Deere & Company

Deere sells machines to people who grow things, build things and cut things down, and then lends those same people the money to buy the machines. That is the whole company. It has been in business since 1837, when a blacksmith in Grand Detour, Illinois, built a plough out of a broken steel saw blade because the cast-iron ploughs of the day clogged with the sticky soil of the prairie and had to be scraped clean every few yards.

There is a certain irony in that origin, which we will come back to in Part V. The company was founded on a maintenance problem, and its largest legal fight of the last two years was about who is allowed to do the maintenance.

Today it reports in four pieces. Three of them make things. One of them is a bank, and understanding that is worth more to you than anything else in this analysis.

We held this analysis deliberately. Deere's fiscal third quarter ended 2 August and the results landed five days ago, and publishing before them would have meant writing about a cycle without the one piece of evidence that mattered: whether farmers had started ordering machines for next season. They had.

The segment detail is where the shape of this cycle shows itself, and it is not a single story.

Look at Construction & Forestry for a moment. Sales up 18%, operating profit up 84%, margin from 7.7% to 12.1%. That is what operating leverage looks like when volume returns to a factory that has been running below capacity — and it is a preview, in miniature, of what the large-agriculture segment does on the other side of this cycle.

The company that reported this quarter is not the company most people have in their heads. Large agriculture is 37% of revenue, not 80%. Two of the three manufacturing segments grew double digits. The diversification is real, and it is new.

We have done this four times now. For REITs we explained why earnings per share is the wrong line and funds from operations is the right one. For business development companies it was net asset value and net investment income, and why an Altman-Z or a price/earnings ratio on a BDC is a category error. For banks it was net interest income and return on tangible common equity. For regulated utilities it was the difference between the return a company earns and the return it is awarded.

This is the cyclicals-with-captive-finance instalment, and it is the one that fools the most people, because unlike a REIT or a bank, Deere looks like an ordinary industrial company. It files like one. It is in the industrial indices. And then you point ordinary industrial ratios at it and they return nonsense.

The Altman-Z fault is worth showing you in full, because it is arithmetic rather than opinion.

The score is a weighted sum of five ratios. One of them is retained earnings divided by total assets, and it carries a weight of 1.4 — the second-largest in the formula. Our feed publishes the inputs it used, and the retained-earnings input is zero.

We are not telling you to use 2.95 either. The corrected number is still an artefact, because the underlying formula does not know what a captive finance company is. We show you the correction to make a narrower point: the number that appears on screens, in stock reports and in a great many articles about this company is not merely the wrong tool — in this case it was computed wrongly as well.

One score, incidentally, does survive: Piotroski at 8 out of 9. It is a checklist of nine yes/no financial-health tests rather than a formula weighted on a 1968 sample, and it is far more robust to this kind of structure.

Deere does something admirable here, and almost nobody reads it: it publishes a full consolidating balance sheet splitting Equipment Operations from Financial Services, every quarter, in the press release itself. Here is what it says as at 2 August 2026.

The tractor company carries about $2.6 billion of net debt against $21 billion of equity and roughly $5 billion of annual earnings. That is not a levered balance sheet. That is a fortress with a bank standing next to it.

And the bank is not a distressed one. Its debt is matched against $42.9bn of financing receivables, $6.3bn of securitised receivables and $7.4bn of equipment on operating lease — roughly $56.6 billion of earning assets against $54.5 billion of borrowings. A 7.3× debt-to-equity ratio inside a captive lender is unremarkable; it is less levered than most banks.

There is a second reason the finance arm deserves attention, and it is the best early-warning system Deere has. If American farmers were genuinely in distress, it would show up first as rising delinquencies in this loan book — long before it showed up in machine sales, because a farmer stops paying before he stops farming. It is not showing up. That is worth more than any sentiment survey.

One honest note on our own numbers. The net debt/EBITDA figure our feed returns is 0.74×, which is as wrong in the other direction as 4.8× is in this one. We have computed 4.8× ourselves from the reported balance sheet so you can see the arithmetic. The lesson is not that one provider is careless; it is that when a ratio's answer depends this heavily on which debt you count, the ratio is telling you something about your data rather than about the business. Use the consolidating table. Deere publishes it.

Everything about Deere as an investment reduces to one question: where are we in the farm cycle? Get that right and the multiple, the margin and the dividend all fall into place. Get it wrong and no amount of analysis of the moat will save you.

Start with the thing that dollar revenue hides. Deere's sales fell about 24% from the fiscal 2023 peak. Unit volumes fell far more than that, because price and mix cushioned the reported line.

Combine sales across the whole of the United States fell from 7,349 machines to 3,579 in two years — a 51% collapse. Tractors are down 38% from their 2021 peak. This is a depression in volume terms, and it is considerably worse than the revenue line suggests.

Which is precisely why the recovery, whenever it comes, has such violent operating leverage. Deere's factories have been running at volumes that cannot be sustained indefinitely by replacement demand — a combine wears out eventually, and the American fleet is ageing while almost nobody buys.

All quotations above are from Deanna Kovar, President of Worldwide Agriculture and Turf, on the Q3 fiscal 2026 earnings call of 21 August 2026.

The company's own summary of the print, printed as a headline bullet on the release, was this: "Order book trends reinforce 2026 as the bottom of the ag equipment cycle."

One important qualification, and it comes from the company itself. Kovar, on the same call: "underlying fundamentals continue to support a measured recovery, rather than a sharp rebound in 2027."

That sentence should be read carefully by anyone buying this stock today on a snap-back thesis. Deere is telling you the bottom is in and that the climb out will be gradual. Those are two different statements and only one of them is bullish.

Here is the most useful frame in this entire analysis, and it costs nothing to apply.

Deere is not experiencing something new. It is experiencing the thing it experiences roughly every decade. The last farm downturn ran from fiscal 2013 to fiscal 2016. Compare the two.

Minus 47% over three years, both times, with a multi-year dividend freeze in both. The amplitude, the duration and the capital-allocation response are almost identical. That is not a coincidence; it is what an agricultural equipment cycle looks like.

And there is a precedent inside the precedent that deserves your attention. Going into fiscal 2017 — the recovery year — Deere's own initial guidance was for net income of roughly $1.5 billion, essentially flat with the trough. It delivered $2.159 billion. Management, at the bottom, materially understated its own recovery.

We are not going to build a thesis on the assumption that history repeats to the decimal. But if you are wondering whether the current guidance of $4.75–5.00bn and the 2027 consensus of $22.22 are conservative, the last time Deere stood at exactly this point in exactly this cycle, they were.

Deere's competitive position rests on three things, and they are not equally strong.

Picture a farmer two hundred miles from a city, mid-harvest, with a weather window closing in thirty-six hours and a combine that has stopped. Downtime in a harvest window costs far more than any repair. What he needs is a part and a technician, today.

Deere's dealer density is unmatched, and it is not replicable. A competitor cannot buy it; it would need decades of capital and relationships to build it. This is the moat, and it has nothing to do with technology. It is also, not coincidentally, precisely what the right-to-repair fight was about.

This is the newest source of advantage and the most interesting. Deere sells subscriptions attached to machines — guidance, data, autonomy, and a system called See & Spray that identifies individual weeds and sprays only those, rather than blanketing a field.

The See & Spray economics are the mechanism, and they are worth understanding. A 50–60% cut in chemical spend is a hard-dollar return a farmer can compute on the back of an envelope in about a minute. That is what converts a technology into a switching cost — and it accrues per acre, per season, which is how a one-time machine sale becomes an annuity.

Now the honest part. Under the "Leap Ambitions" framework announced in 2022, Deere set out to reach roughly 10% of total revenue from recurring sources by 2030, and — as reported at the time — around 1.5 million connected machines by 2026. It is 2026. The company reported nearly 1.2 million. That is a shortfall of roughly a fifth against a target now due.

We flag two caveats. The 1.5 million figure comes from Deere's 2022 investor communications as reported, not from a current filing, and companies quietly restate such targets. And Deere does not disclose a recurring-revenue percentage at all — it discloses renewal rates and user counts instead. That non-disclosure is itself information. If the number were running ahead of plan, it would be a headline slide rather than an absence.

For years the bear case on Deere's parts-and-service annuity was regulatory. In January 2025 the Federal Trade Commission, joined by the attorneys general of Arizona, Illinois, Michigan, Minnesota and Wisconsin, sued Deere over restrictions that forced farmers back to authorised dealers for many repairs.

It is settled. A stipulated order was entered on 8 July 2026. Separately, a consolidated private class action was settled for $99 million.

Bulls should not read "settled" as "resolved in Deere's favour." The overhang is gone, which is worth something. But what replaced it is a decade of mandated independent access to the highest-margin, most counter-cyclical revenue in the company. That is a genuine, if slow-acting, erosion of the annuity.

A counterpoint worth stating, because it cuts both ways: repair advocates themselves regard the settlement as weak. Weaker enforcement means less erosion of the annuity — and also means the legislative pressure does not go away.

There is a symmetry here that is almost too neat. The company was founded in 1837 on a plough that did not need to be scraped clean. Its defining legal fight of the 2020s was over who is permitted to do the scraping.

Deere is not a toll booth, and the analysis is stronger for saying so plainly.

Compare it to a confectioner. Customers buy chocolate every year regardless of the economy; pricing power is annual; capital requirements are trivial. Deere's customers can simply not buy a combine this year, or next year, or the year after — and they have been doing exactly that. Combine unit sales down 51% in two years is a customer base exercising an option that a confectioner's customers do not have.

The counter-argument is real, and it is the bull case in one sentence: Deere is a cyclical with an annuity growing inside it, and the question is how fast the annuity grows relative to the cycle's amplitude. Over-90% subscription renewals, 450,000 monthly software users and a parts business that holds up while machine sales collapse are evidence that the annuity is genuine. The right-to-repair settlement is precisely a ten-year constraint on it. The two central threads of this analysis meet here.

There is a cost sitting inside Deere's current earnings that has nothing to do with agriculture, and in the short term it is a larger factor than the cycle itself.

Two things follow, and they point in opposite directions.

The optimistic reading: roughly a billion dollars of annual cost is a policy artefact rather than an operating failure. It is not evidence that the business has deteriorated, and it is reversible in a way that lost market share is not. The Section 232 rate on imported goods has already dropped to 15% from roughly 25%, and $382 million has come back through the door this year.

The cautionary reading: management guides the net cost up in 2027. If you are modelling a recovery year, roughly a billion dollars of tariff is sitting in front of it.

★ And there is a third thing, which we think reflects well on management. Deere has absorbed this cost rather than surcharging its customers — in a year when its customers are already under pressure. It costs a fifth of a year's earnings. It buys decades of dealer and farmer loyalty. That is a long-horizon trade, and it is the sort of decision that is very easy to criticise in a quarterly earnings model and very hard to argue with over twenty years.

This is where Deere has behaved genuinely well, and it is visible in three numbers moving in three directions.

Deere protected the recurring commitment and flexed the discretionary one. That is the textbook cyclical capital-allocation stance, and remarkably few companies actually run it. The dividend is a promise; the buyback is an opinion. Deere kept the promise and suspended the opinion.

It also means a large lever is pre-loaded for the other side. The repurchase authorisation announced in December 2022 was for up to $18 billion, and roughly $7 billion of it remains unused — around 6% of the current market capitalisation, sitting on the shelf.

Two further pieces of behaviour belong here. Deere deliberately under-produced retail demand for two years, accepting factory under-absorption and the margin damage that comes with it, in order to force the dealer channel clean. And it absorbed roughly a billion dollars of tariff rather than surcharging farmers. Both cost real money in the current year. Both are the right decision over a decade.

Deere yields 1.0%. Nobody buys it for income, and this section might therefore look like a formality. It is not, for one reason: the freeze is the most commonly misread fact about this company.

Seven consecutive quarters at $1.62. The last increase was declared on 3 December 2024 — twenty months ago. Before that, Deere had raised the dividend six times in three years, including three times in calendar 2023 alone ($1.25, then $1.35, then $1.47). Peak-cycle behaviour, followed by a full stop.

An income investor arriving cold might read a twenty-month freeze as a distress signal. It is the opposite of a distress signal, and the proof is in the arithmetic.

Our house requirement is that coverage is tested on free cash flow, not earnings. For Deere that requires the same correction as Part II: the consolidated cash flow statement counts the finance arm's lease originations as capital expenditure, which understates the manufacturer's cash generation dramatically. Use the Equipment Operations column that Deere publishes.

Two-and-a-half times covered by free cash flow, at the bottom of the worst downturn in a decade. The consolidated figure of 0.46× is the same artefact we dismantled in Part II, and it would tell you the dividend is uncovered. It is not.

Very little, and that is an honest answer rather than a complacent one. On the current shape of the business, a cut would require a downturn materially deeper and longer than the one Deere is already in — earnings falling by roughly half again from a level that is already down 48% from peak, sustained for years, while the finance arm's credit book deteriorated enough to require capital support from the parent.

The genuine risk to watch is not the dividend but the freeze extending. If the recovery Deere calls "measured" turns out to be slower than that, the dividend stays at $1.62 into 2028, and an investor who bought expecting the historical pattern of increases resuming gets a flat cheque for another two years. That is a real cost, and it is far more likely than a cut.

Here is the trap, and it catches people in both directions.

Deere trades at 36 times trailing earnings. The instinctive reaction is that this is expensive. In a cyclical, that instinct is backwards. A cyclical's price/earnings ratio is at its lowest when earnings are at their peak — which is the worst moment to buy — and at its highest when earnings are at their trough, which is the best. A low multiple on peak earnings is a warning. A high multiple on trough earnings is the correct sign.

So the honest question is not "is 36× expensive?" It is: are these trough earnings, and how much of the recovery is already in the price?

Part III answers the first half: yes, almost certainly. This part answers the second half, and the answer is less comfortable.

Put the two halves together and the position is uncomfortable rather than contradictory.

The cyclical logic says buy a great cyclical when the multiple looks absurd because earnings are depressed. Deere qualifies on the first condition and on the quality test. But the second half of the trade — the part nobody puts on a slide — is that you have to buy it before the market agrees with you. The moment for that was somewhere near $433. Today the shares have risen half, sit within 4% of a 52-week high, offer 6.8% to consensus, and carry six sell ratings.

You are no longer being paid to be early. You are being asked to pay for being right.