Mike Rivers' Blog Headline Animator

Thursday, October 30, 2014

John Deere: expected returns and potential downside

(Full disclosure: my clients and I own shares of Deere.)

In preceding articles, I've covered Deere's (DE) general situation, competition, economics, management, and opportunities & risks. Now it's time to put those thoughts together with some math to figure out what kind of returns can be expected from John Deere.

Before I jump in, I want to make it clear that my expected return discussion is based on the long run. For that reason, it is important to read the full article and see the second half, where I talk about how bad valuation can get in between now and the long run. Caveat emptor.

Long term expected return

My approach to projecting long term returns is to look at long term trends and normalize that for cyclical factors. I want to know what long term, normalized sales per share, net margins, growth and multiples are so that I can estimate a five year price (not necessarily as a five year price target, but a normalized level for price in five years).

Sales per share

In Deere's case, sales growth from 1982 to 2013 (using the exponential fit function in Excel) is quite stable (96.7% R-squared function, Excel). If it weren't, I wouldn't use it. Deviating from this fit would have to assume a secular change in the farming or farm equipment market different from anything seen from 1982 to 2013. A fit from 1982 to 2013 shows a $33.8 billion normalized sales level a year from now. Applying 377 fully diluted shares (I take basic shares and add all options, restricted stock, etc. to that number) to that sales level implies around $90 in sales per share. 

To adjust for the ethanol boom, I also did a fit from 1982-2004, and that showed sales per share of $75. To estimate what things would look like if the last 10 years were the trend going forward, I also did a fit from 2004-2013, and that showed sales per share of $95. Now, I have estimates for normalized sales per share with low, average and high trends in mind.

Net margin

Net margins at Deere have moved a round a lot over the last 32 years. The median net margin over that time was 5.9%, but it has also been steadily trending up (due both to Deere being better managed and a nice tailwind from farming growth scaling up). Below are the the longer to shorter term median net margins:

  • 30 year: 6.1%
  • 25 year: 7.2%
  • 20 year: 7.7%
  • 10 year: 7.7%
  • 5 year: 8.2%
  • 3 year: 9.2%
With these numbers in mind, I'll base my estimates on a low end net margin of 6%, mid point of 7.5%, and high end of 9%.

Growth

I break growth into three parts: sales growth, margin growth, and share growth/buybacks. For Deere, the historic growth trend has been 7.5% (the first fit referred to above). Looking at the trend from 1982-2004 (pre ethanol boom), the trend was 6.7%. These numbers were confirmed by looking at long term averages as well, which show and average of 7.6% and a median of 9.8%. For my estimation, I will use a low end sales growth estimate of 5%, a mid point of 6.5%, and a high end of 8% (I'm being conservative on this because I know the ethanol boom of the last 9 years won't be repeated).

Margin growth has varied widely over the last 32 years, but has generally trended up at a median rate of 1.6%. I think it would be imprudent to assume that Deere can recreate that accomplishment in the coming 5 years, so I will use a low end of 0% margin growth, a mid point of 0.5%, and a high end of 1% (I'm still assuming management can bring margins up with scale, manufacturing efficiencies, etc.).

Share count has also varied a lot over the last 32 years. In the more distant past, share count actually grew, but as management has refocused on building shareholder wealth, and been incentivized to do so, share count has declined at a median rate of 2.3% over the last 18 years and 3.9% over the last 10 years. I don't expect that high rate to continue assuming the agriculture market cools off, but I do expect a low end of 0% buybacks, a mid point of 0.5%, and a high end of 1%.

Putting together these pieces, I'm estimating 5% (5+0+0), 7.5% (6.5+0.5+.05) and 10% (8+1+1) growth rates at the low, mid and high ends.

Multiple

What multiple of earnings has the market been willing to pay for Deere? That has fluctuated widely, too. Because Deere is a cyclical business, investors have been willing to pay high multiples when earnings were low and low multiples when earnings were high. Multiples have also trended up over time as Deere has become a better business with wider profit margins. Given that, the median, low and high multiple to earnings over the last 32 years has been 10.5x and 16.5x, with 13.5x in the middle, so that is what I will use.

Expected returns

If you put together all the low, medium and high end assumptions above over a five year period, plus dividends growing at the same rate as sales per share and an $85 price tag on Deere, you get return expectations (annualized) of -2.9% 11.2% and 23.2%. Now, I assign a range of probabilities to those returns to come up with expected returns. Assuming a probability of 45% and 20% for the low end, 50% to 65% for the mid range, and 5% to 15% for the high end, I come up with a return expectation of 5.5% to 10.2%. (If you plug different numbers into the framework above, you can come up with vastly different results, so a lot depends on your assumptions being valid, or at least reasonable.)

This may not be the barn-burning return you expected, but it looks good compared to my projection of a 3.4% annualized return from the S&P 500 (at $1,982.30) over the next give years.

Keep in mind that my 5-10% return expectation on Deere is a long term projection. The path to that return may be bumpy, as I highlight below in my section on how bad things can get.

How low can you go?

To buy a cyclical company like Deere, it's not enough to have an idea what average future returns may be. You must also be ready to ride the cycle down to an uncomfortably low point, and be willing to buy more on that difficult trip down. This is particularly important with Deere because a long agricultural boom is coming to an end and farm equipment sales are clearly already tumbling. So, how bad can things get for Deere's stock price in between now and the long term?

One way to look at how low Deere's stock price can go is to look at multi-year sales per share (I use sales per share to account for the fact that earnings per share can get so low as to make earnings multiples meaningless) compared to the lowest multiples that have been experienced historically. Looking at an average of 3 year of sales per share relative to lowest annual prices, I can see that Deere got down to a 0.2x multiple of sales per share in 1986. Looking at 5 year average sales per share, 7 year, and 10 year, I see multiples of 0.4x, 0.5x and 0.5x. Below are the prices that Deere could get to, accordingly, from around $85 today:

  • 3 year sales per share, 0.2x multiple: $17
  • 5 year sales per share, 0.4x multiple: $29
  • 7 year sales per share, 0.5x multiple: $33
  • 10 year sales per share: 0.5x multiple: $29
I'm not predicting such low prices, but I am saying that Deere could get that low if history is a guide and an equivalently bad downturn occurs.  As I said above, caveat emptor. It should be noted, though, that I don't think the 1986 scenario is likely because this farm boom did not include the debt binge of that period (Kansas City Fed study), but it is best to consider all empirical evidence.

Another way to think about how low Deere's price can get is to look at prior peak to trough sales declines and apply low end multiples. Between 1982 and 1986, Deere's declined peak to trough by 24%. From 1990-1992, 16%; from 1998-1999, 19%; from 2008-2009, 21%. The 1982-1986 scenario, the worst I have on record, would see Deere's 7/31/13 LTM peak sales go from $35,250 to $26,920, or to $71 per share. The 1990 drop, the least bad drop, would pull sales down to $29,573 or $79 per share. Applying the 25th-percentile low multiple (0.5x in both cases) to those figures gives share prices of $35.50 and $39.50. 

A final way to prepare for low prices is to look at my normalized sales fits above and compare them to the lowest multiples of sales seen historically. The trough multiples on normalized sales were 0.2x sales per share in 1986, 0.4x in 1992, 0.6x in 1999, and 0.4x in 2009. Applying those multiples to the fitted sales per share above of $75, $90 and $95 gives price bottoms of $15-19, $30-38 and $45-47.  (Once again, keep in mind I consider the 1986 scenario quite unlikely.)

As I hope I've made clear, although I expect Deere to provide good long term returns, the path to those returns may be quite uncomfortable. Such is the nature of cyclical companies.

The upside is that Deere's price getting that low would likely generate truly amazing returns going forward (as they did for smart investors who bought in 1986, 1992, 1998 and 2009). Prices may never get that low, but it is best to prepare for such an eventuality even if it never occurs. Forewarned is forearmed.

Nothing in this blog should be considered investment, financial, tax, or legal advice. The opinions, estimates and projections contained herein are subject to change without notice. Information throughout this blog has been obtained from sources believed to be accurate and reliable, but such accuracy cannot be guaranteed.

No comments: