Two “frames” for political discourse

Guest Post by Ed Stephan.

Carpetbagger asked me to expand on my post about the Italian economist/sociologist Vilfredo Pareto (1848-1923). There I talked about the cycle Pareto said most governments went through: obtaining power through actual or promised problem-solving, followed by increasing concern about merely staying in power. Since problems aren’t getting solved, governments turn first to lying, then more lying, and when the lies are no longer believed, relying solely on force. At some point the problems mount up, opening the way for those not in power to claim the ability to solve them … and the cycle begins anew. Pareto saw the “circulation of elites” through democratic elections as a way of softening the extremes of this cycle.

The notion of “elites” tends to make most liberals’ hair stand on end. All of us, I think, would agree that there are elites within specific fields. There are very few world-class violinists, a much more sizeable group of players of professional quality, still more of amateur skill, and a virtual army of us who can’t play violin at all. The same is true with most other skills: ability to play football, write literature, rob banks, getting people to vote for us or read our blogs.

Pareto argued that “ability to make money” was just like those other abilities. The very rich are very few, the well-off are more numerous but still rare, most people are well down the income scale. He pictured these types of distribution as in Fig. A (he also developed a famous formula for a continous line of this shape called the “Pareto curve” – there are many later subtle variations, but this is good enough for our purpose).

graph of Pareto and normal distributions

This fact of unequal distribution, of wealth particularly, has always been disturbing to liberals who see such a curve as reflecting some kind of conspiracy to keep the poor in their place. In contrast, we have been taught to think more in terms of Fig. B, roughly modeling what is called the “normal distribution” or the “bell-shaped curve.”

That curve has an interesting history all on its own, much too long to do more than sketch here. It was developed to describe errors of observation in astronomy. I say the star is here, you say it’s there, someone else picks another spot and so on. Where is it? God only knows. What we have are a bunch of all-too-human estimates, i.e., errors of varying degree. The medieval view was for someone (preferably infallible) to pick “the best” guess. The Belgian statistician Quetelet suggested taking them all into account and averaging them, the more extreme errors canceling each other out. He came to believe that all of nature was distributed this way (lengths of ears of corn, heights of people, etc.), extreme circumstances producing highs and lows around some middle value. The “curve of normal error” came to be called the “normal curve”.

Political Implications. The normal curve happened to fit the style of political thought which was emerging in Europe. Rather than the aristocatic elites (Fig A) with “lower levels” looking to “their betters” for guidance, in a democracy there would a distribution of opinion, with some few leaning left, a similar few leaning right, with most of us in the middle (Fig B). Quetelet even envisioned a time when social scientists would be able to measure such distributions, a procedure we now take for granted in the form of measuring “public opinion”, in order to know what the French revolutionaries called the “national will”, what Quetelet called the will of the “average man”.

When I was an undergraduate there was a popular sociology textbook which showed the two distributions, Fig B representing the distribution of intelligence (I.Q. scores) and Fig A showing the distribution of income in the U.S. The title under the display was “Is This Fair?” implying that since our “native abilities” were normally distributed, our incomes ought to be, too … unless some conspiracy was at work to skew the “normal” distribution in this “abnormal” way.


I.Q. Tests. As it happened, when I first saw that display I had just been reading about the history of the I.Q. test. Intelligence, generally, is distributed like any other ability (Fig A) — extremely rare Einsteins, fairly rare bright/pedant/nerds, a much more numerous rest of us. In fact, it took fifty years of development (bowing to Quetelet’s faith in the universality of his curve) to generate a system which would produce I.Q. scores of having a “normal” distribution. And that, in addition to fitting our democratic sensibilities, got enshrined in our school systems: “special” and “gifted” programs for the extremes, with most of us “normals” taking the standard fare.

Pareto argued that Fig A describes what happens “naturally”, i.e., without governmental interference. Countering the Marxists of his day, he said: Suppose you equalized incomes for every household, a little random variation producing something like Fig B. If you went away for a year, leaving things to take their own course, when you came back the incomes would look like Fig A. That is because some people would be very good at making the most of their opportunites and would move up; most, lacking such talent or interest, would drift down the scale. You’d be right back where you were before you equalized the incomes.

Conservatives have used precisely this model to argue that widespread poverty is normal, that markets should be free of governmental interference. It seems to me that liberals need to take this model into account, too. Not as a “natural” condition but as part of the nature of the markets we create and can modify … with political effort. Such expenditures include knowledge about what’s going on in the market (as opposed to Cheney’s secret meetings with Enron), regulation of markets to make sure that there isn’t unfair manipulation of them by entrenched elites, “safety nets” for those who (the model predicts) will fall below some standard of living.

Social “Facts”. Proving (or positing) that something is “natural” doesn’t mean it has to be so in all its effects. If I want to grow a garden on the side of hill, it may be depressing to learn that water “naturally” runs downhill. I could, of course, pray that it were otherwise and ask the gods to make it so. I could accept my lot and bemoan the wealth of those who live down by the stream. Or I could learn something more about the “natural” world – including how to make a pump, e.g., to get the water where I want it. I think something like that needs to be done on the left. Our markets and their consequences are not simply “natural” — they can be studied and politically adapted to meet the needs we specify. We need not, to quote Bryan, be “crucified on a cross of gold” (or any other economic “fact”).

Almost no important social or economic variable is “distributed normally”, like a bell-shaped curve. All them are more or less skewed. Einsteins and Namaths and Ronald Reagans and JFKs are rare — as a matter of policy, you shouldn’t count on them to show up. This raises serious and fundamental questions about how we “frame” poitical issues. It also raises important questions about how to study the elusive, and possibly illusory, notion of “public opinion”.

If Pareto’s model is a better mirror of reality, the whole notion of measuring public opinion, either through large scale national random samples or a simple show of hands in a focus group, comes into question. Are we looking at an apparent number of independent votes, or are those rather dependent on, and reflective of, very few “opinion leaders”, what Pareto called the statistical “elite”? By the logic of the “normal” distribution, it is unlikely that a random sample or a focus group will contain such “ab-normal” opinion leaders. If they don’t, what use are such procedures? what are they really measuring?

Pareto has often been ignored by those on the left, or condemned as an enemy of the (Marxist) left. He was, in fact, offered a seat in the Italian Senate by the fascist Mussolini. I’m pleased to say that he declined the honor and moved, along with his mistress and 40 angora cats, into retirement near Lake Geneva.

Excellent post! I love statistics and this will give all of us more ammunition to make and support better policies.

  • Models and curve theories like the bell-shaped and Pareto can describe the course of a trend, but taking a longer view one could argue that these trends cause an action and reaction to the trend, and hence the continuation of a cycle.

    When wealth accummulates in an increasingly skewed fashion, as it is now, the critical mass of the non-wealthy at the other end of the scale increases. When the social discontent reaches a breaking point, revolution occurs and a redistribution of that wealth will take place and then the cycle will start over again.

    The right may advocate for the Pareto curve being the way things are, but I would argue it has its social consequences and is not a static situation. Using governmental forces to artificially maintain a bell-curve scenario would naturally be better for keeping the peace.

  • “If you could separate causes from results, if you could know that Paine, Marx, Jefferson , and Lenin, were results, not causes, you might survive.”

    -John Stienbeck The Grapes of Wrath

  • Might it be that the Pareto curve is actually just the Bell curve with the left side of the bell counted with the middle? Abilities in a lot of areas do seem to be distributed with a few greats at the top, a few doofuses* at the bottom, and most of us somewhere in between.

    *Let’s face it, just about everyone is a doofus along some axis or other.

    And abilities are fundamentally different from, for instance, money. You can take my money from me one way and another. You really can’t take away my ability to write or dance or dunk a basketball (in two of which areas I am definitely a doofus).

  • The argument you make is perhaps observant, since I have used both these tools in several applications, and concur that we should analyze distribution of wealth in order to obtain a fairer society. The question will always be ambivilant however, because one cannot seperate themselves from the question. The truth and bias require analysis themselves, especially when you try to define what is fair. Ones interpretation of fairness is truly based on what needs are being met. I have a detailed extrapolation of what I term fairness based upon ones wealth and will attempt to post it on my Blog. (http://www.thezonker.blogspot.com)

    Additionally, it should be mentioned that the Liberals attempted to create a fairer society in this country through the organization of labor, Unions.

    I do believe if we postulate that one’s IQ is the sole factor in determining one’s wealth however, we have too much evidence that indicates the contrary proof. (i.e., George W. Bush) Obviously there are exceptions to every rule, and those exception do perhaps cancel out in a natural order, but when we set the limits of that order through government and economic policy, the natural laws cease to apply.

    Your argument that we could use our analysis to improve the lives for all based on fairer government and policy is a sound one. As a Liberal I believe that this is the role of government. History has taught us all too well that unbridled wealth and power only brings out the worst of humankind. Total Freedom is Anarchy, which is no freedom at all.

    I scoff at those who believe all taxes are a sin. Taxes are a necessary part of any government that choses to deal in commerce. Certain guarantees in society can not be privatized because they are essential to a functioning government (i.e. roads, waterways, armies, etc.) Privatization of these essential services only destablizes the functions of government and leaves it vulnerable to coruption. The debate over which services of governement are essential and necessary to ensure a minimum level of acceptable standards has been the question of the last century. It would be nice to answer that question by the end of this century.

  • Good stuff. My colleague, Prof. Matski, has made a similar point: That even if you’re a very classical free-market “liberal” (in the old sense of the word), you should acknowlege that markets were created by man, and not God.

    Matski writes:

    The core of my liberalism is my belief that government has only two proper purposes: (1) To correct market failure and (2) To enforce contracts, including social contracts… My own twist on this classicism is that I believe that MARKETS ARE SOCIAL CREATIONS, and, therefore, government has the right and responsibility to regulate them in whatever way it sees fit…

  • Pretty heavy stuff for the likes of me but it leads me to a question.

    Pareto’s belief that “the ability to make money” falls under the same category of distribution as talents stikes me as somewhat unrealistic.

    It seems from my point of observation that generational wealth does not figure into this theory very well. This inherited type of money grows under it’s own momentum without the recipients having to do much for the growth to continue.

    I am not totally discounting that there are some individuals that seem to have a knack for making money, I have known a few pretty industrious individuals in my lifetime. But when considering all the wealth that is earned in this country during a year I wonder if these types of individuals account for a very large percentage of the total as opposed to the amounts earned by the leisure class.

    Somehow as I’m writing this with the little understanding that I have of such subjects, it looks like there are some major flaws in the logic of supply side economics also.

    Those of you have posted on this thread I would appreciate your help, if there are flaws in my logic on this I would be happy for you to point them out.

  • The comparision between something that falls into a Pareto distribution
    and something that falls into some form of Gaussian (bell curve) distribution
    is comparing apples and aardvarks.

    The reason that wealth follows a pareto distribution has to do with the fact
    that wealth allows you to build wealth. Once you have accumulated a certain
    degree of wealth, you can gain much more. You can control more resources
    both political and economic that you can use to build additional wealth.

    The factors that create the Pareto distribution are also related to
    things like locking in market share (think Microsoft), market presence
    and distribution channels. All this can be summarized by the term
    “the rich get richer”.

    The idea that the government could force wealth into a Gaussian curve is
    entirely argued against by history. The Pareto distribution of wealth
    is constant since capitalism emerged from feudalism. There is nothing,
    per se, wrong with this distribution
    as long as it does not become too extreme. What government CAN do is
    push on the distribution so that the wealthy don’t control quite as
    many resources and they are more distributed to everyone else. This policy
    is not limited to taxation. Building a middle class through educational grants
    and support for unions will do this too. This was government policy until
    about 1980. Government policy is now to shift wealth from everyone else to
    the few in the fat part of the Pareto curve. For example, getting rid of
    the inheritance tax is one way to make sure that the place of the rich in
    the distribution remains unchanged.

    As to economic features not falling into a Gaussian curve: they do actually
    fall into a curve, but the curve has fat tails and is skewed. It shows
    extreme events are much more common than a Gaussian normal would suggest. The classic
    example is profit and loss in a real market like the New York Stock Exchange.

    These statistics, especially the distributions of profit and loss, are of
    great interest to investors and to market regulators. But I think that they
    are of less interest when it comes to social policy. We don’t need the Pareto
    distribution to tell us that the middle class is struggling while the rich
    get richer.

  • Ian Kaplan is essentially correct. What this analysis misses is that we’re measuring two different things–the normal distribution which is much more about the raw material, and the Pareto distribution, which is more about social distribution under certain conditions common to hierarchical societies with significant material surpluses and long histories of elite, anti-democratic governance. Social distribution is affected by both network and iterative effects (overlapping phenomena, the former more about why there is no “next McDonald’s,” the latter about why people with marginal advantages receive disproportionate rewards over time), as well as such mundane phenomena as inheritance, and other social customs.

    The confusion climaxes with the statement, “Almost no important social or economic variable is ‘distributed normally’, like a bell-shaped curve. All them are more or less skewed. Einsteins and Namaths and Ronald Reagans and JFKs are rare — as a matter of policy, you shouldn’t count on them to show up.”

    But Einsteins are rare in the normal distribuion curve, as are IQs of 25.

    As for the matter of elite vs. mass opinion, this has been debated for some time. It’s quite clear that the two diverge in important ways. Elites have often believed in slavery, for example. The slaves, not so much.

    Closer to home, public opinion in the US has remained relatively stable over the past 30 years, fluctuating within a fairly narrow range on basic economic policy (and a much narrower range on abortion), growing more liberal on race and gender matters, and slightly more conservative on crime, while elite opinion has shifted sharply to the right. The different trajectories alone are enough to show that the two are out of kilter.

    Pareto’s lens helps show us that elite opinion dominates, but we knew that already. It hardly constitutes an argument that it should dominate, much less that we should stop caring about the consent of the governed.

  • Ian and Paul,

    Both excellent comments. This really isn’t the place to go into a formal (mathematical) elaboration about the difference between the normal and Pareto (or better, lognormal) model of social reality (it involves additivity vs multiplicativy, and independence). Suffice it to say that if “nothing succeeds like success” then the highly skewed, non-normal distirution results.

    The kid who seems athletic in grammar school is likely to get more attention (coaching, encouragement, etc.) in high school; the best predictor of a scholar’s lifetime publication rate is the age at which s/he publishes the first one; at the same rate of interest, Bill Gates will make a hell of a lot more money than I will because he starts with more.

    I one time made up a little “thought experiment”. Fill an urn with marbles, half red ones and half green ones. Draw a marble from the urn. If it’s green move on to the second urn, red stay where you are. At the second urn the marbles are 60% green. Drawn again. At the third urn the marbles are 70% green. In other words, the going gets easier the further you go. If you carry this process out (simulate it on a computer, with thousands of drawers through a set of such urns), you’ll generate a lognormal (close to Pareto) distribution.

    I think it mirrors much of life. In any field of the endeavor, the hard part is getting started. After that each move up becomes easier, which seems to fly in the face of common sense.

    I actually tried to get data from the Pentagon (number of people by military rank) to try to fit this model, but I was told that data is classified (I hve no idea why). That’s the last I’ve thought of it, till now.

  • Comments are closed.