The social impact of different professions

Posted by on

Economists and Harvard and Chicago recently published a paper1 that contains a number of estimates of the social value produced by different professions per dollar of salary. The estimates aren’t the core aim of the paper, but are nonetheless fascinating.

The first set of estimates are by one of the authors of the paper, Lockwood, and aims to stick to views that would be typical based on the the economics literature:

Profession Lockwood’s estimates (additional social value produced per $ of salary at the margin)
Academia/research 2
Advertising/marketing/sales -0.3
Agriculture 0
Arts/Entertainment 0
Business operations 0.1
Engineering/technical 0.4
Entrepreneurship 2
Financial Services -0.5
Law -0.5
Management -0.2
Medicine 0.1
Public service/military 0.4
Real Estate -0.1
Teaching 4

Looking at the first figure, a value of 2 for academia means that for every $1 of salary an academic earns, a further $2 of value is created for the rest of society. One way to understand this figure is that the rest of society should be willing to pay a further $2 for the services delivered by the academic, meaning the academic only captures 33% of the value they create for themselves.

On the other hand, in financial services for every $1 of salary, $0.50 is lost by the rest of society i.e. it’s extracted as rents. For arts/entertainment, each person captures all the value they create, and there’s no broader spillovers.

What are these estimates based on? The paper says:

These views arise from taking very seriously, and perhaps to a bit of an extreme, arguments made commonly within the economics community for externalities from various professions. These include the enthusiasm for entrepreneurship and engineering and skepticism of law pushed by Murphy et al.2, the emphasis on the importance of research central to the modern growth literature,3 evidence that teachers are dramatically under-compensated,4 concerns about the social value of advertising,5 a widespread recent concern that much activity in the financial sector is zero-sum racing over information6 or may actually be value destroying78 and, most importantly, worries about managers devoting significant time to defrauding their shareholders rather than creating value for them9 and shareholders failing to correspondingly reduce pay due to competitive pressures.10

(We’ve listed and archived all these papers in the notes in case you’d like to read more).

Nevertheless, the authors caution that these estimates are not the aim of the paper and are speculative. They say:

We do not believe compelling data exist that allow the size of externality shares to be estimated in a way likely to lead to broad agreement about these. We therefore treated these as subjective estimates likely to vary widely across individuals. We therefore did not seek, as with our other data and assumptions, to find a single, “approximately correct” assumption.

For this reason, they provide three further sets of estimates. One set of estimates they believe reflect the extreme right of American politics – the Tea Party – and another set for the extreme left – Occupy Wall St. They also provide a set of estimates by Nathanson, another author of the paper, who is much more sceptical than Lockwood that large differences in the social value of different professions actually exist.

Nathanson’s position reflects largely a general Coasian skepticism of the size of externalities that persist, based on the notion that, while some externalities may persist, they most are largely internalized already by various social, public or civic mechanisms. For example, prestige that is costly either because it is limited in supply or requires scarce attention from others in society, government subsidies on various sectors and regulations on others and voluntary tithing and selective loan repayment to universities conditional on income may partly or fully offset externalities. Thus apparent lack of correspondence between material rewards of different professions, in this view, masks other rivalrous non-pecuniary or unmeasured pecuniary goods being given to different professions that internalize most externalities.

In Nathanson’s world, you should basically just take whichever job offers the highest personal rewards.

Profession Nathanson’s estimates Tea Party estimates Occupy Wall Street estimates
$ of social value produced per $ of salary at the margin
Academia/research 0.5 -0.3 0.6
Advertising/marketing/sales 0 -0.2 -0.9
Agriculture 0 0 0
Arts/Entertainment 0 -0.5 0.8
Business operations 0 0.3 -0.2
Engineering/technical 0 0 0.1
Entrepreneurship 0.5 1.5 -0.5
Financial Services -0.1 0.2 -0.8
Law -0.3 -0.8 0.3
Management -0.2 0.3 -0.6
Medicine 0 0.1 0.2
Public service/military 0.1 -0.8 1.5
Real Estate 0 0 -0.3
Teaching 0.2 0.3 2

As you can see, there’s substantial disagreements, which shows the need to make your key assumptions clear whenever attempting this kind of assessment.

Some of the largest disagreements include:

  • The Tea Party and Lockwood think entrepreneurship is strongly positive, while Occupy Wall Street think it’s negative.
  • The Tea Party think public service is strongly negative, while Occupy Wall St think it’s strongly positive.
  • Occupy Wall St and Lockwood think teaching is very positive, while the others are more skeptical. The same is true of academia to a lesser extent.

However, in other areas there’s substantial agreement. In particular, most professions are pretty close to zero, which is what you’d expect in an efficient market.

How much do these differences matter?

Another striking point is that the size of the differences are fairly small. If the differences are this small, then it suggests that for supporters of 80,000 Hours, judgements about the average social value produced by different professions could easily be overwhelmed by considerations like (i) your different chances of success in different jobs (personal fit), (ii) different career capital and option value from different jobs, (iii) and differences in donation or advocacy potential between different jobs.

For instance, if you think that you can donate your income to a charity with a cost-benefit ratio of 10, then how much you’re able to donate could easily overwhelm the differences in direct impact.

This is especially true when you consider how speculative the estimates of the social value of different professions are today.

Notes and references

  1. Lockwood, Benjamin B., Charles G. Nathanson, and E. Glen Weyl. Taxation and the Allocation of Talent, 2012.↩
  2. Murphy, Kevin. M, Andrei Shleifer, and Robert W Vishny. The Allocation of Talent: Implications for Growth, 1991.↩
  3. Acemoglu, Dawn. Introduction to Modern Economic Growth. Princeton University Press, 2009.↩
  4. Chetty, Raj, John N. Friedman, and Jonah E. Rockoff. The Long-Term Impacts of Teachers: Teacher Value-Added and Student Outcomes in Adulthood, 2011.↩
  5. Galbraith, John Kenneth. The Affluent Society. New York: Houghton Mifflin, 1958.↩
  6. Hirshleifer, Jack. The Private and Social Value of Information and the Reward to Inventive Activity, 1971.↩
  7. Brunnermeier, Markus K., Alp Simsek, and Wei Xiong. A Welfare Criterion for Models with Biased Beliefs, 2012.↩
  8. Posner, Eric and E. Glen Weyl. An FDA for Financial Innovation: Applying the Insurable Interest Doctrine to 21st Century Financial Markets, 2013.↩
  9. Shleifer, Andrei and Robert W. Vishny. A Survey of Corporate Governance, 1997.↩
  10. Bénabou, Roland and Jean Tirole. Bonus Culture: Competitive Pay, Screening, and Multitasking, 2012.↩
  • pliny

    “If the differences are this small, then it suggests that for supporters
    of 80,000 Hours, judgements about the average social value produced by
    different professions could easily be overwhelmed by considerations like
    (i) your different chances of success in different jobs (personal fit),
    (ii) different career capital and option value from different jobs,
    (iii) and differences in donation or advocacy potential between
    different jobs.”

    What if you just value certain goods produced by particular jobs more highly than other folks, and want to see more of them produced as opposed to other goods?

    • Evan Gaensbauer

      Effective altruism seems to hinge upon an assumption that a consensus of those dedicating our careers to creating lots of social value reach a more or less common consensus of what ‘good(s)’ we’re trying to produce. Obviously, there will be differences among individuals. However, I believe most of effective altruism values improving the welfare of the worst off among persons, who are no less deserving of a quality life than anyone else. Givewell quantifies these as Quality-Adjusted Life Years (QALYs).

      Increasing QALYs may be represented in different ways, for example, through treating diseases, or donating money to research into preventing future catastrophes. The delivery of vaccinations for neglected diseases vs. the payment of salaries for crucial researchers may be perceived as different economic goods. However, from the perspective of effective altruism, they’re generally perceived as proxies for producing an ultimate good, e.g., QALYs.

      So, who is the ‘you’ in your example? From a practical perspective, who among us might be the one who values what one career produces so differently from what the majority values?

    • Benjamin Todd

      I’m not sure I understand the question. Some jobs will have more valuable direct impacts than others (as shown by the different numbers for ‘e’ above), and it’s partly a question of values. The direct impact of your job is an important consideration, but so is personal fit, career capital, donation potential etc.

  • Owen Cotton-Barratt

    Thanks, this is fascinating.

    You said:

    > In particular, most professions are pretty close to zero, which is what you’d expect in an efficient market.

    Note that it’s not clear what direction causality goes in here — the estimates were provided by economists, so this feature may reflect their view on how efficient the labour market is.

    We might like to distinguish between trade surplus and other externalities. When a firm employs someone, this is a form of trade and should be good for both parties. In an efficient market the employee can get essentially just as good a job elsewhere, so the firm doesn’t capture much value, but as the market isn’t entirely efficient we should expect the firm to capture some value. So (if I’m interpreting the numbers correctly) this effect would push expect typical professions to have numbers above 0. The other aspect is true externalities (i.e. how it effects people not involved in the employment). These could be positive or negative, but an assumption of efficiency wouldn’t make them go away.

    • ESRogs

      > In an efficient market the employee can get essentially just as good a job elsewhere, so the firm doesn’t capture much value

      I’m not sure I follow this. Would it also hold that the firm can hire just as good of an employee elsewhere and so the employee doesn’t capture much value?

      • Owen Cotton-Barratt

        The distinction (in this idealised model) is that the supply of labour is fixed, whereas the supply of firms can vary easily. If you reversed that (fixed number of jobs, much greater number of possible employees) then the employees wouldn’t capture much value.

  • Tom Stocker

    if academia is 2 overall as a marginal estimate, then particular types of academic research that push the production function further out and allow for more growth (this is the basis of the estimate) might be much more marginally beneficial. biotech research, research aimed at reforming the way we develop technologies or other similar avenues might be really worthwhile? Is this type of contribution conceptualised as ‘advocacy’ in your framework?

    • Benjamin Todd

      I agree there should be more variation within each profession. Likewise, you might expect the worst bits of finance to have a more negative value of ‘e’ than -0.5.

      I’d say advocacy is separate: that’s your effect on the behaviour of others through setting social norms, persuading people and promoting ideas through a public platform. Whereas working on a different research field is still within direct impact.

  • Benjamin Todd

    I spoke to the author and a more up-to-date version of this paper has been published, which has different figures and based on more in-depth reasoning. A further update is expected in about a month.
    http://scholar.harvard.edu/bblockwood/publications/taxation-and-allocation-talent

  • Michael Vassar

    GIGO. This is SO FAR short of the level of depth of thought/analysis required to be informative.

    • Benjamin Todd

      The new version of the paper is better: http://scholar.harvard.edu/bblockwood/publications/taxation-and-allocation-talent
      I think it’s interesting to see some best guesses, but of course I don’t think you should choose a career based on these estimates.
      The authors admit it’s highly speculative (and isn’t the main purpose of their paper) – their aim is to provoke more people to do this kind of work.