VISTA·RESEARCH™
How it worksThe LibraryResearchHeritageAboutTalk to the Desk
Inquiry No. 1 · Mondelez International (MDLZ)

The Chocolate Truce

Cocoa is falling, and Mondelez should be the winner. Whether it keeps the winnings depends less on the cocoa market than on whether the world's chocolate makers keep the peace.

By Russ W. Rosenzweig · Vista Research · 3 October 2026 · Why I write with AI
Updated 6 November 2026: what a former revenue management director at a global confectioner told us. See Primary research.
The answer first

At $58.19, its close on October 2, Mondelez is already priced for a full recovery. Put the business back at its 2024 margin, grow it 4% a year for a decade, and the shares offer about 8.5% a year. The strongest bull case we could build with discipline offers 9.1% to 9.4%. Only one version clears 10%, and it quietly assumes that every dollar Mondelez reinvests will earn 43% forever. Meanwhile the company says it is spending its cocoa relief on its brands, prices in Europe are already falling, and Lindt has just told investors it will use easing cocoa to reprice for volume. Two numbers decide this, and neither appears in any filing: how much of the relief survives the promotion calendar, and what the new emerging-market volume earns.

The idea

The bull case fits on an index card, which is part of its charm. Cocoa spiked, squeezed every chocolate maker's margin, and is now coming down. Mondelez, owner of Cadbury, Milka, Toblerone and Oreo, should get its margin back, while its brands and its reach in emerging markets keep the top line growing. The repair pays for the patience, and the growth pays for everything after.

Its latest advocate is Ben Shuleva, Fidelity's consumer-sector leader, in a Barron's interview with Evie Liu published October 1. Mondelez is among the top holdings of the staples fund he co-manages, and his case is the index card: chocolate and cookies are indulgences where brands matter and private labels struggle, the emerging-market business grows faster than America does, and cocoa, which squeezed margins on the way up, is on the way down.

It is a good hypothesis because it can be wrong in specific, checkable ways. This inquiry goes looking for them.

The business, as a machine

Shuleva summed up the whole industry in a line: "Staples companies buy commodities and sell brands to capture the spread in between." Adrian Slywotzky and David Morrison catalogued the ways a business can capture a spread like that in The Profit Zone (1997). Several of their profit models describe a piece of Mondelez. One describes the whole of it.

Brand profit. The same chocolate earns more in a purple wrapper. Slywotzky's favorite demonstration was a single California plant that built one car under two nameplates; the Toyota badge fetched about $300 more than the GM one (The Art of Profitability, 2002, p. 132). Cadbury is that badge, and so are Oreo and Milka.

A model that fits is only half the question. The other half is whether the company is winning at it, and for a brand the test is old and unforgiving: can it raise its price and keep its volume?

winning while Δp / p > 0 and Δq / q ≥ 0

Here Δp / p is the change in price and Δq / q the change in volume and mix. Mondelez's second quarter, region by region:

RegionPriceVolume and mixThe brand test
North America+2.2+1.2Winning
AMEA+1.9+5.2Winning, though operating income fell 0.4%
Latin America+7.9+0.5Winning, flattered by about 1.5 points of trade inventory
Europe−1.4−2.1Losing: price cut, volume still down
Mondelez+1.5+0.7Winning, narrowly

Percentage points of organic revenue growth, second quarter of 2026, from the company's release.

The brand is winning nearly everywhere except the one region where the chocolate truce is decided.

-20+2+4+6-20+2+4+6+8Price, percentage points of organic revenue growthVolume and mix, pointsWinning: price up, volume heldLosing: price cut,volume downNorth America (+2.2, +1.2)AMEA (+1.9, +5.2)Latin America (+7.9, +0.5)Europe (-1.4, -2.1)All of Mondelez (+1.5, +0.7)
The brand test, region by region: the change in price along the bottom and in volume and mix up the side, in points of organic revenue growth for the second quarter of 2026. The shaded corner is the winning one, price up and volume held; every region is in it except Europe, which cut price 1.4 points and still lost 2.1 points of volume, and the company as a whole (the ring) wins narrowly. Source: Mondelez, second-quarter 2026 earnings release.

The profit pattern the company has been living through comes from Slywotzky's sequel, Profit Patterns (1999, pp. 107 to 109): the value chain squeeze, in which suppliers on one side and customers on the other both grow stronger and take more of the value, leaving the step in between unable to cover its cost. For two years the step in between was the chocolate maker, caught between the cocoa market and Europe's grocers. The squeeze is now releasing on the supplier side. This inquiry is really about one question: who collects the release, Mondelez, the retailer, or the shopper at the shelf?

So far, nobody: the release has not reached the income statement. Higher raw material costs were still among the reasons the company gave for lower operating income in the second quarter, because a large maker buys its cocoa well ahead. The scoring starts when it arrives, and the test is the same one in reverse: whoever's share of the chain's profit rises as cocoa falls is collecting it.

Underneath the pattern sits one of the oldest models in game theory, the repeated game. Chocolate makers do not agree on prices, and they don't need to, so long as each expects the others to answer a price cut with one of their own. The truce holds while

G ≤ δ / (1 − δ) × (Vtruce − Vwar)

Here G is what one maker gains, once, by cutting price first while cocoa falls; Vtruce is each year's profit if everyone holds price; Vwar is each year's profit after a price war; and δ is how heavily each maker weighs next year against this one.

Falling cocoa raises G: there is suddenly room to cut price and still earn more per bar than last year. Weak volume raises it again, because the shelf space a price cut wins is worth more when shoppers are scarce. And on September 29, Lindt, the category's premium leader, said it expects 2027 volume growth from "increased brand investments, innovations, and an adjusted pricing strategy, supported by cost savings and easing cocoa prices." The same day Lindt cut its 2026 organic growth outlook to 0% to 2%, from 4% to 6%, and Bloomberg reported that it is cutting prices on its Christmas range and plans to lower them across the board from January. In the language of the model, someone has just mentioned G out loud.

The second piece of math decides what Mondelez's growth is worth. A business earning NOPAT and growing at g is worth

V = NOPAT × (1 − g / RONIC) / (r − g)

where RONIC is the return on the new capital that growth consumes and r is the return the owner requires. Growth adds value only when RONIC is above r; at RONIC equal to r, growth is worth exactly nothing. For Mondelez the new capital is outlets, production lines, acquisitions, and the advertising that the accounts treat as an expense.

What the price assumes

We rebuilt the ten-year owner-cash model behind this inquiry line by line and turned it around, in the manner of expectations investing (Mauboussin, 2006). Instead of asking what the shares are worth, we ask what $58.19 offers a year on each set of assumptions. Vista does not publish what a share is worth. What a price assumes is the more honest question anyway.

The caseWhat it assumesWhat $58.19 offers a year
Downside1.5% growth; margins stuck at 13%5.0%
Base, with growth earning 20% on new capitalThe base case below, with its terminal value built from an explicit 20% return on new capital8.3%
Base4% growth for a decade; margins back to 16.2% by Year 4 and held there forever8.5%
Base, working capital tied to growthTwo treatments of supplier finance8.7% to 8.9%
The bull case at full strength5% growth, 17% margins, 20% return on new capital, about $375 million of interest9.1% to 9.4%
Upside, disciplined5% growth, 18% margins, 20% return on new capital9.9%
Upside as first builtThe same, with a terminal value implying 43% on new capital forever10.3%

Annual return at which each case's owner cash equals the October 2 close. Year-end cash timing; the bull case range spans year-end to mid-year timing. Inputs and method are at the end of this inquiry.

Four things stand out.

The base case is already generous. It returns margins to 16.2% by Year 4 and keeps them there forever. 16.2% was 2024, the better of the two pre-spike years we checked: 2023 was 15.9%, 2025 fell to 13.2%, and the second quarter of 2026 printed 13.1%. It also grows revenue 4% a year for a decade, inside the company's long-term target of 3% to 5% but above the 2.2% it managed in the latest quarter.

10%12%14%16%18%20%22%Reported results, then the model's Years 1 to 10 (Year 1 begins October 2026)Adjusted operating margin, %202320242025Q2 20261234567891015.9%16.2%13.2%13.1%Needed for 10% a year: 20.05% in Years 2 to 10Base case: back to 16.2% by Year 4 and held$58.19 offers about 8.5% a year
The margin the price already assumes: the base case climbs back to 16.2%, 2024's level, by Year 4 and holds it for good, and with revenue growing 4% a year, $58.19 offers about 8.5% a year on it. To offer 10% with everything else unchanged, the margin would have to average 20.05% from Year 2 to Year 10, nearly four points above any full year we checked. Reported margins are from the company's results; Years 1 to 10 are an illustration from the model.

To offer 10%, the base needs a margin nobody has seen. With everything else unchanged, margins would have to average 20.05% from Year 2 to Year 10, nearly four points above any full year we checked.

The terminal value is doing quiet heavy lifting. Nearly 55% of the base case lives beyond Year 10, and the shortcut that produces it implies that new capital earns about 27%. Pin that return at 20% and the base offers 8.3%; at 12%, 7.9%. The upside case is the only one above 10%, and its terminal value implies 43%.

7%8%9%10%11%12%20%27%43%What new capital earns after Year 10, % a yearWhat $58.19 offers a year, %10% a year7.9%8.3%8.5%, the shortcut9.9%10.3%, as first builtUpside caseBase case
What $58.19 offers a year, depending on what new capital is assumed to earn after Year 10, the assumption that carries the terminal value. Pinned at 20%, the base case offers 8.3% and the upside case 9.9%; the shortcut behind the base case assumes about 27% (8.5%), and the only case above 10%, the upside as first built, assumes 43% forever (10.3%). An illustration from the model.

Year 1 asks for a fifth more cash than the company is promising. The base case's first twelve months produce about 21% more owner cash than the company's own 2026 free cash flow guidance of about $3 billion implies, once stock compensation, lease payments and an acquisition allowance are taken out.

This reminds me of Warren Buffett's test of a business, given to the Financial Crisis Inquiry Commission in 2010: "if you have to have a prayer session before raising the price by 10 percent, then you've got a terrible business." This fall's chocolate question is the mirror image: who can keep a price when the cost underneath it falls. Judging by Europe's second quarter, the prayer meetings have started.

What the company is telling us

In the second quarter, adjusted gross profit rose 3.0% at constant currency and adjusted operating income fell 6.1%. The gap is the reinvestment. The company names the reasons for the decline itself: higher raw material costs, higher other selling, general and administrative expenses, and higher advertising and consumer promotion costs. Its chief executive describes "reinvesting behind our brands to enable sustained performance for years to come." That may prove wise. It is not, yet, cocoa savings kept.

The regions tell four different stories:

One more line belongs in any model. Mondelez runs on negative working capital, but part of it is financed: $2.9 billion of supplier finance sits inside its payables and $585 million of receivables have been sold, and both balances were lower in June than in December. That is a stress to disclose, not a gift to capitalize.

The literature is not kind to permanent margins. Fama and French (2000) found that profitability reverts toward the average at roughly 38% a year, and Chan, Karceski and Lakonishok (2003) found sustained high growth rare and impossible to pick in advance. Companies that grow their investment or their assets fastest have tended to earn less afterward (Titman, Wei and Xie, 2004; Cooper, Gulen and Schill, 2008), which is a reason to pin the return on new capital rather than let a formula assume it. And the restructuring and system costs that companies leave out of adjusted earnings tend to come back (Doyle, Lundholm and Soliman, 2003), which is why the model keeps charging for them.

This reminds me of Proverbs 27:23-24 (BSB): "Be sure to know the state of your flocks, and pay close attention to your herds; for riches are not forever, nor does a crown endure to every generation." A 16.2% margin is a crown. The base case wears it for eternity.

The bull case, at full strength

Here is the best case we could build without cheating. Margins climb to 17.0% by Year 5, eighty basis points above 2024. Revenue grows 5% a year from Year 2, the top of the company's long-term range. Interest runs at about $375 million, the company's 2026 planning figure. Working capital grows with sales, without assuming the supplier-finance programs grow too. And new capital earns 20% in perpetuity, which is generous but not magical. On that case, $58.19 offers 9.1% to 9.4% a year, depending only on when in each year the cash is assumed to arrive.

That is not a poor return from a company that sells biscuits and chocolate on every continent. Claus and Thomas (2001) estimated that the equity premium implied by prices and analysts' forecasts is closer to three percent than to the historical average. An investor who agrees can defend a 9% hurdle on a steady consumer staple, and at that hurdle the bull case works. An investor who requires 10% will find that no disciplined case we built gets there. Which hurdle is right is the reader's decision, not ours.

The data that we're missing is...

Primary research

Likely, we can get some of it from primary research, which is what Vista is all about. The filings will tell us, a quarter at a time, whether the truce held. People who ran these businesses through earlier cocoa cycles can tell us now what usually happens, and how fast.

A former commercial or revenue-management lead at a European chocolate maker who worked through an earlier cocoa downcycle

We would askWhen cocoa fell after a spike, how much of the gross margin recovery did the business keep after six quarters of promotion, price-pack resets and retailer funding?

The answer that would change the view"Less than half."

What we heard, November 2026

A former revenue management director at a global confectioner answered the question above in a written interview with Vista's research desk on 6 November 2026. He began with the brand test: the regional figures mix chocolate, biscuits, candy and chilled across countries with very different brands, so a region's price and mix can be skewed. "There is no headline model." Inside a maker, he said, the cocoa saving is not planned as a share kept or given back. It is one cost line among many, beside energy and the supply chain, and the plan is set so that the total margin rises once every line, volume included, is counted. His rule: keep the saving where the market is shrinking or a brand and country earn below-average margins; reinvest it where share is at risk or margins are above average. He gave no share, because the question is not framed that way inside a business. Asked to apply the rule to European chocolate today, he said, "It would keep less."

Makers do not always wait for the buyer to ask, and they tend to tie any funding to specific activity with a return attached: "It is not simply a transfer of profit." What is given stays given: "Costs are rarely recovered by clawing back a concession." When costs rise again, a new price increase or savings found elsewhere recover them. A rival's price cut is usually answered first with more promotional spending, which is easier to pull back than a supply price. Shoppers buy from a repertoire of about seven brands, mostly on impulse, taking "whichever they see first at a perceived competitive price". "Even some of the strongest brands are less than 15% household penetration and are 80% promoted." The Buffett test quoted above is sound guidance for an investor choosing where to invest, he said, but for a business already in a category like this one, "applying a 10% price increase doesn't require a prayer, but careful planning considering elasticities and shopper behaviour."

What it does to the viewIt cuts against the base case, which puts the margin back at 16.2% and holds it there: a maker in Europe today would keep less of the relief, what it gives up is not clawed back, and the next rise in costs waits on another price round. It also qualifies the brand test above. Those figures are region-wide, and in a category bought on promotion and on impulse, a quarter's price and volume can reflect the promotion calendar as much as brand strength. The inquiry's figures and verdict are not recomputed here. The five tests stand, and under the checkpoint below we name two measures he gave us to watch beside them.

A former general manager or route-to-market head for a multinational snack maker in Asia, the Middle East or Africa

We would askWhen volume grows 5% and profit doesn't, is the gap front-loaded distribution that pays back within three years, or the permanent cost of competing with local brands?

The answer that would change the view"The new volume settles at a structurally lower margin."

A former cocoa procurement or commodity-risk executive at a large confectioner

We would askAs a matter of past practice, how far forward do the large makers buy, and how long before a fall in futures reaches the cost of a delivered bar?

The answer that would change the view"Most of 2027's cost is already locked near the peak."

What we heard, October 2026

René Koeppel, former Head of Price Risk Management at Nestlé, answered this question for us by email. The large makers normally hedge 9 to 12 months ahead, because supply follows the West African main crop. On the way up, each announced price rise sent retailers to restock at the old price, so the makers' best cover was used up early and replaced at higher prices; in a long rise, he wrote, a maker ends up underhedged, because its cover was sized for a normal cycle while retailers sped up the drawdown. The usual response is cover cut to an operational minimum and buying from hand to mouth, and the more sophisticated buy out-of-the-money call options, which leave a gap between spot and strike but cap the next rise. On the way down, he told us, there were months in the spring of 2026 when buyers could cover 2027 delivery at close to no forward premium. Falling cocoa does restore margins, in his words, "but with a delay."

What it does to the viewIt is the opposite of the answer that would have changed it. Much of 2027's cocoa could be bought near the lows, not locked near the peak, which supports the base case; and the restocking loop explains why the squeeze on the way up was worse than the hedges promised.

A former confectionery buyer at a large European grocer, and a retail scanner-data specialist

We would askWhen a supplier's costs fall, do retailers take it as price, as promotional funding, or as their own margin, and how quickly? Are this year's share gains bought with promotion?

The answer that would change the view"The retailer takes it, within two quarters."

Every conversation would draw on past, general experience only: no current employees of Mondelez or of its customers and suppliers, nothing confidential, and every Advisor screened for conflicts before a word is said. For a client engagement, Vista would combine this inquiry with those interviews: two or three Advisors, structured conversations, and one written brief that says where they agree, where they split, and what would change the answer.

The checkpoint, set down now

We will score this inquiry in February 2027, after Mondelez reports its fourth quarter, and publish the result on the scorecard whether it flatters us or not. The tests are fixed today, so they cannot drift:

Four or five passes, and the relief is being kept: the base case's margin path starts to look earnable. Two or fewer, and the truce is not holding. Three is too early to say. We will also report what the price then offers on each case, and read Lindt's January sales report for what "an adjusted pricing strategy" turned out to mean.

Added 6 November 2026. The five tests above stay exactly as fixed on 3 October. After the interview under Primary research, we will also watch two measures a former revenue management director at a global confectioner named: market share, and promotion depth, meaning the share of volume sold on promotion and the split of base from incremental volume. They were added after publication and are not scored. Both come from retail scanner data that market research firms sell, which we do not have, so we will report them only where public data allows.

Willy Wonka, in the 1971 film, described the investor's position better than any strategist: "The suspense is terrible. I hope it'll last." It will last at least two more quarters. Until then, the price already assumes the happy ending.

The bottom line

The bet. That Mondelez keeps the cocoa windfall: margins back to 2024's 16.2% within four years and held there, with revenue growing 4% a year for a decade.

The payoff. If it does, about 8.5% a year; the strongest disciplined bull case offers 9.1% to 9.4%; if margins stay near 13% and growth slows to 1.5%, about 5%. No disciplined case reaches 10%.

Our read. The price already assumes the happy ending, while the evidence points the other way for now: the company is spending the relief on its brands, European prices are falling, and Lindt says it will reprice for volume.

What settles it, and when. The fourth quarter, reported in February 2027: whether pricing net of commodity costs stays positive and the relief reaches operating income. Before then, we would talk with a former revenue-management lead at a European chocolate maker and a former confectionery buyer at a large European grocer about what happens to promotion calendars when cocoa falls.

Questions this inquiry answers

What does Mondelez's stock price assume?

At $58.19, its close on October 2, 2026, Mondelez is already priced for a full recovery: with margins back to 2024's 16.2% within four years and held there, and revenue growing 4% a year for a decade, the shares offer about 8.5% a year, and the strongest disciplined bull case offers 9.1% to 9.4%. No disciplined case reaches 10%, and if margins stay near 13% and growth slows to 1.5%, the shares offer about 5% a year. This inquiry is research, not a recommendation.

Why haven't falling cocoa prices lifted Mondelez's profits yet?

In the second quarter of 2026, adjusted gross profit rose 3.0% at constant currency but adjusted operating income fell 6.1%, and the company named higher raw material costs, higher other selling, general and administrative expenses, and higher advertising and consumer promotion costs. A large maker buys its cocoa well ahead, and the company says it is spending its cocoa relief on its brands.

What happened to Mondelez's prices and volumes in Europe?

In the second quarter of 2026, Europe cut price, with pricing down 1.4 points, and still lost volume, down 2.1 points, while adjusted operating income fell 22.9% at constant currency. It was the one region losing the brand test, which asks whether a company can raise its price and keep its volume.

Why could falling cocoa prices start a chocolate price war?

The note models the chocolate truce as a repeated game: makers keep their prices while each expects the others to answer a price cut with one of their own, and falling cocoa and weak volume both raise what one maker gains by cutting first. On September 29, 2026, Lindt said it expects 2027 volume growth from "increased brand investments, innovations, and an adjusted pricing strategy, supported by cost savings and easing cocoa prices."

What margin would Mondelez need for its shares to offer 10% a year?

With everything else in the base case unchanged, margins would have to average 20.05% from Year 2 to Year 10, nearly four points above any full year the note checked. The margin was 15.9% in 2023, 16.2% in 2024 and 13.2% in 2025, and the second quarter of 2026 printed 13.1%.

What are the checkpoint tests for this Mondelez note, and when are they scored?

Vista will score the note in February 2027, after Mondelez reports its fourth quarter, and publish the result on its scorecard. The tests: the "pricing net of commodity costs" contribution in Mondelez's gross profit bridge is positive for the second half of 2026; in the fourth quarter, adjusted operating income grows at least as fast as adjusted gross profit (both at constant currency), European pricing is zero or better with volume and mix no worse than flat, and AMEA adjusted operating income grows at constant currency at least as fast as its volume and mix; and the full-year 2026 adjusted operating margin is above 2025's 13.2%. Four or five passes means the relief is being kept, two or fewer means the truce is not holding, and three is too early to say.

How the figures were made

Owner cash each year = (revenue × adjusted EBIT margin − net interest) × (1 − tax rate) + depreciation and amortization − capital spending − working-capital investment − after-tax ERP and restructuring cash − finance-lease principal − an acquisition allowance + equity distributions − minority distributions − legacy pension cash. Year 1 runs from October 3, 2026 to October 2, 2027. Year-10 cash grows into perpetuity at the terminal rate, or, where stated, the terminal value is built from an explicit return on new capital: NOPAT × (1 + g) less the investment that growth at that return requires. Shares are fixed at 1.290 billion; no buybacks are assumed. The return each case offers is the discount rate at which its owner cash, per share, equals the $58.19 close. Every figure in this inquiry was recomputed in code from these inputs.

InputDownsideBaseUpsideBull case
Year-1 revenue ($bn)39.540.541.040.5
Annual growth after Year 11.5%4%5%5%
Adjusted EBIT margin, Years 1 to 412.5, 13, 13, 13%14.5, 15.5, 16, 16.2%15.5, 17, 18, 18%14.5, 15.5, 16.2, 16.7%
Margin from Year 513%16.2%18%17%
Tax rate27%25%24%25%
Net interest ($bn)0.600.450.400.375
Capex, D&A (% of revenue)4.0, 3.43.8, 3.43.7, 3.43.8, 3.4
Working capital0.7% of sales0.6% of sales0.5% of sales6.4% of new sales
ERP and restructuring cash ($bn)0.35 falling to 0.150.30 falling to 0.100.25 falling to 0.0750.30 falling to 0.10
Lease principal, acquisitions ($bn)0.17, 0.200.16, 0.200.16, 0.200.16, 0.20
Terminal growth2%3%3.5%3%

Every case also adds $0.05 billion a year of distributions from equity investments and deducts $0.016 billion of minority distributions and $0.05 billion of legacy pension cash. The working-capital variants of the base case apply either the June ratio of trade working capital to sales (negative, with the financing programs in place) or the same ratio with the programs added back (6.4%), to each year's new sales.

Sources

Disclosures

This inquiry is research, not investment advice. It states what a market price assumes under labeled assumptions; it is not a recommendation to buy, sell or hold any security, and the decision belongs to the reader. As of October 3, 2026, Russ W. Rosenzweig, Vista's founder, owns no Mondelez shares. Holdings through mutual funds and exchange-traded funds are not counted. Lindt's guidance and price cuts were added on October 6, 2026; the first version quoted only its outlook. What a former cocoa risk manager told us was added on October 8, 2026, under Primary research. How this inquiry was made: written by Russ W. Rosenzweig with Vista's AI research desk. It began as a research question he put to the desk, which challenged the model over three rounds; the model and every figure were then rebuilt in code and checked against the company's filings. Why I write with AI. Why I publish these inquiries.