29

In this animated video of his talk, Dan Pink claims performance incentives do not work, and can actually lead to worse performance.

In summary, he quotes two studies (which he unfortunately does not name) where people are asked to do a task, and receive a higher bonus if they do it better. What they both find is that as long as the task requires at least rudimentary cognitive skill, the bonus does not lead to improved results. In fact, if the bonus is high enough, it actually leads to people performing worse.

Does this view present the current scientific consensus? Does anyone have any links to good studies (preferably meta-studies) on this subject?

Christian
  • 33,271
  • 15
  • 112
  • 266
knatten
  • 391
  • 2
  • 4
  • 1
    How do you define and measure "if they do it better"? Basically this means introducing metric, and smart people will play the system to maximize these metric. Yet metric don't show you whole picture, not even significant part of it. For example it's been commonly known, that rewarding software developers based on metrics totally hinders productivity. Not sure if there are any formal studies to prove that though.. – vartec Jun 21 '12 at 10:03
  • @vartec: I haven't watched the video for a couple of years, but my recollection is that they failed according to the specified metric - i.e. the one that the person should have been explicitly maximizing. – Oddthinking Jun 21 '12 at 11:47
  • Non-animated version [here](http://www.youtube.com/watch?v=_mG-hhWL_ug) where he gives references. – Oddthinking Jun 21 '12 at 11:57
  • 7
    Paper #1: D. ARIELY, U. GNEEZY, G. LOWENSTEIN, &. N. MAZAR, [Large Stakes and Big Mistakes](http://www.bos.frb.org/economic/wp/wp2005/wp0511.pdf) Federal Reserve Bank of Boston Working Paper No. 05-11, July. 2005; NY TIMES, 20 Nov. 08 – Oddthinking Jun 21 '12 at 12:01
  • 7
    Paper #2: URI GNEEZY and ALDO RUSTICHINI, [A Fine is a Price](http://rady.ucsd.edu/faculty/directory/gneezy/docs/fine.pdf), Journal of Legal Studies, vol. XXIX (January 2000) – Oddthinking Jun 21 '12 at 12:05
  • @Oddthinking post an answer, already :-) – Sklivvz Jun 21 '12 at 12:08
  • @Sklivvz: :-) I am fleshing out the missing links, because it struck me as odd than Dan Pink wouldn't provide references. However, the question is much bigger - it asks about scientific consensus and I remain ignorant about that. – Oddthinking Jun 21 '12 at 12:29
  • Are you a starving Person in India? How accurate a study reflects in regard to you is highly dependent on the group being studied (cohort vs cross-sectional study). – user1873 Jun 21 '12 at 12:43
  • @vartec: My question really is: Will it provide value to the organization? There are two issues I see here: 1. Is it possible to make measurable goals that measure the right thing? 2: If so, will tying these to bonuses help people reach them? I think Dan in his talk says that even if 1 is possible (it is at least for the simple studies) 2 is still false. In software, my experience is that 1 is also false, making the situation even worse. My area is software development, and I would really like to see some studies on this. – knatten Jun 21 '12 at 13:37
  • @Oddthinking I think Dan didn’t mention the papers because this doesn’t fit the format of a TED talk, and as far as I know the papers are incredibly well-known in the relevant fields anyway, so no need to mention them. – Konrad Rudolph Jun 21 '12 at 16:18
  • @Konrad: The animation was actually based on an RSA talk, not a TED talk (see link above.) In that talk, Pink had some slides which included the references in small print (which is how I found them). So, he originally did the right thing. The animator edited down the talk and also left out those details. – Oddthinking Jun 21 '12 at 16:22
  • @Oddthinking I had assumed that RSAnimate just used the voice-over from the TED talk (which apparently they even did). I didn’t think of the slides. – Konrad Rudolph Jun 21 '12 at 16:26
  • 7
    @vartec - the "smart people will game any metric" is an axiom for anyone who found StackExchange by following Joel Spolsky's blog :) What more studies do you need if Joel says so? :) – user5341 Jul 11 '12 at 10:52
  • The problem is in measuring performance. It is incredibly complex to measure the value of an employee. And when it comes to these type of studies, is is invariably done wrong. The researchers want to measure "objective" performance which can't be done. If this kind of thing is to be attempted at all, the only possible way to do it is to find a top notch manager who knows that business and ask him how employees are performing. Of course this won't be done because it is not "objective". – ax123man Mar 20 '15 at 13:21

1 Answers1

4

The relationship of financial incentives to performance quality and quantity is cumulated over 39 studies containing 47 relationships in the study 'Are Financial Incentives Related to Performance? A Meta-Analytic Review of Empirical Research' (http://www.researchgate.net/profile/Atul_Mitra/publication/232559540_Are_financial_incentives_related_to_performance_A_metaanalytic_review_of_empirical_research/links/0f317533773cf0d1ea000000.pdf). Per this metaanalytic study, financial incentives were not related to performance quality but had a corrected correlation of .34 with performance quantity. Setting (laboratory, field, experimental simulation) and theoretical framework moderated the relationship, but task type did not.

Overall, this study underscores the generalization of positive relationship between financial incentives and performance. It emphasizes the wisdom of designing incentive systems carefully; it also highlights the utility of including financial incentives as integral components in theoretical frameworks of organizational behavior and the management of human resources.

pericles316
  • 22,676
  • 2
  • 84
  • 161