A sticky floors model of promotion, pay, and gender

Abstract

According to raw data from the British Household Panel Survey, full-time women are more likely than men to be promoted. Controlling for observed and unobserved individual heterogeneity, we find that women are promoted at roughly the same rate as men, but may receive smaller wage increases consequent upon promotion. To help explain these phenomena, we construct a new “sticky floors” model of pay and promotion. In our model, women are just as likely as men to be promoted but find themselves stuck at the bottom of the wage scale for the new grade.

Introduction

It is well established that women fare badly in the labour market relative to men. But how does the status of women change over the course of their careers? Do ‘glass ceilings’ exacerbate the male–female wage differential, or are women able to take advantage of promotions to catch up partially to male wage rates?

Recent evidence from the British Household Panel Survey (BHPS) shows that women are just as likely as men to be promoted. Over 1991–1995, full-time male workers had only a $\text{9}\text{percent}$ chance of receiving a promotion each year compared to $\text{11}\text{percent}$ for full-time female workers. However, the BHPS data also show that the wages of promoted men grow $\text{4.7}\text{percent}$ faster than the wages of non-promoted men, while the wages of promoted women grow only $\text{1.3}\text{percent}$ faster than those of non-promoted women. These stylised facts suggest that standard views of glass ceilings may not tell the full story about gender and promotions. The disadvantage women face in the promotions process may be in the salary rewards to promotion, not in the likelihood of being promoted.

To examine these issues, we develop a model of promotions, pay and gender that distinguishes between the initial pay increase upon promotion and subsequent pay increases. We assume that the initial pay increase is set by the firm to induce workers to put in additional effort to acquire specific human capital. Once the worker is promoted and paid this higher wage, however, the firm has no reason to continue to increase wages in the absence of outside offers. Women may receive fewer outside market offers than men or may be less able to take advantage of them due to family commitments. Firms may respond differently to outside offer threats from women and men owing to discrimination. On the other hand, as in Lazear and Rosen (1990), women may have better non-market opportunities than men. Our framework is sufficiently general to allow us to develop the implications of these differing assumptions for gender differences in promotions, in pay for those promoted who remain at the firm, in pay for those promoted who leave the firm to other market opportunities, and to quit rates to other firms and to non-market alternatives. We then use our theoretical framework to inform the empirical analysis of data from the first five waves of the BHPS, 1991–1995. The empirical analysis allows us to draw conclusions about the validity of the different assumptions about outside opportunities and firms’ reactions to those outside opportunities.

In our model, under the Lazear and Rosen assumptions that women have better non-market opportunities than men, they are less likely to be promoted than men but more likely to receive higher wages if promoted, although this second result depends upon assumptions about the functional form of the distribution of outside opportunities. Women are more likely to quit to non-market opportunities. However, if instead the driving assumption is that women have worse market opportunities than men, they are more likely to be promoted than men, less likely to quit to new jobs, and are likely to receive lower wage gains than men. Alternatively, if the driving assumption is that firms are less likely to match outside offers for women, men and women are equally likely to be promoted, women are likely to receive lower wage gains than men if they remain at the firm or if they leave, and are more likely to quit. We use the term ‘sticky floors’ to refer to the situation—arising either due to worse market alternatives or to less favourable firm responses to outside offers for women—where women are promoted as often as men, but receive lower wage gains consequent upon promotion. In firms with formal wage scales, women remain stuck to the lower wage points on the wage scale of their new, higher job grade.

For the purposes of this paper, a representative panel data source such as the BHPS is particularly appropriate. This is because it covers a large number of workers employed in different firms and occupations, so that the information on promotions and wages arises from a wide range of current practices in the British labour market as a whole. Moreover, the data contain a wealth of individual and work-related characteristics to be used as controls, and the panel nature of the survey allows us to test our model of pay and promotion while accounting for individual unobserved heterogeneity. In contrast, the use of personnel records of a single firm—even if the data set had sufficient individual characteristics recorded—would limit us to the behaviour of that firm, and not allow us to make inferences about average behaviour in the labour market. For example, it is possible that some firms discriminate while others do not. There may also be objective reasons for different promotion practices. These factors may explain why studies based upon single firm data show conflicting results. Cobb-Clark (1998) has recently surveyed the empirical literature, which consists predominantly of studies based upon personnel records from a single firm.¹ She finds that the literature contains no clear results and, depending upon the sample, women are promoted more, less, or the same as men.

The estimates presented in our paper suggest that existing theories about gender differences in promotion need to be supplemented by the sticky floors models of promotion. We find that there is no significant difference in the promotion rates between men and women, an outcome that runs counter to the predictions of some traditional models, but is explicable under the assumptions of the sticky floors models. Using suitable controls, and correcting for unobserved heterogeneity, we also find some (albeit insignificant by conventional statistical standards) support for the sticky floors claim that women gain lower wage increases from promotion. There is, however, some support for the Lazear and Rosen model from the evidence about quits. The only significant gender differences for quits are for those out of the labour force, as that model would predict. Overall, however, it is clear that the promotions process does not systematically mitigate the general disadvantage women face in the labour market. Importantly, however, the mechanisms operating are not the simple glass ceilings ones of an unfavourable promotions rate for women, but involve lower wage returns to promotion for women.