Monday 22 January 2018

Productivity is the Wrong Argument for Diversity

If you look around your workplace and everyone, or least all the managers, look the same - same sex, skin colour, social class, age - then your company has a diversity problem. But why is it a problem?

Because the most obvious explanation is a failure of meritocracy. Such features as the colour of one's skin or sex are arbitrary and irrelevant to people's ability to do a job. Therefore the fact that people of certain skin colours or sex are missing from your workplace relative to the wider society presents a prima facie challenge to the fairness of your company's criteria for employment and promotion. To assume otherwise - for example that people of certain colours, sex, class, age, happen to have different (inferior) career preferences or different (inferior) talents has no credibility. It is to assume the exact set of facts most convenient to make a problem someone else's, rather than to take responsibility for investigating and fixing it.

Call this the negative argument for diversity: If you don't have internal diversity in line with the wider society then you are probably treating people unfairly and you need to investigate and try to fix it. For example by identifying and mitigating biases in how job applicants are evaluated and structural impediments to their career progress. It leaves a lot of details still to be argued out, but I think it is the right way to go.

But there is another kind of argument that is now much more common, the positive argument that organisations should promote diversity because it pays off. This is the argument I want to criticise, on the grounds that it jeopardises the negative argument from fairness; reduces individuals to stereotypes about groups; and perpetuates unjust stereotypes and social relations.

The positive argument for diversity in the workplace is that it produces productivity gains for the group as a whole. A large number of empirical studies seem to support this. A group with more variety of racial, gender, class, and sexuality types will tend to have more variety of life experiences and ways of looking at the world. Such groups will be able to bring more different perspectives to problems and will therefore be more likely to come up with better, more innovative solutions. An additional important mechanism is friction. Because of the social distance between members of different groups, everyone ups their mental game in comparison with the mutual complacency that might pervade a group where everyone is very comfortable with everyone else. It is like the difference between the way you clean your house when a stranger is coming to dinner and the way you clean up when it is just your friends coming over for a beer.

This article by a business ethicist seems to be typical of the genre: 'How Diversity Makes Us SmarterBeing around people who are different from us makes us more creative, more diligent and harder-working'. And here's an interview by a professor of complex systems that says roughly the same thing: How Diversity Powers Team Performance.

I. Say Goodbye to Fairness

The first problem with supporting diversity because it pays off is that it undermines the genuinely ethical concern about fairness to workers. The productivity argument at its simplest goes like this:
  • Premise: Empirical research has shown that (at least some forms of well-managed) diversity increase employee productivity. 
  • Conclusion: Therefore, corporations should promote (those forms of) diversity. 
However, there is an implied but unstated premise needed to complete the argument.
  • Missing premise: If an action has higher productivity than alternatives, corporations should do it
Thus the productivity argument introduces a particular standard by which to judge diversity policy or any other action: will it raise productivity? This is an entirely amoral but conventional approach to business. It is not about which values we should have but only about the most efficient way to advance the material interests of the corporation. Diversity policies are to be evaluated in just the same way as schemes for minimising tax exposure. 

But the productivity argument for diversity is actually anti-ethical because it undermines any concern for a genuinely moral principle like fairness and the perspective of (would-be) employees in general. The productivity argument does not provide the corporation with an additional reason to support workplace diversity as well as fairness. Instead it offers an entirely different way of reasoning about diversity that competes with fairness. Diversity is now a right that companies have over their workforce as part of their general right to maximise profitability. It is not a duty that companies owe to people as part of their general duty to respect every person's inherently equal dignity. In this logic fairness cannot even be understood as a reason for action. It can only appear indirectly in the corporation's calculations - if enough people were upset about it to affect work output.

Moreover, diversity policies that follow from a concern with productivity are likely to be quite different - worse - from those that follow from fairness. Because the productivity argument for diversity is purely instrumental it is entirely dependent on how facts turn out. Suppose it turns out that including people with some skin colours, or some combinations, doesn't work as well or has more expensive management costs. (This seems consistent with the literature.) Then the productivity argument for diversity says companies ought not to hire people from those groups unless they are the only choice. People of those types will face systematic employment discrimination whatever their personal achievements.

More specifically, the productivity perspective argument does not instruct the corporation to support diversity as a general principle at the same level as 'pursue profit', but rather to use empirical findings about diversity to achieve desired outputs. Managers are supposed to operate like gardeners, micromanaging the precise combination of species most likely to generate the sort of harmonious interactions they are looking for, and at the level of maintenance they are willing to commit to.

Thus, a company might say to unfortunate job candidates something along these lines, 
"We are confident you could do this job very well. However, research has shown that statistically people of your ethnic background are 5% less stimulating to team productivity than people of type x". 
"Our diversity recipe calls for 1 part x to 3 parts y to 4 parts z. We are already full up with your kind"

"The work here is routine: creativity is not needed and is strongly discouraged. Therefore we keep diversity to a minimum and only hire people who look the same. We find it much easier to manage our employees that way."

II. Judging People by Stereotypes is Degrading

The second problem with the productivity argument for diversity is that it reduces individuals to statistical generalisations about the groups they happen to be members of. This is part of what is troubling about the diversity rejections above. But it is even insulting to be selected on the basis of how your skin colour, sex, class background fits a corporation's preferred diversity mix. For example
"Welcome to the company! Your precise work assignments will be determined on a weekly basis according to which teams need a black male perspective for their projects."
What is insulting is not that people are hired and assigned to work teams on the basis of how well their particular skills or perspectives complement each other. That seems quite reasonable. What is insulting is that corporations are supposed to look past these features of job applicants as individuals, the items they are proud to list on their resumes and ask to be evaluated by. Instead the focus is on categorising applicants into groups mostly defined by social injustice.

Assignment to these minority groups comes with the burden of negative stereotypes that bias employers' perception of talents and achievements when hiring or promoting. Now that (certain kinds of) diversity are seen as productive, what changes?

Features such as skin colour that were once seen as negative are now seen as positive. But judgements of employability are still based on features one cannot change. Moreover, an implicit bias has been converted to an explicit justification. Before, you had the challenge of managing to get hired in spite of bosses' biases. Now you are hired explicitly because you fit a stereotype and your challenge is to play that role successfully. You can't just be 'an engineer'. You must play the part of a hyphenated character in a play written by other people: the female-engineer, the Asian-engineer, etc.

The absurdity is that insofar as the benefits of diversity come from bringing different epistemic perspectives together, diversity of things like skin colour are only a proxy for what is actually important. If you have a lot of diversity of colours, gender, and so on but they all studied the same software engineering programme at Stanford then epistemic diversity will be low: they will all solve problems in similar ways. Even from the narrow instrumental logic of corporate productivity there doesn't seem a good reason to focus on assigning applicants to different minority groups rather than directly analysing the interestingly different backgrounds and skills each individual actually possesses. After all, you already have their resumes! 

III. Useful Diversity Entrenches Injustice

To the extent that diversity drives productivity further than mere epistemic variety (such as having a poet and a mathematician on the same team) it seems to be due to the social distance that exists between members of groups. Thus, apparently white men will pay more attention to different ideas or objections raised by the black member of their team than to each other. Because the black guy is the other on the team, he is more carefully watched. Workplace diversity reproduces the social tensions of wider society, and this disrupts the complacency of teams where everyone is very comfortable with each other.

This mechanism is plausible, yet its functioning clearly requires the continued existence of social divisions and stereotypes. If we stop finding people of different races/gender/sexuality 'difficult' to work with then this benefit of diversity will disappear.

This seems to have happened in the past. For example, these American studies don't bother to assess the diversity gains from teams made up of people with Jewish, Irish, Italian, or Catholic features because being tagged as a member of those groups no longer raises a red flag as a suspicious 'other'. But this transformation accompanied the ending of discrimination against members of these group, which seems like a success for them and for America. The loss of one particular source of workplace productivity seems more than counterbalanced by the wider gains of treating these people the same as anyone else (i.e. fairness).

The gains of workplace diversity fueled by mutual suspicion or tacit hostility between members of different groups may be real. But they are the byproduct of toxic and unjust social relations that do enormous harms to society as a whole and to minority groups in particular. For example, being seen as the 'other' might help your team by making them pay more attention to your heterodox ideas, but how much good does that do you? Will being the 'other' help you get a raise or a promotion, or to drive a nice car without getting pulled over by the police?

More worryingly, diversity policies based around productivity may not simply make passive use of a positive byproduct of a toxic phenomenon. They may also sustain and reinforce racist and sexist relations and stereotypes by giving them a new significance. 

Categories like race are not properly scientific because they don't exist outside particular cultural frameworks (unlike, say, biological units like species). Specifically, they were originally inventions used to justify treating some people differently - as less than equal. But they will continue to exist as long as they serve a social function. When large numbers of corporations use these invidious categories as the basis for hiring decisions, it reinforces the social significance of such labels to people's lives - how they see themselves and others. Hence categories invented to justify oppression remain part of our social reality. as job categories When corporations consciously hire people to play the role of the 'other' they reinforce the underlying negative stereotypes and the social distance between people.

IV. Saving Business Ethics

I have tried to show that the way one argues for diversity has significant consequences for how people are treated. Arguments matter. In my last words I want to relate this particular dispute to two broader divides in how to think about the ethics of business.

First, a foundational problem of business ethics is how to incorporate the profit seeking interest of corporations within moral analysis. The productivity argument for diversity is an example of attempting to justify moral conclusions from the prudential perspective of a corporation, i.e. to provide an amoral justification for doing the right thing for people unable to follow genuinely moral analysis. It thus resembles the argument in the ethics of war that goes, It is wrong to bomb civilians and anyway it doesn't work.

The problem with this strategy is that instead of supplementing the properly moral case you may undermine it by endorsing an alternative instrumental problem solving logic. Instead of demanding that corporate leaders review their values and acknowledge the moral reality of others and their duties to them, we allow them to continue blindly pursuing the prudential logic of profit maximisation. My view is that business ethics should seek to transform values - of business people and the wider public; not merely try to influence corporate actions.

Secondly, academic business ethics is divided between those who investigate right and wrong using the methods of moral philosophy and those who collect and analyse facts about matters relating to right and wrong using social scientific methods. I am sceptical of the general quality of the empirical research in this area, which seems to me even sloppier than the already low average of the social sciences (a lot of significance chasing). But my larger concern is that empirical methods are being mistakenly used to investigate what is morally right or wrong in itself. Worse, this flawed kind of business ethics sells much better! Business people seem to find factual claims, however shoddy, more comfortable and persuasive than actual moral analysis.

Of course facts about how the world works can be extremely useful for making moral decisions. For example, knowing that a person can live a normal life with only one kidney is useful for deciding whether to donate a kidney to someone who would otherwise die. However, even here facts play a subsidiary role and do not substitute for moral analysis. They don't tell you the value of a life.

There is a division of labour that should be respected. Empirical inquiry is very good at answering some questions - such as what happens to variable x when variable y is increased. If social scientists are allowed those are the questions they will run after. But the hard-won answers they bring back will be irrelevant or a distraction from moral inquiry. Take the idea I consider central to the moral analysis of diversity: fairness. What empirical fact is this about? How can the Kantian idea that all persons should be alive to the moral reality of others be looked for in the data? At best, one could survey what people think of fairness and how much it matters to them (and how much that affects productivity).

Such facts about human psychology might be very helpful in framing a diversity policy so that employees would perceive it as fair and it would have the best chance of success - and that is no small thing. But that is a different exercise from moral inquiry. Whether women adapt their career aspirations to the constraints of prejudices about femininity doesn't make those constraints any less unfair. Facts about the psychology of fairness cannot help us decide whether fairness itself requires diversity any more than facts about productivity can.