The limits of the scientific method in economics and the world

November 11, 2011

By Roger Martin
The opinions expressed are his own.

Part one of this essay was published Thursday. This is part two.

As the power of the scientific method has encroached further than its applicability warrants into fields such as economics and business, its predictions of the future become ever more erroneous. In this list, we can include virtually every economic prognostication from the first half of 2008, and countless market research studies that misjudge consumer interest in the new product concepts that they test.

Were he alive today, Charles Sanders Peirce, a turn-of-the-20th-century American pragmatist philosopher, would be unsurprised at the shoddy results of these analyses. A brilliant thinker, Peirce is less famous than his peers William James and John Dewey, mainly because he was an ornery and unpleasant character.  However, Peirce had an almost entirely overlooked yet extremely insightful theory applicable to modern scientific work. He concluded that no new idea was ever derived from the analysis of the past using inductive and deductive logic – the two forms of logic our modern scientific method utilize.

These forms of logic can only prove things in the past, so whether business people, economists and scientists realize it or not, the only thing about the future that the application of  deductive or inductive logical analysis can predict is a simple extrapolation of the past.  That is very useful for studies of things in the Aristotelian camp of cannot be other than they are. If we find some property of granite by studying existing hunks of granite, we can safely extrapolate that in the future, hunks of granite will have those properties.

However, deductive and inductive logical analyses aren’t so hot for things that can be other than they are – like economy for example.  These things change constantly due to the interactions of the people and organizations. The fall of 2008 wasn’t an extrapolation of the past – it was discontinuous with the past.

If economists were being Peircian, when they saw an anomalous situation of the unprecedented economic meltdown of fall 2008, they would have taken all of our disparate knowledge and mingled it together creatively to ask: what on earth could be going on here? Rather than forgetting entirely that their theories were demonstrated to be totally lacking, and then going on to analyze some more and predict more based on those theories, they would have created a new hypothesis to explain what just happened.  This would be what Peirce evocatively called ‘a logical leap of the mind’ and ‘an inference to the best explanation.’

That is the merger of science and innovation.  It is what Peirce called ‘abductive logic’.  It is the formation of a new hypothesis.  Deductive and inductive logic can’t make the future anything but the logical extension of the past and what occurred in the fall of 2008 simply wasn’t a logical extension of the past.  Peircian abductive logic could have created a new hypothesis that could be analyzed going forward, as the data streamed in.

This is certainly not the universal or even the majority instinct in our modern world. Rather, there is misplaced trust in the analysis of data to reveal some truth. Analysis certainly can: if you torture the data enough, it will give some answer, if only to make the pain go away. The problem is with the low information content of the answer. To Peirce, if an existing deductive rule is applied to the analysis of historical data, it may confirm that the rule is operative, in which case nothing new has been learned.  If instead the rule is dis-confirmed (for example that in the back half of 2008, the U.S. economy didn’t perform the way we would have expected it to do), then we have learned that our deductive rule is not operative – a good thing but not a clue to the future.  If inductive logic is applied using a large and sufficiently homogenous sample size, we will wait a heck of a long time to get a sample size that will enable us to confidently make an assertion that what we see is true.

Aristotle would argue that the Peircian logical leap is a product of rhetoric: the dialogue between inquiring minds that attempts to create a future that does not now exist, rather than mindlessly crunching the numbers that do exist.  Rather than ask whether the new idea can be proven in advance analytically, these inquiring minds would ask how this novel idea could be tested going forward. If we listened to all of Aristotle (rather than just half of him) and mixed it with a little Peirce, we would accept the limitations of science to shape that which can be other than it is, and would spend more time and energy engaging in the sort of dialogue that creates new models that can be tested going forward.

Of course, I am not arguing that all science is devoid of novel hypotheses or that all natural and social scientists eschew abductive logic entirely.  Rather, I am arguing that we need to reign in faux science, which is appropriate only for understanding things that cannot be other than they are, using the tools of deductive and inductive logic.  And we need to release the energy of a broader conception of science and innovation that helps us to shape those aspects of our world that can be other than they are, using abductive logic.  It won’t always be clear in advance which is which, but it is important that we not default reflexively to analysis rather than innovation. Restoring that balance is the modern – or perhaps more appropriately – post-modern challenge of our world.

Part one of this essay was published Thursday. This was part two.


We welcome comments that advance the story through relevant opinion, anecdotes, links and data. If you see a comment that you believe is irrelevant or inappropriate, you can flag it to our editors by using the report abuse links. Views expressed in the comments do not represent those of Reuters. For more information on our comment policy, see

[…] This is part one of this essay. Read part two here. […]

Posted by The limits of the scientific method in economics and the world | The Great Debate | Report as abusive

you will never be able to predict the future. The meltdown although unpredictable, conformed to classical economic theory. Assets were overvalued and demand plummeted. No big deal, basically bubbles happen because of a lack of transparency and overvaluation of assets.

The root cause of the meltdown of 2008 was exactly the “solution” you are suggesting. We should not dare to take up your suggestion. New economic models were applied, which only mathemeticians understood. These models assumed that housing would continually increase in value, and were the basis for over-rating morgage-backed securities. This model was disproven -but didn’t we know all along that housing values cannot rise infinitely? The fraud was commited by the rating agencies.

The only truly innovative model as an alternative model to capitalism was presented by Marx. And just like capitalism , it has never been successful at completely removing all the problems that people have.

The discussion should be more about what we need to acheive as a society and how economic theory can be applied to solve problems.

Posted by JFG | Report as abusive

Tommyrot. You, sir, clearly don’t understand the scientific method which doesn’t start until you have a hypothesis. The scientific method does not allow you to prove or disprove a hypothesis via any form of logic. It requires you to devise a test – a physical, real world experiment – to see if your hypothesis accurately predicts the results of that experiment. That experimentation and testing allows you to either prove (and/or refine) or disprove your hypothesis. But in no case does the scientific method create your hypothesis. Plausibly you’re thinking of epidemiology which, despite it’s name, isn’t science at all but merely statistics and does exhibit the flaws you describe.

It’s so very kind of you to admit that not all science is devoid of novel hypotheses considering the huge advances in our understanding of how the world works and the natural processes that underlie it all that occurred in the last century (for example) – advances that dwarf anything economics has managed to make. But the scientists who advanced these novel hypotheses were not creating a future that did not exist. The quantum nature of the universe existed all along; it did not suddenly arise because scientists proposed that it might be so.

I find your arguments rather devoid of merit. The only bit in your argument that does have merit is rather trivial. That would be “Rather than forgetting entirely that their theories were demonstrated to be totally lacking, and then going on to analyze some more and predict more based on those theories, they would have created a new hypothesis to explain what just happened”. Any scientist will tell you that when you are presented with results that are not explained by existing theory, you have no choice but to propose hypotheses which will explain it. You then must begin the arduous task of determining the hypothesis which best explains both the phenomenon you observed as well as all other phenomenon previously explained by your theories.

Posted by majkmushrm | Report as abusive

I’m glad I read this article, because now I know to avoid this author. What a load of steaming tripe! The scientific method is: the worth of a theory rests on its falsifiability by experiment. it’s not about ab/in/deduction.

rather unfortunate that a dean doesn’t have someone to call him out when he’s blathering.

Posted by markhahn | Report as abusive

“Deductive and inductive logic can’t make the future anything but the logical extension of the past and what occurred in the fall of 2008 simply wasn’t a logical extension of the past.”

Sure it was – we just didn’t have a model complex enough to predict it. But in hindsight we can see that it all started in the late 70s, intensified greatly during the 80s, floated/plateaued in the 90s, and crashed rapidly back down to Earth in the 00s. We can see that now. And our analysis of the past can now take the repercussions of deregulation, etc into account when we make forecasts for the future.

The scientific method says only this: we can never know everything, no conclusion is final. We only have a working hypothesis.

The scientific method is inherently forward-looking but since data cannot arrive from the future, we can only get data from the past and present. The only thing that will work around the problems of deductive/inductive logic under discussion here is a time machine. Nobody can predict the future, all models are just models.

Posted by Nullcorp | Report as abusive

Majkmushrm (love the name, but maybe the majk got in the way while reading these articles),

I fear you’re talking past Ol’Roger — go easy on business tarts. Worse, you may be missing his point — and its a worthwhile point (although an unoriginal one).

Roger understands the scientific method. Perhaps he didn’t make that clear enough in Part 2 of this article, but it’s beside the point. He wants to be saying that the scientific method, precisely the one you attempt to clarify, is applicable in hard natural sciences. Physics – in a nutshell. The scientific method works in physics, and it belongs in physics, and it works in other natural science investigations too, but that’s about it.

Things get absolutely hilarious (for thinking people) when you get university departments that use “the scientific method” to attempt to answer questions and prove hypotheses that concern human beings; and their thoughts, and their decisions, and their actions, and their future actions no less! The scientific method does not apply here, and the problem is that IT IS STILL REGULARLY APPLIED HERE. This happens everyday in political “science,” and education, and psychology, and sociology. But the single worst offender remains economics. In short, the “professors” that due this (misapply the scientific method) are fools, but no one recognizes them as that, and this is a problem.

Most economists are fools simply because most try to act like physicists when they simply are not. They have condemned themselves to study things effected by and created by human beings. They do not have the luxury of studying electrons. For economists this is a problem: they want to apply the scientific method in their discipline (maybe they think this makes them big boys) even though such a method requires objects that behave just like electrons in order to work. If you want to test a hypothesis with the scientific method, or better yet, if you want to make predictions using “the scientific method,” you better be damn sure you are dealing with objects that are INVARIANT over time, and that are RESILIENT TO CHANGE. Otherwise you’ll find yourself testing perpetually wrong hypotheses and making consistently wrong (laughably wrong) predictions — just like economists. To throw a wrench in the scientific method, all you need is some uncertainty, some variance, a little bit of change. (Note here that humans and their emotions and their lives are full of change. Why would a thinking person rely on methods and theories that, BY DESIGN, require us humans to behave like electrons?)

I think Roger is trying to say economics needs new methods. Maybe he should of just said “economists and the people that listen to them would benefit a great deal from thinking.” But since he is the dean of a tart factory, such clear and honest words risk crushing fellow tarts.

Posted by MisterMarket | Report as abusive

You state “As the power of the scientific method has encroached further than its applicability warrants into fields such as economics and business, its predictions of the future become ever more erroneous. In this list, we can include virtually every economic prognostication from the first half of 2008, and countless market research studies that misjudge consumer interest in the new product concepts that they test.”

I submit that instead of “economics and business” you substitute “religious belief” and it will answer your quandary as to why the scientific method has seemingly failed.

It failed now, just as it always has in the past in the face of intractable religious dogma, especially when it began to present answers that refuted the basic tenets of that religion.

Capitalism is in many ways similar to a religion, and those who practice it do not want answers that tend to challenge their cherished beliefs.

Yes, it is really as simple as that.

There is nothing wrong with scientific method, but there is certainly something wrong with Capitalism.

Posted by Gordon2352 | Report as abusive

Hi Roger, Albert Einstein came to a similar conclusion when he wrote an article on this topic in the ‘Monthly Review’, 1949.

Here’s the link, enjoy: socialism

(extract) “It might appear that there are no essential methodological differences between astronomy and economics … But in reality such methodological differences do exist.”

Posted by scythe | Report as abusive

Well, MisterMarket, I agree with you (I’m a physicist) but since I also have an MBA, I think some aspects of economics are amenable to a scientific approach but you first have to understand it to be able to effectively apply it. I question whether or not Mr. Martin does. The social “sciences” and economics typically fail for a couple of reasons. One, they haven’t identified all the factors that affect a given event and, two sometimes those factors are not constrained to follow a verifiable set of rules (like human behavior). The presence of the first factor will simply cause you to fail. The presence of the second factor will deny you the use of the scientific method and force you to use statistical methods. But I’m not the first to say that. A gentleman by the name of Asimov comma Isaac in his “Foundation” series talked about that long before me.

Posted by majkmushrm | Report as abusive

Nice set of articles!

But generally, don’t we have to wait till the current model fails before we embark on the creation of a new model?

Isn’t that, in essence, how all the Natural Sciences operate?

That is, even the sciences that deal with things that “cannot be other than what they are” normally push their models (i.e. theories) to their breaking point before there’s any attempt to frame replacement theories.

For example, Relativity was conceived when Classical Physics was found to be unable to explain the results of the Michelson-Morley Experiment.

Posted by jrpardinas | Report as abusive

Economics as a science? That is laughable. If it were there’d be more of a consensus on fundamental theories like Keynes or Hayek.

Alan Greenspan is the poster child for failed economics. For nearly 10 years the housing bubble grew from Alan Greenspan’s proclamation that the derivatives markets had diversified risk, and the statistical evidence of a very recent look to the past. Narrow minded greed took down the protections provided by the Glass Steagal act. This wasn’t the scientific method, this was politics based on the weak assumption that the moral hazard was a deterrent.

I also laugh at the models that predict stock prices on the historical bumps in stock price curve as if price was more a reflection of symmetry and price psychology than events in reality like profits or deals.

Scenario planning is the best method for dealing with the futurer. Risks, costs, and benefits of multiple possible futures must be assessed with weighted values agreed upon by a team of experts from diverse disciplines. Then contingency plans must be devised. But that’s not all, constant monitoring with updates to the risks, costs, benefits, and weighted values are needed to keep the assumptions from going stale. This isn’t exactly science, it’s a practice like due diligence. Its an art form. It’s also a huge undertaking, so it is not done except in M&A and planning for large corporations. So, instead we make educated guesses and gamble on the future based on the two to three standard deviations of what has happened in the recent past. As long as it works until the next quarter, we’re golden. When it doesn’t we dust off our pants and move on.

The operative word here is gamble. We just apply mathematics to give it the appearance of science.

Posted by LEEDAP | Report as abusive

This article is a joke.
The big banks not only have control over the money supply, they also have access to just about every piece of financial data on the planet.
They have super computers crunching the numbers and placing trades in all kinds of markets.
Just because the people in the know don’t share the information with you doesn’t mean they don’t have it.
Did Goldman Sachs and JP Morgan get caught unawares during the financial crisis? Hardly.

Posted by RandomName2nd | Report as abusive

Wow, you guys are so smart. Here’s one thing I know about economics. As long as the current administration continues with their Keynesian tinkering, my 401K will languish in the toilet. And I don’t need any economic models or the scientific hypothesis to prove it. Any one of you smart guys want to bet.

Posted by RReyes | Report as abusive

[…] The limits of the scientific method in economics and the world […]

Posted by News: Oscars come to London to honor Vanessa Redgrave | News 25/7! Delivering news in real time | Report as abusive

[…] Roger Martin’s “The limits of the scientific method in economics and the world” (part one and part two), recently published on […]

Posted by Should economists be “imagineers” of our future? | The Great Debate | Report as abusive

The limits of the scientific method are only limited by the person.

From what I’ve read from this article the author seems to argue that predicting the “future” of markets is impossible with logic.

Science encompasses logic in that the investigator will try to prove his/her hypothesis through experimentation of variables by using logic. Such as looking at why does water boil at a lower temperature at higher elevation? Logic would suggest that you try to boil water at varied elevations and take a temperature. The variables are few and easily controlled.

When we come to something complex like “economics” we are looking at a myriad of variables, but the majority of these variables seem to be human. The human condition has been the one thing we have been trying to figure out since the dawn of our existence and we still cannot figure it out.

I do believe in the scientific method and maybe it can work for economics, but the investigator needs to account for all of the variables. In my mind the reading Keynes and Hayek would be as important as reading the writings of Socrates, Faulkner, and J.K. Rowling for an economist.

Posted by Maiorana | Report as abusive

[…] thought that there would be a response to this article by Roger Martin (see the original posting here). Mark Thoma’s excellent response is here, makes a required […]

Posted by The economists fight back! « Proeconomia | Report as abusive

Physics doesn’t apply to the demand side because economists can ignore the depreciation of automobiles owned by consumers but the depreciation of identical cars owned by Hertz and Avis do depreciate.

So in economics the Laws of Physics are affected by semantics. So the depreciation of hundreds of millions of cars around the world do not affect the Net Domestic Product of any country. ml

Posted by psikeyhackr | Report as abusive

[…] 11/11/11/the-limits-of-the-scientific-me thod-in-economics-an… […]

Posted by Interesting Reuters opinion: ‘the limits of the scientific method in economics and the world’ | i-write-melons-in-all-sizes-and-forms | Report as abusive

Yeah, ignoring the annual depreciation of all of the cars purchased by consumers in every country for the last 50 years. Running the planet on defective algebra is so scientific.

Consumer cars get added to GDP when purchased but their depreciation is not subtracted to computer NET Domestic Product. But then economists don’t mention NDP much.

Posted by psikeyhackr | Report as abusive