Industry Structure: Porter & Value Chains

Phase 4 · Week 38 · ~20 hours

Where it sits. Third module of the strategy-and-moats phase. The first taught you to name a company's charging machine; the second taught you to weigh its smallest repeatable unit. Both modules stared at the company. Now the eyes lift to the neighborhood, because the single most reliable thing an analyst can know about a company before reading a word of its filings is what industry it lives in. Some industries hand their members high returns on capital almost regardless of execution; others grind even brilliant operators down to their cost of capital or below. Michael Porter's two instruments, the Five Forces and the value chain, are the standard equipment for reading that neighborhood, and the treatment here follows the way working analysts actually use them: as questions answered with filing evidence, scored, and converted into a prediction about industry ROIC, never as a five-box diagram recited from memory. You will run the full apparatus end-to-end on a structurally blessed industry (Indian decorative paints and adhesives) and a structurally cursed one (US airlines), map who captures the profit in a full value chain (smartphones, supplier to customer), learn the three generic strategies and the "stuck in the middle" trap, and learn to define the industry correctly in the first place, because every industry analysis lives or dies on that. The next two then narrow from industry structure to firm-level defenses (the five moat sources, Greenwald's collapse of the forces into entry barriers, Helmer's 7 Powers); the one after them adds the time dimension (capital cycles, disruption). In Phase 8's teardown, all of it becomes question 5 ("who competes and who has the right to win") answered in about ten minutes.

Suggested hour budget. Core content: 8h · worked examples: 2h · practice set: 4h · applied mini-project (the Industry Map, this module's gate): 4h · flashcards + mastery check: 2h.

App link. This module seeds the Industry & Business Model branch: the five-forces scorer (evidence packets in, force scores and justifications out, graded against ground truth), the profit-pool allocator (drag revenue and margin blocks along a chain until the pool adds up), the industry-definition cases ("same evidence, two definitions: does the verdict flip?"), and the rubric-graded Industry Map submission that is this week's gate.

Learning objectives

You will be able to:

1. State fairly what the evidence says about how much industry structure explains. The variance-decomposition findings (Schmalensee, Rumelt, McGahan & Porter), the persistent spread in industry-level ROIC, and the correct analyst conclusion: firm effects are larger, but industry is the strongest ex ante base rate and sets the distribution a firm draws from.
2. Run each of the Five Forces as a question, not a checklist item. For rivalry, entry threat, substitutes, buyer power and supplier power, name the structural determinants that drive it and the specific filing and dataset evidence (segment notes, ≥10%-customer disclosures, DSO/DPO trends, gross-margin pass-through, capacity and share tables, risk-factor language) an analyst uses to score it 1–5.
3. Compute and interpret concentration measures, CR4 and HHI (with the DOJ/FTC thresholds), and explain why concentration is an input to a rivalry judgment, never the verdict itself, and why the relevant market (route vs national, regional vs national, segment vs sector) changes the number.
4. Run a complete five-forces analysis end-to-end on a real industry, producing a scored scorecard, a verdict on expected industry ROIC vs WACC, the binding force, and the structural watch items, as demonstrated on Indian decorative paints and US airlines with the profit-pool numbers that make the contrast concrete.
5. Classify a firm's competitive posture into Porter's three generic strategies (cost leadership, differentiation, focus), diagnose "stuck in the middle" from margin and share data, and state honestly where the framework's trade-off claim holds and where dual advantages have genuinely existed.
6. Draw Porter's value chain (five primary, four support activities) for any business and use it two ways: to locate where cost and willingness-to-pay are actually created, and to compare two rivals' chains activity by activity.
7. Map a profit pool across a full supplier→firm→customer chain. Estimate each link's revenue and operating margin from pure-play filings, compute each link's share of chain profit, explain why the pool sits where it sits (the forces, run per link), and say which way it is migrating, as demonstrated on the smartphone chain.
8. Define the industry correctly before analyzing it. Apply demand-side and supply-side substitution tests, avoid the too-broad and too-narrow traps, and use the "run it twice" discipline when the boundary is genuinely uncertain.

Prerequisites & connections

Builds on. M4.01: you drew the value chain from input to cash (section 4.2 there) for one company. Here you industrialize it, drawing the same chain across every participant, with a margin attached to each seat. M4.02: contribution margins and operating leverage come back as structural facts about whole industries. An industry of high-fixed-cost firms selling a commodity behaves differently from an industry of variable-cost firms selling differentiated goods. M2.01–M2.03: margins, DuPont, and the blank-sheet ROIC build. Every force score you write is in the end a claim about someone's ROIC, and you'll keep checking those claims against computed returns. M2.05: benchmarking, because comparing a company with the right peer set is the industry-definition problem, formalized. M3.03: the WACC hurdle (~12% INR for India, ~8–9% USD for the US, illustrative as of mid-2026, check against your own builds) is the line every industry verdict is measured against. M3.05: reverse-DCF thinking comes back as "what industry structure is this multiple assuming will persist?"

Feeds forward. M4.04–M4.05 are the firm-level sequel. Once you can say this industry's average member earns X, the moat question becomes why does this particular member earn more, and what protects that? Greenwald's move, collapsing the five forces into the single question of entry barriers, arrives in M4.05 and will feel inevitable by then. M4.06 adds dynamics: the capital cycle is what happens to your force scores when an investment boom hits. Phase 5's seventeen sector playbooks run this same apparatus with sector-specific instruments, and the airline structural anatomy you build here becomes the foundation under the Phase 5 airline playbook's RASK/CASK tooling. Phase 8's teardown question 5 and the Day-2 industry map of the deep dive are this work under a stopwatch. Competency C4 (assess competitive advantage) and C3 (business model cold) both draw on the gate skill certified here.

The short version. An industry's structure sets the base rate of returns for everyone in it. That structure is how rivals fight, how easily outsiders enter, what else does the customer's job, and who upstream and downstream holds the pricing pen. Your job is to score those forces from filing evidence and find the profit pool in the chain. Only then do you ask whether a particular firm can beat its neighborhood.

Core teaching content

4.1 The evidence: how much does the neighborhood matter?

Industry structure determines profitability more than management quality does. You will hear that claim in every strategy course. It is repeated often enough that you should immediately want the evidence, and the evidence turns out to be more interesting than the slogan.

Start with the spread. It's real, and it's enormous. Porter's 2008 update of his framework ("The Five Competitive Forces That Shape Strategy," HBR, January 2008) carries an exhibit worth keeping in your head for good. It shows the average ROIC of US industries from 1992 to 2006. The all-industry median was around 15%. Soft drinks and prepackaged software were at the top, around 35%+, with pharmaceuticals around 30%. Dead last were airlines, at roughly 6%, below any sensible cost of capital, averaged over fifteen years. Pull Damodaran's data library ("Margins by Sector," "ROE/ROIC by industry," updated every January, free) and you'll see the same pattern in any recent year in both the US and emerging-market datasets. The gap between a top-decile and a bottom-decile industry median is routinely 20+ percentage points of ROIC. That isn't noise. Whole industries stay above the value-creation line for decades while others stay below it for decades.

Now decompose the variance, and the picture sharpens. Three famous studies asked the same question. Of the total variance in business-unit profitability, how much attaches to which industry you're in, and how much to which firm you are? Schmalensee (1985, American Economic Review) found industry effects of roughly 20% and could barely detect firm effects in his one-year FTC dataset. Rumelt (1991, Strategic Management Journal) used panel data and reversed the emphasis. Stable industry effects fell to under 10%, while business-unit effects explained roughly 45%. McGahan & Porter (1997, SMJ) had the broadest dataset and landed in between: industry ~19%, corporate parent ~4%, business-specific ~32%, year ~2%, the rest unexplained. They also found that industry effects persist longer than firm effects. Here's the plain summary, then. Firm-specific effects explain more of the variance than industry effects do, roughly 30 to 45% against 10 to 20%, in every serious study since Rumelt.

The slogan, taken literally, is false.

So why does every great investor still put industry first? Three reasons, and they're what a working analyst should take from all this:

1. Industry is the strongest ex ante observable. You can score an industry's structure before you know anything about the specific firm. It's the base rate, the outside view. Firm effects are larger, but you can only measure them after deep firm-specific work, the kind the moat material and Phase 8 teach. A base rate you can read in an hour is worth more at the start of an analysis than a bigger effect you can't see yet.
2. Industry sets the distribution the firm draws from. The best US airline over a cycle earns roughly what a mediocre branded-consumer company earns. In Indian decorative paints, the fourth-place player, Akzo Nobel India (ROCE historically well above 20%; illustrative, pull screener.in), has out-earned the first-place Indian airline in most years. When the industry mean is 25% ROIC, execution decides whether you earn 18% or 35%. When the mean is 5%, execution decides whether you earn 2% or 9%. Both gaps are "firm effects." Only one of the two distributions contains outcomes worth owning.
3. Attribution to management is systematically inflated. Phil Rosenzweig's The Halo Effect documents the mechanism. When results are good, we retro-fit a story about visionary leadership. When the cycle turns, the same executives become cautionary tales. Two observations from the Berkshire letters compress fifty years of this evidence. In the textile post-mortem, Buffett noted that when a brilliantly regarded manager takes on a business with terrible economics, the business's reputation is the one left standing. (A later lesson on judging management quotes him and works through what it means.) And in the 2007 letter, reviewing the airline industry's economics since Kitty Hawk, he suggested that a farsighted capitalist on the spot would have done investors a favor by shooting Orville down.

Where does that leave you? Somewhere between determinism and hero-worship. Industry structure is the prior, and firm analysis is the update. Set the prior properly. Never skip it, and never stop there.

4.2 What "structure" means: and why five forces, exactly

Five isn't an arbitrary number. Once you see where it comes from, the forces stop being a list to memorize.

Every industry creates value. Customers' willingness to pay exceeds suppliers' opportunity cost, and the gap is the value created per unit. The industry's profit is whatever slice of that gap its members keep. Anything that takes a slice away is a "force":

• Buyers take a slice by pushing price down toward cost. Their power decides where in the willingness-to-pay range the price settles.
• Suppliers take a slice by pushing input costs up toward the customer price. Labor is a supplier, and so is Boeing.
• Substitutes cap the price from outside. A customer's other way of getting the job done sets a ceiling that no amount of restraint inside the industry can lift.
• Rivalry competes the remaining slice away among incumbents, usually through price, sometimes through cost-inflating arms races (advertising, capacity, discounts).
• Entry sets the duration of whatever slice remains. If outsiders can come and take it, today's profit buys tomorrow's competition.

Put that together and you get industry profit = value created − buyer's cut − supplier's cut − substitute's ceiling effect − rivalry burn, sustained only as long as entry barriers hold. The five forces are simply an exhaustive inventory of who can take the money, one force per claimant: customers, input providers, outside solutions, existing competitors, future competitors. So there are five, and not seven or three. Some writers add a sixth, either complements (from Brandenburger and Nalebuff's value net) or government. Treat government by what it does. It builds barriers, supplies, buys and referees rivalry, so it works through the five. India will give you constant practice: licensing, PLI schemes, tariff walls, price caps on drugs and MDR caps on payments.

Two rules before we take the forces one at a time.

Rule 1. Every force gets scored with evidence, and the score is a number. The scale throughout is 1 to 5 per force. A 1 is benign (the force takes almost nothing), 3 is moderate, and 5 is severe (the force takes almost everything). You'll attach at least two pieces of filing-grade evidence to every score. The sum, somewhere from 5 to 25, is a crude index. As shorthand, below ~10 you're usually looking at a structurally blessed industry and above ~18 at a cursed one. But the sum isn't the output. The output is three sentences: the expected band of industry ROIC against WACC, the binding force, and what could change it. The binding force is whichever one most constrains profits. Forces don't average, and the binding one governs.

Rule 2. The forces explain the average member's economics. A five-forces verdict is about the industry's center of gravity. It says nothing about any single firm. IndiGo earning real profits inside cursed Indian aviation is a firm effect. So is a lazy regional paint brand earning mediocre returns inside blessed Indian paints. Firm effects belong to the moat work that comes next. Keep the two layers apart and you'll avoid the commonest confusion analysts run into in this whole phase.

4.8 From scores to a verdict: the discipline that separates analysts from checklists

A completed five-forces analysis fits on one page, in this shape:

1. Industry defined in one line, with the boundary choices stated (see 4.15).
2. Scorecard. Five forces, each scored 1 to 5, each backed by 2 to 3 cited evidence items (a filing, a dataset, a computed number, never "seems competitive").
3. Verdict. State the expected band of industry ROIC relative to WACC, something like "this structure should let the average member earn WACC + 8–12 points" or "this structure should pin the average member at or below WACC". Then check it against the actual computed ROICs of the players. If your structural verdict and the observed returns disagree, one of them is wrong, and finding out which one is the analysis.
4. The binding force, the one that governs. Forces don't average. Four benign forces and one severe force make a constrained industry. Indian multiplexes are concentrated, yet content suppliers and the streaming substitute bind them. State the constraint in a sentence a 15-year-old could repeat.
5. Structural watch items. These are the two or three developments that would move a score: a merger, a regulation, an entrant with a distribution adjacency, a substitute crossing a price-performance threshold. Structure is durable, but it isn't permanent. Every score therefore carries a "what would change this" line, and the cases in 4.9 to 4.11 show you a 25-year-stable structure being stress-tested as well as a structure that flipped twice in a decade.

Why insist on questions over a checklist? A checklist user writes "rivalry: high, many competitors" for both restaurants and airlines and learns nothing. A question user asks how incumbents fight for the marginal customer and finds two different fires. Restaurant rivalry burns as localized monopolistic competition, where differentiation resets the fight every block. Airline rivalry burns through fare-matching engines within hours, across every overlapping route. It's the same word with different physics and a different verdict. The question drags the mechanism into view, and the mechanism is what you score.

4.9 Case 1: Indian decorative paints: anatomy of a blessed structure (and a live stress test)

All figures in this case are illustrative, modeled on ~FY24/FY25 scale. Pull the actual filings: the ARs and investor decks of Asian Paints, Berger, Kansai Nerolac, Akzo Nobel India, Indigo Paints and Grasim, plus screener.in for the ratio histories.

Start with the definition. The industry is Indian decorative paints (wall paints, enamels, putty, waterproofing for homes and buildings). It deliberately leaves out industrial and auto coatings, which have concentrated B2B buyers and different physics. It's roughly ₹75,000 to 80,000 crore in size, ~70 to 75% organized. Organized shares (pre-2024, illustrative) are Asian Paints ~53%, Berger ~18%, Kansai Nerolac ~13%, Akzo Nobel ~7%, Indigo ~3% and a tail of ~6%. That gives HHI ≈ 53² + 18² + 13² + 7² + 3² + ~6 ≈ 2,809 + 324 + 169 + 49 + 9 + 6 ≈ 3,366. That's deeply concentrated, with a stable leader at ~3× the #2.

Before you score a single force, look at what the structure delivered. Asian Paints compounded revenue at low double digits for decades with ROCE in the 30s. Berger earned ROCE in the low-to-mid 20s, and even #4 Akzo earned well above 20%. Gross margins across the top four were around 38 to 43%, with EBITDA margins of 15 to 23%. Everyone was above the hurdle. Twenty-plus percentage points of ROCE separated this industry's median from Indian aviation's. Now explain it:

• Rivalry: 2/5. One dominant leader with the most to lose from a price war holds the umbrella. Whenever the leader announces a price increase, the others follow within weeks. Watch any input-inflation year. In FY22, ~20%+ RM inflation was passed through in staged hikes by all four, and gross margins recovered within 4 to 6 quarters, visible in every player's quarterly P&L. Share moved glacially, a point or two per decade between the majors. Price isn't the weapon. Majors fight on distribution reach and painter relationships. Demand grew at ~1.5× GDP as repainting cycles shortened from ~7 to 8 years toward ~4 to 5, so nobody needed to steal volume to grow.
• Entry threat: historically 1.5/5, now the live question. Entry faces a four-layer barrier stack. First comes distribution. Asian Paints alone has ~1.5 to 1.6 lakh retail touchpoints, the majority holding its tinting machine. That's a dealer-level switching cost. Each machine (plus its colorant stock and software) commits the counter to one brand's system, and counter space is finite. Second comes logistics density. Paint is heavy, low-value-per-kg, seasonal, and sold in thousands of SKUs. Asian Paints resupplies dealers up to 3 to 4× a day with no distributor layer, a six-decade data-and-depot asset an entrant can't buy quickly. Third is brand as sunk cost, decades of accumulated A&P (all the majors spend steadily on brand) that an entrant must match before earning a rupee. Fourth is the entry ledger. For thirty years entrants either stayed niche or exited bruised. Indigo Paints is the niche case: it built ~3% share over 20 years on differentiated products and tier-2 towns first, profitably but slowly. Several global majors sold out or stagnated. Notice what's absent from the stack. Technology isn't there, because paint chemistry is licensable. Neither is capital as such, since a plant costs ₹500 to 1,500 crore, which is trivially financeable. These barriers are demand-side and distribution-side, the durable kind.
• Substitutes: 1.5/5. Wallpaper is niche. So are tiles and bare-concrete aesthetics, and none of them is improving its price-performance against paint. There's no trajectory to date a crossover on. Over the longer run, keep an eye on pre-painted and factory-finished construction materials.
• Buyer power: 1/5. The buyers are millions of households, each buying every 4 to 8 years. Paint is ~3 to 5% of a renovation's cost while the labor to apply it is ~50 to 60%, so the household optimizes on the painter's recommendation instead of the paint's price. The painter is the real decision unit, fragmented across lakhs of contractors, and the majors invest heavily in painter loyalty (apps, training academies, points programs). No aggregator re-concentrates this demand… yet. Watch paint-plus-service offerings and quick-commerce hardware channels.
• Supplier power: 2/5. Paint's inputs are titanium dioxide (the largest single pigment cost) plus crude-oil-derivative monomers and solvents. All are global commodities from multiple sources, cyclical but not concentrated against the paint makers, who are large customers of their local suppliers. Input spikes get passed through with a 2 to 4 quarter lag, and FY22 proved the mechanism live.

Scorecard: 2 + 1.5 + 1.5 + 1 + 2 = 8/25, blessed. Structure should let the average organized member earn WACC + 8 to 15 points, and observed ROCEs (20s to 30s against a ~12% INR hurdle) agree. No force binds, though the umbrella depends on rivalry staying oligopolistic. Then come the watch items, and with them the case turns from textbook to live ammunition:

The Birla Opus stress test (2024–). In February 2024, Grasim (Aditya Birla group) entered decorative paints with a committed ~₹10,000 crore investment. It planned ~6 plants toward ~1,300+ MLPA capacity, a scale comparable to the #2 incumbent built in one stroke, and backed them with a national ad blitz. The structurally interesting part was a distribution adjacency. Birla White (wall putty) is already stocked in tens of thousands of the same paint and hardware counters, so Opus had a dealer relationship to convert instead of one to build. Management claimed high-single-digit organized share within its first full year (check in Grasim's FY25/FY26 disclosures). Look at how the incumbents responded. Asian Paints' FY25 revenue fell ~4 to 5%, to ~₹34,000 crore. Its EBITDA margin compressed from ~22 to 23% toward ~18%, PAT fell ~25 to 30%, and its stock de-rated. Meanwhile dealers received better terms, tinting machines were subsidized, and consumers got price.

Now read it as an analyst would. Capital alone didn't do this. Capital plus a distribution adjacency plus a conglomerate's patience did. That barrier stack was real, but it had a mapped flank. Next, the first casualty of entry is the price umbrella, so industry economics can deteriorate for everyone even if the entrant never reaches #2. And the five-forces scorecard is a photograph. Your one-pager carries watch items for that reason. Will paints settle as a five-player oligopoly with slightly lower economics, or tip into a share war? As of mid-2026 that's an open question, and you should track it in the filings each quarter.

Nobody knows yet, and your write-up should say so.

Adhesives are worth a sidebar, because they run on the same physics, only stronger. Pidilite's Fevicol holds roughly two-thirds of organized white-glue adhesives (illustrative). It's an even lower-ticket item, at ~1 to 2% of a furniture job. The carpenter chooses it entirely, and carpenters have been courted for decades. Fevicol's carpenter clubs predate the internet. It moves through the same dense counters, and the brand is so dominant it is the category's name. Consolidated EBITDA margins run ~20 to 24% and ROCE ~25 to 30% (illustrative, FY24 scale). When the decision-maker is a fragmented influencer, the ticket is trivial and the brand equals the category, buyer power approaches zero. And no entrant has found the flank. Compare it with paints force by force and you'll see why Pidilite's structure may be the more finished of the two.

4.11 What the contrast teaches

Put the two scorecards side by side and five lessons fall out:

1. Structure shows up in the accounts, reliably. Twenty-plus points of ROIC separated these industries' medians for decades. No management story was needed to predict either. The structures were legible in advance, from public evidence.
2. When the same force carries opposite signs, the cause is physics. Culture doesn't explain it. Paint rivalry is gentle because the product differentiates at the painter's counter and nothing perishes. Airline rivalry is savage because seats spoil at door close above near-zero marginal cost. Score the mechanism and leave the adjectives out.
3. Within-industry variance is real, and it's bounded. IndiGo, the cost leader you'll meet properly in the next section, is the best airline story in the world and has earned real profits in most years (FY24 PAT ~₹8,200 crore, ~64% domestic share; illustrative). Kansai Nerolac has trailed its paint peers for years. Yet the best airline still had COVID wipe out several years of accumulated profits, and the laggard paint company still out-earned most airlines that ever existed. The neighborhood bounds what execution can deliver.
4. Structure moves, in both directions, on events you can watch. Airlines improved through consolidation and then partly relapsed. Paints deteriorated because of one determined entrant with a flank. The scorecard is a photograph, and the watch items are the film.
5. The profit pool needn't be where the headlines point. In aviation it's with the GDSs, lessors, engine aftermarkets and card networks. In the paint job it's with the manufacturer, while the painter gets wages and the dealer gets turns. So the next instruments you need are the value chain and the profit-pool map.

4.13 The value chain: Porter's activity map

Now drop from the industry to the firm's anatomy. Porter's value chain (Competitive Advantage, 1985, chs. 1 and 2) breaks any business into nine generic activities. Five are primary, the flow of making and selling. It runs from inbound logistics to operations to outbound logistics, then marketing & sales, then service. Four are support, the scaffolding over all of them: firm infrastructure, human resource management, technology development, procurement. Margin is what's left of the customer's price after all the activities are paid for. You've already walked one company's chain from input to cash. Porter's contribution is to make the walk comparative and causal. Every cost the firm incurs and every rupee of willingness-to-pay it creates has to belong to some activity, so strategy becomes visible as which activities you perform differently, or don't perform at all.

There are two ways to use it.

Use 1. Find where the money actually happens. Allocate the cost structure to activities, where the P&L plus the MD&A's cost commentary usually gets you 80% of the way, and ask which activities create the willingness-to-pay. For Asian Paints the answer is startling, and strategic. The factory (operations) is maybe a fifth of the story. The decisive activities are outbound logistics, with 3 to 4 deliveries a day to 1.5 lakh counters and no distributor layer, done at a level no rival matches, and marketing & sales as painter-relationship infrastructure. For IndiGo, the chain's crown jewels are operations scheduling (utilization, turnarounds) and procurement (mega aircraft orders timed to cycle troughs). Know which two activities carry the firm and you know what to monitor for erosion.

Use 2. Compare two rivals' chains, activity by activity. Lay IndiGo's chain beside a full-service carrier's. Inbound logistics (fuel, airframes) are the same. Operations are radically different: fleet commonality against a mixed fleet, 35-minute turns against 55. The service chain differs (none at all, against lounges, meals, two cabins), and so does the distribution mix. Now every line of the CASK gap has an address. "Low-cost carrier" stops being a label and becomes a list of ~9 deliberate activity choices, each of which you can check for durability. It's also your bridge back to 4.12. A generic strategy is a coherent set of activity choices, and "stuck in the middle" is an incoherent one. And, as Porter argued in 1996, the durability of a position rises with the number of reinforcing links an imitator must copy at once. Copy IndiGo's turnaround time without its fleet commonality and network density, and you've copied the timetable without the economics.

Don't let the tool turn into paperwork. The nine boxes are a checklist for not missing activities, and you shouldn't treat them as a form to fill. For a services or platform business, rename freely. A marketplace works better as demand acquisition, then matching, then fulfillment, then retention. Your working routine is costs to activities, willingness-to-pay to activities, then a look at which activities differ from rivals and why that difference survives.

4.15 Industry definition: the trap you fall into before you begin

Every instrument so far takes "the industry" as its input. Define it wrong and everything downstream comes out confidently and neatly wrong. There are two ways to fail:

Too broad means averaging the structure away. "Indian FMCG" blends soaps, cigarettes, paints, and noodles. They don't share buyers or channels. Their force profiles differ too, so the average describes nothing. "US retail" mixes Costco's economics with a mall jeweler's. Structure is visible only at the level where the forces are actually shared. Decorative paints works as an industry. "Coatings" doesn't. Domestic ULCC air travel works, and so do luxury hotels in Indian metros, while "transportation" and "hospitality" are too wide to mean anything. Porter's own rule is that where structural differences between segments are large, you analyze the segments separately.

Too narrow means leaving out the constraint. Define the industry as "premium multiplexes in South Mumbai" and you exclude the streaming service and the ₹250 single screen that cap your pricing. Define it as "sit-down casual dining" and you miss delivery platforms rearranging the whole demand pool. Theodore Levitt's "Marketing Myopia" (HBR, 1960) is the classic warning. Railroads defined themselves as being in the railroad business instead of transportation, and missed trucks and planes. Watch for the overcorrection, though. Defining Coca-Cola's market as "share of stomach," or every media company's as "the battle for attention," flatters TAM stories and excuses bad economics. You want the boundary where pricing constraints actually operate, however uninspiring it looks.

You have two working tests.

• Demand-side substitution. Would the marginal customer switch between the two products in response to a modest relative price change, say 5 to 10%? If so, they're in the same market. A Rolex buyer doesn't cross-shop a Titan on price. A Vistara economy passenger absolutely cross-shopped IndiGo, which told you Vistara's "premium market" was mostly a hope.
• Supply-side substitution. Could producers of the other product redeploy their assets into yours quickly and without large sunk cost? If so, they belong in your entrant set even if they're absent today. Birla White's putty distribution made Grasim a supply-side neighbor of paints all along, and the map should have flagged it.

Geography is part of the definition too. Freight intensity sets the radius. Cement travels ~300 km economically, so "Indian cement" is really five regional markets with different HHIs and price behavior. Paint is denser in value and travels further, and software travels free. Airlines price by route. Nationally, HHI is ~1,250 (illustrative; you compute it in the practice set) and reads "unconcentrated," while individual city-pairs are duopolies at HHI > 4,000. Even those duopoly routes earn poorly, which correctly tells you that in airlines the binding constraints are cost structure and suppliers. Local concentration isn't what binds. When a boundary is uncertain, run it twice, narrow and broad, and see whether the verdict flips. If it does, the boundary is the analysis, and you should say so explicitly in your write-up.

4.17 Compression: what survives into the teardown

In Phase 8 you'll have roughly ten minutes for all of this inside a 60-minute teardown. Here's what survives the compression:

1. The base rate comes first. Which industry is it, which structural archetype, what do members earn? Damodaran's industry tables or your Phase 5 playbooks give you the number in seconds.
2. Name the binding force, mechanism included, in a single sentence.
3. Check the entry ledger. Who tried in the last decade, and what happened? It's the fastest test of barriers you can trust.
4. Place the company in the pool. Which link of the chain is it, and is that link's share of the pool growing or shrinking?
5. Check the definition. Are you scoring the market where this company actually competes?

Greenwald's shortcut, which you'll meet with the moat material, compresses even further. If there are no barriers to entry, skip the rest, because returns will mean-revert to cost of capital regardless of the other forces. You now have the full apparatus that makes that shortcut trustworthy.

Common mistakes & how experts think differently

1. Reciting the checklist instead of finding the binding force. The novice writes five paragraphs of "moderate-to-high" and concludes nothing. The expert writes one sentence per force plus a verdict, and the verdict names the constraint: "concentrated exhibition, but content suppliers and the streaming substitute govern, so expect single-digit ROCE." If your five-forces write-up could describe three different industries, it describes none of them.
2. Treating concentration as the verdict. "Top two players have 60% share: great industry!" US airline consolidation, Indian multiplexes, and memory chips all falsify that. Concentration alongside commodity products, powerful suppliers or strong substitutes still earns poorly. What you actually want is fragmentation among buyers. HHI is one input into one force.
3. Scoring the forces without filing evidence. "Buyer power seems low" is an opinion. "No customer exceeds 10% (segment note), DSO stable at 32 days across five years, and FY22's 20% RM inflation was fully passed through within four quarters (gross margin 41.2%, then 38.9%, then 41.8%)" is an analysis. Every score in your scorecard should survive the question "where, specifically, did you see that?"
4. Confusing the industry's fate with the firm's. Five forces describe the average member. The expert keeps two layers in view. One is the neighborhood's base rate, which you score here. The other is the firm's deviation from it, which the moat work measures. Buying "the best house in a terrible neighborhood" can work, as IndiGo shows. But it works only when the firm's advantage is structural (cost leadership). A narrative like "well-managed" won't do.
5. Attributing structure to management, in both directions. Executives of paint companies gave confident interviews about excellence for thirty years while the structure carried them. Airline CEOs were called fools for outcomes their structure dictated. That's Rosenzweig's halo effect. The expert asks what would a mediocre operator earn in this seat? Management gets credit only for the residual.
6. Missing the intermediary who actually decides. Households "buy" paint, but painters choose it. Doctors prescribe what patients "buy," and developers choose the app store's ecosystem. Score buyer power at the decision-maker, whoever happens to pay. Misidentify the decision unit and the whole force is mis-scored.
7. Capital intensity mistaken for an entry barrier. "It costs ₹1,400 crore to build a plant" deters no one when the plant is financeable and the output is a commodity, because supplying that money is what capital markets are for. You'll see later in the phase that the capital cycle is this mistake industrialized. The expert asks whether the capital is sunk and unrecoverable (brand-building, R&D, dealer networks). Only sunk cost deters.
8. Static reading of substitutes. The substitute that matters is rarely competitive today. It's the one whose price-performance curve is steeper than yours. Experts date the crossover, while novices note "substitutes exist" and move on. Everyone who valued multiplexes on 2019 footfalls learned this the hard way.
9. Analyzing the firm's own chain but not the industry's. The value chain (4.13) tells you where your costs are. And the profit pool (4.14) tells you whether your link is worth occupying. Foxconn executes assembly brilliantly, at 2.5% margins, forever, because the link is structurally starved. Before you admire a company, check its seat at the table.
10. Defining the industry to fit the thesis. Bulls define narrowly ("premium urban quick-commerce has no competition"). Bears define broadly ("it's just retail"). The expert defines the industry by where pricing constraints actually operate, uses the substitution tests, and runs it twice when uncertain, reporting whether the verdict flips.

Worked examples

Worked example 1: Structure flips in real time: Indian telecom, FY16 → FY25

(All figures are illustrative, rounded to make the mechanics clear and modeled on TRAI quarterly reports and Bharti/Vi filings. Pull the actuals.)

Setup, FY16 (pre-Jio). Subscriber shares are Bharti 24%, Vodafone 19%, Idea 17%, Reliance Comm 10%, BSNL 9%, Aircel 8%, Tata Docomo 6%, Telenor 5%, others 2%. Blended ARPU ≈ ₹190/month for the leader. Sector ROCE is already below WACC for most players. It's a capex-heavy business, spectrum auctions extract the surplus, and government acts as a supplier.

Step 1, concentration before. HHI = 24² + 19² + 17² + 10² + 9² + 8² + 6² + 5² + 2² = 576 + 361 + 289 + 100 + 81 + 64 + 36 + 25 + 4 = 1,536. That's moderately concentrated, but with eight materially sized, strategically diverse players (state-owned BSNL, distressed RCom, subscale foreign entrants). The "diverse rivals" red flag is therefore at maximum.

Step 2, the shock. In September 2016, Jio enters with a nationwide 4G network built in advance, capital used as a chosen weapon (~₹1.5+ lakh crore). It offers free voice forever and data at a fraction of prevailing prices. Notice which barrier failed. Telecom's classic barriers are spectrum licenses and network capital, but policy had made spectrum buyable, and one conglomerate's balance sheet made capital irrelevant. Retaliation was impossible, because incumbents carried legacy 2G/3G cost bases and debt.

Step 3, the carnage, in force language. Rivalry went from 4 to 5+, because price became the only variable. ARPU fell from ~₹190 toward ~₹100 by FY19. Exit barriers proved low for the weak, which was a mercy. RCom, Aircel, Telenor, Tata exited or collapsed. Vodafone and Idea merged in desperation, and their combined share then shrank, which is merger arithmetic under fire. Industry revenue pool shrank while capex requirements rose with the 4G buildout, and ROCE went deeply negative for all but Jio and Bharti.

Step 4, concentration after (FY25, illustrative). Jio 40%, Bharti 33%, Vi 18%, BSNL 8%, others 1%, so HHI = 1,600 + 1,089 + 324 + 64 + 1 = 3,078. That's two funded players + one zombie + one state player. Tariff repair followed as structure predicts, with coordinated-looking hikes of ~20 to 25% each in Dec 2019, Nov 2021, Jul 2024. Leader ARPU was back to ~₹245 to 250 by FY25. Re-score rivalry at ~3 and improving. But two things keep the structure from fully healing. One is supplier power, the spectrum + capex treadmill. Another is the zombie's survival, with Vi's ~₹2 lakh crore obligations kept alive by dilution and dues moratoria.

Lessons. HHI doubled in nine years, so structure is dynamic, and its direction was readable from the moment Jio's capex was announced. Jio destroyed ~₹4+ lakh crore of incumbent value by resetting the price deck. Being better had nothing to do with it, since entry threat is about capacity to harm, whatever the product quality. After the shock, the equilibrium (3 players, price umbrella restored) was just as readable. Analysts who scored the new structure in 2019 to 20 caught the ARPU-repair cycle early. Both halves of the decade were structure at work.

Worked example 2: The profit pool hiding inside an airline: loyalty economics

(Synthetic composite, modeled on the scale of Delta's disclosed Amex remuneration; pull Delta's 10-K, loyalty note, for the real figures.)

Given. An airline has flying revenue (passenger + cargo) of $54bn at a 5.5% operating margin. It also earns co-brand card remuneration of $7.2bn, the payments a card partner makes for miles sold, brand access, and acquisition. On that, the incremental operating margin is ~52%, because the miles' marginal cost is a seat that would often fly empty, plus program costs.

Step 1, flying profit: $54bn × 5.5% = $2.97bn. Step 2, loyalty profit: $7.2bn × 52% = $3.74bn. Step 3, pool shares: total operating profit = 2.97 + 3.74 = $6.71bn. Loyalty share = 3.74 / 6.71 = 55.8%. Revenue share of loyalty = 7.2 / 61.2 = 11.8%.

Now read it with the profit-pool lens. A line of business with ~12% of revenue and ~56% of operating profit is a profit pool inside the firm, and its force profile explains why. In effect, the card partnership is a three-party network. It sells a differentiated product (aspirational travel rewards) to fragmented cardholders with real switching costs (banked miles, status), and it faces none of flying's suppliers. So the "airline" is two businesses: a cursed one that generates the loyalty program's product, and a blessed one that monetizes it.

This map forces three implications on you. First, the equity's value driver is partly interchange regulation. A credit-card fee cap would hit >half the profit, so score it as a watch item. And notice the definition trap, because this risk is entirely outside "the airline industry" as normally defined. Second, comparing airlines on EV/EBITDA without separating loyalty economics compares different animals. Third, United's COVID-era MileagePlus appraisal (~$22bn against a ~$10bn market cap) was the market being told, in one number, where the pool was.

Worked example 3: Generic strategies, computed: one route, two cost structures

(Synthetic, calibrated to illustrative Indian CASK/RASK levels.)

Given. A 1,000-km route. Carrier L (cost leader): CASK ₹4.50, 186 seats, 87% load factor. Carrier F (full-service): CASK ₹6.20, 158 seats (two-class), 78% load factor. The market-clearing average fare, set by competition (i.e., by L's pricing plus demand), is ₹5,300 per passenger. F achieves a 12% yield premium (₹5,936) through corporate contracts and cabin mix.

Step 1, cost per passenger. Cost per flight = CASK × seats × km. For L, that's 4.50 × 186 × 1,000 = ₹8,37,000, and per passenger at 87% load (162 pax) it's 8,37,000 / 162 = ₹5,167. For F, it's 6.20 × 158 × 1,000 = ₹9,79,600, and per passenger at 78% load (123 pax) it's 9,79,600 / 123 = ₹7,964. Step 2, profit per passenger. L: 5,300 − 5,167 = +₹133. F: 5,936 − 7,964 = −₹2,028. Step 3, the strategic reading. L earns a thin but positive spread at the market price its own pricing holds down. That's what cost leadership means in a commodity industry. Your advantage is denominated in the only currency the customer respects. F's 12% yield premium is real, and customers demonstrably pay more, but its cost premium is 54%. Differentiation fails when customers won't pay enough more than the differentiation costs, even if they will pay more. That gap (premium 12%, cost +54%) is "stuck in the middle" written as arithmetic, and it's the P&L Jet Airways and Vistara lived inside for a decade. Can F escape? Only in two ways. It can move to (a) segments where the yield premium widens toward the cost premium, like long-haul international with business cabins, where full-service economics do work. Or it can (b) collapse the cost gap, which means dismantling the very activity system (4.13) that produces the premium. The trade-off is the strategy.

Worked example 4: Scoring one force from a filing excerpt

(The excerpt is synthetic but assembled from standard 10-K risk-factor and note language. It's the drill you'll run on real filings in the mini-project.)

Excerpt (components supplier "Veltron," 10-K). Item 1A: "Our two largest customers accounted for approximately 43% and 24% of net revenue… Our agreements generally do not contain long-term volume commitments and are terminable on 90 days' notice… We have agreed to annual productivity price reductions with certain OEM customers… Our customers are increasingly dual-sourcing components we currently sole-supply." Segment note: Customer A = 43%, Customer B = 24%. Receivables: DSO 71 days (prior years: 58, 63). Gross margin went 31.4%, 29.0%, 27.2% over three years, while the 10-K notes raw-material indices were "broadly stable."

Now score buyer power, mapping each piece of evidence as you go. (1) Two buyers hold 67% of revenue [segment note], so leverage is near-maximal. (2) There are no volume commitments and a 90-day termination clause [Item 1A], so the leverage is contractually live. (3) Annual price-downs are institutionalized [Item 1A], which means leverage is being exercised as policy, year after year. (4) DSO went from 58 to 71 days with no revenue surge [receivables note]. Buyers are stretching terms, and the supplier is financing its customers. (5) Gross margin fell 420bp on stable inputs [P&L + MD&A], so the transfer is visible in the accounts, which closes the loop. (6) Dual-sourcing [Item 1A] shows the buyer deliberately dismantling the remaining switching costs, the sole-supply positions. Score: 5/5. Before you finalize, check one mitigant. Does any product line carry re-qualification switching costs or aftermarket revenue? Nothing in the excerpt says so, so the score stands. Verdict sentence: "Buyer power is the binding force: two OEMs holding two-thirds of revenue are converting Veltron's margin into their own procurement savings at ~140bp/year, and the filing shows the mechanism in four separate places." Look at the method. Every clause of the conclusion has a filing address, and that's the standard your mini-project will be graded against.

Worked example 5: Concentration is not pricing power: the import border as the real ceiling (India)

(All figures are synthetic or illustrative, scaled to a realistic tradable-commodity industry. Call it Indian float glass, with synthetic producers "A/B/C". Check any live case against the producers' filings, DGCI&S import data, and the customs tariff schedule.)

Say you're handed a domestic commodity industry and told "it's a tight oligopoly, so pricing power is strong." Three domestic producers plus a sliver of imports supply the market. Shares of domestic consumption are Producer A 46%, B 34%, C 12%, imports/others 8%. The prevailing domestic price is ₹41,500/tonne. Before you accept the claim, run the two numbers that decide it.

Step 1, concentration measured. HHI = 46² + 34² + 12² + 8² = 2,116 + 1,156 + 144 + 64 = 3,480 and CR3 (domestic) = 92. Under both the 2010 and 2023 DOJ/FTC thresholds this is highly concentrated (>2,500). A checklist-reader stops here and writes "rivalry benign, pricing power strong." Hold that verdict until the second number is in.

Step 2, the landed cost of the marginal import. A good that can be shipped answers to the price at which a container clears customs, whatever the domestic share table says. Build the import-parity price step by step. Ex-works $360/t + ocean freight & insurance $45 = CIF $405/t. Add basic customs duty of 10% to get $445.50. Add the social-welfare surcharge (10% of the duty, = $4.05) to get $449.55. Add clearing and inland freight to market of $18 to get $467.55/t. At ₹86/$ that comes to ₹40,209/tonne.

Step 3, the verdict flips. The domestic price (₹41,500) is only (41,500 − 40,209)/40,209 = 3.2% above the landed cost of imports. A "tight oligopoly" that actually used its pricing power, say with a 10% hike to ₹45,650, would price itself ~13.5% above the border and pull in every importer with a warehouse. So the binding constraint is the threat of entry-by-import, an outside force the share table never shows. Internal rivalry, the thing the HHI measured, isn't what binds. The concentration was real, and nearly irrelevant to the pricing question.

Step 4, what actually moves this industry. Because the ceiling is a landed cost, the two variables that move it are the exchange rate and the tariff. Rivals' conduct doesn't move it. If the rupee strengthens to ₹80/$, the same $467.55 lands at ₹37,404/t. Suddenly the domestic price is ~11% above parity, and margins compress with no change in structure. A safeguard or anti-dumping duty works the other way. Lifting the effective rate to ~25% pushes the landed cost to ₹48,917/t (17.9% above today's domestic price), which hands the oligopoly real pricing room overnight. Here's how to read it. For any tradable, add a sixth line to your rivalry evidence, the import-parity price, and put the customs tariff and the currency on your structural watch list. "Highly concentrated" describes the size of the room. But the border decides how high the ceiling is. In India this is a live and constant game. In the paints case, the FY22 pass-through story worked because paint is dense and travels. A light, shippable commodity would have imported the deflation instead.

Worked example 6: Same 24% ROIC, different persistence: why the industry prior is set first (US)

(Synthetic firms "H" and "F". Persistence coefficients here are illustrative. They're chosen to reflect the McGahan & Porter finding that industry effects persist longer than firm effects, and they don't measure it. Check the direction against that literature, and treat the specific half-lives as a teaching model.)

Remember the uncomfortable pair of facts from the evidence on industry effects? Firm effects explain more one-period profit variance than industry effects, yet every great investor scores the industry first. This example turns that resolution into arithmetic. Two synthetic US businesses each carry $1,000m of invested capital, and each earns 24% ROIC today against an 8% WACC, the same 16-point spread. Firm H's advantage is industry-structural, because it's in a blessed neighborhood. Firm F's identical return comes from a firm-specific edge in an otherwise competitive industry.

Step 1, this year they're indistinguishable. Economic profit = (24% − 8%) × $1,000m = $160m for both. A one-period snapshot, which is what a variance decomposition of a single year sees, can't tell them apart. Schmalensee's one-year dataset "found industry" and little else for that reason.

Step 2, fade the spread. Advantages mean-revert as arbitrage does its work, but at different speeds. Model the spread as spreadₜ = spread₀ × fᵗ. Industry-structural advantages decay slowly (f = 0.88, illustrative), and firm-specific ones decay fast (f = 0.70). By year 5, H's spread is 16 × 0.88⁵ = 8.4 pts and F's is 16 × 0.70⁵ = 2.7 pts. By year 10 it's 4.5 pts against 0.5 pts. The identical twins have separated completely.

Step 3, the half-life makes it portable. The spread's half-life is ln(0.5)/ln(f), which gives 5.4 years for the industry-anchored edge and 1.9 years for the firm-specific one. So "industry effects persist longer" means a longer half-life, and half-life is what a discounting investor is paid for.

Step 4, capitalize the difference. Discount each excess-economic-profit stream at the 8% WACC over 25 years: Σ (spread₀ × fᵗ × $1,000m) / 1.08^(t+1). Firm H's stream is worth ~$795m and Firm F's ~$421m. That's 1.89× more value for the identical starting ROIC, and the whole gap comes from persistence. Set the industry base rate first, then, because it's the durable component, the one you can read ex ante. You can score H's neighborhood in an hour from public structure, and most of the capitalized value comes from it. Firm effects are larger in the cross-section of a single year and smaller in the present value of a decade. Prior first, update second. This is the number behind that rule.

Practice set

Problems 1 and 2 are guided, with the method shown in the prompt. Problems 3 to 9 are independent. Problems 10 and 11 are timed. Write your full answer before you open any solution (working agreement #3). Synthetic data is labeled, and you should treat all real-company figures as illustrative.

Problem 1 (guided). (a) Compute HHI and CR4 for organized Indian decorative paints, pre-2024 shares (illustrative): Asian 53%, Berger 18%, Kansai 13%, Akzo 7%, Indigo 3%, six ~1% players. (b) Compute HHI and CR4 for US domestic airlines (illustrative shares of revenue passenger miles): AA 17.5%, DL 17.5%, WN 17%, UA 16%, AS 6%, B6 4.5%, NK 4.5%, F9 3.5%, fragmented tail 13.5% (treat as contributing ~20 HHI points). (c) Classify both under the 2010 and 2023 DOJ/FTC thresholds. (d) One paragraph: why does the more concentrated industry here also happen to be the more profitable one, and why would that inference be unreliable as a general rule?

Solution. (a) HHI = 53² + 18² + 13² + 7² + 3² + 6×1² = 2,809 + 324 + 169 + 49 + 9 + 6 = 3,366. CR4 = 53 + 18 + 13 + 7 = 91. (b) HHI = 306.25 + 306.25 + 289 + 256 + 36 + 20.25 + 20.25 + 12.25 + ~20 ≈ 1,266. CR4 = 17.5 + 17.5 + 17 + 16 = 68. (c) At >2,500, paints are highly concentrated under both guideline sets. Airlines at 1,266 are unconcentrated under the 2010 thresholds (<1,500) and moderately concentrated under 2023 (1,000 to 1,800). (d) In paints, concentration coexists with fragmented buyers, weak substitutes, and passive suppliers, so the leader's umbrella holds and concentration expresses a healthy structure. But concentration is one determinant of one force. Airlines at CR4 = 68 (route-level HHI often >4,000) still earn poorly, because commodity product + perishable capacity + supplier power bind. Memory chips, multiplexes, and telecom towers all show concentration coexisting with weak economics. Concentration is an input to the rivalry score, and profitability is the output of all five forces.

Problem 2 (guided). Score buyer power (1 to 5) for the synthetic auto-component maker "Vertex Autocomp" and name the two strongest tells. Here's the packet. In the segment note, Customer A is 41% of revenue and Customer B 27%. Contracts include 2 to 3% annual "productivity price-downs." An RM pass-through clause covers 60% of raw-material cost with a one-quarter lag. DSO went 58, 66, 74 days over three years while revenue grew 6%/yr. Gross margin went 29.8%, 28.1%, 26.4% while the RM index was flat. Both OEMs dual-source every component by procurement policy. Then name the two business changes that would most reduce the score.

Solution. Score: 5/5. Start with leverage. There's 68% two-customer concentration [segment note], plus dual-sourcing as policy, which keeps switching costs deliberately at zero. Then look at sensitivity. OEMs are thin-margin assemblers for whom components are the dominant cost, so they must squeeze. And the power is being exercised, visibly, in the accounts. Price-downs are contractual, and nothing about them is episodic. DSO is up 16 days on modest growth, so Vertex increasingly finances its buyers. Gross margin is down 340bp on flat RM, which quantifies the transfer at ~115bp/yr. The two strongest tells are (i) margin compression against a flat input index, which isolates price/terms as the cause and kills the cost-inflation alibi, and (ii) DSO drift, terms of trade worsening in the quietest corner of the balance sheet. Read what the pass-through clause covers. The 60%, one-quarter-lag clause protects against input risk and does nothing about price-down risk. To reduce the score, Vertex could (1) win proprietary, re-qualification-heavy content (safety-critical, software-integrated parts) that makes dual-sourcing costly, since switching cost is the antidote to concentration, or (2) diversify the customer base (or grow aftermarket/exports) so no buyer exceeds ~15 to 20%. Full-coverage pass-through with zero lag would help too, though less.

Problem 3. Two synthetic industries. In K, a greenfield plant costs ₹1,400 crore, which is 4% of national capacity. Freight economics cap the sales radius at ~300 km, the product is chemically identical across producers, and incumbent gross margins are 32%. K's entry ledger shows six entries in ten years, and four of them reached >5% regional share within eight years. In L, a plant costs ₹40 crore (~1% of capacity), and incumbent gross margins are 55%. Incumbents have spent 12% of sales on advertising for 20 years, and products reach 4 million outlets through layered distribution. L's entry ledger shows ~40 brand launches in ten years, and none crossed 2% national share. Only two entrants gained real share, and they did it by acquiring the #4 incumbent. Which industry has higher entry barriers? Say where specifically they are, and what each entry ledger proves.

Solution. L, by a wide margin, and the barrier isn't the plant. K looks "capital intensive" (₹1,400 crore against ₹40 crore), but the capital buys a recoverable, financeable asset that produces a commodity. Any period of ROCE above the hurdle attracts a new plant, and the ledger confirms it, with 4 of 6 entrants succeeding. So returns are structurally mean-reverting. K is a capital-cycle industry, and its true HHI is the regional one (300-km radius), so the national number is the wrong one to use. L's plant is trivial, but entering means matching two decades of sunk, unrecoverable spending. There's the accumulated brand (12% of sales × 20 years), an invisible balance-sheet asset an entrant must replicate before earning anything. And there's access to 4 million outlets whose finite shelf space is already allocated to incumbents with proven velocity. This ledger proves the wall: 40 attempts and zero organic successes. Only one strategy worked, buying an incumbent's brand-and-distribution stack, which is entry in name only (capacity and competitive intensity unchanged). Here's the general law. Capital requirements deter only when sunk. Demand-side and distribution barriers, built over decades, deter even when the price of admission looks small.

Problem 4. Substitute trajectory. A multiplex visit costs a family of three ₹390/head (₹260 ticket + ₹130 F&B) for a 2-hour film. A streaming bundle costs the household ₹2,400/year and delivers 480 household viewing-hours/year. (a) Compute cost per person-hour for each (assume all three watch together for streaming). (b) The theatrical exclusivity window has fallen from 8 weeks to 4. Streamers' content budgets are rising ~15%/yr, multiplex ticket prices rise ~5 to 7%/yr, and footfalls remain ~15% below pre-2020 levels (all illustrative). Score the substitute force and state its direction. (c) Name two responses available to the exhibitor and which customer cohort each defends.

Solution. (a) Cinema: 390 / 2 = ₹195 per person-hour. Streaming: 2,400 / 480 = ₹5 per household-hour, and ÷3 viewers ≈ ₹1.7 per person-hour. Ratio ≈ 115:1. (b) Score ≈ 4.5/5 and rising. The static gap alone isn't the verdict, because the products do different jobs, an out-of-home event against convenience viewing. But every trajectory marker points one way. Streaming's library value is improving (content spend +15%/yr) while its price is roughly flat, so its price-performance curve is steeper. Cinema's own price rises 5 to 7%/yr into it. A window shrinking from 8 weeks to 4 cuts the "see it first" half of cinema's value proposition, the differentiation half. And a footfall gap that persists years after reopening is the erosion showing up in volume instead of price. (c) One response is event-ization and premium formats (IMAX-type screens, blockbusters as spectacle). It defends the experience-seeking cohort by widening the differentiation the substitute can't replicate. A second is F&B and pricing architecture (cheaper Tuesday tickets, subscription passes), which defends the price-sensitive frequent-viewer cohort, the one substitution claims first. Notice what neither response fixes: dependence on hit supply. That belongs to a different force (suppliers), and the exhibitor's scorecard needs both.

Problem 5. US airline consolidation, in numbers. Domestic shares, 2007 (illustrative): AA 13, WN 12, UA 11, DL 10, CO 8, US 8, NW 7, B6 4, AS 3, F9 2, fragmented tail 22 (contribute ~30 HHI points). Shares, 2019: AA 17.5, WN 17.7, DL 17.3, UA 15, AS 5.5, B6 5, NK 4, F9 3, tail 15 (~30 points). (a) Compute both HHIs. (b) Industry net margin went from deeply negative (2001–09 cumulative losses ~$55–60bn) to ~5–9% in 2015–19, while jet fuel roughly halved in 2015–16. A colleague says "the profits were just fuel." What evidence would you check to apportion credit between structure and fuel, and what did that evidence actually show?

Solution. (a) 2007: 169 + 144 + 121 + 100 + 64 + 64 + 49 + 16 + 9 + 4 + 30 = 770. 2019: 306.25 + 313.29 + 299.29 + 225 + 30.25 + 25 + 16 + 9 + 30 ≈ 1,254. HHI was ~1.6× higher after the mergers of Delta and Northwest (2008), United and Continental (2010), Southwest and AirTran (2011), and American and US Airways (2013). (b) Three tests separate structure from fuel. (i) Capacity restraint. Did seat growth (ASMs) run below demand growth? It did, for most of 2010 to 19. The consolidated big four, each with hub dominance to protect, stopped dumping capacity. That's rivalry structurally improving, independent of fuel. (ii) The 2018 stress test. Fuel rose sharply in 2018, and a purely-fuel story predicts a return to losses, but the industry stayed solidly profitable (margins compressed, restraint held). (iii) Ancillary revenue. Unbundled bag and seat fees, a structural price-discrimination change, grew to tens of billions industry-wide, independent of fuel. A fair apportionment says both mattered. Structure explains why 2018-level fuel no longer meant losses, and fuel explains why 2015 to 16 was the peak. One coda. Post-2021 ULCC capacity growth and labor-cost resets (supplier power) eroded domestic economics again. Improved doesn't mean immunized, and your scorecard's watch items exist for this.

Problem 6. Auto chain profit pool (synthetic, per vehicle over life). Compute each link's operating profit per vehicle, each link's share of the chain's profit pool, and each link's share of chain revenue; then identify the two largest revenue-share/profit-share divergences and explain each with a one-line force profile.

Solution. Profits per vehicle are suppliers 7,20,000 × 7% = ₹50,400, OEM 11,60,000 × 9% = ₹1,04,400, dealer 12,20,000 × 1.6% = ₹19,520, financier 2,60,000 × 22% = ₹57,200, insurer 1,50,000 × 4% = ₹6,000, aftermarket 1,40,000 × 18% = ₹25,200. Pool = ₹2,62,720. Chain revenue = 7.2 + 11.6 + 12.2 + 2.6 + 1.5 + 1.4 = ₹36.5 lakh. Shares (profit | revenue) come out as OEM 39.7% | 31.8%, financier 21.8% | 7.1%, suppliers 19.2% | 19.7%, aftermarket 9.6% | 3.8%, dealer 7.4% | 33.4%, insurer 2.3% | 4.1%. Two links diverge most. Dealers take 33% of revenue and 7% of profit. They're a pass-through link, squeezed between a concentrated OEM that sets invoice and margin terms and perfectly informed, comparison-shopping buyers. They're fragmented among thousands of outlets and survive on volume bonuses, F&I commissions, and workshop traffic. Financiers take 7% of revenue and 22% of profit. Concentrated lenders sell a differentiated product (credit approval, convenience at point of sale) to buyers who rarely comparison-shop the loan while they're negotiating the car. Switching hassle and information asymmetry both favor the lender. As an extension, an EV transition shrinks the aftermarket block (fewer wearing parts) and re-concentrates supplier power into cells, so the pool migrates and the dealer/service link is the first casualty. This is the Gadiesh and Gilbert map doing real work. This chain's biggest revenue taker is its second-smallest profit taker.

Problem 7. Classify each synthetic firm's generic strategy (cost leadership / differentiation / cost focus / differentiation focus / stuck in the middle), citing the numbers that decide it. (a) Unit cost 12% below industry average, prices 6% below, share from 31% to 38% over five years, ROCE 22%. (b) Prices 30% above average, gross margin 65% against a peer 45%, A&P + R&D = 18% of sales, share stable at 12%, ROCE 24%. (c) Sells only to cardiac-care hospitals in tier-2 cities, with 3% national share but 45% share of that segment, prices 15% above segment alternatives, ROCE 21%. (d) Prices 10% below the premium player, unit costs 25% above the cost leader, gross margin 38%, share drifting from 14% to 11%, ROCE 9%. (e) The industry's largest player, with the lowest unit costs and a premium brand. Its gross margin declines 60bp over five years while share, sales-per-outlet, and ROCE all rise, with marketing flat.

Solution. (a) Cost leadership. Its structural cost gap (12%) funds under-pricing (6%) and superior returns, and share gains plus high ROCE together are the signature. Price cuts without a cost gap would show share gains with ROCE erosion. (b) Differentiation. Its 30% premium is bigger than the extra cost of being different, as the 20-point gross-margin gap shows, and the 18% A&P/R&D is the differentiation's ongoing price. Share stable at a premium means the premium is being defended, and nobody is milking it. (c) Differentiation focus. It serves a narrow buyer segment, dominates within it (45%) and prices at a premium. National share is irrelevant to the classification. (d) Stuck in the middle. It's neither cheapest (+25% against the leader) nor premium (−10% against the differentiator), and the P&L renders the verdict: mid margin, eroding share, sub-WACC ROCE. Each customer type has a better option. (e) This is the claimed dual advantage, and a fair analysis names the engine instead of the label. Falling gross margin with rising share, throughput, and ROCE on flat marketing is the scale-economies-shared signature. Scale generates a cost advantage that is spent on price or quality, which buys more scale. Classify it as cost leadership whose surplus funds the brand. And note (per 4.12) that such positions are rare and engine-driven, and treat them as hypotheses that need this fingerprint before you believe them.

Problem 8. Fix the industry definition, using the demand-side and supply-side substitution tests: (a) An analyst evaluates a 10-minute grocery app against "Indian retail (₹80 lakh crore)". (b) An analyst benchmarks Titan's Tanishq against Rolex and LVMH because "both are luxury." (c) An analyst concludes Indian aviation is "unconcentrated" because the national HHI is moderate, so no route can sustain fares.

Solution. (a) Both traps at once. "Indian retail" is uselessly broad, since kirana, jewelry, and electronics share no forces. But "10-minute delivery apps" is too narrow, because the marginal basket cross-shops kiranas (phone orders), DMart/scheduled e-commerce, and quick commerce on modest price and convenience differences. By the demand-side test they're one market, urban convenience grocery, with quick commerce as a channel within it. E-commerce majors and organized grocers can also redeploy dark-store assets quickly, which passes the supply-side test, so they belong in the entrant set. Run it twice (with and without kirana) and report whether the pricing-power verdict flips. (b) This fails the demand-side test. A Tanishq buyer facing a 10% price move cross-shops the family jeweller and never looks at Rolex. Most Indian jewelry is sold by the unorganized sector, and that's both the constraining competition and the opportunity, the organized-formalization story. Define it as Indian branded jewelry retail, where Tanishq's forces are trust against the local jeweller, an entrant-and-rivalry story with little to do with global luxury dynamics. Borrowing LVMH's multiples through a "luxury" label is definition-shopping. (c) This inverts the geography trap. Airline pricing happens route by route, and a business-heavy metro pair served by two carriers at HHI 4,000+ absolutely sustains fares (watch corporate-route yields) even while the national number reads 1,250. But complete the thought structurally. Even concentrated routes stay contestable (aircraft are mobile, and the threat of redeployment holds fares down). Airline profitability therefore needs the cost and supplier story on top of concentration. Define the market at the level where prices are set (route), aggregate for cost and fleet analysis (national), and say which question each level answers.

Problem 9. Mirror-image supplier scoring. (a) Score supplier power over US airlines from this packet. Airframes come from a duopoly with 7-year backlogs. Engines come from three makers whose service contracts (the dominant share of engine lifetime cost) carry contractual escalators. Pilot contracts were renegotiated in 2023 at ~+35 to 40% cumulative. Hub airports charge regulated, rising, take-or-pay fees. Fuel = 25 to 30% of opex, globally priced. (b) Score supplier power over Indian paint majors. TiO2 (~15 to 20% of RM cost) comes from six global + two domestic producers. Solvents and monomers are crude-derivative commodities from many sources. Paint majors are among the largest customers of their domestic suppliers. FY22's input spike was fully passed through to consumers within four quarters. Write the one-sentence mechanism for each.

Solution. (a) 5/5. Mechanism: every critical input (airframes, engines, cockpit labor, gates, fuel) is sold by a party more concentrated than the airline industry itself, none depends on any single airline, and several hold contractual escalators, so each upcycle's profits are claimed upstream before equity holders are paid. The evidence anchors are these. Duopoly + backlog = no negotiating alternative at the fleet level. Engine aftermarket = switching locked for the asset's life. Union = concentrated supplier of a non-substitutable input with strike leverage. Fuel = pure price-taking. (b) 2/5. Mechanism: inputs are globally sourced commodities with multiple suppliers in every line, the paint makers are large customers of the sellers and not the other way round, and demonstrated pass-through converts input inflation into a timing problem instead of a margin problem. The residual 2 (not 1) reflects real input-cost cyclicality and the 2 to 4 quarter pass-through lag. Supplier prices can hurt quarters even when supplier power is low. Notice the asymmetry that decides both scores. It's relative concentration and relative dependence, and the size of the input bill doesn't come into it.

Problem 10 (timed, 10 minutes). An airline reports flying revenue of $48bn at a 4.5% operating margin and co-brand remuneration of $6.0bn at a 55% incremental margin. (a) Compute the loyalty share of operating profit. (b) A regulator caps interchange, cutting remuneration 35% at unchanged margin. Compute the fall in total operating profit. (c) One sentence: what does this imply about how the equity should be analyzed?

Solution. (a) Flying: 48 × 4.5% = $2.16bn. Loyalty: 6.0 × 55% = $3.30bn. Total = $5.46bn, and loyalty share = 3.30 / 5.46 = 60.4%. (b) New loyalty profit: 6.0 × 0.65 × 55% = $2.145bn. New total = 2.16 + 2.145 = $4.305bn. Decline = (5.46 − 4.305) / 5.46 = 21.2%. (c) Most of this "airline's" profit comes from a payments and rewards business whose binding risks (interchange regulation, card-partner concentration) are outside the aviation industry map. So the analyst must run two structural analyses and value two economics, or be wrong about both.

Problem 11 (timed, 25 minutes). Full five-forces from a data table: Indian multiplex film exhibition (anchored on the merged leader; figures illustrative). Screens: the leader has ~1,750 of ~3,200 multiplex screens (~55%), and there are ~6,000 single screens nationally. Occupancy is ~25%. An average ticket is ₹260, rising 5 to 7%/yr. F&B spend per head is ₹132 at ~75% gross margin. Film rental (the distributor share) is ~45 to 50% of net box office in release weeks. Top 10 films ≈ 35 to 40% of annual box office. Leases are mall revenue-share deals with minimum guarantees. Theatrical windows run 4 to 8 weeks, streaming costs ₹150 to 300/month, and capex is ₹2.5 to 3 crore per new screen. Mall developers actively court anchor multiplexes. Produce five scores with one evidence line each, the verdict (expected ROCE band against a ~12% hurdle), the binding force, and two watch items.

Solution (model answer; your scores may differ by ±0.5 with good reasons; the verdict logic should not). Rivalry 2.5. Exhibition itself is consolidated, with the leader at ~55% of multiplex screens, and ticket prices rising above inflation show an umbrella. Single screens do cap pricing in price-sensitive markets. Entry 3. Capex per screen is modest and financeable, but prime mall anchor slots are scarce and pre-committed to the leader, which is a distribution barrier. Regional entry keeps happening. Substitutes 4.5, the Problem-4 trajectory: a ~100:1 cost-per-hour gap, a shrinking window, footfalls below pre-2020. The substitute binds on frequency and on the marginal film. Buyer power 3.5. Buyers are fragmented, which is good, but they're perfectly price-elastic with a free alternative at home. Occupancy at 25% despite blockbusters filling halls shows it. Demand is hit-driven, and habit has little to do with it. Supplier power 4.5, because the two suppliers that matter both hold the pen. Content is hit-concentrated (top 10 films = ~35 to 40% of box office), and the distributors of those same hits set rental terms (~half of box office). Mall landlords take revenue share with guaranteed minimums, a fixed-cost floor under a variable business. Sum ≈ 18/25. Call it structurally strained. Expect through-cycle ROCE in the mid-to-high single digits, below the ~12% hurdle, with profitable years driven by content cycles. (Check it: listed exhibitors' actual ROCE history looks like this; check on screener.in.) Binding force: supplier power (hit content) compounded by the substitute. Concentration in exhibition can't fix either. A 55% "share" of showing other people's films, to audiences who now have the films at home eight weeks later, prices like the weak link it is. For watch items, track (1) theatrical-window contracts, where re-lengthening improves the substitute score and further shrinkage worsens it, and (2) the F&B + premium-format revenue mix, the exhibitor's owned, high-margin pool, whose growth is the de-commoditization escape route. As a bonus, (3) any exhibitor move into content co-production is an attempt to integrate toward the pool. Once more, the meta-lesson: market leadership isn't a thesis. Structure is.

Problem 12 (guided). Import-parity ceiling (synthetic; all figures illustrative). A domestic commodity industry (three producers, HHI ≈ 3,100) sells at ₹47,000/tonne. The marginal import has a CIF price of $520/tonne; basic customs duty is 7.5%; the rupee is ₹85/$. Method: landed cost = CIF × (1 + duty) × FX, ignoring minor clearing costs. (a) Compute the landed cost and say whether imports cap the domestic price. (b) The rupee strengthens to ₹80/$. Recompute, and say what happened to the ceiling and to domestic margins. (c) What effective duty would lift the landed cost to 15% above the current domestic price? (d) One sentence: what should this industry's five-forces scorecard list as its top two watch items, and why is neither about the three producers?

Solution. (a) 520 × 1.075 × 85 = ₹47,515/t. The domestic price (₹47,000) is ~1.1% below landed. The industry is already pricing right at the import-parity ceiling, so despite HHI 3,100 it has essentially no room to raise prices without inviting imports. Imports are the binding constraint. (b) At ₹80/$: 520 × 1.075 × 80 = ₹44,720/t. The ceiling fell ₹2,795/t purely from FX. The domestic price is now ~5.1% above parity, so imports undercut and domestic margins compress. A currency move did the damage, with no help from any competitor. (c) Need 520 × (1 + d) × 80 = 47,000 × 1.15 = ₹54,050 → (1 + d) = 54,050 / (520 × 80) = 1.299 → d ≈ 29.9%. (d) "The exchange rate and the customs/anti-dumping tariff." For a tradable good the ceiling is a landed cost, so the border (FX × duty) sets how high the industry can price, whatever the domestic share table says. Concentration measures the room. It doesn't set the ceiling.

Problem 13 (guided). Merger arithmetic (illustrative). A market has six competitors with shares A 21, B 19, C 18, D 16, E 12, and a fragmented tail contributing ~14 HHI points. Firms C and D propose to merge. Method: HHI = Σ sᵢ²; the increase from a merger of two firms is exactly ΔHHI = 2·sᵢ·sⱼ. (a) Compute pre-merger HHI. (b) Compute post-merger HHI two ways (rebuild it, and via the ΔHHI shortcut). (c) Classify pre and post under the 2010 and 2023 thresholds. (d) Under the 2023 guidelines' structural presumption (post-HHI > 1,800 and ΔHHI > 100), is this merger presumptively anticompetitive, and what would you check before treating that as the whole story?

Solution. (a) HHI_pre = 21² + 19² + 18² + 16² + 12² + 14 = 441 + 361 + 324 + 256 + 144 + 14 = 1,540. (b) Merged C+D = 34. Rebuild: 34² + 21² + 19² + 12² + 14 = 1,156 + 441 + 361 + 144 + 14 = 2,116. Shortcut: ΔHHI = 2 × 18 × 16 = 576, and 1,540 + 576 = 2,116, which checks out. (c) Pre 1,540: 2010 moderately concentrated (1,500 to 2,500), 2023 moderately concentrated (1,000 to 1,800). Post 2,116: 2010 moderately concentrated (still <2,500), 2023 highly concentrated (>1,800). (d) Yes. Post-HHI 2,116 > 1,800 and ΔHHI 576 > 100, so it clears the structural presumption, and the merged 34% share > 30% also trips the 2023 share-based presumption. Before you treat that as the verdict, check the idea this whole argument rests on: concentration is an input to one force. Ask whether powerful suppliers, a cheap substitute, or import parity already cap the industry's returns. If they do, blocking the merger protects competition that structure was suppressing anyway. Antitrust asks "does this lessen competition?" An analyst asks "does the industry earn its cost of capital?" Those two questions are related, but they aren't the same.

Problem 14. ROIC persistence (synthetic). Firm earns 26% ROIC against an 8% WACC; model its excess spread as spreadₜ = spread₀ × 0.85ᵗ (0.85 illustrative). (a) Compute the spread (in points) and the implied ROIC at years 3, 7, and 10. (b) Compute the half-life of the spread. (c) In which year does the spread first fall below 5 points? (d) A rival firm starts at the same 26% ROIC but its edge decays at 0.70/yr; without recomputing the whole table, state which firm an acquirer should pay more for and why, and connect the answer to why the evidence on industry effects says to score the industry before the firm.

Solution. (a) spread₀ = 26 − 8 = 18 pts. Year 3: 18 × 0.85³ = 11.05 pts (ROIC ≈ 19.1%); year 7: 18 × 0.85⁷ = 5.77 pts (≈ 13.8%); year 10: 18 × 0.85¹⁰ = 3.54 pts (≈ 11.5%). (b) Half-life = ln(0.5)/ln(0.85) = 4.27 years. (c) 18 × 0.85⁷ = 5.77 (still ≥5) and 18 × 0.85⁸ = 4.90 (<5) → year 8. (d) The first firm (decay 0.85). A slower decay means a longer half-life, so more of the excess-profit stream survives to be discounted into today's value. The present value of a spread that fades at 0.85 is far larger than one fading at 0.70, even though this year's profit is identical. That's the evidence on industry effects in miniature. Usually the durable, slow-decaying component of returns is the industry-structural one, so the analyst reads the neighborhood (the persistent base rate) before crediting the firm. Persistence is what capitalizes. Starting numbers don't.

Problem 15. Profit pool for a ₹250 branded-coffee cup (synthetic, per cup). Compute each link's operating profit per cup, its share of the chain profit pool, and its share of chain revenue; then name the largest revenue-share/profit-share divergence and give its one-line force explanation.

Solution. Profits per cup are farmer 18 × 7% = ₹1.26, processor 9 × 5% = ₹0.45, roaster 40 × 24% = ₹9.60, café 183 × 7% = ₹12.81. Pool = ₹24.12/cup. Chain revenue = 18 + 9 + 40 + 183 = ₹250. Shares (profit | revenue) are café 53.1% | 73.2%, roaster 39.8% | 16.0%, farmer 5.2% | 7.2%, processor 1.9% | 3.6%. The largest divergence is the roaster and brand owner. It touches only 16% of the cup's price but captures ~40% of its profit, because the brand + roast-and-blend IP is a differentiated, hard-to-substitute link, a mini-Apple at the "brand" end of the smiling curve. The café is the mirror image, revenue-heavy and profit-thin. It carries the chain's operations cost sink (rent, labour, perishable milk) at a commodity-service margin, which makes it the "assembly" seat of this chain. The lesson repeats: the link that touches the most money is rarely the one that keeps it. Watch the migration too. Single-serve and at-home formats (pods, instant) shift volume from the café seat toward the roaster and brand seat, so the pool moves toward the fatter end.

Problem 16. Substitute crossover (synthetic; US$). A US industry sells a service at $100 per unit-of-job, rising ~5%/yr with inflation. A technological substitute delivers the same job at $180/unit today, but its cost is falling ~18%/yr. (a) Write each trajectory and solve for the crossover year t* (when the substitute becomes cheaper). (b) Compute the first whole year in which the substitute is cheaper, showing the two prices. (c) Is a snapshot ("the substitute is 80% more expensive today") a reason to score the substitute force low? Score it and justify in one sentence. (d) Name the cohort that defects first.

Solution. (a) Incumbent 100 × 1.05ᵗ; substitute 180 × 0.82ᵗ. Set equal: t* = ln(100/180) / ln(0.82/1.05) = 2.38 years. (b) Year 2: substitute 180 × 0.82² = $121.0 > incumbent 100 × 1.05² = $110.25. Year 3: substitute 180 × 0.82³ = $99.2 < incumbent 100 × 1.05³ = $115.8 → year 3. (c) No. Score it high (≈4/5) and rising, because the decisive question is the trajectory and two curves closing at a combined ~23%/yr cross in under 3 years despite today's 80% gap. A static "80% more expensive" reading is the same trap that valued multiplexes on 2019 footfalls and print classifieds on 2004 revenue. (d) The price-sensitive, low-switching-cost cohort defects first (leisure before contracted corporate, casual before power-user), so the erosion shows up in the mix and in volumes years before the headline share moves.

Problem 17 (timed, 20 minutes, stretch). Full mini-analysis from a data table: Indian tyres (synthetic packet; all figures illustrative). Shares (replacement + OEM, by value): MRF 24, Apollo 21, CEAT 15, JK 15, and imports + global brands ≈ 25 (treat as ~five ~5-point players). Natural rubber + carbon black + crude-derivative inputs ≈ 60–65% of COGS, imported/priced globally, cyclical, historically passed through with a 2–3 quarter lag. OEM channel: a handful of concentrated automakers, annual price-downs, ~35% of volume. Replacement channel: fragmented, brand-influenced, ~65% of volume, higher realisations. There's no substitute for a tyre, though retreading exists in commercial. Plants are capex-heavy but financeable, and global majors are already present. Produce (i) HHI and CR4, (ii) a 3-link profit pool from the table below (profit share of each), (iii) five force scores (one evidence line each), the binding force, and an expected through-cycle ROCE band against a ~12% hurdle, and (iv) two watch items.

Solution. (i) HHI = 24² + 21² + 15² + 15² + (5 × 5²) = 576 + 441 + 225 + 225 + 125 = 1,592 and CR4 = 75. That's moderately concentrated under both the 2010 (1,500 to 2,500) and 2023 (1,000 to 1,800) thresholds. (ii) Pool = 38×10% + 48×11% + 14×5% = 3.8 + 5.28 + 0.70 = 9.78. Profit shares: manufacturer 54.0%, inputs 38.9%, dealer 7.2%. The dealer's 14% of revenue collapses to 7% of profit, because it's the fragmented, price-taking replacement conduit. (iii) Rivalry 3. It's moderately concentrated: four balanced majors, a semi-commoditised product. It's restrained in good years and share-fought in the OEM channel. Entry 2.5. Plants are capex-heavy but financeable, and global majors are already in. Entrants are really blocked by the replacement-channel dealer network and brand (MRF), and not by the factory. Substitutes 1.5. Nothing replaces a tyre, and retreading nibbles at the commercial segment only. Buyer power 3.5, bifurcated. OEMs (concentrated, annual price-downs) score ~4.5 and the fragmented replacement market ~2, weighted by ~35/65 volume. Supplier power 3.5. Rubber + carbon black + crude derivatives are 60 to 65% of COGS. They're globally priced and cyclical. Power is moderate, but cost volatility is high, and pass-through lags 2 to 3 quarters, so margins whipsaw. Sum ≈ 14/25. Binding force: input-cost cyclicality (supplier side) compounded by OEM buyer power. Manufacturers hold the biggest profit slice but earn it on the industry's most volatile cost base. ROCE is therefore a cyclical high-single to low-double digits, oscillating around the ~12% hurdle instead of sitting durably above it. Verify against listed tyre-makers' 10-year ROCE on screener.in, where the swings are the signature. (iv) For watch items, take (1) the natural-rubber and crude price cycle plus the ₹/$ rate, which set the cost base and the pass-through lag, and (2) the OEM against replacement mix. A shift toward the fragmented, higher-margin replacement channel is the de-commoditisation escape route, while EV adoption (fewer service touchpoints, different tyre wear) reshapes the aftermarket. Here's the meta-lesson again: a mid-HHI industry with a fat manufacturing profit share can still be a cyclical, hurdle-hugging business, because the seat that holds the profit also holds the volatility.

Problem 18 (life cycle, three industries). Classify each of the three by stage (embryonic, growth, shakeout, mature, decline) and name the two pieces of evidence that decide it. Then say which of the five forces the stage tells you to score first. Figures are illustrative and shaped to the public record; verify each against the sources named before you use any of it on a live name.

• A. Indian organized decorative paints. Volume growth is around high single digits, and the top four hold roughly 90 points of share. Capital expenditure has run near depreciation for years across the incumbents. One entrant committed about ₹10,000 crore from 2024, and industry margins compressed inside a year. Sources: company annual reports and investor presentations, and the industry chapter of any recent sector DRHP.
• B. Indian wireless telecom, FY16 to FY25. Subscriber shares moved 100 points in total across five lines over nine years, which is 2.2 points per firm per year on the share-stability metric. Two operators entered insolvency, one survives on state support, and the field went from a dozen names to three plus a state carrier. Capital expenditure ran far above depreciation throughout. Sources: TRAI's monthly subscription reports, free at trai.gov.in.
• C. Indian quick commerce. Revenue is compounding at rates well above 50% a year. Contribution margins are crossing from negative to slightly positive, on mature dark stores only. New entrants are still arriving, and they're funded. Capital expenditure is several times depreciation, and nobody's share is stable for two consecutive quarters. Sources: the listed players' quarterly investor presentations and the DRHPs of the unlisted ones.

Solution. A is mature. It's decided by the frozen share table (90 points in four hands, movement under one point per firm per year on the share-stability metric) together with capital expenditure at roughly one times depreciation. That's the signature of an industry replacing capacity instead of adding it. Growth near nominal GDP backs it up. Score threat of entry first, because a mature structure's whole verdict rests on whether the barrier holds, and a funded entrant is testing it right now. That 2024 margin compression is the first datapoint of the test. It doesn't answer it. B is a shakeout that has just ended, arriving at mature. Look at the 2.2 points per firm per year of share movement, four times the mature threshold, alongside the collapse in player count with two insolvencies. Capital expenditure far above depreciation puts it in the capacity-arriving phase of the capital cycle, which covers the same years. Score rivalry first, because a shakeout is rivalry doing its work. Ask whether the surviving three now price like an oligopoly or keep fighting, because that answer decides whether the next decade is mature or a second shakeout. C is growth, not yet shakeout. Its deciding evidence is the pair of 50%-plus revenue growth and capital expenditure at several times depreciation, with entrants still arriving. Unstable share confirms nobody has built anything defensible yet. Score threat of entry first and score it high, and note the caution the stage imposes. Rising share here is close to worthless as moat evidence, so the only useful question is what the mature-store contribution margin looks like at scale. That's a unit-economics question. The trap in C is granting a moat during a growth stage. Evidence that would matter hasn't been generated yet, because the shakeout hasn't happened.

Applied mini-project: the Industry Map (this module's gate)

Run the full apparatus on ONE industry you haven't studied deeply before. Choose from (a) Indian QSR (Jubilant FoodWorks, Devyani, Westlife, Sapphire, plus the delivery platforms' disclosures), (b) Indian cement in one region (UltraTech, Shree, Dalmia + a regional player; define the region first), (c) US home-improvement retail (Home Depot, Lowe's 10-Ks), or (d) Indian hotels (Indian Hotels, EIH, Lemon Tree + industry supply data). They're on the list because each hides a definition question, a non-obvious binding force, and a mappable chain.

What to pull (all free). Start with the last two annual reports or 10-Ks of the top 3 to 4 players: Item 1/1A or MD&A, segment notes including the ≥10%-customer disclosure, and the cost lines for supplier analysis. Add the most recent investor presentation of each (share and capacity data). Pull 10-year ratio histories from screener.in or stockanalysis.com, which are your ROIC ground truth. Get one regulator or industry dataset (RBI/CRISIL summaries, cement dispatch data, STR/HVS hotel supply summaries, NRAI/platform filings for QSR). And take Damodaran's industry margin/ROIC tables for the base rate.

Deliverable (max 2 pages + exhibits).

1. Industry definition, in one paragraph: boundary chosen, both substitution tests applied, and (if the boundary was close) the run-it-twice result.
2. The map, one exhibit, Mauboussin-style: suppliers → rivals (with shares) → buyers, plus entrants, substitutes, and complements, with each participant's approximate operating margin or ROIC annotated (pure-play proxies where needed).
3. The scorecard: five forces, each scored 1–5 with 2–3 cited evidence items (filing page/note or dataset named), each with a "what would change this" line.
4. The profit-pool table, ≥4 links: revenue, margin, profit, share of pool. Add one paragraph on why the pool is where it is and where it's migrating.
5. The verdict: the expected industry ROIC band against WACC, plus the binding force in one sentence. Then reconcile it against the actual computed ROICs of the players. If structure and returns disagree, resolve the disagreement, because that's the heart of the exercise.

Scoring rubric. Self-score after writing. A pass is ≥13/16, per the program's written-deliverable protocol.

File the map in your knowledge system. Phase 5 will reuse it, and Phase 8's Day-2 industry map is this same deliverable at speed.

Case Lab: run case CL-4.2 in the app's Cases tab: a guided, gradeable industry-structure lab (score the forces from a supplied evidence packet, isolate the binding force, and reconcile your verdict against the players' computed ROICs).

Reading & resources

• Porter, "The Five Competitive Forces That Shape Strategy," HBR, January 2008. [Paid article; often accessible via library databases] [Beginner→Intermediate]. The definitive statement, by the author, 29 years wiser. Read this before the 1980 book. The ROIC-by-industry exhibit alone is worth the price.
• Porter, Competitive Strategy (1980), chs. 1–3 and 7. [Paid] [Intermediate]. Ch. 1 structural analysis in full; ch. 2 the generic strategies and "stuck in the middle"; ch. 7 structural analysis within industries, the segment-level analysis of section 4.15.
• Porter, Competitive Advantage (1985), chs. 1–2. [Paid] [Intermediate]. The value chain, from the source.
• Porter, "What Is Strategy?", HBR, November–December 1996. [Paid] [Intermediate]. Activity systems, fit, and why straddling fails: the mature version of the generic-strategies argument, with the Southwest map.
• Mauboussin & Callahan, Measuring the Moat (Morgan Stanley Counterpoint Global; latest edition). [Free PDF, search the title] [Intermediate]. Read the industry-analysis and industry-map sections now (pp. on lifecycle, share stability, and the map); the moat sections are M4.04–M4.05's text.
• Greenwald & Kahn, Competition Demystified (2005), chs. 1–2 now. [Paid] [Intermediate]. The deliberate simplification of all this: entry barriers as the only force that matters. Full treatment in M4.05.
• Gadiesh & Gilbert, "Profit Pools: A Fresh Look at Strategy" and "How to Map Your Industry's Profit Pool," HBR, May–June 1998. [Paid] [Intermediate]. The method of section 4.14, with the original auto-industry map.
• McGahan & Porter, "How Much Does Industry Matter, Really?" SMJ, 1997; Rumelt, "How Much Does Industry Matter?" SMJ, 1991; Schmalensee, "Do Markets Differ Much?" AER, 1985. [Free via Google Scholar] [Advanced]. The variance-decomposition trio of section 4.1. Skim it for the tables and leave the econometrics.
• Rosenzweig, The Halo Effect (2007). [Paid] [Beginner]. The corrective for management-attribution bias. Chapters 1 to 5 are enough.
• Buffett, Berkshire Hathaway shareholder letters, 1985 (the textile post-mortem) and 2007 ("the great, the good and the gruesome," with the airline/Kitty Hawk passage). [Free at berkshirehathaway.com] [Beginner]. Structure-first investing in its native prose.
• IATA, "Profitability and the air transport value chain" (Economics Briefing No. 10, 2013) and the updated value-chain profitability analyses (2024). [Free at iata.org/economics] [Intermediate]. The airline profit-pool numbers of section 4.10, with methodology.
• Damodaran's data library: "Margins by Sector," "ROE/ROIC by Industry" (US + emerging markets; updated each January). [Free at pages.stern.nyu.edu/~adamodar] [All levels]. Your standing base-rate tables; pull them for every industry you ever map.
• Counterpoint Research quarterly handset profit-share notes. [Free summaries] [Intermediate]. The smartphone pool's live scoreboard.
• Levitt, "Marketing Myopia," HBR, 1960. [Paid] [Beginner]. The too-narrow trap, in the original.
• Filings & datasets for the cases here: Asian Paints, Berger, Kansai, Akzo India, Indigo Paints, Grasim/Birla Opus ARs + investor decks (BSE/NSE, company IR); Pidilite AR; Delta/United/American/Southwest 10-Ks (EDGAR: Item 1, 1A, MD&A, the loyalty notes); IndiGo (InterGlobe Aviation) ARs; DGCA monthly domestic traffic data; US DOT/BTS T-100 and Form 41; TRAI quarterly performance-indicator reports; screener.in and stockanalysis.com. [All free]
• Business Breakdowns (Colossus) podcast. [Free] [All levels]. Search the archive for the paints, adhesives, semiconductor, and airline-adjacent episodes. Each is a one-hour oral industry map. Listen after writing your own, so it calibrates your view instead of anchoring it.
• Stratechery (Ben Thompson): the free essays on Aggregation Theory. [Free tier] [Intermediate]. Profit-pool migration in internet value chains, and where the buyer-power force went digital.
• Going deeper: India's Directorate General of Trade Remedies (DGTR) "final findings" and the CBIC customs tariff. [Free at dgtr.gov.in, cbic.gov.in] [Intermediate]. Real anti-dumping and safeguard investigations lay out the landed-cost, normal-value, and injury-margin arithmetic behind worked example 5 and problem 12. These are the documents in which policy sets (and resets) an import-exposed industry's price ceiling, a "structural watch item". Read one steel or bulk-chemical case end to end and you'll never again call a tradable industry "safe" on its HHI alone.
• Going deeper: Pankaj Ghemawat, Commitment: The Dynamic of Strategy (1991); with Wiggins & Ruefli, "Sustained Competitive Advantage… Persistence of Superior Economic Performance" (Organization Science, 2002, free via Scholar). [Paid book] [Advanced]. The theory and the evidence behind the persistence argument and worked example 6: why some return advantages persist for years while others fade in months (commitment and sunk, hard-to-reverse choices), and how the fade rate, not this year's spread, is what capitalizes into value. The rigorous version of "score the durable industry prior first."

The Modern Analyst's Addendum

Everything above teaches this skill from first principles, by hand — that is how you learn it, and the mastery check still tests it that way. This addendum shows how a working analyst amplifies the same skill today. It adds; it never replaces. (R1/R10)

AI-Augment this skill

{ "skill": "Building and re-running an HHI/CR4/delta-HHI concentration table from a filing or a merger notification instead of re-deriving the arithmetic by hand every time", "use": "Feed an LLM a filing's market-share disclosures (or a competition-authority order) and have it extract a clean share table, then run it through a small concentration function so you can instantly re-score alternative entry or merger scenarios (who loses share to whom) instead of re-squaring numbers on a calculator each time.", "tools": ["Claude or ChatGPT to extract share tables from annual reports, investor decks, or CCI/DOJ merger orders", "a notebook with a small HHI/CR4/delta-HHI function (the block below)", "CCI combination-order PDFs (cci.gov.in) or DOJ/FTC complaint exhibits, which often publish their own HHI workings to check the function against"], "prompt": "Here is a market-share table for [industry] extracted from [source]. Compute HHI, CR4, and classify concentration under the 2010 and 2023 DOJ/FTC thresholds. Then model two entry scenarios: (a) the new entrant's share is taken proportionally from every incumbent, (b) it is taken mostly from the smallest/tail players. Report the resulting HHI under each and flag which incumbents are most exposed in each scenario.", "verify": "Cross-check the AI's share table against the actual filing or order line by line -- market-share figures are exactly the kind of number a model can misquote confidently -- and recompute the HHI and at least one delta-HHI scenario by hand before using either in a memo.", "diy": "The mastery check hands you a share table with no tool in the room and asks for HHI, CR4, and a concentration classification by hand. You must also be able to state, unaided, which parts of a five-forces scorecard are numbers and which are judgment -- the function above only speeds up the arithmetic half." }

Modern Data Analysis

By hand first. HHI/CR4 arithmetic and the merger delta-HHI shortcut (ΔHHI = 2·sᵢ·sⱼ) haven't changed, and they're what the mastery check gates.

Today's workflow. Once you can compute HHI by hand, stop re-deriving it for every scenario a client or a memo asks about ("what if the entrant takes share from the leader instead of the tail?"). Write it once as a function instead, then re-run it against as many share tables and entry assumptions as the analysis needs. That's how the Birla Opus stress test turns into a quantified range instead of a single illustrative paragraph.

Tools & sources (IN + US). In India, start with the Competition Commission of India's "Combination Orders" (cci.gov.in). CCI publishes its own HHI/CR calculations for every merger it reviews, which gives you a free, real, checkable answer key. MCA21 and annual-report segment notes cover customer and supplier concentration (the Ind AS 108 "single external customer ≥10% of revenue" disclosure). In the US, the DOJ/FTC 2023 Merger Guidelines and the Antitrust Division's public complaint exhibits (justice.gov/atr) show real HHI workings in live cases. EDGAR 10-K Item 1 "Competition" and the ASC 280 major-customer note cover customer concentration, and company investor decks carry supplier and vendor concentration disclosures.

# recomputed in-session -- R3
def hhi(shares_pct):
    return sum(s**2 for s in shares_pct)

def cr(shares_pct, n):
    return sum(sorted(shares_pct, reverse=True)[:n])

def delta_hhi_merger(s_i, s_j):
    return 2*s_i*s_j

def entry_scenario(shares_dict, entrant_share, mode="proportional", flank_from=None):
    """mode='proportional': entrant dilutes every incumbent equally.
       mode='flank': entrant's share comes entirely out of flank_from (a list of keys)."""
    shares = dict(shares_dict)
    if mode == "proportional":
        scale = (100 - entrant_share) / sum(shares.values())
        shares = {k: v*scale for k, v in shares.items()}
    else:
        remaining = entrant_share
        for k in flank_from:
            take = min(shares[k], remaining)
            shares[k] -= take
            remaining -= take
    shares["Entrant"] = entrant_share
    return shares, hhi(shares.values())

# --- anchor: the module's own Sec 4.9 decorative-paints shares, tail modeled as six ~1%
# players -- matching Sec 4.15 Problem 1's own resolution of "tail ~6%", not one 6% block ---
paints = {"Asian Paints": 53, "Berger": 18, "Kansai Nerolac": 13, "Akzo Nobel": 7, "Indigo": 3,
          "tail-1": 1, "tail-2": 1, "tail-3": 1, "tail-4": 1, "tail-5": 1, "tail-6": 1}
hhi_paints = hhi(paints.values())          # -> 3,366 (matches Sec 4.9 and Sec 4.15 exactly)
cr4_paints = cr(list(paints.values()), 4)  # -> 91

# --- extension: the Birla Opus entry, two ways ---
prop_shares, hhi_prop = entry_scenario(paints, entrant_share=8, mode="proportional")
tail_keys = ["tail-1", "tail-2", "tail-3", "tail-4", "tail-5", "tail-6"]
flank_shares, hhi_flank = entry_scenario(paints, entrant_share=8, mode="flank",
                                          flank_from=tail_keys + ["Indigo", "Akzo Nobel"])
# hhi_paints -> 3,366 ; hhi_prop -> 2,913 (-453) ; hhi_flank -> 3,416 (+50)

Verify. hhi(paints.values()) returns 3,366, the Sec 4.9 figure, once the "tail ~6%" is modeled the way Sec 4.15's Problem 1 solution resolves it: six ~1% players, each contributing 1² = 1, and not one block contributing 6² = 36. cr(..., 4) returns 91, the CR4 the paints case doesn't state explicitly but which follows directly from its own share table. Those two entry scenarios turn the Birla Opus paragraph into two concrete numbers instead of one qualitative "open question," and the contrast between them is the lesson. If the entrant dilutes everyone proportionally, HHI falls 453 points to 2,913, which is still highly concentrated, though visibly less so. If it flanks the fragmented tail and Indigo first, which is closer to the paints case's own account of converting Birla White's counters instead of fighting Asian Paints head-on, the big four are untouched, yet HHI rises 50 points to 3,416. HHI rewards fragmentation among small players, so consolidating six tiny competitors' share into one 8%-share entrant is itself a concentrating move, whatever it does to competitive intensity among the leaders. Both are real, computable consequences of one filing's share table. Which one is actually happening gets answered by dealer-count and channel-check evidence. No formula can tell you.

Quantitative lens

Here's what you can measure and what stays judgment, stated plainly. That distinction is the modern contribution here. A bigger spreadsheet isn't.

Concentration recomputed as the verification anchor. Take the decorative-paints shares (53/18/13/7/3/tail~6, with the tail modeled as six ~1% players per Sec 4.15's own Problem 1 resolution) and square every term. You get HHI = 3,366, which matches the figure stated in the paints case. It's still worth running the check by hand every time, even when it matches. Had the tail been one undifferentiated 6% player, the same formula would give 3,396, a 30-point (~0.9%) difference. That happens not to change the "highly concentrated" classification here, since both numbers are deep past the 2010 >2,500 and 2023 >1,800 DOJ/FTC lines. On an industry closer to a threshold it easily could, so always state how you modeled a fragmented tail as well as its total share. CR4 = 91 follows from the same table, however the tail is split.

Supplier concentration, made numeric. In the airline case, the Boeing/Airbus airframe duopoly is part of the supplier-power force. You can put an HHI on that claim too. Illustrative global narrowbody and widebody delivery shares of Boeing ~45%, Airbus ~50% and others ~5% give HHI ≈ 4,550 and CR2 ≈ 95. That's a more concentrated market than either the paints or the airlines case it feeds into. You've turned "supplier power is high" from an adjective into a number you can re-check against updated delivery-share data (Boeing/Airbus quarterly delivery reports, EDGAR 10-K Item 1). It's the same move as the merger-delta exercise above, applied one link further up the value chain.

Limits. HHI and CR4 measure the revenue-share structure of a market as it's currently drawn. They say nothing about the assets' mobility or financeability. Think back to the contestability caveat, with US airlines at HHI>4,000 still earning below WACC across cycles. They say nothing about buyer- or supplier-side counter-concentration either. And everything they report depends entirely on how the "relevant market" was defined. Compute the same paints shares on "all coatings including industrial" or "organized plus unorganized" and they'd read very differently, and the index can't tell you which definition is right. The entry-scenario model above is a mechanical sensitivity exercise. It shows what the arithmetic would say under two assumptions about who loses share. It can't show which assumption will actually hold at Asian Paints, Berger, or Akzo Nobel in FY26 to FY28. That's still the open question the paints case says it is.

Where this goes next: galaxy cross-links

• G1 · EC1.01 (microeconomics-market-structures): the market-structure theory (oligopoly, contestability) underneath HHI's DOJ/FTC thresholds.
• G3 · CN3.01 (competitive-and-industry-frameworks): the consulting-side five-forces and structure-analysis toolkit that the method here mirrors.
• G2 · DA1.02 (pandas-data-wrangling): scales the concentration functions above from one filing's share table to a multi-year, multi-industry panel.

Teach it back & journal

Feynman prompt. Explain industry structure to a smart 15-year-old using two lemonade stands. Stand one is on a crowded beach with thirty identical stands. Anyone can set up tomorrow (entry). Every customer sees every price (buyer power). Ice and lemons come from the beach's only kiosk, which charges what it likes (supplier power). Sea breeze is free and cold (substitute). And by sunset unsold lemonade is worthless (perishable capacity), so there's a price war at 5pm. Stand two is inside the only gate of a park. Here the gate is the barrier, the stroller-pushing parents don't comparison-shop, suppliers compete to stock it. Same lemonade, same effort, same talent, yet different fates, decided by the seat and not the seller. Then have her answer the killer questions: who takes the money at each stand before the owner sees it, and if you could make one structural change to the beach stand (not a better recipe, but a change to the beach), what would it be? If she says something like "get the town to license only three stands" or "buy the ice kiosk," she's understood barriers and supplier power better than most adults. Finish with the one-page explainer: "An industry is a neighborhood. Five claimants stand between the value a business creates and the profit it keeps. Score each claimant with evidence, find the one that binds, check the answer against the returns everyone actually earns, and only then ask whether this particular firm can beat its neighborhood."

Journal reflection. You work in, or have worked closely with, an industry too. Score the five forces on it from lived experience plus anything public. Work out who the buyers are and how concentrated they are, whether a handful of large accounts or many small ones. Ask what switching costs really amount to when a client changes providers. Note which substitutes are improving (in-house tooling, software platforms, AI) and on what trajectory. Sketch the entry ledger by thinking of every experienced person who left to start a competitor, and whether they took share. And find where the profit pool is along the chain from independent operator to boutique to global firm. Write the scorecard, the binding force, and the verdict. Then record (1) which force you feel every week but had never named, (2) where you would have mis-scored your own industry before now, and what evidence corrected you, and (3) one sentence on what this implies for any services firm you ever analyze. File it. Phase 5 will ask you to do this for seventeen industries you don't live in, and knowing what the forces feel like from inside is calibration no dataset provides.

This module's flashcards and mastery quiz are wired into the app: see the node's Quiz and Reviews.