Winner of the New Statesman SPERI Prize in Political Economy 2016

Wednesday 6 August 2014

Bonus Culture

The comments I received on my post about a maximum wage were interesting for many reasons. In this post I just want to focus on one common misperception. This is that a maximum wage, because it interferes with the market, must be distortionary and therefore something that should tend to be avoided. This in turn assumes that the actual economy approximates the competitive efficient allocation students learn about in Econ 101, which is what neoliberals would like people to believe.

To see why this might well not be true when it comes to high pay, I want to briefly discuss a paper (NBER version here) by Roland Bénabou and Jean Tirole, which has the title of this post. I will try and make my discussion as non-technical as possible.

The basic idea is as follows. Jobs involve two types of activity or output, one of which is measurable and one is not. To reflect this, pay involves two components: a fixed component, which the worker always gets, and a performance element (the bonus) which depends on the worker’s measured output. The firm wants the worker to undertake both types of activity, so it sets the appropriate relationship between the two elements of pay to make sure the worker puts the right amount of effort into each type of activity. I will talk about what these two types of activity might be in practice below.

So far, so good. But now we add an information problem. There are two types of workers: the ordinary and the talented. The talented worker is much better at producing the measurable output. The firm cannot tell the two types apart, but it would naturally like to attract some of the more talented workers. One way it can do this is to offer two types of remuneration package: a ‘low bonus’ type and a ‘high bonus’ type. Talented workers will be attracted to the high bonus package, because their talent means that they can achieve high measured output (and therefore high pay) with relatively little effort.

This is useful for the firm, but it creates a problem. The bonus payment is now doing two jobs: allocating worker effort between activities, and attracting talented workers. In these circumstances, the bonus payment can depart from its efficient level. In particular the paper shows (p13) that in a competitive labour market, bonus payments designed for talented workers will be too high, in the sense that they lead to these workers putting too much effort into the measured activity, and too little into the other activity.

So what might these two activities (one measurable, one not) be in practice. The paper suggests, for measurable activities, things like sales, output, trading profits, and billable medical procedures, and for immeasurable activities things like intangible investments affecting long run value, financial or legal risk-taking, and cooperation among individuals or divisions. So the problem is that, in an effort to attract talented workers, the firm over incentivises effort on achieving tangible short term goals at the expense of work on intangible, longer term objectives.

The relationship to my discussion of a maximum wage should now be clear. To quote from the paper: “Turning to policy implications, we show that a cap on bonuses can restore balance in agents’ incentives, and even re-establish the first best, as long as it does not induce employers to switch to some alternative “currency” to screen employees.”

If you want to think about how this idea relates to alternative models of executive pay and competition for talent, I would encourage you to read section 1.1 of the paper, which is not too technical. The paper also contains a lot more that will be of interest to economists.

The general point I want to make is this. We can think about the minimum wage as an unfortunate interference in the market which can nevertheless be justified on equity grounds, or as a means of reducing poverty. However we can also see it as a way of increasing the efficiency of the economy, because many employers of low paid workers can exploit their monopoly power to pay wages that are below the efficient level (monopsony). Exactly the same may be true of the maximum wage. It could be that top pay is inefficiently high because executives have monopoly power, or it could be as this paper suggests because the firm wants to attract unobservable talent. I see no reason to presume that the dramatic increase in top pay reflects increases in the productivity of those workers.



  1. Off topic of this particular posting I know, but in case you didn't realise, an evidently economically illiterate sophist on Conservative Home webpage has attempted to take you to task on your previous post regarding UK GDP figures and falling productivity.
    Stating that as one of those neo-Keynesian economists who have been now proved so wrong about macro policy, you are apparently desperately scrabbling around for some bad news to seize on from the deluge of good economic data, and there is no reason at all to be concerned about low productivity at all!

  2. Not all talented people respond to bonuses well; some find them anathema. As for top end salaries, it could be argued that what they attract isn't talent, but the most greedy. Is that really a good thing?

  3. This sort of modelling is fun. I've done similar stuff myself. But many have drawn this correct conclusion about the perverse effects of high pay and bonuses without the modelling. Why was this common-sense reasoning ignored? Are economists and their clients (all of us) handicapping themselves by demanding mathematical proofs of the 'bleedin' obvious' before acting?

  4. You can of course set up theoretical models like this to 'show' almost anything.

  5. This was long so I have truncated it but I have lost the link, it was from the New Scientist some years ago.

    Editorial: "The bonus pay paradox"

    If you want to boost people's performance, don't bank on bonuses

    BONUS culture has come under intense scrutiny since the ongoing financial crisis began in 2007. Many people have been outraged by the way some bankers and top executives seem to have been rewarded for failure. Others find the idea of multimillion-dollar bonuses morally abhorrent.

    But few have asked whether performance-related bonuses really do boost performance. The answer seems so obvious that even to ask the question can appear absurd. Indeed, despite all the fuss about them, financial incentives continue to be introduced in more and more areas, from healthcare and public services to teaching and academia.

    "Economists and workplace consultants regard it as almost unquestioned dogma that people are motivated by rewards, so they don't feel the need to test this," says Alfie Kohn, a teacher turned writer. "It has the status more of religious truth than scientific hypothesis."

    So it may come as a shock to many to learn that a large and growing body of evidence
    suggests that in many circumstances, paying for results can actually make people perform badly, and that the more you pay, the worse they perform.

  6. If there was an efficient market in executive pay then surely pay levels would fall in response to increased supply?

    Executive pay is an insiders racket......

  7. What you'll get here is the complaint that footballers will be driven from the UK (see Krugman blog Piketty Day Notes April 16, 2014 that culture, media, and sport makes up only about 3% of top earners in the US).

    In the US United for a Fair Economy put an imaginary $1,000 dollars 1993-2000 in each of the ten S&P 500 companies with the highest paid CEOs: general investment of that $10,000 in 1993 in the S&P 500 by 2000 had risen to just over $32,000, whereas those rolling ten companies with the highest CEO pay would have led the investment to be worth in 2000 just over $3,500 (Leviathans (2005), pp153-4).

  8. "So the problem is that, in an effort to attract talented workers, the firm over incentivises effort on achieving tangible short term goals at the expense of work on intangible, longer term objectives."

    What you are saying here is that bonuses themselves are not bad, but rather they target the wrong things. Instead of using convoluted logic to argue for a market intervention in wages, why not target the structure and incentives created by bonuses? Longer deferral and better selection of targets could more effectively align "longer term objectives" with the workers own incentives.

    Of course, you also argue that these long term objectives are "intangible." This is a red herring. With a long enough deferral/clawback period, it is perfectly possible to target less tangible goals such as long term safety of a bank. For example, greater emphasis on manager discretion. As for completely un-measurable "intangible" objectives - output that is completely unobservable and unmeasurable is probably not output at all.

  9. A lot of the problem, it seems to me, lies in measurement. It's relatively easy, as we have noted before, to measure the financial value of e.g. pop stars: their records/concerts either sell or they don't. Harder to measure sports stars: are improved performance/ticket sales/sponsorships due to the individual or the team around the star? Harder still to measure the investment or trading banking star, with the added problem of distinguishing between short and long term results, taking account of potential externalities (toxic securities, collapsed banks ...) and nearly impossible for major company CEOs - literally thousands of support staff who may be better directed by the 'star' but may also be busy compensating for his/her errors.

    A rather long note on measurement problems follows:
    I've spent a lot of my professional life trying to set SMART objectives for policy interventions, investments etc. and I'm much less confident in the approach than I was when I started.

    As your summary of the article says, work (also policy, investment, everything really) comprises measurable and non-measurable components. As we cannot, by definition, objectively measure non-measurable components we should (the argument goes) focus on identifying those measurable elements that are the best proxies for the overall objectives.

    So far so good, and an approach I have used a lot BUT I am more and more convinced that if we focus on the measurable we ignore the immeasurable. The result is a kind of generalised application of Goodhart's Law (any indicator of money supply used as an instrument loses its value as an indicator) in that the more incentives are focused on specific indicators the less value those indicators have as proxies for the overall purpose.

    What, in the end, this means to me is that while objective and even SMART indicators have their place, quality is less measurable but still critical. This is not, in fact, a counsel of despair. Despite the inherent subjectivity, my (limited) experience in recruitment and in exam marking suggests that experts and even lesser experts will generally reach a pretty close consensus on all but the outlier perfomers. This is not perfect, and there is a ream of literature on the various biases involved, but if we accept that there is also bias, or at least error, of a different form in objective assessment then it is possible to move to a 'least bad' outcome. In the case of high salaries/bonuses, this should lead to a more modest approach that recognises that the individual will not, and should not, capture all the value he/she is hopefully adding but that there is still room for some differentiation.

  10. "In this post I just want to focus on one common misperception. This is that a maximum wage, because it interferes with the market, must be distortionary and therefore something that should tend to be avoided. This in turn assumes that the actual economy approximates the competitive efficient allocation students learn about in Econ 101, which is what neoliberals would like people to believe."

    The market is amoral. Maximum economic output does not automatically mean maximum utility for members of society (believing that it does would be the real misperception). Even if CEOs were so productive (on their own, not just through holding a strategic position in a productive organization) that it actually justifies making many millions per year, it could still be that the net effect of giving them so much money is bad for society, in fact it probably is bad (it leads to political corruption and reckless risk-taking). Whether bonuses work or not and how productive CEOs actually are, it all doesn't matter since a high level of income inequality is simply a threat to a democratic society, that's all the reason anyone should need.

    P.S. "neoliberal" is nothing but a pejorative. There are no people who call themselves neoliberals, it is an ill-defined term that has come to mean something along the lines of "some bastards who are to the right of me on economic matters".


Unfortunately because of spam with embedded links (which then flag up warnings about the whole site on some browsers), I have to personally moderate all comments. As a result, your comment may not appear for some time. In addition, I cannot publish comments with links to websites because it takes too much time to check whether these sites are legitimate.