Ashwini Versus the A.F.L.-C.I.O.

A modern variant of David versus Goliath is unfolding in academic economics.

Ashwini Agrawal is David. He is a graduate student getting a Ph.D. at the University of Chicago Graduate School of Business.

The A.F.L.-C.I.O. is Goliath.

Here is the background, as I understand it:

The A.F.L.-C.I.O. manages huge pension funds for its members. These pension funds are invested in publicly traded companies. When it comes time for shareholder votes on issues like who to appoint to the board of directors, the A.F.L.-C.I.O. pension fund managers get to cast votes just like any other shareholder.

The question Ashwini asks is: how do the A.F.L.-C.I.O. pension fund managers vote? To answer that question, Ashwini exploits a neat natural experiment.

In 2005, the A.F.L.-C.I.O. splintered, with a big chunk of the members leaving to form the Change To Win coalition. Ashwini looked at whether the shareholder votes cast by the A.F.L.-C.I.O. pension fund tended to differ when the union workers at the firm were no longer A.F.L.-C.I.O. workers.

In his dissertation, Ashwini found that the behavior of the A.F.L.-C.I.O. pension fund did seem to change. In particular, it looks like the A.F.L.-C.I.O. pension funds were voting in ways that might help workers when they were A.F.L.-C.I.O. union workers, but not after they switched union affiliations.

My own reaction to the findings is that they make sense. Why shouldn’t the A.F.L.-C.I.O. pension fund use its clout to try to make things better off for A.F.L.-C.I.O. workers?

I guess some would argue that the responsibility of the pension fund managers is to maximize the value of the pensions, although I don’t really see why. I would think the goal of the A.F.L.-C.I.O. would be to maximize the welfare of the union members. Sacrificing a little on the portfolio returns to help workers in other ways makes good sense to me.

The A.F.L.-C.I.O. doesn’t seem to share my view.

When Ashwini’s results got discussed not once but twice in the Wall Street Journal and on a Harvard Law School blog, the A.F.L.-C.I.O. had an angry reply.

Ashwini responded, as did his advisor Steve Kaplan.

Which led the A.F.L.-C.I.O. to fire back again.

Now, I have to confess I’m not a completely unbiased party to all of this — to begin with, I like underdogs. Secondly, Ashwini took one of my classes, and I first found out about this when he came to my office terrified that the A.F.L.-C.I.O. was going to sue him and find ways to ruin his career (hopefully not the way the Teamsters ruined Jimmy Hoffa‘s career).

I told him that as long as the university would pay his bills, getting sued by the A.F.L.-C.I.O. would be the best thing that could ever happen to his career.

Having read through this mess, I think the A.F.L.-C.I.O. just isn’t being very smart about this. If you want to show that an academic paper is wrong, saying that a graduate student is lying about contacting your organization — as the A.F.L.-C.I.O. said — is completely the wrong approach.

What the A.F.L.-C.I.O. needs to do is hire an economist to reproduce what Ashwini has done. If he made some mistakes, then the A.F.L.-C.I.O. can make these public and the issue will go away.

If he didn’t make mistakes, then either the A.F.L.-C.I.O. has to admit that they use their formidable power in every way possible to help workers, or they need to institute a different set of rules about how they vote on shareholder issues.

Round one goes to David.


Prof. Levitt, #20 VXN answers your question most directly about why the pension trustees can't seek to maximize worker welfare through proxy votes. #11 coyote and #14 scott also get at it.

I don't know the details for pension fund managers but I work at an environmentally responsible mutual fund where we have the same issue. We want to vote the proxies in an environmentally responsible way even though those votes may differ from votes that maximize shareholder value. Yet our trustees are bound by fiduciary duty to our fund shareholders. In the mutual fund world we have some leeway: we may publish a set of proxy voting policies and then we may vote in line with the stated policies. I don't think the pension fund managers have similar leeway, though, as the situation is entirely different - the workers aren't able to choose among pension funds, so they haven't opted-in in any meaningful way.

Someone who knows the rules should post. I suspect it's in ERISA and elsewhere.

I hope the SEC leaves this one alone. I don't want to go through another seventeen rounds of fighting over proxy voting rules right now, I'm exhausted from last summer still. And I wish all the trustees could make political decisions about their votes with the pension fund money. I wish there were some mechanism by which that could be ethically supportable (have workers, former workers & retirees opt-in to a politically active sub-fund? But what an administrative nightmare!)


Kevin H

I think this is an interesting example of the difference between politics and science. The political answer to this problem try to smear your opponent, where as the scientific answer is to actually deal with the facts themselves.


Daniel @4 and @8:

Agrawal's paper clearly delineates the method by which the results can be obtained. Last time I checked, in this country, the burden is on the "accuser" to prove wrong doing. Hire an economist, follow the recipe, and if the results don't match call him back.

The approach of the AFL-CIO has shown that their primary intent is to discredit Agrawal at all costs. Not only have they been malicious in their approach, they have also proved themselves to work in a dishonest fashion.

This was a monumental blunder on the part of the Union -- especially over an issue which seems to indicate that they are doing a reasonable job for their constituents. The issue should have been a non-starter.


Ivan @5: Great point. I would add that having your paper vetted on the global academic job market is probably a deeper assessment than peer reviews. It's really just a "Wisdom of crowds" argument: if Agrawal presented this work at many top places, as I understand he did, that certainly is a better set of collective eyes than the 3-4 reviewers who would be assigned to a journal article peer review.


The fiduciary responsibility of the pension plan trustees is to maximize the investment return of the plan while minimizing the risk to the plan. Their responsibility is not to more broadly maximize the well being of the AFL-CIO membership. That is for the union leadership overall and is not a role of a pension plan trustee.


Is it just me, or does the post by MS seem awfully union-y in grammar and tone? I think it might be the first time that I have seen anyone use the word ain't on this message board.


I think there is a fundamental missunderstanding of what the AFL CIO is, it is not a labor union, it is rather an association of labor unions. They may advise labor unions who have established pension funds and health and welfare funds with investments in the various financial markets around the world. The employees of the AFL CIO may contribute to a pension fund managed by another union, like the OPEIU, but I would be surprised if the AFL CIO had their own fund, and I don't actually know if they do. It would seem to me to be an expensive proposition to operate a fund for a small number of employees. With respect to other funds they don't make decsions directly about what actions boards of trustees take on their pension funds. I know I am an officer of a labor union and a trustee on a pension fund. The board of trustees makes decisions that will, consistent with our fiduciary duty, maximize returns to be spent for the benefits our fund has promised the pensioners and participants of the fund.

I find all this bit about the motives of the "AFL CIO" amusing and annoying at the same time. It displays the general ignorance of the public at large about organized labor.



"hire an economist" -- or someone else who can handle data collection and multivariate regression. replicating this work would be time consuming but not that hard, and by no means requires someone to know what the envelope theorem is.


I hope the AFL-CIO leadership gets into real trouble because of this. That will be a more than deserved punishment for using false argumentation against the FTA with Colombia. There migth be good-real arguments to go against it; but to play dirty with the reputation of a whole country that is trying really hard to overcome problems they can't even dream about, is very irresponsible.


Two thoughts:
Non-disclosure agreement. If the AFL-CIO wants to review the work, they must sign a non-disclosure agreement. Such an agreement would protect Mr. Agrawal's work from being "stolen" from him.

I am reasonably sure that with the media this has already gotten, the AFL-CIO would not want to face the penalties written into a non-disclosure agreement (in fact, a crafty lawyer could write in a few million dollar penalty if they disclose any of the information).

Then again, a monetary figure built into a non-disclosure may prompt the AFL-CIO to violate the non-disclosure agreement. Hence, ensure that the court fees, the lawyer fees, and a "lottery" is built into the NDA. That way, lawyers will want to win it for you, courts would have to award it to you (viable contract), and for all of that trouble, might as well retire in luxury!

On the other hand, the AFL-CIO person may be able to release data for the research and actually disprove it. If that is the case, ensure that no NDA is signed by Mr. Agrawal. While this would be a total surprise, it still has a remote possibility of happening.



The statement "we need to review the accuracy of Mr. Agrawal's data and statistical model, and when given the opportunity to talk to him, inform him of the serious flaws in his research." seems odd as the AFL-CIO rep comes to a pre determined conclusion about the research. Where is the open mind?


My guess is that they fear federal rules for pension fund managers. I am not at all familiar with these rules, but it would not be a stretch to imagine that such rules require pension fund managers to focus solely on the risk/reward of the portfolio managed, without reference to how such decisions might affect organizations with which she/he is associated


Great topic.

I would love to hear a response from Richard Posner or Wesley Wildman.


I think the AFL-CIO is worried because if this paper is correct, they may be getting an SEC/FINRA/Relevant Authority investigation. If the pension accounts of former members perform worse in a statistically significant way, then there may be a breach of fiduciary duty--they're playing politics with a retirement fund. That's one of those things that people get in lots of trouble for.


re: Luke #6

Fair enough, but it is the way in which one chooses to assemble statistics and produce the appropriate algorithms that discrepancies may arise. If I really wanted to, I could find statistics that show reading the Freakonomics blog makes you more likely to eat spam . Hopefully, someone would question the validity of my methodology.


The idea that the AFL-CIO would not be able to compile the data needed to reproduce the analyses presented in Agrawal's paper is beyond ludicrous. It is THEIR OWN DATA. Every director vote, every bit of information about union representation at the companies included in the AFL-CIO pension portfolio -- all this is information that the AFL-CIO should have readily at its disposal. Not necessarily all compiled nicely and ready to work with, but surely not terribly difficult to obtain. All intra-office phone calls/emails.

Asking the researcher to release his dataset, when they HAVE the data in question is just silly. If he made mistakes and got some numbers wrong (say, when he relied on information about union representation from the Investor Relations dept), they can rebut him on facts. They don't need his data set to do that.


I think you may have mischaracterized things a bit. I glanced at the study. It appears your colleague only looked at the STAFF pension fund. This is very different than (I'm guessing) the MEMBERS of the AFL-CIO (of course some staff probably are members, and some are not as the paper mentions). Presumably the AFL-CIO has many members that are covered by other pension arrangements.

Secondly, I wonder how your colleague controlled for the fact that many pensions were likely changing the types of asset classes during this time period (2003 was a pretty bad year for most pensions, though things generally improved over this period)?

Anyway, an interesting look at proxy voting, but I think your post may have overstated the impact or generalizability of the study.


re daniel @ 4:

the problem with releasing the data, as stated by steve kaplan, is that mr. agrawal put in a lot of work converting the data from a raw public source to a usable format. as a result, he is entitled to profit his work, most likely in the form of getting the paper published in a journal.


AFL-CIO is acting as expected. Not attacking the actual methods and numbers in A.Agrawal's work, but instead attacking the credibility of his findings. It has been clearly stated that A. Agrawal acquired his data from publicly available sources so AFL-CIO has no merit about the "non-disclosure" of data.

Secondly, with lack of peer review, it is true, these findings have not been published yet, but A. Agrawal's work has been examined by nation's top economists. This perhaps does not add it the amount of credibility that a publication in a peer review journal would have, however it would be inaccurate to say that this data has not been carefully scrutinized by A. Agrawal's peers.


To Andrew:

Not necessarily. The pension plan covers those who are currently employed in union positions, and those who are not. If the pension plan votes in board members who are not union friendly, it could have a negative impact on currently employed union employees. While this could be good for the company (and thus, the share price, and the pension plan), it may not be in the best interest of the current union workers.

Of course, if the board which is not friendly to unions keeps the company in business where it might have otherwise failed, the employee is still better off.