Pension protection in the United Kingdom (UK)

Pension income can be received from a number of different sources, including the state, the private market, the labour market and the family (Price Reference Price2008). The British pension system has historically relied on a relatively low-value State Pension, which is contributory, flat-rate and worth approximately 18 per cent of the national average earnings, and a relatively small public earnings-related scheme, topped up by means-tested benefits for those on low incomes and by private pensions for those with middle and high incomes (Pensions Policy Institute 2013). As a result, occupational/private pensions have been considered critical in securing financial adequacy for older people (Bardasi and Jenkins Reference Bardasi and Jenkins2010). The means-tested benefit in the form of Pension Credit has ‘topped-up’ the incomes of poorer pensioners, but has also been associated with persistent non-take-up by approximately one-third of eligible recipients (Department for Work and Pensions 2012). The eligibility criteria for receiving each of these pension incomes are different; the discussion here refers to criteria in place at the time when the analysed data were collected. The State Pension is a flat-rate amount received by individuals who, in this analysis, had made at least 39 years' worth of contributions (this was changed to 30 years of contributions and more recently to 35 years starting from 2016); Pension Credit is a means-tested benefit which is composed of a Guarantee and a Savings element, and which takes into account one's income from pensions, benefits, earnings and other sources; and finally occupational/private pension income is received by individuals who have paid into an employer's pension scheme or a private pension account during their working life.

In order to capture the relative importance of different sources of pension income for older people from BME groups, three types of pension income are examined in the analysis: the State Pension, occupational/private pensions and Pension Credit. The distinction between the three types of pension income is critical when studying ethnic differences, as existing research has exposed a gap in the percentage of individuals qualifying for a State Pension or saving through a private pension between BME groups and the majority White population (Steventon and Sanchez Reference Steventon and Sanchez2008).

Conceptualising the concept of ethnicity

Examining the relationship between ethnicity and pension protection in later life requires an understanding of how both concepts have been operationalised in existing empirical analyses. In many nationally representative surveys, including the one used for the analysis in this paper, ethnicity is self-defined by respondents. However, both the definition of the concepts of ethnicity and ethnic group, and the categorisation of responses are contested and dynamic areas, reflected in changes in the phrasing of questions, the available responses and the final categorisation of responses by researchers across time. For example, an ethnic group might refer to a group sharing common characteristics, such as religious affiliation, language and cultural norms, while at the same time, such characteristics may be used to distinguish a specific group from others, thereby denoting difference (Wimmer Reference Wimmer2008). Similarly, ethnicity might be related to, but not be defined exactly as, other concepts such as one's race, nationality, migration history and heritage. Burton, Nandi and Platt (Reference Burton, Nandi and Platt2008) emphasise the difference between conceptualising ethnicity as one part of an individual's identity – an approach favoured by psychologists; and perceiving ethnicity as contributing to social stratification through attaching particular outcomes to groups with particular characteristics – an approach which tends to be used by other researchers and by policy makers. The same authors point at the apparent contrast between the subjective measurement of ethnicity through self-defined questions, and the aim to obtain a reliable or ‘objective’ definition of individuals' ethnicity (Burton, Nandi and Platt Reference Burton, Nandi and Platt2008). In this paper, we consider ethnicity to be one of the self-defined demographic and socio-economic characteristics which contribute to an individual's or group of individuals' identity, and which can help explain particular outcomes for such individuals or groups – in this case relating to pension protection in later life.

The use of pre-designed categories in the self-defined question on ethnicity, and its implications for the measurement of ethnicity over time, has also been studied to a significant extent, especially since 1991, when the question on ethnicity was first included in the UK Census. Both the wording of the question and the available response categories have changed between the 1991, 2001 and 2011 Censuses, making comparable research more challenging, on the one hand, albeit with the aim of providing more meaningful questions for the UK's increasingly diverse population (Jivraj Reference Jivraj2012). For example, the 1991 Census question on ethnicity enquired ‘which ethnic group [individuals] descend from’, while the 2001 Census asked after ‘individuals' ethnic group in terms of cultural background’ and, finally, the 2011 Census asked individuals to simply define their ‘ethnic group or background’. In spite of such changes, and challenges associated with them (e.g. self-definition of children in inter-racial unions), the evidence on the UK's increasing ethnic diversity is undisputable (Office for National Statistics 2012). The proportion of the UK's non-White population, which in broader terms could be taken to refer to ethnic minorities, increased from 7 per cent in 1991, to 9 per cent in 2001 and 14 per cent in 2011 (Jivraj Reference Jivraj2012). Challenges associated with the changing nature of ethnicity questions are not dealt with in this paper, which presents cross-sectional analysis of ethnic minority groups. For the purposes of this paper, individuals are simply asked ‘what is your ethnic group?’, and provided with 18 possible responses (including ‘other’ which is further analysed), which are then grouped into eight categories.

Pension protection among minority ethnic groups in Britain

Empirical evidence comparing pension protection among older individuals from different ethnic groups shows significant differentials not only between the White majority and the ethnic population as a whole, but also between different ethnic groups and the different genders. According to the 2013/14 Pensioner Income Series, the mean gross income of pensioner units between 2010 and 2014 was £492 per week for White British individuals, £423 per week for Asian/Asian British/Chinese individuals and £350 per week for Black/African/Caribbean/Black British individuals (Department for Work and Pensions 2015: table 5.4). Examining the composition of this income reveals further gaps, with 98 per cent of White British pensioners receiving a State Pension compared to 84 per cent of the Asian and 88 per cent of Black elders, and conversely 27 per cent of White British receiving income-related benefits compared to 44 per cent of Asian and 45 per cent of Black minority elders. In addition, less than one-third of Asian elders (30%) and 47 per cent of Black elders received income from an occupational pension compared to 63 per cent of White British pensioner units (Department for Work and Pensions 2015: table 5.4). This contributes to 47 per cent of Indian pensioners being in the bottom two quintiles of the income distribution for the pensioners' population, compared to 45 per cent of Black/African/Caribbean/Black British pensioners and 36 per cent of the majority White population (Office for National Statistics and Department for Work and Pensions 2015). Earlier analysis of the Family Resources Survey between 1994 and 2008 indicated that individuals up to the age of 65 from Pakistani, Bangladeshi and Black heritage consistently report having the lowest incomes as well as the lowest savings for retirement (Gough and Adami Reference Gough and Adami2013). In addition, Pakistani and Bangladeshi pensioners have been shown in both quantitative and qualitative research to be more likely to rely on means-tested benefits than pensioners from other BME groups (Berthoud Reference Berthoud1998; Burton Reference Burton1996; Gough and Hick Reference Gough and Hick2009).

Such differentials may be the result of a complex set of reasons, including the timing of arrival and commencing social insurance contributions in the UK, employment patterns, relatively low-paid earnings across the lifecourse when in work, as well as cultural reasons. Low income over the lifecourse can be directly affected by low participation rates in the labour market for particular ethnic groups, and research has consistently revealed a disadvantage amongst Pakistani and Bangladeshi communities in this respect (Ginn and Arber Reference Ginn and Arber2001). Between July and September 2013, 78 per cent of White men aged 16 and over were in the labour market, compared to 62 per cent of Pakistani and 64 per cent of Black/African/Caribbean men of the same age, while the equivalent rates for women were 69 per cent among the White group, 31 per cent among Pakistani women, 58 per cent among Black/African/Caribbean women and 34 per cent among Bangladeshi women (Office for National Statistics 2013). Other research has shown that Pakistani and Bangladeshi pensioners are less likely than White pensioners to contribute to, and be aware of, an employer's pension scheme (Nesbitt and Neary Reference Nesbitt and Neary2001). Finally, gender can further contribute to the ‘ethnic pension penalty’, with Bangladeshi and Pakistani women being the least likely among all men and women from BME groups to be benefiting from a private pension scheme (Ginn and Arber Reference Ginn and Arber2001; Gough and Adami Reference Gough and Adami2013).

The policy significance of examining differentials in terms of pension protection also relates to the fact that particular ethnic communities experience further dimensions of disadvantage which can affect wellbeing in later life. Research has consistently shown that Pakistani, Bangladeshi and Indian individuals aged 50 and over are more likely to report a limiting long-standing illness, which can affect their capacity to engage in paid work and contribute to their pension entitlement (Evandrou Reference Evandrou2000a ). At the same time, women from these three ethnic communities are more likely to provide care of 50 hours or more per week than their counterparts from the majority White population or other ethnic groups (Office for National Statistics 2005), which can have an adverse effect on their labour market participation. Other factors associated with one's living arrangements may also contribute to such differentials. For instance, older persons from BME groups are more likely to live in large households, which may be advantageous in terms of family support and the pooling of resources within the household, but on the other hand, may have an adverse impact in terms of overcrowded accommodation and the inadequacy of pooled resources (Pensions Policy Institute 2013).

Data and methodology

The research employs Wave 1 data (collected between January 2009 and March 2011) from Understanding Society, which is a longitudinal survey of the members of approximately 40,000 households in the UK. The sample of Understanding Society used here includes the General Population (GP) and an Ethnic Minority Boost sample (EMB). The addresses of the sample were randomly selected from the Postcode Address File in Great Britain and the Land and Property Services Agency list of domestic properties in Northern Ireland. In each household, all individuals aged ten years and over were eligible for interview. Computer Aided Personal Interview was used to administer the household and individual adult questionnaires. The response rates for the GP and EMB were 82 and 73 per cent, respectively (Boreham, Boldysevaite and Killpack Reference Boreham, Boldysevaite and Killpack2012). Although Wave 1 is the not the most recent wave in this survey, significant attrition in the ethnic groups between Waves 1 and 3 render Wave 1 the most appropriate for yielding adequately large cell counts for older individuals from ethnic groups. The data-set is ideal for this study, as it includes an EMB sample, designed to include at least 1,000 individuals from five ethnic groups: Indian, Pakistani, Bangladeshi, Caribbean and African. The analytical sample for this paper includes all adults aged from the State Pension Age and over (60 for women, 65 for men at the time of data collection), totalling 10,382 respondents, of whom 601 came from the five ethnic groups above. This relatively large sample size allows the study of a sub-group of the population, i.e. elders from minority ethnic heritage, which has been difficult in the past due to low cell counts.

The analysis examines pension protection in later life using three separate indicators: whether the individual is currently in receipt of a State Pension; whether they are currently in receipt of an occupational or private pension; and whether they are currently in receipt of Pension Credit. Table 1 shows the descriptive statistics for the sample. An individual's ethnicity was analysed using a self-reported variable with eight response categories (White British, Other White, Indian, Pakistani, Bangladeshi, Caribbean, African and Other Ethnic groups (includes Mixed, Other Asian and Other Ethnic group). This facilitated analysis of differentials between BME elders and the White British majority, as well as between individuals from different minority ethnic groups.

Table 1.

Descriptive statistics of study population (aged State Pension Age and over)

Source: Understanding Society (2009 – 11).

Figures 1–3 show the percentage of men aged 65 and over, and women aged 60 and over from different ethnic groups reporting being in receipt of income from these three sources. Key differentials are evident; between White British elders and BME elders; between elders from different BME groups; and between older men and women. In terms of receiving the State Pension, White British older men are most likely to be in receipt of the State Pension (94%), compared to 67 per cent of African men. Among older women, the ethnic differences appear to be more prominent, with 91 per cent of White British women receiving the State Pension, compared to 62 per cent of African women (Figure 1).

Figure 1.

Percentage of older individuals receiving the State Pension, by ethnic group and gender, 2009 – 11.

Note: SPA+: State Pension Age and over. Significant at p < 0.001 Source: Understanding Society (2009 – 11).

Figure 2.

Percentage of older individuals receiving an occupational/private pension, by ethnic group and gender, 2009 – 11.

Note: SPA+: State Pension Age and over. Significant at p < 0.001 Source: Understanding Society (2009 – 11).

Figure 3.

Percentage of older individuals receiving Pension Credit, by ethnic group and gender, 2009 – 11.

Source: Understanding Society (2009 – 11).

Among older men, White British are most likely to be in receipt of an occupational/private pension (84%), compared to 19 per cent of Pakistani men, 25 per cent of African men and 33 per cent of Bangladeshi men (Figure 2). Older women are overall less likely to be receiving an occupational/private pension compared to men, however, important ethnic differences can also be noted. Older women from the Caribbean are the most likely to be receiving such pensions (54.5%), reporting higher levels of receipt than White British women compared to 30 per cent of Indian women, while there are no Pakistani women in the sample in receipt of such pensions (Figure 2). For those being in receipt of Pension Credit, both men and women of Bangladeshi origin are the most likely to be in receipt of Pension Credit (20%), while Caribbean men and Pakistani women are the least likely to be receiving this benefit (7 and 6%, respectively) (Figure 3). Although indicative of patterns for the receipt of this benefit, probably due to low cell counts for this variable, the cross-tabulation of this relationship is not statistically significant.

Such differentials go some way towards measuring the impact of ethnicity on older persons' chances of having adequate pension protection, or highlighting the risk among particular groups of men and women, of having inadequate pension protection. Multivariate analysis using three separate binary logistic regression models helps to disentangle the relative importance of ethnicity in terms of older individuals' pension protection, taking into account key characteristics at the same time. As such, this analysis addresses the research question ‘What are the characteristics (including ethnicity) associated with receiving a State Pension or an occupational/private pension or Pension Credit?’ Building on existing literature and Figures 1–3 which show significant gender differences in pension protection, the three regression models were also run separately for men and women, in order to understand the determinants of receiving a certain type of pension income in later life.

The explanatory variables used in the multivariate analysis reflect demographic characteristics (age, gender, marital status, living arrangements); health status (self-reported general health, report of a limiting long-standing illness); and socio-economic status (highest educational qualification, housing tenure, whether one has ever worked). In addition, the analysis controls for two factors which can affect individuals’ employment status and history, and by extension, their eligibility for receiving different kinds of pension income. The first is whether an individual has been born in the UK, and if not, the length of time they have been in the UK (less than 10, 10 – 39, 40 – 49, 50 or more years). Such an indicator is important to take into account, as existing literature shows cohort differences in the employment and savings behaviour of individuals from particular BME communities, such as the Pakistani and Bangladeshi (Nesbitt and Neary Reference Nesbitt and Neary2001), and second-generation Indians and Pakistanis being more likely to set up dual-earner households than their parents (Barnes and Taylor Reference Barnes and Taylor2006). The second factor is whether an individual has difficulty speaking English on a day-to-day basis, drawing upon the evidenced link between fluency in English and employment prospects among non-White individuals (Dustmann and Fabbri Reference Dustmann and Fabbri2003). The choice of other variables included in this part of the analysis is based on evidence included in existing literature showing, for example, the association between pension protection and gender (Foster Reference Foster2013) or pension protection and working life history (Sefton et al. Reference Sefton, Evandrou, Falkingham and Vlachantoni2011); and the use of educational qualifications and housing tenure as proxies for older people's socio-economic status (Evandrou Reference Evandrou2000b ). Most of the variables were used in the analysis following minor merges of categories of the original variables. However, the derivation of certain variables is worth detailing further. In order to gain a balanced understanding between positive and negative self-reporting of general health, the original five-category variable (excellent, very good, good, fair, poor) was used to derive a three-category variable, which merged the three ‘positive’ categories into one ‘good’ category, and maintained the ‘fair’ and ‘poor’ categories. The construction of the variable indicating the reporting of a limiting long-standing illness was achieved using two separate variables, i.e. whether the respondent experiences limitations in their activities due to health and whether they face a long-standing illness in their everyday life. As the two variables are asked separately, the derived limiting long-standing illness variable includes a category of respondents who reported no long-standing illness but who still reported limitations in their everyday life. Tables 2–4 present the final models of the regression analysis, which include the explanatory variables exhibiting a statistically significant relationship with the outcome variable in each case, i.e. the receipt of different types of pension income.

Table 2.

Determinants of currently receiving a State Pension (final model)

Notes: N = 10,393. Ref.: reference category. Highest educational qualification was excluded from the final model as it was not statistically significant.

Source: Understanding Society (2009 – 11), authors’ calculations.

Significance levels: * p < 0.05, ** p < 0.01, *** p < 0.001.

Table 3.

Determinants of currently receiving an occupational or private pension (final model)

Notes: N = 9,637. Ref.: reference category. GCSE: General Certificate of Secondary Education.

Source: Understanding Society (2009 – 11), authors’ calculations.

Significance levels: * p < 0.05, ** p < 0.01, *** p < 0.001.

Table 4.

Determinants of currently receiving Pension Credit (final model)

Notes: N = 10,390. Ref.: reference category. GCSE: General Certificate of Secondary Education. Self-reported general health was excluded from the final model as it was not statistically significant.

Source: Understanding Society (2009 – 11), authors’ calculations.

Significance levels: * p < 0.05, ** p < 0.01, *** p < 0.001.

The association between ethnicity and the receipt of pension income

Table 2 shows that ethnicity has a significant effect on older individuals’ chances of being in receipt of the State Pension. The model for both men and women indicates that Pakistani and African individuals, and individuals from Other Ethnic groups are significantly less likely to receive a State Pension than the White British majority. Beyond the impact of ethnicity, the model shows that one's chances of being in receipt of the State Pension are also affected by other demographic, health and socio-economic characteristics. Such chances increase with age, as women are 23 per cent less likely than men to be in receipt of the State Pension, while divorced/separated individuals are 30 per cent more likely than married individuals to be receiving a State Pension. In terms of socio-economic variables, older individuals’ odds of being in receipt of a State Pension are highest for owner-occupiers and those who have ever worked. Crucially, when considering the impact of ethnicity, individuals who have spent less than ten years, or between ten and 39 years, in the UK are 94 and 64 per cent less likely, respectively, than those born in the UK to be in receipt of a State Pension, indicating a ‘migration pension penalty’. Finally, the report of fair or poor health reduces one's odds of receiving the State Pension, while individuals who reported a limiting long-standing illness are 48 per cent more likely than those not reporting an illness, to be in receipt of a State Pension. Education was not statistically significant in the final models (both in the model for men and women together, and in the separate gender models), and as such, education is not added in the final models for this outcome variable.

When the model was run separately for men and women, the adverse effect of ethnicity among men became more evident. Indian, Pakistani, Bangladeshi and African men, as well as men from Other Ethnic groups, are less likely to be receiving a State Pension than White British men, while such an adverse effect is not statistically significant for women from BME communities. No significant effects are found for men in terms of age and marital status, whereas among women, older age is associated with higher odds of receiving a State Pension, and the odds of receiving State Pension income among divorced/separated women are 1.39 times the odds among married women. In terms of socio-economic variables, men who rent their property privately and women who own their property with a mortgage are less likely to be receiving a State Pension than men and women owner-occupiers, respectively. The results also show the importance of employment: older men and women who have never worked are 41 and 46 per cent, respectively, less likely to be receiving a State Pension. Finally, health appeared to have an effect in the women's model, but not in the men's model. The odds of receiving a State Pension among older women reporting fair or poor health are 0.73 and 0.52 times, respectively, the odds among older women reporting good health, while older women reporting a limiting long-standing illness are 63 per cent more likely to be receiving State Pension income than those not reporting an illness. Finally, the disadvantage faced by migrants is also evident, especially among women: men and women who have spent less than ten years in the UK are 97 and 94 per cent, respectively, less likely to be receiving a State Pension, compared to their counterparts who were born in the UK, while even women who have spent more than 50 years in the UK are 46 per cent less likely than women born in the UK to be receiving a State Pension.

The results indicate that ethnicity also has a significant effect on older individuals’ chances of being in receipt of an occupational or private pension (Table 3). The model including both men and women shows that most non-White British groups are less likely to receive an occupational or private pension. Pakistani and Bangladeshi persons are the least likely to receive an occupational or private pension compared to the White British majority (0.10 and 0.19 times, respectively). Being over 75, female, with educational qualifications lower than a degree, renting from the social or private sector, reporting fair or poor general health, having never worked, having spent up to 39 years in the UK (compared to having been born in the UK) and, finally, having difficulty speaking day-to-day English, are associated with lower odds of receiving an occupational or private pension.

When the analysis was run separately by gender, the results were found to be broadly similar (e.g. in terms of the effect of education, housing tenure and whether one has ever worked), although certain differences between men and women are evident. Older age appears to be a negative factor for women, but not for men (odds ratio (OR) = 0.77 compared to the younger group). Being single never married, divorced/separated or widowed is associated with higher odds of receiving an occupational or private pension for women, but with lower odds for men. For men, all BME groups (except for the Caribbean and Other Ethnic groups where results are not significant) are less likely to be receiving an occupational or private pension than the White British, whereas for women such low odds are only statistically significant for Pakistani women (OR = 0.08). Finally, the negative impact of having difficulty with speaking English on a day-to-day basis is evident for older men, but not women.

The results of the different factors associated with the receipt of Pension Credit between the White British majority and ethnic minorities are illustrated in Table 4. From the model including both men and women combined, only Indian individuals are significantly more likely to be receiving Pension Credit than the White British majority (OR = 2.30 compared to White British). It is important to note that the differences in the eligibility rules for receiving Pension Credit, which was designed to ‘top-up’ the income of poorer pensioners, means that the groups who were more likely to receive State Pensions and occupational pensions (e.g. White British) were shown to be less likely to be in receipt of Pension Credit, while the opposite is the case for many individuals from BME communities. There are no age and gender effects; while single never married, divorced/separated or widowed older people are more likely to be receiving Pension Credit than those who are married (2.90, 3.84 and 2.42 times, respectively). In terms of socio-economic variables, having educational qualifications lower than a degree, and either owning one's home with a mortgage or renting (privately or socially) are associated with higher odds of receiving Pension Credit, while having never worked is associated with lower odds. Finally, the report of a limiting long-standing illness increases one's odds of receiving Pension Credit (the self-reported health status variable was not statistically significant in this model).

The separate models for men and women show that demographic, socio-economic and health determinants have a similar effect on men's and women's chances of receiving Pension Credit, although the strength of such effects varies between the two genders. Being single or divorced/separated is associated with higher odds of receiving Pension Credit, but such an effect is stronger for women than for men, while widowhood has a positive effect on such odds for women only. Slight differences in the strength of the coefficients also characterise the impact of socio-economic characteristics: for example, having no educational qualifications is associated with higher odds of receiving Pension Credit among women than among men, while renting privately is associated with higher odds among men than among women. Having never worked is associated with lower odds for men only, while the report of a limiting long-standing illness is associated with higher odds of receiving Pension Credit for both men and women. The effect of ethnicity appears to be positive on Other White, Indian and Other Ethnic men's chances of receiving Pension Credit, but not on women's chances.

The calculation of predicted probabilities for each of the three outcome variables (State Pension, occupational/private pension, Pension Credit) helps to distil the effect of ethnicity further. The calculation uses the strongest predictors associated with the outcome variable in each case (Figures 4–6). The three figures show the clear differences between ethnic groups in terms of receiving pension income from different sources. For example, in Figure 4, White British individuals, followed by Other White, Indian and Caribbean individuals, show the highest likelihood of receiving the State Pension, while Pakistani and African individuals, followed by the Bangladeshi group, show the lowest likelihood. The receipt of an occupational/private pension is most likely among the White British, Caribbean, Other White and Other Ethnic groups, closely followed by the Indian group, whereas the Pakistani and Bangladeshi groups are the least likely to be receiving such pensions. Finally, Figure 6 shows that Indian and Bangladeshi individuals, closely followed by the Other White, Pakistani, African and Other Ethnic groups, are the most likely to be receiving Pension Credit.

Figure 4.

Predicted probabilities and 95 per cent confidence intervals of receiving a State Pension by ethnicity.

Note: Reference category: aged below 75, married, owning home outright, educated to degree level, reporting positive self-rated health, no limiting long-standing illness, having worked and born in the United Kingdom. Source: Understanding Society (2009–11), authors’ calculations.

Figure 5.

Predicted probabilities and 95 per cent confidence intervals of receiving an occupational/private pension by ethnicity.

Note: Reference category: aged below 75, married, owning home outright, educated to degree level, reporting positive self-rated health, no limiting long-standing illness, having worked and born in the United Kingdom. Source: Understanding Society (2009 – 11), authors’ calculations.

Figure 6.

Predicted probabilities and 95 per cent confidence intervals of receiving Pension Credit by ethnicity.

Note: Reference category: aged below 75, married, owning home outright, educated to degree level, no limiting long-standing illness, having worked and born in the United Kingdom. Source: Understanding Society (2009 – 11), authors’ calculations.

The analysis in this paper includes a number of limitations which should be taken into account. Firstly, the data-set does not distinguish information on the precise amount received from different sources (e.g. amount received only from State Pension). As a result, it is not possible to further our understanding of differences in the disadvantage faced by particular ethnic groups. Secondly, the receipt of Pension Credit suffers from significant under-reporting, which is reflected in the survey data we have analysed. Although it was not possible to distinguish for different ethnic groups, when we compared administrative data from the Department for Work and Pensions on the caseload for Pension Credit receipt (including both Guarantee and Savings credit) with Census 2011 data on the older population, we found that such data points to 22.83 per cent of women aged 60 or over and 32.83 per cent of men aged 65 or over receiving Pension Credit, compared to 9.5 and 7.5 per cent in our analysis. However, when we compared the administrative data for the receipt of the Guarantee element of Pension Credit only, which is less likely to be under-reported than the Savings element, the survey data which we analysed was more consistent with the administrative data (7.45% of women; 5.49% of men). These figures confirm a significant degree of under-reporting in the receipt of Pension Credit in the survey data, which in all likelihood underestimates the percentage of older people receiving Pension Credit and therefore the percentage of individuals on low incomes in later life.