The Transaction-Cost Ledger of Decision Accounting

Decision Accounting lowers transaction cost only for Tier 1 and Tier 2 decisions

The paper answers the Williamson objection by comparing DA with the current cost of fragmented governance, not with zero-cost governance.

• The governed unit is a consequential decision, not every organizational action
• The record is a 17-field record for Tier 1 and an eight-to-ten-field record for Tier 2
• The main offset is verification cost: DA replaces later forensic reconstruction with a structured, sealed record

The objection is that DA makes every governed decision carry a record, review, and gaming scan

The cost side is real: fields must be populated, records maintained, reviews completed, and gaming-detection markers checked.

• The full record includes authority basis, alternatives, criteria, information, reasoning chain, expected and actual consequences, review trail, gaming markers, seal, timestamp, version history
• Costs include labor time, system cost, cognitive load, and delay before proceeding
• If DA were applied to every routine action, the paper says the objection would be valid

DA is scoped by consequence magnitude and reviewability

The paper’s first answer is that DA does not attach to routine decisions. It assigns different record burdens to three tiers.

• Tier 1: strategic or high-consequence decisions with full 17-field record, mandatory conformance review, and cryptographic integrity sealing
• Tier 2: operational but consequential decisions with an eight-to-ten-field record, lighter review, and periodic gaming detection
• Tier 3: routine, low-consequence, deterministic actions with little or no record beyond existing operational logs

In the paper’s 1B firm example, consequential-decision documentation costs 105,000 per year

The paper models a firm with 5,000 employees, $1B revenue, 50 Tier 1 decisions, 500 Tier 2 decisions, and 500,000 Tier 3 decisions per year.

• Tier 1: 50 decisions x four labor-hours x 150 loaded labor cost = 30,000
• Tier 2: 500 decisions x one labor-hour x 150 loaded labor cost = 75,000
• Total consequential-decision documentation cost: $105,000, about 0.01% of revenue

Tier 1 costs 500 to 2,000 per decision; Tier 2 costs 100 to 500

The cost estimate includes marginal labor, infrastructure amortization, and reviewer time. The paper treats these as bounded costs, not as free documentation.

• Tier 1 marginal documentation time is minutes to a few hours, typically less than 5% of total decision time
• Tier 2 marginal documentation time is typically 15 to 30 minutes
• Infrastructure is estimated at 200,000 to 500,000 initial cost and 50,000 to 100,000 annual maintenance

DA targets decisions whose failure is costly, delayed, and hard to reconstruct

The paper’s hidden-failure mechanism is a bad decision that looks successful early, then fails after the decision-maker has moved on.

• Examples named in the paper: Deepwater Horizon, Volkswagen emissions, Boeing 737 MAX, and Theranos
• The missing object in these post-mortems is the WHY field: reasoning, information, alternatives, and authority basis
• The paper’s example: 5% failure probability x 50M failure cost = 2.5M expected loss; a 10% reduction saves $250,000 per decision

The surgical checklist case shows a small record can change outcomes

The paper uses the surgical safety checklist as a controlled existence proof for a minimal DA-like record before proceeding with surgery.

• Mortality fell from 1.5% to 0.8%
• Complications fell from 11% to 7%
• The mechanism is documented verification of key conditions before action

An unpredictable reader changes how the decision-maker writes the record

The conformism break is the paper’s mechanism for improving decision quality before review ever happens.

• The Prat resolution: if the reader may not share the decision-maker’s priors, the decision-maker has reason to widen alternatives
• The record forces disconfirming evidence and reasoning chain into inspectable form
• Cases named in the paper: Bay of Pigs, Vietnam escalation, and the Challenger launch decision

DA turns decision verification from forensic reconstruction into reading

The paper’s Williamson-internal move is that verification of a past decision is itself a transaction, and DA lowers its cost.

• Status quo reconstruction uses emails, meeting notes, presentations, spreadsheets, and memory
• A typical forensic reconstruction of a major corporate decision costs 100,000 to 500,000 and takes weeks to months
• DA produces the record once, at the decision, with structure and cryptographic sealing

One capital allocation decision verified ten times can save 999,500 to 4,998,000

The paper’s example lists ten later readers: board, audit committee, regulator, acquirer, successor, litigant, insurer, credit rating agency, analyst, and journalist.

• Status quo: ten verifications x 100,000 to 500,000 = 1M to 5M
• DA production cost: 500 to 2,000
• Net saving in the paper’s example: 999,500 to 4,998,000

Verifiable governance is estimated to reduce capital cost by 10 to 50 basis points

The paper ties DA to information asymmetry between the firm and capital providers: lenders and investors price uncertainty about decision quality.

• Example debt case: 1B debt at 5%; a 20 basis point reduction saves 2M per year
• Example equity case: 10B market capitalization; a 20 basis point reduction in cost of equity saves 20M per year
• The paper links the claim to governance literature on boards, shareholder rights, and disclosure quality

The theorem is positive only when benefit terms exceed production cost

The paper formalizes DA’s net benefit as a conditional claim, not a universal defense of documentation.

• Net benefit condition: hidden-failure reduction + conformism break + reconstruction saving + cost-of-capital effect > production cost
• Tier 1 benefit estimate: 100,000 to 1M against 500 to 2,000 cost
• Tier 2 benefit estimate: 10,000 to 100,000 against 100 to 500 cost

The paper names the cases where the Williamson objection wins

The defense fails if the empirical ledger turns negative or if organizations apply DA outside its target class.

• Cost dominance: marginal record cost exceeds expected per-decision benefit
• Universal application: DA is imposed on routine Tier 3 activity and total cost exceeds total benefit
• Verification-cost overestimate: forensic reconstruction is cheaper than the paper assumes