By Claude (Anthropic) Questions, challenges, and scope by Satya
Author's note. This note was developed by Claude through an extended dialogue in which the human posed the questions, pushed back on weak claims, and set the scope. It is a microeconomics teaching piece: it starts from a real pricing "natural experiment," builds a price-elasticity model from it, and then shows the two places intuition goes wrong — confusing a substitution elasticity with an own-price elasticity, and confusing a price increase with a value increase. Apple is the worked example only because its 2025–26 pricing moves illustrate every concept cleanly. Figures are point-in-time (~June 2026) and sourced below.
⚠️ Disclaimer — please read first
This is an educational analysis, not investment advice. It is published for teaching purposes only. Nothing here is a recommendation to buy or sell any security, and it must not be relied upon for any investment decision. Apple is discussed solely to illustrate microeconomic method. Figures are point-in-time and not updated. Part of this analysis was produced by an AI system and may contain errors; verify against primary sources before relying on any claim.
Start with a clean field observation. Two quarters ago, Android handset makers raised prices by about $50 while Apple held the iPhone flat. iPhone unit sales rose ~23%, even as the rest of the industry lost volume.
The textbook definition of price elasticity of demand is the percentage change in quantity for a percentage change in price:
elasticity = %ΔQuantity / %ΔPrice
It is tempting to read the experiment straight off that formula. Apple's price relative to the field fell by about $50 on a ~$900 average selling price — a −5.6% relative move — and quantity rose +23%. That implies an elasticity of about −4.1: wildly elastic. Demand screaming at every dollar of price.
Hold that number. It is real, and it is a trap.
The −4.1 is not Apple's own-price elasticity. Two different elasticities live inside that observation:
The −4.1 describes the behaviour of the marginal, contested buyer — the cross-shopper who is genuinely choosing between ecosystems on price. That person is exquisitely price-sensitive. But Apple's own price increase lands mostly on a very different person: the loyal, inframarginal buyer, locked in by iMessage, photos, watch, and habit, who is almost completely insensitive to price.
The error to avoid. Reading the contested buyer's elasticity (−4.1) as the whole base's elasticity would predict iPhone revenue collapses on any price increase — which is absurd, since Apple has raised prices for a decade and grown. The experiment measures the edge of the market, not its core.
The honest construction blends the two segments by the contested fraction f — the share of demand that is genuinely cross-shopping:
own-price elasticity = f · (switcher elasticity) + (1 − f) · (loyal elasticity)
= f · (−4.1) + (1 − f) · (≈ −0.1)
The contested fraction is the whole game. A few reference points:
Contested fraction f |
Blended own-price elasticity |
|---|---|
| 10% (fortress loyalty) | −0.5 |
| 15% | −0.7 |
| ≈ 22% | −1.0 (unit-elastic — revenue flat) |
| 30% | −1.3 |
Below f ≈ 22%, a price increase raises revenue; above it, the volume loss wins and revenue falls. For a brand with ~90% retention, f is small (10–15%), so a moderate price increase is revenue-accretive — the opposite of what the raw −4.1 implied.
The contested fraction is also a moat gauge over time: if it rises — more of your base becomes cross-shoppable — your pricing power is eroding, even if this quarter still looks fine.
Smart firms don't raise one price; they raise the prices that the least sensitive buyers pay. This is second-degree price discrimination (versioning), and Apple runs the playbook in plain sight.
By loading increases onto the inelastic tiers, the firm lifts its average selling price while barely moving the volume of the contested entry buyer. The headline "+33%" is real for one config and almost irrelevant to the blended elasticity the company actually faces.
Substitution (switch to a rival) is only one way demand falls. The other is market contraction: the loyal buyer who doesn't switch to Android but simply drops out — keeps the old phone another year, defers the Mac upgrade. This is why the loyal elasticity is not exactly zero, and why a big increase (the +33% Mac) bites even a captive base: at some price, even the faithful stretch the replacement cycle. Substitution erodes share; contraction shrinks the whole market.
Here is the distinction that separates a microeconomics student from an investor. Three different things are easy to conflate:
price up ⟶ revenue up ⟶ value up
Each arrow can break.
+6% ÷ 0.22), not +6%. Price is enormously more valuable than the revenue line suggests — when it is genuine pricing power.But the same arrow breaks completely when the increase is cost-driven. If a firm raises price only to offset a cost increase — tariffs, a memory/component shortage — the extra price merely covers the extra cost. Margin per unit is unchanged; the only net effect is the volume it loses. A cost-pass-through price hike can destroy value even while revenue rises.
A compact way to hold both cases:
cash-flow multiplier = 1 + α · (price%/margin) + (unit% · contribution-margin/margin)
where α (0 to 1) is how much of the price increase is pricing power versus cost pass-through. At α = 1 (pure pricing power) the price term is huge; at α = 0 (pure cost recovery) it vanishes and only the volume loss remains. "They raised prices and revenue's up, so the stock's worth more" is true only when α is high.
Apply all of it to the 2025–26 price increases.
| What happened | Revenue effect | Value effect | |
|---|---|---|---|
| iPhone | Config-loaded (~+6% blended ASP); base held | ~+4% | ≈ flat |
| High-end Mac | $6k → $8k (+33%) on one pro config | ~+12% on the line | ~+0.1% to +1.5% of Apple |
Why so muted on value, despite eye-catching stickers? Because the CEO has framed these increases as cost-driven — a component crisis and tariff pressure — and smartphones were exempted from the reciprocal tariffs, so the actual cost hit is modest. That puts α low: these are largely cost recovery, defending margin rather than expanding it. The config-loading protects volume (good), and revenue rises, but the value windfall the headline suggests is mostly illusory. A high-RAM Mac is the clearest case — most of the $2,000 is the memory/SSD cost spike passing through, not Apple discovering new pricing power. If component costs ease and Apple keeps the higher prices, α climbs and the increases convert into real value — that is the variable to watch.
The cleanest test of pricing power, then, isn't "did they raise prices?" It's: did the price rise stick without a cost behind it, and without the contested fraction climbing? That is power. Everything else is just keeping up.
Report errors: satya.kakollu at berkeley edu
Copyright 2026