Radical
Statistics

The Journal

The Subjects

The Books

News

Links

About

Home

What does the future hold for women? Connecting gender, pension scheme participation and earnings.

Debora Price

The real pension problem for women: ignoring the gender dimension

Pension scheme participation during the working life is a matter of growing interest to policy makers, economists and other academics. This is particularly so with the closing down of salary-related occupational pension schemes, and increasing reliance in the UK on schemes based on contributions, which build up a fund from which an annuity will be purchased at retirement. Between 1993 and 2002, the Analytical Services Division of the Department of Social Security (1) published no less than fourteen reports on the subject, with others of peripheral relevance (DWP 2002). Within many of these, distinctions based on gender are marginalized. For example, Hedges, in an exploration of retirement planning, devotes two out of 123 pages to the subject, noting principally that women - even young women - seem to depend heavily on their husbands for pension provision; a finding that some might think warranted more emphasis (Hedges 1998: 121-122). In a report on the incomes and living standards of older people, despite the known poverty of older women, gender is hardly mentioned (Whiteford and Kennedy 1995).

Where gender is specifically covered, the results are always of concern. Women are much more likely than men not to belong to a pension scheme or to have lost their rights in a pension scheme; and while during the 1990s, occupational scheme receipts increased for men with each cohort, the same was not true for women (Stears 1997: 170). Indeed, Falkingham and Rake have commented that concern about whether women will have enough income to support themselves in retirement " emerges starkly from virtually every report on pension income" (Falkingham and Rake 2001: 67).

Those concerned about or directly addressing gender issues tend to report on pension coverage - that is numbers and proportions participating in different types of pension schemes (see for example Burton 1997; McKay, et al. 2000; Ginn and Arber 2000). This type of research tends to show that pension coverage between women working full time and men working full time is not that dissimilar (Disney, et al. 2001). But pension coverage is only part of the picture. Women may be covered in the sense of participation, but are they accumulating sufficient pension to support themselves in old age? In simulation models for pension accumulation over the working life using gender-based, age-specific profiles, Falkingham and Rake estimate that if comparing full time female and male workers each working from 18 to 64 and earning the average earnings for their sex throughout their working lives, the woman's total pension income will be only 74 per cent of the man's (Falkingham and Rake 2001: 77-78). With persistent low pay, gaps in employment, or part-time work, the differentials are even greater.

Accumulation of pension depends ultimately on size of contributions (2) . Yet, there has been little attempt to connect pension coverage and size of contributions. In a recent article using panel data (3) to examine who contributes to what type of pension scheme in Britain, and how much they contribute, the authors comment:

" Such information is not generally publicly available, somewhat surprisingly (since it is basic to an understanding of the UK's pension programme), and this represents, to our knowledge, the first attempt to compile these data."
(Disney, et al. 2001: 71)

This study shows, as others had before, that women are more likely to work part-time, and that this factor led to more women being without a private pension - i.e. the 'pension coverage' issue. In the analysis that then follows - an analysis of the size of contributions to pension schemes that is 'basic to the understanding of pension provision' - gender is not included as a differentiating variable. This trend in research has led the Equal Opportunities Commission to comment:

" Lack of a gender perspective in data collection, research and policy evaluation compounds existing inequalities between women and men as the differences are invisible in policy analysis."
(Equal Opportunities Commission 2002:2)

In this essay, I argue that in the data-imperfect world in which researchers operate, it is not necessary to have panel data to contribute to the pool of knowledge about the size of contributions to pension schemes, nor is it necessary to have details of the size of contributions. Sufficient information for some reasonable and reliable conclusions to be drawn can be gleaned from cross-sectional datasets - such as the General Household Surveys - which use large probability samples and ask reasonably detailed questions about pension coverage.

Women's Earnings and a Structural Relationship with Pension Provision

The pension system in the United Kingdom is horrendously complex in its detail (4), but in broad terms it can be simply stated.

1. Until very recently (5) those without earnings from paid work could by law make no contribution to a pension, other than to the basic state pension(6) . The basic state pension now pays well below the lowest threshold for means tested benefits.
2. For those earning, if they are earning below a government-defined limit called the 'Lower Earnings Limit' (" LEL" ), they do not automatically make pension contributions to the State pensions, are not required to do so, and are also not compelled to make any pension contributions to a private pension. The LEL is so low that even if these individuals were contributing voluntarily to a private pension at 15 per cent of their earnings(7), they would be accumulating hardly any pension. For data collected in the years examined here, the Lower Earnings' Limits were, per week:

1. Those earning above the Lower Earnings Limit must contribute to the Basic State Pension. In addition,
   • They must also contribute to a State Second Pension at a prescribed rate (until recently, SERPS, now the State Second Pension or S2P); or
   • Their employers must guarantee this element as part of an occupational pension scheme(8) ; or,
   • They are permitted to take out a compulsory second pension, a private pension, which is intended to replace SERPS. The prescribed rate is again, a percentage of earnings(9) .
2. It is open to paid workers to make voluntary contributions over and above these compulsory contributions. Their employer might, if they are lucky, make these on their behalf; or, both they and their employer might contribute; or they alone might contribute. In each case, contributions are only up to a legally permitted maximum. Employed people may contribute up to 15 per cent of gross pensionable earnings, but this maximum is rarely contributed(10) . Where employers contribute to occupational schemes, their average contribution is 5 per cent(11) ; about 20 per cent of these schemes have no further contributions from employees. Where employees contribute, their average contribution varies between 4 per cent and 6 per cent depending on the type of scheme (Smith and McKay 2002: 4); for those making voluntary contributions to private non-occupational schemes, the average is about 5.5 per cent (Disney, et al. 2001:78).

Therefore, even if these voluntary contributions are being made by employees and/or employers, the distribution of contributions is in reality quite concentrated between 5 per cent and 11 per cent of earnings. We can have a reasonable idea of the distribution of pension contributions, simply by knowing the earnings of those contributing. We can therefore make reasonable, if somewhat broad, inferences from the data.

Methods

Three years of the General Household Surveys are combined and examined: 1993-4, 1994-5, and 1995-6 (OPCS 1996; 1997; 1998) for men and women aged 20 to 59: n=36,698.

Earnings are gross weekly earnings from work(12). Some analysis includes those with zero earnings (where shown). Although it is conventional to analyse employment status by full or part time employment, here a categorisation according to whether a person is earning above or below the Lower Earnings Limit is substituted because of the important relationship with pension provision (as explained above). Those below this threshold will be mostly part-time workers. But part-time workers who are professionals, managers or self-employed, may earn well above this limit. The Lower Earnings Limit therefore measures the low paid, and importantly measures workers who are not even acquiring credits to the basic state pension let alone a state or other second pension.

Following the description of pension schemes above, five categories are analysed:

1. Those with no earnings
2. Those earning below the lower earnings limit (the 'LEL').
3. Those who have only a State Second Pension or a legally permitted private alternative (with no additional contributions above those that are compulsory). This categorisation differs from conventional analysis of pension coverage, by grouping those with only a 'minimum' private pension with those who have only State pensions, rather than with those who have private pensions. This is because the numbers who have these 'minimum' private pensions are an artefact of the UK pension system; these schemes(13) are nothing more than a replacement for SERPS - benefits are not necessarily likely to be any better. In fact, for significant numbers (mostly women), benefits are likely to be worse.
4. Those who participate in private (non-occupational) schemes who are indeed making pension contributions that are in addition to their 'state' pension entitlements, whether this 'state' pension entitlement comes from the basic state pension, a state second pension, or a legally permitted but compulsory private alternative to a state second pension.
5. The final category is those who belong to an occupational pension scheme. While mostly, but not necessarily, providing better benefits than other private schemes, an occupational scheme has other characteristics that are valuable to members(14).

Gender Inequalities in Pension Provision

Figure 1 shows that by far the majority (66 per cent) of those not earning are women. While the paid workforce is split almost half and half between men and women, the gender imbalance according to whether a worker is earning above or below the lower earnings limit is starkly shown: 80 per cent of those earning below the LEL are women, but only 45 per cent of those earning above it.

In the two columns on the left of Figure 1, effectively no pension is being acquired. The vast majority of those acquiring no pension are female.

Click HERE for Figure 1: Those not earning, earning below the National Insurance Lower Limit and Earning above the National Insurance Lower Limit by gender; men and women aged 20 to 59

Figure 2 shows the proportions of men and women in each of the five categories of pension scheme participation outlined above. Grouping those who have only minimum private pensions in substitution for state second pensions with those who have a state second pension is revealing. Whereas the proportions of men and women with either a state earnings related pension or private 'equivalent' both hover around 20%, far higher proportions of men than women have 'additional' private (non-occupational) pension schemes.

Far higher proportions of men than women also have current occupational scheme membership - the far right of the graph at Figure 2(a) (the top of the graph at Figure 2(b)). The two

Click HERE for Figure 2: Participation in Pension Schemes by Gender, men and women aged 20 to 59, percentages (a) Bar Chart (b) Stacked Bar Chart

Columns labelled 'pension rich' are people who either have 'additional' private pensions, or are in an occupational scheme, or both. The proportion of men is much higher than the proportion of women - 56 per cent of men are in these two categories, compared with only 31 per cent of women. These categories combined are referred to in this essay as 'making contributions to an additional pension scheme'.

Connecting Earnings

To make inferences about the extent of difference between the size of men's and women's contributions to pension schemes (and therefore pension accumulation) the relationships between earnings, gender and pension contributions are examined next. Excluding those with zero earnings (mostly women - see above), earnings are classified by decile. The proportion of those in each earnings decile making contributions to an additional pension scheme are shown in Figure 3. Not surprisingly, the relationship is almost perfectly linear. Over 90 per cent of people earning in the top decile make additional pension contributions, and fewer than 5 per cent in the bottom decile.

But it must always be remembered that ageing is a progressively gendered experience. Older people tend to be women, particularly the very old. It is therefore not sufficient, either in

Click HERE for Figure 3: those making contributions to a pension scheme in addition to the state schemes/ compulsory private schemes as a proportion of earnings deciles*, men and women aged 20 to 59

The examination of pension purchasing behaviour, or in projecting to the future pension or means tested benefits needs of the country, to look at all people. Pension coverage and its link to earnings should always be examined according to gender.

Figure 4 shows the identical graph to Figure 3, but now the gender profile within those making additional pension contributions in each decile is shown. The first four deciles are composed almost entirely, or mostly of women. And those here shown within these deciles as making additional pension contributions are also mostly women. They may be making additional contributions, but those contributions are being made as a small proportion of the lowest 40 per cent of earnings - in these groups, everyone earns less than £ 191 per week: a 5 per cent contribution of even the top wage is less than £ 10 per week. In the higher deciles (comprised mostly or almost entirely of men), those making additional pension contributions, as an equally small proportion but of much higher earnings, are predominantly men.

Click HERE for Figure 4: those making contributions to a pension scheme in addition to the state schemes/ compulsory private schemes as a proportion of earnings deciles* according to gender, men and women aged 20 to 59

This leads to an obvious conclusion: it is insufficient to examine pension coverage as such. If the women within any particular category of pension coverage are earning much less than the men, then their contributions will be much lower - perhaps so low that they will not be lifted out of means tested benefits when older, despite participating in additional pension schemes.

To elucidate this issue, the earnings distributions for those men and women who participate in each category of pension scheme outlined above are shown in Figure 5. Here, the extent of women's double disadvantage can be seen. Within each category, the women who contribute to that type of pension scheme earn significantly less than the men. The top of the box (showing the 75th percentile) on the darker box-plots (women) is either well within the range of the third quartile (25th to 50th percentile) of the lighter box-plots (men), or at best, equivalent to men's median earnings for that category.

This graph illustrates four main points:

• A clear hierarchy of financial advantage represented by the categories of pension scheme membership shown.
• At each category of pension membership, the women who are making the same type of contribution as the men are doing so from much lower average earnings.
• Even within the same category of contributions, since pension contributions are usually a fixed percentage of earnings, it is reasonable to infer that the women will be making on average lower contributions than the men and thus, while acquiring the same type of pension provision, will be acquiring a smaller amount of pension.
• Women appear to enter additional pension schemes at lower earnings than men. This finding alone shows that analysis of data by pension scheme membership obscures the severity of the pension disadvantage for women.

Click HERE for Figure 5: Earnings' distribution according to gender and type of current pension contributions, age 20 to 59

Participation may depend on earnings, but clearly, not earnings alone. Apart from compulsory contributions, a conscious decision must be made to enter a pension scheme. This is so even when a pension scheme accompanies employment, for since April 1988 it has been possible for employees to elect not to join their employer's pension schemes (Social Security Act, 1986). The remainder of this essay examines how women's and men's choices appear to differ within their earnings constraint.

For those in paid work, the differential effect of the earnings gap on pension scheme participation of men and women is large. Figure 6 shows the distribution across earnings deciles of those who make additional contributions to pension schemes, by gender. For men, the highest proportion contributing earn the most. The proportions then move in a step-like fashion down to the lowest proportion contributing, who earn the least. About 70 per cent of men contributing to additional pensions are in the highest 30 per cent by earnings. This is not so for women. The distribution is far more even through the deciles. Half of all women making additional pension contributions are in the top 50 per cent of earnings, and half in the bottom 50 per cent of earnings.

Click HERE for Figure 6: Distribution of those making some form of additional pension contribution across earnings deciles*, men and women aged 20 to 59

Figure 7 shows the distributions for men and women across earnings deciles of those who are not making additional contributions to pension schemes. Over 55 per cent of women who make no additional contributions to their pensions are in the bottom 20 per cent of earnings - earning less than £ 102 per week. This is not so for men, who are distributed fairly evenly across earnings deciles.

Click HERE for Figure 7: Distribution of those making no additional pension contribution across earnings deciles*, men and women aged 20 to 59

These gender effects are largely because of the concentration of women in lower deciles and of men in higher deciles of earnings. This raises the question: what are the differences within each decile between the women and the men? Table 2 lists the probabilities by gender within each decile of making additional pension contributions. There is no gender difference in the proportions making contributions among the lowest 20 per cent of earners. But among those in the 3rd to 7th deciles, this analysis reveals the perhaps surprising result that within each, the women are more likely than the men to be making contributions to an additional pension scheme. Beyond that, men and women are again similar, but it is relevant to note that only 14 per cent of women earn in the top 30 per cent of earnings.

Click HERE for Table 2: Percentages making additional contributions to pension schemes by gender and earnings decile*

Overall, men are much more likely than women to be making additional pension contributions (71 per cent as opposed to 47 per cent). But this is simply because of women's numerical concentration in the lower deciles of earnings. In the third decile (£ 101 to £ 150), for example, women are almost one and a half times as likely as men to be making additional pension contributions. Table 3 shows the results of a logistic regression analysis, which defines the odds of a woman making additional pension contributions as 1, and gives the odds ratios for men making additional pension contributions. Model 1 gives the unadjusted odds ratios, and then Model 2 gives the odds ratios holding earnings decile constant (here the highest decile is defined as having odds of contributing to an additional scheme = 1).

Click HERE for Table 3: Odds Ratios for contributing to an additional pension scheme according to gender and earnings decile†; men and women age 20 to 59

While the odds of a man making additional pension contributions are almost three times the odds of a woman doing so, once earnings deciles are held constant, the odds ratio reduces to 0.84. This reversal of direction is usually obscured because of the concentration of women in low paid jobs.

Conclusions

While those who contribute to additional pension schemes may well have the best chance of escaping means testing in old age, policy analysis that is limited to pension coverage only tells part of the story. There should be no complacency in reading statistics that suggest either that women's pension coverage is increasing, or that full time working women are almost equivalent to full time working men in pension coverage.

The amount which can be contributed to additional pension is limited by earnings, and by the percentage of earnings which it is legally possible to contribute - a maximum of 15 per cent for employees, but with most contributing far less than this. The very large disparities shown in this essay between the earnings of those women who are contributing to additional pension schemes, and the earnings of those men who are contributing to additional pension schemes, indicates the scale of pension inequality by gender in the UK. In the mid-90s, half of the women making additional contributions to their pensions were earning less than £ 228 per week, whereas 82 per cent of men who were making additional contributions were earning more than this.

The earnings gap is the fundamental cause of the 'pension problem' for women. This research shows that women who do earn would appear to be trying quite hard to make pension provision - harder than men, but that their persistent low earnings, and their low earnings relative to men, prevent them from doing so effectively.

REFERENCES

Burton, D. 1997 'Ethnicity and Occupational Welfare: A Study of Pension Scheme Membership in Britain', Work, Employment & Society 11(3): 505-518.

Disney, R., Emmerson, C., and Wakefield, M. 2001 'Pension Reform and Saving in Britain', Oxford Review of Economic Policy 17(1): 70-94.

DWP 2002 ASD Research Division Reports Index, Last updated 26 Mar 2002 http://www.dss.gov.uk/asd/asd5/rrs-index.html

Equal Opportunities Commission. 2002 Response to Modernising Annuitities: A Consultative Document, February 2002, London: Equal Opportunties Commsission.

Falkingham, J. and Rake, K. 2001 'Modelling the gender impact of British pension reforms' in J. Ginn, D. Street, and S. Arber (eds.) Women, Work and Pensions, Buckingham: Open University Press.

Ginn, J. and Arber, S. 2000 'Personal pension take-up in the 1990s in relation to position in the labour market', Journal of Social Policy 29(2): 205-228.

Hedges, A. 1998 Pensions and Retirement Planning: Department of Social Security Research Report No, 83, Leeds: Corporate Document Services.

McKay, S., Heaver C., and Walker R. 2000 Building up Pension Rights: Department of Social Security Research Report 114, Leeds: Corporate Document Services.

OPCS. 1996 General Household Survey, 1993 - 1994 [computer file]. Colchester: The Data Archive.

---. 1997 General Household Survey, 1994 - 1995 [computer file]. Colchester: The Data Archive.

---. 1998 General Household Survey, 1995 - 1996 [computer file]. Colchester: The Data Archive.

Smith, A. and McKay S. 2002 Employers' Pension Provision 2000: Department for Work and Pensions Research Report No. 163, Leeds: Corporate Document Services.

Stears, G. 1997 'Occupational and Other Non-State Pensions' in R. Disney, E. Grundy, and P. Johnson (eds.) The Dynamics of Retirement, London: The Stationery Office.

Whiteford, P. and Kennedy S. 1995 Incomes & Living Standards of Older People: Department of Social Security Research Report No. 34, London: HMSO.

NOTES

1 Now the Department for Work and Pensions.

2 This is a simplification of the way that salary-related schemes work, but it is true to say that the better the benefits offered the more must be contributed to the scheme.

3 The Research used the first nine waves of the British Household Panel Survey (1992 to 1998), analysing those aged 20 to 59 present in all nine waves, excluding the self-employed.

4 Too complex to describe here.

5 The change was through the introduction of 'stakeholder pensions' in October 2001, which allowed non-earners to acquire a non-state pension scheme for the first time.

6 By a system of 'deemed credits' for certain categories of people out of work including the unemployed, the sick and those caring for children.

7 The maximum permitted for employees; the percentage limits for self-employed people are higher, but they do not acquire any state second pension or any compulsory equivalent; in reality only a tiny proportion of workers in this category make any pension contributions and these are highly unlikely to be at 15% of their gross earnings.

8 Technically, an employer could provide a pension only at the level of SERPS and nothing more (at the time of data collection called a Guaranteed Minimum Pension), but this is very rare and has been discounted for the purposes of this essay; note also that the strength of this guarantee has since reduced.

9 The mechanism is complex through National Insurance Contributions and/or a rebate of National Insurance Contributions and there are slight variations in the contributions/rebate according to circumstances.

10 For employees, the maximum is subject to an earnings cap. The rules for self-employed people are far more complex and also age-related; relatively few women are self-employed and most of these earn very little.

11 Employers' contributions can be much larger in certain circumstances, but this is rare.

12 Earnings have not been adjusted over the three years.

13 APPs, or 'Appropriate Personal Pensions'.

14 For example, discretionary uprating, dependants' benefits, life and disability benefits, and for most in this survey, salary-related schemes as opposed to defined contribution schemes.

Debora Price,
Centre for Research on Ageing and Gender,
Department of Sociology,
University of Surrey