The Gender Gap in Private Pensions

Older women in Britain receive considerably less private pension income than older men, on average. We analyse this differential by examining differences between the sexes both in private pension coverage and in pension income conditional on receipt. Using regressionbased decompositions, we show that both gaps are associated mainly with gender differences in returns to personal characteristics rather than with gender differences in personal characteristics per se. In particular, although there are marked differences between elderly men and elderly women in their lifetime employment histories, these differences account for only a small fraction of the overall private pension income gap between the sexes. NON-TECHNICAL SUMMARY Differences in income from occupational and personal pensions and annuities (‘private pension income’; PPI for short) are one of the most importance sources of income inequality among British pensioners. There is also a substantial difference between the average income of older men and of older women, much of which has been attributed to differences between the sexes in private pensions. Only one third of female pensioners over 65 received any PPI in 1993–94, compared with two-thirds of male pensioners. Moreover, according to the former Department of Social Security, ‘[t]he main difference [in the net income of men and women] occurs in occupational pension income. In 1997/8, single men received £44 a week on average from this source, compared with £29 for single women. Average incomes from the other sources were very similar for single men and women. In this paper we investigate the sources of the differences in PPI between British men and women aged 66 and older, addressing several related questions. To what extent is the gender gap due to women having lower coverage rates than men or due to having lower pension incomes conditional on coverage? To what extent does the gender gap arise because women’s characteristics are less well rewarded in pension terms than men’s? And to what extent does the gap arise because women have less advantageous personal characteristics than men? In particular, what is the role played by women’s lower lifetime labour market participation rates and their interrupted working lives? Answers to these questions are relevant to the formulation of pension policy, including whether measures should be directed at the labour market or at pension schemes per se. For example, even if personal pension coverage rates among women rise, their subsequent PPI receipts will remain low if also earnings remain relatively low or working lives are relatively short or interrupted. In the past, this was not necessarily seen as a problem, as it was assumed that women would benefit from the pension entitlements of their husbands. This supposition is increasingly less appropriate, given rising divorce rates and growing expectations of financial independence for women regardless of marital status. We examine the gender gap in private pensions using a regression-based decomposition framework, based on a joint model of the probability of PPI receipt and of PPI amounts conditional on receipt. Using data from the British Household Panel Survey, we fit the model separately for older men and women and use the estimates to decompose the differences between the sexes on both outcomes, combining applications of methods due to GomulkaStern and to Blinder-Oaxaca. We show that there is a substantial gender gap in private pensions in Britain, with two distinct components. First, there is the shortfall for women in the likelihood of receiving any PPI at all (only one third of elderly women receive any PPI, compared with about threequarters of elderly men) and, second, there is the lower PPI received by female recipients compared to male recipients (about £40 per week on average compared with £80 per week). Both components of the overall gender gap arise mainly because women’s characteristics are less well rewarded than men’s, rather than because women have less advantageous personal characteristics than men. In particular, differences in returns account for at least four-fifths of the PPI gap among recipients. Put another way, the role of differences in characteristics is greatest in accounting for the gap in PPI receipt probabilities but, even in this case,