What is Moneyball for AgTech Automation?

I have written about “Moneyball” for AgTech Automation before, and wanted to dig in and start providing some details on what it means and what it could mean for ag operators, specifically as it relates to the portfolio of labor and automation solutions. The premise emerged from conversations I had with Jeff Morrison at Grimmway, who is one of the best at analyzing labor costs, identifying innovation that could help reduce labor costs and potential labor availability risks in an operating environment where labor is definitely getting harder to find and the cost is always going up because of both regulatory pressures (minimum wage, overtime, and break rules, as well as annual AEWR updates for H-2A immigrant labor wages) and market pressures such as general wage inflation (it’s a lot harder to get farm workers onto the farm when they’re working at McDonald’s and pulling down $20 or more an hour).

In the case of the actual Moneyball, Billy Beane used analytics to identify the baseball statistics that were often under-valued, and looked at the ones that were generally premium priced. What Beane found was interesting – big Sportscenter friendly stats like home runs (HR) and runs batted in (RBIs) were known historically and could be presented in the context of a highlight segment easily. On the other hand, emerging baseball nerd stats like wins above replacement (WAR) player, which compares that position to comparable positions and then looks at salary differential, and on-base percentage (OBP), which factors in how good a hitter is at getting on base in all ways (not just hits, but walks), were not well understood and a lot harder to make Sportscenter friendly. Beane decided to focus on lesser known hitters who didn’t hit the ball out of the park that often, but also didn’t strike out and drew a fair number of walks. The Moneyball component was finding the undervalued asset (WAR and OBP stats) and then trading players with better premium stats (HR and RBI) for seemingly “lesser” players that commanded lower salaries. The scenario played out well for Billy, and this kind of rationale helped both the Boston Red Sox and the Chicago Cubs win World Series rings.

So how do you translate it to AgTech automation? First, you identify an undervalued asset – in this case the impact automation can have on your labor portfolio. Second, you look at the premium product you are paying for now – in this case increasingly it’s immigrant labor that comes with large additional costs over and above domestic labor – housing ($10-15M for a 600-800 bed dorm complex in much of California depending on the amenities and location – and that’s a pre-inflation # that’s likely going up) and transportation (you need a way to get workers to and from the dorm complex and the work site every day – that’s a vehicle, gas, and insurance for the entire crew the entire time they’re participating in the program). Third, you look at your labor mix – how many domestic workers are you able to hire each year and is that number increasing or decreasing and do you need to bring in some immigrant labor to supplement the domestic labor force (or, in many cases, to make up for the short fall of labor you don’t have and can’t find – sometimes at any price). So those are the three big components to Moneyball for AgTech – labor usage, labor mix, labor cost, and innovation impact.

Now, let me add one last piece to the puzzle – how do you optimize the mix? It turns out this is the key to the model. In Billy Beane’s case, he needed a target to establish the trades he was willing to make. He started by looking at offensive trends over years relative to regular season win totals and win percentage. He determined the likely number of baseball games the team had to win to win the division (or at least to make the playoffs). Then he looked at the average number of runs per game, set a target of runs needed to get to that win total, and ran the analysis of every trade, free agent signing and draft pick on what he thought they would contribute to that run total for the season. Anytime he could do a deal that would add to the run total and get closer to the magic #, they looked hard at the deal and made a lot of transactions with that rationale. Similarly, anytime he could do a deal and reduce salary while still staying above the target season run total, he knew that left money in his pocket for other deals. Sportscenter was not kind to some of these deals – “how can Billy give that big name player with a huge salary up for these unknown guys when he’s in a playoff hunt?” they would scream coming out of a commercial? Billy knew the high profile stats guy was overpaid relative to value delivered and that money was better used on a couple up and comers who could deliver regular walks, higher OBPs, and get to the right run total.

So, knowing that, how do we take that part of the model to AgTech? Well, you start by looking at the total amount of production needed to hit sales totals for the year to get a annual yield total (sounds a lot like a run total for the A’s). Then you look at the current labor mix needed to get there and project it across the entire growing season. How many domestic workers for how many hours across each activity (planting, weeding, thinning, harvesting, field prep)? How many immigrant workers for how many hours for the same activities? Are you hiring them directly, through a farm labor contractor (FLC), or a mix of both (it impacts the price – and the risk)?

Then the key analysis – what are the innovation options that could reduce the availability risk or the overall cost of the labor portfolio? This requires field trials and validation of each of the innovation solutions. Then it requires some capital stack development – how long before that $1.4M laser weeder ROIs to the operation and how long for the same question on 5 $20,000 Burro units. As important, what’s the cost of capital for the capital expenditures of buying the equipment? That has to be factored into the analysis. It’s not enough to just buy the automation solution – it has to meaningfully advance things for the operation or you’re not doing Moneyball right if the capital costs aren’t factored in. You can optimize the portfolio by bringing in automation to do some or all of a key task, and you can bring in things like the Burro to increase the crew efficiency. Both can improve the portfolio – doing the analysis determines which one should be the investment priority because no grower operation I work with has unlimited capital to spend on innovation.

We have been pretty busy working at the intersection of these principles – right between the grower, the FLC, the worker, and the innovators. Ben Palone has done some fantastic work in digging into the details of what these costs look like for WG members. Lynn Hamilton and Mike McCullough are leading a team at Cal Poly focused in on the current and projected labor trends in terms of mix (domestic v H-2A) and cost by key crop types. Finally, all of us are starting to work towards a second variant of Moneyball. Each of the growers is like the Oakland A’s and optimizing their portfolio for their team.

But Western Growers wants to try and help develop Moneyball for the whole league (i.e. all the growers). So we are heading down multiple paths at the same time. First, Ben is building out field trial data with automation startups that will include some of the most detailed economic analyses I’ve seen to date – very thorough and completely validated by the grower. Second, one we have permission from the grower and the startup involved, Western Growers will put out WG-branded case studies that are open sourced and shared with WG members and other growers so everyone can see how the math actually works. Third, we are going to turn the case study details into general-purpose AgTech economic templates all can use. Fourth, and here’s where it gets really interesting – we are going to engage some contractor resources to work with growers and help them do the same discovery and economic analysis that Ben has been doing for their own operations. It turns out not every grower has a Jeff Morrison – some do, but a lot don’t. It’s a big commitment for an ag operator to make, and many don’t have the need or the capital to put into that position. Fifth, and equally important, that same template should be used by investors and startups so that when they want to perform due diligence on the startups or pitch to investors better, everyone’s using the same playbook.

So those are the key steps we are going to be doubling down on in automation for 2024:

1)???? Field trial data – we have some initial data, we will be pushing hard for a lot more.

2)???? Case studies – tell the story of how the innovation was able to help the operation and what the economic impact was for the organization with WG-branded case studies.

3)???? AgTech Economic templates – publish these as open sourced documents available for all with documentation explaining how to use them.

4)???? Add some resources to the team to help other WG members use and get the benefit of the templates.

5)???? Roll out the same templates to AgTech investors and startups so everyone can start looking at and thinking about the innovation impacts the same way.

So while individual growers will continue to play Moneyball for their operations teams, Western Growers is gong to be playing Moneyball for the whole league (or at least the whole league of WG growers). We will have a lot of great stuff to share this quarter – news on how we’re scaling field trials, 2 case studies are in draft mode (and the next 3 are in the hopper), the template is almost built, and we will be adding some resources to work with growers next quarter.