The snowballing penalty effect: Multiple disadvantage and pay
UK research, 2013
It is sometimes argued that setting right the wrong of gender pay inequity will “cure” more than just the pay gap between men and women, namely it will resolve other lines of inequity such as white versus non-white and disabled versus non-disabled. This argument rests on the logic that separating employees into male versus female represents the largest possible grouping, and remedial strategies will thus incorporate benefits for subsets such as those women who have a disability. This logic also assumes that the pay inequity experienced by the larger group (ie female) is homogenous for those within that group (ie all women experience the same type of pay inequity), rather than assuming there are greater levels of variability within the group. Further the logic assumes that if there is variation within the group it is simply additive (ie female + disabled), rather than qualitatively different, and therefore reducing one element is as powerful as reducing another element. All of this is thrown into doubt by Dr Woodhams (University of Exeter), Dr Lupton (Manchester Metropolitan University and Professor Cowlings’ (University of Exeter) recent research which found that those employees who are least demographically similar to a white, male, middle-aged/mid-career and non-disabled employee, experience the greatest level of pay inequity, and this is not a simple matter of addition but more like a snowball of disadvantage. If this is correct, then it implies a much more sophisticated approach to pay inequity is required, than a simple focus on gender. Insightfully this study reveals that pay, as a mechanism by which we bestow “value” on employees, provides the clearest indicator of the “ideal” employee, and this has much less to do with merit than conformity to a demographic norm.
To identify the relationship between multiple levels of demographic “otherness”, ie sex, disability, age and ethnicity, on pay outcomes in comparison to single levels of difference; and to consider whether current single dimensional remedies are appropriate to redress disadvantage.
A decade of HR records, for over half a million employees (n=513,741) in a private sector company, were analysed according to age, race, sex and disability, as well as mode of work (full-time/part time), tenure, location and pay. Each employee group was assigned a number which indicated their “level” of difference from the most highly paid group (male, mid-career (aged 31- 45), white, non-disabled), for example a white male, aged 46, non-disabled employee was categorised as having one level of disadvantage (age) in comparison to a white female, aged 46+, non-disabled who was characterised as having two levels of disadvantage. In this way the employees were separated into 16 groups or combinations, with the maximum level of disadvantage being four. The median and average pay were calculated for each group to identify the variation within and between groups.
The study made findings in relation to the overall pay differences between single dimensions (eg pay) as well as multiple dimensions.
(i) Overall: Overall the study found that women were paid 12% less than men, people with a disability 12% less than those without a disability, and those from ethnic minorities 6% less than whites. Age categories show that older workers (46+) earned 2% less than workers aged 31-45, and employees aged 18-30 earned 84% of that earned by employees 31-45. Moreover pay is more variable as employees age.
(ii) Variations by levels of disadvantage:
a. From the high to the low: The group with the highest level of pay was that characterised as male, white, mid-career and without a disability (median pay £22,650, with the average £25,693). This group was therefore characterised as having no disadvantage. In contrast, the group experiencing the greatest pay penalty were those (as expected) with four levels of disadvantage (women, ethnic, 46+, with a disability) whose median pay was £17,479, and the average £17,474. Two things stand out from these extremes, firstly multiple points of disadvantage come with a significant price penalty (on average £8,219) and secondly there is greater variation within the “ideal” employee group, compared with multiple disadvantaged, as indicated by the gap between the median and mean pay within a group
b. Just one disadvantage: The largest group of employees had one disadvantage (56.6%), with 36.4% of the sample having no disadvantage, and 11.3% two levels of disadvantage, but not all disadvantages were the same. The group who most closely resembled the “no-disadvantage” group had, as expected, just one difference, namely age (they were slightly older being 46+), but this one level of disadvantage carried only a small penalty, namely whilst they also received a median pay of £22,650, the average was slightly less than their peers at £25,548. In contrast, the group of white men, mid-career who were disabled (ie with just one disadvantage – disability), received significantly less pay than their peers (£20,937 as a median and £21,324 on average) whereas white women, mid-career with no disability ranked higher with a median pay of £21527 and £23,428 on average)
c. Variability within groups: If the characteristics within groups (age, race, disability) were ignored and the lens of analysis reduced to just one dimension, namely sex, it missed the significant variation within groups. In particular there was an 18% variation within the male group and a 28% variation within the female group, ie not all men, and not all women are the same, and a single lens analysis masks these differences. Critically age represents 33% of the variation across mid and late career groups, which is a single dimension carrying greater weight than even sex
d. Snowballing effect: There was an exponential relationship between the number of levels of disadvantage and pay, ie it was not an additive relationship (eg sex+ race) but a multiplier (sex x race), which meant that the pay of multiple disadvantaged groups was significantly lower than that of groups with one or two levels of disadvantage.
Focussing on single dimensions of pay equity, and particularly a gender based analysis, can be misleading and reinforces a narrow belief that gender is a dominant prism, and separating employees into groups based on sex provides the most meaningful insights and leads to the most impactful strategies. In fact this study demonstrates that there is greater variation within male and female populations than between them. Further, that experience of multiple disadvantage are qualitatively different than a simple addition of another layer. This means that employees are being viewed through multiple lenses simultaneously when being evaluated in terms of pay, and yet the checks and balances in our HR systems only capture a single dimension, and may lead those who “pass” this test to miss the other ways in which bias is being played out.
This study suggests that a much more sophisticated approach to pay equity is needed to eliminate inequities, included multiple checks and balances. Moreover, this study demonstrates a critical insight, namely that pay reflects value, and a greater “value” is placed on those employees who more closely represent the ideal employee. This insight complements the research on the demographic composition of leadership, representing another visible outcome of who is valued.
To read the full article see Woodhams, C., Lupton, B., and Cowling, M., (2013) The snowballing penalty effect: Multiple disadvantage and pay, British Journal of Management, published online ahead of print.