The Journal of Things We Like (Lots)
Select Page

Katharina Eckartz, Oliver Kirchkamp, & Daniel Schunk, How Do Incentives Affect Creativity (CESifo Working Paper Series, Paper No. 4049, 2012), available at SSRN.

The classic justification for intellectual property laws was perhaps stated best by Abraham Lincoln, who, in speaking of the patent system, characterized its function as “adding the fuel of interest to the fire of genius.” Put less poetically, IP aims to encourage creativity by granting creators exclusive property rights in their creations. That way, if a patented invention or copyrighted work turns out to be worth money, the creator will benefit, rather than a copyist.

That sounds entirely sensible in theory. We think that people generally respond to incentives. Make gasoline more expensive by taxing it, and people generally use less of it. Give people a tax break on home mortgages, and they build more and bigger houses. Make creativity a little less risky, and the payoff a bit more certain, and we’ll get more investment in creative labor.

But is creativity really like gas and houses? Can the amount that we get of creativity be raised by providing incentives? The answer is bound to differ a lot depending on what sort of creativity we’re talking about. Poetry and drug discovery are both creative endeavors, but they are more different than alike. It is difficult to imagine piracy leading to poets throwing down their pens. But we may more readily imagine a drug company cutting its research budget if others are permitted to copy the pills the company spends hundreds of millions to develop. In between poetry and pharma there is a lot of territory. Where in this spectrum of creativity do intrinsic incentives to create leave off? And where do we need to start to introduce extrinsic incentives—either through prizes, or IP laws, or other schemes like government funding of basic research? On these questions, we know surprisingly little. Do incentives work? If so, when? And what sort of incentives are best?

A new paper by three European economists sheds some light on those questions. In How Do Incentives Affect Creativity?, Katharina Eckartz and Oliver Kirchkamp of Friedrich-Schiller-University Jena, and Daniel Schunk of the University of Mainz present the results of an carefully-designed and admirably creative experiment (hereinafter, the “E-K-S Experiment”).

The authors are interested in whether compensation schemes that pay people more when they perform better provoke creative effort, relative to payment of a flat fee that does not vary based on performance. That’s a narrow question, but, as so often is the case, one which is surprisingly difficult to test.

The E-K-S Experiment employs a word task as its measurement of creativity. Participants were presented with a series of alphabetically-ordered lettersets consisting of 12 letters—an example would be “accdeeeginst”—and asked to construct as many words as they could, using only the letters in the letterset, in five minutes. Longer words were deemed more valuable than shorter, in an amount disproportionate to the number of letters in the word.

Is assembling words from a letterset really a “creativity task”? The authors contend that the task is measuring a specific sort of creativity; namely, the ability to combine known elements within a set of pre-determined rules. Of course this is far from the only form of creativity, and it is possible that different sorts of creators might respond to incentives differently. But that’s not really a criticism of the E-K-S Experiment. No one experiment can capture all the facets of the huge range of activities we group as “creativity”. To be a valuable contribution, the E-K-S Experiment must only describe how incentives affect subjects’ performance in a credible model of one sort of creativity, and then later papers, or other researchers, can expand the findings to model many different types of creative work.

Back to the design of the E-K-S Experiment. The experiment was run with 216 student subjects. Subjects were first required to pass a German-language pre-test. Subjects were also asked to self-report their interest in participating in a range of creative tasks. This was done to assess subjects’ intrinsic motivation to engage in creativity.

Once these preliminaries were completed, subjects performed the creativity task three times, in “rounds” lasting five minutes. Each time the subject performed the task, the payment scheme shifted. The experimenters tested a flat fee scheme, a pay-for-performance scheme—where the amount of money earned increased along with performance—and a “tournament” scheme, in which subjects were divided into groups of four and the highest-scoring in the group earned the lion’s share (but not all) of the prize. Subjects undertook the creativity task within each of these conditions. The order in which the conditions were presented was varied to make sure that performance was not dependent on which condition the subjects completed first (that is, before they either got better at the task, or became tired and bored with it). After performing the creativity task under one of the payment conditions, the subjects were asked to perform a “control” task—i.e., a task, such as a number-adding test, that involves work but not creativity—under the same payment condition. In the final stage of the E-K-S Experiment, subjects were asked to pick one of the payment schemes they wished to repeat. This last stage was included to see whether there would be any significant differences among subjects regarding their choice of “favorite” payment scheme.

So, what were the results? They can be stated very simply: Variation in subjects’ performance was driven almost entirely by subjects’ individual characteristics—i.e., how good they were at the task. The payment incentive schemes had almost no effect on performance in either the creativity or control tasks. Subjects in the flat fee payment condition performed about the same as when they were in the pay-for-performance or tournament schemes. This was true whether subjects’ performance was assessed for complexity of the words they produced, or their originality. Aside from a small, positive effect on word length produced by the pay-for-performance scheme, all effects were too small to be statistically significant.

Subjects’ general indifference to incentive payment schemes carried through, moreover, into the final “self-selection” round. Flat payment schemes—i.e., those not related to performance—were subjects’ favorite on average, with 40.74% of subjects choosing this scheme. (Females favored flat payment more than males, who actually favored pay-for-performance very slightly.) The tournament condition was the least popular, with approximately 27% of subjects choosing it overall, and with women and men choosing tournament at roughly the same rate. This cuts against previous findings that male subjects, supposedly more risk-seeking than their female counterparts, favor tournament settings where most if not all of the rewards go to a winner, leaving the losers with little or nothing. So if the conventional wisdom says that men are from Mars and women from Venus, the results of this experiment suggest that, at least as far as creativity incentive preferences go, the sexes occupy pretty much the same territory right here on Earth.

What did stand out in the E-K-S Experiment was the tendency of higher-performing subjects to sort themselves in the “self-selection” stage into either the pay-for-performance or tournament scheme. This means that average performance was marginally higher in these two conditions vs. flat fee. Not, however, because the conditions created some differential incentive effect. Rather, because subjects had a relatively accurate sense of their own relative performance (even though the experiment was not designed to give them information about relative performance) and chose performance-based schemes when they were confident of their relative ability. But there is very little normative significance to higher-performing subjects’ self-selection. When these subjects were confined to a flat payment condition, they performed about the same. In short, this subgroup’s preference has little effect on their creative output.

So, what’s the payoff of the E-K-S Experiment for our thinking about IP law? Well, it’s complicated, but here’s a take. You can think about both patent and copyright as analogous to the E-K-S Experiment’s “tournament” payment condition. That is, if you set aside for a moment other compensation options—in particular the prospect of being paid a salary to engage in creative work—an inventor who patents his new machine, or a writer who copyrights her screenplay, both are entering into a market where there are a few (potentially big) winners and a lot of losers. Put differently, only a few patented inventions and copyrighted works are worth real money. The vast majority are worthless. Most people probably realize this. Does it blunt their willingness to enter into tournament markets?

There is a good deal of research into these “tournament” markets, and most of it suggests that people are quite optimistic about their chance of success, and are attracted to the prospect of a big payoff, even if any individual competitor is very unlikely to actually win it. The E-K-S Experiment, however, suggests otherwise. Subjects were no more creative in the tournament market setting. Nor, in general, did subjects favor that form of compensation over flat fee or linear performance-based incentives.

Which leads to the following hypothesis—and it is only a hypothesis, because the E-K-S Experiment suggests it but does not test it directly: In at least some settings—and at least for individual creators (the behavior of corporate creators is not modeled in the E-K-S Experiment)—getting paid a flat salary to do creative work is every bit as effective an incentive to perform as the prospect of striking it rich via a valuable patent or copyright. The hard part is figuring out when this is true. And why.

Download PDF
Cite as: Christopher J. Sprigman, Creative Incentives, JOTWELL (October 18, 2013) (reviewing Katharina Eckartz, Oliver Kirchkamp, & Daniel Schunk, How Do Incentives Affect Creativity (CESifo Working Paper Series, Paper No. 4049, 2012), available at SSRN),