Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Analyze market structures across perfect competition, monopolistic competition, oligopoly, and monopoly to predict firm behavior and market outcomes. Use this skill when the user needs to classify a market's competitive structure, predict pricing behavior, evaluate antitrust implications, or understand why an industry behaves the way it does — even if they say 'why can they charge so much', 'is this market competitive', or 'will prices come down'.
.claude/skills/asgard-ai-platform-econ-market-structure/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-13 | ✗→✓ | ▲ Improved | -12% | 0% |
| case-01 | ✗→✓ | ▲ Improved | -44% | 0% |
| case-03 | ✗→✓ | ▲ Improved | -35% | 0% |
| case-17 | ✗→✓ | ▲ Improved | -27% | 0% |
| case-19 | ✗→✓ | ▲ Improved | -28% | 0% |
Market structure determines how firms compete, set prices, and earn profits. The four structures — perfect competition, monopolistic competition, oligopoly, monopoly — predict increasingly different behaviors as concentration rises and differentiation increases.
IRON LAW: Structure Determines Behavior, Not Vice Versa
Classify the market structure FIRST based on structural characteristics
(number of firms, barriers, differentiation), THEN predict behavior.
"This company charges high prices" does not mean it's a monopoly —
high prices can occur in oligopolies and even monopolistic competition.| Feature | Perfect Competition | Monopolistic Competition | Oligopoly | Monopoly | |---------|-------------------|------------------------|-----------|----------| | Firms | Very many | Many | Few | One | | Product | Homogeneous | Differentiated | Homogeneous or differentiated | Unique, no close substitutes | | Entry barriers | None | Low | High | Very high | | Price power | None (price taker) | Some (limited by substitutes) | Significant (interdependent) | Full (price maker) | | Long-run profit | Zero (economic) | Zero (economic) | Positive possible | Positive | | Examples | Agricultural commodities, forex | Restaurants, clothing | Airlines, telecom, auto | Utilities, patents |
Perfect Competition: Price = marginal cost. Firms are price takers. No advertising needed. Long-run economic profit = 0.
Monopolistic Competition: Short-run profits possible through differentiation. Long-run: entry erodes profits to zero. Firms compete on brand, quality, location.
Oligopoly: Firms are interdependent — each watches rivals' moves. Game theory applies. May collude (tacitly or explicitly). Kinked demand curve or Cournot/Bertrand models.
Monopoly: Price > marginal cost. Deadweight loss exists. May be regulated (utilities) or temporary (patents). Natural monopolies occur when average costs decline with scale.
markdown# Market Structure Analysis: {Industry} ## Classification - Structure: {type} - Evidence: - Number of firms: ... - Product differentiation: ... - Entry barriers: ... - Interdependence: ... ## Predicted Behavior - Pricing: {price-taking / markup / strategic} - Long-run profit: {zero / positive} - Competition type: {price / quality / advertising / innovation} ## Policy Implications {Antitrust concerns, regulation needs, consumer impact}
Scenario: Taiwan's telecom market
references/oligopoly-models.md| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-11 | pass→pass | 17,342 | 10,833 | -38% | 1 | 1 | 0% | 2,721 | 2,726 | +0% | 0 | 0 | — |
case-12 | pass→pass | 20,544 | 8,962 | -56% | 1 | 1 | 0% | 3,185 | 2,360 | -26% | 0 | 0 | — |
case-13 | fail→pass | 26,950 | 15,819 | -41% | 1 | 1 | 0% | 3,999 | 3,505 | -12% | 0 | 0 | — |
case-14 | fail→fail | 23,056 | 15,507 | -33% | 1 | 1 | 0% | 3,683 | 3,515 | -5% | 0 | 0 | — |
case-01 | fail→pass | 51,943 | 22,006 | -58% | 1 | 1 | 0% | 7,281 | 4,043 | -44% | 0 | 0 | — |
case-02 | fail→fail | 30,607 | 12,196 | -60% | 1 | 1 | 0% | 4,715 | 3,083 | -35% | 0 | 0 | — |
case-03 | fail→pass | 34,048 | 18,474 | -46% | 1 | 1 | 0% | 5,559 | 3,608 | -35% | 0 | 0 | — |
case-04 | pass→pass | 5,597 | 4,128 | -26% | 1 | 1 | 0% | 901 | 1,633 | +81% | 0 | 0 | — |
case-05 | pass→pass | 14,053 | 17,282 | +23% | 1 | 1 | 0% | 2,161 | 4,038 | +87% | 0 | 0 | — |
case-06 | pass→pass | 9,408 | 8,526 | -9% | 1 | 1 | 0% | 2,150 | 2,838 | +32% | 0 | 0 | — |
case-07 | fail→fail | 12,864 | 12,758 | -1% | 1 | 1 | 0% | 2,094 | 2,865 | +37% | 0 | 0 | — |
case-08 | pass→pass | 19,710 | 18,619 | -6% | 1 | 1 | 0% | 2,941 | 4,013 | +36% | 0 | 0 | — |
case-09 | pass→pass | 17,018 | 9,025 | -47% | 1 | 1 | 0% | 2,542 | 2,359 | -7% | 0 | 0 | — |
case-10 | pass→fail | 21,213 | 7,832 | -63% | 1 | 1 | 0% | 2,950 | 2,151 | -27% | 0 | 0 | — |
case-15 | pass→pass | 13,400 | 12,888 | -4% | 1 | 1 | 0% | 2,047 | 2,769 | +35% | 0 | 0 | — |
case-16 | pass→pass | 25,453 | 22,722 | -11% | 1 | 1 | 0% | 3,521 | 3,933 | +12% | 0 | 0 | — |
case-17 | fail→pass | 28,755 | 14,440 | -50% | 1 | 1 | 0% | 3,997 | 2,903 | -27% | 0 | 0 | — |
case-18 | fail→fail | 30,966 | 17,215 | -44% | 1 | 1 | 0% | 4,346 | 3,606 | -17% | 0 | 0 | — |
case-19 | fail→pass | 20,617 | 9,517 | -54% | 1 | 1 | 0% | 3,291 | 2,382 | -28% | 0 | 0 | — |
case-20 | pass→pass | 17,873 | 12,582 | -30% | 1 | 1 | 0% | 2,863 | 3,060 | +7% | 0 | 0 | — |
case-21 | fail→pass | 19,691 | 14,693 | -25% | 1 | 1 | 0% | 2,885 | 3,323 | +15% | 0 | 0 | — |
case-22 | pass→pass | 22,902 | 18,203 | -21% | 1 | 1 | 0% | 3,564 | 3,795 | +6% | 0 | 0 | — |
case-23 | pass→pass | 20,566 | 11,436 | -44% | 1 | 1 | 0% | 3,289 | 2,938 | -11% | 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. 23 cases were attempted. The headline lift of +22 percentage points is the difference between those two pass rates over the 23 comparable cases. 1 case got worse with the skill loaded, and it is included in that figure.
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.