Kumu / Micro/Macro Econ / Imperfect Competition

Imperfect Competition — Pricing Power Analysis

You are advising the seed company — a real firm, deliberately unnamed — which sells two products out of the same factory. One is commodity corn seed, where it is one supplier among many and takes the market price. The other is patented GMO seed, where it faces the entire market's demand curve alone. Same crop, same $130M of fixed costs — and roughly an 83-fold difference in profit.

Your job is to model both markets, explain why the gap exists, put a dollar figure on what society pays for it, and take a position on whether that price is worth paying.

market power · MR = MC 2 stages · spec-driven build Optional · ungraded · self-paced

Optional · ungraded · self-paced

This engagement is optional. Case 1 — Perfect Competition — is the graded case this term; this one is here for anyone who wants to keep going. Nothing is submitted, nothing is due, and you work it at your own pace. The stages, the repo, and the method are identical — only the obligation is gone.

The economics, briefly

In the first case you advised a price taker: the market handed it a price and the only question was how much. Here the question changes shape. A firm facing the whole market's downward-sloping demand does not get to sell one more unit at the going price — it has to cut the price to move that unit, and the cut applies to every unit it sells. So the revenue from one more sale is less than the price of that sale. For linear demand the arithmetic is exact: marginal revenue falls at twice the slope of demand.

That single fact rewrites the decision rule. Produce while marginal revenue exceeds marginal cost, stop where they meet — and then read the price off demand, because demand is what tells you the most buyers will pay for the quantity you chose. Firms that read the price off marginal revenue instead price themselves at about a quarter of what the market would have borne, which is the most common error in this case and one an AI will make confidently if your specification is vague about it.

The two markets

Non-GMO corn seed (price taker)GMO corn seed (patent monopoly)
DemandFlat at P = $120/bag — one seller among manyP = 525 − 0.0000067·Q — the firm is the market
CostsBoth: TVC = a·Q + b·Q² → MC = a + 2b·Q, with a = $1; fixed costs $130M. GMO's b is 15× smaller — the biotech scales.
RuleP = MCMR = MC, then price off demand
Real anchor~10M non-GMO acres ÷ 2.5 acres/bag ≈ 4M bags~86M GMO acres ≈ 34.4M bags at ≈ $270 — the model lands within 1%

Grounded in validated facts: GMO corn seed really runs $250–305 a bag against ~$85–150 conventional; the seed company's traits covered ~80% of U.S. corn and 90%+ of soybean acres; and the 2018 acquisition that folded it into a major agrochemical company required the largest antitrust divestiture in U.S. history — about $9B of assets sold to a competitor. Sources are in the project README.

Why the company is not named

The firm is real, and its identity and every source are in the project README — one click, openly signposted. The scenario withholds the name because you are being asked to advise this company, not to report on it. Once you know who it is, the model you built stops being the source of truth: there is a real number one prompt away, produced from parameters that are not yours, and comparing your $2.99B against it is not a check — it is a substitution. Every case on this site follows the same rule. The market is named, quantified, and cited. The protagonist is not.

What you'll be able to do afterwards

How the engagement runs

Two stages, same rhythm as the first case: specify the model before it exists, have AI build it from your spec, audit what comes back — then explain what it means and recommend something.

0 of 2 stages complete

The lab that makes the welfare geometry visible

Econ Policy Lab

The Econ Policy Lab draws consumer surplus, producer surplus, and deadweight loss as areas you can watch move. This case's hardest number — the roughly $3 billion a year of surplus that simply vanishes — is one of those areas. Open the lab, impose a distortion, and watch the triangle appear; then go find the same triangle in your own model. It is also the independent cross-check for your Stage 1 audit, the way the Farm Profit Lab was in the first case.

Where AI fits, and where it doesn't

Same line as every engagement this term, applied to this case's artifacts:

Good uses — log themYours alone
Explaining MR, MC, markup, and deadweight loss until they click; quizzing youThe hypothesis, written before you model anything
Building the workbook from your committed spec; debugging what it returnsWriting the spec — and auditing the result against it
Attacking your draft argument on the patent tradeoff and finding its weakest claimWriting the analysis, the memo, and the reflection

Every AI-supplied number is a draft until you have checked it — and in this case there is a specific thing to check for. A model that has not been told marginal revenue is its own series will happily build you a price taker.