Posts Tagged ‘prediction’



You know the story: Xi and his San tribe are “living well off the land.” They are happy because of their belief that the gods have provided plenty of everything, and no one among them has any wants. One day, a Coca-Cola bottle is thrown out of an airplane and falls to Earth unbroken. But the bottle eventually causes unhappiness within the tribe, leading the elders to believe it’s an “evil thing” which the gods were “absent-minded” to send them. Xi then travels to  the edge of the world and throws the bottle off the cliff. He then returns to his tribe and receives a warm welcome from his family.

I wonder if Paul Krugman expects to receive a warm welcome from the economics family after throwing the prediction bottle over the cliff.

Hardly anyone predicted the 2008 crisis, but that in itself is arguably excusable in a complicated world. More damning was the widespread conviction among economists that such a crisis couldn’t happen. Underlying this complacency was the dominance of an idealized vision of capitalism, in which individuals are always rational and markets always function perfectly.

I actually agree with Krugman on this point. Economic prediction is, in fact, impossible and the really crazy feature of mainstream economic models is the fact that endogenous crises simply can’t occur. Exogenous factors, sure, but nothing internal to the models can lead to a crash. Their idealized vision of capitalism, absent an external event (such as a credit crunch or an increase in the price of oil), simply leads to a full-employment, price-stable equilibrium.

But, wait, doesn’t the entire edifice fall when—on its own terms—the ability to correct predict is dispensed with? The whole rationale of giving up realistic assumptions about the economic system has been the ability to accurately and correctly predict the movements of the economy. That’s the mantle of predictive science that has been used, since at least the mid-1950s, to expunge all other economic theories and approaches from the discipline.

Mainstream economists can’t have it both ways: to celebrate their models for their predictive ability and then to dispense with prediction when, as in 2007-08 (just as in 1929), their models clearly failed. We need something better.

As for their track record since the crisis broke out, well, they haven’t fared much better—at least to judge by where we stand right now. Krugman, for his part, wants to stick with the hydraulic mechanisms of the textbook economic models, which “did a pretty good job of predicting how things would play out in the aftermath,” and declare that “too many influential” economists must be crazy.


The economists at Goldman Sachs have now come in with their predictions for the 2014 World Cup finals. And, by a wide margin, Brazil are the favorites to win (3-1 over Argentina in the final match).

The problem, of course, is that football is a low-scoring game and, therefore, quite unpredictable. Thus, as the Goldman Sachs team admits, when looking at how their model would have done in predicting the goal difference in each game of the 2010 World Cup finals,

Overall, there is a positive and statistically significant relationship between the actual and predicted outcomes. However, the fit of the relationship is not particularly tight with an r- squared of 0.24, because football is ultimately a pretty random game.

As for me, with only the most informal of statistical analyses (in my head, based on what I know of the various national teams and the history of World Cups), I actually agree with the prediction: a final four of Germany, Spain, Brazil, and Argentina, and Brazil raising the trophy.

Then again, anything can happen. . .


This semester, I’m teaching a course on Marxian economic theory. It’s been a real eye-opener for the the students, who seem a bit surprised to learn that there is such a wholesale critique of the mainstream economics they’ve been learning. Some are even intrigued by this new way of thinking about the economy, which led one of them to pose the following question: did Marxists predict the crisis better or more accurately than mainstream economists?

Well, I explained, that’s setting the bar pretty low, since mainstream economists simply failed to predict the crash of 2007-08. But, I explained, Marxists did no better. And that’s because economic forecasting is like selling snake oil: lots of folks earn lots of money promising the ability to predict economic events but all they’re doing is selling the promise, not the actual ability, to get the forecasts right. (And, of course, they pay nothing for their failures, since they’ve left town long before people discover the magic elixir doesn’t work.)

And that’s what has happened to the students: they’ve been told mainstream economics is superior to all other approaches, that it’s a “real science,” because of its predictive power. And they’re willing to jump ship, as it were, if an alternative theory offers more predictive power.

The problem is, as Sir David Hendry explains, forecasting only works if the future behaves the same as the past, if it follows the same rules and falls under the same normal distribution. If it doesn’t, then all bets are off. What that means for me (and for Chris Dillow) is that Marxists are no better at predicting the future than mainstream economists. In fact, economic forecasting, of whatever sort, is a false promise.

But then I went on in my response to the student’s question: what really distinguishes different groups of economists is whether or not they include the possibility of a crisis in their theories and models—and what they would suggest doing once such a crisis occurred (including measures to prevent future crises). And there the difference between mainstream and Marxian economics couldn’t be starker: mainstream economics simply doesn’t include the possibility of crises (except as an exogenous event) whereas Marxists start from the proposition that instability is inherent (and therefore an endogenous tendency) in an economy based on the capitalist mode of production. That’s one fundamental difference between them. The other is that, once a crisis occurs (such as in 2007-08), the two groups of economists offer very different solutions: whereas mainstream economists spend their time debating whether or not any kind of intervention is warranted (based on neoclassical versus Keynesian assumptions concerning invisible and visible hands), Marxist economists presume that interventions are always-already being made (in terms of determining who pays the costs of the crisis) and that it’s better both to help those who are most vulnerable and to put in place the kinds of institutional changes that would prevent future crises.

So, no, I don’t put a lot of stock in economic forecasting, whether promised by mainstream economists or others. It’s a promise of control that is a lot like selling snake oil. But I’m willing to throw in my lot with an approach that, first, actually includes the possibility of such crises at the very center of the theory and, second, is willing to move outside the paradigm of private property and markets to help those who are hurt by the crisis and to change the rules so that those who created the crisis in the first place no longer have the incentive and means to do it again in the future.

And you don’t need a crystal ball to know that, if such changes are not made, another crisis is awaiting us just around the corner.


Here’s the graph Bruce is referring to in the comments on this post:


And here’s the same series going back earlier:



Really, Noah Smith? The best you can do is go after Steve Keen for failing to successfully predict when, where, and how the crash of 2007-08 would break out?

Now, maybe Keen deserves a bit of stick for loudly proclaiming more “loudly and confidently than just about anyone else on the planet” that he predicted the global financial crisis. Perhaps that’s a bit brash.

But mainstream economists are the ones who dominate economic discourse. And they’re the ones who claim the scientificity of their approach to economic analysis is based not on the realism of their assumptions but on the predictive power of their models. And, finally, they’re the ones who, with few exceptions (like Nouriel Roubini, aka Dr. Doom), failed to predict the more recent crisis.

At least Keen and other heterodox economists use theories that contain the possibility of crises occurring based on the endogenous tendencies of capitalist development (I know, and we’ve successfully predicted 15 of the last 5 crises). Mainstream economists don’t even admit of that possibility, although Smith has shown that at least a few of them have been able to successfully recalibrate one of their models (by adding financial frictions) and then to have successfully predicted the crisis—AFTER THE FACT.

Well, that simply doesn’t cut it. Either admit that mainstream economics is a failure because it didn’t successfully predict the crisis or give up on the idea that predictive power is one of the key criteria of economics, which has served as an excuse for attempting to demonstrate that what mainstream economists are doing is science and what the rest of us are doing is non-science. You just can’t have it both ways.

And beating up on Steve Keen is simply the coward’s way out.

Readers of this blog know that I have sung the praises of Nate Silver’s statistical work for quite some time, long before the right-wing attacks on his electoral predictions started.

My appreciation stems not from Silver’s predictions per se but, rather, from his appreciation of uncertainty, as this review by Samuel Popkin makes clear.

One of the biggest problems we have in separating signal from noise is that when we look too hard for certainty that isn’t there, we often end up attracted to noise, either because it is more prominent or because it confirms what we would like to believe. This is a worse problem in politics than in baseball or poker. If most polls are reporting a tight race, an outlier showing a bigger gap will be the poll that makes news, thus getting more of our attention. Partisan TV pundits try to assuage the worries of the faithful on their side instead of making accurate predictions. When Silver analyzed 1,000 predictions on The McClaughlin Group, he found them no more accurate than flipping a coin. . .

In his analysis of fascinating examples ranging across all the areas in which we try to predict future outcomes, Silver stresses the gap between what we claim to know and what we actually know. In 1997, the National Weather Service predicted that heavy winter snows would cause North Dakota’s Red River to flood over its banks in two months, cresting at 49 feet. The residents of Grand Forks were relieved, since their levees were designed to withstand a 51-foot crest. If the Weather Service had mentioned that the margin of error for its forecast was five feet, the three feet of water that poured over the levels in an eventual 54-foot crest might not have destroyed 75 percent of the town. Happily, the Weather Service now provides that information—an example of an easy reform to forecasting.

Special mention