Second-order preferences

A reader recently forwarded me a paper given by Australia Institute Director Clive Hamilton to a Productivity Commission retreat (Clive has kindly given me permission to post the paper). Here’s the gist of it:

The core of my argument draws on the distinction between first and second-order preferences. First-order preferences are the ones mainstream economics recognises, the ones that are revealed by our behaviour in the marketplace. Second-order preferences are the preferences we have about our own preferences. This has been explored by US economist David George in a series of papers and a book titled Preference Pollution (University of Michigan Press, 2001). 

George gives the example of his consumption of fast food. As he sits eating a greasy hamburger he is wishing that he was eating something healthier. His second-order preference is a preference for his food preferences. It is not just another preference but a deeper order of preference. One reason for his first-order preference prevailing is the circumstances under which we make decisions. Snap decisions are more likely to follow first-order preferences while more considered judgments are more likely to see us following our second-order preferences.

My general view is that this is an interesting theory, but because it’s entirely non-tractable, it’s difficult to see how we should allow it to inform our work. Clive tells stories in which our second-order preferences are more altruistic than our first-order preferences. In his case, I’m sure that’s true. But equally, I could tell stories in which people’s second-order preferences are more selfish than their first-order preferences. If second-order preferences are never revealed, the two theories are indistinguishable.

This entry was posted in Economics Generally. Bookmark the permalink.

12 Responses to Second-order preferences

  1. derrida derider says:

    It’s not that it is non-tractable that’s the biggest problem, its that its non-falsifiable. Because they are not revealed, we are free to posit any “deep preferences” that suits us (or, more precisely, that suits our preferences).

    Not just a feeble doctrine, a sinister one to boot.

  2. harry clarke says:

    I’ve seen the idea used in the economics of addiction. I think the second-order preferences are seen to reflect desires of the pre-frontal cortex (rational, caculating) while the first-order preferences are those of the mid-brain (emotional, impulsive). So consistent with dual brain models of neuroeconomics.

    These two sets of preferences fight it out. If the first-order preferences win you can explain hyperbolic discounting.

    I am unsure DD if not refutable. You might argue that non-impulsive people will not be consistent with these preferences.

  3. Ben says:


    Are you the last believer in the fully-informed rational human being acting in his own self interest? It’s nice for economic modelling but not a true and honest way of describing human nature/the world.

  4. Russell Hamilton says:

    “because it’s entirely non-tractable, it’s difficult to see how we should allow it to inform our work.”

    Well, awareness of it can help save economists from simplification. Take the referendum on late night shopping in WA (so much enjoyment to be had from this). Economists would tell us that because the market revealed that where late night shopping was introduced it was popular – the consumers’ preference, then the citizens in WA would no doubt want it too. Trials of it were popular. In the pockets where it does exist in WA it’s popular.

    And they do want it – so convenient. But the referendum questions were, unusually, aimed at those second preferences – how do you prefer shopping to be part of community life – and a surprising result obtained.

  5. Geoff R says:

    Good example Russell. People have been arguing about this for 1000s of years. Sometimes economists do forget that other disciplines have often been there before. But from a economic perspective alone see the ANU book Rationality, Indivdualism & Public Policy.

  6. Christine says:

    Could it somehow be identified in people restricting their opportunities to choose at some later point? Thus – I want to exercise rather than read blogs (really truly I do). But I won’t do it unless I have some commitment mechanism. Thus I ask a friend if we can set up a regular gym-going appointment. (Or at least I could! And maybe I will! So there.) Presumably there are more compelling stories re drug or alcohol addiction.

    But the way preferences are typically defined (revealed) will make it impossible to identify these types of situations as first/second order preferences. But I think it doesn’t really handle situations where someone voluntarily constrains their choices particularly well.

  7. derrida derider says:

    Let me put it another way. I think public policy should be based on evidence. And as Andrew pointed out, in the absence of revealed second-order preferences we can have no evidence that they exist. Assumptions about preferences which are not revealed are inherently unfalsifiable.

    The danger is that there is a rich history of assuming such preferences in others, and that these assumed preferences have usually coincided with what the assumer would like to believe they are. It rapidly leads to the position of “I know what’s good for the people better than they do” – a motif underlying the more miserable parts of 20th C history.

    None of this means that the behavioural economics stuff is all invalid – quite the contrary in fact. But before you assume, say, that people deep down really, really want to save for their old age but are unable to (and therefore require coercion “for their own good” from their friendly Government) I want more than just some theory about how this could happen. I want the hard evidence from their revealed behaviour.

  8. Ben says:


    Aren’t second order preferences “revealed” by people’s responses to questioning (such as in Russell’s referendum example above) ?

    Just because something is difficult to measure does not mean it does not exist.

    Are we to judge people purely on their behaviour and not their expressed wishes and desires. For example do we assume obese people are acting rationally in buying unhealthy food? Are they maximising their utility? Or do we allow that they might be accepting of us as a society (government!!) to consider making healthy choices easier (eg. a fat tax)?

    I certainly agree that evidence is critical to public policy decisions. But we have to ask the right questions. And where evidence is laking or ambiguous we still have to make judgement calls. The pure sciences have moved on from a “clockwork” view of the universe and the study of economics will slowly make this transition too.

  9. Christine says:

    DD: very good point on possible abuses of the notion of second order preferences, and the need for some actual evidence on them. Saving for retirement is an interesting issue, with an increasing amount of evidence that, for instance, the status quo matters. I gather that if you have a default amount taken out of your pay for super but allow people to change if they like, people will nonetheless tend to stick pretty close to the default amount. That’s not particularly evidence for second order preferences, as such, but it does provide evidence that a simple revealed preference approach might miss some things.

    That said, one reason for government coercion for people to save for retirement is not that it’s for their own good, but that given the public pension system its in society’s interests that they save something themselves to reduce public spending on pensions.

  10. Russell Hamilton says:

    “That said, one reason for government coercion for people to save for retirement is not that it’s for their own good, but that given the public pension system its in society’s interests that they save something themselves to reduce public spending on pensions.”

    So, if you have children, some of your income should be garnished and put in a fund to pay for your children’s education – in order to reduce public spending on education? How much is it in society’s interest that everybody looks after themselves individually ?

  11. Christine says:

    Russell, I didn’t say it’s a good reason, just a reason – public pensions mean less incentive for you to save for your own retirement even if you can, meaning that government thinks they need to force you to do it so you don’t put too much pressure on the pension system. In response to your point on education, no I woudn’t agree with your statement. Parents already bear most of the costs of raising kids, and there are externality benefits to education for society as a whole that don’t apply with the same force to retirement saving. But yes, I would say that it’s in society’s and individuals’ interests to have individuals look after themselves and their kids to a fairly large extent. Not exclusively, but, and it wasn’t particularly the point I was trying to make.

    Mostly I was just trying to suggest that (a) there might be some evidence of these ‘second order preferences’ in retirement saving, which could justify some types of government policies that look like they’re trying to override ‘first order preferences’; and (b) the reasons for forced retirement saving are not limited to the argument that the government knows best what your true second order preferences are anyway, so it might not be the best example. (Reading it like this, though, it sounds confused – probably because I really am being very speculative here.)

  12. Russell Hamilton says:

    “I really am being very speculative here” – you say that like it could be a bad thing …….

    Sorry for the education example – I knew it was crummy as I typed it, just too lazy to think of a better one. I agree with what you say; I was just pleased to hear of this secondary preferences thing because economists seem to think of economics as science, rather than social science, and data and ‘revealed preferences’ as allowing for easy conclusions – when in fact a bit more speculation and doubt would be a good thing.

Comments are closed.