Co-Compensation

Economists spend immense amounts of time ranking journals, partly to decide on monetary and non-monetary professional rewards, partly as pure gossip. There is some imperfect agreement on rankings.

Given that agreement, how should we credit coauthored publications (the overwhelming majority of papers)? For the same quality of paper, with N authors does credit get apportioned as 1/N-does an author of a paper with two authors get 50 percent credit; or is the credit more than that? I would think that in equilibrium credit would be apportioned as 1/N; otherwise I would put a friend’s name on my papers, she would put mine on hers, and we’d both get more credit in total. Two young guys I know say that they do exactly that.

Credit at 1/N seems reasonable and is consistent with an old piece of empirical research (Raymond Sauer, Journal of Political Economy, 1988). But one institution I know of gives you 0.71 credit for a two-authored paper, 0.58 credit if you are one of three authors, 0.5 if one of four (using a credit rule of 1/square root(N)). It just started offering a bonus of $23,000 for a sole-authored publication in one of five top journals. But if you are one of two authors, you get $16,250.

This seems absolutely nuts to me, as it gives explicit monetary incentives to multiply co-authors. Why write one paper and get $23,000 when you can share co-authorships with a friend and get $32,500? Since economists always rise to the challenge of responding to incentives, I would bet that this institution sees a lot more coauthored papers by its faculty than before. Better quality, maybe, although I’m dubious; more coauthors for sure.


Christopher Strom

An institution paying cash based on the 1/square root(N) formula?

Oh, this is a system just begging to be gamed...

Unless they feel that a paper's value increases with the number of authors.

BTW - To any authors at DH's mentioned institution: I can make my name available for a hefty kickback. How about a paper on ethics and incentives?

Eric M. Jones

This scheme, to have lots of names on a paper, is often handled by putting the chief author(s) first, then everybody else who had anything to do with it.

The whole business becomes more clear if you do something like what Google does: The publication is more valuable as other people quote it or link to it. If you publish it but nobody in your field ever thinks enough of it to refer to it, then it is not worth much.

keith

I roughly recall that in physical science, you really only credit the top three authors, and you give primary credit to the first author.

Martin

An interesting related question is: How is credit shared when one author is famous and another is not? For example, if some relatively unknown economists publishes with, say, Greg Mankiw, does he get less credit than he would have had he published it with a lesser known economist.

Perhaps readers would assume Mankiw did all of the great research, since everything Mankiw does seems to be good. On the other hand, maybe the unknown prof did all the real work--why else would Mankiw publish a paper with him?

A similiar question could be raised about graduate students publishing with professors. If the paper is great, do readers assume the prof did most of the research; and if it's horrible, do readers assume the grad student did most of it? If so, then publishing with a prof as a graduate student would be a horrible deal.

noah

@Martin

Of course the graduate student did all of the work, that's what they're there for. Without this "exploitation" by the professor, the student would never get published, and couldn't go on to become a professor and have her own graduate students to do all the research. I think untenured professors also do some research on occasion.

Caliphilosopher

Better quality through collaborative efforts? What's wrong with that?

DH is setting up a straw man argument - if economics is supposed to have influence on these things, then there will be a point or diminishing return - where the increase in authors does nothing significant to the quality (or decreases the quality) of the paper.

Jeff Brandt

The scheme of putting your name on a friend's paper, and your friend's name on your papers, only works if both people consistently release a similar amount of quality papers. Otherwise, the more prolific author would get less credit than he/she would have received and the less prolific author would get more credit. Even though there would be more overall credit given, the person who wrote more would lose out.

However, given credit of 1 for a sole author , and 2/3 for each of 2 coauthors. It would make sense for the more prolific writer to go along with the scheme as long as the ratio of quality papers written did not exceed 2:1.

Joe Smith

The economists in the group may not be familiar with the famous Alpher Bethe Gamow paper which raised interesting questions about "authorship" and credit:

http://en.wikipedia.org/wiki/Alpher-Bethe-Gamow_paper

As a student I had the privilege of working for a young math professor who was very generous, with me and others, in giving co-authorship credit and I have always appreciated it. Whatever it cost him in the short run it has paid off over his career in the gratitude and respect of his peers.

Caliphilosopher

#8 -

Extremely well said.

Paul

Reminds me of a friend's idea for the academic chain letter...add the person at the top as a co-author on your next paper, add your name to the bottom, and send off to 5 colleagues (preferably in wildly different fields).

science minded

It is not always easy working with others. I tried collaborating with a friend (with whom I had been talking about science for more than 30 years. I had developed writer's block and so met one on one with him for about 2-3 days and the conversation was helpful in the long run. But in the end, I found myself unable to do science this way. My friend's personality was too overbearing and my own need to work this out for and by myself too powerful. So the collaboration ended. I still have reason to consider my friend's contribution to my own work as important (necessary) and will credit where credit is due.

This is not to say that all collaborations don't succeed and that they all are the same. . One of my first research jobs was in the research department of an insurance company. There were three of us and because we worked so well together, we were able to write a paper that was the result of our throwing around our different ideas and we succeeded. Perhaps it is a matter of the circumstances. I know of another successful collaboration between two colleagues.. But In that case, the author's worked independently and held quite different views. They would send each other materials by mail and, despite their differences, apparently had developed a good working relationship .

And then my husband and i have collaborated on numerous occasions- of writing ads and promotional materials based upon the research that I did or that we both did..I must admit- it has always been a great experience.

And come to think of it, if I had to list everyone who made a contribution to the book that I am now writing , there would be too many people to name whose ideas in some way influenced my own. So at what point does the individual take credit for the work they did whcih is somewhat based upon what others already have accomplished..

Read more...

Andy

I have a comment and a question.
Comment: The idea of cooking up an ultimate citation formula for measuring or incentivizing scientific research seems to me largely fruitless. The main issue is this: What if you had perfect data about the whole future of every detail of the universe and could calculate anything? What then would define scientific impact? That is, all of these crude citation counting measures are presumably attempts to measure some approximation of that ultimate thing, but no-one has a good idea of what that ultimate thing is. Whatever the ultimate measure, probably H-index is closer to it than is the total number of papers, say. But resolving finer than that with co-author counts and so on seems fruitless.

Largely unrelated question: Now that you are on to the economics of academics here is a question. Does anyone know a reasonable scheme for partitioning teaching effort within a department? Some courses are easier to teach than others. Some are more fun than others. Is there a reasonable bidding scheme for partitioning teaching and other departmental service?

Read more...

jorge

The incentives from the university's standpoint are often quite clear. They prefer to have a professor with 20 joint publications to the same prof with 10 singly authored ones. Indeed, even the ability to co-author with other successful researchers is itself an important signal about networking ability and will increase the fame of the professor and his institution disproportionately to any actual contribution.

Norm

Hehe... reminds me of this story from the Annals of Improbable Research

http://improbable.com/2009/01/22/guardian-column-141/
More Scientists Join Gangs

More and more, more and more scientists are ganging up to write research studies. It's no longer unusual to see a paper that lists more than 500 - that's five hundred - co-authors.

...

If there were a prize for the largest number of co-authors, it would have gone to the 2,512 people credited with writing Precision Electroweak Measurements on the Z Resonance, which appeared in the journal Physics Reports in 2006. That's a mild elevation from the previous record of 2,458 co-authors, attained just two years earlier…

brian

This is one market that will, much sooner than later, prove efficient.

Juan-Camilo Cardenas

I'd like to see the evidence supporting the scheme of academic friends adding each other's names in papers to augment the bonus of each other.

It is plausible, yes!, it is just wrong and have never heard of it in academia. Smart people who publish in top journals have more interesting or easier ways of making money.

Besides, we build and maintain social norms to make collaboration an co-authorship a productive scientific exercise where there might be some increasing returns to scale to co-authors.

Juan-Camilo Cardenas

Jennifer

As pointed out by a couple other commenters, in most other fields, one can tell who to give the most 'credit' to by the order of author names. However, in economics, the convention is to list authors alphabetically, rather than by contribution. Only if I see author names OUT of alphabetical order do I assume it means the 1st author really did a significant amount of the work. Of course, I've also always wondered if putting author names in order or contribution creates a lot of tension for people in other fields that economists are able to avoid.

mbp

responding to 4 and 5

I was once working in a field where the custom was always to list the graduate students first, then the professors, with any well known professor listed last. I don't know what (if any) purpose that custom served, particularly if it was a well known convention.

If I was looking to hire someone, I think I would rather have someone who made a serious contribution collaborating with colleagues on two different problems, rather than someone who did one thing by himself/herself. That of course assumes that the co-authors actually did the research together.

Casey Barker

I work within a very similar (easily gamed) incentive system, but this one has interesting legal implications: Patents.

I'm an engineer, and I'm paid a good salary to solve problems and implement solutions. We're always coming up with new inventions in the line of work. However, my employer wants us to file for patents, which they can use to fend off competitors, so they created an incentive payment system.

The incentives vary at each step of the process (initial disclosure to a filtering committee, filing with the USPTO, and patent grant), with the bulk of the payouts at the filing step. This actually makes sense because, once filed, patents can take many years to grant, if they grant at all, and much of the work is just in getting the invention written up and filed.

The issue comes with determining inventorship. Inventorship has a legal implication. Each of the listed inventors must be able to identify some portion of the patent that they contributed (usually a "claim"). Likewise, everyone who contributed a claim must be listed as an inventor. If the inventorship is not correct, the patent could be invalidated under prosecution, making it worthless to the company.

This is a hard problem: How do you design an incentive system that encourages inventors to A) file for patents, B) include everyone who contributed to the idea, and C) not include anyone who didn't contribute?

My current company pays each inventor the same bonus for up to three inventors per patent filing. For filings with more than three inventors, the inventors evenly split a pool of money equivalent to three inventor payouts.

Now, these are engineers we're talking about -- people who spend their lives figuring out how to optimize feedback loops. How many inventors do you suppose end up on virtually every filing? Yeah, three. If four or more inventors really contribute, they usually end up on the filing. (Most folks know better than to leave out genuine contributors, even if it means a fractionally smaller payment, for fear they themselves might be ostracized on future filings. Sometimes, filings get packed with many inventors for fear of leaving someone out, especially if the idea "evolved" in a large meeting.) But virtually nothing gets filed with fewer than three inventors. I've even seen deals brokered amongst engineers so that a "minor" contributor gets to claim inventorship in exchange for handling the mundane task of writing and reviewing the filing documents with the lawyers.

Adding to the maddness, when an engineer is granted a 10th or 25th patent, they get a large (or very large) bonus and special recognition. Plus, once a year, the "patent of the year" is selected, and all the inventors on it receive substantial bonuses and recognition.

This whole thing gets gamed pretty badly, but if you *really* want to see a gamed system, imagine the multi-department committee of peer engineers who review the initial invention disclosures to determine which inventions will get pursued for filing!

Unfortunately, I don't know how to design a better incentive system, but for now I'm happy to benefit from this one.

Read more...

a prof

In my previous econ department, we used (1.5)/N to reward each co-author. Two authors would get 75% of the benefit of a single authored paper. Four authors would each get 38%. I have to think about why this didn't lead to phantom coauthors...