With Apologies to Dr. Johnson, I Refute it Thus

Over the past few days much has been made about a study published by a couple of professors, one at Princeton and the other at Northwestern.  They’ve conducted some sort of study about “public policy” initiatives and changes from 1981 to 2002, which income quintiles supported or opposed them, and whether they were actually adopted.  Their conclusions, at least as presented by the BBC, are to the effect that the U.S. is an “oligarchy,” in which “economic elites and organised groups representing business interests have substantial independent impacts on US government policy, while average citizens and mass-based interest groups have little or no independent influence.”  Their conclusions are based on what is described as a “multi-variate analysis” and they do in fact reports some interesting findings.

Before we get into the substance of the study, let’s note a couple of ironies.  The authors of the study are professors at two universities both of which can legitimately describe themselves as elite.  In the U.S. it so happens that professors from elite universities, and the institutions themselves, exercise a fascination with the policy-making industry out of all proportion to their numbers.  Secondly, the BBC link above is to a blog that calls itself “echo chambers,” which is about as unconsciously self-descriptive of the left-wing mainstream media (but I repeat myself) as anything I’ve heard.  They too exercise an influence out of all proportion to their numbers or their intellectual or moral stature.  It is as much due to them that we have Dear Leader as our president now, a man about whose moral and intellectual antecedents we still know next to nothing, seven years after he burst onto the scene and was adopted as the Messiah by precisely operations such as the Beeb.  It is the singular lack of interest in what happened on September 11, 2012, that is the reason why we still don’t know who left a United States ambassador to die in Benghazi.  It is to the nattering classes that we can thank our lack to this day of an answer to the question so not-famously posed by the Secretary of State whose job it was to protect that ambassador: “What difference does it make at this point?”

The full study is available here.

The authors themselves describe their data and methods as “imperfect.”  “Shoe-horned into a mold” might be a better description, but for the moment let’s leave it as it is.  One of the authors has been for many years compiling a data set that consists of answers to polling questions.  In order to be usable at all for the present purposes, they needed to identify answers that met four criteria:  “dichotomous pro/con responses, specificity about policy, relevance to federal government decisions, and categorical rather than conditional phrasing.”  In other words, respondents had to give a yes/no answer to a policy question that was specifically identified and not hedged about with if/and/but/whether.  Moreover, the surveys from which the answers came needed to break the respondents out by income/wealth level.  The authors identified 1,779 such data points.  Of course, since the polls in question weren’t all conducted by the same outfits, they had to massage the income/wealth data a bit to reduce them to a common denominator, so to speak.  Through sundry statistical techniques they were able to isolate the “affluent” (90th percentile), the median (50th), and the very poor (10th percentile).  The paper doesn’t describe how they accomplished that but I’m willing to accept they got it right enough to be useful.

Right up front they acknowledge one problem with their definitions: in 2012 dollars you can report $146,000 and still be in the 90th income percentile.  That sounds like a great deal more than it is.  Let’s start with the fact that it’s $146,000 in reported income, which (to borrow a line from molesworth) as ane fule kno is an entirely different animal than cash you get to see, even if only to pay tax on it chiz chiz chiz.  Just by way of personal example:  For 2013, for every dollar in gross income I reported on my 1040 and paid tax on, I had cash gross income of just over $0.685.  At lower income levels that difference is going to get materially smaller, although for self-employed lower income quintiles it will still be significant.  Add in (subtract, in other words) the tax lick, a not extravagant house payment, a couple of kids in private school (for any of several reasons) or with some sort of special need or special talent which needs addressing, a car payment or two for a mid-range sedan and a “family” car, expenses of commuting, some self-employment taxation, and the other expenses of self-employment, and you can burn through $146,000 very fast and never be free of hoping you can find the dollars to pay next quarter’s tax payment.

The authors don’t think their characterization of modesty as wealth to be a problem, because the policy preferences of the 98th percentile (the top 2%) correlate more closely (r=.91 vs. r=.69) with the rest of the top ten percent than they do with the 50th percentile.  Interesting, but they’ve left out the answer to a question your ordinary cross-examiner would pose to them:  How do the preferences of the next 8% (that is, the 90th through 97th percentiles) correlate with the 50th percentile?  I’ll leave to the reader’s own opinion just how meaningful a 22% gap in correlation coefficient is when you’re measuring something like “policy preferences.”  I’d also point out that r=.69 means that over 47% of the variation in the top 2% preferences can still be explained by variation in the 50th percentile’s preferences.  I find it difficult to suggest that when two groups that far apart on the income spectrum Agree on Stuff almost half the time, in a society as enormous and heterogeneous as America’s, it’s at all helpful to characterize that as evidence of predominating “elite” policy preferences.  This is of course just my opinion, but the authors’ data highlight to me what is an extraordinary amount of convergence about policy preferences across enormous swathes of the American population, a convergence that is nothing short of astounding when you consider the fantastic ranges of social, economic, geographic, and ethnic variation among us.

Measuring individuals’ preferences in relation to enacted/non-enacted policies is problematic enough.  The authors here get much further off the reservation when the issue becomes measuring the effect of organized groups of individuals.  And by the way, I lump into “organized groups of individuals” all voluntary associations, from the American Bar Association to the Chamber of Commerce to the National Education Association to the Sierra Club to the AFL-CIO to the Roman Catholic Church to General Motors and Sun Microsystems.  There is no logically meaningful basis on which to distinguish a shareholder from a union card holder from a congregant from a dues-paying member of the WWF.  Each willingly associates himself and his efforts (represented by his money or some portion of it) with those of other people with whom he imagines himself to share some commonality of interest.  Period.

The authors do not seem to realize that.  They (in Appendix 1, on page 30) divide the organizations they look at into “business and professional organizations” and “mass-based organizations.”  They then take a big ol’ swig of Kool-Aid and recognize a third category of interest group, which they do not code as being either business- or “mass”-based.  Those three are the National Education Association, the National Governors Association, and “universities” (you know, like their own).  Really?  The NEA is somehow different from the AFL-CIO, the UAW, the Teamsters, all of which are lumped into “mass”-based interest groups?

The authors’ blind spot about the nature of organizations goes deeper.  They lump the U.S. Chamber of Commerce into “business and professional interests,” implying of course that it’s Citigroup driving the cart.  Had they drilled down to find out more of whom the C of C is representing they might well have observed how much of the membership is of tiny businesses, mom-and-pop outfits.  Even the vaunted American Trial Lawyers Association (or as it now tries to call itself, the American Association for Justice) has for its members overwhelmingly small firms and solo lawyers.  Yes, there are some enormous plaintiffs’ shops out there, and some huge insurance defense firms.  But the vast majority of its membership consists of the three guys with their shingles out two blocks off main street.  And by the way, the median incomes of lawyers in private practice isn’t that far off what your garden-variety public school teacher makes, especially when you factor in the economic effect of summer vacation (I saw a number of years ago a figure of just under $40,000, at a time when our little country school system was paying roughly that for a teacher with six or eight years’ experience).  And so forth.

On an even more basic level, in determining how to characterize these groups’ efforts — are they to be counted with the oligarchs or with the “masses”? — they might have asked themselves a basic question about the American population:  What percentage of working Americans are at least partially self-employed?  About a decade or so ago I saw the figure 35%.  Hmmm.  That puts a bit of a different color on the horse, doesn’t it?  If you actually ask yourself whose interests are these organizations representing (you know, kind of a critical threshold question for whether these data make any sense at all), you find that while they might be of assistance to the kind of people who backed President Wall Street, they’re also very, very much of assistance to the family of five sitting two pews behind you in church.  Because when you make it harder for ordinary people (you know, that 50th percentile) to buy a house, what you’re doing is making it harder for the members of the National Association of Realtors (one of our authors’ “business interest groups”) to make a living, which means that you’re making it harder for the mother of that family to use her realtor’s license to make some money on a part-time basis while the kids are in school.  Here we might also observe that close to half of all small businesses are owned by women.  Yep, boys and girls:  “Business interests” are women’s interests.  And while not a woman, I’d wager that most women would trade being able to make a decent living being their own bosses for having to shell out $25 a month for birth control pills, Sandra Fluke notwithstanding.

But how did the authors come up with their list of associations to score in the first place?  That was really scientific.  Rather than doing the grunt work of figuring out the total resources devoted to promoting or fighting a particular issue, they used a proxy.

“Fortunately, however, Baumgartner et al. found that a simple proxy for their index – the number of reputedly “powerful” interest groups (from among groups appearing over the years in Fortune magazine’s “Power 25” lists) that favored a given policy change, minus the number that opposed it – correlated quite substantially in their cases with the full interest group index (r=0.73).”

Let’s ponder that.  They outsourced that portion of their thinking to Fortune magazine.  Sort of as if the International Panel on Climate Change decided that getting to the raw data was too much work, so they’d just see what Popular Science had to say about it, and go from there.  But there’s more:  This proxy index “correlated quite substantially” with the full interest group index.  R=0.73, after all, guys.  Wait?  Remind me what was r as between the Top 2% income group and the 50th percentile?  Wasn’t that r=0.69?  Why yes, yes it was.  So we may say that the policy preferences of the Top 2% “correlated quite substantially” with the policy preferences of the median, no?  I refer Humble Reader to my remarks about the convergence of opinion across income groups.  Does that support or contradict the theory of America-as-oligarchy?  Let’s hear it from the authors themselves:

“It turns out, in fact, that the preferences of average citizens are positively and fairly highly correlated, across issues, with the preferences of economic elites (see Table 2.) Rather often, average citizens and affluent citizens (our proxy for economic elites) want the same things from government. This bivariate correlation affects how we should interpret our later multivariate findings in terms of “winners” and “losers.” It also suggests a reason why serious scholars might keep adhering to both the Majoritarian Electoral Democracy and the Economic Elite Domination theoretical traditions, even if one of them may be dead wrong in terms of causal impact.”

I’ll let the reader slog through the specific findings of the paper.  My quibble is not with the outcomes of their statistical model, but rather with the usefulness of the model itself.  It confines itself to asking whether, when various income groups and interest groups supported or opposed a particular policy, that policy actually came to pass.  “Our dependent variable is a measure of whether or not the policy change proposed in each survey question was actually adopted, within four years after the question was asked.”  But how do you determine whether, if something “passed,” it was the question that was posed?  “We want healthcare reform!”  OK.  Do you want socialized medicine, or the “Affordable” Care Act as originally proposed, or the act as it was actually passed, or some fundamental changes to the current system not involving a federal-level take-over?  Homosexual marriage?  How many different potential “policies” are there actually out there?  “Immigration reform”; let’s think of all the possible ways in which you could “reform” America’s present bastard system.  Unfortunately our authors don’t do a very good job of explaining how they coded whether a specific “policy” was enacted or not.  They also don’t explain how they accounted for whether a specific part of a policy that might have been of critical importance to a particular segment of their income/interest groups got enacted.

Finally, and in terms of drawing useful conclusions about whether “elites” are running the show or not, the entire model ignores a central fact about the United States, viz. its divided sovereignty.  It makes a tremendous difference where you live in the United States on a host of issues that are of tremendous importance in the day-to-day lives of Americans.  In fact it makes not a small difference where within a state you live.  I’m going to suggest that in order to be useful, in the sense of getting an idea of whether “ordinary” Americans actually get to run their own show, or are buggered around by Them Awful 1%ers, you can’t weight all of the authors’ 1,779 issue points the same, and you can’t weight those the same as the issues that occur at the state and local levels.  What matters more to our 50th percentiler in his life today, after all:  Whether CAFE standards are increased to 42 m.p.g. by 2025, or whether Common Core is crammed down his local school system, or whether his local grocery store sells wine, or whether, as a solo house-framer he’s now required to pay $8,500 per year for worker’s compensation insurance for himself?

Yesterday morning on my way in to work, I saw an interesting sight.  A song bird, a tiny little song bird, was dive-bombing a crow.  The battle paralleled the center stripe of the street, so I got to see it play out.  This crow must have outweighed his attacker by a factor of at least ten.  I mean, the crow just dwarfed the song bird.  And yet that tiny little song bird ran his country ass off.  It was obvious what was happening.  The song bird had a nest of eggs to protect.  He had a Big Dog in the Hunt.  The crow was just after breakfast, and if he couldn’t crack open that nest’s eggs, somewhere else he’ll find others.  Or even something to eat other than eggs.  I’m going to suggest that metaphor is not unhelpful in trying to make sense of the authors’ data.  Individuals and organizations will pour tremendous effort into attending to what is important to them.  They may agree or disagree with any particular thing to a greater or lesser degree, but their determination to carry their preferences is going to be a function of its importance to them.  I may strongly disagree with federal policy X (or mildly disagree).  But whether I put my back into getting my way is going to be a function of what difference it makes to me.  I just don’t see that these authors’ data or model accounts for that fact of life, and that failure to account for it seriously undermines its utility in understanding the world I see around me.

Now, let’s see if we can come up with some data points on Undeniably Major Issues that might tend to falsify the authors’ hypothesis that elites exercise a significant independent influence on federal policy-making and the rest of us proles don’t.  Because after all, as Popper teaches us, you can’t say something is true if it cannot be false, but you can say something is not true if you can show it to be false.  Since what we’re talking about is something as squishy as “policy preferences” and degrees of support/opposition, we’re not going to get a scientific refutation.  But we might look around for something that is both Extremely Important on Any Objective Basis, and which is inconsistent with the authors’ conclusions.

One which immediately comes to mind is the current battle over amnesty for all the illegal immigrants tooling about the place.  The wealthy and the business interests favor it.  After all, Juan from Ciudad Juarez is not going to be competing with the authors of this paper for a comfortable seat in the faculty lounge; he’s going to put Joe Six-Pack out of work.  Juan will go to work for the homebuilders, not in competition with them.  Granted, we don’t at the moment have an outcome on this, but I think you can derive some meaningful thoughts on our authors’ hypothesis from the number of congresscritters who are running from amnesty like scalded dogs.  If the paper’s hypothesis is correct, that elites and “business interests” exercise a major moving force over federal policy decision, while Us Proles and “mass-based” groups don’t, then we’ll see some significant form of amnesty passed, and passed soon.

One day Dr. Johnson and some friends were discussing one of the then-recent philosophical arguments that the physical world isn’t real, in the sense that what we see we only think we see.  It’s all conception, not reality.  The question arose of how to refute that argument.  Johnson observed, “I refute it thus,” turned, and kicked a nearby large stone so hard his foot bounced off it.  If the United States was truly an “oligarchy,” in which “elites” ran the show for their own benefit, how do you explain that, since 1986 the share of all federal tax dollars paid by that nefarious, oligarchic top 10% of income earners has increased from 54.6% to 70.6% in 2010?  And it’s not that they’re paying a larger share of a smaller tax bill, either.  By 2010 the federal government was consuming a proportion of the country’s GDP (over 23%) not seen since 1946, when we’d just started de-mobilizing from fighting a two-front world war.  That nasty ol’ top 10% is paying a 29.3% greater burden of tremendously larger bill.  In fact, as of 2009, a year in which continued a trend of a declining share of total income by the top percentiles, “The top-earning 5 percent of taxpayers (AGI equal to or greater than $154,643), however, still paid far more than the bottom 95 percent. The top 5 percent earned 31.7 percent of the nation’s adjusted gross income, but paid approximately 58.7 percent of federal individual income taxes.”  Those top-fivers are paying in taxes nearly twice their proportion of the income.  By the way, note further that in 2009 it took only $154,643 in nominal dollars to get you into the 95th percentile.  We see further evidence of the “oligarchic” tendencies of American society in the fact that,  in 2009, “[T]he top 1 percent no longer pays a larger percentage of total income tax than the bottom 95 percent.”  Aw.  Isn’t that just awful?

That’s a pretty sorry-assed oligarchy we’ve got here, if it can’t keep itself from getting increasingly plundered over years and decades.  I defy the authors of this article, and I even more defy all the out-of-breath reporters, at the BBC and elsewhere, to demonstrate any other recognized oligarchy in which the “ruling elite” that runs rough-shod over the toiling masses permits itself to be used so shamelessly as a piggy-bank.

I could go on about all the other respects in which the authors’ data and model do or do not help us think about what kind of society we have in the U.S. or what kind of society we ought or want to have.  But I’ve got to go make some of that $0.6855-per-dollar-of-reported-income.  Suffice it to say, thus, I refute it.

Leave a Reply