Whats Wrong With Economics?

What's Wrong With Economics?

Add to Reading List
Add to Reading List

The question of whether mainstream (neoclassical) economics as a discipline is fit for purpose is well-trodden ground, particularly on the internet. However, popular debates often focus on either policies or on specific theories, instead of economist's approach as a whole. While both simplistic 'free market' ideas and unrealistic theories are surely worthy of criticism, in my opinion this approach does not get to the crux of the issue. Economists will simply argue (somewhat correctly) that free market ideas are just a perversion of neoclassical models, and that certain theories are not important or can be modified to incorporate the critic's objections (again, somewhat correctly).

However, I think the problems with economics are broader and more general than any theory or policy. Economic theory is flawed, not necessarily because it is simply 'wrong', but because it is based on quite a rigid core framework that can restrict economists and blind them to certain problems. In my opinion, neoclassical economics has useful insights and appropriate applications, but it is not the only worthwhile framework out there, and economist's toolkit is massively incomplete as long as they shy away from alternative economic theories, as well as relevant political and moral questions. In other words, the problems with economics cannot be dismissed as isolated problems with, say, areas of undergraduate economics, or attributed to a few Chicago School loose cannons. In this post, I will outline the problems I see with economics as currently practiced.

Economics lacks pluralism 

Unlike other social sciences, which are generally characterised by competing schools of thought, mainstream economics has dominant way of thinking: neoclassical economics. Neoclassical models generally start from an individual agent facing clear choices under scarcity: a firm maximising profit subject to costs; a consumer maximising utility (satisfaction) subject to a budget; even a politician aiming to get reelected by choosing policies which will most appease different interest groups. Economists often take parameters such as technology, preferences and so forth as a given, and then ask what happens if agents in the model optimise their choices, meeting through a market to interact (not always in equilibrium, as is commonly thought, though often this can be the case). This approach roughly characterises a lot of mainstream economics, from theories of the firm to the dominant macroeconomic Dynamic Stochastic General Equilibrium (DSGE) models to the canonical supply-demand.

Economists also have a stock of reasons that their approach is superior. First, they believe economics must explicitly model agents who make decisions, which is what differentiates it from, say, physics, where atoms have no autonomy. Second, economists believe that having these agents modeled on the "deep parameters" of human behaviour means that the agents' behaviour will not change when policy changes, an approach which stems from Robert Lucas' 'critique' of naive aggregate modeling. Third, economists believe their approach allows them to state their assumptions clearly so they know exactly what they're modeling and can understand and isolate causal links.

However, there are important counterpoints to these arguments. First, there is no reason to believe that the economy as a whole should necessarily be modeled from individual choices: people's choices may interact and depend on each other to such an extent that each individual is rendered irrelevant, and we can proceed from a structural standpoint. Second, there is no reason to believe that there are any "deep parameters" of human behaviour that will not change with policy (ask an anthropologist), and we cannot immunise ourselves from this problem by modeling things a certain way; rather, we simply need to be aware of the ever-evolving relationship between policy and the economy. Third, rather than merely allowing economists to state their assumptions clearly, the complexity of modelling from individual agents all too often means economists have to make a plethora of assumptions, implicit and explicit, which are not only unrealistic but come to form a straitjacket over their models.

There are alternative approaches to economics that are equally valid. Sraffian economics takes the classical view that societies must reproduce themselves - that is, outputs in one period must be sufficient for inputs in the next period - and uses it as a benchmark for analysis. Marxism, generally regarded by economists as 'discredited' (mostly due to misunderstandings over what is meant by the word 'value') models the business cycle as a function of the labour put into production. Evolutionary Economics and Econophysics model the economy using tools from biology and physics, respectively. Institutional Economics tends to take nothing as a given, and first looks at things like history, laws and culture to see what makes firms and economies tick. Models such as Steve Keen's, or the Modern Monetary Theory (MMT) school, take banks and debt as central, and as such are highly suitable for modelling financial crises. Economists sometimes take elements of these approaches into account, but it's pretty difficult to get a model with a completely different framework accepted in a major mainstream journal.

Despite the existence of alternatives, economist's insistence on using a certain approach means they can find it hard to accept any other way of thinking, which leads another problem: 

Economic theories persist despite empirical and logical problems

Given such uniformity of approach, one would expect economic theory to be remarkably consistent with available evidence, free of substantial controversy and to offer clear explanations of phenomena such as business cycles, income distribution and firm behaviour. But this is not so: economists have not  found a universal explanation or collection of explanations for, say, the business cycle, and also differ on how best to counter it. Economic theory is historically rife with controversies over its internal consistency and relevance, and often problems are conceded or even pointed out by influential theorists. Paradoxically, though a large degree of economic work nowadays is empirical, key theories are rarely (or never) truly falsified.

One such example is the IS/LM model, which remains a cornerstone of modern undergraduate macroeconomics, and is used by influential economists in policy discussions. This is despite the fact that the model's creator, John Hicks, criticised and disowned the model decades ago. He believed that the model - which was supposed to be an easily digestible interpretation of John Maynard Keynes' famous General Theory - did not include a role for irreducible uncertainty, which was fundamental to the concepts IS/LM was trying to espouse. The result was that it was not possible to interpret the time period of the model in a way that rendered it internally consistent. Now, the mere fact that it was the model's creator who disowned it doesn't mean he was right, but it should at least make us sit up and pay attention - after all, the creator of a model is surely best placed to understand it.  Despite this, the criticism is rarely mentioned (or even known) by IS/LM's modern adherents, or taught on courses.

There are many such examples of persistent, but flawed, economic theories: production functions and the Solow growth model are still commonly used, despite the Cambridge Capital Controversies, in which the creators of the models conceded key points about the validity of their approach. DSGE persists despite the fact that that dominant versions failed to help central banks foresee and ameliorate the 2008 financial crisis. Neoclassical theories of the firm persist despite the fact that studies of the firm show firms barely even understand economist's approach, let alone adhere to it. Game Theory is widely loved by economists, despite being falsified in experiments and having no useful application in the real world so far, something admitted by notable game theorist Ariel Rubenstein. Utility maximisation persists as an explanation of consumer behaviour, despite a myriad of evidence that contradicts it: most notably time inconsistency, which shows that people's decisions change so much over time that they cannot really be said to be acting on coherent preferences.

The standard economist's response to such criticism is "oh it's just a tool, a simplification, not reality". At worst, this roughly translates as "there is no event, argument or alternative that could make me abandon my theory completely". This mentality is effectively a crude version of Milton Friedman's Methodology of Positive Economics, where Friedman argued that economic theories should not be judged by their assumptions but by their predictions. Yet Friedman's essay lacked coherence, as he failed to offer any falsifiable predictions, and actually twisted his logic to dismiss early evidence contradicting neoclassical theories of the firm. Friedman's essay has met many refutations, one of which was by influential economist Paul Samuelson, who argued that a theory is really a collection of hypotheses, all of which are refutable and hence subject to continual revision, and - contrary to what Friedman implies - there is not a clear cut that can be made between the internal mechanics of a model and the 'predictions' it makes.

It would be completely wrong to suggest that economists never tackle interesting problems like behavioural biases or institutions: in fact, both of these problems are quite in vogue in mainstream economics. However, they almost always retain the methodology outlined above as a starting point, and try to annex a particular anomaly to their core framework: the 'frictions' approach described by Noah Smith. The result is that it's hard to find an economic model that does not retain a few unrealistic characteristics in areas that are not deemed to be under investigation, something economists will readily admit. But if we know the concepts or assumptions we are using are largely wrong, why continue to use them? Science progresses by changing its framework to make it as consistent with available evidence as possible, not by exploring any every documented departure from its framework as an isolated 'anomaly'. Economists seem to think the neoclassical approach is the default starting point for economic analysis, but I don't see how this role is deserved.

The irony here is that, though economics is often touted as the king of social sciences due to its mathematical and scientific approach, economists seem to take their theories - their 'way of thinking' - as a starting point through which they interpret reality, rather than as a truly falsifiable framework. The result is that the way economic models are used has more in common with something like Karl Marx's historical materialism than science. However, unlike marxists, economists are (erroneously) wont to see their approach as neutral, which is my final problem: 

Economic theory too often tries to be ahistorical, amoral and apolitical

Economists try place their analysis above ethical and political considerations, and purport to engage in purely technical, 'value free' debates. However, the fact is that economic models inevitably contain judgments: what should we study, and under which criteria should we evaluate it? It may be true that the statement 'the minimum wage increases unemployment' will be interpreted differently by people who have different values, but the fact that we are even studying the minimum wage, and its effect on unemployment, implies that the relationship between the two is important. Furthermore, it is a stretch to suggest that the value judgment is much of a leap from the analysis: someone who had been taught that the minimum wage unambiguously created unemployment would be likely to be against it (economists also generally favour things like 'efficiency' and 'competition', opinions which rely on value judgments).

Ignoring a particular ethical or political concern does not make it irrelevant; it simply reflects a judgment on the part of economists that the concern is not important. And this judgment may be mistaken - consider the work of the philosopher Michael Sandel, who has recently highlighted that economic theory assumes buying and selling a commodity is independent of the commodity itself. This is generally not borne out by real world examples, as buying or selling something can change how people perceive it, which can have far-reaching impacts on how markets function. Is systemically ignoring such considerations in one's models a value-free position, or just a blatant denial of reality?

The nature of economic models also lacks historical perspective. As Marx pointed out some time ago, economists usually take capitalism, and the institutions and social relations it entails, as a given for their analysis. For example, the common analysis of markets versus governments takes markets and the institutions required for them - such as property rights - as a given, and terms further institutions or laws an 'intervention' into the market system. This creates a false dichotomy where interventions past a certain arbitrary line are open to criticism and debate, while those that are within the bounds of what is 'necessary' for the basic market model are not. A similar problem emerges with the behaviour of economic agents. These agents are usually fairly materialistic, 'maximising' their profits or consumption, but questions about the desirability of this behaviour are not asked. Surely this silence can only serve as a mark of implicit approval and legitimisation.

A final area where economists are seemingly blind is politics. In fact, the political blindness of even the most well-intentioned economists can be surprising. I was discussing monetary policy with economists Miles Kimball and J.P. Koning, and they both endorsed a specific approach, advocating various measures to achieve their favoured outcome. However, these policies were completely unacceptable politically: one of them was suspending 1:1 conversion of bank deposits into cash; another was intermittently declaring random notes and coins worthless to stop people holding money. I think I speak for many when I say 'no thanks' to such ideas, yet their feasibility or acceptability wasn't even deemed an issue. Surely, though, the true test of whether an economic policy or system is desirable is whether people want it? After all, 'the economy' is really just an abstract idea; what ultimately lies beneath the models are people, and if they don't like your idea, chances are you're doing something wrong.


Ethics, history, politics and alternative theories can and should be a bigger part of being an economist. By trying to make economics an isolated sphere of study so that they don't have to deal with such considerations, economists do themselves no favours, as they only end up boxing out important and potentially productive debates. Economists may dismiss such concerns (particularly alternative theories), but as economics continues to lose its stature among students, employers and the general public, they're really only doing a disservice to themselves, and the discipline as a whole.


Keep up to date with the latest thinking on some of the day's biggest issues and get instant access to our members-only features, such as the News DashboardReading ListBookshelf & Newsletter. It's completely free.


Please read our Community Guidelines before posting
Unlearning Economics

Eh, only just saw this, and it's really almost not worth replying to - it's mostly an incredulous rant, punctuated by occasional uncharitable interpretations of things I may or may not have said previously, all in a vein attempt to 'catch me out' and show my logic is incoherent or what have you.

" Tricky, without a appealing to a production function involving capital and a decision theory."

Oh, right, you've asserted it so it must be correct. "Oh my god, a difficult question! Let's revert to neoclassical economics."

"Decision theory" does not mean "neoclassical decision theory". You don't need a production function to derive an estimate for whether an investment decision will be profitable; you simply calculate expected returns and costs, and proceed accordingly. Firms certainly don't use neoclassical concepts, so according to you how do they ever make decisions!? It's impossible!!

"But on another thread you told me that people looking to store value wouldn't switch into consumption if the returns on assets fell. So how are you going to account for people choosing to invest less if the return on capital falls?"

Wow, this is a very convoluted way of trying to suggest I've contradicted myself. If the return on assets falls, people may choose to hold cash instead. They may also choose to invest; and, if the return on this falls, they may again hold cash.

"And what do you mean by "falsifying" a production function? It's a description, which is going to be inaccurate (false) to some degree because it's obviously a highly simplified picture of a much more complex reality."

Falsification is fairly straightforward: if sufficient evidence is gathered to refute the predictions of a theory, it should be abandoned. And no, you can't pull your typical economist catch all of "no, it's just a simplification" to avoid any and all criticism - the type I outline in this very post. There HAS to be some criteria that would make you abandon the PF, else it's not science. It's faith.

Anyway, I'll ignore the rest of your rant about how I would do economics without doing it your way, and answer this question directly:

"What "actual causal mechanism" are you going to appeal to?"

Case studies of firms, investors and consumers which show how and why they make the decisions they do. You can build models based on these observed decision processes, as Institutionalists and post-Keynesians have done (though not enough, I agree). You could then use these mechanisms to interpret the data.

A PF simply implies that a higher price of capital would result in less investment in capital. This observation seems trivial to me/i>

trivial? oh dear me no. At least, not for you. It might be straightforward to explain in context of utility max decision makers facing a known production function, with suitable data to check against your theory. But how are you going to proceed?

First, let's be clear: Prices aren't part of a production function, which relates quantities of physical inputs to outputs. The prices of inputs and outputs are a separate question.

How do capital prices affect investment decisions? You need to explain what determines the return on investment and how people take decisions. Tricky, without a appealing to a production function involving capital and a decision theory. Of course you wouldn't merely want to come up with a verbal version of what you decry in economics.

Mainstream theory has no problem explaining why people would choose to invest less, facing any given production function, if capital prices were higher, making current consumption relatively more attractive. But on another thread you told me that people looking to store value wouldn't switch into consumption if the returns on assets fell. So how are you going to account for people choosing to invest less if the return on capital falls? (a higher capital price means you have to pay more , in terms of consumption goods, per unit of future output returned). You desire coherent theory, don't you.

But you have deeper problems. Didn't Cambridge UK point out that higher capital prices might just reflect higher returns on capital? If so, perhaps higher prices could even be associated with higher investment. You need a theory of capital price determination. And what data are you going to observe? You can't use estimates of aggregate capital stock, based as they are on assumptions you reject. You can't infer anything about how expenditure on capital goods translates into productive capacity and hence into output, if you cannot appeal to commensurate units of physical capital some stable production relation (function).

And even supposing you find some acceptable data, how are you going to "observe" it? Look at a correlation matrix? How do you interpret that? Don't tell me your going to run linear regressions on aggregate data! You haven't explained how you would distinguish between hundreds of competing explanations simply by observing correlations in the data.

Let's return to the idea of production functions. Are you going to assume production relations in aggregate are similar across countries? If not, perhaps capital prices vary across countries because the nature of production varies across countries, and it's the variation in production relations that explains variation in GDP, and in no sense are higher capital price "resulting" in lower GDP via causing less investment. There are many other possible factors that might be correlated with capital prices that in fact explain their correlation with GDP, without capital prices having any causal role at all.

What "actual causal mechanism" are you going to appeal to? You haven't said. I might hypothesise one, but I know I'd have a devil of a time testing it, although at least I might be able to use a model to generate some quantitative predictions.

And what do you mean by "falsifying" a production function? It's a description, which is going to be inaccurate (false) to some degree because it's obviously a highly simplified picture of a much more complex reality. However I might very well decide it's a poor description, and "reject" it, if I can show it makes seriously misleading predictions (with "serious" being context specific). And what do you know, there's a strand of research doing just that, which you seem to be unaware of, being under the impression economists have closed off production theory to "falsification".

So here's your problem. You spend all your time looking for ways to knock economics, but show no sign of appreciating why economics has evolved as it has, and you don't seem to realise how your critiques undermine your own ability to address economic problems. You're like somebody doing cost-benefit analysis without any interest in the benefits.

Unlearning Economics

I'm aware that the production function is pervasive in a lot of econometric work, and it's a problem.

I see a need for theories but ones based on actual causal mechanisms (which themselves can be verified) rather than logically incoherent theories. You can't just assume a theory and claim that your approach is somehow superior for estimating causal relationships, then close your theory off to falsification (how would a PF be falsified for you?) and logical criticisms. If this is what economists 'do' it is not worthwhile.

A PF simply implies that a higher price of capital would result in less investment in capital. This observation seems trivial to me. As far as quantifying it goes, I don't see how a PF gives us any more information than those observed in the data. I don't see how any of your examples cannot be observed by looking at data.

sorry, that was unnecessarily rude. but really, you write all this stuff about economics and then you don't see the need for theory if we ever want to do more than observe correlations.

you do appreciate, by the way, that many of the regressions one might run to investigate the relationship between output and investment are estimated aggregate production functions?

Right. Well I'm sorry, and I know you won't like this, but I think that reply reveals that you really don't understand the first thing about what economists do. We can observe hundreds of things that are correlated with wealth/poverty. Which of these explain poverty? Which of these are explained by poverty? If policy interventions are available to change one of these things, should it be undertaken? We might be able to tell a theoretical story in which a high relative price of capital goods might predict lower GDP, but can we quantify the relationship? How important is this versus the hundreds of other potential explanatory stories? If we think higher capital prices result in lower investment and a lower capital stock, what is the relationship between the capital stock and output? None of these things can be done by observing correlations.

Unlearning Economics

Solow seems to me to be suggestion PFs are at best a way-station to something better. Economists don't seem to be overly concerned with reaching that something.

I don't see any reason to invoke a PF in your example. If high costs are associated with lower growth in the data, isn't that enough? You'd simply run regressions on the relevant data such as investment, the price of certain types of goods (perhaps a weighted average), growth and so forth. I don't see what a PF would add.

right! well Solow isn't saying economists should abandon aggregate production functions because they are "flawed" - illuminating parables and a means of organizing the data puts it very well. I'd add potentially useful approximations too, for those occasions where you want a model to make predictions.

as often when I read your critiques, I think I'd like to see you go off an actually attempt some economics can then come back in a few years time and see if your attitude towards useful conveniences and approximations has changed.

take capital for instance. we know capital goods exist and are used in production. So how are you going to proceed, now that you are "not having it"? Take an interesting hypothesis, motivated by the data, that the relative price of capital goods is higher in poor countries and that partly explains lower output. If you have a simple production function with capital goods you can write down a simple model to explore the idea that the high relative price of capital goods could explain lower levels of output. You could then take your model to the data, as they say. There are lots of, to my mind, interesting papers on this topic. How would you approach the problem?

Unlearning Economics


Thanks for the comment. As always Solow is on the money, though once respected economists start disowning the mainstream, they are ignored or even ridiculed (admittedly I'm using econjobrumours as a reference point for the latter, but it does seem that generally, mainstream people don't pay much attention to Solow's critiques).


I appreciate that the Lucas Critique is a spectrum, not a binary. But we need to be more aware of it: I have seen repeated references to the idea that microfoundations can 'immunise' us to it (indeed this paper:, which you once linked me to, contains such a term). I do not think that we can dismiss or approve a model out of hand based on the LC; we have to examine the posited relationships in the model carefully and see how they will change with policy.

Re: production functions, to quote Bob Solow yet again:

"I have never thought of the macroeconomic production function as a rigorously justifiable concept. … It is either an illuminating parable, or else a mere device for handling data, to be used so long as it gives good empirical results, and to be abandoned as soon as it doesn’t, or as soon as something else better comes along."

You're right that input-output models are desirable. But this 'capital' stuff - I'm not having it. We need to have coherently defined units, and without them we just end up with questionable 'properties' like TFP:

I'm willing to accept production functions insofar as they deal with physical units of something, and I appreciate that a lot of advanced economics uses set theory to do just this. But we need to be more willing to abandon the old models.

Re: Keen, the model is flawed and incomplete, as Keen himself will readily concede. But as Kuhn pointed out, the incumbent will always be more developed than the challenger. The question is whether Keen's framework is more promising than, say, DSGE. I think that it is; you don't. But I digress.

some small points.

first, don't get too hung up on this deep parameters thing. You don't need to call in the anthropologists. Economists really aren't supposing the existence of eternal universal human constants. They just need descriptions of behaviour that are sufficiently constant in a time and place to do a reasonably job of explaining how people will respond to different policies or shocks or whatever. I think people get the wrong impression from the phrase "deep parameters" , it really means just one layer "deeper" than the estimated parameters in a regression that shows last time interest rates when up by X investment fell by Y.

second, you write things like how economics still uses production functions and the Solow growth model despite the Cambridge Capital Controversies .... really you think economics should have no models for how inputs are combined to produce output? I think economics just need to know that they are dealing with simplifications, they need to know about aggregation problems (hint: lots of them do!) but they don't need to stop using these things. The Solow model remains very useful, despite the fact you could fill a side of A4 with a list of things it ignores. You really seem to think that pointing out "flaws" in a model somehow means they should not be used. That's wrong. Every model has flows, the prophet Keen's included.

Solow's concern from 1986 rings truer than ever today:
‘My impression is that the best and brightest of the profession proceed as if economics is the physics of society. There is a single universally valid model of the world. It only needs to be applied. You could drop a modern economist from a time machine … at any time, in any place, along with his or her personal computer; he or she could set up in business without even bothering to ask what time and which place’.
(From 'Economics: Is Something Missing?', p. 25 in Economic History and the Modern Economist).
While I question what Solow perceives as 'the best and brightest' for the most insightful work does not necessarily come out of so-called ‘the best and brightest’ schools, I concur that there is great need for change --modeling frictions won't cut it-- but there remains little to no demand within the establishment circles to embrace pluralism: the interwar period saw pluralism while the postwar period saw the rise of the neoclassical orthodoxy and the eventual exclusion of non neoclassical analyses from major journals.
The use of mathematics should add colour to the analysis not lifelessly deduct from it and, as Sidney Weintraub has pointed out in 'How Economics Became a Mathematical Science', the goal posts of what is deemed scientifically rigorous and mathematically demanding changes with the time and would not meet with approval from mathematicians or physicists.
If there is to be a revolution then it won’t happen anytime soon; as you have articulated the bounds of the expressible are exceedingly narrow: Paul Krugman disagreeing with John Cochrane isn’t a debate, it is farce. There are some very intelligent people, such as Michael Kumhof, who uses a neoclassical framework to flesh out some great ideas on banking and inequality that can make the orthodoxy relevant again but they are the exception rather than the rule.
Love him or loathe him, I’ll finish with a Chomsky quote on public affairs: ‘Commentary on public affairs in the mainstream literature is often shallow and uninformed. Everyone who writes and speaks about these matters knows how much you can get away with as long as you keep close to received doctrine. I'm sure just about everyone exploits these privileges. I know I do.’ Replace ‘public affairs’ with ‘economics’ and you have our current plight.

Twitter Feed

RT @SMerler: This, by @NickMalkoutzis and @YiannisMouzakis is a must read on Greece. Myths vs reality.