Behavioural economics and the Lucas critique

There's been a few recent posts on the Lucas critique in the economics blogosphere (ex: Noahpinion), so I'm going to take advantage of this to make a point that I don't recall seeing made elsewhere. (If I'm re-inventing the wheel here, please let me know in the comments and I'll add the appopriate attribution.)

A key insight of behavioural economics is that people don't always and everywhere re-optimise whenever their environments change. Instead, they will often - or even usually - make use of various rules of thumb and/or passively accept the default option. The costs of re-optimising every time you face something new don't always offset the benefits from making what may be only a slightly better choice. This the idea behind 'nudges': you can alter people's behaviour by making minor changes to the frames in which people operate: if people have the habit of choosing the default option, then you can change choices by changing the default option.

But this only works if the change is subtle enough to not attact the full, direct attention of the decision-maker. If the change is big enough, people will haul out the full artillery of their rational selves in order to try figure out what optimal decision is. This means that behavioural economics is unlikely to be of much use in policy-making.

This is essentially the point of the Lucas Critique. Behaviourial economics' prescription of nudges work precisely because peoples' attention are not called to them, or how these measures are supposed to work. But if you announce a measure to change people's environments in order to attain a stated policy goal, peoples' attention will (presumably) focus on the problem at hand and they will re-optimise. This reaction - the focus of attention and recalibration in the new environment - is exactly what nudge prescriptions are designed to avoid.

Okay, you may say, that may be a problem for people like you and me, but most people won't be paying attention to the policy discussion in the first place, so the nudge story still goes through. But I'm not sure how much I'm willing to get behind a policy whose effectiveness depends on citizens' not knowing what their government is doing or why.

Comments

[Kevin]

" if the change is subtle enough to not attact the full, direct attention of the decision-maker"

If people can be 'tricked' about such items as their retirement savings (which will have a huge impact on their future wellbeing) or eating / exercise habits (which again will have a large impact on life outcomes), then I don't think it is a stretch to think that a guy can be 'tricked' when thinking about his or her wage offers and how they relate to inflation.

So, my counterclaim is that people seem to make big repeated errors on things that are much more important to their lives than 2 vs 3 percent inflation or other issues where the Lucas Critique is invoked.

[Stephen Gordon]

Sure, but if, say, due to a lack of information, people are making big repeated errors, wouldn't the policy response be to make a really big deal to explain what is going on and why? That seems like the opposite of 'nudging'.

[Roger Magi]

This is if rational agents can follow what is going on and predict the most probable outcome. I think that the first part is probable, but if agents would consider the economy as a black box (no consistent expectation) even if they focus on policy the Lucas critique cannot be relevant. This second condition is unlikely to hold, first expectations are model dependent, and second even with homogeneous modelling (DSGE) the predictions are radically different.
At least we should say that the Lucas critique holds only under consensus(or at least as aggregate belief) and simple prediction.

[Kevin Andrew]

Here is how I sometimes think of rationality in economics:

I often make irrational decisions about small things. However, usually when making a decision about something big I try and do the best with the information I have. As well, a lot of big decisions are made by experts or other people that do make optimizing decisions based on the best available information. In aggregate, the big decisions by experts or that individuals think about dominate.

However, sometimes we will see what appears to be irrational behaviour. There are two things we could do with this: either accept that people are irrational and that we can't learn anything or try and model why the decisions may actually be a rational response to some perceived information or constraints.

I don't deny that a lot of irrational decisions are made in life! However, as economists I think that we would be better served trying to understand what appears to be irrational behaviour better as opposed to just giving up and claiming that people are irrational. As well, when looking at the aggregate economy there are a lot of instances where rationality isn't too bad an assumption as the key decision makers behave this way.

[rsj]

But this only works if the change is subtle enough to not attact the full, direct attention of the decision-maker. If the change is big enough, people will haul out the full artillery of their rational selves in order to try figure out what optimal decision is. This means that behavioural economics is unlikely to be of much use in policy-making.

Woah Nelly. The core criticism here is that fully optimizing requires infinite cost. It is a non-polynominal problem. Therefore it is always impossible to "fully" optimize, and households will always need to use rules of thumb. It's not that they sometimes use rules of thumb and sometimes solve the Hamiltonian problem fully.

Because the Hamiltonian problem requires infinite costs to solve, it is never solved.

And the real criticism here is of knife-edge solutions that require the hamiltonian be fully solved, and an assumption that "almost" optimizing gives you a solution close to the fully optimizing solution.

That requires a convergence result which you don't have.

In fact, whenever a research paper is published that shows new effects as a result of tweaking a model, that is proof that solving an approximation to non-linear model is not the same thing as approximately solving the original model.

Without such a convergence result, there is not much point in even bothering to come up with a model. Every model is an approximation of a more complicated model.

Therefore models which assume that households are optimizing are unlikely to be useful for policy guidance.

[rsj]

On the other hand, if you could come up with a convergence result -- say define a tower of models of increasing complexity -- and prove that the solution to the less complex model is a good a approximation to the solution of the more complex model along each step -- then that would be really cool! I'm not sure if it would be very useful, but it would certainly be cool.

Another nice convergence result would be to partially solve a hamiltonian problem, and show that if everyone partially solves it to N steps, then in the limit as N -> infinty, the partial solutions converge to the full solution.

If you could combine both of the above, then the result could be claimed to be robust. I'm not aware of a single robust model -- but perhaps some of the modeling pros on here could point out an example.

[Sandwichman]

"I'm not sure how much I'm willing to get behind a policy whose effectiveness depends on citizens' not knowing what their government is doing or why."

And how would that be any different from what we currently have?

[marris]

> In fact, whenever a research paper is published that shows new effects as a result of tweaking a model, that is proof that solving an approximation to non-linear model is not the same thing as approximately solving the original model.

I've read this five times and still don't understand what it means.

What is the NP hard problem that needs to be solved? The Hamiltonian of what?

[Shangwen]

Is there some well-known central tendency in consistently inefficient individuals, upon which we can design policy with certainty? I know of many experiments in health care delivery where people have attempted subtle changes to decrease inefficient utilization--increasing a co-pay by $1, relocating a clinic to a slightly more inconvenient site--and obtained dramatic but temporary effects. In our hospital we tried to speed up post-surg recovery times using incentives to participate in pre-surg weight-loss and exercise programs; it worked for a small segment of motivated people, but after a few months the effect died out. People want hands-on time with a physio after the OR, not a sheet with exercise instructions that just sits on their desk at home before the surgery. The people whose behavior we really hoped to change didn't budge. Now, who is the fool in this context?

At a really applied level, the hubris in a lot of nudge policies is this: all people have inefficient behaviors, which troubles those who study economics (or public admin). We can't order them to choose efficiently, but we wish they would. So we will put on our magic Behavioral Econ Overlordship hats, and implant the illusion of free will into people's heads. And, in manipulating these people we have acknowledged are unreasonable, we can somehow predict what they will then do. Maybe this works with pension contribution defaults, and chocolate bars hidden way down the lunch counter line, but those are unnaturally linear choice contexts. The big problems of the day are not like that.

[Lord]

"If the change is big enough, people will haul out the full artillery of their rational selves in order to try figure out what optimal decision is. "

Why would that be considered a failure? If they don't think about it they get a reasonable default and if they do they get what they consider best. It works either way, and likely better than not thinking about a poor default.

[Neil]

This critique underestimates the cost of change. The big "nudge" that policy makers are talking about relates to retirement savings. A large number of people realise they should be saving for retirement, but put off signing up for employer match plans because it requires wading through piles of literature, at very high cost. If enrollment in a conservative investment, then avoiding dealing with the issue still works just fine, even if you're well aware of the policy change. It's not that you didn't want to enrol, it's that you didn't want to think about it.

Nudges don't require ignorance or lack of understanding to work. They just have to eliminate barriers between the default option and the policy goal option.

[Neil]

Argghh. Have to stop rephrasing 10 times before posting. What I was trying to say was

This critique underestimates the cost of change. The big "nudge" that policy makers are talking about relates to retirement savings. A large number of people realise they should be saving for retirement, but put off signing up for employer match plans because it requires wading through piles of literature, perceived to be a very high cost, at least relative to a benefit in the distant future. People are known to put a surprisingly low value on the future. If enrollment in a conservative investment is the default, then avoiding dealing with the issue still works acceptably, even if you're well aware of the policy change. It's not that you didn't want to enrol, it's that you didn't want to think about it.

Nudges don't require ignorance or lack of understanding to work. They just have to eliminate barriers between the default option and the policy goal option.

[Daphne Millar]

Lots of things wrong with this but above all with "I'm not sure how much I'm willing to get behind a policy whose effectiveness depends on citizens' not knowing what their government is doing or why."
So that means you are prepared to get behind a policy which assumes people do know what the government is doing and why. You should get out more.

[TJ]

"Nudging" is more a question of re-framing the choice you give them.
If you have the best policies available but they take a lot of effort to use as an individual, chances are your great policies won't be effective and it will appear as though no one _wants_ your great policies, and you scratch it down as a perverse preference of some sort, a bad policy, etc.

>If the change is big enough, people will haul out the full artillery of their rational selves in order to try figure out what optimal decision is.

In most cases, even when faced with very large changes, people will do their best to not have to pull out their full artillery. Assuming that decisions on Healthcare, mortgages and pensions are big changes that will have a very large impact on your life, people still consistently make uneducated (choosing the default or blindly accepting recommendations) choices on those topics. I have seen people put more research into what kind of grill to get than what kind of pension they have.
If you have a sub-optimal default option and people consistently choose the default option over optimizing, changing the default option for the better wouldn't make everyone re-optimize from the default, removing the default option and forcing people to choose may make people optimize.

>But if you announce a measure to change people's environments in order to attain a stated policy goal,...

Let's use the infamous supermarket example. If a policy is announced to everyone that the candy is moved away from check-out counter to make people buy it less - will people re-optimize by buying more candy when they're in the candy aisle? Or purposely going there to pick up that candy bar they kind of want but know they shouldn't get? Or let their kids roam further away from them when they are waiting in the queue?
One of the core tenets put out for the nudges is that they shouldn't be hidden or done on the sly, but that you should be able to say "We have structured this choice in this way, because of this" to everyone, including policy-users.

Disregarding "Nudge", you have a whole host of cognitive biases, hyperbolic discounting and loss aversion being the most familiar, that influence economic decision making, including reactions to policy changes, this stuff just very quickly becomes very iffy to model, particularly in a way that can be applied as a micro-foundation.

[UnlearningEcon]

"But this only works if the change is subtle enough to not attact the full, direct attention of the decision-maker."

Actually many biases are surprisingly resilient - for example, it is well known placebos still work even when the patient is told they are receiving a placebo. Furthermore, a policy that incorporates a bias does not necessarily involve referencing the bias itself through a bullhorn.

"But I'm not sure how much I'm willing to get behind a policy whose effectiveness depends on citizens' not knowing what their government is doing or why."

Most people do not know this. Designing policy around how people actually behave is no different ethically to designing it as if they are rational utility maximisers (actually it's better, because it's based on the real world).

[Andreas Maaløe Jespersen]

"If the change is big enough, people will haul out the full artillery of their rational selves in order to try figure out what optimal decision is. "

This is a nudge in itself. If we as a species are misers when it comes to applying our cognitive abilities to their fullest, then a policy maker can (and should) figure out when these abilities are most likely to help us. And don´t forget, even when we "pull out the full artillery" we are still suggestible to biases like loss aversion and future discounting.

[Mike Smitka, Washington and Lee University]

I've not looked at GEQ in a long time, but unless something has changed, there remains no convergence or approximation solution. [Is someone up to date on this?]

If so, then this not only implies that (dumb luck aside) macro models are very dependent on their particular structures (and this is news?) but that micro models also remain, at some level, acts of faith: without GEQ there's no reason to think that most prices are (approximately) "right" in conveying ... well, you pick!

At the same time, unless we treat it simply as a game, we engage in economic analysis because we believe that, despite theory, prices somehow reflect social costs and benefits. We can reassure ourselves that markets in general seem quite stable (whereas the lack of robust GEQ provides no reason to think that small impulses result in small changes).

Using very simple models to "get at" very complex social behavior remains an art, informed by empirical evidence (when done honestly). It's strange that economists could think otherwise -- not that those in the "hard" sciences are any more enamored of the philosophy of science and deep issues of methodology. Grad school socializes us into particular habits (dare I say to employ standard rules of thumb?) on how to behave. We do respond to "prices", what is easy to publish, what sub-disciplines look likely to generate a job offer. But even with that, in grad school did we rationally pick our sub-fields, or did a series of random, cascading events land us somewhere? For most of my classmates, I think the latter dominated; certainly it did for me.

[whitfit]

Stephen writes: "But if you announce a measure to change people's environments in order to attain a stated policy goal, peoples' attention will (presumably) focus on the problem at hand and they will re-optimise. This reaction - the focus of attention and recalibration in the new environment - is exactly what nudge prescriptions are designed to avoid."

I thought that nudge policies were designed to make the optimal decision the default, and allow people to do something else if they want. When the non-optimal decision is the default, people often don't optimise, and don't make the effort required to switch to an optimal path, even where that path is available to them. Why does that imply that if people are placed on the (generally) optimal path by default, when individuals run the optimization, they will switch to another path?

Stephen also says: "Okay, you may say, that may be a problem for people like you and me, but most people won't be paying attention to the policy discussion in the first place, so the nudge story still goes through. But I'm not sure how much I'm willing to get behind a policy whose effectiveness depends on citizens' not knowing what their government is doing or why."

I don't see why the nudge depends on not being publicized, and if the decision that the nudge policy creates is optimal for most people, as discussed above, there is no reason to think that people won't be happy to maintain that path, even while knowing that it is a nudge. I don't think the policy is dependent on people not knowing what their government is doing or why, but it is to nudge people in a direction that is better where they aren't the type of people that sit down and carefully optimize things that may have significant future effects. I agree that a policy that is only effective if citizens are ignorant of their government's decision is problematic - but I also am willing to believe that nudge policies could be effective without such ignorance.

[Ryan]

First, explain to me why we want to "nudge" people, other than that we are self-important would-be dictators who think we know better than other people.

Then, explain to me how gods like us know which of the people we're nudging have re-optimized, and which have not.

[rsj]

First, explain to me why we want to "nudge" people

We want to nudge people in order to make better outcomes easy, and worse outcomes more difficult. For example, we can make it harder to buy cigarettes, or restrict the ability to smoke in many places. We make it more difficult to purchase guns, and require waiting periods. We make retirement plans a default option, and removing money from the plan costly.

other than that we are self-important would-be dictators who think we know better than other people. Well, yes, there are professionals who know more. Even a disiniterested person is going to make a more rational choice, and even the same person will exhibit time inconsistency, choosing the "right" thing for their future selves to-do tomorrow, but often choosing the wrong thing today. I'm going to quit smoking next month, too, and next year I will start saving for retirement.

Then, explain to me how gods like us know which of the people we're nudging have re-optimized, and which have not. You don't care *which* person re-optimizes. You just look at aggregate smoking rates or aggregate levels of senior poverty and see if they are rising or falling.

[Ryan]

We want to nudge people in order to make better outcomes easy, and worse outcomes more difficult.

You mean like when people think they want to own cigarettes, guns, and asset liquidity, but really they don't?

Well, yes, there are professionals who know more.

There are professionals who know more about my preferences than I do?

Even a disiniterested person is going to make a more rational choice, and even the same person will exhibit time inconsistency, choosing the "right" thing for their future selves to-do tomorrow, but often choosing the wrong thing today.

So because people value the same thing differently in different at different points in time means they act irrationally?

You just look at aggregate smoking rates or aggregate levels of senior poverty and see if they are rising or falling.

*ahem* What I meant was, (gasp!) what if smoking is a perfectly optimal behavior for someone who enjoys it?

[rsj]

You mean like when people think they want to own cigarettes, guns, and asset liquidity, but really they don't?

Yup. In fact, I make it a point to talk to people about finances, and you would be shocked at their ignorance. There are several people who have retirement accounts set up now as a result of me just telling them that this was an option available to them and (more importantly) being willing to walk them through the steps of setting one up. It was mostly hand holding. But if this was a default option then they would take it.

You seem to believe that every possible aspect -- the literally billions of choices available to us at any moment -- are the result of deliberate optimizing behavior and trillions of hours of research.

Allow me to dissuade you of this belief. There is no such thing as individual preferences in a vacuum. Individual preferences only exist in the context of a larger society that shapes those preferences.. People's behavior has nothing to do about optimization in some sort of vacuum. They are socialized. There is a reason why if someone grows up in a community of doctors, they are more likely to become a doctor. That is a possibility that was made available to them as a result of their environment. Without those role models, few would realize which career choices are available to them and which are not.

People have to learn what their choices are. People even have to be taught how to optimize, and no one does it well. Optimization does not happen automatically, or for free, and it rarely happens well. I often talk to small businessmen, and most of those in retail choose keystone pricing (they double their wholesale price). Why do they price this way? Because that is how they were taught to do it. Someone has to teach them how to set prices, how to run a business. A convenience store with hundreds of products and is not going to calculate some estimated demand curve for each product. It is too costly to gather that information, and much easier for the owner to just double the price.

Similarly, a culture of entrepreneurship does not spring out of the air -- the German Princes had to be "led by the nose to their profits" by the emperor, because they didn't know how to be entrepreneurs. Obviously once they were receiving profits, many realized that this is what they wanted, but the fact remains that the German industrial culture was the result of state policy, policy to fund trade schools, to create pools of craftsmen, as was the resulting set of preferences that arose from that culture.

So there is an interplay in the preferences spring out of the larger society, which in turn is also shaped by preferences. Both interact with each other.

You seem to believe that there is some individualistic "pure" self which has its own preferences that either cannot or should not be affected by institutions outside of it.

That is a psychotic, infantile view of human nature.

[bamboo investments]

It seems to me that "nudging" people is simply creating market-based incentives. Its better than NYC where Bloomberg just banned certain kinds of soft drinks.

[K]

Stephen: "But I'm not sure how much I'm willing to get behind a policy whose effectiveness depends on citizens' not knowing what their government is doing or why."

Maybe it doesn't. Maybe some people actually expect their governments to be looking out for their interests, and therefore are inclined to accept the judgement of government experts. We defer to experts on medicine, auto repair, whatever, without, for the most part, ever finding out whether the services provided were valuable or not. Any idea what your kids' teeth would have looked like *without* those orthodontics? There is simply no basis on which to form a rational choice. Lots of people would probably prefer to defer to government experts on money issues like savings and investment. Not crazy.

Presumably, if the choice is left up to the individual, it's because there are several valid options. If someone can be "nudged" just by changing the default value, they presumably didn't have sufficiently information to make a rational choice. In which case, trying to provide a sensible default is the least we can do for them. Other, expertly computing intertemporal optimizers, will check the box that best suits their preferences. If a significant number of people are influenced by the default option in matters as significant as how much money to put aside for retirement, that is an astonishing indictment of the Lucas Critique.

[Andrew F]

Wasn't Bloomberg's soda ban just a big shove? A $200 fine for selling a soda larger than a certain size. A de facto ban, perhaps.

[dwb]

i recently read The Hour Between Dog and Wolf, written by a hedge fund trader turned neuroscience. I definitely think there is feedback between brain and body, particularly in the area of risk taking. However, i find it hard to extrapolate this to say that biology can be responsible for coordinating a "bubble" in asset prices between disparate trading floors in NY, San Fransisco, London, Sydney, Hong Kong...

I think its more useful to understand these phenomenon to understand how to design incentives and compensation schemes to direct desired behavior. In the case of trading, clearly biology and poor heads-i-win-tails-you-lose compensation is like a double barrel to the financial markets, at times.

[Nathan W]

Well if you're going to nudge someone you may as well have a default option that is optimal in some way, but perhaps the default option has distributional or other effects that vary by type of individual. The extent to which these differentiated results will affect the nature and magnitude of feedbacks within the economy is a tough one though. If it causes different groups to consume or invest more or less effort and/or resources, then their use and/or leverage of their specific endowments/abilities and opportunities can also be expected to shift in response to the new environment (here, more thinking of the change in the default option than in the aggregate economic situation).

Say, you change the default option from "contribute minimum CPP" to "contribute more than minimum CPP, with minimum payments requiring a single simple extra form" then for reasons that are not worth discussing now, I assume that many lower and middle income groups will save/invest more and that higher income groups will tend to save/invest a similar amount. If the assumption is correct, then should we assume that the lower and middle income classes are necessarily better off? Within those groups are there further differences in how people will reallocate spending, and if so does that matter?

The relevance of the example is not the specifics of a model, but just the principle that differentiated impacts across groups will affect the impacts on the economy and the welfare of individuals … and, I suppose, that it's interesting to think about this when taking a more deliberative approach to what constitutes optimality, depending on social values among the population, or relatedly, depending on the choice classifications for the purpose of disaggregation. Here, I think we could expect lots of disagreement about the best way to classify things, and which type of optimality should be sought under which type of social valuation. I.e., if it impacts inequality or has sector-specific impacts, this would be good to know.

In any case, I especially like the final sentence, but in this case I think it's not too big of a deal because I assume we’re talking about cases where a) people still have the option, hopefully presented to them and b) there is at least a minimum of transparency, so that if it's a big enough deal for enough people someone will eventually catch on a make it a big deal.

[Ryan]

rsj - I'd love to debate this with you further, but not only do you know more about my preferences than I do, but you also know whether or not I'm psychotic. Clearly, you are the authority here.

[rsj]

Ryan,

There is a difference between a psychotic view and a psychotic person.

The fact that I need to point this out is telling.

The belief that individuals are completely isolated from society so that their own preferences are not shaped by the institutions around them is the view that I was criticizing. The arrow of influence goes in both directions:

1) individuals preferences --> society/institutions
2) society/institutions --> individual's preferences

When someone says that the government (which is a social institution) should not shape individual preferences or the resulting behavior, then they are asserting that 1) is legitimate, but 2) is illegitimate.

And yes, this is an infantile, psychotic view of the relationship between Man and Society. We are social animals.

[Ryan]

The fact that I need to point this out is telling.

Is this the kind of statement you expect to foster an open, productive debate?

Your argument is not just weak, but horrendously so. Let me see if I can count the number of fallacies you have committed:

1) Strawman: I never claimed that individuals don't shape society or that society doesn't shape individuals.
2) Ad Hominem: Implying that I adhere to psychotic beliefs.
3) Genetic Fallacy: Implying that psychotic people's arguments are incorrect by virtue of the fact that they are being made by psychotic people.
4) Bandwagon: Claiming that our social nature implies that any institution that professes to act in the social interest is worthy of its cause.
5) Appeal to Authority: "Well, yes, there are professionals who know more."
6) Ambiguity: Blurring the line between "society" and "institutions" by using the phrase "society/institutions"
7) Loaded Question: Presenting a two-directional model involving the above ambiguity.
8) Begging the Question: Assuming that, because institutions "influence" individuals, there is sufficient grounds for deliberately distorting markets in order to manipulate market preferences.
9) Appeal to Nature: Suggesting that because men are "social animals," your opinion is correct because it happens to involve an acceptance of "society."

I thought for sure you'd have more than ten, but I guess nine fallacies are still sufficient to discredit your point here. But don't mind me, I'm just a psychopath.

[Frank Dean]

Admirers of the Lucas Critique generally react indignantly when it is suggested that all the Critique amounts to is the observation that mis-specified models are unreliable, but it is one way of looking at it. In particular, in-sample performance of an economic model doesn't indicate much about out-of-sample performance.

The response has been micro-foundations, in the hope that by specifying behaviour at a lower level, the model would always be operating in-sample.

Behavioural economics presents convincing evidence that human behaviour deviates substantially from that of the rational-expectations-agents favoured by Lucas and his followers. So on the micro-level, these RE models are never operating in-sample, which suggests they will be unreliable. As far as I know, the evidence confirms this. (Perhaps this explains the insistence that the Lucas Critique is much deeper, important, and more subtle than simple mis-specification.)

Some are now suggesting that the Lucas Critique is irrelevant because if people follow rules-of-thumb, policy-makers in fact can rely on statistical relationships observed in aggregate data. (Presumably because the statistical relationships are outcomes derived from robust rules-of-thumb. Which sounds a lot like a claim that the behavioural relationships are in fact structural for all practical purposes.)

Stephen Gordon appears to be suggesting that the movement of rules-of-thumb from the macro-level to the micro-level will still leave models vulnerable to the Lucas Critique. This is because he doesn't believe the rules-of-thumb will be reliable guides to behaviour; when conditions change sufficiently, behaviour will deviate from the micro-behavioural rules-of-thumb. They are rules-of-thumb, so by definition they would seem to be behavioural rather than "structural" relationships. I suppose the point of disagreement concerns how robust the rules-of-thumb are.

Seems like a stalemate. Empirical results from behavioural economics show that people follow rules-of-thumb, which means that modelling their behaviour as if they were solving an optimization problem will not uncover the "deep structural parameters" required to satisfy the Lucas Critique. Yet perhaps some of the time people do "focus on the problem at hand and ... re-optimise". But presumably this doesn't happen often enough to generate the data required if researchers are to uncover the "deep structural parameters." We want models that can handle out-of-sample events, but we can't actually build them.

To demonstrate that "behavioural economics is unlikely to be of much use in policy-making", we would need to figure out what kind of change is "big enough" for people "to try figure out what optimal decision is."

Arguably this is what Lucas et al. did. Except the result has been that rather than believing in an illusionary unemployment/inflation trade-off, policy-makers now believe in an illusionary low-inflation/optimal-growth relationship. Which is weird, because these days a low-inflation/optimal-growth relationship can't be seen anywhere.

[Lorenzo from Oz]

Why would not rational being optimise on cognitive effort? Do we really think that cognition is not bounded by scarcity like everything else.

[rsj]

Why would not rational being optimise on cognitive effort? Do we really think that cognition is not bounded by scarcity like everything else.

LOL

How much effort should he spend, figuring out how much effort to spend?

[vimothy]

It's interesting how today the left and the nominally right divide up into two almost identical political philosophies, where one side believes that the social order can be transformed into a technical problem best solved by an aristocratic class of quote-unquote "experts" and the other believes that it can be transformed into a technical problem best solved by the limitless genius of the decentralised market. Exciting times.

[Lorenzo from Oz]

rsj by using cognitive inertia: don't pay attention unless one thinks you have to. It is what we all do; that is why we have habits, routines and rules of thumb, otherwise there is not enough time in the day. They are all cognitive rationing systems.

[W. Peden]

Vimothy,

Yes; all the more odd, given that the correct answer is probably that (iii) the problem of the social order is unsolvable, and there are no satisfying outcomes, such that the Enlightenment ambitions of both right and left are unachievable and the way to peace is a philosophical acceptance of human cognitive limitations.

I don't believe in cyclical history, but I do believe in a cyclical history of economic thought, at least since the Enlightenment: we are dissatisfied with the way things are; advocates of laissez-faire promise a better alternative; this alternative is largely implemented, but we are dissatisfied with the way things are and the reduction of power to abuse creates a fairly competant civil service and political class; statists promise a better alternative; this alternative is largely implemented; then the limitations of all-powerful state controllers becomes apparent and we become dissatisfied with the way things are; advocates of laissez-faire promise a better alternative; this alternative is largely implemented, but we are dissatisfied with the way things are and the reduction of power to abuse creates a fairly competant civil service and political class...

I think that the laissez-faire option is less dissatisfying, but I don't think it will ever provide sufficient contentment to end the cycle. I suspect that the cycle could only end with a global and sufficiently systematic/brutal statist movement in the manner of Stalinism/Fascism, which could write laissez-faire theory and history forever and impose a counter-Enlightenment.

Behavioural economics, like macroeconomics in the 1930s, is an interesting move of the cycle, and quite apart from the fact that it will never lead to lasting satisfaction with the social order, it's a fascinating and fruitful research paradigm.