Abstract
In the twentieth century, the Irish-born population in England has typically been in worse health than both the native population and the Irish population in Ireland, a reversal of the commonly observed healthy migrant effect. Recent birth cohorts living in England and born in Ireland, however, are healthier than the English population. The substantial Irish migrant health penalty arises principally for cohorts born between 1920 and 1960. In this article, we attempt to understand the processes that generated these changing migrant health patterns for Irish migrants to England. Our results suggest a strong role for economic selection in driving the dynamics of health differences between Irish-born migrants and white English populations.
Introduction
Migration research has traditionally examined the relative role of pre-migration selectivity of migrants, including their health as well as impacts after arrival of assimilation into receiving countries on migrant health (Borjas 1985; Schultz 1998). This research has typically emphasized the “healthy” migrant effect, pointing out that, on average, migrants are healthier than populations in sending countries from which they are drawn (Jasso et al. 2004). This research has also argued that migrants often move to countries where either the stress of migrating to a new land or new adopted health behaviors may be worse than in sending countries, such that the health advantage of migrants dissipates over time (Antecol and Bedard 2006). Both in terms of initial selectivity and subsequent assimilation, demographic and economic literatures focused almost exclusively on physical health issues. Far less research exists on how pre-migration mental health influences initial migration selection. In this article, we examine a case where selectivity of migrants appears to be negative and pre-migration mental health issues are very important in that selection.
We examine health and mental well-being of Irish immigrants living in England in the twentieth century. Health of migrants and its evolution over time are key to economic outcomes of migrants, including wages and labor supply, given that health is a central determinant of both (Grossman 1972; Smith 1999). Historical interactions between Ireland and England underwent several shifts during the twentieth century that changed incentives to migrate as well as composition of flows. The countries are geographically proximate, and England represented the main destination country for Irish migrants in the twentieth century. Irish migrants to England had well-documented physical and mental health problems (Marmot et al. 1984). While the United States was the main destination for Irish migrants in the nineteenth century, England was the overwhelming destination of choice of Irish migrants and remained so for much of the twentieth century. We aim to understand these issues in the context of a selection model incorporating pre-migration mental health.
Our data point to an Irish-born population living in England less healthy in physical and mental health than the English-born population in England and the Irish-born population living in Ireland. This masks substantial heterogeneity. Among Irish migrants to England, individuals in the pre-1920 Irish birth cohort who migrated were slightly healthier, migrants in birth cohorts between the 1920s and 1960s were markedly less healthy, and recent migrants to England are healthier than comparable English and Irish-born populations still living in their countries of birth.
If Irish migrants to England were randomly selected from the Irish-born population and their health did not suffer from living in England, health differences between Irish migrants to England and English-born populations would reflect only differences between the Irish-born and the English-born populations. Health differences between Irish migrants and those born in Ireland or England arise for two reasons. First, Irish migrants experience more stressful conditions in England that may lead to worse health. Second, Irish migrants to England select themselves based on a bundle of characteristics, including health. Irish migrants living in England throughout most of the twentieth century experienced many periods of political stress and occasional violence between the two countries, which may have affected their mental health. This second explanation suggests that health of Irish migrants to England relative to the Irish-born who did not migrate and relative to the English-born living in England should deteriorate with duration of stay in England.
Conditions in Ireland for many children during this period were not stress-free, and a large fraction of Irish children who were abused in Ireland appear to have migrated to England. This suggests that pre-migration conditions in Ireland may have been key in shaping selectivity of the Irish who did migrate to England. Our analysis uses a combination of micro-survey data collected in England in the late 1990s and early part of the current decade combined with a comparable data surveying Irish populations in Ireland from the same period.
This article is structured as follows. The second section describes datasets documenting comparability of key outcome variables. Following that, we outline the background of twentieth century Irish migration, providing figures on numbers of migrants and the current stock; review prior knowledge of health patterns of Irish migrants living in England and summarize evidence on physical and sexual abuse of children in Ireland; and present descriptive models of health differences of successive cohorts of Irish migrants in England. Then we outline a theoretical model of migrant selection recognizing that migrants may have low or even negative costs of migration because of physic costs associated with place of origin. We model selection effects by examining education of migrants, assuming that this is determined prior to migration and is related to adult health (Lleras-Muney 2005). In this section, we also investigate whether assimilation into England may have generated Irish migrant health patterns, using length of time spent in England of Irish migrants as an exposure variable. The final section summarizes our main conclusions.
Data and Descriptive Statistics
The three populations of interest in this research are (1) the Irish-born who migrated to England, (2) the English-born still living in England, and (3) the Irish-born still living in Ireland. We use micro-data from the 1999 and 2004 Health Survey for England (HSE) to examine health status trajectories of the Irish-born population living in England and the English-born population living in England. To describe the Irish-born population still living in Ireland, we use two data sources. First, we use Wave 7 of the 2000 Living in Ireland Survey (LII) to include Irish born in Ireland still living in Ireland in our analyses. Second, we use the 2011 (The Irish Longitudinal Study on Ageing) TILDA survey to include, for ages 50 and older, attributes of the Irish-born who stayed in Ireland and those who migrated but subsequently returned to Ireland, including the extent to which both suffered sexual and physical abuse in childhood.
The Health Survey for England (HSE)
The HSE is an annual survey aimed at monitoring demographic health trends in England. Questionnaires are consistent from year to year, enabling us to merge survey years obtaining larger cross-sections of data. Each year incorporates a different theme designed for a specific issue. The 1999 and 2004 surveys focused on health of minority ethnic groups. These surveys are chosen for our pooled cross-section because they enumerate respondents’ country of birth, which is not available in other survey years.
We compose our working data sample by combining the 1999 and 2004 data samples. We take Irish-born population living in England from each survey. The sample of white English-born population still living in England is obtained from the 1998 survey. The 1998 survey was the closest survey to the 1999 and 2004 surveys, and simultaneously collected information on both country of birth and ethnicity. The indigenous English population is not present in the 1999 and 2004 surveys, and there is no clear way of identifying this group in other surveys.
Living in Ireland (LII)
Micro data from Wave 7 of the Living in Ireland (LII) survey are used to examine Irish non-migrants. All non–Irish-born are excluded from our sample, resulting in 7,883 observations. Wave 7 of the LII collected several variables, allowing us to examine key differences between Irish-born populations in Ireland and in England. The LII survey design is similar to the HSE. Both are conducted by personal interview, contain common demographic variables, and are nationally representative. The 2000 LII wave was fielded at a similar time to the HSE micro data used.
The Irish Longitudinal Study on Ageing (TILDA)
TILDA is a nationally representative survey of individuals and spouses aged 50 and older living in the Republic of Ireland. The first wave was successfully carried out between late 2009 and mid-2011. Data on more than 8,500 people were collected with a response rate of 62 %. The computer-assisted personal interview (CAPI) covered topics such as demographics; incomes, wealth, and consumption; transfers to children and parents; job and migration histories and labor force status; and health status (physical and cognitive), health behaviors, and health care utilization. The self-completed questionnaire covered topics such as mental health, quality of relationships, and histories of childhood physical and sexual abuse. All TILDA respondents were asked whether they were physically abused by parents or by others and whether they were sexually abused by parents or by others all before they were 18 years old (Barrett and Mosca 2011).
Measures Utilized From the Surveys
Key demographic characteristics, age, sex, and marital status exist in all surveys and are defined similarly. Years in full-time education is computed from respondents’ answers to questions about age when leaving full-time education. A cardinal measure of education based on highest level of education completed was constructed.1
Our main health outcome measure is self-rated health, assessed very similarly in the LII and HSE.2 Three health-dependent variables are constructed only for the Irish-born population living in England and white English populations living in England both from the HSE. We use the HSE to construct three variables, taking the value 1 if the respondent had cardiovascular, muscular, or respiratory conditions, respectively.
Mental health is a central outcome. The General Health Questionnaire is a 12-question scale based on factors such as concentration; losing sleep; confidence; under strain; feeling depressed, happy, worthless; self-perceived capability. This questionnaire has been favorably tested for reliability and validity (Hannan 1970). For each of the 12 questions, responses are measured on a Likert scale and can score from 0 to 3, representing not at all, no more than usual, rather more than usual, and much more than usual, respectively. Total score is summed across 12 questions, with higher values implying worse mental health. Our other variables are specific to a particular analysis and described when we present that analysis.
Background and Literature
Background of Irish Migration to England
Because our data from the HSE refer only to England, we examine the Irish-born population living in England. However, we occasionally reference data relating to Britain (combined countries of England, Scotland,3 and Wales) or the United Kingdom (Britain plus Northern Ireland) because that is the form in which the original data were published. Still, England represents the large majority of Irish/UK population. Of 390,000 Irish-born population estimated to be in the United Kingdom in 2009, more than 330,000 were in England (about 85 %).
Figure 1 depicts by calendar year post–World War I net migration in Ireland, which was mostly a net outflow from the Irish Republic. Britain was the major destination for most post-1940s Irish migrants, with most going to England. A report by the National Economic and Social Council (NESC 1991) estimated that Britain accounted for 79 %, 86 %, and 54 % of migration outflows from the Republic of Ireland in the 1946–1951, 1971–1981, and 1981–1990 periods, respectively. The 2001 census enumerated 460,287 Irish-born people living in England, a figure representing a sizable proportion (13.5 %) of total Irish-born population living in Ireland.4
Several studies (e.g., O’Grada and Walsh 1994) have examined macroeconomic determinants of twentieth century Irish migration. The results of these time-series econometric models are consistent; a close link existed in relative differences between Irish and British labor markets. The post–World War II, pre-1980s cohort was typically young and unskilled, from economically underdeveloped regions of Ireland (NESC 1991). This profile began to change in the late 1970s to a higher-educated, more-skilled group from a wide spectrum of Irish regions. The 1960s and 1970s were a period of relative wage convergence between Ireland and England (Barrett 1999; O’Grada and Walsh 1994), which is important in understanding changing health composition of migrants. This wage convergence may have been amplified by increasing demand for skilled labor during latter part of the twentieth century (Barrett 1999).
The period we study was marked by periodic violence between the Irish Republic and England. Following the War of Independence from 1919 to 1921, a treaty was negotiated, which created Irish Free State (later, the Republic of Ireland) and retained six Ulster counties (Northern Ireland) under British rule in the United Kingdom. During the 1960s, civil rights marches in the North developed into widespread political violence. The period of political violence in Northern Ireland5 from the late-1960s to the mid-1990s—known as the “Troubles”—created more 3,000 casualties and included a bombing campaign on the British mainland conducted by the Irish Republican Army. Casualties intensified in 1970s and 1980s, with many killings by both Republican and loyalist groups. Such attacks were highly publicized in England, and although they are difficult to map directly onto anti-Irish sentiment, they are clearly a measure of tensions that Irish migrants to England might have faced (Ferriter 2004).
Literature on Health of Irish Migrants in Britain
Public health research has consistently found Irish migrants to be in worse health than their English counterparts. Mortality is the most commonly used health barometer of the Irish-born population living in England. The Office of Population Censuses and Surveys (OPCS 1990) Mortality and Geography review in the 1980s reported high mortality rates among Irish-born men and women in both the 20–49 and 20–69 age cohorts who were living in England. This Irish population had the highest mortality rates of all groups in this study, excluding African women in the 20–49 age category. Marmot et al. (1984) analyzed mortality disparities between the Irish-born living in England and the English and Welsh. After socioeconomic status was controlled for, Irish men still have higher mortality rates: for Irish migrants, the standard mortality ratio (SMR) was 122 in total compared with 88 for European migrants. Similarly, Raftery et al. (1990) highlighted comparatively high death rates from accidents, poisoning, and violence in this group. Adelstein et al. (1986) examined SMRs and showed that Irish migrant health patterns were not typical of the Irish population in Ireland.
Several researchers have explored mental well-being of the Irish in Britain. Cochrane (1983) highlighted large proportions of Irish migrants in Britain seeking hospital consultations for issues of mental health. Leavey et al. (2007) analyzed Irish migrants in the United Kingdom’s mental health system. They conducted interviews with Irish migrants, some suffering from depression, to explore reasons for psychological distress. They reported a strong likelihood that physiological distress was established in the pre-migration period.
Although health disadvantages of Irish migrants to England are well documented, there is little agreement on the source of these disadvantages and reasons for heterogeneity. Studying Irish migrant populations in England offers an opportunity to understand roles of changing selection, assimilation, and mental health in generating migrant health differences over time.
Childhood Abuse in Ireland
Table 1 presents data by 10-year birth cohorts on prevalence of childhood abuse before age 18 of Irish residents in Ireland as derived from TILDA.6 Table 1 contains total abuse numbers with separate data on physical and sexual abuse during childhood years. Besides the Irish-born living in Ireland at the time of the TILDA survey, separate figures are given in Table 1 for “stayers” (who never migrated from Ireland) and for those migrants who returned to Ireland. Note that migrants in TILDA refer to those migrants who eventually returned to Ireland. Although we have no direct data on migrants who remained in England without returning to Ireland, it seems likely from data described later that returning migrants understate abuse among Irish migrants who did not return to Ireland. There were not strong differences in childhood abuse for boys and girls, so genders are combined in Table 1. Online Resource 1 contains data on abuse by gender.
About 11.5 % of TILDA respondents reported some type of abuse during childhood, with about 7 % reporting childhood sexual abuse and 8 % reporting physical abuse, indicating strong overlap in experiencing both sexual and physical abuse during childhood. Most importantly, childhood abuse rates are 50 % higher among prior Irish migrants compared with stayers who never migrated from Ireland.
Both physical and sexual abuse rates are statistically higher among prior Irish migrants compared with stayers. Abuse rates are particularly high in 1941–1960 birth cohorts of return migrants from England. Among those born between 1951 and 1960, 1 in every 5 return migrants suffered some form of childhood abuse, with almost 1 in 10 experiencing sexual abuse. If, as the narrative in the next few paragraphs indicates, prior Irish migrants who did not return to Ireland had even higher rates of abuse, then childhood abuse rates among Irish migrants living in England are very high indeed.
Other evidence supports the hypothesis that abuse and neglect of children in twentieth century Ireland shaped migrant health trajectories in England. The 2002 Sexual Abuse and Violence in Ireland report (SAVI), based on telephone interviews with more than 3,000 people in Ireland, revealed high rates of childhood abuse incidents among men and women (McGee et al. 2002). Some 20.4 % of female respondents reported abuse as children, 10 % reported noncontact abuse, and 5.4 % reported penetrative sexual abuse. For men, 16.2 % reported contact sexual abuse as children, 7.4 % reported noncontact abuse, and 2.7 % of men reported penetrative sexual abuse as children. Almost one-third of women and one-quarter of men reported some form of sexual abuse as children.
Forty percent of cases involved multiple abuses. For girls, 24 % of perpetrators were family members, 24 % were strangers, and 52 % were nonfamily members known to the girl. For boys, 14 % were family members, one-fifth were nonfamily members, and more than one-half were nonfamily members known to the child. Of the abused, about one-third of women and one-quarter of men reported moderate or extreme effects on their adult lives, with 25 % of women and 16 % of men reporting symptoms consistent with posttraumatic stress disorder (PTSD).
Other evidence comes from investigating abuse in industrial or reformatory schools in Ireland (schools for children with excess truancy or crime involvement, or where families were deemed incapable of caring for children). The Ryan Commission interviewed more than 1,200 people who attended these schools, and more than 30 % were currently living in the United Kingdom. An estimate that the Commission employed was that approximately 50 % of children attending these schools migrated.7 To get a sense of scale, the Commission estimated that approximately 100,000 people attended these schools from 1930 to 1970.8 Based on net migration flows in Fig. 1 and a 0.80 % migration to the United Kingdom, we estimate a net migration of 790,000 Irish to England between 1927 and 1970. Attendees of these schools would then represent 6.3 % of these migrants.
Pre-migration mental problems of Irish migrants to England are not limited to issues associated with these schools. Several epidemiological studies find that pre-migration factors were operating in setting mental health of Irish migrants into the United Kingdom. Leavey et al. (2007) found that men with pre-migration mental health conditions were seven times more likely to have an existing psychological problem. Such effects exist for women but to a smaller degree (1.6 times more likely). Leavey et al. (2007) also reported that 40 % of Irish migrants in their depressed Irish migrant cohort reported experiencing some degree of emotional abuse and neglect before coming to England during their childhood in Ireland.
Cochrane and Bal (1989) reported higher rates of admission to psychiatric institutions for a range of conditions of Irish migrants in both 1971 and 1981. Contradicting an exclusive assimilation story, rates of admission among Irish migrants subsequently declined, consistent with a steady improvement in overall health of the Irish birth cohorts entering England. Raleigh and Balarajan (1992) reported that young (ages 20–29) Irish migrants in England and Wales from 1979 to 1983 had relative suicide rates of 174 for women and 267 for men (compared with 100 for the general population).
Basic Health Differences
Basic health disparities are displayed in Fig. 2, which plots fractions in poor health by birth cohort years for three groups: Irish born in Ireland now living in England, Irish born in Ireland still living in Ireland, and white English born and living in England. Compared with Irish born and living in Ireland or the white English population, a much higher proportion of Irish born in Ireland and living in England were in poor health. That the Irish born and living in Ireland are in better health is consistent with a long literature finding that people in Ireland have among the highest self-rated health scores in the world.
Figure 2 highlights considerable heterogeneity in health disparities by birth cohort. The poor health excess of Irish born in Ireland who live in England compared with the two other groups is particularly large for birth cohorts born between 1921 and 1960, but is smaller for pre-1921 birth cohort and the more recent post-1960 birth cohorts. Explaining this heterogeneity in the health penalty for the Irish-born who live in England is a major aim of this article.
Explaining mental health disparities is another central theme. Figure 3 displays by birth cohort year scores on the GHQ-12 psychological distress scale for men where most of the Irish-born living in England excess mental health illness lies. A higher score in GHQ indicates worse mental health. The sample of male Irish-born migrants to England displays substantially higher levels of psychological distress than the sample of Irish-born still living in Ireland and higher levels of psychological distress than the sample of white English-born living in England. These excess mental health problems are concentrated in the 1931–1950 birth cohorts.
To set the stage for analyses to follow, Table 2 displays a set of descriptive ordinary least squares (OLS) models describing birth cohort variation in health measures in the HSE for the populations of Irish-born and English-born both living in England. Table 2 documents birth cohort-specific health differences between these two populations. Our health measures include the probability that respondents are in good health (very good or good) or in bad health (bad or very bad), and whether respondents report three specific types of physical illnesses: cardiovascular disease, muscular disease, and respiratory disease. Besides a set of 10-year birth cohort dummy variables (with born pre-1921 as the excluded group) to test for patterns of health differences between the Irish-born and English-born groups both living in England, we interact three birth cohort groups with being Irish-born: born pre-1921, born between 1921 and 1960, and born between 1961 and 1980.
Because earlier-born cohorts are older in any calendar year, earlier-born cohorts are characterized as being in poorer health in all health measures. More important, our estimates indicate a statistically significant and substantial pattern of excess poor health for those born in Ireland between 1921 and 1960 who were living in England. With regard to specific health conditions, this group of Irish-born living in England report a statistically significantly higher probability of having a cardiovascular condition but neither muscular nor respiratory conditions, suggesting that stress and inflammation may play a role in this excess illness. When we estimate separate models for men and women, we find excess poor health in good, bad, and cardiovascular health for men and women, although effects are larger for men than for women.
To test health differences between the Irish-born who live in England and the Irish-born population who stayed in Ireland, Table 3 combines samples of Irish-born living in England in the HSE and LII surveys, which consist of Irish-born living in Ireland. Once again, among the Irish-born of the 1921–1960 birth cohorts, we find a statistically significant and substantial pattern of excess poor health and a greater degree of mental problems among those Irish-born who migrated to England compared with the Irish-born who stayed in Ireland. When we estimate separate models by gender, the excess poor subjectively reported health for the Irish-born 1921–1960 cohort was about the same for men and women, but the worse mental health for these birth cohorts was concentrated among men only.
The results in Tables 2 and 3, which document excess poor health among Irish migrants to England born between 1921 and 1960, cannot speak to reasons for this health penalty—namely, how much can be attributed to migration selection and how much to factors associated with assimilation into England. We address those issues in the next sections.
Models of Migrant Selection and Assimilation
We document next the amount of education selection of Irish migrants to England and develop a theoretical model of migrant selection that includes effects of pre-migration mental health issues. We test the implications of this model and show that, as predicted by the theory, pre-migration childhood abuse leads to a more negative selection in schooling of migrants. Finally, an empirical model of post-migration assimilation of Irish-born migrants living in England is tested.
Documenting Migrant Selection in Education
To assess migration selection, we need a pre-migration marker of human capital of migrants compared with others in their birth cohorts. We use educational attainment as our marker of pre-migration selection. Education is a well-documented determinant of adult health (Lleras Muney 2005), so education selection should translate into subsequent health selection. To ensure that our comparisons reflect conditions in Ireland and not in England, we consider Irish migrants who completed some schooling in England. The 1999 and 2004 HSEs contain data on year of migration. We assume that individuals received some education in England if their years of education variable exceeds the age at which they migrated minus 5, essentially assuming that individuals commence schooling at age 5. Employing this method on the 1999 and 2004 subsamples reveals that 23 % of Irish migrants to England received some education in England. In our analysis, we use a sample of Irish individuals educated exclusively in Ireland, so that education reflects pre-migration conditions in Ireland.
Table 4 shows education comparisons by birth cohort for white English-born and living in England, Irish-born living in England, and Irish-born living in Ireland. Data are presented for men and women separately in Table S4 of Online Resource 1. The Irish-born living in England are the reference group for these tests. The final three rows provide summary education data for three key broad periods of Irish migration to England: 1921–1950, 1951–1960, and 1961–1980.
Especially in birth cohorts between 1921 and 1950, Irish-born living in England are less educated than Irish-born living in Ireland, indicating on average, negative education selectivity among Irish-born migrants to England during this time period. During the 1950s, the education of Irish-born migrants to England and those Irish-born still living in Ireland are about the same. However, for the most recent birth cohorts in Table 4 (1961–1980), selectivity actually reversed with Irish-born migrants living in England now more educated than Irish-born who stayed at home.
When we turn to gender-specific education comparisons in Table S4 (Online Resource 1), we see that the general patterns just described are similar for men and women. However, the magnitude of the negative migration selection of Irish-born migrants to England among oldest birth cohorts and the positive migration selection of Irish-born migrants to England in the youngest birth cohorts is larger for men compared with women. At the peak of the negative education selection of Irish migrants during the 1940s, the size of selection effects is more than a year for men, twice the rate displayed among women.
Theoretical Model of Migration Selection
will be positive. 
measures a relative skill price premium between the home country and the receiving country. If wages of unskilled are relatively high in a receiving country, 
:
is less than 1; 
would indicate that migrants’ human capital is more relevant in the receiving country than in the home country (e.g., someone trained in computer technology with no industry in the home country).
Equation (5) provides us with several insights into selection mechanisms that influenced migration flows between Ireland and England in the twentieth century. Higher migration costs will require higher levels of human capital kiH to justify migration. Because health is an important form of human capital, this is the theoretical cost justification of the healthy migrant effect. Because monetary costs of migration from Ireland to England were very low historically (free migration with low transportation costs), this implies that Irish-born migrants to England will be less positively selected on skills and health than will Irish-born migrants to America, Canada, or Australia. Irish migrants to the United States had much more education than Irish migrants to England.
Unless migration costs are negative, any parameter lowering the size of the term in parentheses in Eq. (5) implies that migrants must be more skilled to justify migration because it will take more skilled migrants to offset migration costs. The lower the uniform wage premium (
) of receiving to sending country, the more skilled and healthy migrants relative to the average skill and health in the sending country will be. As income differences between the two countries become smaller, this model predicts that Irish migrants will become more positively selected. The greater the wage skill premium (
) in the receiving country relative to the sending country, the more negative migration selection. Although migration rates will increase if skill transferability is high, migrants who come will be less skilled and less healthy.
We now consider how well our model predicts twentieth-century patterns of Irish emigration to England. Historical and geographical links between Ireland and England hold vital explanatory power. Our context is low relative costs (dis-utilities) of migration (
) and high levels of human capital and skill transferability (
). Physical transport costs between the two states are low because of geographical proximity institutional and cultural similarities, including a common language. High skill transferability between the two states is facilitated by the ease with which an individual can enter the labor market: no visa or permit is required. Thus, we see less selectivity among the Irish migrating to England, resulting in lower average migrant skill levels.
There was a significant improvement in the profile of Irish migrant cohorts in the twentieth century. Without any dramatic changes in migration costs or skill transferability, our model predicts that this change was driven by an increase in Irish wage rates, 
. During the latter part of the twentieth century, Ireland experienced rapid economic growth relative to England. A buoyant Irish labor market halted net outflows of migrants, and those who chose to migrate were more likely to have more schooling. Our model provides a simple explanation of individual heterogeneous selection mechanisms characterizing Anglo-Irish migration in the twentieth century.
Psychic costs associated with location are usually not emphasized in economic migration models and are our main focus here. Economic models of migration generally suggest that psychic costs of migration are positive and associated with destination location because the migrant is moving to a less familiar environment away from family and friends. For some potential migrants, there may be a high psychic cost of staying in the home-country environment. Those who suffered mental distress as children or young adults associated with their family or school life in Ireland may feel less mental distress from migrating to a country that is not associated with these problems. This situation characterizes a meaningful fraction of Irish-born migrants to England. If so, the costs of migration to England could actually be negative. Negative psychic costs with place of origin imply, according to Eq. (5), that increasingly less-skilled and less-healthy Irish migrants would migrate to England.
Empirical Model of Migration
Our outcome Edmci is the education of an Irish migrant mi to England from birth cohort c. First and foremost, this is influenced by mean gender-specific education of the Irish birth cohort Edci, of which each respondent is a member. The other variables on the RHS of Eq. (6) reflect possible reasons for education selection of Irish migrants to England, as discussed earlier. Our main economic migrant selection drivers are real income differences (wIt – wEt) and unemployment rates differences (UIt – UEt) between Ireland and England at time of migration. We also include an index of the extent of violence (Violencet), measured by number of deaths resulting from Irish political violence in Britain two years before and two years after year of migration. Our tests rely on the notion that Irish-born populations living in England experience more stress caused by political violence in Britain than Irish-born populations living in Ireland. Finally, we measure the fraction (expressed as a percentage) of abuse (Child Abusec) during childhood years experienced for each birth cohort obtained by combining abuse numbers for stayers and migrants in TILDA. To mitigate against measurement error, we take a three-year cohort moving average of our cohort abuse index.
Table 5 displays results of our education selection model. Not surprisingly, education levels of Irish birth cohorts are highly predictive of education of the Irish-born living in England from the same cohort, suggesting that as education levels improved in Ireland throughout the latter half of the twentieth century, so too did education of Irish-born migrants migrating to England. Because the estimated coefficient is not statistically different from 1, this implies that any migrant education selection is captured through other variables.
Although differences in unemployment rates between the two countries at the time of migration do not influence education selectivity of migrants to England, differences in log of real wages at the time of migration has a statistically significant and quantitatively meaningful effect on education of Irish migrants in England. This result is consistent with our selection model because as wage differences between the two countries converge, migrants should be more skilled relative to the sending country mean. In a separate analysis by gender, impacts of real wage differences of education of Irish migrants are more than twice as high for men than for women. Greater selectively for men compared with women is expected given that, on average, male migration is more focused on labor market reasons than is female migration.
Our measure of political violence at time of migration does not influence education selectivity of Irish migrants to England. Finally, we estimate that an increase in the fraction of the birth cohort who suffered from childhood abuse does lead to a more negative education selection of Irish migrants to England. This result is consistent with the implications of our theoretical model that abuse during childhood makes migrants more negatively selected.
These selection effects are strong, given the rapid convergence of Irish real wages to English real wages over the time period. Real wages at the start of the period is log −0.5 lower in Ireland and converges fully over the time period. This alone would predict a 1.25-year contraction of the difference in education relative to an unadjusted difference of 1.5 years, controlling for changes in base education in Ireland.
Empirical Assimilation Results
The possibility that Irish migrants in England have poorer physical and mental health because of a stressful process of assimilating into a foreign country must be considered. Migration is often viewed as stressful, and life in a different land can add to that stress. Although transportation costs were low so that periodic return trips were not difficult, political tensions were high during much of the twentieth century, particularly during the 1970s and 1980s, when the conflict surrounding Northern Ireland was at a peak. We find no evidence that this political violence influenced selection of Irish migrants to England, but a cumulative impact of living in England during those years may have carried with it a physical and mental health cost for Irish migrants.
Our data permit indirect tests of the assimilation hypothesis as a reason for the negative health disadvantage of Irish migrants to England. The HSE asks migrants in ethnic booster samples (that increase the size of the immigrant ethnic samples) the age at which they migrated to England. If exposure to an English environment leads to stress and a greater propensity to adopt unhealthy behavior patterns, this should intensify with more years of exposure to this environment. Time since immigration to England is an indicator of “exposure” to an English environment.
We basically test here whether duration of exposure to England has large implications for health conditions and health behaviors of the Irish-born migrant group. To examine this, we include time since migration in models of good health, GHQ score, the probability of marriage, and smoking and drinking behavior.9 These models include our marker for migrant selection effects—namely, Irish migrant respondents’ years of schooling relative to birth cohort means. Results are displayed in Table 6.
Controlling for migration selection effects and birth cohort dummy variables, Table 6 shows that length of time spent in England does not predict any outcomes. This result may be due to relatively small sizes of the Irish in England available in the HSE, but the size of estimated effects look small as well. We also include a variable measuring age of departure from Ireland, but this variable is statistically insignificant.
Tables 7 and 8 show a closer look at birth cohort differences in relevant behavioral outcomes of Irish migrants compared with white English and the Irish-born living in Ireland. Separate data by gender are in Online Resource 1. Irish migrants are less likely to be currently married, are more likely to be divorced or separated, and are much more likely (especially in older cohorts) to be widowed. High rates of widowhood are a vivid indicator of poor health of Irish migrants to England, while high rates of marital instability may reflect psychological problems migrants faced—many with origins before migration.
Irish-born migrants living in England are characterized by the highest rates of having ever smoked, but their current smoking behavior is more in line with that of the other groups. This indicates high rates of smoking cessation among the Irish in England, most likely for health reasons. Smoking among the Irish-born living in England was largely initiated prior to migration. Among Irish migrants in the HSE, 68 % started smoking before their migration to England, and more than one-half were smoking at least five years before migration to England. Thus, smoking among Irish migrants appears largely to reflect pre-migration conditions. Finally, among those who drink, Irish migrants to England are heavier drinkers than the white English, but a much larger fraction of Irish migrants have stopped consuming alcohol, which is another likely indicator of problem drinking. (Note that the LII does not include drinking variables.) The data in Tables 7 and 8 support the view that key bad health behaviors of Irish migrants had their origin in Ireland before migration.
Conclusions
We developed and tested a model of immigrant selection that explicitly incorporates high physic costs associated with staying in the sending country. The implication of these high physic costs indicates not only a greater likelihood of migration but also more negative selection of migrants on education and health. This model is tested using data on Irish-born migrants to England in the twentieth century.
The existing literature clearly documents that Irish migrants to England were in very poor physical and mental health compared with the English-born and compared with the Irish-born who remained in Ireland. What is in dispute is why. A common view is that conditions in England were very stressful for these Irish-born migrants, especially in light of sporadic political violence between the two countries. Consistent with our theoretical model, we argue instead that factors leading to poor mental health in Ireland prior to migration increased the propensity to migrate and that these Irish-born migrants to England were less educated and healthy than their counterparts remaining in Ireland.
We present evidence that many children born in Ireland suffered high levels of childhood physical and sexual abuse and that rates of childhood abuse were far higher among Irish migrants than those staying in Ireland. We also show that for cohorts born before 1950, Irish migrants to England were negatively selected on education. Our empirical model indicates that high levels of childhood abuse in a birth cohort is associated with less-educated Irish-born people migrating to England, even when we exclude those Irish-born migrants who came to England as children. Finally, many poor health behaviors, such as heavy smoking, have their origins pre-migration from Ireland to England.
We also provide a test of a model of effects of assimilation into England by Irish-born migrants on a set of health outcomes and poor health behaviors. We find little association of time spent in England on these health outcomes or behaviors. One possibility is that the Irish of this generation not only were unharmed by going to England but were possibly helped. National Health Service (NAS) mental and physical health services were of far higher quality than those in Ireland during this period and were used highly by the Irish-born migrant group. Although we do not dismiss the possibility of some effects resulting from assimilation factors, explanations that rely exclusively on a troubled assimilation in England may be missing a significant part of the story.
Although we focused on the Irish/England case, bringing in psychic costs associated with staying in the sending country may have far greater applicability. Many potential migrants suffered from persecution in the sending country because of their ethnicity or other attributes. Refugees and asylees, an important subcategory of migrants around the world, are not primarily economic migrants. The extension of migrant selection to include psychic costs associated with sending countries applies to those cases as well.
Acknowledgments
Thanks to Christian Danne and Iva Maclennan for excellent research assistance. Delaney acknowledges generous support from the Fulbright Commission and the Center for Health and Wellbeing, Woodrow Wilson School, Princeton University. RAND research was funded from a grant by the National Institute on Aging.
Notes
Our assumptions are spelled out in Table S1 of Online Resource 1. We correct for participants stating “no education,” by using the question, “How old were you when you completed this?” Table S2 (Online Resource 1) displays those in the “no education” category who have some formal schooling.
In LII, respondents were asked, “In general, how good would you say your health is? Would you say it is (1 very good, 2 good, 3 fair, 4 bad, 5 very bad)?” For HSE, respondents were asked, “How is your health in general? Would you say it was (1 very good, 2 good, 3 fair, 4 bad, 5 very bad)?” Questions used in both surveys are highly comparable on outcome measures.
There is a long history of migration of Irish people to Scottish cities, particularly Glasgow, so the Irish in Scotland are an interesting subject of study. We do not have access to data that allow a detailed health comparison as in with the HSE. The 2001 census figures show 21,774 people from the Republic of Ireland living in Scotland.
Data are from the Irish and UK censuses. The number of Irish people living in England declined from 2001 to 2010. The 2010 population figures list 398,000 Irish-born people living in the United Kingdom.
Migration of those born in the Republic of Ireland to Northern Ireland was of a much smaller scale than migration to England. In the 2001 census, migrants to Northern Ireland represented 2.3 % of the total Northern Ireland population. In total, there were 39,051 such individuals living in the North from a total population of 1,685,267 in the North.
TILDA respondents were asked four questions as part of a self-completion “drop-off” questionnaire: “Before you were 18 years old, were you ever physically abused by either of your parents?; . . . were you ever physically abused by anyone other than your parents?; . . . were you ever sexually abused by either of your parents?; . . . were you ever sexually abused by anyone other than your parents?” Parental and non-parental sexual abuse were combined into one sexual abuse variable, “sexual abuse”; similarly, parental and nonparental physical abuse were combined into one variable, “physical abuse.” A “total abuse” variable was constructed that takes a value of 1 if any abuse was reported across the four categories.
See documentation on the Ryan Commission website (http://www.childabusecommission.ie/).
See volume 1, chapter 3 of the Commission Report for a detailed description of numbers and reasons for admission.
Marriage and smoking are measured in the HSE and LII, so we construct comparable measures of proportion currently married, proportion divorced or separated, and proportion widowed across each cohort. For smoking, we construct two variables: current smoker and ever smoked. Our measure of alcohol volume is constructed using HSE data by combining average units consumed and frequency across several different alcohol classes. Respondents who answered that they never drink were asked, “Have you always been a non-drinker, or did you stop drinking for some reason? (1 Always a non-drinker, 2 Used to drink but stopped).” We use this to construct the variable Fraction who stopped drinking. There are no comparable drinking variables in the LII.
