Is Economics Science?

The most recent issue of Science magazine (from May 23, 2014) focuses on economic inequality.  Most likely this is a response to Thomas Piketty's book Capital in the 21st Century, which purports to explain economic inequality in terms of return on capital investment relative to labor.  The question we will address here is whether or not this subject belongs in a publication titled Science.

For something to be scientific, it must predictively explain phenomena, hypotheses must be experimentally testable, and one must be willing to discard hypotheses that clearly do not match or predict experimental data.  Hypotheses and explanations for experimental data that continue to predict future data are considered to represent reality.  In some disciplines (notably physics) the explanations are written as mathematical expressions, which are often extrapolated into domains not yet probed experimentally and yield predictions that can be further tested. As a note, the quality of a given description is determined by its accuracy relative to measurement accuracy and certain descriptions (known as "theories" in the physics community) can only be considered accurate within a given experimental space (e.g. low velocity, small mass, long time scale). 

At some level, modern economics is attempting to be like physics - economists attempt to write down rigorous mathematical formulas that describe the flow of money, ideas, commodities, etc.  The mathematical predictions can then be compared to what has actually happened to test for accuracy.  Experiments can be devised to test these "theories"in small scales, with limited resources and individuals, and such results are often extrapolated to populations at large.  This is fine, as science goes, and economics can be considered science under these circumstances.

The problem arises when economists treat economics as a deductive philosophy: given certain assumptions (sometimes mathematically rigorous, sometimes not), one can use deductive logic to predict what will happen in the future.  If the assumptions are correct, and the mathematics accurate enough, the future can be predicted.  Much of modern economic commentary resides in this treatment.  When we read about, for example, how changing the minimum wage will affect the labor market, the conclusion is often based on deductive logic given a premise and a few assumptions.  The problem is often that the assumptions are not based on any experimental evidence but are essentially made-up axioms (e.g. the "people act rationally" assumption), used to make predicting easier or to appeal to some basal desire for humanity to be predictable. 

We can see the effects of this type of thinking in modern day economic analyses.  One would hope that economists would understand the weaknesses of their assumptions and mathematical descriptions (most physicists certainly do), but this often does not seem to be the case.  This brings us back to Piketty's book, which relies on actual long-term macroeconomic data to make statements about the past, and possibly predict the future.  This is scientific - developing predictions and hypotheses which can be tested (though the test will be how the actual economy behaves over the upcoming years).  We do not know if the predictions will be accurate (given that most previous predictions of large-scale economic behavior, for example those of Malthus, Karl Marx, and others have failed, I am not optimistic for Piketty), but we will see.  Somewhat problematically (but interestingly), we cannot devise controlled macroeconomic experiments - society in the past, right now, and into the future is one ongoing experiment with parameters continually changing.  Perhaps one day we can write down a good description (even if only statistical predictions are possible).