In 1874, Max Plank’s physics professor explained to him that “in this field, almost everything is already discovered, and all that remains is to fill a few holes.” I heard the same thing in 1980 from a Nobel Laureate in Economics — almost verbatim. Didn’t he know? But how could he? Optimistically, I concluded that Economics was at a similar stage to physics in 1874.
My transition from physics to economics was baffling. Fellow graduate students explained that economics was unscientific because economists could not do experiments. Did they think Astronomers were busy experimenting on stars? (Blinder uses this excuse, NYbooks.com 12/18/2014.) And besides, economists could do experiments. Of course, parts of economics were scientific even back then. But times have changed and economics is now making steady progress towards becoming scientific thanks to behavioral economics and people like Esther Duflo. I hope this site can make a small contribution to this progress.
Here I will discuss problems caused by the unscientific nature of economics in three otherwise unrelated areas: the design of markets for electricity-generation capacity, international negotiations on climate policy, and the foundations of neoclassical economics.
What is science?
Several popular misconceptions about science have long confused the discussion of whether economics can be a science. The two most common misconceptions are:
- All scientific conclusions must be validated by experiments.
- Scientific theories must yield numerically precise predictions.
A moment’s reflection on actual science reveals these to be poppycock. First, there may not be a single theory in astronomy that has ever been checked experimentally. For example, the theory of planetary formation was not tested by forming planets in the laboratory.
Similarly, the theory of evolution does note allow us to predict, even roughly, what new species will evolve in any particular ecosystem. But that does not make the theory of evolution unscientific.
The heart of the scientific method is actually quite simple:
- Form a hypothesis about what causes something to happen in the real world.
- Check that hypothesis by observing the real world.
This is completely different from mathematics:
- Make any set of logically consistent assumptions you like.
- Use them to prove something you find pleasing.
Neoclassical economics seems to mistake math for science, apparently due to physics envy and the fact the physics uses hard math. The difference is physicists just use math as a tool to do science. They don’t mistake math for science.
For a scientific hypothesis to make sense, the hypothesis must be “falsifiable,” that is, it must be possible, at least in principle, to prove that the hypothesis is false.
The standard way to check a hypothesis is to ask what it predicts, and then check the prediction. For example, if we were to hypothesize that most businesses behave as perfect competitors, and we wanted to check this scientifically, we would first ask what predictions the theory of perfect competition makes about the behavior or nature of businesses. One obvious predictions is that businesses will operate at their efficient size (or, if we add a little realism, mistakes will cause half to be over and half under).
To check this prediction, we could ask businesses if they were trying to increase their volume of business, decrease it, or keep it the same. If most are found to be trying to expand their business, we would reject the hypothesis that most businesses behave as perfect competitors.
Later, I will return later to discuss the role of experiments in economics.