In class, I made the argument that economics is not science, because it cannot undertake repeatable experiments. Someone rebutted: this would mean that I am ignore an emerging body of work, some by prominent Nobel Prize winners, that is transforming economics into a field firmly grounded in fact.
How would you justify whether economics is scientific or not?
Welcome to the demarcation problem of science. What is this thing called science? In lower levels of education, one is often given the impression that 'science', whatever that may be, exists as a monolithic entity. There is no sufficiency and necessity definition of what science is. It's better to say 'sciences' or 'scientific' when speculating as to this question. While physics has often been touted as a role-model of the sciences with its focus on measurement and manipulation of variables, in fact, each of what are often seen as fundamental sciences, physics, chemistry, biology, psychology, and sociology, have different methodologies and are roughly lumped into the natural sciences (the first three) and the social sciences (the last two), though newer tools for exploring mind and data have "hardened" the "softer" sciences. Today, one can use fMRIs, for instance, to peer directly into the operation of the brain when trying to diagnose abnormal psychological states.
Wikipedia isn't inaccurate in it's definition of economics:
Economics (/ɛkəˈnɒmɪks, iːkə-/) is the social science that studies the production, distribution, and consumption of goods and services.
So, economics is considered a science, but how is this so? For starters, your premise that variables can't be manipulated is wrong. Behavioral economics, for instance, uses some hard science:
Behavioral models typically integrate insights from psychology, neuroscience and microeconomic theory.
In fact, Daniel Kahneman and Amos Tversky with their prospect theory have helped to reshape our notion of economics and make it even more scientific than ever by killing the idea sometimes known as Homo economicus, the rational economic agent. Behavorial economists often manipulate variables in line with their psychological training.
How then is it a science? Well, there's more to science than manipulating variables. For instance, it's possible to draw causal inferences in nonexperimental research and sophisticated statistical modeling can be used for prediction. Statistical or probabilistic explanation has value too. Operational definitions are used to define abstractions, and theories still have mathematical and logical aspects that can be explored rationally. In general, like the hard sciences, the soft sciences can use deduction, induction, and abduction to eke out certainty from the world around us. While complex phenomena do not reduce as easily as simple physics experiments, nonetheless they are subject and object of scientific discourse.
REFERENCES
Blackwell's Companion to the Philosophy of Science
Blaclock, Hubert M. Causal Inferences in Nonexperimental Research
Kahneman, Daniel. Thinking, Fast and Slow
The Popperian view of science is that a claim is "scientific" if it can be falsified. Science cannot prove that a hypothesis is true, only that it is manifestly false. If economics can make falsifiable claims, then I think it is justified to say such claims are "scientific", at least on some level (the degree of repeatability is certainly a basis for ascribing a non-binary "scientific" label).
Thus, even if one cannot control the variables, it is still possible to "do science", as another answer noted for astronomy (at least until we master astro-engineering ;). An example of a "scientific astronomical claim" is that inter-galactic cosmic rays cannot exceed 5x10^19 eV, known as the "GZK limit". Since cosmic ray detectors can measure the energy and direction of cosmic rays, we only need to plot histograms of energy vs. source to determine whether the claim is falsified or not.
When a claim can be tested merely by accumulating observations under both the experimental and control groups without intervention by experimenters, you have what is called a "natural experiment". In the cosmic ray case, the experimental group are cosmic rays emanating from a known galaxy, while the control group are cosmic rays not emanating from a known galaxy (i.e., coming from a source within the Milky Way). Even though we cannot control, let alone cause, the emission of cosmic rays from any stellar source, we can still "conduct" the natural experiment just by watching the sky for a while.
As it happens, such experiments also occur in economics. As you may know, there is currently debate about the effects of a high minimum wage. It would be unethical and really, unthinkable, to conduct a randomized controlled trial (RCT) in which half of the citizens in a city got the high minimum wage, while the other half got the federal wage. However, it is possible to infer the effects (and even make falsifiable predictions) by observing before-and-after economic effects in a city which did implement a high wage and comparing it to a similar city which did not. Economists from the University of Washington did exactly that. While this study was primarily observational, it would have been just as easy to start with one of the partisan claims, such as: "Average restaurant prices will increase by more than 5%" or "Unemployment levels will remain the same, to within the margin of error." These are clearly as falsifiable as the claim: "No high-energy cosmic rays can reach us from another galaxy."
Another answer observed that economics and politics are intimately entwined. We would like any policy with an economic impact to be informed by the best knowledge we have available. Thus, whenever Congress proposes a new law, the CBO is asked to create forecasts of the costs and impact of the proposed law. That forecast is essentially a falsifiable prediction, based on economic models which serve as the "economic theory" on which Congress debates. Thus, as a democracy, we act as if economics is a "science" whenever we weigh the CBO forecast for the cost of an entitlement program like the ACA or a tax break like the Tax Cuts and Jobs Act.
In some cases, economists act like experimenters when they propose policy with an explicitly economic effect. For instance, consider the "Cash for Clunkers" program proposed by economists and signed into law by Obama. Of course, the economists had no control over who received the rebate and who didn't, so again, RCT is infeasible for pretty much any national economic effect. Apparently, no economists dared to make quantified claims about the results before the program went into effect, so this isn't quite as scientific as other programs.
A much better example is the American Recovery and Reinvestment Act (ARRA, but you can call it the "Obama Keynesian Stimulus Package"). As expected, the CBO provided a forecast for the program cost, as well as impacts on GDP and employment. The very notion of government economic stimulus in the first place is due to Keynes and can be considered a fairly scientific theory, even if it is not as precise as one may be used to from the hard sciences.
Of course, the way results are interpreted is open for debate, and much controversy surrounding the UW minimum wage study revolved around whether they had an appropriate control city for comparison. These are the weaknesses of not having a fully experimenter-defined control group. But, not being able to control every aspect of the universe should not prevent us from attempting to pose and test scientific claims about how those phenomena work. Is economics as scientific as Newtonian mechanics? No, it is not. But science is a continuum, and I think it can make sufficiently detailed falsifiable claims to put it squarely on the "science" side of the continuum.
Wikipedia says
Natural science is concerned with the description, prediction, and understanding of natural phenomena based on empirical evidence from observation and experimentation. It can be divided into two main branches: life science (or biological science) and physical science.
Social science is concerned with society and the relationships among individuals within a society. It has many branches that include, but are not limited to, anthropology, archaeology, communication studies, economics, history, human geography, jurisprudence, linguistics, political science, psychology, public health, and sociology. Social scientists may adopt various philosophical theories to study individuals and society. For example, positivist social scientists use methods resembling those of the natural sciences as tools for understanding society, and so define science in its stricter modern sense.
Economics is a "social science". Should we also categorise it under the bare label of "science" as well?
I think it would make sense to do so. Most social sciences are based on the first process of the natural sciences: observation. While observation would not suffice for most natural sciences, some like astronomy are arguably predominantly observation based, as humanity is not currently capable of affecting star formation etc.
Experimentation is tricky for the social sciences because of the ethical questions around using human subjects. But with willing volunteers experiments can be made. For example, some people create their own languages, called conlangs, in order to test theories of language, and then teach them to other people.
Economics clearly has a strong observational and theoretical history, but there are also times in which people have experimented with economic ideas. I would say that Basic Income trials would count as economic experiments for example. With observation, analysis and theorising, and experimentation, I think economics should be considered a science.
I fear, there is a misunderstanding about the meaning of "scientific". An experiment is one method in a lot different methods as a scientific approach. (And there are a lot experiments in economics like in game theory.) Even a repeatable experiment is no guaranteed proof, that a theory is true. There might be edge cases in an experimental setting, which might not be covered (read about the famous example about Newton and Einstein). To find these edge cases is an explorative task and I strongly suggest this is scientific, too.
On how to do science, there are a lot of different views on this subject. The most common way today is Empiricism. The "repeatable" is not meant, that an experiment can be repeated in the same way. The meaning is, that other people can repeat the same experience in observing defined situations or objects (like white and black swans).
In social sciences like economics, experiments have an ethic element. Doing experiments with humans is usually a bad idea (see for example the Stanford Prison Experiment). Usually, experiments are done in mathematically models using probability theory. But there are a lot of problems in a sense of repeatable results. To improve these models and methods is a scientific task.
You wrote,
... I made the argument that economics is not science, because it cannot undertake repeatable experiments.
I'd like to take issue with the view that the ability to undertake repeatable experiments is either necessary or sufficient (or both?) for a discipline to qualify as a science.
First off, if your view were tenable, then astronomy could not qualify as a science. Many (most?) astronomers may well be surprised to be informed of this view.
Second, associating "science" with the ability to undertake repeatable experiments misses a fundamental purpose of conducting repeatable experiments. (Aside: I will assume, for the sake of argument, that there is some basic agreement as to what the words "repeatable", "experiment", "theory" and "causality" mean. I will also assume that you consider physics and chemistry to qualify as branches of science.) Repeatable experiments in physics and chemistry are certain impressions notwithstanding, not carried out to create loud bangs and dense smoke. Rather, they are designed, undertaken, and evaluated in order to test a scientific theory and, in particular, to investigate causal relationships, or causes and effects, among the variables considered by the theory. In this view, the outcome of repeatable experiments is to either disprove or to fail to disprove a theory. (Some people seem to consider a failure to disprove a theory to be more or less the same as a proof of that theory; I don't subscribe to this view.) Science advances by building new, "better" theories and discarding disproven theories.
As one of the other answers has already pointed out, the branch of inquiry known as behavioral economics does conduct repeatable experiments. Would you be willing to concede, then, that behavioral economics is a science?
To truly ascertain, then, whether economics is (in part...) a science, one needs to look into (a) what the (scientific) tenets of economic theory about causes and effects among economic variables are and (b) how economists go about testing economic theories.
A nice introductory-level online resource for delving more deeply into the issue of whether economics is (in part) a science is the webpage Nature of Economics: Economics as a Science and an Art. Note: I have no affiliation of any kind with either this webpage or the author of the article.
And, yes, I'm an economist.
To me, the lack of being able to predict (at least on a national level) tells me that economics cannot be a science (at least at that level). It might suggest things about consumer behavior but it can't predict basic things that most people things that economists should know.
Ask an economist to predict ANYTHING that will happen 1 year out on a national level: What will be the average unemployment rate for 2020? How about the inflation rate? When will the next stock market crash happen?
Economists on TV act like they know the answers but they don't. No economist can predict more than a couple of months out.
So no matter how you define economics as as a science or not, politicians and the general public act like it is a science and most economists seem to be OK with that assumption.
Humanities is art not science, but based on science.
In humanities, which economics is the study of humans trading, there are many theories and no facts.
Experience, exposure, practise, and training allows you to choose the most useful theory to the specific circumstances. That is the art.
The goal is to integrate theory and self because as a human you cannot be removed from the practise.
A useful theory predicts something. That's the test.
Welcome, Aleesha
If economics is a science, we have I think to concede that it is does nor have the explanatory scope or depth of (say) particle physics. But this sort of damaging comparison, though common, is not what is needed. Biology is also a science which lacks the explanatory scope or depth of particle physics yet its status as science is secure.
Popper may be right; there might be a reliable way of demarcating science from non-science. I am undecided about this but even if there are demarcation criteria it does not follow that while we can distinguish science from non-science, we can define the nature of science in terms of necessary and sufficient conditions: 'X is a science if and only if ...'.
#Kinds of science
A useful approach is to mark out different kinds of science. Here I'll settle for a distinction between experimental/ non-experimental science. The grounds on which economics is denied scientific status is frequently that it is a non-experimental science. There might be some a priori truths in economics but linked as it is with the management and control of the economy - within free-market or central control limits - the subject's value is likely to be a function of predictions and explanations derivable from whatever experiments we can manage. It is not clear, the putative and minuscule a priori aside, where else the predictions and explanations could come from.
#The case against economics as an experimental science
Here's a useful introductory discussion, just to set out the issues:
Paul Samuelson, Milton Friedman and Friedrich Hayek declared long ago that economics as a discipline was not amenable to the experimental method, just as Aristotle thought physics to be non experimental. Only with 20-20 hindsight do such judgments seem surprising. To understand their basis is to understand the very nature of scientific experiments and their contributions to science. The essence of scientific experiments is control over the environment in which phenomena of interest take place. Of course, by control we mean true manipulative control, not the mere "quasi-control" sought in analyses of data gathered about naturally occurring phenomena. Whether a discipline is judged to be amenable to useful experimentation depends on whether the variables that are believed to be critical to the relevant phenomena are amenable to experimental control. The identity of these variables depends on the state of the extant theory. The state of extant theory, in turn, depends on the available evidence, experimental as well as non-experimental. If the variables that are crucial to the extant theory of a discipline are currently judged to be beyond experimental control, it seems reasonable to conclude, for the time being at least, that such a discipline is not experimental. The state of extant theory defines what the discipline is at any given point of time. Received theories change over time. Paradigms shift when a complicated set of explanations based on many variables are replaced by a simple explanation based on fewer variables. Some variables that were believed to be "crucial" in the old paradigm may be absent from the new. If the lack of controllability of such variables held up the discipline from being "experimental" in the old paradigm, the discipline may become "experimental" in the new paradigm. The double helix model of dna and the plate tectonic theory in geology are recent examples of previously non-experimental disciplines becoming amenable to certain types of experiments because of shifts in paradigms and the accompanying simplification of theory. ('Emergence of Experimental Economics':41.)
The amenability of a discipline to experiments is not inherent in it; it depends on the current state of its theory. When Aristotle wrote about physics not being subject to experiments, he may have meant that this was so with reference to the theories of physical phenomena prevalent in his time, or their subset that was accept able to him. Perhaps a similar interpretation would be appropriate for the judgments passed down by so many leading economists about economics not being an experimental science. How ever, if change in theory is a precondition for the conversion of a discipline from non-experimental to experimental, how could this occur in economics, where no obvious paradigm shift is known to have occurred in recent decades? How does one reconcile the judgments of eminent economists such as Samuelson, Hayek and Friedman with the increasing acceptance and use of experiments in economics in recent years? Why did so many economists believe their discipline to be non-experimental in the first place? ('Emergence of Experimental Economics':41-2.)
#The counter-argument - economics as an experimental science
The scale of macroeconomic phenomena precludes most kinds of controlled experimentation. By the same criterion, astronomy might also be regarded as a non-experimental science. Due to their inability to manipulate planets, stars or galaxies, astronomers had to devise ingenious, sometimes spectacular, quasi-experiments on naturally occurring phenomena to adduce convincing evidence to reject contending theories (for instance, an experiment during the 1914 solar eclipse verified Albert Einstein's prediction about the curvature of space in the neighbourhood of heavy celestial objects). It would be politically and ethically difficult, if not impossible, to conduct macroeconomic experiments that manipulate monetary and fiscal policies to gather observations to verify or reject various macroeconomic theories. To this extent, the non-experimental tag assigned to economics seems justified. But these arguments do not apply to price theory or micro economics, just as the arguments about astronomy are inapplicable to the physics and chemistry of most terrestrial phenomena.
In his introduction to Price Theory, Friedman wrote,
Economic theory, like all theory, may be thought of in two ways. It may be thought of as a language or filing system, or it may be thought of as a set of substantive empirical propositions. With respect to theory in the first meaning, the relevant question to be asked is usefulness and not Tightness or wrongness. The proposition that price is determined by the interaction of demand and supply forces is merely an attempt to set up a useful filing system within which there can easily be placed under the headings "demand" or "supply" any one of the forces affecting price. The usefulness of this filing system will in turn depend on the substantive fact whether a listing of the forces operating on demand contains few elements in common with the listing of the forces operating on supply. Economic theory as a list of substantive empirical propositions contains propositions which are, in principle, capable of being tested because they attempt to be predictive. The definition of a demand curve is theory as language. However, the statement that the demand curve slopes downward to the right is theory as a substantive empirical proposition. It has empirically observable consequences, whereas the definition of a demand curve does not. Theory as language coincides with Marshall's engine of analysis. His objective, as well as that of any other investigator, is to construct a language that will yield as many substantive propositions as possible (1962: 8).
The experimental method has proved to be a powerful engine for generating substantive empirical propositions, and for distinguishing between competing propositions for prediction of the same phenomena - precisely the function Friedman attributed to a large part of economic theory.
We conjecture that the reasons for the residual distrust of experiment methods among economists are some of the same reasons that have given economics, among all social sciences, an extraordinary degree of coherence and power - its willingness to abstract from reality and its ability to use the mathematical technique of optimisation as a fundamental organising principle. Abstraction from reality is necessary for assumptions such as atomistic competition among fully informed agents, who are assumed to optimise for the market to arrive at an equilibrium. However, such abstraction and optimisation are devices of convenience to be used to build and solve simple models that can be tested for their predictive ability; they are not to be taken seriously in themselves. Absent experiments, and given the limitations of quasi-experiments that depend on naturally observable phenomena alone, abstract assumptions and optimisation came to be taken far too seriously by economists themselves. Since we know that no laboratory market could possibly meet all abstract assumptions and real human beings do not have the ability to optimise complex problems by intuition alone, it is easy to reject the potential of experiments to contribute to economic theory. This is why the results of Smith's experiments, conducted in the late 1950s with a mere handful of students, came as a surprise not only to economists at large, but also to Smith himself who was trained as an economist and continued to regard his experiments as an avocation until well into the mid-1970s (Smith 1991).
The case pursued ...
About the same time experimental techniques made inroads into price theory, game theory and industrial organisation theory, microeconomic theory began to be used to build a new macroeconomic theory. Experimental techniques have followed microeconomics into the new macroeconomics to identify substantive empirical propositions about expectation formation, monetary policy and inflation. Years ago, physics, with its sound experimental foundations, had made inroads into astronomy in the form of astrophysics. Reinhard Selten suggested a different explanation for the mid-20th century spurt of interest in experimental economics. The appearance of von Neumann's and Morgenstern's The Theory of Games and Economic Behaviour (1944), oligopoly theory, and cooperative game theory offered a multitude of solution concepts that could not be resolved by application of a priori reasoning alone. So those who had contacts with experi mental psychology turned to experimental methods for selecting solution concepts. This explanation pointed to what is, perhaps, a unique role economics experiments can play - a role in selecting from multiple equilibria. It is consistent with the recent interest of theorists in experiments with overlapping generations of economies with multiple equilibria.
#References
'Emergence of Experimental Economics', Economic and Political Weekly, Vol. 46, No. 35 (AUGUST 27-SEPTEMBER 2, 2011), pp. 41-46 Published by Economic and Political Weekly.
Decision processes. Edited by R. M. Thrall, C. H. Coombs, and R. L. Davis, John Wiley & Sons, Inc., 1954, 332 pp.
M. Friedman, Price Theory, ISBN 10: 0202060748 / ISBN 13: 9780202060743 Published by Aldine Transaction, 1962.
J. von Neumann & O. Morgenstern, The Theory of Games and Economic Behaviour, Princeton: Princeton University Press, 1944.
Vernon L. Smith, Papers in Experimental Economics. New York: Cambridge University Press, 1991.
Certainly some economics qualifies as science. Here is an example of "hard science" as used in economics.
The Black-Scholes options pricing model is used by exchanges and traders to compute fair prices for options contracts traded on derivative instruments.
The model combines statistical methods (used to compute a measure of underlying volatility) with the methods of calculus (functions, derivatives, and integrals) to compute a "fair price" based on the parameters of the underlying contract.
These methods are applied by the exchanges at the end of each trading day to establish an official settlement price. Conversely, during the trading day, traders use the model by inputting current prices and computing the "implied volatility" - looking for arbitrage opportunities across associated markets.
The Black-Scholes model is a mathematical model, as is most of physics these days. The authors received the 1997 Nobel Memorial Prize in Economics.
