Living on The Rock

On this day in 1938, Orwell was in Gibraltar and wrote:

Standard of living apparently not very low, no barefooted adults and few children. Fruit and vegetables cheap, wine and tobacco evidently untaxed or taxed very little (English cigarettes 3/- a hundred, Spanish 10d. a hundred), silk very cheap. No English sugar or matches, all Belgian. Cows’ milk 6d. a pint.

Jeez. In that one paragraph he beat Krugman to the “love of varieties” thing by about 40 years and he beat Broada and Romalis to the “price deflate using the basket of goods actually bought” thing by 70 years.

20 Responses to “Living on The Rock”

  • notsneaky says:

    Well, except Broada and Romalis are wrong – you can’t deflate using the basket of goods actually bought since baskets and prices are both endogenous. You gotta pick a basket, some basket and use it consistently for everyone.

    I guess you can do some cultural adjustment, which is what it looks like Orwell’s doing.

  • Gabriel says:

    Notsneaky, fine, I pick as my basket, the price of wet weather tires sets. That’s fixed and exogenous, right? :-)

  • notsneaky says:

    Broada and Romalis pretty much assume that the poor and the rich are different in some fundamental “they have a different utility functions” sense. This pretty much goes how Economics has been done since Bentham and is completely ad hoc.

    Obviously the poor consume the goods they consume because they are poor – their real incomes are low so they consume stuff with low prices.

    Yes, picking wet weather tire sets as a basket would make more sense then their approach.

  • pushmedia1 says:

    B&R’s results are contingent on the extent to which rich and poor are actually different types of people. What informs your prior that they’re not in fact different? Any prior is ad-hoc as we can’t observe people in both states of wealthiness.

    Doesn’t the behavior of lottery winners prove the rich are really different from the poor (or formerly poor)?

  • pushmedia1 says:

    That’s funny. B&R’s egalitarian results rely on a very inegalitarian assumption.

  • notsneaky says:

    Well, you’d start with the whole “equal capacity for pleasure” thing.

    My consumption basket used to give a heavy weight to Ramen noodles and cheap beer. Now my consumption basket gives a heavy weight to steaks at Appleby’s and well, slightly less cheap beer. Have my preferences changed? Am I fundamentally different than I was? Nope, just the constraints I face have changed. While there may be some income-independent, class-specific goods people consume (and go on consuming as their income changes (country music?)) there’s plenty of goods – most of them – that people only buy because they can’t afford to buy better ones. If you were to compare the ‘inequality’ between my old self and my present self it makes no sense to use the different baskets, as the old me would’ve very much liked to consume my present basket had he been able to do so. To assume that poor people want to consume fundamentally (i.e. not related to income and prices) different goods than rich people, well, that way lies madness. Or worse, sociology.

    Of course in a deeper way, since utility – even if one allows interpersonal comparisons of it – is ordinal and not cardinal it doesn’t make sense to talk about ‘inequality in utility’. Things which are not ordinal don’t have distributions which is what inequality measures. And even if you make that leap to cardinal utility, then you better be assuming identical utility functions.

  • notsneaky says:

    And how many rich people win lotteries – or even more accurately, how many of them win lotteries that increase their income by the same factor as when poorer people win them? I’m not sure what those lottery studies show except that when people get an unexpected huge windfall, sometimes they go a little crazy.

    As an aside – and I think some confusion arises from this – it’s perfectly legitimate to talk about welfare gains or whatever to poor people when the prices of the “poor people goods” goes down. What’s not kosher is using different sets of goods to compare “inequality” between the rich and poor. The number of wet weather tires a given person can buy has a distribution. Utility doesn’t.

  • pushmedia1 says:

    I can’t decide if you’re saying we shouldn’t study inequality or if you’re saying its hard/impossible.

    Why do you have such a strong prior that utility functions are structural? Why can’t preferences be different between people over space or within people over time? Isn’t this the basic argument of Clark (but I don’t know your take on him).

    I’ll grant that this is a convenient and disciplining assumption… sociologists explain everything and nothing with their shifting frames or social construction or whatever… but its only that, right? To me, its the same sort of assumption as the assumption of materialism in science. Things may be spooky and supernatural or we may be brains in vats, but we can’t explain those kinds of things so we make a disciplining assumption of materialism.

    Back on the subject, its at least possible that “poor people” have different preferences (not just different constraints) than “rich people”. In particular, I can think of a couple consumption patterns that don’t seem to emerge from budget constraints, e.g. obesity.

  • notsneaky says:

    We can certainly study inequality and yes, sometimes it is hard, but that’s not the issue here. The issue is that studying inequality requires one to study, well, inequality, and not something else. This actually goes beyond this talk of identical or not utility functions and just to the basic units you measure stuff in. The study of inequality is the study of the distribution of some variable. But that means that this variable needs to be measured in common units. If I have a variable x, then divided some portion of observations of x by y and the remainder by z and then talk about the distribution of … well, what exactly x/y and x/z simultaneously? That doesn’t make sense, which is the point here.

    Or think about it this way – the BR approach is as if one was examining height inequality but half the observations (the short people) were measured in centimeters, while the rest where measured in feet. Then the actual numbers were used to calculate some overall measure of “inequality”. Obviously this would be wrong. In fact, BR is even worse, since at least there’s a linear mapping between centimeters and feetses while there’s no obvious mapping (except the one they’re trying to avoid – converting everything into just straight up nominal incomes) between the units of “poor people baskets” and “rich people baskets”.

    Of course if utility was cardinal and interpersonally comparable then we wouldn’t have to worry about all that and just talk about the distribution of utilis. If it was also observable. But since it ain’t, even with cardinality and ipc, any assumption on what the distribution of utility functions is actually like is gonna be ad hoc and in that case there is strong reasons to treat everyone the same – assume identical utility functions.

    As far as Clark, at least he has a story (borrow from Braun or whoever) which explains why the preferences of people 100,000 years ago would be different from that of today. Broda and Romalis just have an ad hoc assumption/justification.

    And you pretty much answer why I have a such a strong prior on this in your next to last paragraph. Either we’re doing science or we’re simply explaining poor people by their intrinsic “propensity for poverty” and the rich by their intrinsic “propensity for affluence” – i.e. engaging in circular arguments disguised and obscured by fancy language (I was being quite unfair to sociology too)

    Explain the obesity thing. I have no idea how this doesn’t follow from budget constraints. Poor people prefer to be fat? Why did poor people prefer to be skinny in the past? Or is it that the used-to-be-poor-people-who-preferred-to-be-skinny all became rich, and the used-to-be-rich-people-who-preferred-to-be-fat (like the “Capitalists” from Marxist cartoons) all became poor?

  • Gabriel says:

    Off-topic: I gather that classes didn’t start yet at your schools?

  • pushmedia1 says:

    I dig. I just wonder if sometimes economists aren’t too constrained by the assumption of constant preferences. There must be some economic phenomenon for which Occam’s razor suggests slight differences in preferences is more likely than a hyper-complex rationalization. Or is economics just the science of rationalization.

    I’m actually cool with that view of economics. Rationalization seems to explain a ton of human behavior. Let rest of the social sciences pick up the scraps.

    Re obesity: what are the common preferences between rich and poor that would suggest differences in wealth would induce *more* consumption (of calories) by the poor? The poor want to eat cardboard tasting health food, but can’t because its too expensive? They want to run around the block but they don’t have perfect access to credit markets?

    BTW, I’ve struggled with obesity before… this is just the first example that came to mind of differences in behavior that doesn’t seem to have much to do with wealth budget constraints.

  • pushmedia1 says:

    Gabriel, I’m a fourth year… what are these “classes” of which you speak? :-)

  • notsneaky says:

    Sure. Sometimes preferences change and people have different preferences. But in each case I think it is important to have a story or a model of why this is so. Assuming whatever ad hoc preferences you want to get your conclusions is weak (and it’s also an instance of when the critics of rational choice theory are right – it is circular). And if you can explain some economic phenomenon without a recourse to differences in preferences, then that is preferable.

    The thing with obesity is straight forward as soon as you stop thinking of food as a good which is meant to provide calories, rather than a good which is meant to provide, well, pleasure and entertainment. I mean if all we cared about was making sure we get some calories without the bad stuff we’d all be eating… I dunno, that hippy food whatever it is. But we consume food not just for the nutritional content but for the taste and pleasure. And as it happens tasty, pleasurable food makes us fat.

    Now, poor people care about pleasure, fun and entertainment as much as other people, despite what that fool Maslow said and what snooty puritanical rich people think poor people SHOULD care about. And when you’re poor a lot of ways to have pleasure, fun and entertainment are expensive – but the price of tasty but bad for you food has been dropping. So this is just people reacting to changes in relative prices of pleasure, fun and entertainment, as embodied in various kinds of goods. Which also explains why it hasn’t only the poor people who’ve been getting bigger.

  • pushmedia1 says:

    So the assumption of homogeneous preferences isn’t disciplining (just have a good model of heterogeneous preferences*), its just less distasteful?

    That’s an odd way to do science.

    * although, it might be the case that these models are hard to come up with so homogeneous preferences makes discipline easier

  • notsneaky says:

    In the absence of strong evidence or reasons for it, one should treat one’s model children equally – assume same preferences – particularly when there’s welfare implications. What’s unscientific about being safe in the face of our ignorance?

  • Gabriel says:

    OK, let’s try another approach… an explanation based on “different preferences” is just one step higher from no explanation at all, and any explanation based on a different assumption is strictly better. That could be a principle…

    Also, the alternative to “different preferences” is not “the same preferences” but rather not having preferences drive the result, i.e. the result holds with “same preferences” and “different preferences” too.

  • notsneaky says:

    “OK, let’s try another approach… an explanation based on “different preferences” is just one step higher from no explanation at all, and any explanation based on a different assumption is strictly better. That could be a principle…”

    That would work if you were trying to EXPLAIN something, rather than MEASURE something.

    “Also, the alternative to “different preferences” is not “the same preferences” but rather not having preferences drive the result, i.e. the result holds with “same preferences” and “different preferences” too.”

    Sure. But if you can’t have both and you don’t have a good reason to assume people really are different in a way that is not circular, go with same preferences.

  • Gabriel says:

    Notsneaky, I suspect that we’re talking about different things…

    Are people different? If we assume that we can observe their budget set and if we see people with the same budget set pick different things then people are different (as a matter of fact, whatever that might mean).

    I see no reason or wisdom in assuming that my yet unborn child will share my preferences re: less onions and more olives on pizza. This is why I’d rather get him/she/it a trust fund in numeraire rather than options on particular pizzas (although I could get more of those, by taking on some tastes risk).

    The only situation in which I see “same preferences” making sense is as a crutch for when general aggregation results are not available. Other than that, I see no redeeming feature.

    Regarding measurement vs. explanation, I’m not sure I understand…

    With price indexes you want to measure people’s ability to buy stuff. Problem is that “stuff” changes over time and across places. And there are never markets for all the stuff that once was, everywhere.

    Ideally, you’d have *everything* in that price index, even WW2 battleships and highways and Louisiana-s but that won’t work out and what Broada and Romalis do is take that one step further and point out that if it makes sense to exclude Louisiana-s from everyone’s CPI calculation, then it might make sense to exclude caviar-filled Ferraris from some people’s CPI. I don’t think that’s necessarily related to preferences, but rather with the share/size of the budget set.

  • pushmedia1 says:

    “In the absence of strong evidence or reasons for it, one should treat one’s model children equally – assume same preferences – particularly when there’s welfare implications.”

    But the economist doesn’t look for strong evidence or reasons for heterogeneous preferences. Other than because the search would be distasteful, why don’t they?

  • notsneaky says:

    “With price indexes you want to measure people’s ability to buy stuff. Problem is that “stuff” changes over time and across places. And there are never markets for all the stuff that once was, everywhere.

    Ideally, you’d have *everything* in that price index”

    Yeah, so why not use the basket bought by a “typical household” as a compromise? And again, what matters in this case is that you use the same basket for everyone so that everything’s in consistent units.

    As to why economists don’t look for heterogenous preferences. Well, at the end of the day what you want to estimate is a demand function, with price and income and then some constants. Does it matter what kind of heterogenous preferences go into that?
    I don’t know, it depends on the area and the question under study. But in the Broada Romalis case it’s most certainly wrong.