BY UNLEARNING ECONOMICS 07/10/2014
What is the way forward for economics and homo economicus following the financial crisis? To answer this I will consider a debate which is important in theory, yet often sterile in practice: is economics a ‘science’? In this essay I will argue that economics contains elements of both science and philosophy, and that these must both be embraced if it is to move forward. However, embracing this will also entail making major changes to current practice.
Economics certainly stands head and shoulders above other social sciences in its use of mathematical and statistical methods, attracting many bright minds and giving it an aura of technical difficulty that fields like sociology lack. However, despite this, the lack of laboratory experiments and the inherent unpredictability of social behaviour make economic theories hard to subject to the same empirical rigour as the natural sciences. Critics are quick to point out that economic theories are often built on unrealistic foundations yet have remarkable staying power (Keen, 2011). However, economists have formed a standard line in response to these criticisms. Although they may appreciate that economics is not, and probably never can be, on the same plane as natural sciences in terms of the accuracy and predictive power of its theories, economists maintain that formal, mathematical modelling serves the purpose of ensuring valid logic and provides a platform to test assumptions and conclusions statistically (Rodrik, 2013; Debreu, 1986).
While partially appealing, this is clearly not enough to establish the scientific status of economics. It is not enough for a theory to be clear, precise and logically coherent: it must be well established that its major implications are consistent with available evidence. In order to test a theory’s assumptions and conclusions properly, the mathematical relationships and quantities it depicts must represent measurable, real world properties. Properties in science such as weight, speed and temperature can be measured and used to generate falsifiable propositions; economics is rife with unobservables such as utility, propensities, expectations and elasticities. At best, we can infer the value of these things from observed behaviour, but there is no way to observe them ex ante in order to make predictions, rendering them somewhat tautological: any observed behaviour can have some value or mathematical function assigned to it ex post.
The result of this is that economic models often represent simplified versions of reality, used to think through problems systemically, rather than truly ‘scientific’ theories. In their paper ‘Economic Models As Analogies’, Gilboa et al (2013) distinguish the ‘rule-based’ approach of natural science – where theories are seen as generally applicable within a given domain, and where contradiction by evidence even once is enough to raise serious questions about a theory – and the ‘case-based’ approach of economics, where each theory is seen as a story from which we can draw lessons, and “empirical data, experiments, or theoretical analysis [have] the same epistemological stature.” In the latter formulation there is no reason to abandon a theory when it is contradicted by empirical evidence, because “cases do not make any claim to generality, and therefore they cannot be wrong.”
Not all economists agree with such a view – Larry Summers seemed to imply the exact opposite when he asserted that “the laws of economics are like the laws of engineering. One set of laws works everywhere.” Although that is not an opinion representative of the whole economics discipline, it would be inaccurate to describe all economic models as existing only in the theoretical vacuum of ceteris paribus given the way they are applied. The DSGE models used by central banks are often fitted to real world data, while canonical theories such as demand-supply and the Efficient Markets Hypothesis (EMH) are held to be generally applicable, describing the usual workings of goods markets and financial markets respectively. There is obviously some unspoken internal disagreement between economists about the role of their theories.
Is economics philosophy?
The ‘case-based’ approach used in economics resembles philosophy in a number of ways. First, logicians make use of basic algebraic labels to think things through, even though these labels may not apply to ‘scientific’ properties or categories. It can clarify our thinking to say that statement A implies statement B, even if statement A is that Jeff likes jam and statement B is that Jeff likes jam on toast. Second, political philosophy also makes use of unfalsifiable thought experiments to inform understanding, such as the classic example of the runaway train which can be diverted to kill one person, saving 5 others in the process, or Robert Nozick’s (1974, pp.160-164) Wilt Chamberlain example. Finally, economic theories cannot avoid embracing ethical considerations whatsoever, as even the choice of how to evaluate economic institutions and policies itself contains judgments about tradeoffs between different outcomes. In fact, much of economics could be interpreted as an attempt to mathematise utilitarian philosophy, with economists often asking which policy or set of institutions will produce the maximum total welfare (Pareto optimality).
Can this approach be considered a sound way to explore economic issues? Defenders of the approach would presumably argue that the social world is irrevocably complex, and by isolating key features one at a time, we can fruitfully inform our understanding of the whole. Those who mistake the map for the terrain may be in the wrong, but the approach itself is not. Consider the seminal Akerlof (1970) Lemons model. If this were a scientific model, it should probably be discarded or modified: its central prediction that used car markets will fall apart is obviously wrong. However, if we construe this model as a mere analogy, used to inform our understanding of car markets, then we can approach it from a different perspective. Used car markets do not fall apart because intermediaries such as autotrader, or methods for uncovering information about cars, or simple trust, prevent them from doing so. By helping to understand why these things are needed, the Lemons model can inform our understanding of how information asymmetry manifests itself and how it can be overcome.
However, this still leaves us with a few unanswered questions, most notably which analogy to apply and when. The major danger here is that we fall into an “anything goes” methodology: as long as a theory is logically coherent, intuitive and fits a few stylised facts, it is accepted into the pen with countless other theories. If we have no objective criteria by which to rule out certain theories, how do we avoid the choice of theory and assumptions being subject to the whims of the practitioner, in a similar vein to Leamer’s (1983) worries? On the other hand, once we open the door to empirically reviewing or rejecting theories, we are moving back towards the ‘rule-based’ approach and there is no reason to uphold the contention that a theory cannot be wrong. To refer back to the Lemons model, we may want to modify it to include the real-world factors that make used car markets work, dropping the original version in the process. If the model subsequently becomes too convoluted or its predictive power does not improve, it may be better to restart from an alternative foundation.
…or is it science?
Is it even possible for economics to go in a more scientific direction? Lawson (2013) has argued that due to the ontological nature of economics, and social science in general, we generally cannot hope to rely on the “event regularities” depicted by mathematical equations. Previously observed social relationships are not guaranteed to continue as people adapt to changes in policy or other circumstances. The result is that while even a simple econometric relationship such as Y = ? + ?X + ? could be true today, or for a significant period of time, there’s no guarantee it will continue to be in the future, particularly if the model itself becomes public knowledge or a guide to policy (Lucas, 1976). This means any mathematical model is at best limited and historically contingent, and the more relationships depicted, the higher the chance the model will break down.
If Lawson is right, mathematical economics – at least the level of maths we currently see in mainstream journals – could be a dead end. This critique would also apply to a lot of heterodox economics – particularly post-Keynesians, Sraffians & Marxists – which also makes use of mathematics, and which often seeks to uncover the rules which govern the economy. As Lawson himself stresses, this doesn’t rule out any use of maths whatsoever, but it does mean that theories are always historically, politically and ethically contingent, and must evolve along with social reality. It’s also the case that theories in economics can be partially true: for example people may change their behaviour in response to a tax, but not by as much as standard theory predicts. This stands in contrast to physical science where there is a sharper line between right and wrong, and along with the aforementioned political, historical and ethical contingency of economics, this explains why competing, even contradictory theories can all have a place in the economists’ toolkit.
However, if economics evolves in the direction of the rule-based, scientific path, there must be more willingness to abandon theories which are consistently shown to be at odds with the evidence. Fundamental but largely unexamined questions such as “is the canonical demand-supply wrong?”, “do people act on preferences?” and “is long-term growth demand (output) driven?” should be put back on the table. Furthermore, when theories are abandoned, there must be a willingness to embrace new theories, bearing in mind that a promising new theory will not come fully equipped, with all the kinks ironed out and with every possibility accounted for. A choice of this nature would ideally entail an immediate shake up of existing theories and an honest, empirical consideration of new (or old) ones, such as post-Keynesianism (Lavoie, 2007), Marxism (Kliman, 2012), Evolutionary Economics (Potts, 2000) and many more. Consideration of these theories must take them into account as a whole, rather than simply referencing them in the title of a paper which puts one aspect of their ideas into a standard general equilibrium model (such as Eggertson & Krugman, 2012).
Economics resembles philosophy in that there is room for competing theories, ethical considerations are inescapable, and economic theories can inform our understanding without necessarily being as accurate as theories in the natural sciences. Nevertheless, thoughtlessly embracing the idea that all models are wrong leaves us unable to see the analytical forest for the trees, and there must be more room for outright falsification of some theories, if the discipline is to learn from its mistakes. Due to its complexity and evolutionary nature, economics cannot and should not be practised as a natural science. However, if it so chooses, it can etch out a unique place for itself among fields, taking some of the best features from both the natural and social sciences and regaining respect as a worthwhile endeavour.
Akerlof, George A. “The Market for “Lemons”: Quality Uncertainty and the Market Mechanism.” The Quarterly Journal of Economics 84.3 (1970): 488. Web.
Debreu, Gerard. “Theoretic Models: Mathematical Form and Economic Content.”Econometrica 54.6 (1986): 1259. Web.
Eggertsson, G. B., and P. Krugman. “Debt, Deleveraging, and the Liquidity Trap: A Fisher-Minsky-Koo Approach.” The Quarterly Journal of Economics 127.3 (2012): 1469-513. Web.
Gilboa, I., Postlewaite, A., Samuelson, L. and Schmeidler, D. (2014), “Economic Models as Analogies”. The Economic Journal, 124: F513–F533. doi: 10.1111/ecoj.12128
Keen, Steve. Debunking Economics: The Naked Emperor Dethroned? London: Zed Book, 2011. Print.
Kliman, Andrew. The Failure of Capitalist Production: Underlying Causes of the Great Recession. London: Pluto, 2012. Print.
Lavoie, Marc. Introduction to Post-Keynesian Economics. N.p.: Palgrave Macmillan, 2007. Print.
Lawson, T. “What Is This ‘School’ Called Neoclassical Economics?” Cambridge Journal of Economics 37.5 (2013): 947-83. Web.
Leamer, Edward E. “Let’s Take the Con Out of Econometrics.” The American Economic Review 73.1 (1983): 31-43. JSTOR. Web. 29 Aug. 2014. <http://www.jstor.org/stable/10.2307/1803924?ref=no-x-route:f0875385d47d40334893b30a853bb5b1>.
Lucas, Robert E. “Econometric Policy Evaluation: A Critique.” Carnegie-Rochester Conference Series on Public Policy 1 (1976): 19-46. Web.
Nozick, Robert. “How Liberty Upsets Patterns.” Anarchy, State, and Utopia. New York: Basic, 1974. N. pag. Print.
Potts, Jason. The New Evolutionary Microeconomics: Complexity, Competence, and Adaptive Behaviour. Cheltenham, UK: E. Elgar Pub., 2000. Print.
Rodrik, Dani. “What Is Wrong (and Right) in Economics?” ‘Dani Rodrik’s Weblog’ Dani Rodrik, 07 May 2013. Web. 29 Aug. 2014.