The overstated business case for having more female managers

Westpac released a report this week on gender representation in business leadership positions. It's an important topic, and rightly received a lot of media attention (see here and here, for example). However, the media focused attention on the one part of the report that is not worth the (virtual) paper it was written on:

New Zealand's economy has a nearly $900 million annual economic hole because of low numbers of women in management roles, new research suggests.

Let me break that $900 million down, and if at the end of this post you still believe it, I know some Nigerian princes with millions of dollars that they can't get out of their country that you can help.

First, the $900 million is actually $881 million, and it comes from two sources. From the report, (emphasis is theirs):

First, having more women in leadership positions can change perceptions about female competency and skills, and this effect can increase female labour force participation. We estimate that increased female participation at manager level and above would be worth an additional $196 million (or 0.07% of GDP).

Second, women in leadership roles tend to be more supportive of flexible working policies, which in turn also increases labour force participation.

We estimate that if all New Zealand businesses were to achieve gender parity in leadership, it is likely to lead to an increase in the number of businesses offering flexible work policies. The associated benefit resulting from more businesses offering flexible working policies is an additional $685 million (or 0.26% of GDP).

Let's look at where they get each of those two numbers from, starting with the $196 million. They first estimate a model that relates female labour force participation to the proportion of employees that are female managers. They use OECD data for four years only (2011-2014). The results of that analysis suggests that:

...a 1% increase in the share of female employees who are managers is associated with a 0.09% uplift in the female labour force participation rate.

They then extrapolate that number (based on Australian data that 13.3% of male employees are managers, but 8.9% of female employees are managers) and claim that the female labour force participation rate would increase by 0.39% if gender parity in management were attained. The problems with this analysis are many. First, the model only shows a correlation, not cause-and-effect. Nothing in their analysis proves that changing the number of female managers would cause any change in the female labour force participation rate. Second, it's based on only four years of data, during which most countries percentage of female managers and female labour force participation rates will not have changed by much. So the extrapolation will likely be well outside the observed data. Third, the data on the proportion of managers is for Australia, not New Zealand. While our two countries are similar, they are not the same, as the report even notes:

Relative to Australia, New Zealand performs equal to or better in nearly every respect, including pay gaps.

So it seems unlikely that that first number can be believed.

The second number ($685 million) is even less plausible. They first run a model that relates the number of flexible working policies to the proportion of female managers. They find that the:

...marginal effect of increasing the availability of flexible working policies by 13% if the average share of female management increases from the status quo calculated in Australia (22.4%) to parity (50%).

They then use that result in an additional calculation that is summarised in the following table:

The first problem here is the model, which again shows correlation, not cause-and-effect. This model is even more problematic than the previous one though, because the causation clearly runs in both directions - having more flexible working policies will attract more female managers, as well as more female managers being more likely to pressure for implementing flexible working policies. That means that you can have very little confidence in the coefficients from the model because they are biased. Second, again they are using Australian data in part of their calculation.

Third, look at the top row of the table: "Percentage of people not in the labour force citing flexible working policies as a very important incentive to join the labour force". They then assume that 13% of the 23% of people would join the labour force if flexible working policies were increased by 13%. There is no consideration that these people have only been asked about incentives, and not about actually joining the labour force. On top of that, if you roll out flexible working policies in 13% more businesses, that doesn't mean an increase in the availability of jobs with flexible working policies of 13% (unless you also assume that the firms now offering flexible working policies have the same average size as all firms collectively). Maybe it will be larger firms that do this, or smaller firms? This hasn't been considered.

Finally, both numbers ($196 million and $685 million) assume that the increase in labour force equates to an increase in employment. That is by no means a given. If more people enter the labour force, but there are no additional jobs for them, that increase in the labour force becomes additional unemployment, with no increase in GDP at all. Alternatively, the increase in labour supply might reduce average wages, either directly or through an increase in part-time (compared with full-time) work. So, even though GDP might increase, other workers in the economy are made worse off.

There is another bit of analysis in the report that associates each one-percentage-point increase in female managers with an increase in return on assets of 0.07%, and then argues that raising female management to parity would increase return on assets by 1.5%. However, if you believe that, why would you stop at parity? If you went all the way to 100% female management, you could increase return on assets by nearly 5%!

Overall, the Westpac New Zealand Diversity Dividend Report does make some good points. However, the economic impact and business case are extraordinarily oversold as they are clearly flawed, and they do no credit to the overall argument that a more equal representation of women in management would be a desirable result.