Cognitive Biases and Improved Decision Making for Agribusiness Leaders

Note: For Upstream Ag Professional members, the full Report is available for download below.

Cluster #6 Strategic and Organizational Inertia

How to Implement and Final Thoughts

Every strategic decision in agribusiness passes through a human brain before it becomes action. That brain, regardless of the education or the decades of experience informing it, is running on cognitive software that evolved to avoid predators, and prioritize short term constraints, like food management or safety, not to optimize for being different, thinking long term or building systems that create success in business.

These sub-optimal approaches to thinking and decision making are known as cognitive biases.

They are patterns deviating from rational judgment and are not occasional mistakes or signs of low intelligence, however, they are predictable, measurable, and universal tendencies that affect how we process information, evaluate risk, remember outcomes, and interact with others.

They have long been central to my considerations when decision-making — though I remain far from perfect at managing them.

Early in my career, when I was training sales agronomist on agronomics or customer relationship building, I would open presentations by walking through the specific biases that would undermine the concepts I was about to teach, or that were driving the problems we were trying to solve.

The idea was to help individuals prime their minds to understand why how we think can work against us and bring a novel approach to catalyzing the training or conversation.

I think agribusiness professionals operating in high stakes environments should be even more aware of these biases – for those allocating capital and resources, managing a team, or setting strategic priorities. The higher the stakes and complexity of a role, the more cognitive biases can distort our judgment in ways that are hard to detect in the moment.

Warren Buffett’s long term business partner, the late Charlie Munger is known for bringing theminto the  mainstream, in part thanks to his famous speech from 1995 at Harvard called The Psychology of Human Misjudgement.

However, one of the individuals who was most influential in researching and discovering them, Daniel Kahneman, did more work than anyone surrounding these concepts. Thinking Fast and Slow introduced millions to a framework for understanding how and why we consistently make suboptimal decision. Yet what always stood out to me was that even Kahneman acknowledged he remained susceptible to the very biases he spent his career studying. 

I highlight this because full elimination shouldn’t be the goal, but awareness, combined with knowing when a particular bias is likely to surface, can meaningfully reduce the frequency and severity of mistakes — plus, enables professionals to alter environments for their teams and staff to be able to better manage them from making suboptimal decisions, too. There are more than 200 recognized cognitive biases, however, over the last decade I have collected some of my favorite that I review consistently to try and improve my awareness of them. I wanted to take my rough document and piece it together with some coherence and apply it to agribusiness and share it as a useful resource for those interested in improving their psychological understanding and decision making abilities.

The focus on this overview is making better decisions. However, there is an inverse value — understanding these biases also allows you to exploit them. For example, in pricing, or idea positioning or in sales or program design. Knowing how people tend to think gives you an advantage in many areas of business.

This breakdown examines 40 cognitive biases through the lens of agribusiness professional decision-making, for the likes of boardrooms, R&D labs, commercial teams, and executive suites where strategic decisions can influence billion dollar of outcomes.

For each bias, we cover four dimensions:

The biases are organized into six clusters based on the cognitive domain they primarily affect. For example, many biases interact with and reinforce each other. Loss aversion feeds the disposition effect. Confirmation bias amplifies groupthink. The narrative fallacy makes survivorship bias feel like insight. Understanding these interactions is as important as understanding the individual biases and I encourage you to keep in mind the relationships as you review them.

The sunk cost fallacy was popularized by behavioral economist Richard Thaler, drawing on numerous individuals' work over decades prior.. The bias is a commonly referenced one in economics classes.

The core finding is that humans treat spent resources as ongoing obligations rather than what they are, which is irretrievable costs that should have zero weight in forward-looking decisions.

Sunk cost fallacy is a problem because it converts every business decision from "what produces the best outcome from here" into "how do I justify what I already spent," which is a fundamentally different and worse question when trying to optimize future outcomes.

An ag retailer invested $4 million over three years building a proprietary digital agronomy platform rather than partnering with a 3rd party to license their software internally on an annualized basis. The intended budget was for $7 million total.

By year two, it was clear the platform couldn't match the functionality of established competitors, adoption among retail teams was low, and development costs were accelerating. But the leadership team kept funding it, reasoning that walking away would "waste" the $4 million already spent, and the two years (time is also a sunk cost). They eventually shut it down at $4 million total, having delayed a partnership with an existing platform that would have been operational 18 months earlier. The $7 million was gone either way. The real cost was the $3 million in additional spending and the lost revenue from the delayed partnership.

Pre-commit to decision review triggers before launching any major initiative.

Set specific milestones (revenue targets, adoption metrics, cost ceilings) and agree in advance that missing them triggers a formal kill-or-continue review.

Some research has shown that separating the people who made the initial investment from the people who evaluate its continuation significantly reduces sunk cost bias. In practice, this means having someone other than the project champion run the review.

Frame every evaluation around "if we were starting from scratch today with this information, would we fund this?" If the answer is no, the project should stop.

Loss aversion is one of the foundational discoveries of prospect theory, published by Daniel Kahneman and Amos Tversky. Their research demonstrated that losses are psychologically weighted approximately twice as heavily as equivalent gains, meaning a $100,000 loss feels roughly twice as painful as a $100,000 gain feels good. This asymmetry distorts rational decision-making because it causes people to take irrational risks to avoid losses and to be excessively conservative when potential gains are available.

In business settings, loss aversion is responsible for a massive amount of value destruction through inaction and missed opportunity.

Consider a major seed company evaluating whether to exit a legacy trait platform and consolidate R&D resources into a next-generation biotech pipeline. The legacy platform still generates $80 million in annual revenue but is declining 8-10% per year and consuming significant R&D and regulatory maintenance budgets. The new pipeline has higher expected value but won't generate revenue for three to four years. Loss aversion makes the $80 million in visible, current revenue feel more "real" than the projected future value. Leadership delays the transition repeatedly, allowing the legacy platform to consume resources that would accelerate the replacement, ultimately resulting in a weaker competitive position in both the declining and emerging markets.

Notably, this is a core component of “disruption theory” by Clayton Christenson and what drives businesses to optimize for what they already have, rather than what is possible.

Kahneman's own recommendation is to adopt a policy-based approach rather than evaluating decisions in isolation. When you evaluate each decision independently, loss aversion dominates. When you aggregate decisions into a portfolio, the math overwhelms the emotion.

Practically, this means establishing decision frameworks with pre-set criteria for market exits, product sunsets, and resource reallocation before the emotional weight of specific losses enters the picture. Research in 2009 also showed that deliberately reframing losses as costs (which feel routine) rather than losses (which feel painful) reduces loss-averse behavior by about 50%.

The endowment effect was identified by Richard Thaler. Using the "mug experiments" showed that people demanded roughly twice as much to sell an object they owned as they would pay to acquire it. Ownership itself, independent of any rational valuation, inflates perceived value. The business implications come back to things lik overvaluing their own assets, divisions, brands, intellectual property, and market positions simply because they own them, leading to poor sell/hold/invest decisions across the portfolio. On an individual basis, this means we also place a higher value on our own ideas, leading to another bias frequently referenced called “Not Invented Here Bias.”

A mid-size agricultural equipment dealer group owns a network of 12 locations across three states. Market analysis consistently shows that three of these locations are underperforming, dragging down returns on capital. A competitor offers to acquire all three at fair market value. The dealer group's leadership declines, convinced the locations are "worth more than that," despite the fact that their own financial analysis supports the offered price. The endowment effect is inflating their valuation of what they own.

Implementing what behavioral economists call the "trading paradigm" in asset evaluations. For any asset the company holds, regularly ask: "If we did not currently own this, would we acquire it today at its current carrying cost?"

This reframes the question to eliminate ownership bias.

Research suggests that creating formal valuation processes using external benchmarks (comparable transactions, market multiples, replacement cost analysis) rather than internal assessments reduces the endowment effect substantially. 

Having third-party valuations as a standard part of annual asset reviews provides a check against inflated internal estimates.

Mental accounting was described by Richard Thaler as the tendency to categorize and treat money differently depending on its source, intended use, or the mental "account" it sits in, even though money is perfectly fungible. A dollar of profit from grain trading is economically identical to a dollar from custom application services, but companies routinely treat them differently, investing more carefully from some "accounts" and more loosely from others. We see this in our personal lives with money coming from a tax refund, or a company bonus, viewed as something to be spent differently than we would regularly do with our core salary.

Mental accounting creates inefficiencies in capital allocation because resources don't flow to their highest-return use, they flow according to arbitrary categorical boundaries.

An agribusiness cooperative receives a significant patronage distribution from a joint venture investment. Rather than allocating this capital through their normal investment evaluation process, the board treats it as "found money" and approves a facilities upgrade that had previously been rejected as below their hurdle rate. The economics of the project didn't change. The source of funding changed, and because the money felt like a windfall rather than operating capital, it faced a lower decision-making standard. Over a five-year horizon, that capital deployed at the originally rejected project's actual return cost the cooperative roughly $2 million in opportunity cost versus their standard investment options.

Thaler's own prescription is to make all capital allocation decisions through a single, unified evaluation process regardless of funding source. Every dollar should compete for deployment on the same terms. Practically, this means eliminating categories like "windfall funds," "insurance proceeds," or "one-time gains" from capital allocation vocabulary and routing all investment decisions through identical hurdle rate requirements. Some research shows that simply making people aware of their mental accounting tendencies reduces the bias, so training leadership teams on this concept has direct value.

The disposition effect was identified by Hersh Shefrin and Meir Statman, drawing on Kahneman and Tversky's prospect theory. It describes the tendency to sell winning positions too early and hold losing positions too long. This is loss aversion applied specifically to portfolio and position management. The mechanism is surrounding the fact that selling a winner locks in a gain (which feels good), while selling a loser realizes a loss (which feels bad), so people systematically do more of the former and less of the latter. The result is a portfolio that gradually concentrates in losers and sheds winners, the exact opposite of optimal position management.

A large grain trading firm's merchandising desk holds both a profitable long position in soybeans (up $1.2 million) and an underwater short position in corn (down $800,000). Market fundamentals suggest the soybean rally has room to run while the corn position is likely to deteriorate further. The desk liquidates the soybean position to "lock in the win" and holds the corn position hoping for a recovery. Over the next 60 days, soybeans gain another $600,000 of value and corn losses deepen to $1.4 million. The desk left $600,000 on the table and absorbed $600,000 in additional losses, a $1.2 million swing from a single behavioral error that had nothing to do with market analysis.

Set pre-defined targets and stop-losses for every position at the time of entry, and execute them mechanically is one commonly referenced approach to mitigation. Position management software that automates exits removes the emotional component entirely. Some firms use a "blind review" approach where position managers evaluate whether to hold or exit based on current market fundamentals without being told whether the position is currently profitable or not. This eliminates the reference point that triggers disposition-driven behavior.

Scope insensitivity (also called scope neglect) was demonstrated by Kahneman and colleagues  through studies showing that people's willingness to pay to solve a problem is remarkably insensitive to the magnitude of the problem. Participants would pay almost the same amount to save 2,000 birds as to save 200,000 birds. The emotional response to "birds in danger" is what drives the reaction, not the scale. In business, this manifests as agonizing over small decisions with the same intensity applied to decisions that are orders of magnitude more consequential, consuming scarce leadership attention on trivial matters while under-analyzing transformative ones.

An agribusiness holding company's executive team spends 90 minutes in a board meeting debating the design and cost of a new $80,000 employee break room renovation. The same meeting allocates 30 minutes to reviewing a $12 million acquisition of a specialty fertilizer manufacturer. The break room triggers detailed opinions from every board member because it is tangible, understandable, and emotionally engaging. The acquisition involves complexity that is harder to discuss, so the board defers to the CFO's recommendation. The emotional engagement is roughly the same for both decisions despite a 150x difference in financial magnitude. Parkinson's Law of Triviality describes a similar dynamic, but scope insensitivity is the underlying cognitive mechanism.

Research suggests that joint evaluation (presenting decisions side by side with their relative magnitudes made explicit) significantly reduces scope insensitivity. Structure agendas and meetings to explicitly weight discussion time proportional to financial impact. Some companies implement a simple rule: minutes of discussion must be proportional to the dollar value at stake, with a visual timer in the room. In Mindware: 33 Mental Models for The Modern Agribusiness Leader I highlight Jeff Bezos’ “One-Way vs. Two-Way Door” framework for prioritizing decision making as well.

Confirmation bias is arguably the most pervasive and well-documented cognitive bias in the literature. Peter Wason first demonstrated it in 1960 with his "2-4-6 task," showing that people systematically seek information that confirms their existing hypotheses while ignoring disconfirming evidence. Decades of subsequent research have reinforced its universality and I know I commonly have the biggest issue with this bias. The problem is not that people lack access to good information, but that we unconsciously filter information to reinforce what they already believe. In complex business environments with abundant data, confirmation bias worsens because there is always some data point that supports whatever position you want to defend.

A crop protection company's product development team is convinced that their new fungicide formulation is superior to competitors based on early-stage trial data. As field trial results come in over two growing seasons, the team emphasizes the sites where performance was strong and attributes poor-performing sites to "unusual conditions," late application timing, or tank mix incompatibility. An objective analysis of the full dataset shows the product is statistically equivalent to existing market options, not superior. But by the time this analysis reaches leadership, the product has been branded, priced at a premium, and launched. First-year sales disappoint, and the company spends two additional years adjusting positioning before accepting the product's actual market fit.

Wason's original research and some subsequent work converged on the same prescription: actively seek disconfirming evidence. Institute formal "red team" or "pre-mortem" reviews for major decisions where a designated team is tasked specifically with building the case against the proposed action. Research by Koriat, Lichtenstein, and Fischhoff (1980) showed that simply asking decision-makers to list reasons their preferred option might be wrong reduces confirmation bias measurably. In product development, require that trial data analysis be conducted by a team separate from the development team, with explicit instructions to identify conditions under which the product underperforms. Make disconfirming data as visible as confirming data in review presentations.

Note: There is risk of generative AI to supercharge this bias. However, I believe that there is also a way to exploit generative AI to mitigate. One way I do this is when I am making a statement within Upstream Ag Insights, I often ask Claude (my current generative AI system of choice) to argue the other side, tell me what I am missing and tell me where my thoughts are inadequate given disconfirming evidence.

Anchoring was first described by Tversky and Kahneman as part of their foundational heuristics and biases research program. They demonstrated that when people make numerical estimates, they are heavily influenced by whatever number they encounter first, even when that number is arbitrary or irrelevant. Subsequent research has shown that anchoring operates even when people are explicitly warned about it and even when the anchor is clearly random. The bias is remarkably resistant to debiasing, which makes it one of the most dangerous in business settings where historical figures, initial offers, and precedent prices are constantly present.

A large agribusiness is negotiating the acquisition of a specialty seed company. The target's investment banker opens with an asking price of $120 million. The acquiring company's internal valuation, completed before the ask, estimated fair value at $75-85 million. Despite this analysis, the negotiation team finds itself working to "get the price down to $100 million," which feels like a significant win relative to the $120 million anchor. They close at $98 million and celebrate the $22 million "discount." They overpaid by $13-23 million relative to their own pre-negotiation valuation, but the anchor shifted their reference point so effectively that the overpayment felt like a bargain. The $120 million number was never a valuation. It was a strategic anchor.

Research by Mussweiler, Strack, and Pfeiffer demonstrated that the most effective counter to anchoring is generating your own anchor before encountering external ones. In negotiations, this means completing rigorous independent valuations before any engagement with the counterparty and physically writing down your range before seeing their number. Epley and Gilovich showed that anchoring from self-generated values is significantly weaker than anchoring from external values. In pricing and procurement contexts, train teams to recognize anchor-setting behavior and to reset discussions around independently derived fundamentals. "What is this worth based on our analysis?" must always precede "what are they asking for?"

The framing effect was demonstrated by Tversky and Kahneman with their "Asian disease problem," showing that identical outcomes described in terms of lives saved versus lives lost produced dramatically different choices. The same expected outcome framed as a gain triggers risk-averse behavior, while framed as a loss it triggers risk-seeking behavior. This can be problematic for businesses because virtually every strategic decision can be framed either way, and the frame rather than the substance often determines the outcome. Whoever controls the framing controls the decision.

An agricultural lender presents two identical loan restructuring options to a struggling agribusiness client. Option A: "This restructuring preserves 70% of your current operating capacity." Option B: "This restructuring requires giving up 30% of your current operating capacity." They are the same restructuring. But Option A framing leads the client to accept readily, while Option B framing triggers resistance and counter-negotiation. Internally, the same dynamic plays out when leadership presents strategy changes. "We're reallocating resources to our highest-growth divisions" and "We're cutting investment in our legacy businesses" describe identical decisions but produce different organizational reactions and different levels of resistance.

Tversky and Kahneman's own recommendation is to evaluate decisions under both frames and check for consistency. If a decision changes when you reframe it, the frame is driving the choice rather than the substance. Later research categorized framing effects and found that attribute framing (how a single characteristic is described) is the most resistant to changing the bias, while risky choice framing (the original Tversky and Kahneman variety) responds well to forcing consideration of both framings simultaneously. In practice, requiring that all major proposals be presented in both "gain" and "loss" language and flag any decision where the two framings produce different preferences.

The availability heuristic was identified by Tversky and Kahneman as the tendency to estimate the probability of an event based on how easily examples come to mind. Events that are vivid, recent, or emotionally charged feel more probable than they actually are, while events that are abstract, distant, or undramatic feel less probable. This is one of the most practically important biases because it distorts risk assessment, causing organizations to over-prepare for dramatic but rare events and under-prepare for mundane but frequent threats.

Following a major cybersecurity breach at a competing agribusiness company that made industry headlines, a large grain merchandising firm's board directs $2.5 million in emergency spending toward cybersecurity upgrades. The vivid, recent, and dramatic nature of the competitor's breach made cyber risk feel urgent and probable. Meanwhile, the firm's internal audit had identified chronic inventory management system failures that were quietly generating $1.8 million in annual reconciliation losses, ongoing for three years. This slow-moving, undramatic risk never commanded the same urgency because it didn't produce a memorable event. The cumulative $5.4 million in inventory losses dwarfed any realistic cyber exposure estimate, but availability drove the capital allocation.

Research has shown that base-rate information, when made concrete and salient, can counteract availability bias. Build enterprise risk assessments on data (historical frequency and magnitude) rather than recent events. Require that any risk assessment triggered by a specific event also include a comparative analysis of all risks in the same category, ranked by expected annual loss rather than recency or vividness. Sunstein research recommends institutional "availability audits" that compare organizational attention allocation to statistical risk exposure. When the two diverge significantly, it signals availability-driven rather than evidence-driven resource allocation.