must we give up understanding to secure knowledge in economics?
By Alex Rosenberg and Tyler Curtain.
In physics our knowledge exceeds our understanding. In economics the reverse is true. Seeing why helps us make sense of both of these disciplines.
In physics we’ve reached the point where we know the nature of things to twelve decimal places. We reached that point by starting with common sense and correcting its predictions until we arrived at quantum mechanics—a theory that we literally can’t understand, despite the fact that we know it to be as close to the truth as any theory we have in any science. It took about 350 years—from Newtonian gravity to Schrödinger’s cat—to get to the point where knowledge exceeds understanding in physics.
Economics is harder than physics. It must be. It’s not much younger a science, having gotten its start with Adam Smith in 1776, a good 83 years before Darwin was able to put biology on a scientific footing. If economics were as easy as physics, it would have made more progress by now.
As far back as John Stuart Mill philosophers of science have been trying to figure out exactly why economics is harder than physics. They have given a variety of answers.
We think a large part of the reason is that unlike the physicists, the economists have been unwilling or unable to let go of the notion that their understanding of economic affairs counts as knowledge about economic behavior. Like physics, economics starts with common sense—in this case firm convictions about how we are driven to make choices by our desires and our beliefs, which the economists label preferences and expectations.
Giving up firmly held convictions isn’t just a problem for economics. Physics had the same problems: humans have difficulty relinquishing the conviction that motion requires force, that there is a preferred direction in space, that every event has a cause. But progress—as measured by prediction—required relinquishing our sense that we already know how things work. In the social sciences, it’s been almost impossible to give up trying to explain things by making sense of them in the form of stories we understand.
Economics is the social science that has most relentlessly tried to be scientific about turning understanding, which gives meanings to events, into knowledge. Instead of moving beyond understanding in the way physics or biology has, a significant part of the discipline has sought to formalize, mathematize, generalize the sense we so easily make out of economic exchange.
Economists point to this approach, developing mathematical models of the ways we make sense of our own economic decisions, from the inside so to speak, when they insist that their discipline is a science. Their models are designed to make sense of economic events as the outcomes—intended and unintended—of many individual human choices, each of which we can understand—make sense of—by putting ourselves in the shoes of one or a million consumers or producers or bankers or brokers or treasury officials. It’s tempting to treat anything that can be made sense of by a convincing enough story as explained. Convincing stories enable economists to feel vindicated in their claim that economics explains. It makes sense of events—that’s what stories do—and it does so by using a general theory. Which theory? That depends on the economic school one subscribes to. Every school tells a different story. The mathematics, the modeling is supposed to make the stories into science.
Where should prediction fit into this picture? If economic theorists converged over time on increasingly accurate predictions, we’d be confident that economic explanations are on the right track. If some economists were regularly more reliable in prediction than others, we’d know whose explanations to credit, to improve, to extend. If some economists were right sometimes, wrong other times, and no one was regularly right, we’d conclude that when economists get their predictions right—as they sometimes do—they are getting things right by accident, without really knowing why.
Can we have confidence that economics is on the right track, as a science, when it looks like no economic theory shows much of a sustained pattern of predictive success? Many economists will say, Yes! They’ll cite other disciplines, often biology, claiming that it lacks a track record of convergence on successful predictions. Biologists will challenge the notion that biology only explains and doesn’t predict. Just consider advances in clinical medicine, public health, molecular genomics and agriculture.
Economists insist that their explanations constitute advances in knowledge even without predictive improvement: 80 plus years after the Great Depression we have good explanations or at least better ones for why it happened. Trouble is economists can’t agree on which explanation that is.
We agree with the economist that prediction isn’t the sole aim of science, that thick descriptions and even after-the-fact explanation are also important. But the crucial question here is how to tell whether we’ve got a good explanation, or even just a better one that we had before. This is where a requirement of at least some increasing predictive success has a role to play. A new or improved theory should have some direct or indirect testable consequences that will provide evidence that its explanations are built on more solid ground than its competitors and predecessors. This is a very modest requirement of predictive success.
Why has economics not been able to meet this modest requirement?
Unlike the physical processes, the domain of economics includes a wide range of social “constructions”: institutions—markets, rules—about money supply growth, objects—currency, shares. Even when idealized these factors don’t behave uniformly. They are made up of unrecognized but artificial conventions that people persistently change and even destroy in ways that no social scientist can really anticipate. We can exploit gravity, but we can’t change it or destroy it. No one can say the same for the socially constructed causes and effects of our choices that economics deals with.
Another factor in its domain that economics has never been able to tame is science and technology itself. These are the drivers of economic growth, the “creative destruction” of capitalism. But no one can predict the direction of scientific discovery and its technological application. That was Karl Popper’s key insight. Philosophers and historians of science like Thomas Kuhn have helped us see why scientific paradigm shifts seem to come almost out of nowhere. As the rate of acceleration of scientific revolutions and novel technologies increases, the prospects of an economic theory that tames—endogenizes—the economy’s most powerful forces must diminish further and further. And with it, any hope of improvements in prediction declines as well.
Besides these there are the difficulties that can be traced to economics’ unshakable commitment to understanding—making sense of things, even when that stands in the way of knowledge.
Like all other social scientists, economists can’t really give up trying to show how human affairs emerge from motives and meanings. But if beliefs and desires—preferences and expectations—just change too fast for any one to track them, no predictively successful science that also makes sense of things is on the cards. A prosaic example shows why. Recall the last time you reached into a bag of French fries for the ketchup packet and found only mustard, not something you wanted at all. Now think about how much you’d pay for the same packet of mustard if you find yourself without one at a baseball game with a hotdog in your hand. If circumstances can change our valuations that completely, then preferences and expectations won’t hold still long enough for a theory using them to make predictions about our economic decisions. Economist’s explanations will be able to make sense of them, after the fact, when it’s too late to predict them, of course.
What if markets, industries, whole economies are moved by something other than the summing up of vast numbers of individual choices? This approach–trying to mine “Big Data” for trends that don’t make any sense to participants and observers, even while they make money—is what has made physicists so attractive in the backrooms of so many fund managements. If this approach pans out economics will turn out to be more like natural science, where we secure knowledge by giving up the aim of making sense of things—understanding.
Most economists, we dare say, don’t give data mining much of a chance. They will confidently predict that even if the physicists’ discoveries could be used to make profits, the strategy won’t last long. Why? Well, other people will catch on and begin using the same trick, or exploit it with some new strategy. It wouldn’t make sense for them not to. What economists can’t predict is what we want to know: how long will the new trick work, where and when will it cease to work, what strategy will exploit it?
Economics can’t aspire to much predictive power, and to certifiable scientific advance, if it has to build its models on an approach that makes sense of economic behavior. Maybe it won’t always have to. In The Wealth of Nations Adam Smith wrote “Nobody ever saw a dog make a fair and deliberate exchange of one bone for another….” Give it 350 years. Someday economics may know more than it understands. Then, along with quarrels about Schrödinger’s cat, we’ll have debates about the economic bite of Adam Smith’s dogs.
ABOUT THE AUTHORS
Alex Rosenberg is the R. Taylor Cole Professor of Philosophy and chair of the philosophy department at Duke University. He is the author of Economics — Mathematical Politics or Science of Diminishing Returns, most recently, The Atheist’s Guide to Reality.
Tyler Curtain is a theorist and philosopher with the Department of English and Comparative Literature. He teaches graduate and undergraduate courses in theory, as well as courses in science fiction and fantasy. He co-direct Duke’s Center for the Philosophy of Biology. His research interests include theoretical philosophy, philosophy of biology, evolutionary theories of language, linguistics, philosophy of language, and theoretical computer science.
Rosenberg and Curtain recently asked the question ‘What is Economics Good For?’ in an opinion piece at the New York Times.