Externality

In general economics, an externality is any cost or benefit imposed on parties outside a transaction - pollution, congestion, network effects. In resource allocation, the term narrows: an Externality is the cost your winning imposes on other participants competing for the same scarce resource.

When you win a scarce resource - a Budget line, an ad slot, engineering capacity - someone else loses it. The Externality payment formula quantifies this precisely:

p_i = (optimal total value for others if i didn't exist) - (total value others actually receive with i present)

This number is always non-negative. If your participation doesn't hurt anyone, you pay nothing. If winning the resource means someone else loses $150K in Expected Value, you pay $150K.

This takes the marginal contribution concept - v(S ∪ {i}) - v(S) - and flips the perspective: instead of asking what do you add to the group?, it asks what do you cost everyone else?

Most internal resource allocation at companies is either political (loudest team wins) or egalitarian (split evenly). Both destroy value.

Political Allocation means teams that exaggerate or have senior sponsors get resources regardless of P&L impact. Egalitarian Allocation ignores that different teams generate wildly different Returns from the same resource.

Externality pricing solves both problems:

• •Truthful reporting becomes a weakly Dominant Strategy. Your payment depends only on others' reported values, so lying about your own can only hurt you or leave you no better off. Teams stop gaming.
• •Resources flow to highest-value use. The team with the highest genuine need wins, because the system makes honesty optimal.
• •The price is fair. You don't pay your full value - you only pay the damage your winning causes. This means winners always retain positive surplus.

For a P&L owner managing cost sharing across business units, shared infrastructure costs, or internal ad slots, this is the difference between resource allocation that creates Profit and allocation that breeds resentment.

The system runs in three steps:

Step 1: Collect valuations. Each participant reports their Utility Function for the resource. Because of how payments work (Step 3), reporting truthfully is optimal - not because you trust people to be honest, but because lying is structurally punished.

Step 2: Allocate optimally. Give the resource to whoever maximizes total reported value. This is straightforward Utility Maximization across all participants.

Step 3: Compute Externality payments. For each winner i:

1. 1)Remove i from the problem entirely
2. 2)Re-solve the optimal Allocation without i
3. 3)Compare: how much total value do the other participants get in the world without i vs. the world with i?
4. 4)The difference is i's payment

Why truth-telling is a weakly Dominant Strategy:

Suppose you value a resource at $200K. Three cases:

Report truthfully ($200K). You win whenever your value is the highest. Your payment is determined entirely by others' reported values - it equals the best outcome others could achieve without you. Since you only win when your value exceeds the runner-up's, your payment is always ≤ your true value. surplus is always ≥ $0.

Over-report ($300K). You now win some cases where you shouldn't. Suppose another bidder values the resource at $250K. Reporting truthfully, you'd lose ($200K < $250K) and earn $0 surplus. By reporting $300K, you win and pay $250K - the damage to the $250K bidder you displaced. Your true value is only $200K, so your surplus is $200K - $250K = -$50K. Over-reporting caused you to win at a loss.

Under-report ($140K). You might lose profitable wins. Suppose the runner-up values the resource at $150K. Reporting truthfully, you'd win and pay $150K, netting $50K in surplus. Reporting $140K means the $150K bidder wins instead. Your surplus drops from $50K to $0.

Truth is weakly dominant: there exist cases where misreporting gives the same result (when the margin is large enough that your lie doesn't change the outcome), but whenever the lie actually changes the outcome, it always hurts you.

Externality pricing fits when all four conditions hold:

1. 1)Scarce shared resources. Multiple teams compete for something you can't just buy more of - analyst time, Budget lines, capacity on a shared platform, ad slots in constrained inventory.

2. 2)Private valuations. Each team knows its own Expected Value from the resource better than you do. If you already know everyone's true ROI, just allocate directly.

3. 3)Strategic behavior is a real risk. If teams will exaggerate needs to grab resources, you need incentives that make honesty optimal. Environments with high-stakes Vendor Negotiations or internal competition are prime candidates.

4. 4)You can enforce payments. The system requires that winners actually bear the cost - whether through Budget transfers, Cost Center charges, or reduced Allocation elsewhere.

Don't use it when resources are abundant (no scarcity, no Externality), when you have perfect information about values (just decide directly), or when the overhead of running the process exceeds the value of better Allocation.

One important limitation: total Externality payments are typically less than the total value created, and may fall short of the resource's actual operating cost. If Budget balance matters - if you need payments to fully fund the resource - this system alone won't guarantee it.

Shapley value and Externality pricing both compare outcomes with and without a participant, but they solve different problems. Shapley value averages marginal contribution across every possible ordering of a group to produce a fair cost sharing split - use it when cooperating teams share infrastructure and you need an equitable bill. Externality pricing computes the single payment that makes truthful reporting a weakly Dominant Strategy - use it when competing teams bid for scarce capacity and you need honest valuations to drive Efficient Allocation. In a single-item auction, the Externality payment reduces to 'winner pays the runner-up's bid.' The winner displaces exactly one other bidder, so the payment equals that bidder's reported value. This is why single-item auctions with runner-up pricing produce truthful bidding without Bid Shading - they are a special case of this same formula. A key limitation connects Externality pricing to Budget decisions: total payments are always ≤ total value created, and often substantially less. If the scarce resource has a real operating cost (analyst salary, server capacity, ad slots maintenance), the payments may not cover it. Operators face a trade-off between honest reporting and full cost recovery - strengthening one weakens the other. Each participant's Outside Option of not participating is never preferable to honest participation, because payments never exceed reported values under truthful reporting. No team needs to be compelled to join - the surplus guarantee is self-enforcing.