The plan is usually that of presenting this type of advanced modelling technique through theory and by developing an agent-based model together, using the software NetLogo. This was my third time in the program in Varese. Over the first two years and, over past years, we developed a model of innovation diffusion among an ecology of firms (I will discuss that in another post). This time I asked if someone wanted to develop a different model.
After a first few uncertainties and at the very last minute, someone proposed to model gender pay gap in an organisation. I thought that was a good idea and, once we voted on it, we decided to pursue this model. The ABM took shape as we were modelling it, with inputs from the students and adjustments to the coding so that it was not too complex — this is still an introductory course, after all!
Although simple and intuitive, I believe that the model serves its purpose very well. Of course, when one is up to modelling a social phenomenon, there should be a purpose, to put it with Edmonds (Different modelling purposes, 2017). The idea was to first illustrate the pay gap in an organisation. However, that soon seemed a bit too narrow for an ABM. By proposing dynamic as a central feature, the model was shaped into a way to explore the conditions through which the gap could expand, reduce itself, or disappear. And, on top of this, one has the opportunity to estimate how long it would take for such a gap to disappear.
The Gender Pay Gap (GPG) Model is built on simple features. I have slightly modified — hopefully enhanced — the one developed in class to reach to a more realistic one. Here is a quick summary of the features:
- number_workers — this could be set from 0 to 1000 from the slider;
- women_proportion — the percentage of women in the organisation;
- manag_proportion —how much management the organisation has, 10%, 20%, or 30%; teams are built around managers and some employees are also connected outside of their team;
- hierarchy — how many layers of management are there (0, 1, or 2 levels);
- years — for how long would the simulation run; this goes by 10 years intervals and up to 50; if the slider is set to 0 then it runs potentially forever but, if the income for women at year 100 is the same that there is at year 150, the simulation will stop (...because it makes no sense to keep going!)
- pay_gap — you can set how large the pay gap is, both positive and negative (just to explore alternative realities), where the slider expresses percentages. So, you can go from Italy's 5% to Germany's 20% (yes, for once, Italians are ahead!);
- mean_bias — it is the extent to which a manager discriminates against the other gender when making decisions about salary increase or promotions;
- the three sliders at the bottom right are for visualisation purposes (try to use them... carefully).
 |
| Figure 1. A Model of the Gender Pay Gap in an Organisation |
You can click on Figure 1 above or here to get to the online version of the model: [https://netlogoweb.org/web?https://raw.githubusercontent.com/dsecchi/ABMOsimulations/master/gender-balance.nlogo]
The simulation stops if the distance between men and women's salary is below or at 10% of the original gap, so that we can say there is equality, more or less (I mean, a 5% gap would be a 0.5% gap). What comes out of the model is that the gap is extremely resilient and won't go away very easily. Not even a small gap of 5% would go away easily. And this should make us reflect more on how difficult it really is to make things right.
Now it is time you try the model yourself. Let me know if you think there is something important missing. Thank you.
No comments:
Post a Comment