Why oh why can’t we get a better left academic blogosphere?

I hope I live to see the day that statements like this:

And of course, the poor have done much worse. Household incomes for the bottom quintile have barely moved for decades.

are openly mocked as hopelessly economically illiterate.

“The Poor” (or “The Bottom Quintile”) ain’t nobody ((Wouldn’t it be funny if we talked like this about other transient attributes of people? “The College Town Dwellers” saw decreased incomes over the last couple of decades. “The Sneezers” saw increased tissue use inequality.)) is the obvious rejoinder so feel free to insert such a reply right ***here***. Or ***here***. Add exclamation points if necessary. And make sure to reference the Marxist Fallacy.

Economists usually have a deep appreciation for dynamics in society; intertemporal investment decisions, inflation expectations, growth dynamics and all that. Why is it when it comes to income distribution we slip back into this clunky static framework? As I pointed out earlier this summer, even within the stuffy confines of “income quantiles” thinking, expectations of moving between quantiles, of so-called income mobility, implies agents are ok with increasing income inequality as long as everyone’s incomes are at least weakly increasing (aka “non-zero sum increases in inequality”).

Of course, that result relies on the assumption that people only care about their own level of consumption. Things get sticky if people care about relative consumption. The problem is there just isn’t compelling evidence of such “inequality aversion”. In lab experiments (eg. 1, 2, 3 pdfs), agents do show social preferences, preferences over the consumption level of other agents, but those preferences are expressed as preferences for efficiency (i.e. maximize total consumption) or a preference to maximize the minimum consumption level in the “society” ((In the case of these lab experiments, society is just the other agents in the experiment.)).

If social preferences are for efficiency, then the result still holds because everyone’s expected incomes increase with non-zero sum increases in inequality. Increases in everyone’s incomes satisfy a preference for efficiency by definition. Furthermore, even if social preferences come in the form of maximizing the consumption of the worse off, then the result holds due to the fact that even the bottom quantile’s income is at least stagnant by assumption. This is a key assumption and its reasonable given it conforms to recent experience.

Now, when people are surveyed, they say they don’t like inequality (e.g. this pdf). Besides the usual litany of problems with survey data, a major defect of these studies is that you can’t tell if these stated preferences reflect inequality aversion or maximin preferences or preferences for efficiency. Its next to impossible to ask a nuanced enough question to get at these distinctions and still expect survey respondents to understand what your asking. These distinctions matter a great deal for policy, of course. As mentioned above, social preferences for efficiency and maximin don’t call for redistribution in the face of increasing non-zero sum inequality. Annoyingly, these distinctions are ignored by the survey researchers who interpret their results as measures of inequality aversion and then make the attendant policy recommendations.

Results from the lab have their own litany of problems, but on net they’re a more reliable because they measure the thing we actually care about. Surveys tell us that people have social preferences, but lab experiments tell us people don’t have strong aversion to inequality.

UPDATE: Notsneaky is provoking me in the comments: “I have no basis except my intuition for this claim, but its my blog and I’ll speculate if I want to: quantile based income statistics are normatively biased against more dynamic economies. What do I mean by more dynamic economy? Dynamic economies are those that give more opportunities for individuals to reinvent themselves via schooling, relocation or just through norms that allow people to make dramatic changes in their careers. Personally, I think dynamic economies are better. Anyway, why do I think those economies are normatively biased against by quantile thinking? Dynamic economies will have more individuals that have temporarily low (or zero) incomes and so they’ll have lower incomes in the bottom quantiles.

So we have this funny situation where dynamic economies (good) will be confused for unequal economies (bad). Now, I’m not sure if the American economy is unequal or dynamic, but I am sure that everyone assumes the statistics about “The Poor” are bad.”

15 thoughts on “Why oh why can’t we get a better left academic blogosphere?”

  1. I think you’d be right if he was talking about quintile shares (in fact, then he’d almost certainly be wrong) but he’s talking about bottom quintile incomes. Which roughly means that either the folks who wind up there never leave and never see their income rise, or more optimistically, that when the poorest enter the economy they do at a level no higher than decades ago. Now, of course maybe some of them (lots of them even) leave that low level but it still means they got some catching up to do in terms of their lifetime income (catching up in the sense of keeping up with the average). It’s a statement about poverty, not inequality.

    Sorry, don’t mean to be disagreeing with you this much lately. It’s just somehow happening.

  2. Huh? Your sentiment is hopelessly economically illiterate! :-)

    “The Poor” isn’t a fixed group of people over time. Why does it make sense at all to talk about economic stagnation of a group of people that’s changing from year to year? And why the “where people end up”. People don’t “end up” in transient income quintiles. They move between them with quite a bit of frequency through their lives.

    By this measure, I’m poor. Its just plain silly to categorize me as poor.

    Now if he lamented the fact that there’s quite a number of people who stay in poverty through there lives and he had data showing the trend in that number was increasing, then I’d get it. That seems like an economically important thing to care about. “The Poor” measured by bottom quintiles of transient incomes is a really bad proxy for that measure, though.

    PS – Disagreement is why I’m here… I’m a prick like that.

  3. I’m more appalled by the fact that nearly half the people make below the median income. It’s totally unfair and the government needs to fix it somehow.

  4. So I take it you don’t believe in the poverty rate or even something like the illiteracy rate (since the illiterate aren’t really a fixed group of people over time either)? Sure, the household income of the poor doesn’t tell you everything about a person’s lifetime income you’d like to know, but hey, it’s a summary statistic. Of course it’s gonna miss some information. In the absence of detailed longitudinal data and for the purposes of a blog comment I think it’s perfectly fine to refer to it.

    Take two economies one with higher household income for the poorest 20% than the other. In the poorer economy there’s gonna be SOME people, whoever they are, whose income, for at least a portion of their lives is gonna be lower than ANYONE’s income in the richer one. Analogous statement can be made for an economy in which the income of the poorest 20% is growing vs. one where it is stagnating.

    Of course, yeah, sure, the composition of the poorest 20% in terms of age, job tenure, etc. also matters. But that’s additional information.

  5. Your second paragraph is wrong and beside the point. Less substantively: averages don’t say anything about distributions. The average income could be less, but that doesn’t mean the poorest are poorer. Maybe the “richer” economy has fatter tails in its distribution.

    More substantively: if the “poor” economy has more opportunity for laborers to retool — like going back to school to learn a new trade or a greater ability to relocate — then you may see lots of “poverty” measured in this way. I’m retooling my career as we speak and as I said above, it makes approximately zero sense (s.e. 0.1) to call me poor.

    I have no basis except my intuition for this claim, but its my blog and I’ll speculate if I want to: quantile based income statistics are normatively biased against more dynamic economies. What do I mean by more dynamic economy? Dynamic economies are those that give more opportunities for individuals to reinvent themselves via schooling, relocation or just through norms that allow people to make dramatic changes in their careers. Personally, I think dynamic economies are better. Anyway, why do I think those economies are normatively biased against by quantile thinking? Dynamic economies will have more individuals that have temporarily low (or zero) incomes and so they’ll have lower incomes in the bottom quantiles.

    So we have this funny situation where dynamic economies (good) will be confused for unequal economies (bad). Now, I’m not sure if the American economy is unequal or dynamic, but I am sure that everyone assumes the statistics about “The Poor” are bad.

  6. Where did I say anything about “averages” ? I’m only talking about the tails, or one of them for that matter.

    “Maybe the “richer” economy has fatter tails in its distribution.”

    That’s why I didn’t write that one economy is richer than the other but that one has “higher household income for the poorest 20% than the other”. My subsequent use of “poorer” and “richer” in the rest of the paragraph follows from that and merely reflects the fact that I didn’t feel like writing “higher household income for the poorest 20% than the other” over and over again. Bit sloppy on my part, but come on!

    And btw, this is just a plain ol’ crazy statement: “averages don’t say anything about distributions”. Do variances tell you anything about distributions? Skewness? Kurtosis?
    Of course if you had written “averages don’t tell you EVERYTHING about distributions” then you’d be correct.

    Also I don’t think what you write in your 2nd paragraph was your initial objection – that problem can be fixed simply by changing “household” to “worker” (since folks retooling their stats are not considered to be in the labor force).

    And in your third paragraph, you’re really talking about the number of people – but given the population, the number in the lowest quintile, or any quintile is fixed – so that’s beside the point. This is about the level of income, not the y axis of the distribution but the area underneath it, for the lowest 20% of the population. So it has nothing to do with the amount of income that these “retooling” people get, if that’s who they are (which is a very optimistic guess) – and that’s what’s been stagnating.

    For 4th paragraph – write me down the model, I wanna see it! That’s not a criticism or derision, it’s genuine encouragement (I might give it a go myself). It doesn’t have to be fancy with utility functions or anything, just some flows from one income group to another and some fixed proportions (which could be endogenized later, so reduced form and all). Something like OLG where a portion of young people work, portion “retool” then they become old people and make wages based on retooling and experience. Then do some comparative statics wrt the “portion” parameter. But it should produce the phenomenon that the income of the bottom is lower the greater the portion that are moving up in the world over time.

  7. Your a funny man. Averages say *nothing* about distributions (in the sense there’s an infinite number of fitting distributions) when you’re talking about empirical real-world distributions. When you’re talking about analytical/nice distributions we see in classrooms then the first couple of moments tell you something. In any case, I took you to be talking about the distribution within the 20th percentile and so my criticism was about the distribution within that group.

    My later paragraph isn’t talking about the number of people. Simple point: if dynamics, as I defined, increase, the income of “The Poor” goes down. This is because by defining poor as the bottom quantile (or whatever) you’re picking up all the people that have transient low incomes.

    The model I have in mind is that there’s transition costs when labor moves from one type of labor input to another (there’s many types of labor inputs). Technology in this model doesn’t increase their productivity, it makes it easier to transition inputs. This makes marginal transitions more likely so and we see more transitions in a lifetime. Lifetime income is higher (because transitions are to supplying higher paid labor inputs), but there’s more episodes of low incomes (i.e. higher variance in year incomes). Higher variance means “The Poor”, as defined by the average income in the lowest income years (a definition that corresponds to the cross-sectional definition we’re talking about), is lower.

    If computer techs get paid more all of a sudden (because of production technology changes), people will want to become computer techs. They can’t because there’s transition costs to doing so. The invention of online universities or norms that make it ok to drop your career and move into a new one or whatever make such transitions easier. With more transitions there’s more people with (or more episodes of) low incomes as they transition. This isn’t poverty or inequality. This isn’t badness.

  8. Oy, provoking?!!? Such strong words. I was merely inquiring and all…

    “Averages say *nothing* about distributions (in the sense there’s an infinite number of fitting distributions) when you’re talking about empirical real-world distributions.”

    Ey? Unless you have every single observation in the universe under consideration. there’s always an infinite number of fitting a distribution. Averages tell you the average of the distribution.

    “f dynamics, as I defined, increase, the income of “The Poor” goes down. This is because by defining poor as the bottom quantile (or whatever) you’re picking up all the people that have transient low incomes.”

    Maybe – and I mean maybe in the statistical sense. That is if the group of those who are leveling up is large enough. But suppose you got a bunch of ordinary poor people. Then you got some folks who drop out of the job market to acquire new skills. Since a quantile is based on a number of people (assuming fixed pop), those ‘retoolers’ would have to consist of sufficiently large pool of people in order to push out the regular ol’ po’ folks. And that’s assuming that their income is actually lower than that of those in the bottom quintile (TA salaries I think straddle that edge).

    And
    Probably not – and this I mean in the empirical sense. A quick purview of poverty statistics suggests that the poor are NOT mostly composed of graduate students getting their graduate degrees.

    And like I said, all that can be fixed by just moving to worker income rather than household or individual income, since ‘retoolers’ are no counted among the workers. And my understanding here is that that tells pretty much the same story.

    So in light of all dat, I think John’s comment, while maybe a shorthand for a whole host of nuances and problems, does not constitute “economic illiteracy”.

    I think you’re starting out your model a bit too complicated. I’d just set up the basic transition rates and flows between various kind of worker/individual pools before worrying where those transition rates come from. That should tell you whether the kind of effect you’re describing would really work – and I’m not convinced that it would work out the way you think it would work out, when the maths and squiggles have all been properly laid down on paper.

  9. I have a hard time believing that this (transitions through the income distribution over the course of one’s life) hasn’t been studied. Is it true?

  10. Jason, yes its been studied before. Most of the work is descriptive though like “x% of people move up or down income quintiles per year” kind of stuff. The mechanism has been studied in the context of job switching… I’m waiting for a train so I don’t have cites in front of me… some guy with a russian name.

    notsneaky, on the theoretical point: yes, quantiles are fixed numbers of people, but the average income of a quantile can go down (or up) because different people start getting included (pushing the others out). This is actually the crux of the economic illiteracy claim. These aren’t the same people year after year so increasing/decreasing incomes in a income group have zero normative meaning (given “normal” egoistical preferences or even social preferences that are actually observed in the lab… this was the content of the post we’re commenting on).

    On the empirical point, grad school isn’t the only type of retooling. At this point I don’t hope to account for or make a list of specific ways in which people retool, but I hope to give evidence that they do and they do so more today than before. Also, the technology as transition enabler story suggests technological changes would correlate with this effect.

    Some anecdotal evidence: there are 2 members of my immediate family and a close friend who took extended time off this last year or took low paying part-time jobs to pay the bills while making a career change. These moves were of their own volition. In the income statistics, these people looked poor and they dragged down the average incomes of the lower income groups. Is their behavior evidence of increasing inequality of the American economy or evidence of its dynamism?

  11. Well, here’s classic Krugman:

    “There is considerable income mobility in the U.S., but by no means enough to make the distribution of income irrelevant. For example, Census data show that 81.6 percent of those families who were in the bottom quintile of the income distribution in 1985 were still in that bottom quintile the next year; for the top quintile the fraction was 76.3 percent. Over longer time periods, there is more mixing, but still not that much. Studies by the Urban Institute and the U.S. Treasury have both found that about half of the families who start in either the top or the bottom quintile of the income distribution are still there after a decade, and that only 3 to 6 percent rise from bottom to top or fall from top to bottom.

    Nor is there any indication that income mobility increased significantly during the 1980s. Table 1 shows some evidence from a study by Greg Duncan of the University of Michigan on transitions over a five-year period into and out of a somewhat arbitrary but reasonable definition of the “middle class”. This middle-class category shrank in the 1980s, so that middle-class families became more likely both to rise and to fall; but correspondingly fewer poor families moved up or rich families down into the middle class. (Vanishingly few poor fami- lies became rich or vice versa). The overall picture suggests little change in mobility.”

    Bottom line

    “Income mobility might in principle be an important offset to the growth in inequality, but in practice it turns out that it isn’t. ”

    I’d add that while I don’t care about inequality per se, lack of mobility + stagnating incomes at the bottom I do see as a problem, whatever people in experiments might think about that.

Comments are closed.