Ice cubes, cornflakes, inflation and what caused the sub-prime lending crisis: Why theories are so hard to get right

Podcast: Download

By Navin Kumar
Article ID: 1331

The Phillips Curve is possibly the biggest blow-up in economics that ever happened. Economists – and just about every class of social scientists – are frequently (and rightly) accused of being so infatuated with a theory, that they ignore data if it doesn’t fit in with their model.

The Phillips Curve is a wonderful example of rigorous empiricism. In 1958, A.W. Phillips discovered a striking relationship between inflation and unemployment: periods of high inflation coincided with periods of low employment. Subsequent studies found this result held true across countries and time periods. This led to the belief that there was a “trade-off” between employment and inflation which could be exploited by policy makers: a government could reduce unemployment if it was willing to increase inflation and vice-versa.

The theory behind the data was fairly simple: if unemployment was low, businessses found it hard to hire workers and to increase wages. Higher wages causes goods to be more expensive to produce, so firms increase prices, and this causes inflation. Conversely, if the government caused inflation, there would be a gap in which wages are low in comparison with the price of goods. This is because the workers have not yet negotiated higher wages to compensate for higher prices. During this period, businesses take advantage of low wages by hiring more workers and stepping up production, reducing unemployment.

Higher inflation means increased production and lower unemployment. The empirical evidence and theory were flawless. Policies based on the Phillips Curve enjoyed some initial success. So why is it that in the 1970s the Phillips Curve collapsed, and the world saw “stagflation”: a bizarre situation which combined reduced production with inflation?

An answer came from American economist Milton Friedman: low unemployment, he explained, is the result of unanticipated inflation. If the government started intentionally causing inflation, inflation becomes regular and predictable. If it is predictable, employees started negotiating contracts in which wages increased in tune with inflation and there is no point at which wages are “cheap”.

This is an incredibly subtle difference. Note that the explanation hasn’t changed: the relationship is still the result of wages not keeping up with prices. The only difference is that the second uses a more sophisticated system than the first.

To illustrate what happened, consider the following example (used by Landsburg in his excellent book The Armchair Economist):

“Imagine an economist noticed that people purchased two boxes of cereal per week. Excited, he publishes a paper on the subject and it comes to the governments’ attention. The government – for whatever reason – decides that people should eat four boxes of cereal per week. People already buy two boxes, so if the government sends them [an additional] two boxes every week – yay! – they will eat four boxes a week!”

But that’s not how it plays in real life. After getting the two boxes, this won’t suddenly change consumer habits to four boxes – they’ll instead stick with two. Since the government just gave them two, they’ll buy no boxes at all! These are the perils of not including  people’s behavior when formulating an idea. But it’s even worse when we get a theory wrong. Imagine two economists bumped into the “two boxes a week” fact. One says, “People will always buy two boxes of cereal per week,” while the other says, “People will always eat two boxes of cereal per week.”

How can you tell which one of them is right? From the data, you can’t.  But once you change the rules of the game – by sending them boxes – you can. The first theory implies they will continue buying two boxes per week while the second predicts they will buy none at all. The second will be proven right.

Similarly, in 1958, if you had two theories which stated, “Inflation reduces unemployment,” and “Unanticipated inflation reduces unemployment,” it would be impossible to tell which is one of the two is right because, until now, all inflation has been unanticipated. But in 1980, we suddenly have a lot of data on anticipated inflation. We can now check which of the two theories is right. As it turned out, it was the second: unanticipated inflation reduces unemployment.

Fast forward to 2009. What caused the sub-prime lending crisis? Here are two popular theories:

1) The bankers were greedy. They knew the risks, but went on anyway because of the possibility of large profits that came with sub-prime loans.

2) The bankers were stupid. Despite all the PhD’s and mathematicians on their staff, they didn’t foresee the nationwide fall in house prices.

The first theory implies that banks should be more tightly regulated to prevent greedy bankers from taking risks that will hurt everyone. The second – which requires a bit of explaining – states that the banks created financial products (which are blamed for bringing down the system) in order to reduce the risk. And they did: they reduced the risk that came with a few defaulters (which allowed for cheaper loans) but increased the pain that would occur if a huge event happened – like the fall in housing prices.

However, they calculated that such a huge event was unlikely to occur, so the risk was low. Any government regulator would’ve used the same methods to calculate the risk. They would, in other words, be as dumb as the banks themselves. They would do no good and might end up restricting genuinely good products. So we have two theories, both of which explain the crisis, but both make very different recommendations.

Which of them is right? Your guess is as good as mine. (Incidentally, both these theories agree that banks are now too large and should be trimmed so that a bank failure won’t cause the financial system to collapse with it.

Look at a star-shaped piece of ice sitting in the sun. We know what kind of puddle it will make. But we can’t later look at a puddle and figure out if the ice was originally star-shaped or heart-shaped or a cube.

These examples illustrate how difficult it is to figure out why something happened – and why it’s so important to know for sure. As human beings we must make decisions, but decisions which have a faulty foundation are doomed to fail.

There is nothing as useless as a fact without a theory. But theories themselves are incredibly hard to evaluate. One good measure of a theory is falsifiability. This basically asks the question: how can this theory be proved wrong?

The theory of gravity can be proved wrong if apples started falling up. Take Marx’s theory that “Industrialization alienates workers, and alienated workers will revolt against capitalism.” This can be proved wrong if workers in industrialized countries revolt less often than those in non-industrialized countries. If any theory is presented to you without a test, or is inherently untestable, be suspicious.

The moral of the story is not that we should reject all theories, but that we should look at all ideas on the table with an impartial eye. We should watch out for assumptions and differences between the experimental setting and the real world. We should remain politely skeptical of any theory that’s put on the table without a method of falsification. And we must be especially careful when such theories – whether they talk about religion, science, economics or politics – attempt to plot the direction of humanity.

Other articles related to this topic:

• Can safety regulations kill you? How safe are seatbelts and seatbelt laws?
• Modeling population and technology: Why haven’t you starved to death?
• Global warming and climate change: Why they’re so hard to get right
• What’s the real value of a college education?
• Answers to objections about atheism and evolution