Against laws & "facts"

In a comment here, Unlearning Economics says that there are some people “who consistently beat the market, which contradicts the efficient markets hypothesis.” I agree. But this is not a problem for the EMH so much as for all economics.

I say this because there are, as far as I know, no 100% true laws in economics*. Economics is instead an inexact science in which any “law” has exceptions: as Jon Elster has stressed, the social sciences are about discovering mechanisms, not law-like generalizations. Even the “law” that demand curves slope down has some exceptions.

Rather than ask, “is this theory 100% true?” – a question to which the answer is usually “no” – we must ask: in what contexts is this true? And: how true is it?

As I say, the EMH is false in the context that there are some strategies, such as momentum and defensives, which do beat the market. But it is true – or at least true enough to be useful – in the context of choosing between actively and passively managed funds. 

This question is a purely empirical matter, to be settled by a careful consideration of the evidence. For example, the question of whether minimum wage laws represent an exception to whether demand curves slope down is a question of fact.

I don’t say this to get mainstream economics off the hook by lowering the burden of proof. What I say applies equally to heterodox theories. For example, there are flaws with the labour theory of value. Whether these flaws are great enough to offset its strengths in other contexts is, however, a much trickier question. To claim that the flaws “refute the LTV is the same sort of simple-minded dogmatism as claiming that the defensive and momentum anomalies “refute” the EMH. It misses the point that all theories can be “refuted” by some counter-examples.

The converse is also true. One of my beefs with right-wingers is their tendency to exaggerate the scope of their theories and to turn small truths into big ones. They fail to ask: how true is this?  For example, Tim Worstall might be right that some regulations are bad for the poor. But whether this means a smaller state would increase equality is a trickier matter. Similarly, the claim that immigration has reduced the wages of the low-paid is true, but only slightly so. And, I suspect, trying to defend the 1% by invoking marginal product theory is silly: it’s an attempt to apply a theory in a context where it doesn’t fit.

In fact, it’s a form of laziness. If you think a few simple ideas – Marxism, Econ 101, whatever – can explain everything, you are saving yourself the much tougher job of discovering and questioning the evidence.

But here’s a twist. Not only is the mindless application of general theoretical laws lazy, so too can be an appeal to “facts.” For example, Oxfam’s claim that “62 people own as much as the poorest half of the world's population” looks like a slamdunk case against equality. But it’s not. There must be something odd about a statistic which says that an Indian beggar with one rupee is richer than someone who’s just started working at Goldman Sachs but has lots of student debt**.

Appealing to “facts” in this case is a way of ducking tougher questions: is inequality a problem? Can we do anything useful about it?

I think the answers to these questions is “yes”. But they are hard ones, to which a definitive answer might never be available.

My point here is that many people - critics of mainstream economics, Econ 101ers and some lefties – are guilty of a form of autistic category error. They’re looking to economics for irrefutable big facts and theories which the discipline, by the very nature of our complex social world, cannot provide. Instead, we are in a domain of partially applicable theories and ambiguous or missing evidence. The question is how we deal with this.

 * “There are no lawlike generalizations” is itself a lawlike generalization. Any idiot can play pointless logic games.

 ** The point is that human capital must be wealth as well as financial assets.

Comments

[Pablo Mira]

On EMH, the issue is NOT if you can or can not beat the market. The real danger with that theory is that not beating the market is interpreted as an efficient equilibrium. Even if you are unable to beat the market, you can be part of an awful bubble.

[Matt Moore]

What is the correct response to this state of affairs? To me, the most important implication is that it raises the burden of proof that should be placed on social engineers. If (as I do) you require justification for the use of state power, whether taxation, regulation or whatever, then a highly complex world means the strength of evidence for the benefits needs to be high.

As Hayek said 'the curious task of economic is to teach men how little they really know about what they imagine they can design'.

(PS if you had a different ideology, you might suggest the burden of proof was on anyone who wanted to change the status quo. That's the Burkian position)

[chris]

@ Pablo - I take your point. I usually distinguish between the EMH (which I take to be a description of the cross-section of returns) and the rational market hypothesis, a theory about the overall market. You can believe the EMH whilst rejecting the RMH: Shiller and Samuelson did so:
http://papers.ssrn.com/sol3/papers.cfm?abstract_id=348180

[Stephen]

Re the grad at Goldman Sachs and the Indian with one rupee, I note you have moved from owning to being rich.

Rich is not a well-defined term, but Oxfam's claim is about owing. And the distribution of ownership is concerning, whether or not owning is the same as being rich or having wealth.

[Jason Smith]

Maybe a better way to interpret the EMH is as an equilibrium state where previous conditions are maximally uninformative (not necessarily *completely* uninformative). You can have momentum in cases like that. Instead of the "price state" being described by simply p itself (the EMH says the current price is the best information you have), it is described by a statistical distribution with a drift rate μ and variance σ (but only as long as you are in equilibrium).

When you are not in a market equilibrium (bubbles, recessions, etc), it doesn't apply.

The above is basically the statement that equilibrium is a maximum entropy state with some constraint (the momentum/drift and variance). A max ent state with bounds is a uniform distribution. A max ent state with a specified variance constraint is a normal distribution. A max ent state with a minimum is a Pareto distribution.

[Jason Smith]

Also: "They’re looking to economics for irrefutable big facts and theories which the discipline, by the very nature of our complex social world, cannot provide."

This is a methodological assumption. We have no proof that there do not exist 'big facts' -- not knowing any at present does not mean they do not exist. And there are ways in which extraordinarily complex behavior can be aggregated into simple macro behavior. I'm in the process of writing a talk about this here:

http://informationtransfereconomics.blogspot.com/2016/01/draft-paper-for-talk-this-summer.html

I would also suggest that after hundreds of years of looking into it and not finding any 'big facts' may be a sign we've been going about it all wrong :)

[UnlearningEcon]

I suppose the question now becomes: when can a theory be entirely abandoned or refuted? Do we constantly expand our library of theories, as Dani Rodrik seems to want? Are theories basically never refutable, which is something a paper called 'economic models as analogies' seemed to argue?

I'm not saying you're arguing this but it seems to be the way economics is heading. I've never seen a single theory completely abandoned, while research pushes the basic framework in all sorts of (sometimes contradictory) directions. Personally I find that this is clumsy and creates model selection problems, and am not sure how it is advancing our understanding of the world.

Anyway, I don't disagree with most of your post and I think we may be talking about slightly different things. So I'll stop rambling now.

[Marko]

" For example, Oxfam’s claim that “62 people own as much as the poorest half of the world's population” looks like a slamdunk case against equality. But it’s not. There must be something odd about a statistic which says that an Indian beggar with one rupee is richer than someone who’s just started working at Goldman Sachs but has lots of student debt**. "

The Oxfam claim says nothing that you can extend to comparing a broker and a beggar. The figures are the aggregated wealth of two populations. You could take the mean and say something about the "average" person in each group , but that's about it.

Still , it is a poor choice of groups to compare. The bottom 50% or smaller fractions of the global population will probably always have an aggregate wealth total that can be matched by what will appear to be a ridiculously small group at the top. Better to compare the bottom 90-95% to the tiny fraction at the top with matching wealth. The way things are going , we might get to the point where the bottom 98 or 99% would be a suitable comparator , and some time after that , one rich guy or gal will reach 50% of global wealth.

I don't think we'll ever mix human wealth with the current wealth data. It would make the stark differences between being born to poor vs rich parents too obvious. The one-percenters would never allow it.

[WimNusselder]

The alternative for seeking 'laws' and 'facts' in economics is indeed taking the 'performativity thesis' as starting point.
Contrary to what you wrote 4 years ago (http://stumblingandmumbling.typepad.com/stumbling_and_mumbling/2012/02/non-performativity-in-economics.html) I think economics 'is' ALWAYS performative (if you want to see social 'reality' that way), even if economic 'laws' and 'facts' fail to convince people of their 'reality' and fail to create the conditions that 'prove' their 'reality'.
'Performativity' is itself a perspective, which can gain adherence or fall into oblivion, 'beat' Subject-Object-Metaphysics or replace it.

It helps me to link the 'performativity thesis' to Pirsig's 'Metaphysics of Quality' as described in his book "Lila" (1991), which was a huge improvement on his MoQ as described in his better known "Zen and the art of motorcycle maintenance".
Both Subject-Object-Metaphysics and Metaphysics of Quality, both the performativity thesis and the idea that science can objectify/reify 'truth', are (in Pirsig's terminology) 'stable patterns of intellectual quality' which can be graded (and are evolutionary developed) as being relatively more 'dynamic' (flexible, innovative) than the patterns that (historically) preceded them.

THE question is: will the performativity thesis prove to be more open to Dynamic Quality than verisimilitude theories.
In other words: will we shift to thinking about 'truth' as 'that which performs best in the public arena' & 'that which creates the social reality that we prefer' rather than 'truth' simply 'is' 'representation of Reality'?

[Metatone]

The RMH vs EMH issue is a great instance of the real problem, which I've noticed a lot of econ bloggers "circling the wagons" to not address.

That problem is, what are the "givens" for a typical economics discussion, esp. one relating to policy?

For too many economists, their working mental model includes RMH - it just isn't something that gets examined. Another great example is how "redistribution" is on the one hand assumed when discussing the benefits of freer trade, but then attacked when the details of redistribution are under discussion. Another might be the grab bag of assumptions about when and what kinds of inflation and deflation matter.

The most painful thing about this is that of course when questioned many economists will deny believing in RMH - but will continue to use it as a buried assumption in arguments. Few economists are explicit about their assumptions. Personally I'd link this to the way econ journals have taken a back seat to "working papers" as the currency of status - and of course a general under examination of assumptions in most econ modelling work.

[theOnlySanePersonOnPlanetEarth]

I think we should ask why the wealthy are so inefficient as human beings. For example, the average wage in the UK is between 25 and 30K. And most people are expected to live on this.

Yet I can't help noticing that some people need to be paid above £40k and often above £100k. And sometimes millions.

The economic question is, why are these people so inefficient, that they need double, treble the wealth etc etc than the average?

A further question, why does anyone on planet Earth ever need more than above the average? What is it about those who need more than everyone else that makes them so inefficient as human beings?

Call it the Human efficiency Index. Point out the inefficient humans.

[Metatone]

Here's another pointer on assumptions about separability from Diane Coyle. Once again, something isn't settled - but it becomes part of the regular assumptions framework - and has deep effects on the reality of how the pie is divided.

http://www.enlightenmenteconomics.com/blog/index.php/2016/01/sharing-the-economic-growth/

[Pablo Mira]

Thanks for the reference, Chris. You are making a good point, since rational market lovers often use those two terms as equivalent.

[Magnus]

There are also, as far as I know, no 100% true laws in physics.

[sam]

brilliant post. i've always admired your open-mindedness to economics and i think this is your best attempt yet at putting a fine point on the guiding philosophy behind your equanimity.

i think there are 'laws' and 'facts', but agree that it is misguided to think we deduce answers to most policy questions from them. the laws are seemingly tautological and vague and never uniquely answer the questions economists are burdened with answering. the big facts are koans like 'people respond to incentives' and 'things must add up' and 'choices reveal preferences.' i only think this is an important distinction to make because some non-economists do approach policy questions as if budget constraints don't exist, choice can be disregarded, and perverse incentives ignored.

[Kaleberg]

The EMH always sounded like a tautology to me, but it surely can't mean that no one can beat the market unless there is some hidden assumption that all rational economic entities are alike. Since individuals differ in age, needs and attitude, one would expect a statistical distribution of market results EMH or not.

It's the old 'how to be an amazing stock market tout in ten easy weeks'. Each week send out 1024 forecasts, each time half saying the market will go up and half saying it will go down. At the end of ten weeks there will be some recipients who think you have a clue.

Of course, statistical thinking is troublesome to the ruling class since it implies that so much is luck rather than 'merit'.

[Nanikore]

I think you are a bit patronising to the critics of mainstream economists. The criticism is that "its about the model, not the economy" suggest that they are in more agreement with you than you make out. Really its a criticism of excessive abstraction rather than generalisation. That is why we probably need fewer models, more people working like historians - you build up your understanding purely from the (quantitative and qualitative) evidence, like any other investigator, and what you don't know for the time being, you accept you don't know.

[reason]

Nanikore,
I'm a critic of much mainstream economics and I completely disagree with you. I think the models that have dominated economics (especially macro-economics) are going in the wrong direction, rather than that they are too abstract. The evidence seems to increasingly point towards aggregation of heterogeneous agents as a major source of error - the way to better models is more heterogeneity. Trying to look for single constant parameters in a world a constant demographic and other change is bound to fail. I think we need to think of economics as being much more like ecology (interaction of many diverse actors in perhaps changing proportions, allowing for resource limits and external disruption) than we have up to now. Many of the really important things (the actual process of production, innovation and firm and industry history for instance) are hardly studied at all, while markets and consumers have been perhaps wrongly almost the exclusive realm of focus.

[Bob]

The Labour theory of value is a load of bullshit.

The use value of a machine can be significantly greater than its exchange value. I've worked in factories using machinery that is decades old, but it is the particular combination of machinery and people that create something of greater value.

As Steve Keen has said many times, if Marx had applied is own logic he would have rejected the labour theory of value.

The transformation of value into prices and the tendency of profit to fall are therefore nullified in the same way that long run neutrality of money is nullified. They are logically inconsistent with the premises.

Labour and commodities are joint source of value. It's an emergent property of their particular configuration in an operation.

The Labour theory of value is a belief that people cling onto - just like Samuelson economics. Now the adherents appeal to 'eclecticism' to avoid their entire life's work ending up in the bin where it belongs.