Gender gaps in performance pay

Sara de la Rica , Juan Dolado, Raquel Vegas 03 August 2010

a

A

One of the cornerstones of the standard competitive model of the labour market is the equilibrium condition equating wages to the value of the marginal product of labour. This implies that the final wage distribution represents the equilibrium outcome of demand and supply forces, a result that has proven useful for analysing how changes in the demand and supply of skills affects wages in economies with flexible labour markets, such as the UK and US. But the competitive model is not without its problems.

It is well established that the competitive model can provide a rather misleading interpretation of how wages are actually determined in real-life labour markets when information is asymmetric or search frictions are present (Manning 2003). The reason is that many existing labour market institutions (like unions and minimum wages) tend to compress pay differences between more- and less-productive workers.

While it is reasonable to acknowledge that the competitive paradigm may often lack realism in describing how wages are set, there are some specific forms of wage compensation that could be considered as good proxies of the “wage = marginal product of labour” condition. In particular, if compensation is paid at least partially as a function of performance pay – such as bonuses, commission or piece rates – it seems plausible to assume that this wage component becomes closer to workers’ productivity than the other components that do not depend so closely on individual performance.

Explanations for gender gaps in performance pay

Following this intuitive reasoning, in recent research (De la Rica et al. 2010) we use a newly available cross-sectional dataset (Earnings Structure Survey or Encuesta de Estructura Salarial, 2006) on the detailed breakdown of total wage compensation for Spanish full-time workers aged 18 to 65 into its different components.

Our goal is to analyse whether Spain has a performance pay gap according to gender.1 To do this, we identify performance pay as “variable hourly non-ordinary payments” related to workers' performance, the amount of which is not established in advance since it is dependent on incentives, returns, and extraordinary profits. This measure lumps together bonuses, compensations, and piece rates whereas the non-performance pay components include the ordinary wage and fixed non-ordinary payments, entailing overtime-work pay and participation in firms' normal profits.

We outline two alternative hypotheses regarding gender differences in performance pay. On the one hand, under the presumption that performance pay is determined in a more competitive fashion than the remaining wage components, the performance-pay gap between equally skilled men and women could be smaller than in the non-performance-pay components. In other words, since in theory performance pay responds mostly to meritocracy, equally performing workers should receive the same performance pay irrespectively of their gender. Moreover, if women perceive some forms of (taste and/or statistical) discrimination in non-performance pay jobs relative to men, then they will seek intensively for performance pay in order to ameliorate these disadvantages.

An alternative hypothesis is that performance pay could also provide a clear channel through which women’s greater involvement in household tasks hinders their returns in the labour market relative to men’s (see Becker 1985). Further, the assumptions of free access to performance-pay compensation without search frictions in a competitive setup may not be suitable for female workers for the following reasons:

  • First, as stressed in the occupational segregation literature, women may select themselves into non-performance pay jobs (e.g. public sector jobs) because they anticipate that the latter are more compatible with their larger household responsibilities. Hence, in line with the so-called “mommy track” hypothesis, women may willingly opt for jobs entailing steadier and, possibly, lower pay in exchange for fewer penalties in case of career interruptions.
  • Second, employers may be more reluctant to place women in jobs involving performance pay if they expect lower female work attachment even if they have the same ability distribution as their male colleagues (see Lazear and Rosen 1990).
  • Third, the presence of some monopsonistic features in performance pay jobs, due to women’s lower mobility or lack of alternative job offers, may also play a role even if, contrary to the standard human capital explanation, this does not lead to lower productivity (see Manning 2003).

A striking gender gap in performance pay

Our most striking finding is that the raw gender gap in performance pay in favour of men is much larger than in the non-performance-pay component and in total hourly wages. We find a gender gap of 74% in performance pay compared with 23% and 24% in non-performance pay and total hourly wage, respectively. This is despite the fact that the female workers receiving performance pay in our sample have some better characteristics than men (higher educational attainment but lower tenure). Figure 1 displays these raw gaps in percentage points throughout the wage distribution.

Figure 1. Gender wage gaps (total, non-performance pay and performance pay components)

Note: Gaps in performance pay and non-PP components are denoted by v and f. The dotted line represents the average gap in total hourly wages.

While it is worth noting the performance-pay share in total wage and the incidence are low (about 9% and 23%, respectively) and the contribution of performance pay to the aggregate hourly wage gap therefore only reaches 18% on average and 25% at the top of the distribution, the evidence nevertheless suggests that either:

  • women are exerting lower effort due to housework,
  • women are self-selecting into jobs without performance-pay compensation,
  • or that firms are exerting some monopsonistic exploitation of women receiving performance pay because of their lower outside opportunities.

In order to disentangle these potential explanations, we perform three checks.

  • First, we check whether – controlling for all other productive characteristics – the proportion that performance pay represents over the total hourly wage is lower for women than for men. This sheds light on whether women exert less effort than men. We find that the difference in the proportion not explained by different characteristics is too small to be considered relevant in explaining the performance pay gap.
  • Second we compare the amount of performance pay men and women receive in the same occupations. We still find that a gap of 43 log-points remains within occupations, suggesting the cause is not occupational segregation.
  • Third, we compare men and women in the same firm with identical occupations. If there were no monopsonistic influence, we would find a much lower gender gap. Yet we find there is still a gap of 29 log-points.

Our evidence therefore appears to yield strongest support to a monopsony influence than to the other two explanations.

Monopsony influence

Further evidence on this issue can be obtained from comparing the relative pattern of the gender gap throughout the distribution of the performance pay component. Indeed the available theories on female occupational segregation predict that, if the ability standard for allocation to jobs involving performance pay is higher for women, it should be expected that the relatively few women who are at the top of the performance pay distribution should receive higher performance pay compensation than their male counterparts. In other words, conditioning on observable characteristics, the gender gap should be negative at the top percentiles of the performance pay distribution.

By contrast, monopsonistic theories related to lower female mobility, like Booth et al.’s (2003) “sticky floors” hypothesis, predict that women at all percentiles will be paid less than men since there is a higher rent to be earned by firms due to women having lower outside opportunities because employers perceive that they are less mobile than men.

Our results, summarised in Table 1, using quantile hourly wage regressions accounting for selectivity corrections under the within firm and occupation specification imply a clear “glass ceiling” pattern with the performance pay gap evolving from 20 log points at the bottom deciles to 43 points at the top deciles of the performance pay distribution. Thus, according to this test, monopsony seems again to be the most likely of our explanations.

Table 1. Adjusted gender wage gaps – quantile regressions (with selection correction)
 
Dependent Variable: Log performance pay hourly wage
 
 
 
(1)
(2)
(3)
(4)
(5)
 
 
P10th
P25th
P50th
P75th
P90th
 
 
 
 
 
 
 
 
Female Dummy
 
-0.202**
(0.022)
-0.272***
(0.020)
-0.301***
(0.014)
-0.354***
(0.016)
-0.433***
(0.019)
 
 
 
 
 
 
 
 
Note: Standard errors in parentheses. Estimations also control for the whole set of covariates (age, education, tenure, type of contract) as well as firm and occupational fixed effects.

Conclusions

One striking feature arises from our empirical exercises. Women in those age brackets (30-50) bound to have larger family responsibilities are those who contribute the most to the share of the performance-pay gap explained by age. As a result, despite women’s increasing effort to signal job attachment through higher human capital accumulation, beliefs about unbalanced household tasks generate “a marriage premium” for males and a “penalty” for women even in those components of the wage that should respond more to workers’ performance and less to job characteristics.

References

Becker, G (1985), “Human Capital, Effort and the Sexual Division of Labour”, Journal of Labour Economics, 3.

Booth, A, Francesconi, M, and J Frank (2003), “A sticky floors model of promotion, pay and gender”, European Economic Review, 47: 295-322.

De la Rica, Sara, Dolado, Juan Jose and Vegas, Raquel (2010), “Performance Pay and the Gender Wage Gap: Evidence from Spain”, CEPR Discussion Paper 7396

Dohmen, T and A Falk (2009), “Performance Pay and Multi-dimensional Sorting Productivity, Preferences and Gender”, American Economic Review, forthcoming

Lazear, E and S Rosen (1990), “Male and female differentials in job ladders”, Journal of Labour Economics, 8, 106-123.

Lazear, E (2000), “Performance Pay and Productivity”, American Economic Review, 90(5),1346–1362.

Lemieux, T and D Parent (2009), “Performance Pay and Wage Inequality”, Quarterly Journal of Economics 124(1):1-49.

Manning, A (2003), Monopsony in motion: Imperfect competition in Labour Markets, Princeton University Press.


1 The existing literature on performance pay has mainly focused on analyses of its incentive effects on productivity; see, among others, Dohmen and Falk (2009) and Lazear (2000).

a

A

Topics:  Labour markets

Tags:  Discrimination, gender gap, performance pay, marriage

Sara de la Rica

Professor of Economics at the University of the Basque Country

Professor of Economics at the Universidad Carlos III de Madrid and Coeditor of the journal "Labour Economics". CEPR Research Fellow

Raquel Vegas

Research Analyst at FEDEA