Deere: Planting Precision, Reaping Reversion?

It’s common for investment job interviews to have some stock pitching component—this can take many forms. This is one I’ve used in the past. While I don’t necessarily recommend copying any part of my methodology—nor do I recommend any investment action related to Deere—it’s a good case study in thinking about a new business for the first time, what sorts of things to think about, what questions to ask, etc.

Here’s the piece, with market price updated for relevance. At the end, we’ll give a quick debrief.

Business Summary & Opinion

Note: for financial data & assumptions, see the Excel exhibits attached to this report.

Deere & Co. mostly operates under the John Deere brand. The company manufactures agricultural, home turf, and construction equipment, as well as providing financing for dealers and customers. Sales are generally made through independently-owned dealers.

The company has achieved strong capital efficiency, and reports performance using Shareholder Value Added, a form of economic profit that incorporates a 12% pretax cost of capital. Management measures of operating return have well exceeded this level. In my estimate of invested capital, I sum net PP&E, net working capital, and intangible assets. I compare this with after-tax EBIT (NOPAT) to estimate ROIC, which has averaged 30% over the last ten years, again well above the company’s cost of capital.

Some possible explanations for their strong return profile—each of these would, of course, warrant further investigation:

A strong, independent dealer network is putting up most of the capital (~$40mn average equity to launch a dealership) to fund ultimate point-of-sale. Peers like CAT cite dealer networks as a key competitive advantage, and the higher proportion of wholesale units has a positive impact on margins due to lower selling costs.
Focus on manufacturing expertise, quality, and efficiency. The company has a long history of making top-quality tractors.
Brand value: there are few (if any) North American agriculture equipment makers with the same brand recognition as John Deere.

Deere estimates that unit sales will typically fluctuate 20% above/below mid-cycle in an average peak/trough (the drawdown was ~20% in 2009). They don’t disclose the basis for this estimate, and it may be overly optimistic—but if true, it suggests a cycle less dramatic than one might see in construction or mining equipment.

The company’s mid-term strategy (through 2026) is heavily focused on bringing improved technology to market.

Develop and market an autonomous, electric tractor.
Deliver 20+ electric or hybrid construction models
Increase adoption of several existing software & operating solutions
Grow recurring revenue to 10% of total revenue by 2030

It remains unclear how the company defines recurring revenue—whether or not it includes service and parts on the installed base, information-enabled solutions, etc. The “Other” revenue category into which services & data solutions would fall was only 3% of total revenue in 2022.

Management compensation is driven by Shareholder Value Added (both short- and long-term), as well as 3-year total shareholder return vs. peers. I would consider the focus on SVA to be a positive (given it incorporates cost of capital & is therefore an appropriate proxy for economic profit).

Key Valuation Drivers & Factors

For each of Deere’s key end markets, the following factors should drive performance and valuation over the next few years:

Agriculture: replacement cycle, crop prices, farmer wealth. Bear in mind, this will easily be the most material market for valuation of the company.
Turf: residential improvement spending.
Construction & forestry: general economic cycle, infrastructure spending.

The company’s performance in expanding their information solutions business will also be key. If this is successful, the company should have a less cyclical revenue base with higher margins and improved returns on capital.

Pricing strategy will be critical as well. In the last couple of years, Deere has increased prices significantly, but only so far as to offset inflation in production costs. While this keeps profit levels steady, the higher prices drive revenue up and should bring margins down if this continues.

As electric vehicle technology keeps proliferating, this may increase selling prices and manufacturing costs for Deere—similar to the effects for autos and construction/mining equipment. However, the magnitude of this effect isn’t clear, and ultimately depends on how this technology evolves further.

Additional Considerations

Further data on North America installed base would be helpful: size of fleet, average age, going back a couple of decades. USDA appears to publish this data, but only every five years—it’s possible that sell-side research may have some data, though. Replacement cycle is a critical part of future demand, but it’s difficult to quantify this without up-to-date data.
Curious how the company defines “recurring” revenue. How much of this includes maintenance & service on existing machines? I would assume that a mature company with this size of installed base would have a more robust service operation, but this appears not to be the case. Why is that? Are dealers capturing a lot of this value?
What is their pricing strategy? Is the goal simply to recoup costs of manufacturing? How does this compare to competitors? Is Deere hoping to use cost inflation as an opportunity to gain market share?
What metrics does management use to evaluate potential M&A opportunities? Is this expected to be limited to small bolt-on deals going forward (i.e., not repeating a larger deal like Wirtgen)?
Would be nice to have more years of consistent segment history—plus I’m struggling to understand the rationale for including some of agriculture with turf…breaking it out by machine size doesn’t appear intuitive, since I would think end markets and factors would be similar, while turf appears driven by different sets of factors—some context on how management thinks about this might be helpful.
Thorough economic/end market data outside North America can be hard to find. Fortunately, Deere’s main exposures are in the U.S. and Canada.
Would like to understand the finance segment a bit better. It appears fairly low-risk on the credit side, but highly levered at ~13x D/E (relative to ~6x at CAT). Also, the separate 10K for John Deere Capital doesn’t appear to include the entire finance business, so it’s difficult to get a complete ex-financial services view of the operating business—which is how I’d prefer to value the business if possible. Not highly material to valuation, but would still be good to understand from a risk standpoint.

Initial Assumptions & Valuation

To make an initial valuation of Deere, I reorganize the segments into three principal end markets: agriculture, turf, and construction/forestry. I do this because I’m assuming for now that this makes the most sense in terms of what factors will drive demand and profitability; turf and small agriculture appear to be too different to lump together. As we learn more about the company, I may need to change this approach—given this is a preliminary look, flexibility throughout will be key, and none of the below assumptions should be considered authoritative.

I make a 5-year forecast of revenues & operating margins across these three end markets, using a UFCF approach to value the business during interim years, and apply a multiple to terminal NOPAT to get to enterprise value, which I adjust for debt and book value of John Deere Capital (pensions and OPEB are overfunded, but I don’t give credit for this out of conservatism).

Note: for 2023E, I’m taking company’s latest guidance at face value (given 2 quarters have already passed, and the 3^rd is nearly over). My assumptions below will apply 2024E and on.

For agriculture, it makes sense that higher levels of farm capex should drive demand for agricultural equipment. In this analysis, I’m focusing purely on the U.S. market, since this is most important for Deere. A similar framework can be applied to other regions to the extent that data is available. The goal is to estimate some quantitative basis to predict future equipment sales.

A brief correlation analysis shows that commodity prices and farm incomes correlate strongly with farm capex, which makes intuitive sense, as farmers should spend more when they feel wealthier and when they feel more positive about the crop opportunities that exist. We can create a regression to quantify this relationship using the data we have back to 1982.

It still remains to make some assumptions regarding these variables going forward. For the commodities, I assume that each will reach its long-term average price in 2027E, adjusted for the trough-to-trough CAGR in price achieved over the previous price cycles—this varies by commodity, but see the appendix for the values used. In each year 2024E-2027E, we move 25% of the way to this adjusted average until achieving it in 2027E. For farm incomes, I estimate these using a fixed income margin (which is reasonable vs. history) and incorporate the long-term growth in total farm receipts (see Exhibit 5).

I use my regression to translate these variables into my estimates of farm capex. One advantage of this approach is that the farm capex numbers should already incorporate replacement spending, so this helps mitigate lack of fleet and replacement cycle data.

Since Deere has been aggressive on price, I apply a 2% premium to my estimates of farm capex growth each year to get revenue growth for agriculture.

For the turf business, I look at non-auto durables spending, which has moved in lockstep historically with disposable income in both the U.S. and Canada. This spending has grown at ~4.5% peak-to-peak CAGR between 2007 and 2019 (I ignore 2020-2022 since this likely has some distortions due to high remodel/improvement spending during COVID-19). I take this as my trend growth rate for the turf products.

Choosing a margin for agriculture and turf—I use a single margin across both businesses since they are reported separately for only a few years—is more complex. I don’t consider current 17% margins or 22% guided 2023E to be true mid-cycle numbers. But taking a 10-year historical view doesn’t provide a true trough margin either since we’re excluding 2008-2010. Besides, margins should fall if price increases only offset cost increases, and margins should rise if recurring revenue and scalable information-enabled sales grow. For now, I use a 12% margin, vs. the 10-year average of 13.5%.

For construction & forestry, I use U.S. construction spending as an end market measure. In real terms, construction spending has been flat to slightly down since 1995. Because of this, my revenue forecast is simply price and inflation. I take 2% for each, giving me a 4% revenue CAGR. Although revenue is likely higher than mid-cycle, I don’t apply any penalty to revenue, assuming an offset from infrastructure spending in the U.S.

Similar factors are at play in construction margins as with agriculture & turf margins. I take a similar approach, using a 9% terminal margin vs. the 10-year average of 9.3%.

Use a 25% tax rate, at the high end of the company’s guided range. I apply a 16x multiple to terminal NOPAT, based on a theoretical multiple of 15.2x (Exhibit 12) and the company’s long-term average P/E of 16.1x (Exhibit 11). I’m a bit aggressive on the multiple to capture possible rerating if data-enabled business grows. See Exhibit 13 for valuation summary.

Below the EV line, I deduct net debt and add the book value of John Deere Capital Corp. I value this at 1.0x book value since it’s a captive finance arm. This may be conservative given the low credit risk of the firm’s portfolio, but it’s more highly levered than peers and so I’m willing to err on the side of caution.

These deductions result in my preliminary estimate of value of ~$360/share based on FY22 share count, compared to market price of $402.23 as of 9/13/23.

Below is a summary of my observations/opinions about the business.

Things I like:

Strong returns on capital, not a lot of incremental investment needed
Excluding JD Capital, is not excessively levered
Dealer network is well-developed
Opportunities to improve technological integration of products

Things I don’t like:

Service business appears either underdeveloped or captured by dealers
Risk of low-cost competition, especially in overseas markets
End markets are cyclical—although less so than some other industrial markets

Materials Reviewed

Company filings, presentations, factbooks, and proxy statements.

Review & Feedback

On the whole, I felt pretty proud of my work. Deere publishes a lot of data on their website in Excel form (all the data I used was from company disclosures), which made the project more straightforward. Given that this is essentially working for free, I didn’t bother chasing down every detail or every loose end. I also avoided using any resources available to me as part of my current job—there are really important ethical constraints around this. Still, I had some decent hypotheses, and I at least acknowledged the areas where useful information was either unavailable or hard to find. Here are three key issues I want to highlight in my approach.

Doing a five-year (or any-year) forecast. I generally don’t like doing this at all. The only reason I did in this case was because the prompt very clearly requested a five-year forecast for any/all key metrics that I identified. That said, creating exact numbers over any defined length of time is essentially predicting the future—something you can’t do with precise accuracy. Unless there’s a compelling reason to use a multistage approach—like with a company that’s likely to experience abnormally high or low growth/profitability in the near term—I generally prefer to avoid them. The key is to be honest about what level of detail will actually matter to the long-run performance of the stock, and take seriously the tradeoffs between available information, relevant information, and what assumptions you can reasonably make.

Quantifying drivers. The regression I ran that linked commodity prices and farm incomes to farmer spending was the weak link in my process—one reason why I didn’t give the outputs of the regression directly. While the relationship made intuitive sense to me, the data didn’t bear as strong a relationship as I would have expected. Out of laziness and time constraints, I ignored this. In order to ground my valuation on stronger footing, some better understanding of these drivers was the main thing I missed. A more global dataset might have helped, but this was hard to find. Ultimately there’s no substitute for understanding what compels a firm’s customers to demand the firm’s products—and this isn’t something that a spurious regression is sufficient to accomplish.

Reversion to the mean. To me, reversions to the mean are ubiquitous. We see them everywhere, from business cycles to sports. A key part of my valuation thesis for Deere was reversion to the mean, at least related to commodity prices. This made sense to me given the clear cyclicality of commodity prices (see Exhibit 3 in the appendix). If anything, I felt I was too generous, since I allowed farm incomes to keep growing at trend, even though you could make a case there were corrections in 2009 and 2015 (Exhibit 5). But reversion to the mean can be a tricky concept to accept.

The world is a big enough place—we can find lots of exceptions to mean reversion. But to me, precious few of them obviate the principle. The most dangerous phrase in finance is: “This time, it’s different.” It’s also emotionally difficult to remain mindful of mean reversion—which by definition preaches optimism in the darkest times and pessimism in the rosiest times. If you own a cyclical business such as Deere late in a cycle, it’s natural to try and wish mean reversion away. Besides, mean reversion is given more importance by value-oriented investors than by growth-oriented investors, so it’s important to be mindful of the audience.

Those are a few of the major takeaways from my Deere project. My hope is to do more “business evaluation” posts like this in the future. Let me know if this is something you’re interested in through the comments. And feel free to point out anything else I might have missed, or give ideas on other companies you’d be interested in seeing.