice, 1971
- Cardinal Utility in Welfare Economics and in the Theory of Risk-Taking, 1953