Jump to ContentJump to Main Navigation
Youth Labor in TransitionInequalities, Mobility, and Policies in Europe$

Jacqueline O'Reilly, Janine Leschke, Renate Ortlieb, Martin Seeleib-Kaiser, and Paola Villa

Print publication date: 2018

Print ISBN-13: 9780190864798

Published to Oxford Scholarship Online: January 2019

DOI: 10.1093/oso/9780190864798.001.0001

Show Summary Details
Page of

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (www.oxfordscholarship.com). (c) Copyright Oxford University Press, 2019. All Rights Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a monograph in OSO for personal use (for details see www.oxfordscholarship.com/page/privacy-policy).date: 24 May 2019

Do scarring effects vary by ethnicity and gender?

Do scarring effects vary by ethnicity and gender?

Chapter:
(p.560) 19 Do scarring effects vary by ethnicity and gender?
Source:
Youth Labor in Transition
Author(s):

Carolina V. Zuccotti

Jacqueline O’Reilly

Publisher:
Oxford University Press
DOI:10.1093/oso/9780190864798.003.0019

Abstract and Keywords

Being unemployed or inactive in youth leaves scars, but some people appear to be more successful than others in overcoming an initial disadvantaged situation. This chapter examines how early labor market experiences affect later employment and occupational opportunities for different groups. It compares the outcomes of White British men and women with those of second-generation ethnic minorities (Indian, Pakistani, Bangladeshi, and Caribbean) in England and Wales. It also discusses mechanisms affecting scarring and how they might vary across ethnic groups and genders. The analysis is based on the ONS Longitudinal Study. The examination follows individuals’ labor market experiences from 2001 (aged 16–29 years) to 2011 (aged 26–39 years). Having not been in employment, education, or training (NEET) has a less detrimental effect on later employment probabilities for Asian men than for White British men; the opposite is observed for Pakistani and Caribbean women compared to White British women.

Keywords:   ethnicity, gender, NEET, scarring effects, occupation, unemployment, youth

19.1. Introduction

There is a substantive literature showing that the poor labor market integration of young people can have long-term negative impacts on their adult lives—for example, by increasing the probability of subsequent periods of unemployment or by affecting their income (for the United Kingdom, see Gregg 2001; for the Netherlands, see Luijkx and Wolbers 2009; for Germany, see Schmillen and Möller 2012; Schmillen and Umkehrer 2013; for the United States, see Mroz and Savage 2006). We also know that migrants and their children perform differently in the labor market compared to majoritarian populations. In particular, those coming from developing countries are often disadvantaged in terms of access to jobs, as shown both in cross-national (Heath and Cheung 2007) and in country-specific studies (Carmichael and Woods 2000; Silberman and Fournier 2008; Heath and Li 2010; Kogan 2011; Zuccotti 2015a). However, research that focuses on dynamics into and out of employment, or on the impact of early labor market outcomes on later employment or occupational outcomes for different ethnic groups, is less common (some exceptions are Kalter and Kogan 2006; Demireva and Kesler 2011; Mooi-Reci and Ganzeboom 2015). In particular, surprisingly little is known about how early job insecurity affects different ethnic groups in the labor market over time.

In this chapter, we address this gap in the literature by examining the impact of the early labor market status of young individuals in the United (p.561) Kingdom (in 2001) on their employment probabilities and occupational status 10 years later (in 2011), focusing on how this varies across ethnic groups and by gender. In particular, we are interested in whether an early experience of being NEET (not in employment, education, or training) affects later labor market outcomes. Our analysis is based on the Office for National Statistics Longitudinal Study (ONS-LS), a data set linking census records for a 1% sample of the population of England and Wales across five successive censuses (1971, 1981, 1991, 2001, and 2011). We study individuals who are aged between 16 and 29 years in 2001 and follow them up in 2011, when they are between 26 and 39 years old. The focus is on second-generation minority groups born in the United Kingdom; we also include individuals who arrived in the United Kingdom at a young age.

Understanding how early labor market experiences affect later outcomes for different ethnic groups (and genders within them) is of crucial importance (see Berloffa et al., this volume), especially in countries where the number of ethnic minorities is considerable and increasing. On the one hand, this knowledge enables a better understanding of integration processes over time; on the other hand, it can contribute to the development of more targeted policies, given the dramatic rise in youth unemployment since the 2008 crisis (Bell and Blanchflower 2010; Eurofound 2014; O’Reilly et al. 2015). The United Kingdom represents a valuable opportunity for a case study for this purpose, given its long-standing ethnic minority population, which includes a large and diverse number of second-generation minorities. The groups studied here—Indian, Pakistani, Bangladeshi, and Caribbean (compared with White British)—are also very varied in terms of levels of educational and economic resources, cultural values and religion, and degrees of spatial segregation (Modood et al. 1997; Phillips 1998; Platt 2007; Longhi, Nicoletti, and Platt 2013; Catney and Sabater 2015; Crawford and Greaves 2015; Catney 2016). These differences allow us to explore a range of expectations as to why “scars” related to poor early labor market integration might differ across groups.

We find that the transmission of disadvantage occurs differently across ethnic groups and genders: Some groups/genders perform better (and others worse) in terms of overcoming an initial disadvantaged situation. In particular, Asian men appear to be in a better relative position compared to White British men—a finding that challenges preconceptions about ethnic minorities always performing poorly in the British labor market.

In the next section, we present previous studies on scarring effects and ethnic inequalities, identifying the main mechanisms and discussing why these might vary across ethnic groups and genders. After outlining the data and methods used, we perform the analyses separately for employment and occupational outcomes. Finally, we conclude and discuss our findings.

(p.562) 19.2. Literature review and theory

19.2.1. The long-term effects of youth nonemployment

Experiencing periods of unemployment or inactivity while young has been shown to have both short- and long-term negative effects for labor market outcomes. In the United Kingdom, several studies have addressed this issue (Kirchner Sala et al. 2015). Using the National Child Development Study (NCDS; a UK data set following a cohort born in 1958), Gregg (2001) examines the extent to which nonemployment1 (i.e., unemployment or another inactive situation, excluding students) experienced between the ages of 16 and 23 years (measured when individuals were 23 years old in 1981) has an effect on later work experiences (when individuals are aged between 28 and 33 years). He shows that conditional on background characteristics such as education, family socioeconomic status, and neighborhood unemployment, men who experience an extra 3 months of being nonemployed before age 23 years face an extra 1.3 months out of employment between the ages of 28 and 33 years, whereas for women the effect is approximately half as strong. Kalwij (2004), who follows individuals who turned 18 years between 1982 and 1998 and were registered as unemployed at least once during this period, presents evidence pointing in the same direction. He demonstrates that the longer the previous spell of unemployment, the lower the probability of finding a job later. Specifically, 2 years in unemployment decreases the probability of becoming employed by 31%. Similarly, analyzing the British Household Panel Survey (BHPS), Crawford et al. (2010) show that individuals who were NEET at 18 or 19 years old have an almost 20% greater chance of being unemployed 10 years later, compared to individuals who were either studying or working at the same age. More recently, Dorsett and Lucchino (2014), using the BHPS to study transitions up to age 24 years, show that the longer one remains in employment, the lower the chance of becoming unemployed, whereas the longer an individual remains unemployed or inactive, the less likely he or she is to find employment.

Some authors have examined scarring effects in terms of wage outcomes. For example, using the NCDS, Gregg and Tominey (2005) find that given equal characteristics (including education), 13 months of unemployment between ages 16 and 23 years (vs. being always employed) reduces income by 20% at ages 23 and 33 years and by 13% at age 43 years. They also find that even when individuals do not experience unemployment after the age of 23 years, a wage scar of between 9% and 11% remains. Crawford et al. (2010) demonstrate that individuals who were NEET at ages 18 or 19 years had significantly lower wages when aged 28 or 29 years compared to individuals who were either working or studying at the same age; this held even when they shared similar characteristics, such as comparable education and parental background.

Scarring effects may vary in their intensity depending on the highest level of education achieved or the qualifications of individuals. Kalwij (2004), for (p.563) example, shows that highly skilled men have greater chances of exiting and weaker chances of re-entering unemployment compared to low-skilled men. Burgess et al. (2003), analyzing data from the UK Labour Force Survey (UK-LFS), show that although the effect of early career unemployment is to reduce later employment chances for those with lower or no educational qualifications, the opposite occurs among those with higher educational qualifications. Schmelzer (2011), examining occupational outcomes, arrives at a similar finding. He shows that individuals with higher levels of education do not suffer as a result of early career unemployment; in fact, their stronger resources allow them to stay longer in this situation while waiting for better job offers (see Filandri, Nazio, and O’Reilly, this volume). Individuals with lower education levels, by contrast, are penalized in terms of their future occupations—an outcome that is generally attributed to gaps in their human capital accumulation. It is also possible that these periods outside of employment or education send negative signals to employers.

In summary, a wealth of research on early labor market experiences reveals how crucial these are for later life outcomes. These experiences vary by educational attainment, with the lowest qualified being the most negatively affected later in life. Clearly, such findings are very significant, given the heightened rates of youth unemployment being seen across Europe—both preceding and exacerbated by the 2008 financial crisis (O’Reilly et al. 2015).

19.2.2. Ethnicity and labor market outcomes in the United Kingdom

Western European countries have a long history of immigration, often connected to processes of economic reconstruction. In the United Kingdom, there has been a long-term pattern of Irish migration; however, immigration intensified in the postwar period with the arrival of the first waves of Caribbean migrants in the late 1940s, who were subsequently followed by Indians and Pakistanis and—later—by Chinese, Bangladeshis, and Africans. Today, more than 10% of the population in the United Kingdom self-defines as non-White, and this includes both first-generation migrants and their second-generation children.

In general, studies are in agreement that although problems such as unemployment (Heath and Cheung 2007) and low income (Longhi et al. 2013) are still faced by several ethnic groups in Western European countries, especially the visible non-White groups, the children of immigrants are in a better situation compared to their parents in terms both of education (Brinbaum and Cebolla-Boado 2007; van de Werfhorst and van Tubergen 2007) and of labor market outcomes (Heath and Cheung 2007; Alba and Foner 2015). In the United Kingdom, efforts have been made to develop policies and laws to help these groups integrate (Cheung and Heath 2007), and these initiatives have probably encouraged the processes of social mobility we observe today (Platt 2007). For example, whereas first generations are more often concentrated in low-qualified (p.564) jobs (Zuccotti 2015b), their children have similar (Pakistanis, Bangladeshis, and Caribbeans) or even higher (Indian) rates of participation in professional and managerial occupations compared to White British. Regarding access to jobs, unemployment has historically been one of the main problems concerning ethnic minorities’ labor market integration. However, trends show an improvement in employment levels for all groups in the adult population. For example, the unemployment level for Pakistanis and Bangladeshis declined from 25% in 1991 to approximately 10% in 2011; Indians had practically the same unemployment level as the White British (approximately 6%) in 2011; and the unemployment level of Caribbean men, although still relatively high (16%), has improved since 1991 (Nazroo and Kapadia 2013). Of course, there are also gender differences in this respect, with Pakistani and Bangladeshi women still being characterized by high unemployment and inactivity levels (House of Commons Women and Equalities Committee 2016). Moreover, although studies have shown that some of the differences in employment levels across groups are connected to education (Cheung and Heath 2007), social origins, and neighborhood deprivation (Zuccotti 2015b), discrimination continues to be a key problem faced by ethnic minorities (Heath and Cheung 2006).

19.2.3. A longitudinal view on ethnicity and labor market outcomes

The studies on ethnic inequalities presented so far are either restricted to certain time points or, if applied to several years, do not really discuss changes within individuals or individual-level changes in labor market performance over time. A recent work by Demireva and Kesler (2011) sheds some light on this matter. Using data from the UK Quarterly Labour Force Survey (1992–2008), Demireva and Kesler study transitions into and out of employment for different migrant and native groups. In accordance with previous studies, they corroborate the idea that higher education plays a positive role in these transitions. In terms of ethnicity, they show that men born in the New Commonwealth (which includes the Caribbean, India, Pakistan, and Bangladesh) are more likely than the White British to remain in or to move into unemployment/inactivity between two consecutive quarters of a year. Among second-generation immigrants, the authors note that men are more likely to remain in unemployment compared to equivalent White British; women are also more likely to move from unemployment to inactivity compared to their White British counterparts. However, group differences within second generations are not further developed—a limitation of this work that we address in the current study.

Related studies have been carried out in other European countries (see Reyneri and Fullin 2011 and other articles in the same journal issue) and in comparison with North America (Alba and Foner 2015), with results varying according to institutional factors and labor market characteristics. An analysis of 10 (pooled) (p.565) Western European countries found that non-EU15 immigrants generally have higher probabilities of remaining in unemployment between two years (Reyneri and Fullin 2011). More recently, a study in the Netherlands (Mooi-Reci and Ganzeboom 2015)—a country that, like the United Kingdom, has a relatively long history of immigration—has examined the concept of scars and how these might vary according to the migrant status of individuals. Using the Dutch Labor Supply Panel (covering data between 1980 and 2000), the authors examine income as an outcome and explore how previous unemployment experiences affect re-employment income for native Dutch and foreign-born individuals. They find that individuals born outside of the Netherlands receive lower re-employment income compared to Dutch counterparts with similar unemployment experiences.

Often, ethnic minorities and foreign-born individuals are more exposed to unemployment/inactivity compared to their majoritarian host-country counterparts. Most important, these events seem to have particularly pronounced scarring effects in later life for these groups, including weaker employment chances and lower re-employment income. This chapter focuses on how the early labor market experiences of young people in different second-generation minority groups affect their later outcomes. Although, according to the literature discussed previously, more severe scarring might be expected among second-generation ethnic minorities, the recent improvements in terms of employment and occupation might actually point in the opposite direction.

19.2.4. Highlighting mechanisms: Human capital decay versus stigma

When searching for explanations as to why an early experience of inactivity or unemployment might affect later labor market outcomes, the literature has highlighted two in particular: human capital decay and stigma (Omori 1997; Schmelzer 2011). These explanations focus mainly on employers’ recruitment practices. Human capital decay suggests that in periods of nonemployment, individuals lose vital work experience, which in turn might reduce their future employability and earnings. Stigma-related explanations, on the other hand, suggest that employers judge future employees’ capabilities based on their unobserved trajectory of employment and nonemployment. In other words, they infer workers’ qualities based on their past employment status. In this context, previous unemployment spells have a negative stigma (e.g., when one assumes that individuals are unemployed because they are lazy), which might then affect later employment probabilities and income prospects. However, as suggested by Mooi-Reci and Ganzeboom (2015), stigma might also be related to how employers infer characteristics of individuals based on their ethnic origins. For example, if employers believe that an ethnic minority group has certain negative characteristics in terms of employability—such as an educational degree obtained abroad, language deficiencies, or their concentration in deprived (p.566) neighborhoods—a period of unemployment or inactivity might exacerbate these negative preconceptions and stereotypes, affecting future employment probabilities, type of occupation, or income. These authors’ empirical analysis regarding the Netherlands presents evidence in this direction.

To what extent can we see stigma mechanisms connected to ethnicity occurring in the United Kingdom? First, there is evidence of discrimination in the labor market (Heath and Cheung 2006; Wood et al. 2009). In particular, experimental studies have demonstrated that employers usually prefer White British compared to other ethnic groups, especially Asians and Blacks. Although the reasons behind this preference are still to be explored, we could argue that a period of unemployment or inactivity might affect some ethnic minority groups in particular negative ways and independently of whether they were born in the United Kingdom or abroad. For example, Pakistani and Bangladeshi populations have historically worked in relatively lower qualified jobs and have been spatially concentrated in the most deprived areas (Phillips 1998; Robinson and Valeny 2005). This negative signal in terms of where employers view these populations in the social structure (which could affect their views on these groups’ productivity, for instance) might contribute to how they perceive their nonemployment experiences and thus help create a particularly profound scar for them.

However, for other groups, we might observe other processes taking place. We argued that scars are lighter (or not present) among highly educated individuals, partly because employers do not view a period of unemployment for highly educated individuals particularly negatively (Schmelzer 2011), assuming them to be searching for an appropriately qualified job. In terms of ethnic differences and how employers perceive groups, this might benefit Indians, in particular. This group has very high rates of university achievement, which could be observed as a positive signal for employers in terms of group characteristics. A period of unemployment or inactivity might therefore be more “legitimate” for Indians than for other groups, which would be observed in a lower scarring effect of unemployment/inactivity on this group.

Mooi-Reci and Ganzeboom (2015) also suggest that employers’ perceptions might vary by gender. They argue that immigrant women from poorer countries are more likely to be perceived as more nurturing and obedient, which might weaken the stigma of joblessness. In the United Kingdom, this might apply to Pakistani and Bangladeshi women, who are also embedded in cultural contexts in which women are expected to stay at home (Peach 2005).

The group context or group characteristics, and how employers observe these, are therefore an argument for expecting variation in scars across ethnic groups. In line with this reasoning, Omori (1997) found that individuals who experienced unemployment in periods when unemployment was high were less penalized in terms of future employability compared to individuals who had been unemployed when unemployment levels were low. The context perceived by employers or, in our case, the perceived group context may therefore matter.

(p.567) Until now, we have discussed employers; however, groups’ perspectives, culture, and networks might also affect outcomes. For example, although it is true that Bangladeshi men are usually concentrated in poor areas and have low social backgrounds, there is evidence that second-generation Bangladeshis are doing quite well in the labor market: Not only do they not seem to experience ethnic penalties in employment (Zuccotti 2015b) but also they overperform compared to the White British in terms of the occupations they obtain. This finding might be connected to specific characteristics of Bangladeshis that make them more resilient to adverse situations. Hence, we might argue that they manage to better overcome a situation of early unemployment or inactivity. Similarly, with regard to the arguments concerning gender, given the strong role models in some Asian ethnic groups and the family and community pressures on women to remain out of the labor market (Dale et al. 2002a, 2002b; Kabeer 2002), we could argue that it might actually be particularly difficult for women to become employed if they have had early experiences of unemployment or inactivity. In summary, these arguments suggest that the role of (increased or decreased) stigma might not be the only explanation behind differences in the effect of early labor market statuses across groups.

19.3. Data and methods

19.3.1. The Office for National Statistics Longitudinal Study

Our analysis is based on the ONS-LS,2 a unique data set collected by the Office for National Statistics in the United Kingdom that links census information for a 1% sample of the population of England and Wales, following individuals in 1971, 1981, 1991, 2001, and 2011. The original sample was selected from the 1971 Census, incorporating data on individuals born on one of four selected dates. The sample was updated at each successive census by taking individuals with the same four dates of birth in each year and linking them to the existing data (Hattersley and Creeser 1995). Life-event information has been added to the ONS-LS since the 1971 Census. New members enter the study through birth and immigration, and existing members leave through death and emigration. Some individuals might also exit the study (e.g., someone who goes to live abroad for a period) and then re-enter at a later census point; however, individuals are never “removed” from the data set, nor do they actively “leave” it.

Slightly more than 500,000 individuals can be found at each census point; however, information for people in the 1% sample who participated in more than one census point is more limited. For example, there is information on approximately 400,000 people at two census points, on average, whereas information is available on approximately 200,000 people for all five census points. In total, approximately 1,000,000 records are available for the entire period (1971–2011).

(p.568) One of the most interesting aspects of this data set—in addition to its large sample size—is that both household and aggregated census data for small geographical areas can be attached to each individual and for each census point. This provides a reasonable idea of the “family contexts” and “neighborhoods” in which individuals live at different moments of their lives.

19.3.2. Sample

Our focus is on young individuals aged between 16 and 29 years in 2001, whom we follow through 2011, when they are between 26 and 39 years old. Different definitions have been given as to what it means to be young or to belong to the “youth population.” The Office for National Statistics in the United Kingdom, for example, usually considers an age range of 16–24 years. We decided to use a slightly wider age range for two main reasons. First, we wanted to capture the increasingly lengthy and blurred trajectories into adulthood (Aassve, Iacovou, and Mencarini 2006); second, we could thus cover a larger sample of ethnic minorities. We performed robustness checks excluding individuals aged 25–29 years and found that the results remained robust to the findings shown here.

We constructed our sample in a way that permits more than one measurement per individual. Where individuals had more than one measurement for “family context” and “origin neighborhood” (obtained when they were between 0 and 15 years old, in 1981–1991), we counted these as two units of analysis. For example, we counted an individual twice if he or she was 21 years old in 2001 and had household and neighborhood information in both 1991 (when he or she was 11 years old) and 1981 (when he or she was 1 year old). This structure follows a model used previously by Platt (2007) and is common in works using panel-like data. In order to account for double measurement, we control for “origin year” (1981/1991) and we use clustered standard errors in the regression models. We have also estimated a model in which one origin year per individual is randomly chosen and the results remain the same. The total sample consists of 77,180 cases, out of which 73% are “unique” individuals.

19.3.3. Variables and methods

We study two outcome variables in 2011: employment status and occupational status. These are examined in relation to labor market status in 2001. We observe individuals with different statuses in 2001—NEET (i.e., “unemployed and inactive,” including individuals doing housework, with long-term illness or disability, and other inactive), employed, and students—and ascertain their employment and occupational trajectories in 2011. The focus is on the potential negative effect that being out of employment and out of education might have on later labor market outcomes and how this varies by ethnicity and gender (for a discussion on the concept of NEET, see Mascherini, this volume). Employment in 2011 is a dummy variable that determines whether the person was employed or (p.569) not in 2011 (the reference category is unemployed/inactive, excluding students). Occupational status, on the other hand, is measured using the National Statistics Socio-economic Classification (NS-SEC) (Erikson and Goldthorpe 1992). The NS-SEC includes seven categories ranging from higher managerial/professional occupations to routine occupations. We study the probability of having a Class 1 or Class 2 occupation (vs. any other): Class 1 consists of higher managerial, administrative, and professional occupations, whereas Class 2 consists of lower managerial, administrative, and professional occupations. The occupations within these two classes are often regulated by so-called service relations, where “the employee renders service to the employer in return for compensation, which can be both immediate rewards (for example, salary) and long-term or prospective benefits (for example, assurances of security and career opportunities)” (Office for National Statistics 2010, 3). Note that occupational status refers to the current or most recent job.

We examine these trajectories across five ethnic groups: White British, Indian, Pakistani, Bangladeshi, and Caribbean. In this study, White British are those who identify themselves as White English/Welsh/Scottish/Northern Irish/British3 and have both parents (or one parent, in the case of individuals raised in single-parent households) born in the United Kingdom. Ethnic minorities, on the other hand, are those who identify themselves as belonging to one of the main ethnic groups and have one (single-parent households) or two parents born abroad.4 The parental country of birth is measured when individuals were between 0 and 15 years old in 1981–1991.

In studies of scarring effects, efforts are usually made to measure the actual scar in the best possible way. Often, we do not know all the variables that might affect an outcome. If such variables are present but we do not control for them, then we might be over(under)estimating the size of the scar. For example, if individuals of a certain group have characteristics that make them more likely to be unemployed, this will affect both the 2001 and the 2011 outcomes and will make the relationship between the two unemployment variables at the respective time points stronger than it is in reality. In order to reduce unobserved heterogeneity, we control for a wide range of key predictors of labor market status, including family arrangements and education in 2011 and the socioeconomic characteristics of the households in which individuals lived when they were between 0 and 15 years old. Household-level variables (found in the 1981 and 1991 census files) include number of cars, housing tenure, level of overcrowding in the home, and parental occupation (taking the highest status between the father and the mother). In addition, we also control for current-neighborhood deprivation and origin-neighborhood deprivation (when individuals were between 0 and 15 years old), both measured with the Carstairs Index (Norman, Boyle, and Rees 2005; Norman and Boyle 2014).5 This measure is a summary of four dimensions: percentage male (p.570) unemployment, percentage overcrowded households, percentage no car/van ownership, and percentage low social class.

The inclusion of variables that denote neighborhood characteristics—current and, most important, past—has been a commonly used tool by some authors (e.g., Gregg 2001) to control for the self-selection of individuals into their initial condition (in our case, labor market status in 2001) and hence reduce the impact of unobserved heterogeneity. In terms of our study, neighborhood deprivation when individuals are young is likely to affect labor market status in 2001 but less so labor market status in 2011, except through neighborhood deprivation in 2011 (which we control for). Most important, this variable has the advantage that young individuals probably did not choose the neighborhood where they lived when they were young (rather, their parents did).

Our model has, nevertheless, some limitations. First, we are not able to use (as Gregg (2001) does) more detailed neighborhood unemployment levels or types of jobs available in the area, which would be a better indicator of labor market conditions and availability of jobs. The ONS has restrictions regarding the use of neighborhood variables, and neighborhood deprivation is easy to access and is a commonly used variable among ONS-LS users. Note, however, that because we include students in our initial labor market statuses, neighborhood deprivation is probably a better variable than, for example, neighborhood unemployment alone, given that it includes indicators such as social class and socioeconomic resources of households, which might impact on decisions regarding school attendance. Second, we do not use an instrumental variable approach, as Gregg does: In other words, origin-neighborhood characteristics is not an instrument in our model (as it is in Gregg’s study) but, rather, a control variable. The program we use to analyze our data (Stata 14) has limitations in terms of the commands for instrumental variables, and some tests led us to prefer a classic regression model.6 Finally, a third limitation (that would also be present even with an instrumental variable approach) is that there might be unmeasured parental or group characteristics (e.g., parental aspirations or group preferences for certain areas) that affect individuals’ outcomes as well as their selection of neighborhoods. If present, these unmeasured characteristics will weaken the origin-neighborhood deprivation’s potential ability to randomize the allocation of individuals into areas and, hence, into initial statuses. In summary, we are aware that we cannot fully randomize the selection of individuals into their initial statuses in 2001, which means that we cannot be certain that the relationship between initial status and employment in 2011 is casual. The observed scar might therefore include some unmeasured characteristics of individuals, their parents, or the ethnic groups to which they belong.

Our multivariate analyses are based on average marginal effects derived from logistic regressions. In addition to the previously mentioned variables, other controls include age in 2001, country of birth, and number of census points in which the individual participated.

(p.571) 19.4. Analysis

19.4.1. Descriptive statistics

Table 19.1 shows the percentage of individuals employed in 2011 and the percentage of individuals who declare a high occupational status (either presently or in the most recent job), distinguished by their labor market status in 2001, ethnic group, and gender.

Table 19.1 Employed individuals and individuals with (current or most recent) professional/managerial status in 2011, by labor market status in 2001, ethnic group, and gender (%)

Employed

Professional/managerial status

NEET

S

E

Total

NEET

S

E

Total

Men

White British

58.9

92.0

93.6

89.8

22.8

59.1

42.8

44.4

Indian

77.6

91.8

91.0

90.1

37.7

69.8

55.2

60.2

Pakistani

64.8

86.0

91.2

84.1

21.6

45.8

32.9

37.0

Bangladeshi

64.5

100.0

87.5

87.7

25.0

61.0

35.9

41.8

Caribbean

58.3

76.2

84.9

78.3

40.0

47.2

47.6

46.2

Women

White British

50.2

89.2

85.6

80.0

19.3

62.7

44.5

44.5

Indian

50.9

87.1

82.4

80.3

28.7

74.6

55.5

60.7

Pakistani

29.9

61.6

67.5

52.2

18.4

56.4

38.2

38.4

Bangladeshi

33.7

64.2

68.1

53.5

18.0

53.6

34.8

34.9

Caribbean

41.0

74.6

84.8

74.4

30.3

60.0

51.2

50.2

Totals: Men

White British

3,471

6,878

24,791

35,140

2,768

6,674

24,354

33,796

Indian

85

413

434

932

77

397

422

896

Pakistani

88

222

181

491

74

212

173

459

Bangladeshi

31

60

80

171

40

59

78

177

Caribbean

24

42

86

152

25

36

82

143

Totals: Women

White British

6,875

8,158

23,315

38,348

5,704

7,970

22,988

36,662

Indian

110

357

403

870

87

343

389

819

Pakistani

224

198

203

625

152

172

191

515

Bangladeshi

89

67

72

228

61

56

69

186

Caribbean

39

59

125

223

33

55

121

209

Notes: Labor market status in 2001: NEET, unemployed or inactive; S, student; E, employed. Population: Individuals between 16 and 29 years old in 2001.

Source: Authors’ calculations based on ONS-LS.

For most groups, and as expected, having been employed or in education in 2001 leads to a greater likelihood of being employed in 2011 and to a greater likelihood of having a higher occupational status—compared to individuals who were unemployed or inactive (i.e., NEET) in 2001. In particular, those who were students in 2001 have high proportions in both employment and professional/managerial occupations in 2011, probably attributable to having a university degree. However, the extent to which education and employment in 2001 act as “protectors” in the labor market or, conversely, the extent to which unemployment and inactivity make people more “vulnerable” or generate “scars” varies greatly across ethnic groups and genders.

Having been NEET in 2001 (compared to having been employed) is not particularly detrimental for the labor market prospects of ethnic minorities compared to the White British. Only Caribbean women seem to follow this pattern as regards their employment probabilities (note that among those who were employed in 2001, White British and Caribbean women have similar employment probabilities in 2011, whereas this is 9% lower for Caribbeans among those who were NEET). In contrast, it is White British men who seem to experience deeper scars regarding employment, especially compared to Asian groups (Indian, Pakistani, and Bangladeshi). We observe that among those who were employed in 2001, employment probabilities in 2011 are similar across all groups, but having been NEET has a more detrimental effect on the likelihood of White British men being in employment in 2011. Approximately 59% of White British men who were NEET in 2001 are employed in 2011; for Indians, in particular, but also for Pakistani and Bangladeshi men, the percentage of employed is higher.

Table 19.1 also shows that although, in general, ethnic minority groups do not suffer very strongly from previous periods of unemployment or inactivity, sometimes having been employed in 2001 is not as protective for them as it is for the White British. For example, Caribbean men are similar to White British in terms of their employment probabilities among those who were NEET in 2001; however, they do not benefit from having been employed in 2001 to the same degree as White British men (they have approximately 10 percentage points less probability of being employed in 2011). A similar finding is observed among Pakistani and Bangladeshi men and women when studying occupational status. We observe that although differences with respect to White British are relatively small among those who were NEET in 2001, of those who were employed in (p.572) 2001, White British have higher probabilities of attaining a professional/managerial position by 2011.

Finally, other well-known patterns that emerge from Table 19.1 are the overperformance of Indians in terms of access to high-status occupations and the low employment probabilities of Pakistani and Bangladeshi women (see House of Commons Women and Equalities Committee 2016). In this respect, (p.573) note that although there is no clear evidence of a stronger employment scarring effect for these women (the difference in employment probabilities with respect to White British is approximately 18–20 percentage points among both those who were employed and those who were NEET in 2001), we do observe a particularly strong scar connected to having been a student in 2001: The ethnic gap in terms of employment chances grows to 30 percentage points for this category.

These results, however, need to be studied after we have controlled for a series of factors that might also affect the outcomes. In fact, there is great variation across ethnic groups in terms of educational achievements, socioeconomic backgrounds, and family arrangements, as shown in Table 19.2. (p.574)

Table 19.2 Social origins and individual-level characteristics, by ethnic group

British

Indian

Pakistani

Bangladeshi

Caribbean

Total

Social origins

Parental social class

No earners/no code

5.6

4.9

18.8

29.3

14.5

5.9

Manual (V + VI + VII)

33.4

47.0

56.7

51.7

33.7

34.1

Routine nonmanual (III)

15.1

11.3

3.6

3.2

26.8

14.8

Petite bourgeoisie (IV)

11.8

15.8

12.5

11.3

3.3

11.9

Professional/managerial (I + II)

34.1

20.9

8.5

4.4

21.7

33.2

Cars

No cars

18.7

22.5

39.5

69.0

46.7

19.5

1 car

53.4

57.0

51.8

28.3

45.9

53.3

2 cars

27.9

20.5

8.7

2.7

7.4

27.2

Tenure

Owner

70.4

86.9

86.8

41.9

46.9

70.8

Social rent

22.8

7.7

7.4

42.9

46.4

22.5

Private rent

6.7

5.4

5.8

15.3

6.6

6.7

Persons per room

>1.5 persons

0.7

8.8

22.5

36.2

6.1

1.4

1.5 persons

0.5

3.7

6.5

8.9

5.4

0.8

>1 and <1.5 persons

6.1

20.3

31.2

28.8

13.8

6.9

1 person

16.3

23.8

18.7

12.3

25.5

16.6

≥0.75 and <1 person

29.9

22.0

12.9

9.4

21.7

29.3

<0.75 person

46.5

21.4

8.2

4.4

27.6

45.0

Carstairs quintiles

Q1 (less deprivation)

22.0

7.2

2.1

2.5

2.8

21.2

Q2

21.7

7.7

3.7

3.4

5.1

20.9

Q3

20.8

11.8

5.8

5.9

7.1

20.2

Q4

19.7

21.5

16.8

8.9

22.4

19.6

Q5 (more deprivation)

15.8

51.7

71.6

79.3

62.5

18.0

Individual characteristics

Age (2001)

Mean age

22.6

22.1

21.9

21.8

23.0

22.6

Education (2011)

None and other

10.7

4.9

12.6

12.3

4.8

10.6

Level 1

14.2

9.2

18.5

22.2

14.0

14.1

Level 2

18.8

11.7

16.3

17.0

17.6

18.5

Level 3

18.2

11.5

12.6

12.3

17.1

17.9

Level 4+

38.2

62.7

40.0

36.2

46.4

38.8

Family type (2011)

Single, no children

25.8

37.0

22.1

20.0

45.9

26.1

Couple, no children

22.3

18.3

8.2

8.4

12.0

21.9

Single with children

8.6

8.2

14.6

18.7

22.2

8.8

Couple with children

43.2

36.4

55.1

53.0

19.9

43.2

Country of birth

UK-born

99.0

93.4

81.1

45.6

97.7

98.3

N

74,796

1,830

1,147

406

392

78,571

Note: Population: Individuals between 16 and 29 years old in 2001.

Source: Authors’ calculations based on ONS-LS.

There are two clear and interesting findings from Table 19.2. On the one hand, ethnic minorities tend, in general, to have lower or more deprived social origins. For example, they are more likely to have been raised in areas with high neighborhood deprivation and to have parents with lower occupational status. This is particularly evident for the Pakistani and Bangladeshi populations. These factors might impact negatively not only on their labor market outcomes but also on the transitions they make in the labor market. On the other hand, ethnic minorities also tend to be more educated, revealing their upward educational mobility (given their low parental social backgrounds). For instance, the high percentage of Indians who reach university level (level 4+) is striking. Bearing in mind the positive role that education plays in the labor market, including making good-quality transitions, a higher education level among ethnic minorities might actually help counterbalance their poorer social origins. Recent research (Zuccotti 2015a; Zuccotti, Ganzeboom, and Guveli 2017) shows the (p.575) importance of considering both education and social origins (see also Berloffa, Matteazzi, and Villa, this volume) in the estimation of ethnic inequalities in the labor market. Variation is also observed in terms of family type, with Pakistani and Bangladeshi populations having particularly large shares of households composed of a couple with children. This might be an explanation as to why we see such low employment levels among women from these groups.

The next section examines all these factors together using multivariate logistic regression models. In addition to the socioeconomic, educational, and family variables observed in Table 19.2, we also control for the year in which the origin variables were measured (1981 or 1991) and for the number of census points in which the individuals participated. Finally, note that although the majority of ethnic minorities were born in the United Kingdom, we also consider the country of birth in our analyses (with a dummy as to whether they were born in the United Kingdom or not). Bangladeshis, in particular, have the highest proportion of foreign-born young individuals (see Table 19.2)—a factor that might have a negative impact on labor market transitions.

19.4.2. Multivariate models

This section examines whether the trends found in Table 19.1 still hold after we control for individual and social-origin characteristics, including current and past residential neighborhood deprivation levels. First, we show the average effect of labor market status in 2001 and of ethnic group on labor market outcomes in 2011 (employment and occupational status) before (Model a) and after (Model b) controlling for key individual, social-origin, and neighborhood variables (see Tables 19.3 and 19.4). The results are presented separately for men and women; the coefficients represent average marginal effects derived from logistic regressions (models with all controls are shown in Table A19.1 in the Appendix).

Next, we introduce interactions between labor market status in 2001 and ethnicity in order to study whether scarring varies in relation to an individual’s ethnic group. Models with interactions are used to answer the main question in this chapter: What is the effect of having been unemployed or inactive (NEET), compared to having been employed or in education, in 2001 on the probability of being employed/having a high occupational status in 2011—for different ethnic minority groups and for White British? In particular, to what extent is being out of education and out of the labor market particularly detrimental (or not) for some ethnic groups? Because we work with logistic regression models, we calculated predicted values for the groups from the interaction models (keeping all control variables at their mean; see Table A19.2 in the Appendix) and created graphs.7 Predicted values and graphs serve not only to observe the magnitude of the effects but also to explore at which levels of the dependent variable they occur for an individual with “average” characteristics. Assuming that the variable (p.576) “labor market status in 2001” has a certain “order” in the categories, we explore “slopes” for different ethnic groups: how steep they are and whether they touch or not.

19.4.2.1. Employment scarring

Overall, our findings indicate that having been NEET in 2001, compared to having been employed, reduces by more than 30 percentage points the probability of being employed in 2011—for both men and women (Model a). After we control for social-origin and individual characteristics, as well as for current and past levels of deprivation of the neighborhood of residence (Model b), the effect (p.577) declines, but it is still quite substantive (approximately 17%). Poor labor market integration at a young age creates scarring for both men and women.

Table 19.3 shows that although the effect of having been in education in 2001 on the probability of being employed in 2011 is similarly positive to the effect of having been employed in 2001 (for women it is actually more positive), the education effect becomes negative after we control for key variables. In other words, after we control for the fact that individuals with more socioeconomic resources are usually more likely to continue in higher/university education, and for the fact that higher education levels lead to better employment chances, a situation (p.578) of employment (vs. any other) in 2001 seems to have more positive long-term effects than studying. Although this does not mean that individuals should invest less in education, it does suggest that early experiences of employment—perhaps simultaneously with an educational activity—can have positive long-term effects in terms of accessing work in the UK labor market. As previously argued, this might be connected to the extra skills acquired due to longer lasting work experience but also to sending a positive “signal” to employers.

Table 19.3 Probability of being employed in 2011, by labor market status in 2001 and ethnic group; AME (clustered standard errors)

Men

Women

Model a

Model b

Model a

Model b

Labor market status in 2001 (ref. Employed)

NEET (unemployed or inactive)

–0.338***

–0.175***

–0.357***

–0.173***

(0.0103)

(0.0078)

(0.0079)

(0.0073)

Student

–0.005

–0.041***

0.056***

–0.012*

(0.0045)

(0.0063)

(0.0052)

(0.0071)

Ethnic group (ref. White British)

Indian

0.003

0.004

–0.028

–0.061***

(0.0120)

(0.0109)

(0.0176)

(0.0185)

Pakistani

–0.020

0.007

–0.211***

–0.160***

(0.0164)

(0.0130)

(0.0250)

(0.0223)

Bangladeshi

0.002

0.042***

–0.180***

–0.100***

(0.0243)

(0.0159)

(0.0393)

(0.0325)

Caribbean

–0.067*

–0.023

–0.064*

–0.073**

(0.0347)

(0.0246)

(0.0343)

(0.0324)

N

36,886

36,886

40,294

40,294

Basic controls

X

X

X

X

Individual, social origin, and neighborhood controls

X

X

Notes: Basic controls: Age, country of birth, origin year, and number of census points. Individual, social origin, and neighborhood controls: Education, family type, parental social class, number of cars, tenure, level of overcrowding, and neighborhood deprivation (past and current). Population: Individuals between 16 and 29 years old in 2001.

(*) p < .10.

(**) p < .05.

(***) p < .01.

Source: Authors’ calculations based on ONS-LS.

In terms of average group differences, we observe that men from ethnic minority backgrounds (especially Bangladeshis) have similar or even higher probabilities of being in work in 2011 compared to White British. This finding is similar to previous results obtained for a slightly older age group (aged 20–45 years; Zuccotti 2015b). For women, on the contrary, all ethnic minority groups have lower employment probabilities compared to White British women. Differences that emerge from our analysis range from 6 percentage points lower for Indian women to 16 percentage points lower for Pakistani women.

We identified several statistically significant interactions. For men, having been NEET in 2001 (vs. having been employed) is not as detrimental for Indian and Bangladeshi men as it is for White British men. This denotes lower scarring effects for the ethnic minorities. A similar relative advantage is observed for Indian and Pakistani men when comparing NEET with students in 2001. Among women, the results suggest that Pakistani and Caribbean women have deeper scars connected to having been NEET than is the case for White British women. These findings are better understood by looking at Figures 19.1 and 19.2, which show the predicted values of employment in 2011 for Indian, Pakistani, and Bangladeshi men (vs. White British men) and for Pakistani and Caribbean women (vs. White British women) for each labor market status in 2001 (keeping all control variables at their mean).

Do scarring effects vary by ethnicity and gender?

Figure 19.1 Predicted values of male employment in 2011 (90% confidence interval).

Source: Authors’ calculations based on ONS-LS.

Do scarring effects vary by ethnicity and gender?

Figure 19.2 Predicted values of female employment in 2011 (90% confidence interval).

Source: Authors’ calculations based on ONS-LS.

In visual terms, the weaker detrimental effect of having been NEET, versus having been employed or a student, for Indian, Pakistani, and Bangladeshi men is expressed in the flatter slopes for these three ethnic groups. In particular, for Indians and Bangladeshis, there is a much higher probability of employment among those who were NEET in 2001: This difference is approximately 9 percentage points for Indians and approximately 12 percentage points for Bangladeshis. Note that Bangladeshis are also greatly advantaged among those who were students in 2001. Conversely, these groups have more similar employment probabilities among those who were employed in 2001 (only Indians seem to present a negative and relatively small gap with respect to White British).

The graph for women (see Figure 19.2), in contrast, shows a steeper slope for Pakistanis and Caribbeans than for White British, denoting a deeper scar for the ethnic minority. Looking at the predicted values, we observe, for example, that (p.579) the employment probabilities among those who were employed in 2001 are approximately 74% for Pakistanis and 88% for White British (a 14% gap), whereas among those who were NEET in 2001, the values are 47% and 70%, respectively (a gap that grows to 23%).

Overall, the results on employment scarring show that ethnic minority men are not particularly penalized; On the contrary, being NEET in 2001 has a similar or reduced scarring effect on later employment probabilities compared to White British. Among women, the results suggest higher scarring effects on employment for Pakistani and Caribbean women.

(p.580) 19.4.2.2. Scarring of occupational status

As with the results for employment, the results for occupational status show that having been NEET leads to lower probabilities (approximately 10 percentage points less; Model b) of attaining a high occupational status, even after controlling for key variables. However, having been a student in 2001 is actually better than having been employed as regards future occupational status. Although much of this effect is explained by the education of individuals (introduced in Model b), probably driven by individuals acquiring a university degree, there is still a small residual effect. This might suggest that having been to university provides additional skills on top of the degree itself and/or access to a wider and better qualified network. Following previous findings (Cheung and Heath 2007; Zuccotti 2015b), Table 19.4 also shows that given equality in their labor market situations in 2001 and their individual and socioeconomic background characteristics (Model b), ethnic minorities do as well as or even better, on average, than White British in terms of occupational attainment. In particular, this is the case for Indian and Bangladeshi men. (p.581)

Table 19.4 Probability of having a (current or most recent) professional/managerial occupation in 2011, by labor market status in 2001 and ethnic group; AME (clustered standard errors)

Men

Women

Model a

Model b

Model a

Model b

Labor market status in 2001 (ref. Employed)

NEET (unemployed or inactive)

–0.172***

–0.098***

–0.240***

–0.105***

(0.0105)

(0.0117)

(0.0072)

(0.0088)

Student

0.276***

0.037***

0.278***

0.036***

(0.0085)

(0.0092)

(0.0081)

(0.0087)

Ethnic group (ref. White British)

Indian

0.105***

0.082***

0.105***

0.063***

(0.0208)

(0.0183)

(0.0209)

(0.0187)

Pakistani

–0.113***

–0.021

–0.041

0.010

(0.0270)

(0.0273)

(0.0265)

(0.0238)

Bangladeshi

–0.025

0.075*

–0.040

0.057

(0.0446)

(0.0433)

(0.0429)

(0.0387)

Caribbean

0.017

0.038

0.037

0.041

(0.0510)

(0.0451)

(0.0409)

(0.0358)

N

35,453

35,453

38,391

38,391

Basic controls

X

X

X

X

Individual, social origin, and neighborhood controls

X

X

Notes: Basic controls: Age, country of birth, origin year, and number of census points. Individual, social origin, and neighborhood controls: Education, family type, parental social class, number of cars, tenure, level of overcrowding, and neighborhood deprivation (past and current). Population: Individuals between 16 and 29 years old in 2001.

(*) p < .10.

** p < .05.

(***) p < .01.

Source: Authors’ calculations based on ONS-LS.

Regarding interactions, the results show that for Bangladeshi men, having been a student in 2001 (vs. having been employed) exerts a more positive effect on occupational status than is the case for White British. This can be clearly observed in Figure 19.3, which shows that Bangladeshi men have higher probabilities of achieving a professional/managerial position compared to White British and that this is particularly strong among those who were students in 2001: These have a 70% probability of attaining a higher occupational status (compared to approximately 50% for equivalent White British).

Do scarring effects vary by ethnicity and gender?

Figure 19.3 Predicted values of male access to (current or most recent) professional/managerial occupation in 2011.

Source: Authors’ calculations based on ONS-LS.

Among women, the results are neither substantive nor statistically significant. In fact, the findings show that the general tendency is for the labor market status in 2001 to have a similar effect across ethnic groups. This can also be interpreted in terms of ethnic gaps remaining similar across statuses in 2001.

In summary, the results of the occupational analysis show that for all groups, having been NEET in 2001 leads, in general, to lower probabilities of attaining a professional/managerial position. However, unemployment/inactivity scars do not vary by ethnicity, nor are ethnic minorities particularly disadvantaged if they were NEET in 2001 compared to White British. On the contrary, some groups (Bangladeshi men) are particularly well positioned with respect to White British: In particular, they have higher probabilities of achieving a professional/managerial position if they were a student in 2001.

(p.582) 19.5. Conclusions

This chapter has sought to bridge a gap between two research agendas that have only marginally interacted: ethnic inequalities and labor market scarring effects for young people. A further dimension we have included here—and that is even less evident in previous research—is the systematic comparison of gender differences between different ethnic groups. The use of the ONS-LS enabled us to follow young individuals over time and to have a sufficiently large number of ethnic minority groups, accompanied with rich and detailed information on their socioeconomic backgrounds, including neighborhood deprivation information attached to individuals.

Our results support previous research indicating the effects of early experiences on subsequent labor market outcomes. On average, we found that those who were not in employment, education, or training in 2001 had an approximately 17 percentage points less chance of being employed in 2011 and an approximately 10 percentage points less chance of being in a professional/managerial position compared to those who were employed in 2001; these results were found after controlling for comparable levels of education, social background, and neighborhood deprivation. We also found that whereas having been employed in 2001 leads to the highest employment probabilities in 2011, having been a student in 2011 leads to the greatest likelihood of attaining a professional/managerial position. This is an interesting finding that might indicate different mechanisms playing a role: Although a previous employment experience seems to be crucial for improving future employability, it is participation in the education system (and the additional benefits it may have in addition to the university degree) that makes the greatest difference in terms of acquiring a good-quality job (see Filandri et al., this volume).

Moving to the core question of the chapter, we found that scarring connected to a previous experience of unemployment or inactivity indeed varies across ethnic groups, and it also depends on the gender of individuals. In particular, examining employment probabilities in 2011, the NEET scar is weaker among Indian and Bangladeshi men by more than half compared to White British men. For women, by contrast, scarring appears to be stronger among Pakistani and Caribbean women than among White British women. The nonemployment of Asian women is an issue of current political concern in the United Kingdom (House of Commons Women and Equalities Committee 2016).

Occupational attainment is not affected by ethnic differences for those with a period of being NEET in 2001. However, Bangladeshis have a particularly high probability of attaining a high occupational status if they were students in 2001, even after controlling for their own educational attainment. Interestingly, we also observe these results for Indian and Bangladeshi students when studying access to employment.

(p.583) Overall, our results for men contradict previous findings for the United Kingdom (Demireva and Kesler 2011) and for other European countries (Reyneri and Fullin 2011; Mooi-Reci and Ganzeboom 2015). The penalties associated with coming from an ethnic minority background do not accrue with being unemployed or inactive, as the stigma argument predicted. On the contrary, some male groups actually showed the opposite trajectory. In the case of Indians, it could be argued that their high educational attainment at the group level might compensate for any experience of unemployment or inactivity in the eyes of employers recruiting them. This might be one of the reasons why we observe relatively higher employment probabilities among Indians who were NEET in 2001. Previous findings (Zuccotti and Platt 2017) also show that Indian men benefit in terms of labor market outcomes from being raised in areas with a higher share of coethnics, which might point to networking mechanisms as potential additional causes. The findings are more puzzling for Pakistani and, especially, for Bangladeshi men because these groups have historically been located in the lower sector of the social structure, and we would expect this to send a negative signal to employers. Further research to untangle this puzzle, as well as to explain the advantage found for Indians, might focus on unmeasured characteristics of these groups, including parental aspirations, motivational factors, the role of networks at the neighborhood and the university level (especially for Indians and Bangladeshis), the exploitation of resources such as internships, and the type of university degrees chosen. Note that these factors might be potential explanations for the scar, but they may also belong to the mechanisms of self-selection into initial conditions, given the limits of our model.

Regarding women, youth unemployment or inactivity leads to lower employment probabilities later in life for Pakistanis and Caribbeans compared to equivalent White British. Group stigmatization might be an explanation for the Caribbeans’ disadvantage; this might also be connected with their overrepresentation as single mothers. The result for Pakistanis might be connected to the role in this group of women, who are often occupied with caring activities, and the low value attached to paid work for them (Peach 2005). Evidence suggesting that these cultural values might actually influence labor market transitions is the fact that having been raised in a neighborhood with a higher share of coethnics negatively impacts on Asian women’s employment probabilities as adults (Zuccotti and Platt 2017). White British women, on the other hand, often combine caring with part-time work (O’Reilly and Fagan 1998; Dale et al. 2002a, 2002b). Interestingly, we do not find particularly strong scarring effects for Bangladeshi women (despite the fact that, independently of their origin status in 2001, they have lower employment probabilities compared to White British women). Pakistani and Bangladeshi women therefore seem to be following different transition trajectories—a finding that deserves further examination.

(p.584) Finally, in addition to showing that scars vary by ethnicity, this chapter challenges the idea that ethnic minorities are always disadvantaged in terms of access to jobs. The fact that some ethnic minority groups—especially second-generation men—are less penalized by a previous unemployment/inactivity experience compared to some of their White British counterparts is in part good news in terms of integration processes. Although much of recent UK policy has focused on limiting new immigration, this has gone hand in hand with efforts to promote integration (Cheung and Heath 2007), as well as new legislation to prevent discrimination and to promote “social cohesion” at the local level (Heath and Yu 2005; Rattansi 2011; Cantle 2012; Meer and Modood 2013). Our results are likely to be in part connected to these measures, although the extent to which they imply a decrease in ethnic discrimination in the labor market requires further exploration.

Significant concerns remain regarding employment probabilities for ethnic minority women and young White British men, who are increasingly “left behind” (Organization for Economic Co-operation and Development 2012). The findings raise questions regarding the groups that policymakers should target. Often, being an ethnic minority is equated with being disadvantaged, but our results show that this is not universally the case in the United Kingdom. Among men, scarring connected to having experienced a period of unemployment or inactivity is particularly high for White British. Our evidence is also supported by previous findings showing that––given equality of education and social background––employment probabilities increased for all ethnic minorities between 2001 and 2011 but declined for White British individuals (Zuccotti 2015b). Among women, however, we do observe a clear “ethnic minority disadvantage” in the labor market. Here, the mechanisms behind these disadvantages deserve greater attention: Although discrimination might be part of the story, and here policymaking should definitely have a role, cultural values (especially among Asians) and possibly fewer employment opportunities in their communities might also contribute to the explanation. Policy to address these multiple and complex outcomes clearly needs to be sensitive to the differential effects and outcomes of gender and ethnicity on young people’s employment transitions.8

References

Bibliography references:

Aassve, Arnstein, Maria Iacovou, and Letizia Mencarini. 2006. “Youth Poverty and Transition to Adulthood in Europe.” Demographic Research 15 (2): 21–50.

Alba, Richard, and Nancy Foner. 2015. Strangers No More: Immigration and the Challenges of Integration in North America and Western Europe. Princeton, NJ: Princeton University Press.

Bell, David N. F., and David G. Blanchflower. 2010. “UK Unemployment in the Great Recession.” National Institute Economic Review 214 (1): 3–25.

Brinbaum, Yael, and Hector Cebolla-Boado. 2007. “The School Careers of Ethnic Minority Youth in France: Success or Disillusion?” Ethnicities 7 (3): 445–74.

Burgess, Simon, Carol Propper, Hedley Rees, and Arran Shearer. 2003. “The Class of 1981: The Effects of Early Career Unemployment on Subsequent Unemployment Experiences.” Labour Economics 10 (3): 291–309.

Cantle, Ted. 2012. Interculturalism: The New Era of Cohesion and Diversity. Hampshire, UK: Palgrave Macmillan.

Carmichael, Fiona, and Robert Woods. 2000. “Ethnic Penalties in Unemployment and Occupational Attainment: Evidence for Britain.” International Review of Applied Economics 14 (1): 71–98.

(p.587) Catney, Gemma. 2016. “Exploring a Decade of Small Area Ethnic (De-)Segregation in England and Wales.” Urban Studies 53 (8): 1691–1709.

Catney, Gemma, and Albert Sabater. 2015. Ethnic Minority Disadvantage in the Labour Market: Participation, Skills and Geographical Inequalities. York, UK: Joseph Rowntree Foundation.

Cheung, Sin Yi, and Anthony Heath. 2007. “Nice Work If You Can Get It: Ethnic Penalties in Great Britain.” In Unequal Chances: Ethnic Minorities in Western Labour Markets, edited by Anthony Heath and Sin Yi Cheung, 507–50. New York: Oxford University Press.

Crawford, Claire, Kathryn Duckworth, Anna Vignoles, and Gill Wyness. 2010. “Young People’s Education and Labour Market Choices Aged 16/17 to 18/19.” Research Report DFE-RR182. London: Department for Education.

Crawford, Claire, and Ellen Greaves. 2015. “Socio-Economic, Ethnic and Gender Differences in HE Participation.” BIS Research Paper 186. London: Department for Business, Innovation and Skills.

Dale, Angela, Nusrat Shaheen, Virinder Kalra, and Edward Fieldhouse. 2002a. “The Labour Market Prospects for Pakistani and Bangladeshi Women.” Work, Employment and Society 16 (1): 5–25.

Dale, Angela, Nusrat Shaheen, Virinder Kalra, and Edward Fieldhouse. 2002b. “Routes into Education and Employment for Young Pakistani and Bangladeshi Women in the UK.” Ethnic and Racial Studies 25 (6): 942–68.

Demireva, Neli, and Christel Kesler. 2011. “The Curse of Inopportune Transitions: The Labour Market Behaviour of Immigrants and Natives in the UK.” International Journal of Comparative Sociology 52 (4): 306–26.

Dorsett, Richard, and Paolo Lucchino. 2014. “Young People’s Labour Market Transitions: The Role of Early Experiences.” NIESR Discussion Paper 419. London: National Institute of Economic and Social Research.

Erikson, Robert, and John Goldthorpe. 1992. The Constant Flux: A Study of Class Mobility in Industrial Societies. Oxford: Clarendon.

Eurofound. 2014. Mapping Youth Transitions in Europe. Luxembourg: Publications Office of the European Union.

Gregg, Paul. 2001. “The Impact of Youth Unemployment on Adult Unemployment in the NCDS.” Economic Journal 111 (475): 626–53.

Gregg, Paul, and Emma Tominey. 2005. “The Wage Scar from Male Youth Unemployment.” Labour Economics 12 (4): 487–509.

Hattersley, Lin, and Rosemary Creeser. 1995. “Longitudinal Study 1971–1991: History, Organisation and Quality of Data.” Series LS 7. London: Her Majesty’s Stationery Office.

Heath, Anthony, and Sin Yi Cheung. 2006. “Ethnic Penalties in the Labour Market: Employers and Discrimination.” Research Report 341. Leeds, UK: Corporate Document Services for the Department for Work and Pensions.

Heath, Anthony, and Sin Yi Cheung, eds. 2007. Unequal Chances: Ethnic Minorities in Western Labour Markets. Oxford: Oxford University Press.

(p.588) Heath, Anthony, and Yaojun Li. 2010. “Struggling onto the Ladder, Climbing the Rungs: Employment and Class Position of Minority Ethnic Groups in Britain.” In Spatial and Social Disparities: Understanding Population Trends and Processes, edited by John Stillwell, Paul Norman, Claudia Thomas, and Claudia Surridge, 83–97. London: Springer.

Heath, Anthony, and Soojin Yu. 2005. “Explaining Ethnic Minority Disadvantage.” In Understanding Social Change, edited by Anthony Heath, John Ermisch, and Duncan Gallie, 187–225. Oxford: Oxford University Press.

House of Commons Women and Equalities Committee. 2016. “Employment Opportunities for Muslims in the UK.” Second Report of Session 2016–17. London: House of Commons.

Kabeer, Naila. 2002. The Power to Choose: Bangladeshi Women and Labour Market Decisions in London and Dhaka. London: Verso.

Kalter, Frank, and Irena Kogan. 2006. “Ethnic Inequalities at the Transition from School to Work in Belgium and Spain: Discrimination or Self-Exclusion?” Research in Social Stratification and Mobility 24 (3): 259–74.

Kalwij, Adriaan S. 2004. “Unemployment Experiences of Young Men: On the Road to Stable Employment?” Oxford Bulletin of Economics and Statistics 66 (2): 205–37.

Kirchner Sala, Laura, Vahé Nafilyan, Stefan Speckesser, and Arianna Tassinari. 2015. “Youth Transitions to and Within the Labour Market: A Literature Review.” BIS Research paper 255a. London: Department for Business, Innovation and Skills.

Kogan, Irena. 2011. “New Immigrants—Old Disadvantage Patterns? Labour Market Integration of Recent Immigrants into Germany.” International Migration 49 (1): 91–117.

Longhi, Simonetta, Cheti Nicoletti, and Lucinda Platt. 2013. “Explained and Unexplained Wage Gaps Across the Main Ethno-Religious Groups in Great Britain.” Oxford Economic Papers 65 (2): 471–93.

Luijkx, Ruud, and Maarten H. J. Wolbers. 2009. “The Effects of Non-Employment in Early Work–Life on Subsequent Employment Chances of Individuals in the Netherlands.” European Sociological Review 25 (6): 647–60.

Meer, Nasar, and Tariq Modood. 2013. “The ‘Civic Re-balancing’ of British Multiculturalism, and Beyond . . .” In Challenging Multiculturalism: European Models of Diversity, edited by Raymond Taras, 75–96. Edinburgh, UK: Edinburgh University Press.

Modood, Tariq, Richard Berthoud, Jane Lakey, James Nazroo, Patten Smith, Satnam Virdee, and Sharon Beishon. 1997. Ethnic Minorities in Britain: Diversity and Disadvantage—Fourth National Survey of Ethnic Minorities. London: Policy Studies Institute.

Mooi-Reci, Irma, and Harry B. G. Ganzeboom. 2015. “Unemployment Scarring by Gender: Human Capital Depreciation or Stigmatization? Longitudinal Evidence from the Netherlands, 1980–2000.” Social Science Research 52: 642–58.

(p.589) Mroz, Thomas A., and Timothy H. Savage. 2006. “The Long-Term Effects of Youth Unemployment.” Journal of Human Resources 41 (2): 259–93.

Nazroo, James, and Dharmi Kapadia. 2013. “Have Ethnic Inequalities in Employment Persisted Between 1991 and 2011?” ESRC Series Dynamics of Diversity: Evidence from the 2011 Census. Manchester, UK: Centre on Dynamics of Ethnicity.

Norman, Paul, and Paul Boyle. 2014. “Are Health Inequalities Between Differently Deprived Areas Evident at Different Ages? A Longitudinal Study of Census Records in England and Wales, 1991–2001.” Health & Place 26: 88–93.

Norman, Paul, Paul Boyle, and Philip Rees. 2005. “Selective Migration, Health and Deprivation: A Longitudinal Analysis.” Social Science & Medicine 60 (12): 2755–71.

Office for National Statistics. 2010. Standard Occupational Classification 2010: The National Statistics Socio-economic Classification: (Rebased on the SOC2010) User Manual. Basingstoke, UK: Palgrave Macmillan.

Omori, Yoshiaki. 1997. “Stigma Effects of Nonemployment.” Economic Inquiry 35 (2): 394–416.

O’Reilly, Jacqueline, Werner Eichhorst, András Gábos, Kari Hadjivassiliou, David Lain, Janine Leschke, et al. 2015. “Five Characteristics of Youth Unemployment in Europe: Flexibility, Education, Migration, Family Legacies, and EU Policy.” Sage Open 5 (1): 1–19. doi:10.1177/2158244015574962.

O’Reilly, Jacqueline, and Colette Fagan, eds. 1998. Part-Time Prospects: An International Comparison of Part-Time Work in Europe, North America and the Pacific Rim. London: Routledge.

Organization for Economic Co-operation and Development (OECD). 2012. OECD Employment Outlook 2012. Paris: OECD.

Peach, Ceri. 2005. “Social Integration and Social Mobility: Spatial Segregation and Intermarriage of the Caribbean Population in Britain.” In Ethnicity, Social Mobility and Public Policy, edited by Glenn C. Loury, Tariq Modood, and Steven M. Teles, 178–203. Cambridge: Cambridge University Press.

Phillips, Deborah. 1998. “Black Minority Ethnic Concentration, Segregation and Dispersal in Britain.” Urban Studies 35 (10): 1681–1702.

Platt, Lucinda. 2007. “Making Education Count: The Effects of Ethnicity and Qualifications on Intergenerational Social Class Mobility.” Sociological Review 55 (3): 485–508.

Rattansi, Ali. 2011. Multiculturalism: A Very Short Introduction. Oxford: Oxford University Press.

Reyneri, Emilio, and Giovanna Fullin. 2011. “Ethnic Penalties in the Transition to and from Unemployment: A West European Perspective.” International Journal of Comparative Sociology 52 (4): 247–63.

Robinson, Vaughan, and Rina Valeny. 2005. “Ethnic Minorities, Employment, Self-Employment, and Social Mobility in Postwar Britain.” In Ethnicity, Social (p.590) Mobility and Public Policy, edited by Glenn C. Loury, Tariq Modood, and Steven M. Teles, 414–47. Cambridge: Cambridge University Press.

Schmelzer, Paul. 2011. “Unemployment in Early Career in the UK: A Trap or a Stepping Stone?” Acta Sociologica 54 (3): 251–65.

Schmillen, Achim, and Joachim Möller. 2012. “Distribution and Determinants of Lifetime Unemployment.” Labour Economics 19 (1): 33–47.

Schmillen, Achim, and Matthias Umkehrer. 2013. “The Scars of Youth—Effects of Early-Career Unemployment on Future Unemployment Experience.” IAB Discussion Paper 6/2013. Leibniz: Institute for Employment Research, German Federal Employment Agency.

Silberman, Roxane, and Irène Fournier. 2008. “Second Generations on the Job Market in France: A Persistent Ethnic Penalty.” Revue Française de Sociologie 49 (5): 45–94.

Van de Werfhorst, Herman G., and Frank van Tubergen. 2007. “Ethnicity, Schooling, and Merit in the Netherlands.” Ethnicities 7 (3): 416–44.

Wood, Martin, Jon Hales, Susan Purdon, Tanja Sejersen, and Oliver Hayllar. 2009. “A Test for Racial Discrimination in Recruitment Practice in British Cities.” Research report No. 607. London: Department for Work and Pensions.

Zuccotti, Carolina V. 2015a. “Do Parents Matter? Revisiting Ethnic Penalties in Occupation Among Second Generation Ethnic Minorities in England and Wales.” Sociology 49 (2): 229–51.

Zuccotti, Carolina V. 2015b. “Shaping Ethnic Inequalities: The Production and Reproduction of Social and Spatial Inequalities Among Ethnic Minorities in England and Wales.” PhD dissertation, European University Institute.

Zuccotti, Carolina V., Harry B. G. Ganzeboom, and Ayse Guveli. 2017. “Has Migration Been Beneficial for Migrants and Their Children?” International Migration Review 51 (1): 97–126. doi:10.1111/imre.12219.

Zuccotti, Carolina V., and Lucinda Platt. 2017. “Does Neighbourhood Ethnic Concentration in Early Life Affect Subsequent Labour Market Outcomes? A Study Across Ethnic Groups in England and Wales.” Population, Space and Place 23 (6): e2041. doi:10.1002/psp.2041.

Table A19.1 Probability of being employed and probability of having a (current or most recent) professional/managerial occupation in 2011; AME (clustered standard errors)—full models

Employment

Professional/managerial

Men

Women

Men

Women

Model a

Model b

Model a

Model b

Model a

Model b

Model a

Model b

Status in 2001 (ref. Employed)

NEET (unemployed or inactive)

–0.338***

–0.175***

–0.357***

–0.173***

–0.172***

–0.098***

–0.240***

–0.105***

(0.0103)

(0.0078)

(0.0079)

(0.0073)

(0.0105)

(0.0117)

(0.0072)

(0.0088)

Student

–0.005

–0.041***

0.056***

–0.012*

0.276***

0.037***

0.278***

0.036***

(0.0045)

(0.0063)

(0.0052)

(0.0071)

(0.0085)

(0.0092)

(0.0081)

(0.0087)

Ethnic group (ref. White British)

Indian

0.003

0.004

–0.028

–0.061***

0.105***

0.082***

0.105***

0.063***

(0.0120)

(0.0109)

(0.0176)

(0.0185)

(0.0208)

(0.0183)

(0.0209)

(0.0187)

Pakistani

–0.020

0.007

–0.211***

–0.160***

–0.113***

–0.021

–0.041

0.010

(0.0164)

(0.0130)

(0.0250)

(0.0223)

(0.0270)

(0.0273)

(0.0265)

(0.0238)

Bangladeshi

0.002

0.042***

–0.180***

–0.100***

–0.025

0.075*

–0.040

0.057

(0.0243)

(0.0159)

(0.0393)

(0.0325)

(0.0446)

(0.0433)

(0.0429)

(0.0387)

Caribbean

–0.067*

–0.023

–0.064*

–0.073**

0.017

0.038

0.037

0.041

(0.0347)

(0.0246)

(0.0343)

(0.0324)

(0.0510)

(0.0451)

(0.0409)

(0.0358)

Family type (ref. Single, no children)

Couple, no children

0.092***

0.077***

0.074***

0.035***

(0.0048)

(0.0062)

(0.0077)

(0.0087)

Single with children

–0.010

–0.084***

0.022

–0.094***

(0.0123)

(0.0079)

(0.0199)

(0.0107)

Couple with children

0.076***

–0.079***

0.047***

–0.056***

(0.0044)

(0.0061)

(0.0070)

(0.0078)

Education (ref. Level 1)

No education

–0.149***

–0.289***

–0.046***

–0.041*

(0.0117)

(0.0228)

(0.0152)

(0.0250)

Other

–0.043***

–0.086***

0.076***

0.044*

(0.0095)

(0.0207)

(0.0137)

(0.0234)

Level 2

–0.014

–0.015

0.142***

0.088***

(0.0091)

(0.0203)

(0.0135)

(0.0231)

Level 3

0.022**

0.049**

0.191***

0.154***

(0.0089)

(0.0204)

(0.0137)

(0.0233)

Level 4+

0.044***

0.085***

0.525***

0.497***

(0.0085)

(0.0202)

(0.0130)

(0.0230)

Tenure (ref. Owner)

Social rent

–0.019***

–0.024***

–0.040***

–0.039***

(0.0044)

(0.0055)

(0.0076)

(0.0071)

Private rent

–0.005

–0.014*

–0.008

–0.012

(0.0060)

(0.0080)

(0.0104)

(0.0100)

Number of cars (ref. None)

1 car

0.012***

0.009

0.008

0.016**

(0.0043)

(0.0053)

(0.0077)

(0.0071)

2+ cars

0.017***

0.020***

0.020**

0.038***

(0.0055)

(0.0069)

(0.0094)

(0.0088)

Persons per room (ref. 1 person per room)

>1.5 persons

0.001

–0.026*

–0.030

–0.008

(0.0111)

(0.0153)

(0.0235)

(0.0212)

1.5 persons

0.006

–0.016

0.029

–0.055*

(0.0157)

(0.0181)

(0.0300)

(0.0283)

>1 and <1.5 persons

–0.006

–0.004

–0.015

–0.007

(0.0064)

(0.0074)

(0.0117)

(0.0106)

≥0.75 and <1 person

0.010**

0.003

0.004

–0.001

(0.0043)

(0.0054)

(0.0073)

(0.0069)

<0.75 person

0.009**

0.001

0.025***

0.012*

(0.0044)

(0.0055)

(0.0073)

(0.0069)

Parental social class (ref. Manual [V + VI + VII])

No earners/no code

–0.015**

–0.026***

0.043***

0.013

(0.0064)

(0.0076)

(0.0126)

(0.0114)

Routine nonmanual (III)

0.003

–0.004

0.048***

0.026***

(0.0046)

(0.0058)

(0.0076)

(0.0072)

Petite bourgeoisie (IV)

0.003

–0.001

–0.007

–0.006

(0.0053)

(0.0067)

(0.0088)

(0.0083)

Professional/managerial (I + II)

0.005

–0.009

0.085***

0.053***

(0.0045)

(0.0057)

(0.0073)

(0.0069)

Carstairs quintile in origin (ref. Q1: Least deprived)

Q2

–0.004

0.005

–0.009

0.002

(0.0051)

(0.0064)

(0.0073)

(0.0070)

Q3

–0.002

0.003

–0.003

–0.001

(0.0052)

(0.0067)

(0.0079)

(0.0074)

Q4

–0.008

0.012*

–0.023***

0.003

(0.0054)

(0.0069)

(0.0084)

(0.0079)

Q5

–0.010*

0.010

–0.017*

–0.002

(0.0059)

(0.0075)

(0.0096)

(0.0091)

Carstairs quintile in 2011 (ref. Q1: Least deprived)

Q2

–0.003

0.013*

–0.016*

–0.007

(0.0058)

(0.0074)

(0.0088)

(0.0083)

Q3

–0.015***

0.022***

–0.034***

–0.027***

(0.0058)

(0.0072)

(0.0089)

(0.0084)

Q4

–0.021***

0.018**

–0.047***

–0.020**

(0.0059)

(0.0075)

(0.0093)

(0.0087)

Q5

–0.031***

0.005

–0.042***

–0.022**

(0.0065)

(0.0082)

(0.0105)

(0.0100)

Age

Age in 2001

0.001*

–0.001

0.004***

0.004***

0.016***

0.006***

0.013***

0.005***

(0.0007)

(0.0007)

(0.0009)

(0.0008)

(0.0012)

(0.0011)

(0.0011)

(0.0010)

Origin year (ref. 1981)

1991

0.000

–0.007***

0.005**

–0.004*

0.010***

–0.022***

0.013***

–0.019***

(0.0018)

(0.0020)

(0.0023)

(0.0026)

(0.0030)

(0.0032)

(0.0028)

(0.0030)

Number of census points (ref. 3)

4 census points

0.017***

–0.003

0.031***

0.005

0.098***

0.023***

0.090***

0.022***

(0.0055)

(0.0050)

(0.0069)

(0.0064)

(0.0091)

(0.0084)

(0.0087)

(0.0080)

Country of birth

UK-born

0.008

0.026**

–0.020

–0.017

–0.010

0.035

–0.011

0.015

(0.0135)

(0.0118)

(0.0179)

(0.0170)

(0.0251)

(0.0228)

(0.0237)

(0.0207)

N

36,886

36,886

40,294

40,294

35,453

35,453

38,391

38,391

Notes: Robust (clustered) standard errors in parentheses. Population: Individuals between 16 and 29 years old in 2001.

Source: Authors’ calculations based on ONS-LS.

(p.592) (p.593) (p.594) (p.595) (p.596)

Table A19.2 Predicted values of employment and professional/managerial occupation in 2011, by labor market status in 2001, ethnic group, and gender

Employment

Professional/managerial occupation

Men

Women

Men

Women

Value

SE

Value

SE

Value

SE

Value

SE

NEET

White British

0.79

0.01

0.70

0.01

0.30

0.01

0.30

0.01

Indian

0.88

0.03

0.59

0.07

0.39

0.08

0.33

0.07

Pakistani

0.85

0.04

0.47

0.05

0.33

0.09

0.28

0.06

Bangladeshi

0.91

0.04

0.59

0.07

0.33

0.12

0.44

0.11

Caribbean

0.85

0.07

0.53

0.12

0.45

0.17

0.33

0.10

Student

White British

0.92

0.00

0.87

0.01

0.47

0.01

0.48

0.01

Indian

0.93

0.01

0.82

0.03

0.63

0.03

0.59

0.04

Pakistani

0.91

0.02

0.64

0.04

0.43

0.05

0.55

0.05

Bangladeshi

0.98

0.01

0.71

0.06

0.72

0.08

0.56

0.08

Caribbean

0.90

0.04

0.73

0.07

0.46

0.11

0.55

0.10

Employed

White British

0.95

0.00

0.88

0.00

0.43

0.00

0.44

0.01

Indian

0.93

0.01

0.81

0.03

0.50

0.03

0.51

0.04

Pakistani

0.96

0.01

0.74

0.04

0.40

0.06

0.42

0.05

Bangladeshi

0.95

0.02

0.79

0.06

0.43

0.08

0.47

0.08

Caribbean

0.92

0.03

0.85

0.04

0.49

0.08

0.49

0.06

Notes: Variables set to their mean: Age, country of birth, origin year, number of census points, parental social class, number of cars, tenure, level of overcrowding, neighborhood deprivation (past and current), education, and family type. Population: Individuals between 16 and 29 years old in 2001.

Source: Authors’ calculations based on ONS-LS.

Notes:

(1) In this section, we use nonemployment to identify individuals who are either unemployed or engaged in any other activity that does not involve working or studying. Some studies include students in their comparisons (hence identify NEET populations), whereas others do not.

(2) Some cell counts, percentages, and totals shown in the tables created with ONS-LS data have been modified in order to comply with publication rules (p.585) established by the Office for National Statistics. These modifications, however, do not affect the main findings derived from the regression models. The permission of the Office for National Statistics to use the Longitudinal Study is gratefully acknowledged, as is the help provided by staff at the Centre for Longitudinal Study Information and User Support (CeLSIUS). CeLSIUS is supported by the ESRC Census of Population Programme (Award ref. ES/K000365/1). The authors alone are responsible for the interpretation of the data. This work contains statistical data from ONS, which is Crown Copyright. The use of the ONS statistical data in this work does not imply the endorsement of the ONS in relation to the interpretation or analysis of the statistical data. This work uses research data sets, which may not exactly reproduce National Statistics aggregates.

(3) Ethnicity is measured by a question on self-identification (measured in 2011; when missing, self-identification in 2001 is used). In 2011, the question is formulated as follows: “What is your ethnic group?” The options are White (English/Welsh/Scottish/Northern Irish/British; Irish; Gypsy or Irish traveler; other White), Mixed/multiple ethnic groups (White and Black Caribbean; White and Black African; White and Asian; any other Mixed/multiple ethnic background; open question), Asian/Asian British (Indian, Pakistani, Bangladeshi, Chinese; any other Asian background; open question), Black/African/Caribbean/Black British (African; Caribbean; any other Black/African/Caribbean background; open question), and Other ethnic group (Arab; any other ethnic group). Note that the “Gypsy or Irish traveler” and “Arab” categories were not specified separately in the 2001 census form.

(4) Individuals of whom one parent is born abroad and the other in the United Kingdom are therefore excluded from the analysis. White British with foreign-born parents (or a foreign-born parent in the case of single-parent households) and ethnic minorities with UK-born parents (or a foreign-born parent in the case of single-parent households) are also excluded. African and Chinese were excluded due to the small number of cases.

(5) Neighborhood deprivation is expressed in population-weighted quintiles and is obtained at the ward level. The ward is the key building block of UK administrative geography and is used to elect local government councilors. Wards vary in terms of size and population, with the average population amounting to 4,000. In general, the smallest and most populous wards are in metropolitan areas, where the majority of ethnic minorities are found. The permission of Dr. Paul Norman, School of Geography, University of Leeds, to use the 2011 Carstairs Index of Deprivation he created is gratefully acknowledged. Please see Norman and Boyle (2014) for use of the Carstairs Index in conjunction with the ONS-LS.

(6) “Ivprobit,” which is the command we should use given that our outcomes are dichotomous, does not allow factorial endogenous variables (i.e., status in 2001), but only continuous variables. We have, nevertheless, run a model (without (p.586) interactions) in which a recoded version of status in 2001—being NEET (vs. being in employment or in education)—is used as an endogenous dummy variable, and neighborhood deprivation when individuals were 0–15 years old is used as an instrument. The results are similar to those presented here. Another option would be to use the command “ivregress” and ignore the fact that our dependent variable is dichotomous. We have tried this model as well, but the outcomes are difficult to interpret (predictions are out of range, i.e., they exceed 1, and have very large standard errors). All results are available on request.

(7) To identify relevant interactions (shown in Figures 19.1–19.3), we plotted all interactions in graphs and also created “contrasts,” which show the size of the interaction effect and whether or not it is statistically significant. In the study of employment in 2011, we have identified contrasts that are statistically significant at p < .10 for Indian and Bangladeshi men, for whom the effect of being employed in 2001 versus being NEET is different compared to White British men. We have also found, in the analysis of occupations in 2011, that the effect of being a student versus being NEET is different for Bangladeshi men (p < .10) compared to White British men. Finally, we have identified relevant interactions when the observed effects were quite substantive (but the contrasts were statistically significant at larger p values). In the analysis of employment, the effect of being employed in 2001 versus being NEET is different for Pakistani men and women (p < .14) and for Caribbean women (p < .30) compared to White British men and women.

(8) An earlier version of this chapter was published in the journal Human Relations.