No Such Thing as a Free Lunch: Principles of Economics (Part 173)

Trickle-down economics - it didn't work. The whole idea was supply-side economics: give rich people a lot of money; they'll spend it, it'll go into the economy. Here's what we found out - rich people, really good at keeping all the money. That's how they got rich. If you want it in the economy, give it to the poor people. You know what they're really good at? Spending all their money.

Greg Fitzsimmons

Debates in Macroeconomics (Part H)

by

Charles Lamson

Testing Alternative Macroeconomic Models

You may wonder why there is so much disagreement in macroeconomics. Why cannot macroeconomists test their models against one another and see which performs best?

One problem is that macroeconomic models differ in ways that are hard to standardize. If one model takes the price level to be given, or not explained within the model, and another one does not, the model with the given price level may do better in, for instance, predicting output not because it is a better model but simply because the errors in predicting prices have not been allowed to affect the predictions of output. The model that takes prices as given has a head start, so to speak.

Another problem arises in the testing of the rational-expectations assumption. Remember that, if people have rational expectations, they are using the true model to form their expectations. Therefore, to test the assumption we need the true model. There is no way to be sure that whatever model is taken to be the true model is in fact the true one. Any test of the rational-expectations hypothesis is therefore a joint test (1) that expectations are formed rationally, and (2) that the model being used is the true one. If the test rejects the hypothesis, it may be that the model is wrong instead of expectations not being rational.

Another problem for macroeconomists is the small amount of data available. Most empirical work uses data beginning about 1950, which in 2021 is about 72 years (288 quarters) worth of data. While this may seem like a lot of data, it is not. Macroeconomic data are fairly "smooth," which means a typical variable does not vary much from quarter to quarter or year to year.

To give an example of the problem of a small number of observations, consider trying to test the hypothesis that import prices affect domestic prices. Import prices changed very little in the 1950s and 1960s. Therefore, it would have been very difficult at the end of the 1960s to estimate the effect of import prices on domestic prices. The variation in important prices was not great enough to show any effects. We cannot demonstrate that changes in import prices help explain changes in domestic prices if import prices do not change. The situation was different by the end of the 1970s, because by then import prices had varied considerably. By the end of the 1970s, there were good estimates of the import price effect, but not before. This kind of problem is encountered again and again in empirical macroeconomics. In many cases there are not enough observations for much to be said, hence considerable room for disagreement.

It is difficult in economics to perform controlled experiments. Economists are for the most part at the mercy of the historical data. If we were able to perform experiments, we could probably learn more about the economy in a shorter time. Alas, we must wait. In time, the current range of disagreements in macroeconomics should be considerably narrowed. 

*CASE & FAIR, 2004, PRINCIPLES OF ECONOMICS, 7TH ED., PP. 660-661*

end