Some problems with that productivity paper

Last month a study was released* by Yana Gellen** of the Becker Friedman Institute for Economics at The University of Chicago, “Motherhood and the Gender Productivity Gap.”

Some outlets, like the American Enterprise Institute and Wall Street Journal, have jumped onto the study and claimed this is the reason that working mothers don’t earn as much as men – they aren’t working as much or as productively.  But does the study really show that?  And what does all this mean for working mothers in the academy?

Does this study prove that mothers are less productive?

In short, no.  Digging into the methods – there are some major problems with how this study was done.

  1. Productivity as percentage. The study makes claims about women’s and mothers’ productivity as a percentage of single men’s labor.  This interesting methodological decision automatically norms whatever people coded as single men are doing. That does NOT answer whether or not women are doing whatever they were hired to do, completely and well.  Nowhere does the paper address whether mothers and women produced the number of widgets they were hired to make. This framing is very anti-worker, as it centers the idea that employers should expect infinite productivity from workers and not even ‘settle’ for people doing the jobs they were hired to do well, but only keep the ones going beyond.

This approach could lead to a ‘race to the bottom’ – sure employers want to hire folks who would work more hours or produce more products than they were hired to do, but that doesn’t actually justify cutting people who do their actually assigned work in the hours you’re paying them for. If you want them to do more, pay them more, train them, and/or hire more people to support them.  The paper doesn’t critically engage the problems of norming male output as standard (making women and mothers deviation/deviants by default) or the idea that there can be “good enough” labor and product production.

  1. Types of labor and products The study is using data on employers (firms) and workers from six Danish industries: hospitality, construction, manufacturing, wholesale and retail trade, “other community,” social and personal services, and “real estate, renting, and business activities.” Many of these industries produce very different KINDS of products – and many of those rely on customer interactions. Even the authors acknowledge at one point that sexism itself may drive clients to work with non-women or non-mothers more frequently or for more contracts.  Yet even that admission does not address what supervisors or firms may be doing themselves to steer clients away from mothers (if they are out on leave, did you take their clients away?  Did you not give them the contracts because you assumed they would go on leave? Etc)

Potentially the folks in the manufacturing sector ARE making widgets, but most of the fields represented here involve client interaction that are highly influenced by the biases and perceptions of not just employees’ coworkers, bosses, and firms – but their specific clients and general public.  This must be taken into account. What data exists on Danish perceptions about working women and mothers? If it exists, it should be incorporated.

  1. Is 8% the same as 12%? The study mentions multiple times that it can account for 8 of the 12 residual points using productivity to “entirely” explain wage differences. 8 is not 12. Maybe economics does stats differently than I would.  Even 2/3 is not the same as 3/3.

TL;DR – The study makes bold claims (“Productivity explains the entire pay gap associated with motherhood.”) that major media outlets and think tanks have latched upon, but it’s not entirely supported to extrapolate that far.

Is this study gaslighting us?

But the study doesn’t just find a wage gap for women with children. It finds a wage gap for women who don’t have children, even if those women later had children.  Rather than attribute this to discriminatory practices in pay and promotion, the study claims it’s due to “pre-child effect of anticipated fertility: women may invest less in education or select family friendly career paths in anticipation of motherhood.”   Sure, maybe.  Yet even the authors acknowledge that their study design cannot test for this.  So if it’s a hypothesis, where’s one about the employers?  Even if their hypothesis is true,  women could be making rational decisions to avoid or combat discriminatory employers or fields.   It’s critical to acknowledge the agency of women as economic actors, but this framing does not seem to leave open the possibility that supervisors and employers are actors who can and do discriminate.

Further, the study also claims there is a “post-child effect of realized fertility: women changing their hours worked, occupation, sector, firm, etc., in response to actual motherhood.”  This, again, shifts all the responsibility for wage differences to the women themselves.   This does not leave open the possibility that supervisors and employers may be creating work environments hostile to women and mothers, pushing them out.

What does this mean for academic women and mothers?

We don’t make widgets***.  How such a study tries to code and quantify the work of people doing academic work would be critical to understand. This study problematically normed male labor output as “100% productivity” without addressing the critical question of whether women and mothers were sufficiently performing the jobs they were hired to do.  Especially in a profession – and job market – where there is always MORE that could be done, this framing is especially problematic.   In short, trying to apply this study’s framing to explain away the pay gaps for women and especially mothers in the academy is not appropriate.   Moreover, if adopted, it could seriously harm progress in supporting a diverse professoriate.

*not published in a peer reviewed journal, nor a pre-print server. It’s a “report” released online by academics. Economics seems to do this kind of working paper thing often.
* Mx. Gellen is a graduate student in the program, and I want to be clear that the discussion of the paper is good faith engagement with the methods/approach, not punching down at a student. Especially if this paper really is in working paper form, conceptual constructive feedback is hopefully welcome. These are sensitive subjects that deserve lively debate from the research community.
*** are journal articles widgets? Maybe. But that’s not our only output.  And even if it was, which kinds of articles would count as 1 widget? First author? Major journal? These are SUBJECTIVE and hard decisions when coding labor products.
Advertisements

3 thoughts on “Some problems with that productivity paper

  1. Honestly I don’t know how productive it is to keep re-hashing this question on what is it that causes the gender-pay gap with this kind of imperfect methods. External circumstances are very different for men and women in the workplace in many different ways — some of which you point out but there are many others. Unless they can be made more equal, any broad causal claims of the sort made by this paper are highly dubious. I see this research as being a waste of tax-payer dollars, which would be more productively employed elsewhere.

    • This paper was not funded by taxpayer dollars, FYI. And some people do study wage gaps in order to attempt to fix them – I don’t know the motivation of the researchers here.

      There are no perfect methods for any study but that’s what makes nuanced interpretation and explicit caveats so important – both of which I think are missing from the study and write ups, hence why I wrote the piece.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s