Milton Friedman’s Instrumentalism: Do Unrealistic Assumptions Matter?

Imagine you’re watching an expert billiard player sink shot after shot with remarkable precision. Does the player actually calculate complex angles, velocities, and trajectories using mathematical formulas? Probably not. But what if we could predict their shots by assuming they do? This thought experiment lies at the heart of one of economics’ most controversial debates: whether the truth of a theory’s assumptions matters at all, or whether only its predictions count.

In 1953, economist Milton Friedman published an essay that would become one of the most influential and hotly debated papers on economic methodology. His argument challenged conventional thinking about how economic theories should be evaluated, sparking a controversy that continues to shape economics today. At the center of this debate is a simple but radical question: do unrealistic assumptions matter?

The core argument: predictions over descriptions 🔗

Friedman’s central claim was straightforward yet provocative. He argued that the ultimate goal of positive science is to develop theories that yield valid and meaningful predictions about phenomena not yet observed. According to this view, a hypothesis should be tested solely by the accuracy of its predictions, not by whether its assumptions realistically describe the world.

Think about weather forecasting for a moment. Meteorologists use simplified models of atmospheric behavior that ignore countless variables like individual butterfly wing flaps or the exact temperature of every molecule. These models work because they capture the essential relationships that matter for prediction, even though their assumptions abstract away from the full complexity of reality.

Friedman took this logic further than most scientists would feel comfortable with. He suggested that in economics, we shouldn’t worry too much about whether businesses actually engage in the complex calculations our theories assume they do. What matters is whether the theory predicts market outcomes accurately.

The controversial F-twist: more unrealistic means more significant 🔗

Friedman went beyond merely defending unrealistic assumptions. He made a claim that left many economists and philosophers stunned. In a famous passage, he wrote that truly important and significant hypotheses will have assumptions that are wildly inaccurate descriptive representations of reality, and that generally, the more significant the theory, the more unrealistic the assumptions.

This assertion became known as the “F-Twist,” a term coined by Paul Samuelson, who reportedly chose not to name it after Friedman directly out of courtesy. The logic behind this counterintuitive claim was that powerful theories extract only the crucial elements necessary for prediction, deliberately omitting predictively irrelevant details. A truly abstract theory, by definition, strips away the messy particulars of reality.

Consider economic models that assume perfect information or perfectly rational actors. No real person has perfect information about all market conditions, and humans are notoriously prone to cognitive biases and emotional decisions. Yet Friedman argued that these wildly unrealistic assumptions could still form the basis of useful economic theories, as long as the predictions held up.

Why this matters in practice 🔗

The practical implications of the F-Twist are significant. If Friedman is right, economists can build theories based on simplified assumptions about human behavior without worrying whether those assumptions accurately describe how people actually think and decide. This approach has shaped much of modern economics, particularly in areas like market theory and financial economics where mathematical elegance often takes precedence over psychological realism.

The as-if methodology: acting without calculating 🔗

To defend his position on unrealistic assumptions, Friedman developed what became known as the “as-if” methodology. His most famous illustration involved an expert billiard player who makes shots as if they knew complicated mathematical formulas, could estimate angles accurately by eye, and could make lightning calculations.

Friedman’s confidence in this hypothesis wasn’t based on believing that billiard players actually go through such calculations. Rather, it came from the belief that unless they were somehow capable of reaching essentially the same result, they wouldn’t be expert players in the first place. Natural selection, so to speak, weeds out the bad players.

He applied the same logic to business behavior. According to Friedman, firms don’t need to consciously work through marginalist calculations about maximizing profits. They merely need to behave as if they do for economic theory to have predictive power. Businesses that happen to act in ways consistent with profit maximization survive and thrive, while those that don’t eventually fail. Market competition acts as a selection mechanism, ensuring that surviving firms behave approximately as the theory predicts, even if they don’t consciously follow its logic.

The appeal and the problem 🔗

This argument has intuitive appeal. After all, we often achieve good outcomes through trial and error, habit, or instinct rather than explicit calculation. A child catches a ball without understanding projectile motion. A chef adjusts seasoning without calculating chemical interactions. Why should business decisions be any different?

But critics pointed out significant flaws in this reasoning. The billiard player analogy works only in very specific, controlled conditions. In the complex, ever-changing environment of real markets, the selection pressures Friedman described may not operate consistently or quickly enough to guarantee that surviving firms behave as the theory predicts.

The backlash: measurement without theory 🔗

Friedman’s instrumentalism faced widespread and sustained criticism from economists and philosophers alike. The core concern was that by completely ignoring the realism of assumptions, Friedman’s approach could lead to spurious correlations and what critics called “measurement without theory.”

Consider a simple example. Suppose you notice that ice cream sales and drowning deaths both increase during summer months. A purely prediction-focused approach might lead you to conclude that ice cream causes drowning. The prediction might even hold up for a while. But because the theory ignores the underlying causal mechanism (hot weather causing both), it fails to provide genuine understanding and breaks down when applied to different contexts, like indoor swimming pools in winter.

Samuelson’s logical critique 🔗

Paul Samuelson offered a particularly damaging logical critique of the F-Twist. He pointed out that if a theory’s assumptions and conclusions are logically connected, then false assumptions necessarily imply false conclusions. If we shield assumptions from requiring realism, Friedman essentially admits falsehoods as part of his theory. Moreover, requiring only certain phenomena to be explained could lead to unscientific cherry-picking of results.

Imagine a doctor who develops a treatment based on a completely wrong understanding of how a disease works, but the treatment occasionally produces good outcomes by accident. Would we consider this good medical science? Most would say no, because without understanding the underlying mechanisms, we can’t know when the treatment will work, when it will fail, or how to improve it.

The truth value objection 🔗

Philosophers of science raised another fundamental objection. They argued that Friedman’s instrumentalism ignored what they called the “truth value” of theories. Science, they contended, should aim not merely at predictive success but at genuine explanation and understanding of the world. A theory that makes accurate predictions for the wrong reasons provides neither true knowledge nor reliable guidance for new situations.

This criticism highlights a tension at the heart of economics. Should the field aim to understand how economies actually work, with all their psychological complexity and institutional particulars? Or should it aim primarily to generate useful predictions, even if the theories bear little resemblance to economic reality? Different answers to this question lead to fundamentally different approaches to economic research.

The lasting impact and ongoing debate 🔗

Despite the criticisms, Friedman’s essay became the most influential work on economic methodology of the twentieth century. By the late 1980s, his core claims about focusing on predictions rather than realistic assumptions had become widely deployed in mainstream economics, even as methodological purists continued to object.

The instrumentalist approach Friedman championed has shaped modern economics in profound ways. It provided intellectual justification for highly abstract, mathematically sophisticated models that make little pretense of describing actual human psychology or institutional detail. This has enabled tremendous theoretical sophistication but has also contributed to growing concerns about economics losing touch with reality.

Interestingly, some scholars have noted that Friedman himself didn’t always follow his own methodological prescriptions. In his empirical work, such as the famous Monetary History of the United States, he sought what has been called “thick evidence,” drawing on multiple types of mutually corroborating data rather than relying solely on narrow predictions.

Where do we stand today? 🔗

The debate Friedman sparked continues to evolve. Recent decades have seen an “empirical turn” in economics, with greater emphasis on careful causal identification and realistic behavioral assumptions. Behavioral economics, in particular, has challenged the instrumentalist approach by showing that understanding actual human psychology leads to better predictions and more effective policies.

Yet the core tension remains. Economics deals with enormously complex systems involving millions of interacting human beings, evolving institutions, and shifting technologies. Some degree of simplification and abstraction is inevitable. The question is how much abstraction is too much, and when unrealistic assumptions cross the line from useful simplification to dangerous distortion.

What do you think? Should economic theories be judged primarily on their predictive accuracy, or does understanding the true mechanisms behind economic phenomena matter? When you make decisions about your own career, investments, or business, do you rely more on tested rules of thumb or on understanding the underlying causal relationships?

References

1. https://en.wikipedia.org/wiki/Essays_in_Positive_Economics
2. https://eh.net/book_reviews/the-methodology-of-positive-economics-reflections-on-the-milton-friedman-legacy/
3. https://bcmullins.github.io/economic_methodology_interpretable_ml_friedman1953/