Comparative Advantage

Activate when: user says 'should I do this myself or delegate,' 'I'm the best at X so I should do it,' 'should we outsource or build in-house,' 'how should we divide work on the team,' 'is it worth hiring someone for this.' Do NOT activate when: the situation is genuinely single-task with no trade-off; the user is asking about competitive strategy between companies (use Porter's Five Forces instead).

Overview

Mutually beneficial specialization depends on relative productivity, not absolute. Even if you are better than everyone else at every task, you gain by focusing on tasks where your opportunity-cost ratio is lowest and delegating the rest. The right question is not "am I better at X than them?" but "am I better at X relative to Y compared to how they are?"

Proved by David Ricardo (1817); called "the only proposition in economics that is both true and non-obvious" by Paul Samuelson (1969).

Composes with opportunity-cost, pareto-principle, circle-of-competence, theory-of-constraints.

When to Use

• Founder time allocation across activities; team division of labor
• Build vs. buy / in-house vs. outsource decisions
• M&A decisions; international operations / supply chain design
• Personal career or skill specialization decisions

Not when: situation is genuinely single-task (no B to trade off); transaction costs swamp any gains; alternative producer's capacity is fully constrained.

Coaching Novices (Adaptive Front Door)

• Engine mode: user has a concrete allocation decision → run The Process directly.
• Coach mode: user is unfamiliar or has no concrete case → guide step by step.

In Coach mode, respond one step at a time. Each [WAIT] is a hard stop — output only that step's question, then stop.

1. One-line: the right test is not "am I better than them at this?" but "am I better at this relative to what else I could be doing?"
2. Check fit: if single-task or transaction costs very high, analysis adds limited value.
3. Elicit the case: what are you considering doing? What else could you do with the time? Who else could do it?

[WAIT — do not advance until user responds]

4. One question at a time: what's your opportunity cost of the focal task? What's the alternative producer's OC? Where do ratios diverge? Are transaction costs material?

[WAIT — do not advance until user responds]

5. Close: specialization recommendation + transaction-cost caveat + monitoring plan.

[WAIT — do not advance until user responds]

The Process

Step 1 — Producers and tasks: name Producer 1, Producer 2, Task A, Task B, time horizon.

Step 2 — Productivity: estimate value-per-hour (or units/time) for each producer at each task.

Step 3 — Opportunity-cost ratios: Producer 1's OC of Task A = Task B output forgone per unit of Task A. Repeat for Producer 2. Lower OC = comparative advantage in that task.

Step 4 — Specialization: each producer focuses on their comparative-advantage task. Total output rises.

Step 5 — Transaction costs: add coordination, communication, quality-risk, and information-asymmetry costs. If net benefit positive: specialize. If net negative: produce in-house.

Step 6 — Implement and monitor: document arrangement, set quality monitoring, schedule re-evaluation as productivity ratios or transaction costs shift.

Output Template

markdown

# Comparative-Advantage Analysis: <decision>
Producers: P1 = ___ | P2 = ___   Tasks: A = ___ | B = ___
OC of Task A — P1: ___ | P2: ___   → CA in A: P___ | CA in B: P___
Transaction costs: coordination ___ | quality risk ___ | net ___
Recommendation: P1 specializes in ___ | P2 specializes in ___
Caveats / re-eval trigger: ___

→ Method in Action: Ricardo 1817 England-Portugal + Modern Applications

Pack: Application Patterns

Decision	Absolute-advantage trap	Comparative-advantage answer
Founder time	"I'm best at everything"	Specialize in 2-3 highest-leverage activities; delegate the rest
Senior engineer	"I should code everything"	Architect, mentor, hard problems; delegate routine tasks
Manufacturing	"We could build our own factory"	Outsource to specialist; focus on design/marketing/sales
Customer support	"I know the product best"	Hire CS specialists; handle only strategic relationships yourself

Applying It Well

• Compute opportunity-cost ratios, not absolute productivity
• Transaction costs are real — factor coordination, communication, quality-monitoring
• Revisit as productivity ratios shift or transaction costs change

→ Primary sources: references/sources.md

Common Rationalizations

[D] = designed upfront | [O] = observed in real use. [O] entries are more valuable.

Fake move	Reality
[D] "I'm best at this, so I should do it"	Right question is OC ratio. You can be best at it and still not be the one who should do it.
[D] "I can do it faster than they can"	An hour spent doing it is an hour not spent on higher-OC work. Calculus often favors delegation.
[D] "Quality will drop if I delegate"	Compute: does the quality drop outweigh the OC gain from focusing on higher-CA activities? Often no.
[D] "Outsourcing is risky"	Transaction costs and quality risk are part of the calculus — compute them, don't assume.
[D] "Comparative advantage is academic; real life is different"	The framework underpins modern supply chains, team specialization, and founder time-allocation.
→ Add [O] entries here after each real use — paste the actual failure pattern	What went wrong and why

Red Flags

• Founder doing many tasks personally despite team members who could do them
• "I'm best at X" used to justify allocation instead of OC analysis
• Outsourcing decisions based on absolute productivity rather than opportunity cost
• Transaction costs ignored entirely or used as a blanket excuse

Verification

• Both producers and both tasks explicitly specified
• Productivity or value-per-hour estimated for each combination
• Opportunity-cost ratios computed; comparative-advantage assignment determined
• Transaction costs considered; net benefit computed
• Decision documented; re-evaluation conditions specified

---

Part of deciqAI Knowledge Skills — open-source thinking skills that make rigor executable for AI agents. Built by deciqAI · https://deciqai.com · Contributions welcome — see the template at the repo root.