Front-end costs imply coarseness.
Even arbitrarily small ex ante costs of making legal distinctions can make every optimal evidence structure finite. The result does not rely on complete contracts being impossible or infinitely costly.
Corrao, Flynn, Sastry · NBER WP 34379 · October 2025 · DOI 10.3386/w34379
Why do contracts stay vague when precision is possible?
The paper's answer is not that complete contracts are impossible. It is that the timing of evidence-design costs matters: ex ante drafting and codification costs make finite, coarse contracts optimal; ex post proof-generation costs alone do not.
Front-end
fewer tiers
Costs paid before action: drafting, defining, formalizing distinctions.
Back-end
lower levels
Costs paid after action: observing, proving, certifying evidence.
Core contrast
timing is the hinge, not cost size
Procurement application
Evidence-cost lab
Predicted tiers
4
Formula
1 + β² / 12ακf
menu tiers
continuum benchmark
legally equivalent actions
Stylized rendering of the procurement example, not an empirical calibration. In the paper's closed-form case, front-end cost changes the number of tiers; constant back-end cost shifts payments without changing tier count.
Application formula
In the procurement application, the optimal number of effort tiers lies within one of 1 + beta^2 / (12 alpha kappa_f). This formula is not the general theorem; it depends on the paper's quadratic effort, uniform type, linear payoff, and constant-cost assumptions.
Result map
What the paper proves
Back-end costs alone do not.
Pure ex post proof costs can distort the allocation, but under the paper's conditions they do not make an optimal contract finite.
With both costs, front-end timing drives form.
When the modified virtual surplus has the required shape, optimal menus remain coarse once front-end costs are present.
Procurement becomes a tiering problem.
The worked example produces discrete effort or quality grades, with front-end costs determining the number of grades and back-end costs moving the payment schedule.
Mechanism
Evidence creates contractibility
1Evidence design
The principal chooses what distinctions can be proved.
2Contract menu
Promises and payments are committed.
3Private type
The agent selects a menu item.
4Action
Evidence is generated by behavior.
5Arbitration
Punishment is possible only when breach can be proved.
Review
Interpretation and caveats
Most important contribution
The paper separates "contracts are costly" from the stronger claim that front-end codification costs are the specific force that makes finite, coarse contracts optimal.
Best intuition
When a contract is already precise, one more distinction produces small incremental surplus gains. But drafting that distinction still creates a nontrivial ex ante comparison burden.
Main caution
The model abstracts from renegotiation, relational enforcement, bounded damages, multidimensional evidence, bargaining, and many institutional details that shape real contracts.
Public takeaway
Coarseness can be an optimal design response, not just a legal failure. But the paper does not imply every vague clause is efficient.
Sources
Review basis
Primary paper
NBER Working Paper 34379
DOI
10.3386/w34379
Version context
February 2025 hosted PDF
Local review log
`research/papers/contractibility-design/deep-review.md`
The NBER page states that working papers are circulated for discussion and comment and are not peer reviewed NBER publications. This summary is an independent reading aid.