Welcome to European Tribune. It's gone a bit quiet around here these days, but it's still going.

Best Policy Money Can Buy?

by geezer in Paris Wed Jun 24th, 2009 at 03:49:46 AM EST

Nate Silver of fivethirtyeight.com has done a great job of doing a quick-and-dirty model to prove what we all knew all along- that money buys policy. So why am I excited? Because I'm one of those odd people who question my own certainties-- who likes to find that, once in a while, I was certainly wrong.
Not this time.


Money Rules

"As I lamented yesterday, health care is one of those areas where both popular opinion and sound public policy seem to take a backseat to protecting those stakeholders who benefit from the status quo. But can we actually see -- statistically -- the impact of lobbying by the insurance industry on the prospects for health care reform? I believe that the answer is yes."

Nate Silver:
"I decided to build a model to explain and predict whether a particular senator supports the public option. The variables in the model are as follows:

    -- The senator's ideology, as measured by his DW-NOMINATE score;
    -- Per capita health care spending in the senator's home state;
    -- Lobbying contributions received by the senator from health insurance PACs since 2004.

Below the jump, I explain each of these in a bit more detail.

Ideology. DW-NOMINATE scores measure a senator's ideology on a scale that generally runs from -1 (extremely liberal) to +1 (extremely conservative), although scores slightly less than -1 or slightly greater than +1 are possible under unusual circumstances. These days, the average Democratic senator has a score of -.44, and the average Republican senator a score of +.48.

DW-NOMINATE data has not yet been published for the 111th (current) Congress, so I use data from the 110th Congress instead. For freshman senators, I extrapolate DW-NOMINATE scores by translating (via regression analysis) Progressive Punch scores. Extrapolated scores for freshmen members of the Senate are as follows:

Merkley     -.495
Burris      -.494
Kaufman     -.494
Gillibrand  -.483
Udall, T    -.472
Udall, M    -.420
Warner      -.398
Begich      -.390
Bennet      -.384
Hagan       -.353
Johanns     +.532
Risch       +.534

In addition, a special score is required for Arlen Specter, who recently switched parties. Specter's score is extrapolated based on a recent analysis we did of his voting behavior since becoming a Democrat. Specter's score according to this analysis is -.255. This makes Specter the fifth most conservative Democrat, slightly to the left of Evan Bayh and slightly to the right of Tom Carper.

Per Capita Health Care Spending. As estimated by the Department of Health and Human Services. We use data from 2004, which is the most recent available. Health care spending varied in 2004 from $3,972 per head in Utah to $6,683 per head in Massachusetts.

PAC Contributions. Based on data downloaded from OpenSecrets.org, a.k.a. the Center for Responsive Politics. Contributions were tallied from two industry codes: F3200 (Accident & Health Insurance) and H3700 (HMO's). Data covers the 2004, 2006 and 2008 and 2010 campaign cycles. The fundraising data is adjusted based on the number of cycles that the senator has participated in as a Congressman (including time spent in the House of Representatives) or as a candidate, where 2010 is treated as 1/8th of a cycle since one quarterly report has so far been filed from the two-year period. So, for example, a senator that ran for and won office in 2006 is treated as participating in 2 1/8th out of a possible 3 1/8th cycles: 2006 as a candidate, and then 2008 and the fractional cycle in 2010 as a senator."

In fairness, I'll quote somewhat sparingly,- there's lots- because Nate's work should be read in it's entirety. Link above.

------------------------------ (reading space)

Here's a teaser for those who are short of reading time:


The model employed is a standard logistic regression with these three variables: ideology, lobbying and health care costs. Ideology is statistically significant at the 99th percent level, PAC contributions at the 95th percent level, and health care costs -- senators in states with more health care spending are more likely to support the public option -- at the 90th percent level. The R-squared for the model is .61, which means that these three variables alone give us 61 percent of the information that we need to predict a senator's position on the public option. The model guessed the senator's position correctly in 87 out of 99 instances.

There are several neat things we can do with this.

 One is to evaluate the impact of insurance industry money on a senator's likelihood of supporting the public option, holding ideology and health care spending in the senator's home state constant. The chart below presents these estimates for a liberal (DW-NOMINATE score of -.6), mainline (-.4) and centrist (-.2) Democrat, as well as for a centrist (+.2) Republican:

Lobbying contributions appear to have the largest marginal impact on middle-of-the-road Democrats. Liberal Democrats are likely to hold firm to the public option unless they receive a lot of remuneration from health care PACs. Conservative Democrats may not support the public option in the first place for ideological reasons, although money can certainly push them more firmly against it. But the impact on mainline Democrats appears to be quite large: if a mainline Democrat has received $60,000 from insurance PACs over the past six years, his likelihood of supporting the public option is cut roughly in half from 80 percent to 40 percent.

But does it hold water? James Kwak at Baseline scenario thinks it does pretty well.

Modelling everything, Public Plan edition

Mark Nyhan has a counterargument, and cites the usual raft of papers--

"Nyhan says "studies have typically found minimal effects of campaign contributions on roll call votes in Congress," and cites a Journal of Economic Perspectives paper as backup."

So Kwak looked it over. His conclusions follow:

"OK, Nyhan may be right. But he may not be.

I looked at that paper. First, it cites a stack of papers that support Silver's view (that campaign contributions do influence policy). (Of course, it's common to cite the papers you are trying to refute.) Then it describes a logical argument against Silver's view ("Tullock's Puzzle"), which makes no sense to me, at least as summarized there. The argument is that campaign contributions are pitifully small given the amounts of money at stake, and so firms cannot possibly see contributions as an investment in policy. For example:

    Dairy producers, who since 1996 have had to have subsidies renewed annually, gave $1.3 million in 2000 and received price supports worth almost $1 billion in the Farm Security and Rural Investment Act of 2002.

I dont' see the puzzle; if I can get $1 billion in subsidies by paying $1.3 million, why would I pay any more? It seems to me that the explanation here has to do with special-interest politics; no other constituency is sufficiently mobilized to fight against dairy producers, so they get their subsidy. Here's the conceptual argument: "The figures above imply astronomically high rates of return on investments. In a normal market, with such high rates of return, existing donors should want to increase their contributions." But this assumes that the "investment return" on campaign contributions is a smooth, monotonic (always increasing) function, which seems fundamentally at odds with the way Congress works. But as I said, maybe Tullock did a better job explaining his puzzle.

Finally, they do a regression of "roll call" votes in Congress against campaign contributions and find little influence. But this analysis has serious limitations in the present context.

First, the dependent variable is the aggregate rating of each legislator by the U.S. Chamber of Commerce. (They got similar results using other organizations.) That is, it's an average of a large number of votes made in the course of a session on a large number of issues. So the finding is that corporate contributions only pull a legislator a couple of points toward the Chamber of Commerce's positions overall; but that doesn't mean that on a given issue, he might switch his vote because of one or two large campaign contributors from the affected industry.

Second, it ignores the complexities of the legislative process. On many key issues, there is no roll call. For example, whether the public plan even comes to a meaningful vote will depend on Harry Reid - who may not even want the public plan - and whether he thinks he has the votes. The power of some members of Congress goes well beyond their individual votes.

Third, the data are from 1978-1994. I'm not a student of American politics, but casually reading The New York Times indicates a few changes in politics since 1994: there is more money; there is more money that is not controlled by political parties (weakening bonds to party, which historically explained a lot of votes); and there is more money that gets spent directly on advertising and is not contributed to political campaigns. This last factor could cut either way, but I think it's fair to assume that if a company gave you $50,000, they are probably also donating to soft-money groups that take the same positions.

So here we have: on the one hand, a quick-and-dirty model on this specific question by a guy without a Ph.D. in anything (I think), which correctly picks the positions of 87 out of 99 senators (but it's possible that the model would have done just as well without campaign contributions); on the other hand, a guy with a Ph.D. in political science and a research fellowship in health policy at a very good university, citing a paper by three MIT professors that's more or less on the same general topic, arguing that campaign contributions don't affect policy.

I'd like to refer the reader to my diary, The best and the Brightest , for some thoughts and good links as to why the pedigree James quotes may mean something very different than would appear. I respect Kwak's efforts to place the work in context, but I think it's value is apparent As do the readers at Baseline, as shown by the comments. Seems there's real predictive value here, too. How nice:
What happens if we set the lobbying variable to zero for all senators? That is, suppose that the health care insurance industry were prohibited from making political contributions? In that case, the model predicts, 47 senators would currently support the public option, as opposed to the 38 who actually do. In other words, the insurance industry's influence appears to swing about 9 votes against the public option. Whatever number of senators wind up supporting the public option, add 9 to it, and you'll have a decent ballpark estimate for what the level of support might be if not for insurance industry contributions.
The relationship Nate Silver explores between money and policy is a crucial one, in all quasi-democratic, Western-style political settings, and Nate's work has obvious value in Europe as well as in the US, as at least a good starting point to try to get a handle on it in a comprehensible way.

Display:
Nate has been doing outstanding work in putting numbers and statistical probabilities on all sorts of political hypotheses and theories.  He elevates the whole debate from the usual anecdotal "evidence" to a much more scientific base.  It's not often that the term political science warrants the term science.

notes from no w here
by Frank Schnittger (mail Frankschnittger at hot male dotty communists) on Wed Jun 24th, 2009 at 05:00:36 AM EST
This is Exhibit A in a demonstration of why campaign finance reform is essential PRIOR to substantial reform of ANYTHING impacting large economic interests.  The best Obama can hope for is a system that is comprehensive but that costs far too much.  Else, he must take on campaign finance reform, which I don't see him willingly doing.

"It is not necessary to have hope in order to persevere."
by ARGeezer (ARGeezer a in a circle eurotrib daught com) on Wed Jun 24th, 2009 at 08:39:41 AM EST
Sadly, I agree. Our hero seems reluctant to stand toe to toe and slug it out with any of the real issues, and campaign reform is number one. With three quarters of the population begging for leadership and supportive of a centralized approach, if he can't or won't do it now, he's over.
A medical care system that fails to address the cost issues is an automatic fail. No system based on the insurance industry model will do that. Neither will the "public option", unless some huge holes are plugged. Here's a great link, for those who are fans of Bill Moyers.
The real battle to reform health care

But--look at when this was aired. It's a month old.
Shit.


Capitalism searches out the darkest corners of human potential, and mainlines them.

by geezer in Paris (risico at wanadoo(flypoop)fr) on Wed Jun 24th, 2009 at 12:44:20 PM EST
[ Parent ]
European Tribune - Best Policy Money Can Buy?
So here we have: on the one hand, a quick-and-dirty model on this specific question by a guy without a Ph.D. in anything (I think), which correctly picks the positions of 87 out of 99 senators (but it's possible that the model would have done just as well without campaign contributions); on the other hand, a guy with a Ph.D. in political science and a research fellowship in health policy at a very good university, citing a paper by three MIT professors that's more or less on the same general topic, arguing that campaign contributions don't affect policy.
Well, Nate Silver happens to be a University of Chicago econ[omics] alum[nus]. So he may not have a Ph.D. in anything but he probably has the econometrics training to get one if he so chose.

A man of words and not of deeds is like a garden full of weeds; a man of deeds and not of words is like a garden full of turds — Anonymous
by Migeru (migeru at eurotrib dot com) on Wed Jun 24th, 2009 at 08:54:02 AM EST
you can get PhDs in economics or political science without any econometrics training...

In the long run, we're all dead. John Maynard Keynes
by Jerome a Paris (etg@eurotrib.com) on Wed Jun 24th, 2009 at 09:07:08 AM EST
[ Parent ]
It would probably be a hindrance, actually.

A man of words and not of deeds is like a garden full of weeds; a man of deeds and not of words is like a garden full of turds — Anonymous
by Migeru (migeru at eurotrib dot com) on Wed Jun 24th, 2009 at 09:34:59 AM EST
[ Parent ]
this is not econometrics, is it? It looks like a trivial multiple logistic regression model to me, with a four level factor for ideology and a continuous variable for corruption.

Seriously, any MSc student in ecology would be expected to be able to do this.

Not that the result isn't totally cool!

by PIGL (stevec@boreal.gmail@com) on Wed Jun 24th, 2009 at 10:01:23 AM EST
[ Parent ]
"trivial".

Sometimes it aint the calibre of the weapon that counts, so much as knowing where to point it.

Capitalism searches out the darkest corners of human potential, and mainlines them.

by geezer in Paris (risico at wanadoo(flypoop)fr) on Wed Jun 24th, 2009 at 12:28:24 PM EST
[ Parent ]
The fact is, with real-world data the simpler (the more "trivial") the statistical model, the better. Because a sophisticated model will likely not give significant results.

A man of words and not of deeds is like a garden full of weeds; a man of deeds and not of words is like a garden full of turds — Anonymous
by Migeru (migeru at eurotrib dot com) on Wed Jun 24th, 2009 at 12:32:07 PM EST
[ Parent ]
The most devastating critique possible is one that exposes first order problems.

"It is not necessary to have hope in order to persevere."
by ARGeezer (ARGeezer a in a circle eurotrib daught com) on Sat Jun 27th, 2009 at 01:05:53 PM EST
[ Parent ]
My only question would be what if it isn't the money but rather ideology that drives the voting pattern at least in a significant number of cases. To hear many of our congressmen speak (both parties), one would think they were reared on a steady diet of anti-government sentiment. I can't think of many who are willing to speak out in support of social government programs even after years and years of embarrassing private sector failures and fraud.

I can swear there ain't no heaven but I pray there ain't no hell. _ Blood Sweat & Tears
by Gringo (stargazing camel at aoldotcom) on Fri Jun 26th, 2009 at 12:07:28 AM EST
Your'e right, of course. I think the ideological baggage that cripples their thinking is central to the problem.
But it's crucial to factor into the search for explanations the reality that human events are almost never simple-- if "A" then "B" is just misleading, destructive of real analysis.

What if it's ideology instead of greed?

Nah. It's both, plus a web of other influences.

But as Mig says, a simple model with well selected variables can be a powerful tool to educate, to illustrate, even to predict, in some cases.

Nate does it well.

Capitalism searches out the darkest corners of human potential, and mainlines them.

by geezer in Paris (risico at wanadoo(flypoop)fr) on Fri Jun 26th, 2009 at 02:23:47 AM EST
[ Parent ]
Nate's model controls for ideology. Of course the assumptions that go into determining any given senator's ideology can be debated, but the trend lines are for senators with comparable ideology but different levels of corruption.

- Jake

Friends come and go. Enemies accumulate.

by JakeS (JangoSierra 'at' gmail 'dot' com) on Fri Jun 26th, 2009 at 04:18:54 AM EST
[ Parent ]
Yes, good point. Silver's methodology could fully account for ideology, but to my mind there is the question like which came first, the chicken (money) or the egg (ideology/conservatism).  I can't tell/guess from the information presented.

I can swear there ain't no heaven but I pray there ain't no hell. _ Blood Sweat & Tears
by Gringo (stargazing camel at aoldotcom) on Fri Jun 26th, 2009 at 10:54:45 PM EST
[ Parent ]
That's a bird of a different feather. It's on the edge of one of the central questions for people who study the psychology of human motivation--and the model offers no answer, because it's more complex than the feathered  duality. Nate's model wisely limits itself to illustrating events that vary together- are related in an apparently predictable way, without opening the Pandora's box of "why?".

Capitalism searches out the darkest corners of human potential, and mainlines them.
by geezer in Paris (risico at wanadoo(flypoop)fr) on Sat Jun 27th, 2009 at 03:12:47 AM EST
[ Parent ]
I agree.  His model is quite useful for its purpose and amply demonstrates the alliance of certain birds of a particular feather.

I can swear there ain't no heaven but I pray there ain't no hell. _ Blood Sweat & Tears
by Gringo (stargazing camel at aoldotcom) on Sat Jun 27th, 2009 at 10:26:17 PM EST
[ Parent ]
Nate's model does not prove, or even provide much evidence, that money buys health care policy. All Nate has shown is that monetary contributions from insurance PACs are positively correlated with voting against a public option for health care.  But, as students learn in introductory statistics courses, correlation is NOT causation.  Cause is a much more difficult thing to show from observable data, even in physical science, let alone the more subjective social sciences.  

Nate's analysis really just shows that monetary donations from insurers tends to go more to legislators that vote for things insurers like.  That's no different than saying that monetary donations from progressives tend to go to candidates that progressives like.  It does not at all imply that policy is in any way sensitive to the amount of such donations.  

To show cause you need two things that Nate lacks: a theory and data proving a one-way relationship between the correlated factors.

First, let's talk theory.  Nate implies a common, if cynical, theory about how the world works: that policy responds to financial contributions.  It might be true, and a whole field of economics and political science, called political economy, is devoted to studying the relationship between wealth and policy. (Harvard's Dani Rodrik is the probably scholar best known for his work in this area, particularly regarding whether it can be shown that money buys trade policy.) But in order for it to be true, you have to prove two things:  1) That there is a unique equilibrium where policy outcomes equal financial contributions, and 2) that observable data back up the theoretical existence of such a "market-clearing" equilibrium.  

Nate does neither with his simple regression analysis.  Saying that money buys policy is the same thing as saying that the policy-making process is just a market process, like markets for commodities.  But markets only work when they clear -- when supply and demand for things converge to an equilibrium, usually shown by a price of one thing relative to another thing, like money.  If there is no unique equilibrium, then you usually have little reason to believe that a market will ever tend to clear, which is the same thing as saying that prices (or contribution amounts) have anything to do with supply and demand for things.  That's why academic economics articles spend so much effort on upper-level math before making a usually simple point with basic regression analysis: the calculus proves the theoretical existence or not a single equilibrium in an equation where supply=demand. Nate didn't do that, and much economics lit on the subject doubts whether unique equilbria can be found in most real-life circumstances of so-called policy "marketplaces," making it problematic to conclude anything from his regression.

But even if there was reason to believe that health care policy acts like a marketplace and that there exists a price level of donations for which policy can be influenced, his regression analysis doesn't even show that.  He hasn't proven the absence of endogeneity, which is the problem of both independent and dependent variables affecting each other. In this case, he hasn't proven that donations by insurance companies to legislators weren't "caused" by the legislators' prior relationships and commitments instead of their votes being "caused" by the donations.  The legislators who opposed the public option might very well have always opposed such things, and their donations from insurance companies already reflect that before any of the present discussion of the current policy ever came up.  

Endogeneity is usually an unobservable factor, so only a well-specified and compelling model of how the world works can provide good evidence that endogeity between two variables doesn't exist and causality can be concluded from correlation.  There are, however, statistical techniques that can be used, particularly instrument variable regression analysis, from which direction of causality can be deduced, but Nate didn't use it. A good model for Nate's purposes would require evidence that money is being donated at or around the time of the vote -- that the legislator is being bribed.  But that evidence doesn't exist, and when it does, it's usually part of a criminal investigation such as in the case of the ex-Governor of Illinois.

Most analysis of how policy gets formed and implemented show that money helps, but finanical contributions are only one variable factor among many, and often not the most important ones, that influence public policy outcomes.

by santiago on Fri Jun 26th, 2009 at 12:30:50 PM EST
Santiago--see above.

Capitalism searches out the darkest corners of human potential, and mainlines them.
by geezer in Paris (risico at wanadoo(flypoop)fr) on Sat Jun 27th, 2009 at 03:14:12 AM EST
[ Parent ]

A good model for Nate's purposes would require evidence that money is being donated at or around the time of the vote -- that the legislator is being bribed.  But that evidence doesn't exist, and when it does, it's usually part of a criminal investigation such as in the case of the ex-Governor of Illinois.

I get your points, but I think you read too much into this.

The thrust of his work was, as I understand it, to show a correlation and some possible predictive utility. To expand this to a charge of bribery is excessive- even if the poisonous influence of money on policy is supported by a few centuries of experience. Or is that Millenium?.

Consider as well that, viewed from a purely economic point of view,  a decade or two of corporate contributions made to the right legislator (as defined by -chosen by- the best professionals in the business) is often money well invested in the game of influencing policy, for the corporation. A vote made today, without even a dime in recent payments, to advance that corporation's interests is a good investment for the legislator. Nothing will ever be proved, because the connection is filtered through the human psyche, and tenuous beyond modeling, so far.

As for investigations where evidence exists, ---"where it does, it's usually part of a criminal investigation"---in an alternate universe, perhaps. Or perhaps in Spain, till The Obama finds a way to tighten the screws enough.

Capitalism searches out the darkest corners of human potential, and mainlines them.

by geezer in Paris (risico at wanadoo(flypoop)fr) on Sat Jun 27th, 2009 at 03:42:03 AM EST
[ Parent ]
Consider as well that, viewed from a purely economic point of view,  a decade or two of corporate contributions made to the right legislator (as defined by -chosen by- the best professionals in the business) is often money well invested in the game of influencing policy, for the corporation. A vote made today, without even a dime in recent payments, to advance that corporation's interests is a good investment for the legislator. Nothing will ever be proved, because the connection is filtered through the human psyche, and tenuous beyond modeling, so far.

What this means is that political activism pays dividends, but that is true of all kinds of activism, not just organizing by corporate interests. It does not mean that the amount of financial contributions has any causal relationship to policy outcomes, and it does not even mean that the wealthy have advantages because of their financial resources, even if we all know that it must be true.  My point is that Nate's attempt to use statistical analysis to insinuate a cause-effect relationship between money from insurers and voting on public health care is flawed and shows no such thing for the simple reason that the statistical model he used is unable to show the direction of a relationship between variables. (There are, however, methods that can show such direction.) There is simply no reason to believe that monetary donations have affected voting on this issue based on the data presented.  

It is just as likely, based on Nate's data, that the legislators have always been against public health care and might have even influenced insurers to oppose it too instead of supporting public competition, just like they have somehow managed to convince insurers to donate to their campaigns year after year.  Nate is choosing only one possible interpretation of his regression without explaining why the others can't be possible.  He therefore shows no impact from lobbying on health care voting.

by santiago on Sun Jun 28th, 2009 at 02:29:58 AM EST
[ Parent ]
Corporations depend on arguments like yours that miss the context. They can sell a drug or product that they know is addictive, causes cancer, or whatever, but the burden is placed on the public to prove the causative link, rather than on the corporation to prove their non-existence.

The solution involves actions like properly funding and removing corporate influence on (in the US) the FDA and to engage in campaign finance reform, but as those are non-starters, and given the asymmetry involved, this is one case where I'm willing to make policy based on correlation alone.

you are the media you consume.

by MillMan (millguy at gmail) on Mon Jun 29th, 2009 at 01:48:15 AM EST
[ Parent ]
What matters is truth or fiction.  The facts don't provide any evidence that specific policy outcomes is in any way sensitive to the amount of campaign contributions.  Rather, ideological outcomes of elections that affect a broad array of policies are more sensitive, but that doesn't help explain much about the specifics of a public option or not.  If you want to halt corporate influence, then you've got look elsewhere.  Otherwise you're doing nothing to stop it.
by santiago on Tue Jun 30th, 2009 at 02:34:17 PM EST
[ Parent ]
That's something of a leap there: From saying that there is no evidence (which is true for certain values of "evidence") that contributions do not buy policy, to saying that contributions do not, in fact, buy policy.

The latter implies that there is good reason to believe that contributions do not buy policy, whereas the former merely says that there is no good reason (presented here) to assume that it does.

- Jake

Friends come and go. Enemies accumulate.

by JakeS (JangoSierra 'at' gmail 'dot' com) on Tue Jun 30th, 2009 at 04:05:45 PM EST
[ Parent ]
Fair enough.  The evidence presented by Nate's piece and the diarists' analysis of it don't make the case they believe it does - that money buys policy outcomes.  Other evidence might make that case.
by santiago on Tue Jun 30th, 2009 at 04:34:12 PM EST
[ Parent ]


Display:
Go to: [ European Tribune Homepage : Top of page : Top of comments ]

Top Diaries