Double down or cut? Where the next dollar goes across business units

1. Before you start

Portfolio resource allocation is the decision a group makes when it has one pot of capital and several business units that all want it: where does the next dollar go? The honest answer is not “the biggest unit” or “the one that shouts loudest” — it is the unit where the next dollar earns the most, as long as that beats what the money costs the group. A tiny example: you have $20,000 spare. Unit A would turn it into $24,400 next year (a 22% return); Unit B into $22,200 (11%). Your money costs you 10%. The next dollar goes to A, and it keeps going to A until A’s next dollar earns less than B’s — then it switches. That switching rule is the whole subject.

Three honest statements before you start:

This is a Decide course. You read a situation, learn the idea, and make a call. You do not write or run any code.
The group below, Halcyon Group, and its divisions are composite — an invented conglomerate built from ordinary figures so the reasoning is clean. No number is a claim about any real company.
This is not a certification. It proves, to you, that you can defend where a group puts its capital.

You need arithmetic and the nerve to say no to a powerful division head. The hard part is judgment.

2. The Situation

Halcyon Group is a diversified conglomerate, and next year it has one $60M block of capital to put to work across its divisions — nowhere near enough for everything the division heads are asking for. The head of Hearth & Home, its biggest business by revenue, wants it: “we are half the group, fund us.” The head of Cobalt Chemicals wants it too: “we just sank $80M into a new plant — we can’t stop now.” You sit on the capital committee, and you have to decide where the money goes and defend it to a room of people who each believe the answer is them.

The trap is that the loudest argument and the biggest business both point the wrong way, and the one number that should decide it — what the next dollar earns in each division — is the one nobody in the room has put on the table.

3. What you’ll be able to do

After this course you will be able to:

Rank business units by the marginal return on their next dollar of capital, and put the money where it earns the most above the group’s hurdle rate — not where politics or size point.
Explain why funding the biggest unit or the loudest division head destroys value, and put a number on what that mistake costs.
Use growth-share and attractiveness-versus-position thinking as a frame, and say where each frame stops being the rule.
Decide when to harvest or cut a unit and redeploy its capital, and separate that call cleanly from sunk cost and internal politics.

4. Prerequisites & time box

Prerequisites: arithmetic, and comfort reading a small table of returns. Helpful but not required: the idea that capital has a cost (a hurdle rate the group must clear). No spreadsheet needed. If the terms hurdle rate and return on capital are new, they are defined in section 6.

Time box: about 20 minutes of reading (measured), plus real thinking time on the call — this one rewards arguing the numbers back to yourself. That is under the 25-minute cap for a concept course.

Difficulty: 6 / 8 — a manager-level call: several divisions move at once, the loudest and biggest arguments both point the wrong way, and you have to reason past them to the one number that decides it. A simpler version would hand you two units and the returns; here you weigh several, strip out sunk cost and politics, and defend a cut.

5. The case & where the numbers come from

Halcyon Group is a composite conglomerate: its divisions and every figure attached to them are in-course assumptions chosen for clean reasoning, not drawn from or claimed about any real firm. The ideas — marginal return on capital, hurdle rate, the growth-share matrix, the attractiveness-position grid, sunk and opportunity cost — are standard strategy and finance, cited in section 11. Every percentage and dollar below is an assumption; every ranking and forgone-return figure is computed from these inside the course, so you can reproduce each one.

The three divisions competing for the $60M block:

Division	What it is	Return on its next $20M	Character
Apex Aerospace	Precision components, fast-growing	22%	Small, rising
Hearth & Home	Appliances, mature, half of group revenue	11%	Big, steady
Cobalt Chemicals	Industrial chemicals, declining	6%	Loud, just built an $80M plant

Two more figures the committee works with:

Item	Figure
Group cost of capital (the hurdle rate)	10%
Capital block to allocate this cycle	$60M, in $20M increments

6. The Concepts

Marginal return on the next dollar

The return on capital of a whole division — its total profit over the total money tied up in it — is the wrong number for this decision. Allocation is a decision about the next dollar, so the number that matters is the marginal return on capital: what the next increment of money earns if you put it into that unit. You put the next dollar where that marginal return is highest, and you keep funding a unit only while its next dollar clears the group’s hurdle rate — the return the capital must earn to be worth tying up (here 10%, the group’s cost of capital).

Rank Halcyon’s divisions by the return on their next $20M: Apex 22%, Hearth 11%, Cobalt 6%, against a 10% hurdle. Cobalt’s next dollar earns below the hurdle — it destroys value. Put a number on the temptation to fund Cobalt anyway: a $20M increment into Cobalt at 6% instead of Apex at 22% gives up 16 points of return, or 0.16 × $20M = $3.2M a year of return forgone. That gap is the opportunity cost of funding by politics instead of by marginal return.

(An interactive calculator sits here — enter each division’s next-dollar return, the hurdle, and the size of the capital block, and it tells you whether the legacy unit earns the next increment or should be harvested. Change the returns or shrink the block and watch the recommendation flip.)

Now spread the whole $60M block, not just one increment. Returns fall as you feed a unit — the fourth sorter or the second plant earns less than the first, the plain fact of diminishing returns — so Apex’s increments step down: 22%, then 15%, then 9%. Walk it:

First $20M → Apex at 22% (highest, well above the 10% hurdle).
Second $20M → Apex at 15% (still the best increment left, above Hearth’s 11% and the hurdle).
Third $20M → Apex’s next increment is now 9%, below the hurdle, so it loses to Hearth at 11%. Fund Hearth.
Cobalt’s next dollar earns 6%, never above the hurdle → harvest it; fund nothing there.

Result: $40M to Apex, $20M to Hearth, $0 to Cobalt. The block earns 0.22×20 + 0.15×20 + 0.11×20 = $9.6M — a 16% blended return. Had you instead spread it “fairly,” one $20M increment to each division, you would earn 0.22×20 + 0.11×20 + 0.06×20 = $7.8M. Funding by size and fairness rather than marginal return costs Halcyon $1.8M a year on a single block — and it repeats every cycle.

The oldest frame for this call is the growth-share matrix, drawn up by Bruce Henderson at the Boston Consulting Group around 1970. It plots each unit on two axes — the growth rate of its market, and its share of that market relative to the largest rival — giving four boxes: stars (high growth, high share), cash cows (low growth, high share), question marks (high growth, low share), and dogs (low growth, low share). The prescription is a cash flow: milk the cash cows and use their surplus to fund the stars and the best question marks, and harvest the dogs.

Map Halcyon: Hearth is a cash cow (big, mature, throws off cash); Apex is a star or strong question mark (small but growing fast); Cobalt is a dog (weak position in a shrinking market). The frame gets the shape right — Cobalt as a dog to harvest, Apex as the growth to feed.

Its trap is treating the two axes as the answer. Market growth and relative share are only proxies for the real question, which is the marginal return on the next dollar. A “cash cow” can still deserve fresh capital if its next dollar clears the hurdle; a unit labelled a “dog” can hide a high-return niche. The matrix is where the conversation starts, not where the allocation is decided — for that you need the actual returns.

Attractiveness versus position

The richer frame, built by McKinsey with General Electric in the 1970s, is the nine-box grid of industry attractiveness versus competitive position. Instead of two crude proxies, each axis is a composite score. Attractiveness folds in market size, growth, margins, and how brutal the competition is. Position folds in share, brand, cost advantage, and capability. Score every unit on both, place it in one of nine cells, and read the prescription: invest and grow in the strong, attractive corner; be selective and earn along the diagonal; harvest or exit in the weak, unattractive corner.

For Halcyon, Apex lands top-left — attractive market, strengthening position — so invest. Cobalt lands bottom-right — unattractive market, weak position — so harvest or exit. Hearth sits mid-grid: hold it and take its cash. The nine-box beats growth-share because it carries the qualitative factors a single return number hides: regulatory risk, the option value of a foothold, how defensible a position really is.

Its own limit is that the scores are judgment weightings, and whoever builds the chart chooses the weights. Weight “brand heritage” heavily enough and a declining unit can be dressed up as “attractive.” So the grid is a disciplined way to surface the qualitative factors — but it does not overrule the marginal-return number; it informs it. When a hand-built score and a below-hurdle return disagree, ask which one someone had a motive to bend.

Sunk cost, politics, and when to harvest

Two forces reliably corrupt this decision, and both are on display at Cobalt.

The first is sunk cost. Cobalt’s head argues, “we already spent $80M on the new plant, we can’t stop now.” But that $80M is gone whichever way you decide — it is a sunk cost, and the only question in front of the committee is what Cobalt’s next dollar earns. That next $20M earns 6% against a 10% hurdle whether or not the plant exists, so the plant does not change the call. Throwing good capital after a poor return to justify money already spent is exactly the mistake the frame is built to stop.

The second is politics — funding the loudest voice or the biggest business. Hearth is half of group revenue, and that earns it nothing here: revenue is not a claim on capital. Capital goes to the highest marginal return above the hurdle, whoever runs the unit and however large it is. The biggest P&L and the best lobbyist are not reasons; they are noise.

So when do you harvest or cut? When a unit’s next-dollar return sits below the hurdle with no credible path back above it. Then you stop feeding it, take the cash it still throws off (harvest), or sell it and hand the capital back (cut) — and you redeploy that capital to the units still earning above the hurdle. For Cobalt, at 6% against a 10% bar and falling, the call is to harvest the plant’s output and redeploy the block to Apex, whatever the noise in the room.

7. Your Call

You have seen how marginal return, the two portfolio frames, and a clear head about sunk cost decide Halcyon’s block. Now a different call lands on your desk.

Larkspur Holdings is a family-owned media-and-leisure group — a different company in a different sector from Halcyon’s industrial conglomerate — with a $30M capital block to place this year, in $10M increments, against a 11% hurdle. Its three divisions and the return on each one’s next $10M:

Division	What it is	Return on its next $10M
Streamline Studios	Digital content, fast-growing	24%
Riverside Resorts	Hotels, steady	13%
Ledger Press	Regional newspapers, declining	4%

There is a wrinkle Halcyon did not have: Streamline Studios can only absorb $10M this year — its production pipeline is capped, so it cannot take a second increment however high its return. And the founder, whose father built Ledger Press, is the loudest voice in the room for keeping it funded.

How this differs from the taught case (the transfer): this is a different company and sector (a family media-and-leisure group, not the industrial conglomerate), the figures are different so the arithmetic must be redone, and it adds a constraint the taught case lacked — the top unit has a capacity limit, so you cannot simply pour the whole block into the highest return. The concept being tested is the same: put the next dollar where its marginal return is highest above the hurdle, and harvest what sits below.

8. Self-check

Before you write the memo, make sure you can say each of these in one line:

Why is “we are the biggest division” or “we already spent $80M” the wrong basis for getting the next dollar?
What single number decides where the next increment of capital goes, and what is the hurdle it must clear?
When does a unit stop earning fresh capital, and what do you do with the capital you take back?

If any is fuzzy, reread section 6 — marginal return, the two portfolio frames, sunk cost, and the harvest rule are the whole course.

9. Stretch

Push the decision further on your own:

Back at Halcyon: suppose Cobalt’s head returns with a credible turnaround that lifts its next dollar’s return to 14%. Does the call change, and how much of the $60M block should Cobalt now get before its increments fall back below the hurdle?
The genuinely harder one: Apex’s increments step down 22%, 15%, 9%. If the block were $80M instead of $60M, where does the fourth $20M go — and what does that tell you about the size of the block changing the answer even when no return does?
Write the one sentence you would say to a division head who insists “we are half the group’s revenue, so we should get half the capital.”

10. Ship it — your decision memo

Write a one-page memo to Larkspur’s board. State the call (fund Streamline Studios $10M and Riverside Resorts $10M, return the last $10M to shareholders, and harvest Ledger Press — its next dollar earns 4% against an 11% hurdle). Show the reasoning in two or three lines (rank by the return on the next dollar; fund down the list while each increment clears the hurdle; Studios’ cap and the below-hurdle options mean the last increment is worth more returned than deployed). Name what you rejected (funding Ledger on the founder’s attachment; forcing capital past the hurdle to avoid idle cash) and why. Name the one thing that would change your mind (a real, credible lift in a unit’s next-dollar return above the 11% bar). Keep it to a single page a board grasps in two minutes. This memo is your own argued claim — not a credential.

11. Sources

Halcyon Group and Larkspur Holdings, and every figure attached to them, are composite — constructed for clean teaching arithmetic, not drawn from or claimed about any real company. The ideas used to reason about them are standard strategy and finance; references below.

Concept / claim	Source (publisher)	URL	Accessed
Marginal return on capital and diminishing returns	Wikipedia — Marginal return	https://en.wikipedia.org/wiki/Marginal_return	2026-07-20
Return on capital as the base return measure	Wikipedia — Return on capital	https://en.wikipedia.org/wiki/Return_on_capital	2026-07-20
Cost of capital as the hurdle rate	Wikipedia — Cost of capital	https://en.wikipedia.org/wiki/Cost_of_capital	2026-07-20
Growth-share matrix (stars, cash cows, question marks, dogs)	Wikipedia — Growth-share matrix	https://en.wikipedia.org/wiki/Growth-share_matrix	2026-07-20
Attractiveness-position nine-box grid	Wikipedia — GE multifactorial analysis	https://en.wikipedia.org/wiki/GE_multifactorial_analysis	2026-07-20
Sunk cost should not drive the next decision	Wikipedia — Sunk cost	https://en.wikipedia.org/wiki/Sunk_cost	2026-07-20
Opportunity cost of forgone allocation	Wikipedia — Opportunity cost	https://en.wikipedia.org/wiki/Opportunity_cost	2026-07-20

Next up

Finished this call? Continue the Strategy track:

Price war or not? Responding to a competitor’s price cut → · Browse all courses