Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Apply Transaction Cost Economics (Williamson, 1975, 1985) to analyze governance structure choices — market, hybrid, or hierarchy — based on transaction characteristics. Use this skill when the user needs to decide make-or-buy, evaluate outsourcing vs vertical integration, design governance mechanisms for inter-firm relationships, or when they ask 'should we build this in-house or outsource', 'why do firms vertically integrate', or 'how should we structure this partnership'.
.claude/skills/asgard-ai-platform-grad-strat-tce/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | -10% | 0% |
| case-10 | ✗→✓ | ▲ Improved | 1% | 0% |
| case-13 | ✗→✓ | ▲ Improved | -22% | 0% |
| case-03 | ✓→✓ | = Same ✓ | 42% | 0% |
| case-04 | ✓→✓ | = Same ✓ | 1% | 0% |
TCE explains firm boundaries by analyzing the costs of conducting transactions. Williamson (1975, 1985) argues that governance structures (market, hybrid, hierarchy) are chosen to minimize the sum of production costs and transaction costs, given bounded rationality and opportunism.
IRON LAW: Governance choice depends on transaction characteristics
(asset specificity, uncertainty, frequency) — NOT on firm preference,
tradition, or ideology. A misaligned governance structure will
generate avoidable costs and eventual correction.Key assumptions:
| Dimension | Low | High | |-----------|-----|------| | Asset Specificity | Generic assets, easy redeployment | Dedicated/specialized assets, lock-in risk | | Uncertainty | Predictable outcomes, easy contracting | Unpredictable, incomplete contracts | | Frequency | One-off or rare | Recurrent transactions |
markdown## TCE Governance Analysis: [Context] ### Transaction Profile | Dimension | Assessment | Evidence | |-----------|-----------|----------| | Asset Specificity | [low/moderate/high] — [type] | ... | | Uncertainty | [low/moderate/high] | ... | | Frequency | [low/moderate/high] | ... | ### Governance Recommendation - Recommended structure: [market/hybrid/hierarchy] - Rationale: ... - Required safeguards: ... ### Risk of Misalignment - Current governance: [structure] - Predicted inefficiency: ...
A chip manufacturer assesses outsourcing a custom fabrication process: high physical asset specificity (dedicated equipment), high uncertainty (rapid tech change), high frequency — TCE prescribes hierarchy (in-house fabrication) with clear reasoning per dimension.
Recommending outsourcing "because it's cheaper" without assessing asset specificity or opportunism risk. TCE requires evaluating transaction characteristics, not just spot-price comparisons.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 26,760 | 48,725 | +82% | 1 | 1 | 0% | 4,212 | 3,776 | -10% | 0 | 0 | — |
case-02 | fail→fail | 22,438 | 14,148 | -37% | 1 | 1 | 0% | 3,567 | 3,074 | -14% | 0 | 0 | — |
case-03 | pass→pass | 9,216 | 8,871 | -4% | 1 | 1 | 0% | 1,849 | 2,626 | +42% | 0 | 0 | — |
case-04 | pass→pass | 30,499 | 31,036 | +2% | 1 | 1 | 0% | 5,454 | 5,500 | +1% | 0 | 0 | — |
case-05 | pass→pass | 15,222 | 18,249 | +20% | 1 | 1 | 0% | 3,057 | 4,424 | +45% | 0 | 0 | — |
case-06 | pass→pass | 24,807 | 15,131 | -39% | 1 | 1 | 0% | 3,766 | 3,159 | -16% | 0 | 0 | — |
case-07 | pass→pass | 15,703 | 9,635 | -39% | 1 | 1 | 0% | 2,581 | 2,335 | -10% | 0 | 0 | — |
case-08 | pass→pass | 17,351 | 8,882 | -49% | 1 | 1 | 0% | 2,795 | 2,281 | -18% | 0 | 0 | — |
case-09 | pass→pass | 20,014 | 12,364 | -38% | 1 | 1 | 0% | 2,979 | 2,737 | -8% | 0 | 0 | — |
case-10 | fail→pass | 18,728 | 12,260 | -35% | 1 | 1 | 0% | 2,818 | 2,842 | +1% | 0 | 0 | — |
case-11 | pass→pass | 16,072 | 10,604 | -34% | 1 | 1 | 0% | 2,429 | 2,529 | +4% | 0 | 0 | — |
case-12 | pass→pass | 20,758 | 11,741 | -43% | 1 | 1 | 0% | 3,219 | 2,774 | -14% | 0 | 0 | — |
case-13 | fail→pass | 19,297 | 8,461 | -56% | 1 | 1 | 0% | 2,892 | 2,263 | -22% | 0 | 0 | — |
case-14 | pass→pass | 13,710 | 32,093 | +134% | 1 | 1 | 0% | 2,058 | 2,748 | +34% | 0 | 0 | — |
case-15 | pass→pass | 14,874 | 12,972 | -13% | 1 | 1 | 0% | 2,402 | 2,868 | +19% | 0 | 0 | — |
case-16 | pass→pass | 17,871 | 10,850 | -39% | 1 | 1 | 0% | 2,662 | 2,376 | -11% | 0 | 0 | — |
case-17 | pass→pass | 17,061 | 11,384 | -33% | 1 | 1 | 0% | 2,350 | 2,718 | +16% | 0 | 0 | — |
case-18 | pass→pass | 20,262 | 18,389 | -9% | 1 | 1 | 0% | 2,949 | 3,387 | +15% | 0 | 0 | — |
case-19 | pass→pass | 27,422 | 12,094 | -56% | 1 | 1 | 0% | 4,253 | 2,886 | -32% | 0 | 0 | — |
case-20 | pass→pass | 24,197 | 12,541 | -48% | 1 | 1 | 0% | 3,981 | 2,752 | -31% | 0 | 0 | — |
case-21 | pass→pass | 17,337 | 18,646 | +8% | 1 | 1 | 0% | 2,734 | 3,130 | +14% | 0 | 0 | — |
case-22 | pass→pass | 19,320 | 12,403 | -36% | 1 | 1 | 0% | 2,627 | 2,746 | +5% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted. The headline lift of +14 percentage points is the difference between those two pass rates over the 22 comparable cases.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.