Faculty of Economics and Business
The impact of immigration on the life satisfaction of natives
The case of the United Kingdom
AbstractUsing individual data from the UKHLS and regional characteristics for the period 2009-2018 I explore the association between self-reported life satisfaction and regional immigration across the twelve government office regions (GOR) in the United Kingdom, finding and overall positive and statistically significant relationship in the different specifications, although the results are small in magnitude. Furthermore, I address the channels through which immigration may affect the well-being of native individuals, concluding that the labour market and health are relevant in order to explain this positive association.
Keywords: Subjective well-being, life satisfaction, immigration, diversity, foreign-born, native.
Supervisor: Dr. Padma Rao Sahib
Acknowledgements
Hereby I would like to express my utmost gratitude to my thesis supervisor Professor Padma Rao Sahib, for her guidance and advice has been key in the development of this thesis. Moreover I would like to send my appreciation to my family and friends who always stand by my side through sick and thin for as long as I can remember.
1. Introduction
Immigration has been a widespread topic in the United Kingdom’s political and public sphere in the recent decades, especially prominent were the headlines devoted to immigrant issues right before the Brexit referendum taking place back in June 2016. At the time I am writing these lines, society is facing a global pandemic, which will definitely change the fate of millions of people worldwide. At the frontline of this pandemic are nurses, doctors, care workers and healthcare staff members fighting against covid-19 tirelessly.
UK is no different than other any other country in the world and is currently coping with this pandemic, with National Health Service (NHS) workers battling at the heart of this unprecedented global health crisis. The virus does not differentiate between social classes or even political roles. Indeed, Prime Minister Boris Johnson had to be treated of this illness. In fact, his first words in his very first public appearance after being discharged, as noted by William Booth from the Washington Post (2020), were devoted to praising the NHS health workers who stood by his side during the time he spent at St Thomas Hospital. Those health workers were “Jenny from New Zealand” and “Luis from Portugal,” two of the thousands of immigrants who serve in the NHS. Precisely, as the Office for National Statistics, 2019 (ONS) showcase, from around 1.9 million people who were employed in the healthcare system in the United Kingdom in 2018, non-British nationals added up to roughly 227.000 workers or 12% of the total workforce in the healthcare system.
Despite immigrants have been in the centre of public opinion and turmoil for quite a long time in the United Kingdom (not always fairly treated), now some citizens might start seeing the other side of the coin in these trying times for all of us. Thus, some questions regarding the opinion or feeling of native citizens towards immigration and to what extent this can somehow affect their daily lives may arise.
In this study, I approach the regional effects of immigration on life satisfaction levels of native individuals in the United Kindgom. Hence, answering the following research question;
“How does regional immigration levels in the United Kingdom affect the life satisfaction of natives?”
their life when thinking about it. The subjective well-being aspect assessed in this study is the latter, namely life satisfaction, how do individuals evaluate their life as a whole.
In this study, by native I mean UK residents already born in the country, whereas by immigrant I refer to foreign-born UK resident. Precisely, the aim of the study is to address those effects at the individual level, looking into the 12 Government Office Regions (GOR) across the 2009-2018 time period. In order to gauge those effects it is important to consider what is the immigration share in each regional area (GOR), what role immigrant diversity is playing and to control for both individual and regional variables that could be directly correlated with life satisfaction.
Indeed, international migration is growing bigger worldwide, with OECD countries receiving about 5.3 million new permanent immigrants in 2018, which represents 2% increase from 2017 (OECD 2019). Hence, economists have investigated what the consequences are to both sending and hosting countries, at individual or at country level, of immigration, especially when large inflows take place.
As for the United Kingdom, the foreign-born population in 2018 amounts to roughly 9,2 million people, representing 14% of the total population (ONS, 2019), a peak rate of foreign-born population, comparatively to 2008, as the foreign-born population amounted to 6.7 million people, 11% out of the total. Going further back in time, in 2004, when the United Kingdom experienced a large influx of new immigrants due to the newly agreed EU enlargement of 2004, foreign-born individuals represented 8,9% of the entire population. Hence, one should not disregard those numbers, as they are meaningful and becoming increasingly important in magnitude.
The economic literature on immigration has long been focusing mainly on the impacts immigrants have on natives’ objective welfare, what can be economically measured, such as wages and employment, usually leading to small overall effects in the labour market, either positive or negative. In addition, the arrival of migrants could, for example, decrease the availability of unpriced public goods at the destination, and a degree of cultural homogeneity that may be valued by non-migrants (Clemens 2011). Hence, there are plenty of channels through which immigrant may affect natives’ perceptions of welfare.
However, the approach has been shifting recently and more emphasis is being placed on the subjective well-being of individuals, what people value and how they act in real life, through surveys, people asses their self-reported well being that range from happiness to anxiety or life satisfaction as a whole.
output levels in a given economy. On these grounds, the Sen, Fitoussi and Stiglitz report (2009), strengthened the framework on working with subjective well-being, shifting the emphasis away from measuring people’s well being by means of economic production. Hence, on measuring the quality of life, the notion of subjective well-being views individuals as the best judges of their own conditions.
Regarding the consequences of international migration, Longhi (2014) considers that in many countries local communities are becoming more diverse in terms of country of birth, ethnicity and religion. Therefore, increases in diversity might have consequences for the well being of residents.
This thesis follows the approach of Akay et al. (2014) as a reference, the authors, working with data from the German Socio-Economic Panel (GSOEP) between 1998-2009, examine to what extent the immigration share in a specific local area has an impact on the subjective well-being of the native population.
2. Literature review
2.1 Objective measures of welfare
As previously mentioned, a big part of the economic literature has focused on the impacts immigrants have on the objectives measures of welfare of the native population. Those welfare dimensions include employment, income, wages or the fiscal net balance of immigrants in the host country (the country receiving the migration inflow).
Many studies centred the attention on the United States, because of its large migrant inflows entering the country throughout time.
Borjas (1994, 2003) and Borjas and Katz (2007) found a negative correlation between the presence of immigrants in a local labour market and the earnings of natives. Therefore, Borjas (1994) claimed that immigration is one possible cause affecting the decline in native workers’ earnings between 1980 and 2000.
immigrants have had virtually no effect on wages or unemployment rates of less-skilled workers.
Usually, this skill distribution is not the norm in the labour market, as migrants are usually more educated than native-born in many destination countries. Cross-country evidence points out that migrant stocks are on average more skilled than the native population (World Bank 2018). Indeed, Docquier et al. (2014) on addressing the labour market in OECD countries, also notice the fact that immigrants are more educated than natives. Thus immigration tends to increase the relative size of skilled people in the host countries, leading to increases in the relative wage of low-skilled native-born workers.
The framework provided by Clemens (2011) also helps to understand the rough idea behind migration dynamics. He provides an approach to analyse the gains of greater labour mobility (assuming completely free migration). Overall, increases in labour in the receiving country and a decrease in supply in the sending country, would push down wages in the receiving country and increase wages in the sending country. Furthermore, Borjas (1999) specifies that natives in the host country only benefit from immigration as long as immigrants and natives differ in their productive endowments (characteristics). Natives whose productive endowments are complementary to those of immigrant’s gain, while natives whose endowments are competing with those of immigrants lose.
In addition, Borjas (2003) and Card (2001) assesses the relative supplies of different skill groups on the structure of wages. Both authors consider different elements shaping workers’ skill profile; on the one hand Borjas considers schooling alongside labour market experience, whereas Card solely considers working experience as an element defining the skill profile of a worker. Thus the results are quite different, while Borjas (2003) finds immigration to have a wage impact in lower educated groups, the results from Card (2001) do not depict an overall critical scenario from immigration shocks, even for workers in the bottom of the skill distribution, Card finds relatively modest employment effects of recent immigration flows.
However, the immigration traits between the United States and the United Kingdom are not the same as migration flows into the United Kingdom are especially high skilled (Manacorda et al. 2012), contrary to immigrant characteristics in the United States. Thus, immigrants are more similar in their education and skill distribution relative to the native population (Dustmann et al. 2005). Overall, in the UK, immigration has had little observable effect on the wages of natives (Manacorda et al. 2012; Dustmann et al. 2005).
intermediate education group, for whom the effects consistently suggests a negative effect on their labour market activity and probability of working. However it seems to be offset by increasing employment of the more educated workforce (despite is not statistically different from zero). In addition, Dutsmann et al. (2008) found some negative effect of immigration among natives at the bottom of the wage distribution and a positive effect among those at the top.
Regarding immigrants’ net fiscal position in the host country, the estimates vary, although in most countries it tends to be very small in terms of GDP and is around zero on average across OECD countries (OECD 2013). Generally, migrants are typically younger than natives, representing a lower burden for the fiscal system, and the contributions from short-term migrants may not entitle any social benefits. Moreover, they increase the sustainability of the pension system in their destination country due to the increase in fertility generally associated with migration (Boeri 2010).
However, the picture change across countries and authors; Borjas (1994, 1995) finds that in the United States the new immigrants may have had an adverse fiscal impact because recent waves participate in welfare programs more intensively than earlier waves. Conversely, the OECD (International Migration Outlook 2013) finds that if immigrants have indeed a less favourable net fiscal position, this is driven almost entirely by the fact that immigrant households contribute less in terms of taxes and social security contributions and not by a higher dependency on benefits. On the other hand, Boeri (2010) states that immigrants receive more social transfers when educational attainment and family characteristics are taken into account, especially in rich countries. Nevertheless the overall effect tends to be small.
Furthermore, Dustmann and Frattini (2014) found that for the UK, immigrants from the European Economic Area (EEA) make a positive fiscal contribution, while non-EEA immigrants – not dissimilar to natives in terms of skills and attitudes– have made a negative contribution. Similarly, Dustmann et al. (2010), on assessing the fiscal impact of A8 immigrants – migrants from the Central and Eastern European countries that joined the European Union in in 2004 – found that in each fiscal year since the EU enlargement, new immigrants made a positive contribution to the public finances. In short, the impact immigration might have on objective measure of welfare such as employment, wages or on the government budget tends to be really small. However, the labour markets effects of immigration will differ across countries depending on the structure of the receiving economy as well as the skill set of immigrants relative to the resident population.
2.2 Empirical literature summary on objective measures of welfare Research
paper Scope of the analysis Main finding
Borjas (1994) United States 1980s
Negative fiscal impact of immigration. More participation of recent waves in welfare programs.
Immigration correlated with low-skilled natives workers’ decline in wages.
Borjas (1995) United States 1970-1990
Adverse fiscal effect of immigrants as they have higher dependency on public assistance.
Borjas et al. (1996)
United States 1980-1990
Immigration contributed especially to the decline in the relative earnings of high school dropouts and modestly reduced the earnings of high-school workers relative to college workers. Borjas (2003) United States
1980-2000
Immigration affected the wages of native workers, especially for high school dropouts. Borjas. &
Katz (2007)
United States 1900-2000
Large Mexican migration inflows have an adverse impact on the earnings of less-educated native workers while improves the earnings for college graduates.
Card (1990) United States 1990s
The inflow of newly arrived Cuban workers in Miami had essentially no effect on the wages or employment outcomes of natives in the Miami labour market.
Card (2001) United States 1980s Immigrants’ inflows reduced wages and employment rates for low-skilled natives. Carrasco et
al (2008) Spain 1995-2000 Immigration had no significant negative effect on either employment rates or wages of native workers.
D’Amuni et al. (2010)
West Germany 1990s
The substantial immigration in West Germany during the 1990s had very little adverse effect on natives’ wages and employment levels. Dustmann et
al. (2005)
United Kingdom 1983-2000
Immigration had only a significant impact for intermediate educated people, for whom it decreased their employment rates.
Dustmann et al. (2010)
United Kingdom 2004-2008
Immigrants from the EU enlargement in 2004 have had a positive impact to the public finances in each assessed year.
Dustmann & Frattini (2014)
United Kingdom
2.3. Immigration and Subjective Well-Being
Once some insights are provided on what may be the economic implications for natives’ objective welfare due to immigration, it is time to shift the attention towards the linkages between subjective well-being and immigration, how the economic literature have been addressing subjective well-being and what are the research outputs on the linkages between immigration and life satisfaction are so far.
As briefly mentioned in the introduction, academic studies are attempting to asses economic performance differently, not merely using GDP as a measure of economic performance and social progress but also taking into account individual’s own perceptions and feelings about their own lives. The concept of well-being is intended to capture what is ultimately good for the individual (Angner 2010), hence self-reported or subjective well-being is a proxy to gauge an individual’s utility, as quality of life is a broader concept than economic production and living standards. Thus it includes the full range of factors that influence what we value in living, reaching beyond its material side (Sitglitz et al. 2009).
What definition better suits subjective well-being? There is not a tailor made definition but both Angner (2010) and Brülde (2007) consider four conceptions of subjective well-being; 1) Cognitive views, according to which, subjective well-being is a cognitive state or attitude towards one’s life as a whole, 2) Hedonistic views; Subjective well-being is understood in terms of the presence of pleasure and the absence of pain. 3) Mood or emotions views, where subjective well-being is a certain kind of mood or emotional state and 4) Composite views, (hybrid view) according to which happiness is a complex mental state consisting of both of an affective and cognitive view. The affective view gathers both the hedonistic and mood views. As Diener (1984) states, the economic literature regarding subjective well-being is concerned on how and why people experience their lives in positive ways, what are the drivers behind a self-assessment of people’s own lives. This definition of subjective well-being has come to be labelled life satisfaction and relies on the standards of the respondent to determine what is good in life. Now the question is on how to approach life satisfaction.
Life satisfaction is measured using surveys of individuals, thus each respondent defines happiness, life satisfaction or any other dimension of subjective well-being in his or her own. Questions typically are: How satisfied are you with your life? And respondents usually give an answer based on a 0-10 scale with 0 being utterly dissatisfied with life as a whole and 10 being completely satisfied.
Akay et al. (2014) focus on the effect of immigration on the welfare of natives in Germany, using life satisfaction as the self-reported well-being measure. While addressing immigration, their definition of immigrants is based on citizenship and they narrow the scope of the research on individuals aged between 16 and 64. Overall, the authors conclude that there is an economically and statistically significant positive impact of immigration on natives’ subjective well-being. Natives experience welfare gains as immigration in the local labour market increases. This positive impact is mainly driven through the effect on local labour markets, ethnic diversity and immigrants’ assimilation. Precisely, Akay et al. (2017), investigate how ethnic diversity in Germany influences the subjective well-being of natives. The results from their fixed effects specification suggest that ethnic diversity positively affects the well-being of natives. Moreover, the effect increases if more weight is given to immigrant groups culturally and economically closer to Germany. Some previous studies on ethnic diversity also find a positive impact on the labour market; Suedekum et al. (2014) find a strong positive effect of immigrant diversity on both the regional wage and the employment rates of native workers, this could be driven through skill complementarities in certain production processes (Alesina et al. 2013).
On assessing both the immigration share in a local market and the ethnic diversity, Alesina et al. (2013) define both the intensive and extensive margin. Being the former the composition of the immigrant share (immigrants by country of origin) and the latter the extensive margin, which is the mere share of immigration, without consideration of origin (compositional) effects. The authors find birthplace diversity to be positively correlated to productivity and economic growth, being the effect stronger for skilled immigrants in richer, more productive countries. Similarly, Peri (2012), found positive effects of diversity on the productivity of the US, attributing to unskilled immigrants the promotion of efficient task-specialization on what he terms unskilled-efficient technologies. Interestingly, Hong and Page (2001) show theoretically that a group of cognitively diverse but skill-limited workers can outperform a homogeneous group of highly skilled workers.
Conversely to the aforementioned studies, Longhi (2014) on assessing the impact of ethnic diversity in the UK, finds that diversity has a negative impact on the well-being of UK residents although it differs across groups. On the one hand, white British people living in more diverse areas have lower levels of life satisfaction on average than those living in areas where diversity is lower. On the other hand, the average level of life satisfaction of non-white British people and foreign-born does not seem to be affected by diversity levels. Likewise, Knies et al. (2016) find that diversity is not associated with reduced life satisfaction for Black Africans and UK born Indians and Pakistanis, they even gain utility from being in more ethnically diverse areas.
(UKHLS), finding that immigration has no significant impact on the well-being of natives in the UK. The author, on assessing the time period 2004 – 2016, attempts to identify the impacts of the EU enlargements of 2004 and 2007 in the subjective well-being of the UK natives. Similarly, Howley et al. (2018) study the impact of non-UK born across neighbourhoods in the UK, finding no overall relationship between inflows of foreign-born individuals and subjective well-being levels. However, both Howley et al. (2018) and Ivlevs and Velioziotis (2018) studies find out that the groups experiencing the steepest decline in their own subjective well-being are older individuals, unemployed and those in the lowest income quartile.
The aforementioned studies are solely focused in one country, either Germany or the UK, however, Betz and Simpson (2013) measure the well-being implications of migration in a cross-country study, using the European Social Survey and OECD data, concluding that recent immigrant flows have a nonlinear, yet overall positive impact on the well-being of natives.
Overall, previous literature finds a positive association between subjective well-being measures and immigration and or ethnic diversity, albeit the correlation remains small. This thesis deviates from previous studies addressing subjective well-being and immigration in the United Kingdom (Ivlevs and Velioziotis. 2018; Howley et al. 2018; Papageorgiou 2018), using the most recent available data from the UKHLS from the time period 2009-2018, looking at regional rather than local level and adding the component of immigrant diversity besides the immigrant share.
3. Data and methodology
To approach how immigrant shares in a specific region affect the subjective well-being of the native population, it is of the utmost importance to have a rich dataset of individual characteristics, in addition to regional data for the twelve GORs used in this study.
The UK Household Longitudinal Study (UKLHS), which is a longitudinal survey of the members of roughly 40.000 households for each wave in the United Kingdom, provides information at both individual and household level, collecting both objective and subjective indicators assessed in this study. Households, recruited at the first round of data collection are visited each year to collect information on changes to their household and individual circumstances. This enables of tracking individuals over time.
consecutive time periods (years). Individuals found at selected households in the first wave were designated as Original Sample Members (OMS). The aim of the dataset is to maintain OMS respondents as part of the sample as long as they live in the UK. Some variables, which are time invariant, were only collected in wave 1 and collected in subsequent waves only when new individuals were interviewed. Thus, data on these variables from subsequent waves is added in order to avoid missing observations. As previously mentioned the geographical scope of the study is at the regional level, the so-called Government Office Regions (GORs), which represent the NUTS 1 Eurostat’s nomenclature for the division of UK’s territories. Furthermore, every individual is matched with its geographical identifier, representing where the individual is currently living at the time the interview was carried out.
The GORs used in this analysis are the following; North East, North West, Yorkshire and the Humber, East Midlands, West Midlands, East of England, London, South East, South West, Wales, Scotland and Northern Ireland.
In order to better understand the current association between immigration and life satisfaction, I used the most recent data from the UKHLS covering the period 2009-2018 (waves 1-9). Furthermore, working with an ample set of data increases the data points, leading to more conclusive outcomes. Previously data can be found at the former UKHLS database under the name of BHPS (British Household Panel Survey). Both UKHLS and BHPS have been extensively used in the academic literature linking subjective well-being and immigration in the United Kingdom (Howley et al. 2018; Papageorgiou 2018; Ivlevs and Velioziotis 2018).
The way respondents from the UKHLS are asked about subjective well-being questions does not differ from other databases such as the GSOEP (German Socio-Economic Panel), used in the aforementioned studies from Akay et al. (2014, 2017). Precisely the respondent is confronted with the following statement:
“On a scale of 1 to 7 where 1 means ‘Completely dissatisfied and 7 means ‘Completely satisfied’, how dissatisfied or satisfied are you with your life overall” I use the response to this variable as the dependent variable in the model.
Besides data on individual characteristics, which help control for other characteristics that may affect life satisfaction apart from immigration levels alone, regional characteristics are also included in the model. Especially relevant for this study is the immigration share in each area or the country diversity of immigrants. Furthermore, regional characteristics such as GDP or unemployment are also included in order to help control for regional differences.
I also employed data from the Office for National Statistics (ONS), which gathers vast data at regional and local level. Information available from the ONS brings data on numbers of foreign-born individuals living in each GOR, alongside population estimates across regions, which allows constructing immigrant shares in each GOR. Information from the ONS is available on an annual basis. The ONS bases these estimates on the UK Annual Population Survey, which is the largest survey in the UK consisting of 320.000 respondents.
To be consistent with the definition of immigration, I used data on foreign-born residents of the United Kingdom to construct the immigrant share in each region. Every time I refer to native or immigrant individuals I am considering place of birth, not nationality.
3.1. Immigrant Diversity index
The UK has been for a long time a landmark of cultural diversity, especially if one looks into one of the most cosmopolitan cities in the world, London.
As Khan (2019) points out in his article in The Guardian (07/2019), ethnic diversity is embedded in Britons daily lives. For example, nearly 40 percent of what are considered amongst the top cultural figures comes from either a migrant or minority ethnic background. Similarly, several sports national teams are represented by a large share of foreign-born players. Khan (2019) cites the example of the England cricket squad, where almost half of the players are foreign-born or have some kind of ethnic minority background.
The academic literature has already provided statements both in favour and against cultural diversity, usually focusing on the relationship between racial diversity and social cohesion (Letki 2008; Alesina and Ferrara 2000), which can be ultimately linked to subjective well-being of individuals. On these grounds, Alesina and Ferrara (2000) find that communities with high levels of racial and cultural diversity have lower levels of interpersonal trust and formal and informal networks. Similarly, Letki (2008) on assessing the impact of racial diversity on social cohesion in the context of neighbourhood in Great Britain, states that racial diversity does have a direct negative impact on the perceptions of, and trust in, fellow neighbours.
Alternatively, Ottavino and Peri (2005), on measuring ethnic diversity as the variety of native languages spoken by city residents, conclude that diversity of cultures may have indeed a positive effect as it implies diversity of production skills. The authors find that wages and employment density of US-born workers were systematically higher, in cities with richer linguistic diversity.
Although the measure constructed in this study is about country of birth rather than ethnicities, it is a proxy for the immigrant diversity in each GOR.
Besides the immigrant share in each GOR, I approached immigrant heterogeneity in each region, to see to what extent diversity within a geographical area may play a role on individual’s subjective well-being, in line with the work from Akay et al. (2017). I constructed an index based on country-group of birth, used as a proxy to gauge how heterogeneous the immigrant population in an area could be. In order to do so I follow the approach of previous studies (Akay et al. 2017; Alesina et al. 2003, 2016) linking subjective well-being and immigration, as some of them also consider some sort of ethnic index, either reflecting nationality, ethnic origin or ethnic self-identification. Alesina et al. (2013) define both the intensive and extensive margin. Where the former is the immigrant diversity (immigrants by country of origin) and the latter the extensive margin, which is the share of immigration.
Here, I used nationality groups, as data from the Annual Population Survey (APS) allows doing so. Each foreign-born individual is linked to his/her country reference group based on the country of birth. Similarly, Akay et al. (2017) also consider indices of ethnic diversity constructed using different aggregations of nationality groups. Note that the immigrant diversity index in each region is based on the Herfindahl-Hirschman index, often used in related studies assessing the role of ethnicity or cultural diversity in a specific region (Longhi 2014; Akay et al. 2017; Alesina et al. 2003, 2016; Sturgis et al. 2014).
The Herfindahl-Hirschman index is a statistical measure of concentration used in a variety of contexts. Here, I try to see to what extent do regions differ in terms of diversity of its foreign-born population. The Annual Population Survey (APS) provides information on foreign-born in each region and year by country groups, defined geographically. The baseline for the measure is the size of each country-group relative to the total immigrant population. The Immigrant Diversity is measured as follows:
𝐼𝐷𝑡𝑟 = 1 − !!!" !!!
!
Where 𝐼𝐷𝑡𝑟 represents the immigrant diversity in a region r and year t. The number of immigrants of country group c in region r is depicted as 𝑖𝑚!" and 𝑖𝑚! is the total number of immigrants in each region r.
The index (equation 1) varies from 0, depicting a homogeneous immigrant group, to 1 (perfect heterogeneity). Thus, it approaches 1 when an immigrant population in the region is composed by a large number of groups of relatively equal size and different origins (Akay et al. 2017).
Had I calculated the immigrant diversity index without consideration of country groupings, and calculated it at the individual country level, the numerator would be representing only one country, when divided by the denominator – the total immigrant share in each GOR – the differences across GORs are almost imperceptible and the result will be insignificant.
By means of comparison, Northern Ireland ranks as the region with the lowest immigrant diversity while London subsequently ranks as one of the regions with more immigrant diversity. This heterogeneity (or lack of it) is represented in the Immigrant Diversity (ID) index constructed, where Northern Ireland has an average ID of 0,53, whereas London index goes up to 0,82 over the 2009-2018 period. When taking a closer look we can notice that Northern Ireland immigration is primarily sourced from the EU, representing 65,75% of the foreign-born population in 2018. On the other hand, EU-born share in London is only 31,61% (2018), and other country groups have a significant weight in the foreign-born share, alongside Europeans, thus resulting in a more heterogeneous overall picture. Northern Ireland can be regarded as the outlier, as the overall ethnic diversity across the UK is much closer to 1.
Figure 2: Immigrant Diversity index
Immigrant Diversity index quartiles across GORs for the period 2009-2018. Authors calculations based on data on immigration from the Annual Population Survey.
As can be seen from Figure 2, differences on the diversity index values remain small across the GOR areas, albeit Northern Ireland acts as an outlier and thus assessed in its own due to its relatively low immigrant diversity, mainly driven by the fact that migration towards Northern Ireland is strongly European based rather than the more heterogeneous migration inflows across the United Kingdom GORs. Not surprisingly, London and the surrounding South East GOR, have the highest diversity of its foreign-born population, alongside the North East region.
3.2. Individual characteristics
In order to control for individual characteristics, using UKHLS data, I collected and constructed several variables at individual level. These variables can be found in related studies assessing subjective well-being and immigration or ethnicity such as in Akay et al. (2014 & 2017), Alesina et al. (2013, 2016), Betz and Simpson (2013). Amongst the individual characteristics included are gender, age, income, health status and size of the household or weather the individual resides in an urban or rural area. Some individual characteristics in the model are derived variables, constructed using two or more variables, from the UKHLS.
Other control variables include the monthly income, which takes into account all earnings an individual is receiving monthly, not merely the ones derived due to the primary economic activity but grants, subsidies or income from a second job. Especially relevant is the Easterlin paradox, or income happiness paradox as over the long-term, happiness does not increase as country’s income rises. Indeed, a revised work from the Easterlin paradox (Easterlin et al. 2010) emphasises the relationship in the long-run between income and happiness, including not only developed countries but developing and eastern European countries, strengthening the results for the previously mentioned paradox. Cummins (2000) also emphasises the importance of personal income as an element in the maintenance of subjective well-being, especially for individuals in the lowest income quartiles. People from the poorest background will suffer a sharper decline in subjective well-being measures as in the short term happiness and income go together i.e happiness tends to decline in economic contractions and rise in expansions (Easterlin et al. 2010).
Moreover, I included a series of dummy variables taking into account individual characteristics, starting with the marital status of the individual, which could either be single, separated, married, in a registered civil partnership or widowed. Several studies have addressed to what extent the marital status of individuals could impact subjective well-being, as usually married persons experience more positive emotions and fewer negative emotions, consistently reporting greater subjective well-being than never-married individuals (Diener et al. 2000, Mastekaasa 1993), as having an spouse is an important source of social support being directly related to greater subjective well-being (Haring-Hidore et al. 1985).
the kind of retirement matters as voluntarily retirement has a less detrimental effect on life satisfaction than involuntarily retirement (Abolhassani and Alessie 2013), driven by the fact that voluntarily retirees are better financially and physiologically prepared, thus leading to higher levels of well-being (Bender 2012).
Besides, I also computed individual health status using dummy variables, which range between excellent, very good, good, fair and poor. The aforementioned work from Abolhassani and Alessie (2013), shed light on this relationship between health status and life satisfaction, stating that there is a strong negative effect of having a poor health condition on life satisfaction.
Finally, the academic literature has been assessing the links between subjective well-being measures and the number of children, with either positive or negative outcomes. Myrskyl and Mangolis (2014), using the UK and German longitudinal datasets, find that happiness increases in the years around the birth of a first child. Interestingly, having the first two children increases happiness, but a third child does not improve happiness scores. Angeles (2010) emphasises that the effects of having children will depend on the marital status of the individual, concluding that for married individuals children are positively correlated with life satisfaction and the effect increasing with the number of children, similarly to Ball and Chernova (2008). Nevertheless, Pedersen and Schmidt (2014), find that life satisfaction gains are mainly found with the first child and to a lesser extent from having more children.
Hence, here I included a dummy variable depicting whether or not an individual has one, two or more children, with individuals with no children used as the omitted category. The derived variable children is constructed using both number of adopted and biological number of children per individual.
3.3 Regional characteristics
Figure 3: GORs’ immigration Figure 4: GOR’s unemployment
Figures 3 and 4 provide a visual representation of the aforementioned regional characteristics. Can be seen from the blank areas in Figure 3, the northern part of the United Kingdom (Scotland, Northern Ireland and North East), alongside Wales, have low immigrant concentration, well below the threshold level of 11,03%, which is the average immigrant share in the United Kingdom over the time period 2009-2018. We also find GORs with relatively low immigration levels, represented by the GORs of North West, Yorkshire and the Humber, East Midlands and South West. The darkest areas represent immigrant shares above the previously mentioned threshold, which are West Midlands, East of England and South East. London definitely deserves a special mention, as has an average immigrant share above 36% over the reference time period. As it is clearly a special case, being one of the most culturally diverse regions in the world, it will be addressed separately in the analysis section. As seen in section 4.1, both London and South East also report the highest diversity levels of its foreign-born inhabitants, even though differences across most part of the United Kingdom in terms of migrant heterogeneity remain small.
Figure 4 depicts averaged unemployment levels for the desired time period and GORs. The average cross-country unemployment level is 6,4% which will serve as a reference. The southern parts of the United Kingdom (South West, South East, East of England), with the exception of London, are consistently ranking as the best performers in terms of keeping unemployment levels low, followed by the GORs of Scotland, Northern Ireland and East Midlands, still performing relatively well. Conversely, inner (central) areas such as North East, North West, Yorkshire and the Humber, West Midlands and Wales, alongside London, are subsequently experiencing relatively high unemployment levels. Additionally, albeit London is reporting
Data Source: Annual Population Survey. Author’s calculation. Average immigrant share values for the period 2009-2018. Average immigrant share 11,03%.
relatively high unemployment levels is the GOR with the biggest GDP value over time followed by South East, which contrary to London, also enjoys low unemployment levels. At the bottom of the GDP rank we can find the GORs of Wales, Northern Ireland and North East.
Although we may notice higher immigrant shares in southern areas, which also enjoy lower unemployment levels, one cannot draw any solid conclusion as the figures are merely indicative, further analysis in the following sections is required in order to asses the linkages between the economic performance of a specific GOR and how life satisfaction levels of individuals respond accordingly.
3.4. Summary statistics
Table 1
Low immigration GOR High immigration GOR
Natives Foreign-born Natives Foreign-born Obs (184,31) (16,340) (87,435) (25,996)
Source: UHKLS waves 1-9 (2009-2018), descriptive statistics; mean and standard deviation. “HH size” refers to household size. Health variables depict self-reported health satisfaction.
Variable Mean Std.Dev. Mean Std.Dev. Mean Std.Dev. Mean Std.Dev.
Table 1 presents the summary statistics (mean and standard deviation) for the aforementioned individual variables. In line with Akay et al. (2014) the results are split between GORs with relatively high and low immigrant shares, using the average immigrant share (0.1103) at a particular year as the threshold.
At first glance, we can see that individuals, either native or foreign-born, living in low immigrant GORs experience higher levels of life satisfaction, although differences remain small. Interestingly, natives report higher levels of live satisfaction than foreign-born individuals, in both low and high immigrant GORs.
Additionally, the descriptive statistics for both native and foreign-born individuals enable us to notice some demographic and individual characteristics;
First of all, foreign-born individuals tend to be married in a larger proportion than natives and are less likely to be widowed, divorced or single. Additionally, on average, foreign-born individuals have a higher number of children than natives, either one sole child or more. Native individuals seem to be less likely to raise children than surveyed foreign-born individuals. Thus, this might be partially translated into the relatively bigger household size of foreign-born individuals on average. Moreover, foreign-born individuals are on average younger than natives, being the difference bigger in low immigrant GORs.
Not only do surveyed foreign-born individuals appear to be younger but healthier as well, having a direct implication on the public discourse, as often times immigration is blamed for higher health costs and increased burdens for taxpayers, even though empirical evidence shows that immigrants are typically young, relatively healthy and thus less-likely to overuse the health care system (Giuntella and Mazzonna 2018). On addressing the share of individuals living in urban areas, one can see that foreign-born individuals are overrepresented in urban areas, especially in high immigrant GORs, where up to 95,83% of the foreign-born population surveyed live in an urban area rather than in a rural location. In low immigrant GORs, more natives can be found living in rural areas, roughly 30%. The share of native individuals living in urban areas increases in high immigrant GORs, from roughly 70% to 80%.
In addition, income (log) in these high immigrant GORs is higher for both natives and foreign-born individuals, albeit relatively small differences can be noticed, natives experience higher monthly incomes than foreign-born individuals. Similarly, more individuals report to be employed in high immigrant GORs, which altogether with the relatively higher incomes, suggests better labour market outcomes and agglomeration taking place towards these areas.
degree title/certificate. Studies, such as Docquier et al. (2014), point out at the fact that migrants are more educated than native-born individuals in many destination countries.
This phenomena of relatively better educated migrants being equally or less paid than their native peers is usually linked to the migrants’ slow integration into the labour market and the role of adaptation, often times forcing immigrants to downgrade their occupation upon arrival into jobs unsuited or unrelated to their skills (The World Bank 2018).
3.5. Empirical specification
The data I use is a household survey collected yearly, and a group of cross-sectional individuals (i), are observed over time (over t time periods). Therefore, this forms a panel data set with i>t, also known as microeconomic panel.
Using panel data sets of microeconomic nature brings large amounts of data points. Therefore allowing to account for unobserved individual differences or heterogeneity (Hill et al. 2015). Furthermore, it increases the degrees of freedom reducing collinearity among explanatory variables, improving the efficiency of the estimates. Moreover, panel data models can solve the omitted variables problems as invariant characteristics correlated with the observed ones may be accounted for by the fixed-effects specification.
The dependent variable used in the analysis is the individual’s reported life satisfaction, which is taken to measure the subjective well-being of the individual. The model used is:
𝑆𝑎𝑡𝑖𝑠𝑓!"∗ = 𝛽𝐼𝑀
!"+ 𝑋′!"𝛾 + 𝑍′!"𝛿 + 𝜖!" (2)
Where Satisf*, is the dependent variable measuring life satisfaction of an individual i at time t, which proxies the individual utility, as a measure of subjective well-being. The dependent variable is measured in a scale from 1 to 7.
The key independent variable is the immigrant share in each GOR (IM), measured as the ratio between the stock of immigrants and the total resident population in a given region (r) and point in time (t).
The error term 𝜖!" captures the unobserved, omitted factors in each time period for each individual. Those omitted factors can be split into time-invariant individual characteristics (ui), also know as unobserved individual heterogeneity and the unobserved individual and time-varying factors (eit).
Finally, after omitting missing observations from the dependent variable and sorting out the data, the model is estimated on a sample of 297,147 observations across the period 2009-2018.
Using the cluster-robust standard errors is recommended if N (observations) is large and T small (Hill et al. 2015). Since N is large and T is small in this study, cluster-robust standard errors at the individual level are applied in the different regressions in order to control for a potentially correlated error term (autocorrelation). After estimating both standard and cluster errors regressions in both OLS and fixed-effects specification, the cluster-robust standard errors are somewhat higher than the usual standard errors, thus its inclusion is recommended.
I use the fixed-effects estimation, which is recommended to address the potential endogeneity problem arising from the correlation between the unobserved heterogeneity and the explanatory variables. The results from the Hausman test depict a correlation between the error term ui (unobserved heterogeneity) and the explanatory variables. Therefore I used the fixed-effect estimation as the benchmark model, controlling the unobserved individual heterogeneity.
The fixed-effects specification excludes time-invariant variables such as gender or age (Akay et al. 2014, 2017). As stated in Ferrer-i-Carbonell and Frijters, age can also be written as; ageit𝛽age = agei1𝛽age + (t+1)𝛽age. Thus, agei1𝛽age is considered
time-invariant and will be picked up by the individual effects, with the term (t-1)𝛽age being common to all individuals.
For the interest of the reader I also document both OLS and random effects model, which allow us to see how the excluded variables in the fixed-effect specification behave in this framework.
4. Results
4.1. How do individual characteristics relate to life satisfaction levels?
First of all, notice in columns (1) and (2) that the aforementioned relationship between age and life satisfaction proves to be as expected as well, with the age and quadratic age forms altogether depicting the U-shaped relationship found in subjective well-being related studies (Angeles 2020; Blanchflower 2020; Ferrer-i-Carbonell 2005). This relationship is derived from both the OLS (Ordinary Least Squares) and RE (Random Effects) regressions. Although they do not constitute the reference regression in this analysis due to the aforementioned reasons (see empirical specification), they enable us to see how individual time invariant characteristics such as age and sex – excluded from the fixed-effects (FE) specification – behave in this environment.
Concerning the marital status of the individual, as we can see in the fixed-effects
specification (see column 3), being married is positively correlated with life satisfaction levels, in line with the previously mentioned studies from Diener et al. (2000), Mastekaasay (1993) or Haring-Hidore et al. (1985). Seemingly, being single also leads to higher life satisfaction levels. Precisely, married and single individuals report on average a 0.0879 and 0.0699 unit increase respectively in their life satisfaction level whereas widowed status does not bringing any significant association with life satisfaction levels, as the coefficient is not statistically significant.
Regarding the number of children, despite there is a positive relationship with life satisfaction, the results are not statistically significant. Conversely, other studies have proven this association to be significant and that increases in life satisfaction come mainly from the first and second child, with the effects vanishing with the third (Myrskyl and Margolis 2014; Pedersen and Schmidt 2014).
Interestingly, as for the health status, only self-reported excellent health draws a positive relationship with life satisfaction, with individuals reporting excellent health experiencing on average 0.0944 unit increase in their self-reported life satisfaction. Conversely, individuals reporting bad health – which includes individuals reporting fair and poor health – experience a sharp decline in its life satisfaction levels on average, as they are facing decreases in their life satisfaction by 0.5061 points on average. This negative association between bad health and life satisfaction can be also found in studies such as Abolhassani and Alessie (2013) in Germany. Similarly, Papageorgiou (2018) estimates this decline in life satisfaction for individuals with bad self-reported health in the United Kingdom to be 0.817, a somewhat higher negative effect.
hence easier to meet, they may be facing less disappointment and less frustration, which could drop life satisfaction levels.
However, unemployed individuals face a relevant decline of 0.2619 points on their self-reported life satisfaction levels, holding all variables constant. Moreover, not being in the labour force (disabled individuals and people doing family care) has a small positive impact on life satisfaction.
In addition, despite several studies such as Diener et al. (1995) find a (strong) positive relationship between income and life satisfaction, in this study income and life satisfaction do negatively correlate, albeit the results are small in magnitude. Ferrer-i-Carbonell (2005) on assessing income and subjective well-being linkages in Germany, find that the larger one’s own income is, the happier the individual will be, albeit the reference income group proves to be of the utmost relevance; if increases in family income are accompanied by identical increases in the reference group, then income increases do not lead to significant changes in well-being. Finally, both the number of hours worked (weekly) and higher education draw a positive albeit non-statistically significant relationship with the dependent variable life satisfaction.
4.2 Immigration and life satisfaction
Notice below (Table 3) benchmark results for the fixed-effects specification are reported alongside the OLS and Random Effects regressions, for both the entire sample and London. As previously discussed, London’s immigrant population is well above the United Kingdom’s average and deserves special mention. In order to provide a more complete picture of the analysis, results for foreign-bon individuals are also included. For the sake of brevity only the main outcomes of interest are depicted in the following tables.
The coefficient 𝛽 (equation 2) for the independent variable (immigrant share) of interest is depicted below alongside the immigrant diversity index and the two GOR control variables, log GDP and unemployment levels. Recall all specifications are constructed using clustered standard errors.
Before providing a discussion of the above results it is important to stress the role endogeneity could play in this model as foreign-born individuals could self-select into regions with higher life-satisfaction or well-being levels, if this was the case the model would suffer from reverse causality being the immigrant share a result of life satisfaction.
observed ones. Nevertheless, for the sake of the argumentation, endogeneity concerns have been further addressed using the instrumental variable approach in section 5.3. Table 3
Natives Foreign-born
Satisfaction OLS RE FE OLS RE FE
All GORs IM share -0.0021** (0.0008) -0.0019** (0.0008) 0.0118*** (0.0027) (0.0019) 0.0056*** 0.0070*** (0.0018) 0.0176*** (0.0059) Diversity -0.0010 (0.0008) 0.0014* (0.0008) 0.0063*** (0.0015) -0.0005 (0.0032) 0.0029 (0.0028) 0.0091* (0.0049) GDP (log) -0.0242* -0.0542*** -0.4033*** -0.1021** -0.1374*** -0.4179*** (0.0136) (0.0128) (0.0486) (0.0487) (0.0443) (0.1475) Unemployment -0.0132*** -0.0135*** -0.0292*** -0.0422*** -0.0437*** -0.0491*** (0.0022) (0.0019) (0.0027) (0.0060) (0.0055) (0.0087) R2 0.1442 0.1429 0.0146 0.1190 0.1183 0.0152 N 258,571 258,571 258,571 38,576 38,576 38,576 London IM share 0.0766*** 0.0679*** 0.0634*** 0.0546** 0.0564** 0.0643** (0.0229) (0.0212) (0.0220) (0.0248) (0.0236) (0.0252) Diversity 0.0357 -0.0099 0.0221 -0.0412 -0.0210 0.0113 (0.0928) (0.0677) (0.0694) (0.0871) (0.0849) (0.0889) GDP (log) -1.7243*** -1.4248*** -1.1744** -2.3084*** -2.1812*** -2.2181*** (0.5814) (0.5439) (0.5745) (0.6661) (0.6385) (0.6856) Unemployment -0.0903*** -0.0723** -0.0570** -0.1673*** -0.1639*** -0.1664*** (0.0244) (0.0228) (0.0239) (0.0273) (0.0261) (0.0281) R2 0.1377 0.1367 0.0145 0.1194 0.1186 0.0194 N 19,210 19,210 19,210 14,867 14,867 14,867
The baseline FE (fixed-effects) specification draws a positive and significant correlation (p<0.01) between the immigrant share in a specific GOR and life satisfaction levels on native individuals. The results provided in both the OLS and RE
Source; UKHLS waves 1-9 (2009-2018), the dependent variable is the self-reported life satisfaction. “IM share” is the immigration share and “Diversity” refers to the immigrant diversity index. Results for the OLS (Ordinary Least Squares), RE (Random-effects) and the benchmark specification FE (fixed-effects) estimation. Omitted Dummy variables; “No children”, “Divorced”, “Employed” and “Very good health”.
specification differ substantially compared to the benchmark FE specification. In fact, as noted in Ferrer-i-Carbonell and Frijters (2004), this time-invariant fixed-effects considerably affect the results and their omission could potentially bias in the estimates.
Precisely, for every percentage point increase in the immigrant share for the sample including all GORs, the expected life satisfaction increases by 0.0118 on average, holding all variables constant. Individuals in the London GOR seem to be more sensitive to immigration, as not only the coefficient is also positive and statistically significant but the effect is stronger, albeit it still remains relatively small; one percentage point increase (decrease) in the immigrant share in London, increases (decreases) life satisfaction levels by 0.0634 on average.
Interestingly, the diversity index measurement used in this study is positively associated with life satisfaction across the United Kingdom, with the coefficient being statistically significant at 0.01 percent level. However, the diversity effect is smaller than the immigrant share impact, with life satisfaction levels merely rising by 0.0063 points for every percentage point increase in the diversity index. Conversely, native individuals’ life satisfaction residing in London seems not to be affected at all by a more diverse foreign-born population although the effects (if any) may be positive. This not significant effect of diversity on life satisfaction levels of native Londoners could be driven by the concept of social cohesion as higher levels of diversity could decrease the sense of social cohesion. Precisely, Sturgis et al. (2014) on assessing immigrant impacts on social cohesion levels in London neighbourhoods state that white-British individuals residing in London express the lowest levels of social cohesion in their neighbourhoods, which could ultimately affect life satisfaction levels. Furthermore, Sturgis et al (2014) find that ethnically diverse neighbourhoods in London tend to be also more deprived, thus leading to less social cohesion.
In short, not only the immigrant share in a given GOR has a positive effect on life satisfaction levels but the typology of the foreign-born population. We can expect increased levels of life satisfaction in GORs with higher and more diverse levels of foreign-born individuals after controlling for both individual and regional characteristics. Nevertheless, this positive association vanishes when only assessing the London GOR, as only the immigrant share holds positive and statistically significant.
Furthermore, Table 4 provides FE estimates broken down by sex, age, education and labour market status, which will help assessing the different impacts immigrant shares might have on different population groups.
Overall, these individual characteristics depict a modest, positive impact of immigration on life satisfaction. Female and male individual behave in a similar way, and no meaningful differences are spotted between individuals with or without higher education. Regarding individuals’ age, all results positively draw a relation between immigration and life satisfaction albeit for the cohort of people aged below 40, this association is not statistically significant.
Table 4
Sex Age Education
As for the role of diversity, it showcases a positive and significant relationship with life satisfaction levels for most sub-groups. However, the age cohort above 50 years old does not show a significant relationship between diversity and life satisfaction. Additionally, looking at life satisfaction by labour market status will help on the discussion about what may be the channels though which immigration could affect life natives’ satisfaction levels. Also, studies such as Akay et al. (2014), Suedekum et al. (2014), Howley et al. (2018) or Papageorgiou (2018) emphasize the importance of local labour markets as a mechanism through which immigration and/or diversity may affect natives. In fact, the immigrant share positively correlates with life satisfaction of
Satisfaction Females Males Age < 40 Age 40-50 Age > 50 Higher No higher IM share 0.0116*** 0.0119*** 0.0039 0.0116 0.0228*** 0.0114*** 0.0115*** (0.0038) (0.0037) (0.0034) (0.0081) (0.0055) (0.0043) (0.0034) Diversity 0.0078*** 0.0043* 0.0133*** 0.0100*** 0.0019 0.0063** 0.0060*** (0.0028) (0.0022) (0.0028) (0.0035) (0.0023) (0.0034) (0.0018) R2 0.0151 0.0144 0.0205 0.0170 0.0129 0.0190 0.0138 N 144,499 114,082 83,008 53,251 127,232 64,462 194,120
Labor market status
Employed Unemployed Retired Student IM share 0.0145*** -0.0034 0.0206** 0.0123* (0.0032) (0.0276) (0.0088) (0.0072) Diversity 0.0037* 0.0129 0.0065** 0.0111 (0.0032) (0.0125) (0.0033) (0.0073) R2 0.0122 0.0237 0.0092 0.0188 N 150,761 12,382 67,224 17,592
employed individuals, increasing on average their life satisfaction by 0.0145 points. Furthermore, the diversity measure also draws a positive and significant (at 0,1 percent level) association with life satisfaction levels of employed native individuals.
Interestingly, as retired people are often claimed to be the cohort of individuals mostly opposing immigration, retired individuals’ life satisfaction is positively and statistically correlated with both the immigrant share and the diversity measure, whereas the diversity measure does not have any significant effect for student individuals.
Conversely, this positive effect of either immigration or diversity on life satisfaction becomes not significant for unemployed individuals. Both the positive association for employed individuals and the non-significant association for unemployed people backs up the importance of immigration and local labour markets on the life satisfaction levels of native individuals. This relationship will be further elaborated in the discussion section.
4.3. What about ethnicities?
One of the main concerns embedded in this study is the concept of native individual. The study assesses the linkages between life satisfaction levels and immigration levels for the native group, regardless of ethnic traits. However an individual could be born in the United Kingdom and be considered as a native individual in this study but he/she might have a different ethnic/racial background than the majority. Hence, white-UK individuals may respond differently to either the immigrant share or the diversity measure in their region.
Thus, in this section (see table 5) I examine what is the outcome of equation 1, for the benchmark FE specification, when focusing on individuals being white-UK individuals (Longhi 2014), either English, Scottish, Welsh or Northern Irish.
Table 5
Satisfaction Natives Foreign-born White-UK Non-white UK IM share 0.0118*** 0.0176*** 0.0105*** 0.0177*** (0.0027) (0.0059) (0.0028) (0.0051) Diversity 0.0063*** 0.0091* 0.0075*** 0.0007 (0.0015) (0.0049) (0.0016) (0.0035) R2 0.0146 0.0152 0.0146 0.0155 N 258,571 38,576 241,937 55,690
Source; UKHLS waves 1-9 (2009-2018), the dependent variable is the self-reported life satisfaction. Results for the fixed-effects specification with clustered standard errors. Results for ethnic groups white and non-white British. “IM share” is the immigration share and “Diversity” refers to the immigrant diversity index. Omitted Dummy variables; “No children”, “Divorced”, “Employed” and “Very good health”. Significance levels at *** p<0.01 / **
The above FE results for white-UK individuals (English, Scottish, Welsh or Northern Irish) report no difference when taking into account ethnic traits with respect to the baseline specification, with respect to the immigrant share. The coefficient is extremely close with increases (decreases) in the immigrant share by one percentage point reporting increases (decreases) in life satisfaction by 0.00138 points. Nevertheless the diversity measure becomes statistically insignificant for the cohort of non-white UK individuals. Thus, results from table 5 strengthen to some extent previous findings from section 5.2, regarding the immigrant share although differences can be found in the diversity measure.
4.4. Further specifications
The aforementioned relationship between the immigrant share and life satisfaction levels appears to be positive and statistically significant at the benchmark specification alongside the diversity measure. As previously mentioned, London is clearly a special case in terms of its foreign-born share, averaging 36,10% of foreign-born population inhabitants (Annual Population Survey), and probably being the most ethnically diverse conurbation on the planet (Sturgis et al. 2014). Hence, it might be worth taking a look at the fixed-effects estimation without considering the London GOR. As seen in table 6 (column 2), the results for both immigrant share and the diversity measure hold significant and extremely close to the benchmark specification (column 1) when excluding London from the analysis. Hence, the positive association between life satisfaction levels and the immigration share is not driven by the effects in the London GOR.
In order to further address the aforementioned potentially endogeneity problem of foreign-born individuals sorting into specific areas, I undertook the IV (instrumental variable) approach, using one year lagged immigrant share as an instrument for the current immigrant share in a specific GOR.
A good instrument has a sufficiently strong correlation with the endogenous variable (immigrant share) but should not have a direct link with the dependent variable (life satisfaction, or links other than through the immigrant share. Furthermore, the instrument is not correlated with the error term of the main regression (exogenous).
