Longing for which home: Evidence from global aspirations to stay, return or migrate onwards

Bekaert, Els; Constant, Amelie F.; Foubert, Killian; Ruyssen, Ilse Working Paper Longing for which home: Evidence from global aspirations to stay, return or migrate onwards UNU-MERIT Working Papers, No. 2021-035 Provided in Cooperation with: Maastricht Economic and Social Research Institute on Innovation and Technology (UNU-MERIT), United Nations University (UNU) Suggested Citation: Bekaert, Els; Constant, Amelie F.; Foubert, Killian; Ruyssen, Ilse (2021) : Longing for which home: Evidence from global aspirations to stay, return or migrate onwards, UNU-MERIT Working Papers, No. 2021-035, United Nations University (UNU), Maastricht Economic and Social Research Institute on Innovation and Technology (UNU-MERIT), Maastricht This Version is available at: https://hdl.handle.net/10419/326794 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. https://creativecommons.org/licenses/by-nc-sa/4.0/          #2021-035 Longingforwhichhome:Evidencefromglobalaspirationsto stay,returnormigrateonwards  ElsBekaert,AmelieF.Constant,KillianFoubertandIlseRuyssen            Published14September2021     MaastrichtEconomicandsocialResearchinstituteonInnovationandTechnology(UNU‐MERIT) email:[email protected]u|website:http://www.merit.unu.edu  Boschstraat24,6211AXMaastricht,TheNetherlands Tel:(31)(43)3884400 UNU-MERIT Working Papers ISSN 1871-9872 Maastricht Economic and social Research Institute on Innovation and Technology UNU-MERIT UNU-MERIT Working Papers intend to disseminate preliminary results of research carried out at UNU-MERIT to stimulate discussion on the issues raised.   1 Longing for Which Home: Evidence from Global Aspirations to Stay, Return or Migrate Onwards* By Els Bekaert,1,2Amelie F. Constant,3 Killian Foubert,1,2 Ilse Ruyssen1,2 1 Ghent University, Department of Economics, Belgium 2 United Nations University, Comparative Regional Integration Studies (UNU-CRIS), The Netherlands 3 Princeton University, Office of Population Research, Princeton, NJ 08544, USA, (215) 665 0618; [email protected]; GLO, CESifo; UNU-MERIT (corresponding author) Abstract Aspirations provide the underlying dynamics of the behavior of individuals whether they are realized or not. Knowledge about the characteristics and motives of those who aspire to leave the host country is key for both host and home countries to formulate appropriate and effective policies in order to keep their valued immigrants or citizens and foster their (re-)integration. Based on unique individual-level Gallup World Polls data, a random utility model, and a multinomial logit we model the aspirations or stated preferences of immigrants across 138 countries worldwide. Our analysis reveals selection in characteristics, a strong role for soft factors like social ties and sociocultural integration, and a faint role for economic factors. Changes in circumstances in the home and host countries are also important determinants of aspirations. Results differ by the host countries’ level of economic development. Keywords: Economics of Immigrants; Geographic Labor Mobility; Public Policy; Microeconomic Behavior: Underlying Principles; International Migration; Large Data Sets: Modeling and Analysis JEL codes: J15, J61, J68, D01, F22, C55 * The authors would like to thank Ellen van de Weghe for providing some exploratory analyses, as well as Michel Beine, Frédéric Docquier, Gerdie Everaert, Glenn Rayp and participants to the 2021 EUIA for their useful suggestions. We are also grateful to Robert Manchin and the Gallup Institute for Advanced Behavioral Studies for providing access to the data for the purpose of this project. This paper uses confidential data from the Gallup World Polls. The data can be obtained by directly filing a request with Gallup. The authors are willing to provide guidance on how to obtain the data to interested researchers (Ilse.Ruysse[email protected]) and can provide Stata codes to reconstruct the estimation sample from the raw Gallup data. The authors have nothing to disclose. 2 1. Introduction This paper proffers a thorough global analysis of the factors influencing first generation immigrants’ aspirations to stay in the host country, return back home or move onwards to another country. While there are many theories that explain the initial migration of people, fewer theories exist about the return and onward migration of immigrants as these are relatively more recent phenomena of the late 20th century (Constant, 2021). With the availability of better data, more and more studies try to calculate the probabilities to out-migrate, address selection issues, estimate circular propensities, and determine the steady state, all in the spirit of modelling actual moves (Constant, 2020). Such studies provide useful information that host, home, and third countries are eager to possess in order to effectively and smoothly encounter the movement of people in their territories. Yet, there is no consensus about the selection of those who leave. Due to lack of appropriate data, existing empirical studies have typically focused on a particular host country from which people return back to specific home countries (Constant and Massey, 2002; Jensen and Pedersen, 2007; Constant and Zimmermann, 2011; Monti, 2020; Dustmann and Weiss, 2007). Moreover, all studies on the actual or revealed choice of out-migration rely on ex-post information, that is, information about the characteristics of returnees after they exit the host country. This concrete revealed behavior, however, is based on both the willingness to exit the host country and the ability to do so. An effectuated exit includes unplanned out-migration due to force majeure, in which case the selection of the actual emigrants does not offer any meaningful information about the selection of stayers, which is what host countries care about. Ex-ante information about the selection of out-migrants would be, undoubtedly, more desirable and more valuable to countries, so they can prepare ahead of time and better cope with population shifts related to the exit of immigrants, their return, and onward movement. Such information can be obtained by studying the aspirations, desires, or stated preferences of immigrants. Different than revealed preferences (RP), stated preferences (SP) come from hypothetical scenarios, in which respondents are provided with options and are asked to choose the “best” alternative in line with a utility maximization model. SP are the basis of the utility function, since – simply put – utility is a convenient way of describing preferences or choice behavior. SP are highly informative and uncover different mechanisms than RP. Such is the value of the (re)locational choice, irrespective of the capability to act upon it. 1 1 We are cognizant of the fact that SP and RP are highly correlated but there is no established causation. Due to the endogeneity of choices, SP cannot determine actual (revealed) preferences (Manski, 1990). For example, as 3 Thus, whether SP or aspirations to stay, return, or migrate onwards are realized or not, they merit the attention of the literature. Aspirations illustrate how immigrants perceive their present life in a host country and how and where they envision their future. They capture the state of mind of immigrants and are a window into the immigrants’ potential to shape their present as well as their future. They reveal clues about how immigrants plan to share and dedicate their time or resources, which can make a difference in their integration and socio-economic success in the host country (Constant, 2021). Aspirations to stay signal satisfaction with the host country and can even convey the notion that immigrants want to better themselves, which in turn could shape the communities in which they live. Aspirations to leave, in contrast, can disclose dissatisfaction with the host country, which, in turn, can be manifested through a loose (or lack of) attachment to the labor market, potentially leading to lower earnings, resulting in more dissatisfaction as well as more discrimination from natives. Unhappy migrants who face adversity and anticipate leaving might also be less inclined to invest in human capital and acquire skills that are particularly valuable in the host country such as learning the local language, with detrimental effects on their labor market performance. At the same time, immigrants may invest less in relationships and human skills or assets (Carling and Pettersen, 2014). Unhappiness can be equally manifested as deteriorating health, triggering a lower performance, lower economic stability, and more discrimination from the host society. It can easily end up in a vicious circle. Aspirations or preferences provide powerful information for policymaking decisions in the short, medium, and long-term as well. 2 Knowing the preferences of optimizing immigrants makes policies more desirable and cost-effective. For the host country, aspirations are important when policymakers design integration policies and want to retain their immigrants for whom they may have already made investments. Immigrant aspirations reveal how successful the host country has been in providing for them and incorporating them in the mainstream society and economy. If the majority of immigrants wants to exit, this most likely indicates that immigrants are not comfortable living in the host country. One can argue that in this case, not only the host country has failed in its integration policies, but it cannot recover the investments it has made either. Importantly, selective out-migration can have harmful immigrants optimize their stay in the host country and circumstances change they may prolong or shorten their residence abroad, such that SP and RP may not be the same. SP place an upper bound on the probability that an individual will behave in a given way but do not identify this probability (Manski, 1990). 2 It is possible that preferences to exit the host country capture the swiftness with which immigrants respond to the general situation in the economy. Especially immigrants who have been in the host country for a shorter period of time are more likely to relocate if things change. 4 effects on the economy (labor markets, tax revenues, the pension system), R&D, naturalization rates, and societal norms of the host country (Constant, 2020). For host countries that want to retain their high skilled immigrants, it is important to also know where immigrants are going after exit. This offers deeper insights about why they are leaving. Besides, if large numbers of immigrants return home, this could affect bilateral relationships and agreements. 3 Because out-migrants may be a selected subsample of the initial arrivals, this has serious ramifications in migration research. Selective exit may produce extreme bias in assimilation studies and in studies about the impact of immigrants on the host country’s economy, welfare, and social and cultural realm, thus resulting in a false narrative. Home countries are equally eager to know the characteristics of their returnees. Most alarming is the issue of remittances, which constitute the largest source of foreign income for many developing economies and are a lifeline for development. Because self-selection in specific characteristics is inherent in migration, return migration may contribute to regional imbalances, overwhelm the educational system and/or the labor markets of the home country. Policymakers can accordingly decide on the kind of monetary and non-monetary investments they need to implement in order to counterbalance lost remittance inflows. Home countries can also introduce institutional reforms to accommodate the returnees and speed up their reintegration. This is particularly important because many home countries count on their returnees’ financial investments, their upgraded skills and knowledge, and their enlarged social capital (Constant, 2020). Along the same vein, if selected groups of immigrants aspire to move onwards to another host country, this can have grave consequences for the economy and society of the new host. Yet, comparatively less is known about onward moves. Research on the unconstrained preferences of immigrants can improve the ability to forecast trends in migration and to design better and more appropriate policies. Understanding the extent of out-migration and the characteristics of immigrants who leave compared with those who stay might ultimately contribute to global mobility gains to be shared by home, host, and third countries. Nonetheless, the value of aspirations has been largely overlooked by the literature. In this paper, we make the following contributions. First, we uncover the value of aspirations in the economics of migration research by viewing them as unconstrained preferences that offer insightful mechanisms of immigrant selection for scientists and 3 We do not discount the possibility that sometimes host countries want their immigrants to leave. Knowing what prompts immigrants to leave can be equally powerful when enacting policies. 5 policymakers alike, whether they are realized or not. Using a random utility framework, we empirically investigate the determinants of aspirations to stay, return or move onwards, the three fundamental choices that immigrants face in any host country. We emphasize the modelling of the aspiration to stay, which – while it represents the majority – is often an overlooked choice in studies treated as a corollary. We further model the choice sets return versus stay and onward moving versus stay and underscore the importance of context in these contrasts. Second, we focus on first generation immigrants (the foreign-born) because they are a rather homogeneous group that share many characteristics independent of the host country they live in. They are a distinct group of individuals, different from higher generations and the native-born who may contemplate an initial migration. Third, we make use of the unique 21st Century individual-level Gallup World Polls (GWP), which gather comparable information from over 160 countries worldwide on respondents’ migration status, country of birth, aspirations to migrate internationally, as well as a whole range of individual and household characteristics and opinions. Comparability across countries constitutes the added value of our paper because with the GWP we can carry out a unique global cross-country analysis and explore important but understudied components of migration flows across the world. We go beyond studying the role of traditional determinants and augment our analysis with a novel combination of variables at the individual, community, and country level that could enrich our understanding of out-migration such as individual wellbeing, social and economic integration in the host and home country, and changing circumstances related to exogenous macroeconomic, environmental, and polity conditions in the host and home countries. Our holistic analysis allows us to provide a deeper insight into the complexity of immigrants’ behavior and how immigrants judge their socio-economic, cultural, and political position in the host country compared to other country options. To the best of our knowledge, our paper is the first to make use of the GWP to study aspirations to stay, return or migrate onwards among first generation immigrants on a global and comparative scale, and to bring together several branches of research in migration. Our results reveal stark differences in selection in characteristics and macroeconomic variables among the three comparative choices that immigrants are confronted with. They are robust to several tests, justifying our conjectures, and derive important labor market and migration policy implications. The rest of the paper is structured as follows. Section 2 reviews the various strands of literature to which our paper is related. Section 3 discusses our data and presents the descriptive 6 statistics on immigrants’ aspirations regarding the three hypothetical choices. Section 4 provides the theoretical foundations for our empirical analysis and presents the methodology. Section 5 presents the results. Section 6 concludes. 2. Related Literature: Actual Behavior versus Aspirations Although there has been abundant discussion on the theoretical conceptualisation of return migration over the years (Constant and Massey, 2002; Cassarino, 2004; Dustmann and Weiss, 2007), empirical studies on the determinants of actual return migration are lagging behind. The knowledge gap is even larger when it comes to onward migration. The main reason is the paucity of reliable large-scale quantitative and comparable data across countries on those who out-migrate (Cassarino, 2004; Azose and Raftery, 2019; Constant, 2020). The limited number of host countries that have registers such as the Scandinavian countries and Belgium do not always enforce de-registration upon exit. Researchers typically rely on nationally representative surveys that may not have information about out-migration, and if they do, this information pertains to a specific host country and specific home countries as well as specific populations such as labor migrants or their subsets (guestworkers or the high skilled). Longitudinal surveys may suffer from attrition and/or have a small number of observations to allow for a meaningful analysis. Other surveys that include information about out-migration might not be nationally representative or may lack information on important characteristics, thus limiting their external validity. Another volatility of empirical studies is that – because selection depends on both data and context – their findings may differ even though they pertain to the same home and host countries, and are based on the same datasets. For the actual behavior to be effectuated, many more attributes of the RP are needed such as travel costs and visa costs for instance. Existing studies try to counterbalance data issues with advanced econometric techniques, and by controlling for a variety of observed and unobserved characteristics. 4 Yet, empirical studies are crucial in supplementing the theories, which may sometimes predict mixed outcomes depending on the initial migration conditions between the countries involved (theory à la Roy). A classic example is that, theoretically, returnees could be the best from the worst or the worst from the best. It is then up to empirical research to settle the score in praxis. 4 A comprehensive review about return, onward, and repeat migration in Constant (2020; 2021). 13 The phrasing of Q1 bears the following advantages. First, the word “ideally” conveys an unconstrained maximization problem. Second, the words “move permanently” evoke a serious reaction among first generation immigrants, precluding any superficial, exploratory or whimsical thoughts about relocating. This question, hence, captures aspirations to migrate permanently abroad well. These aspirations are somewhat stricter than mere migration considerations, as used for instance in Creighton (2013), since our Q1 uses a stronger formulation that directly asks for the likely response under ideal conditions. Each choice reveals how much the respondents value it. Moreover, using the respondents’ statements on their preferred destination affords us to exploit the bilateral nature of migration aspirations. We contend that, while Q1 and Q2 are hypotheticals, they are nonetheless meaningful and close to the truth for our sample of immigrants. Because they have already migrated at least once they can adequately evaluate the alternative choices and properly express their preferences. First generation immigrants know the difference between temporary, permanent, and/or circular migration. They are also aware of the perils of the migration journey and they have first-hand experience about the hardships and adversities that come with living in a new country. They are even cognizant of the psychic costs involved. Because immigrants have already migrated once, they are more prone to migrate again either to advance their careers or as a coping strategy. The third country option is, therefore, very real to them. Thus, we evade previous criticisms about the meaning of aspirations or about what aspirations capture (Manchin and Orazbayev, 2018). Another advantage of SP over RP is that individuals are asked to make more than one choice and are presented with trade-offs. In this case, their responses reveal how much immigrants value (or not) the forgone choice or the opportunity cost of not choosing a particular alternative from the choice set. 3.2. Individual and Household Characteristics We control for demographics such as the respondents’ age through four dummy variables to capture non-linearities (18-34 as reference, 35-49, 50-64, and 65-75), gender (dummy for being male), marital status (dummy for being married or cohabitating), and household size (the number of children and adults in the household). We also account for human capital by using three dummies capturing the immigrants’ education levels (secondary, tertiary, and elementary education as the reference i.e., up to 8 years of basic education), a dummy for the duration of stay in the host country (whether the individual has moved to the host country within the last 14 five years), and the GWP's Personal Health Index 8 (measuring perceptions of one’s own health status as well as incidence of pain, sadness, and worry) that varies between 0 and 100 with a higher score indicating an overall better health. We furthermore augment our basic model with rarely available information about the selfreported religion of respondents. Through five dummies for Christianity, Islam, Judaism, other religions, and no religion as a reference category, we seek to examine if religious and spiritual capital plays a discernible role in shaping aspirations to stay, return, or move onwards. Faith in one’s God can provide solace to endure hardships in the host country, hope for better days, and/or confidence in succeeding in a new country. Accordingly, we anticipate to find differences among the denominations. Alternatively, we know that when immigrants move to a new country their religion may clash with the majority of the host country. Such a clash may prompt feelings about out-migration. 3.3. Networks, Economic Independence, and Wellbeing In our analysis we consider additional categories of variables that can best capture what influences the aspirations of immigrants. Namely, we include network effects, the level of integration in the host country, and wellbeing. Strong networks in the host country or abroad are known to impact and sway aspirations. They provide information that reduces the costs and risks of migration, they help with integration in the host country, with migration in a third country, and with reintegration in the home country; they are especially beneficial to the lessselected migrants, women, and the undocumented (Constant, 2020). Having friends to rely on has been found to be the top reason for happiness in studies ranking countries. We construct these variables using direct information from the GWP. Network abroad is a dummy variable capturing whether respondents have friends or relatives abroad whom they can count on when needed (also referred to as “distance-one connections abroad”). 9 The Local Network Index is 8 In GWP, Index scores are calculated at the individual record level. For each record, positive answers are scored as “1;” all other answers are assigned a score of “0”. An individual record receives an index only if it has valid scores for at least a majority of questions asked. A record’s final index score is the mean of valid items times 100. The final country-level index score is the mean of all individual records for which an index score was calculated. Country-level weights are applied to this calculation. 9 Note that this dummy does not account for the country of residence of distance-one connections. Migrants are, however, likely to state having friends or relatives in their country of birth whom they can count on if needed. To test for this, we employ a follow-up question, which tracks up to three countries of residence for distance-one connections, and construct a bilateral network dummy that takes the value one if the respondent indicates having a distance-one connection in the birth country. Indeed, we find that 61 percent of migrants stating at least one country in which they have distance-one connections actually mention their country of birth. This reassures us to 15 constructed from principal component analysis of two questions that proxy the intensity of local social ties: (i) “In the city or area where you live, are you satisfied or dissatisfied with the opportunities to meet people and make friends,” and (ii) “If you were in trouble, do you have relatives or friends you can count on to help you if you need them, or not?”. The index is rescaled to range between zero (no social ties) and one (have full social support in the country). We measure the economic integration of immigrants in the host country using two behavioral and two attitudinal variables. First, we account for employment status (a dummy taking the value one if the respondent is employed part time or full time – including the selfemployed – and zero if (s)he is unemployed or out of the workforce). Second, we include a household wealth index constructed as the polychoric principal component (Dustmann and Okatenko, 2014) of four questions that are available for all countries in our sample. This index captures basic material and financial wealth like having a television, access to the Internet, having enough money to buy food, and ability to provide adequate shelter or housing for one’s family in the past 12 months. Higher values indicate more wealth. Third, we include a dummy measuring whether one lives comfortably on current household income. Fourth, we consider a dummy capturing the respondents’ satisfaction with their standard of living, inclusive of all the things they can buy and do. These aspects are examples of what Sen calls functionings. Finally, we take advantage of two indices in the GWP that assess the wellbeing of the respondents within their communities such as the freedom to make choices or feeling secure and fulfilled. We include the GWP Diversity Index, which measures the community’s acceptance of people from different racial, ethnic, and cultural groups. To measure satisfaction with aspects of everyday life in a community such as education, environment, healthcare, housing, transportation, and infrastructure, we include the GWP Community Basics Index. A higher score on these indices (both varying between 0 and 100) denotes a more accepting community and a higher satisfaction with everyday life in the community (see footnote 8 for the construction of the indices). These types of freedoms and opportunities for fulfilment relate to Sen’s capabilities. 3.4. Macroeconomic, Environmental, and Polity Variables The last set of variables in our model accounts for the extent to which changes in circumstances in the host or home country influence migration aspirations. First, we employ a proxy for continue with the unilateral variable rather than the bilateral one, which is not as relevant for onward migration since we do not differentiate between particular destinations. 16 improving or worsening economic conditions in the host and home country such as GDP growth taken from the World Development Indicators. Higher GDP growth should attract people to stay or return to that country in hopes that they will flourish along with everybody else. Second, we extract information about the occurrence and frequency of natural disasters in the year before the survey date from the International Disasters database EM-DAT. Examples of natural disasters are earthquakes, floods, heat waves, hurricanes, tornados, wildfires, volcanic eruptions, and tsunamis to name a few. They all have devastating effects and can wipe out entire regions. These phenomena are even worse in developing countries. We hypothesize that immigrants should not aspire to stay in or return to such a ruined area. Third, we consider variables that signal political instability in the host and home countries. We construct these variables as dummies from the Polity IV Index. They are indicating a significant change in the Polity IV Index democracy score in the three years preceding the interview. The Polity IV Index goes from -10 to 10, with -10 being attributed to the more autocratic countries, and +10 to the more democratic ones. We contend that more democratic countries with lower instability would be more attractive and will be chosen. 10 3.5. Descriptive Statistics Table 1 provides summary statistics for the fore-mentioned variables employed in our empirical analysis. To better show the composition of our sample and identify potential differences, we disaggregated the characteristics by the aspiring status of stayers, returnees, and onward movers. Our dependent variable shows that 75% of the first generation immigrants in our sample aspire to stay permanently in the host country in which they reside, while only 8% aspire to return to their country of birth, and 17% aspire to move onwards to another country. These raw statistics portray somewhat different profiles of immigrants, depending on their aspiration status. The stayers are, on average, the oldest (47 years old), indicating the “ageing in place” of immigrants. This is 10 years older than those who want to move to a third country (the youngest group) and 5 years older than the returnees. Among the stayers and the returnees we see a predominantly female sample, although the gender ratio balances out among the onward movers. All three groups have a higher percentage of people being married or in domestic partnership, with an average household size of four. Regarding secondary and tertiary 10 Note that much of the host country characteristics are absorbed by the country fixed effects so that the only remaining source of identification is within-country variation over time. 17 education, the groups are not very different. Across all three aspiring groups, over 50% of immigrants have secondary education, over a quarter have tertiary education and close to 20% have up to eight years of elementary education. Table 1: Descriptive Statistics Aspire to Stay Return Move Onwards Mean SD Mean SD Mean SD Dependent variable Aspire to stay 0.75 0.431 Aspire to return 0.08 0.267 Aspire to move onwards 0.17 0.376 Explanatory variables Demographics Age in years 46.689 15.883 41.525 15.371 37.417 13.905 Age 18-34 0.270 0.444 0.385 0.487 0.497 0.500 Age 35-49 0.284 0.451 0.300 0.458 0.287 0.452 Age 50-64 0.271 0.444 0.220 0.415 0.169 0.375 Age 65-75 0.175 0.380 0.095 0.293 0.047 0.212 Male 0.426 0.495 0.407 0.491 0.496 0.500 Married or live with domestic partner 0.637 0.481 0.589 0.492 0.524 0.499 Household size (continuous number) 3.547 2.512 3.733 2.596 4.120 2.987 Human Capital Elementary education 0.200 0.400 0.223 0.416 0.190 0.393 Secondary education 0.524 0.499 0.526 0.499 0.550 0.498 Tertiary education 0.277 0.447 0.251 0.434 0.260 0.439 Residence in host ≤ 5 years 0.163 0.369 0.253 0.435 0.248 0.432 Personal Health Index (0–100) 69.451 28.223 63.995 29.394 65.632 28.005 Spiritual Capital Christianity 0.572 0.495 0.595 0.491 0.540 0.498 Islam 0.205 0.403 0.269 0.444 0.289 0.453 Judaism 0.066 0.247 0.015 0.120 0.033 0.180 Other religions 0.043 0.204 0.027 0.161 0.036 0.187 No religion 0.114 0.318 0.095 0.293 0.101 0.301 Networks Network abroad 0.596 0.491 0.792 0.406 0.662 0.473 Local Network Index (0–1) 0.811 0.295 0.755 0.318 0.743 0.327 Individual Economic Indicators Employed 0.580 0.494 0.564 0.496 0.614 0.487 Basic Wealth Index (0–1) 0.777 0.268 0.718 0.280 0.708 0.298 Living comfortably on household income 0.260 0.439 0.212 0.409 0.185 0.388 Satisfaction with standard of living 0.652 0.476 0.547 0.498 0.494 0.500 Community Wellbeing Community Diversity Index (0–100) 58.411 35.349 53.230 35.474 52.695 34.017 Community Basics Index (0–100) 61.827 27.021 52.507 28.205 51.000 28.425 Macroeconomics GDP growth (Host) 2.476 3.812 2.424 4.210 2.510 4.295 GDP growth (Birth) 2.723 4.778 2.819 4.910 3.323 6.342 Environment Disaster frequency (Host) 1.517 2.472 1.304 2.300 1.286 1.809 Disaster frequency (Birth) 3.258 5.796 2.833 4.622 2.719 5.218 18 Table 1: Descriptive Statistics Aspire to Stay Return Move Onwards Mean SD Mean SD Mean SD Polity Political instability (-10 to 10) (Host) 0.045 0.208 0.055 0.228 0.067 0.250 Political instability (-10 to 10) (Birth) 0.059 0.236 0.052 0.221 0.081 0.273 Observations 21,153 2,174 4,777 Source: Authors’ calculations with GWP data. Notes: The number of observations is reduced when we add Networks, Individual Economic Indicators, and Community Wellbeing variables. The GWP Indices for Personal Health, Diversity and Community Basics vary between 0 and 100, while the first principal components of the Local network and Basic Wealth have been rescaled to vary between 0 and 1. Table 1 also shows a smaller percentage of newcomers among the stayers (16%) and a larger percentage among the out-migrants (25%). Overall, our sample has good health, although the health index is slightly lower among the returnees. Christianity is invariably the majority religion, followed by Islam, no religion, and Judaism. Among those who aspire to move onwards, we find a smaller share of Christians and a higher share of Muslims than among the stayers and the returnees. Noticeable is the higher percentage of Jews who want to stay in the host country, which is Israel for most of them, than to return or move onwards. Interesting are the high scores of the networks indicating how well connected immigrants are both locally and internationally. Among those who aspire to return we find the highest percentage of having networks abroad (79%). Our sample of stayers has an astonishing network in the host country as indicated by the high local network index score of 0.81. Although our entire sample includes the unemployed and those out of the labor force, we still observe high shares of employed immigrants across all three aspiring groups. A larger percentage of the employed is found among those who aspire to move to another host country. All three aspiration groups have a high index for basic wealth. This is because the index covers minimum material and financial wealth. While over a quarter of the aspiring stayers feel comfortable with the household income, only 19% feel comfortable among the onward movers. Among the stayers, 65% are satisfied with their standard of living but among the onward movers only 49% are satisfied. The community indices are higher among the stayers; more people believe that they live in accepting and progressive communities with good infrastructure that care for their residents and provide for their health and needs, compared to the two other groups. Finally, in the group of aspiring stayers we find the highest number for disaster frequency in the birth country. 19 To further elucidate the role of the host country on the three aspiring groups we calculate their shares by country of current residence. Figure 1 illustrates the shares of aspiring stayers (Panel A), returnees (Panel B) and onward movers (Panel C). Darker colors indicate higher shares of individual aspirations to stay, return or move onwards; lighter colors denote lower shares. First, it is evident that the lowest shares of the aspiring stayers are invariably found in Africa and Latin/South America; namely Liberia, Sierra Leone, Peru, and the Dominican Republic. Albania has the lowest shares in Europe. Second, the highest shares of aspiring returnees are also found in Africa and Latin America especially in the Central African Republic, Lesotho, Madagascar, Bolivia, and Venezuela. This is in stark contrast to no aspirations to return among immigrants living in other African countries (Mozambique, Tanzania, Ethiopia, and Angola). Similarly, we do not find aspiring returnees in several Asian host countries (Bangladesh, Bhutan, Indonesia, Laos, Mongolia, among others). Thus the share of immigrants aspiring to return varies widely by host country. Finally, Panel C shows that the highest shares of people aspiring to move to another country are in Albania, Peru, the Dominican Republic, Liberia, and Sierra Leone. Panel A: Aspiring Stayers by Host Country Panel B: Aspiring Returnees by Host Country 20 Panel C: Aspiring Onward Movers by Host Country Figure 1: Share of immigrants aspiring to stay, return, and move onwards by host countries Source: Authors’ elaboration on GWP data Notes: Darker colours indicate a higher share of individuals reporting an aspiration to, respectively, stay, return to their country of birth, or move onwards, while lighter colours denote lower shares. Figure A1 in the Appendix encapsulates the shares of immigrants aspiring to stay and return by birth country (Panels A and B). Immigrants from Bhutan, Cambodia, and Lesotho have the highest shares of aspiring to stay abroad. Those originating from Suriname, Australia, and Canada have the highest shares of aspiring to return to their countries. Panel C shows the share of migrants aspiring to move to another country by preferred destination, over the total number of migrants who want to move to another country. Clearly, most migrants aspiring to move to another country are aiming at OECD countries. The U.S. is by far the most attractive country for onward migration (17.55% declare the U.S. as their preferred destination), followed by Canada and Germany. 4. Theoretical Foundations and Methodology Our analysis is based on the aspirations or SP 11 of first generation immigrants with respect to staying in the destination, returning to their country of birth or migrating onwards to another country. The model that we bring to the data to analyze these migration aspirations is a random utility maximization (RUM) model of migration. Respondents maximize their utility by choosing the best alternative from the choice set. This maximization problem is unconstrained, 11 SP ask for discrete choices and are put in a behavioral choice context. As we argued in the introduction, they both complement and supplement revealed preferences, as it has been amply shown in the literature on consumer choices in market research and commuting behavior in transportation economics (Ben-Akiva et al., 2019). 21 based on free will. That is, if immigrants choose the option to move permanently to another country – either to their birth country or another country – this expresses their willingness to out-migrate, net of their ability to do so. 12 Hence, the choice set D contains three alternatives: staying (k=0), returning (k=1) and migrating onwards (k=2). The immigrants’ choice to stay in the host country is considered the default. It represents a steady state of happiness and contentment. It equally expresses a commitment to make their stay a success even if life in the host country may not be perfect. One possibility is that immigrants choose to stay as a reaction when the alternative of return is not feasible or appealing. It could also be that they do not want to go through the process of obtaining a new visa, go through a new migration journey again, or through a new integration process in another host country. The aspiration to return to the home country expresses a longing and nostalgia. This could be because it was always understood that they will return. Alternatively, they might want to return because they are unhappy in the host country and/or have a negative migration experience. Besides, circumstances in the home country may have changed, so that returning becomes an attractive option. Altruism to go back and help rebuild the country after a disaster could be another motive for the willingness to return. Finally, the aspiration to migrate to a new host country captures an optimism to try something new and different along with some risk loving tendencies. It is well known that immigrants move to places where there is demand for their skills. The move onwards could have been the plan all along, but it was not feasible to migrate to this new host country directly. It is also possible that people are unhappy with their life in the current host country and they hope for a better life in the new country. Or maybe they have already visited the new host country and want to go back to a strong social network. 13 It is important to note that, while this is a hypothetical scenario, for immigrants this scenario is meaningful and very close to reality. Immigrants who have migrated at least once are fully aware of these choices. It is plausible to assume that in the eyes of each immigrant decisionmaker, the three choices of staying, returning home, and moving onwards are distinct and weighted independently. These choices constitute, in other words, the entire universe of choices 12 Constrained utility maximization subject to time and budget constraints is related to RP or the actual move. RP have the advantage of avoiding potential problems associated with hypothetical responses. The drawback is their potentiality to fail to properly consider behavioral and budget constraints. When studying the actual move, the analysis is restricted to this particular observable action. When quantifying preferences for which the attribute cannot be observed or there is no variation, RP are not the appropriate technique. Conversely, these become strengths of SP, which reveal the first unconstrained best while constrained RP may reveal a second best. 13 The underlying assumption is that host countries as well as prospective third countries are more economically developed than home countries. 22 (i.e., a complete set of choices). In addition, once a move has taken place, immigrants are more prone to move again because migration-specific capital builds momentum and causes a selfperpetuating phenomenon (Constant and Zimmermann, 2011; 2012). Let Uikt denote the perceived utility that individual i would derive if opting for alternative k ∈ D at time t. We assume that this alternative-specific utility includes a deterministic component Vikt and a stochastic component εikt, i.e., Uikt = U(Vikt, εikt), conventionally simplified as an additive formulation Uikt = Vikt + εikt. If the stochastic component follows an independent and identically distributed Extreme Value Type 1 (EVT-1) distribution, then the probability pikt that k ∈ D will be the utility-maximizing alternative is given by: 𝑝𝑖𝑘𝑡 =𝑒𝑉𝑖𝑘𝑡 ∑𝑒𝑉𝑖𝑙𝑡 𝑙∈𝐷 . (1) The relative probability of returning to the birth country (k=1) over staying in the host country (k=0) writes as: 𝑝𝑖1𝑡 𝑝𝑖0𝑡 = 𝑒𝑉𝑖1𝑡−𝑉𝑖0𝑡. (2) The relative probability of migrating onwards (k=2) over staying in the host country (k=0) writes as: 𝑝𝑖2𝑡 𝑝𝑖0𝑡 = 𝑒𝑉𝑖2𝑡−𝑉𝑖0𝑡. (3) Hence, the relative probability of aspiring to leave the host country (irrespective of the destination) over staying is given by: 𝑝𝑖1𝑡+𝑝𝑖2𝑡 𝑝𝑖0𝑡 =𝑒𝑉𝑖1𝑡+𝑒𝑉𝑖2𝑡 𝑒𝑉𝑖0𝑡 . (4) Relative choice probabilities are solely determined by the difference in the levels of utility associated to each pair of alternatives (and not by the levels themselves). Hence, we can normalise the utility associated to the baseline option (staying) to zero. Thus, the estimated coefficient for all the regressors gives the differential effect of each variable on the attractiveness of moving versus staying. In the estimation analyses we control for influences that shape aspirations and their relationship to functioning à la Sen (1985). Assuming a Cobb-Douglas function of monotonic, convex, continuous, and transitive preferences, and denoting the home country index by h and the host country index by j, we can write the reduced-form expression for the utility differential between moving options and staying as: 𝑉𝑖ℎ𝑗𝑡 = 𝛼0+ 𝛾𝑋𝑖+ 𝛿𝑊ℎ𝑡−1 + 𝜆𝑍𝑗𝑡−1 + 𝛼𝑡+ 𝛼𝑗+ 𝜂𝑖ℎ𝑗𝑡. (5) where 𝑋𝑖 contains the individual and household characteristics and the variables measuring integration in the host country (network effects, economic integration and wellbeing), and 29 to migrate onwards. One possibility is that immigrants have already found a job abroad. It is also possible that they are confident to find a job when they move there. It is important to keep in mind that our sample is not restricted to working age individuals. 16 Oddly, our index for basic household wealth is not a significant determinant of out-migration, neither is feeling comfortable with the household income in the host country. Lastly, satisfaction with one’s standard of living significantly curtails the odds to aspire to out-migrate (return or move onwards) and thus increases the odds to aspire to stay. In Columns (4) we add the two indicators measuring immigrants’ wellbeing in their community. In line with expectations, both the Community Diversity and the Community Basics Indices have a negative and significant coefficient for either destination out of the host country. Specifically, immigrants are less likely to aspire to leave the host country when they feel they live in a community that is more open and acceptant of people from different racial, ethnic and cultural groups. Likewise, aspirations to return or move onwards are lower among immigrants who are more satisfied with everyday life in the community where they live, including opportunities for education, the environment, healthcare, housing, transportation, and infrastructure. These are important non-economic factors that make a strong difference in the immigrants’ decision-making. They remain significant even after we include more variables in Columns (5). GDP growth in the home country reveals an interesting, and at first sight, counterintuitive finding: it decreases the odds of aspiring to return home and increases the odds of aspiring to move onwards, ceteris paribus. We tie our explanation to the theory of relative deprivation (Stark and Taylor, 1991). Accordingly, when GDP in the birth country increases, everybody is better off in absolute terms. But immigrants rather care about their relative income compared to other households at home. Since their situation in the host country has not improved, they reckon that their status has deteriorated comparatively. Therefore, they choose to stay abroad and earn more money in order to improve their relative position back home. The same rationale explains why immigrants may want to go to a third country in hopes of a better job and better income that will enable them to be positioned comparatively better than their neighbors. Curiously, a higher frequency of natural disasters in the home country significantly increases the probability to aspire to return home versus staying in a safe host country. We advance the following justification. A higher frequency of devastating and catastrophic events 16 Leaving out employment status from the empirical specification does not alter the results for the remaining regressors. 30 may evoke acute altruism and a sense of national pride that prompts immigrants to go back home permanently, help their compatriots, and rebuild their country. Moreover, such a gesture could earn them the gratitude and recognition by their compatriots that sweetens the opportunity cost of return. At the same time, we find that a higher frequency of natural disasters in the home country significantly decreases the probability to aspire to move onwards. One explanation is that immigrants may need to send financial support to the affected country and so they need to stay in the host country and avoid venturing to new lands. As for political instability, we find that only instability in the host country has an influence and only on the aspiration to move onwards. This instability is not enough of a push to aspire a return home. 5.2. Estimates by Development Level Subsequently, we explore the extent to which results differ across groups of host countries that are at disparate economic development levels. Specifically, we differentiate between developing countries (low and middle-income) and developed countries (high-income) following the World Bank country classification. Tables 3 and 4 show the results of MNL for individuals residing in developing and developed countries, respectively, who aspire to return versus stay or move onwards versus stay. Most of the estimated findings are remarkably stable (at least in qualitative terms) across the different country groups. Aspirations to return or move onwards – when the alternative is to stay – among immigrants living in either developing (Table 3) or developed host countries (Table 4) are always lower for older age groups (compared to the 18-34 years old). Overall, this echoes our results in Table 2. In both developing and developed host countries, gender does not significantly influence aspirations to return versus stay (except for a significant and negative effect in Column 1, Panel A of Table 3). But male immigrants are more likely to aspire to move to another country instead of staying where they are. These are the same effects as in Table 2. Being married is not different than not being married in the probability to aspire to return versus stay in both Tables 3 and 4. As in Table 2, marriage makes a difference (negative) when the choice is between moving onwards and stay, no matter the host country. A few differences between Tables 3 and 4 are worth noting. Tertiary education renders immigrants more likely to aspire to return but only when they reside in low to middle-income countries. Newcomers are also more likely to express a willingness to move to a third country when they live in a low to middle-income country (Table 3) but not when they live in a highincome country (Table 4). 31 Table 3: Results by Host Country Development Level: Low to Middle-Income Countries Panel A. Aspire to Return versus Stay Panel B. Aspire to Move Onwards versus Stay Demograph ics, Human & Spiritual Capital (1) Networks (2) Individual Economic Indicators (3) Community Wellbeing (4) Macroeco nomics, Environm ent, Polity (5) Demograph ics, Human & Spiritual Capital (1) Networks (2) Individual Economic Indicators (3) Community Wellbeing (4) Macroeco nomics, Environm ent, Polity (5) Age 35-49 -0.242*** -0.188* -0.218** -0.230** -0.222** -0.394*** -0.387*** -0.387*** -0.384*** -0.376*** (-2.75) (-1.88) (-2.25) (-2.30) (-2.03) (-5.63) (-4.69) (-4.45) (-4.26) (-3.70) Age 50-64 -0.409*** -0.333** -0.384** -0.383** -0.381** -0.932*** -0.882*** -0.865*** -0.863*** -0.880*** (-2.91) (-2.13) (-2.52) (-2.42) (-2.30) (-10.23) (-7.55) (-7.03) (-6.90) (-6.63) Age 65-75 -0.718*** -0.658*** -0.757*** -0.787*** -0.727*** -1.914*** -1.849*** -1.866*** -1.853*** -1.797*** (-4.47) (-2.96) (-3.40) (-3.29) (-2.92) (-11.34) (-11.18) (-10.76) (-10.43) (-9.56) Age 18-34 - - - - - - - - - - Male -0.123** -0.107 -0.051 -0.065 -0.064 0.250*** 0.255*** 0.262*** 0.237*** 0.250*** (-2.08) (-1.21) (-0.54) (-0.68) (-0.64) (4.67) (4.25) (4.13) (3.74) (3.79) Married or domestic partner -0.045 -0.126 -0.132 -0.126 -0.051 -0.296*** -0.316*** -0.305*** -0.294*** -0.290*** (-0.59) (-1.41) (-1.42) (-1.31) (-0.51) (-5.05) (-4.39) (-3.65) (-3.52) (-3.28) Household size -0.014 -0.015 -0.020 -0.020 -0.022 0.013 0.006 0.002 -0.000 0.001 (-0.93) (-0.85) (-1.06) (-1.08) (-1.05) (1.24) (0.54) (0.23) (-0.04) (0.06) Secondary education 0.118 0.086 0.123 0.096 0.033 0.476*** 0.483*** 0.496*** 0.444*** 0.450*** (1.34) (0.83) (1.10) (0.83) (0.29) (7.70) (6.07) (6.40) (5.52) (5.33) Tertiary education 0.271** 0.301** 0.335** 0.338** 0.299* 0.729*** 0.711*** 0.726*** 0.637*** 0.636*** (2.24) (1.96) (2.31) (2.40) (1.96) (7.35) (6.50) (6.69) (5.86) (6.00) Elementary education - - - - - - - - - - Residence in host ≤ 5 years 0.320*** 0.305** 0.308*** 0.316*** 0.297** 0.147** 0.078 0.115 0.147** 0.156** (3.19) (2.54) (2.72) (2.65) (2.31) (2.58) (1.15) (1.62) (2.09) (2.11) Personal Health Index -0.006*** -0.006*** -0.006*** -0.006*** -0.005*** -0.006*** -0.006*** -0.005*** -0.004*** -0.003*** (-4.67) (-5.22) (-5.12) (-4.45) (-3.97) (-7.92) (-6.40) (-4.34) (-3.33) (-2.61) Christianity 0.071 0.227 0.308 0.388 0.379 -0.441*** -0.356** -0.385*** -0.363** -0.407*** (0.33) (0.82) (1.19) (1.47) (1.43) (-3.53) (-2.36) (-2.65) (-2.40) (-2.70) Islam -0.244 0.039 0.191 0.318 0.288 -0.622*** -0.442** -0.428** -0.376** -0.437** (-0.85) (0.13) (0.63) (1.09) (0.97) (-3.84) (-2.36) (-2.37) (-2.06) (-2.41) Other religions (+ Judaism) -0.219 0.114 0.103 0.100 0.125 -0.389* -0.209 -0.191 -0.161 -0.162 (-0.60) (0.28) (0.23) (0.21) (0.25) (-1.82) (-0.93) (-0.85) (-0.69) (-0.70) 32 Table 3: Results by Host Country Development Level: Low to Middle-Income Countries Panel A. Aspire to Return versus Stay Panel B. Aspire to Move Onwards versus Stay Demograph ics, Human & Spiritual Capital (1) Networks (2) Individual Economic Indicators (3) Community Wellbeing (4) Macroeco nomics, Environm ent, Polity (5) Demograph ics, Human & Spiritual Capital (1) Networks (2) Individual Economic Indicators (3) Community Wellbeing (4) Macroeco nomics, Environm ent, Polity (5) No religion - - - - - - - - - - Network abroad 1.094*** 1.111*** 1.064*** 1.062*** 0.350*** 0.417*** 0.397*** 0.415*** (8.70) (8.71) (8.06) (7.49) (4.98) (5.72) (5.29) (5.14) Local Network Index -0.573*** -0.477*** -0.293 -0.294 -0.280** -0.150 0.012 0.024 (-3.56) (-2.78) (-1.55) (-1.41) (-2.48) (-1.26) (0.10) (0.19) Employed -0.125 -0.104 -0.105 -0.020 0.012 -0.000 (-1.23) (-0.99) (-0.97) (-0.29) (0.17) (-0.00) Basic Wealth Index 0.028 0.030 0.062 0.018 0.012 -0.011 (0.16) (0.18) (0.35) (0.11) (0.07) (-0.06) Living comfortably on HH income 0.053 0.125 0.194 -0.042 -0.026 -0.030 (0.35) (0.85) (1.27) (-0.41) (-0.24) (-0.26) Satisfaction w. standard of living -0.376*** -0.335*** -0.363*** -0.390*** -0.299*** -0.322*** (-3.52) (-3.04) (-3.01) (-4.67) (-3.67) (-3.68) Community Diversity Index -0.003* -0.003** -0.001 -0.001 (-1.94) (-2.00) (-0.95) (-0.86) Community Basics Index -0.006*** -0.005*** -0.008*** -0.008*** (-3.20) (-2.92) (-6.06) (-5.69) GDP growth (Host) -0.014 0.023*** (-0.70) (2.60) GDP growth (Birth) -0.023** 0.012** (-2.01) (2.20) Disaster frequency (Host) -0.063 -0.048 (-1.37) (-1.52) Disaster frequency (Birth) 0.039*** -0.015 (3.21) (-1.13) Political instability (Host) 0.099 0.242* (0.47) (1.78) Political instability (Birth) -0.067 0.004 33 Table 3: Results by Host Country Development Level: Low to Middle-Income Countries Panel A. Aspire to Return versus Stay Panel B. Aspire to Move Onwards versus Stay Demograph ics, Human & Spiritual Capital (1) Networks (2) Individual Economic Indicators (3) Community Wellbeing (4) Macroeco nomics, Environm ent, Polity (5) Demograph ics, Human & Spiritual Capital (1) Networks (2) Individual Economic Indicators (3) Community Wellbeing (4) Macroeco nomics, Environm ent, Polity (5) (-0.35) (0.03) Constant -2.096*** -2.878*** -2.872*** -2.760*** -2.614*** -0.431* -0.441 -0.454 -0.217 -0.239 (-4.34) (-5.46) (-5.63) (-5.75) (-5.21) (-1.84) (-1.53) (-1.61) (-0.77) (-0.63) Log likelihood -8917.462 -5928.337 -5469.263 -5256.858 -4790.295 -8917.462 -5928.337 -5469.263 -5256.858 -4790.295 Observations 12,517 8,480 7,855 7,570 6,998 12,517 8,480 7,855 7,570 6,998 Countries 97 97 96 94 89 97 97 96 94 89 Notes: The table displays estimated coefficients for our benchmark model in which we restrict the sample to respondents residing in low and middle-income countries. Each model includes host country and year fixed effects. t-statistics in parentheses. Standard errors are robust to heteroskedasticity and serial correlation and clustered by host country. *, **, *** denote significance at the 1, 5 and 10% confidence level, respectively. Table 4: Results by Host Country Development Level: High-Income Countries Panel A. Aspire to Return versus Stay Panel B. Aspire to Move Onwards versus Stay Demograph ics, Human & Spiritual Capital (1) Networks (2) Individual Economic Indicators (3) Community Wellbeing (4) Macroeco nomics, Environm ent, Polity (5) Demograph ics, Human & Spiritual Capital (1) Networks (2) Individual Economic Indicators (3) Community Wellbeing (4) Macroeco nomics, Environm ent, Polity (5) Age 35-49 -0.149* -0.116 -0.111 -0.111 -0.139 -0.462*** -0.424*** -0.431*** -0.437*** -0.425*** (-1.65) (-0.84) (-0.80) (-0.78) (-0.98) (-7.65) (-4.36) (-4.43) (-4.57) (-4.33) Age 50-64 -0.329*** -0.391** -0.392** -0.413** -0.426** -0.839*** -0.916*** -0.961*** -0.971*** -0.981*** (-3.30) (-2.37) (-2.31) (-2.39) (-2.49) (-9.79) (-9.62) (-9.74) (-10.28) (-9.70) Age 65-75 -0.742*** -0.687*** -0.786*** -0.766*** -0.783*** -1.549*** -1.611*** -1.647*** -1.640*** -1.635*** (-4.63) (-3.60) (-3.74) (-3.56) (-3.71) (-14.04) (-8.55) (-7.90) (-7.90) (-8.07) Age 18-34 - - - - - - - - - - Male -0.050 0.063 0.065 0.063 0.065 0.264*** 0.340*** 0.341*** 0.334*** 0.374*** (-0.82) (0.69) (0.72) (0.69) (0.69) (5.77) (4.97) (4.67) (4.61) (5.45) Married or domestic partner -0.127 -0.175 -0.168 -0.162 -0.144 -0.395*** -0.375*** -0.374*** -0.376*** -0.352*** (-1.59) (-1.36) (-1.32) (-1.22) (-1.04) (-8.75) (-5.25) (-5.25) (-5.06) (-4.82) 34 Table 4: Results by Host Country Development Level: High-Income Countries Panel A. Aspire to Return versus Stay Panel B. Aspire to Move Onwards versus Stay Demograph ics, Human & Spiritual Capital (1) Networks (2) Individual Economic Indicators (3) Community Wellbeing (4) Macroeco nomics, Environm ent, Polity (5) Demograph ics, Human & Spiritual Capital (1) Networks (2) Individual Economic Indicators (3) Community Wellbeing (4) Macroeco nomics, Environm ent, Polity (5) Household size 0.005 0.041 0.041 0.051 0.034 0.023 0.020 0.019 0.026 0.004 (0.22) (1.16) (1.23) (1.54) (1.04) (1.19) (0.57) (0.56) (0.73) (0.11) Secondary education -0.011 -0.092 -0.062 -0.059 -0.044 0.394*** 0.332** 0.333** 0.346** 0.362** (-0.10) (-0.53) (-0.35) (-0.32) (-0.24) (4.77) (2.42) (2.34) (2.41) (2.53) Tertiary education -0.062 -0.101 -0.025 -0.065 -0.067 0.522*** 0.403*** 0.483*** 0.472*** 0.483*** (-0.40) (-0.55) (-0.13) (-0.33) (-0.33) (5.22) (3.60) (4.03) (3.84) (3.89) Elementary education - - - - - - - - - - Residence in host ≤ 5 years 0.368*** 0.216 0.203* 0.232* 0.230* 0.086 -0.020 -0.074 -0.028 -0.039 (5.07) (1.59) (1.65) (1.83) (1.87) (1.50) (-0.20) (-0.84) (-0.31) (-0.42) Personal Health Index -0.008*** -0.007*** -0.006*** -0.005*** -0.006*** -0.009*** -0.006*** -0.003** -0.002 -0.003** (-6.48) (-4.04) (-3.20) (-2.75) (-2.68) (-8.96) (-6.16) (-2.52) (-1.46) (-1.99) Christianity -0.272** -0.384** -0.404** -0.373** -0.410** -0.250** -0.280** -0.268** -0.235* -0.272** (-2.14) (-2.36) (-2.50) (-2.35) (-2.53) (-2.38) (-2.20) (-2.08) (-1.90) (-2.09) Islam -0.063 -0.113 -0.133 -0.124 -0.008 -0.178 -0.167 -0.151 -0.114 -0.195 (-0.23) (-0.34) (-0.43) (-0.44) (-0.03) (-1.48) (-1.21) (-1.08) (-0.73) (-1.12) Judaism -0.573 -1.187*** -1.120*** -1.169*** -1.205*** -0.899** -1.012* -0.928 -1.006 -1.083* (-0.75) (-3.06) (-2.82) (-2.76) (-2.95) (-2.52) (-1.67) (-1.41) (-1.47) (-1.83) Other religions -0.215 -0.295 -0.282 -0.242 -0.353 0.139 0.040 0.022 0.055 0.024 (-0.98) (-1.23) (-1.20) (-1.03) (-1.42) (1.35) (0.24) (0.12) (0.31) (0.12) No religion - - - - - - - - - - Network abroad 1.086*** 1.132*** 1.139*** 1.130*** 0.284** 0.334*** 0.338*** 0.360*** (7.44) (7.76) (7.85) (7.67) (2.42) (3.02) (3.08) (3.02) Local Network Index -0.787*** -0.707*** -0.531*** -0.532*** -1.152*** -1.008*** -0.823*** -0.837*** (-3.91) (-3.59) (-2.89) (-2.82) (-13.22) (-11.60) (-9.07) (-8.67) Employed -0.100 -0.096 -0.089 -0.008 -0.014 0.000 (-1.00) (-0.94) (-0.88) (-0.09) (-0.14) (0.00) Basic Wealth Index -0.387 -0.396 -0.445 -0.028 -0.021 0.004 (-1.30) (-1.40) (-1.58) (-0.12) (-0.08) (0.02) 35 Table 4: Results by Host Country Development Level: High-Income Countries Panel A. Aspire to Return versus Stay Panel B. Aspire to Move Onwards versus Stay Demograph ics, Human & Spiritual Capital (1) Networks (2) Individual Economic Indicators (3) Community Wellbeing (4) Macroeco nomics, Environm ent, Polity (5) Demograph ics, Human & Spiritual Capital (1) Networks (2) Individual Economic Indicators (3) Community Wellbeing (4) Macroeco nomics, Environm ent, Polity (5) Living comfortably on HH income 0.046 0.087 0.086 -0.161 -0.128 -0.102 (0.42) (0.77) (0.80) (-1.56) (-1.28) (-1.11) Satisfaction w. standard of living -0.147 -0.077 -0.073 -0.555*** -0.508*** -0.499*** (-1.47) (-0.83) (-0.76) (-6.81) (-5.78) (-5.62) Community Diversity Index -0.003 -0.002 -0.004*** -0.004*** (-1.20) (-1.10) (-3.80) (-3.61) Community Basics Index -0.011*** -0.011*** -0.009*** -0.008*** (-3.16) (-3.23) (-5.05) (-4.65) GDP growth (Host) -0.010 -0.019 (-0.39) (-0.80) GDP growth (Birth) -0.012 0.018** (-0.64) (1.98) Disaster frequency (Host) 0.052 0.039 (0.81) (0.66) Disaster frequency (Birth) 0.014 -0.014 (1.30) (-1.19) Political instability (Host) 0.000 0.000 (.) (.) Political instability (Birth) -1.144** -0.182 (-2.20) (-0.93) Constant -1.946*** -2.070*** -1.764*** -1.154** -1.327** -1.878*** -1.320*** -1.130*** -0.532* -0.696 (-7.94) (-5.89) (-3.86) (-2.00) (-2.18) (-6.38) (-5.81) (-4.43) (-1.82) (-1.59) Log likelihood -9173.213 -4398.275 -4197.092 -4154.763 -3968.872 -9173.213 -4398.275 -4197.092 -4154.763 -3968.872 Observations 15,587 7,735 7,461 7,451 7,124 15,587 7,735 7,461 7,451 7,124 Countries 41 41 41 41 37 41 41 41 41 37 Notes: The table displays estimated coefficients for our benchmark model in which we restrict the sample to respondents residing in high income countries. Each model includes host country and year fixed effects. t-statistics in parentheses. Standard errors are robust to heteroskedasticity and serial correlation and clustered by host country. *, **, *** denote significance at the 1, 5 and 10% confidence level, respectively. 36 Results on religion are the same as in Table 2 when immigrants live in developing countries. 17 However, the role of religion is somewhat different when immigrants live in highincome countries. We find that Christians and Jews are less likely to choose to out-migrate versus stay, compared to the no religion individuals. Muslims and people of other religions are not significantly different than people with no religion. While local networks act as a retainer when people live in a high-income host country, this is not the case when they live in low to middle-income countries. It is interesting that the effect of local networks is not stable across the model specifications in Table 3. Another notable difference is found in the satisfaction with one’s standard of living, which loses its significant effect on the probability to want to return when one lives in high-income host countries (Table 4). Moreover, the effect of the Diversity Index is not significant for aspirations to move onwards among immigrants in low to middle-income countries and for aspirations to return among immigrants in high-income countries. Remarkably, in Table 3, we find that GDP growth in the host country stimulates aspirations to move onwards but only when immigrants live in low to middle-income countries. Our interpretation is that as immigrants benefit from economic prosperity in the host country, they can afford to move on (or closer) to a better country or their preferred destination. We also remark that GDP growth in the birth country – when one lives in a high-income country – has no significant effect in the choice set return versus stay. Immigrants in developed countries earn sufficiently more so that the relative deprivation theory is not applicable. Finally, it is only in the low to middle-income countries that immigrants have higher aspirations to return home after environmental disasters in their home country. This is probably what drives the result in Table 2. We conjecture that in high-income host countries immigrants earn higher salaries, thus they can send more money home via remittances as a substitute for their presence. Immigrants in developing countries may not have this luxury. The last difference between host countries development levels is related to the political instability in the host country. In developing countries, political instability in the host prompts immigrants to move onwards whereas in developed countries this is not a significant determinant. In sum, preferences to stay or out-migrate are highly selective, and selection changes depending on the context. Worth noting is that employment and other individual economic indicators are not part of the decision-making process and choice of contrasts. 17 Note that due to the very small number of Jewish immigrants living in a developing country we combined these observations with the “other religion.” The overwhelming majority of Jewish immigrants in a developed country lives in Israel. 37 5.3. Robustness Checks A possible concern related to our estimation strategy may be that some countries have only a limited number of observations, which may bias our results. Indeed, fixed effects logit models may produce biased coefficients and standard errors due to the incidental parameter problem (Neyman and Scott, 1948). To mitigate this concern, we re-estimate the empirical specifications presented in Table 2 dropping from our sample countries of residence with less than 16 observations (the threshold to obtain unbiased estimates, defined by Katz, 2001). We present these estimated coefficients in the Appendix Table B1. Reassuringly, we obtain similar results to those reported above in both qualitative and quantitative terms. Additionally, given that we do not differentiate by concrete destination for onward migration, we cannot include destination fixed effects in our multinomial logit setting. As an alternative, we therefore estimate two separate binary logit models; one for aspiring to return versus staying and one for aspiring to migrate onwards versus staying. In the former we can now also include country of birth fixed effects alongside the host country and year fixed effects. Our results are qualitatively robust to this alternative fixed effects structure and allay our concerns. Results from this exercise are readily available upon request. 5.4. Estimates for Aspirations to Return versus Migrating Onwards In this section we seize the opportunity to analyze novel results from the contrast aspire to return home versus moving onwards. In this hypothetical scenario, we study aspiring outmigrants and how they perceive moving to different locations outside the host country. Because selection depends on the forgone choice and context, this analysis helps us check the sensitivity of our results. We apply a standard binary logit framework and present the results from this exercise in Table B2 in the Appendix. The table follows the same format of building up our models as Table 2 and pinpoints stark differences with Panel A in Table 2. Results on age show that older immigrants are invariably more likely to aspire to return home when the alternative is to move to another country, compared to the 18-34 years old. This finding is robust across all model specifications and increases with older age monotonically. This different selection in age (compared to Table 2) reveals the important role of the forgone choice. This is attributable to two facts. First, as one ages it becomes more difficult to relocate to a new country. According to human capital theory, comparatively older immigrants will have lower potential returns from any human capital investment and greater psychic costs. 38 Second, returning to one’s homeland is more appealing and related to retirement. Men are less likely to aspire to return home and would rather move onwards. This is consistent with the riskloving tendencies of men compared to women, as we demonstrated in Table 2. However, marriage and cohabitation strongly increase the probability to aspire to return compared to moving onwards. This result reveals intricate family decision-making that changes when the alternative changes. While first generation immigrants were less likely to aspire to return when the alternative was to stay in the host country, they are now more likely to aspire to return instead of moving onwards. This also highlights risk averse tendencies for married people, ceteris paribus. Next, we find that better educated immigrants, compared to those with elementary schooling, have significantly lower aspirations to return when the alternative is to move onwards. In agreement with human capital theory, educated people seek to move to places where they can match their talents and receive the best remuneration they can. They have a wider vision for the opportunities in a third country and are more marketable. This choice probably also indicates that immigrants know that their education is not valued nor rewarded in their home country, or that the home country is not ready yet to appreciate them. As established in Table 2, newcomers who arrived in the host country within the past five years are more likely to aspire to return than to move onwards, ceteris paribus. This finding shows that those who cannot make it in the host country or cannot cope with the “foreign land” are eager to return rather than seek new opportunities elsewhere. Having close ties and a solid network abroad to rely on (see footnote 9) acts as a booster to return aspirations in this choice set, as it was in Table 2. The power of these networks remains strong in all specifications. In contrast, it is interesting that, in general, economic factors are not significant in this exercise. Shifts in circumstances in the birth country such as GDP growth influence aspirations to return negatively. This is a recurrent finding (Table 2, Panel A) and attributable to the theory of relative deprivation. Accordingly, it is not enough to be successful and enjoy financial security in absolute terms in order to return. One needs to feel successful in relative terms as well. A GDP growth that “raises all boats” in the home country leaves immigrants behind in the income distribution, and unwilling to return. Controlling for everything else, people aspire to return to their homeland when it experiences high frequency of environmental disasters, similar to Table 2. This is irrespective of the alternative choice and explained by the altruistic behavior of immigrants when their birth country is in danger. 45 Snel, Erik, Marije Faber, and Godfried Engbersen. 2015. “To stay or Return? Explaining Return Intentions of Central and Eastern European Labour Migrants.” Central and Eastern European Migration Review 4(2):5–24. Stark, Oded, and J. Edward Taylor. 1991. “Migration Incentives, Migration Types: the Role of Relative Deprivation.” Economic Journal 101(408):1163–1178. Tezcan, Tolga. (2019). “Return Home? Determinants of Return Migration Intention amongst Turkish Immigrants in Germany.” Geoforum 98:189–201. 46 Appendix A. Appendix Figures Panel A. Aspiring Stayers by Birth Country Panel B. Aspiring Returnees by Birth Country Panel C. Aspiring Onward Movers by Destination Country Figure A1: Share of immigrants aspiring to stay or return by birth, and move onwards by host Source: Authors’ elaboration on GWP data. Notes: Darker colours indicate a higher share of individuals reporting an aspiration to, respectively, stay, return to their country of birth, or move onwards, while lighter colours denote lower shares. 47 B. Appendix Tables Table B1: Probabilities to Aspire to Return or Move Onwards versus Stay(after dropping countries with few observations) Panel A. Aspire to Return versus Stay Panel B. Aspire to Move Onwards versus Stay Demograph ics, Human & Spiritual Capital (1) Networks (2) Individual Economic Indicators (3) Community Wellbeing (4) Macroeco nomics, Environm ent, Polity (5) Demograph ics, Human & Spiritual Capital (1) Networks (2) Individual Economic Indicators (3) Community Wellbeing (4) Macroeco nomics, Environm ent, Polity (5) Age 35-49 -0.210*** -0.164** -0.180** -0.183** -0.188** -0.434*** -0.393*** -0.390*** -0.389*** -0.378*** (-3.36) (-2.08) (-2.25) (-2.24) (-2.19) (-9.52) (-6.43) (-6.22) (-6.04) (-5.38) Age 50-64 -0.393*** -0.388*** -0.420*** -0.431*** -0.425*** -0.889*** -0.895*** -0.901*** -0.905*** -0.916*** (-4.90) (-3.44) (-3.78) (-3.77) (-3.63) (-14.16) (-11.78) (-11.41) (-11.38) (-10.90) Age 65-75 -0.770*** -0.720*** -0.828*** -0.837*** -0.806*** -1.717*** -1.731*** -1.758*** -1.748*** -1.715*** (-6.92) (-4.92) (-5.49) (-5.34) (-5.06) (-18.47) (-14.31) (-13.66) (-13.52) (-12.91) Age 18-34 - - - - - - - - - - Male -0.083* -0.024 0.013 0.005 0.007 0.257*** 0.298*** 0.306*** 0.293*** 0.319*** (-1.93) (-0.37) (0.20) (0.07) (0.10) (7.40) (6.66) (6.57) (6.27) (6.95) Married or domestic partner -0.085 -0.141** -0.141* -0.135* -0.095 -0.343*** -0.346*** -0.342*** -0.333*** -0.325*** (-1.62) (-1.96) (-1.89) (-1.74) (-1.18) (-9.12) (-6.72) (-5.95) (-5.72) (-5.34) Household size -0.012 -0.006 -0.009 -0.007 -0.012 0.013 0.007 0.004 0.003 0.001 (-0.88) (-0.39) (-0.57) (-0.46) (-0.71) (1.45) (0.62) (0.41) (0.26) (0.11) Secondary education 0.086 0.051 0.088 0.063 0.031 0.459*** 0.465*** 0.469*** 0.425*** 0.432*** (1.25) (0.56) (0.92) (0.65) (0.31) (8.79) (6.42) (6.65) (5.91) (5.82) Tertiary education 0.096 0.120 0.189 0.151 0.103 0.626*** 0.602*** 0.643*** 0.571*** 0.571*** (0.96) (0.99) (1.56) (1.24) (0.81) (8.98) (7.38) (7.97) (6.92) (7.02) Elementary education - - - - - - - - - - Residence in host ≤ 5 years 0.339*** 0.276*** 0.268*** 0.286*** 0.276*** 0.127*** 0.060 0.070 0.103* 0.107* (5.15) (2.98) (3.10) (3.14) (2.90) (3.07) (1.06) (1.23) (1.82) (1.81) Personal Health Index -0.007*** -0.007*** -0.006*** -0.006*** -0.005*** -0.007*** -0.006*** -0.004*** -0.003*** -0.003*** (-7.90) (-6.64) (-5.89) (-5.10) (-4.77) (-11.39) (-8.92) (-5.23) (-3.89) (-3.56) Christianity -0.175 -0.228 -0.228 -0.199 -0.226 -0.254*** -0.283*** -0.287*** -0.259** -0.307*** (-1.48) (-1.61) (-1.63) (-1.42) (-1.58) (-2.75) (-2.70) (-2.70) (-2.48) (-2.89) Islam -0.244 -0.243 -0.189 -0.126 -0.107 -0.309*** -0.288** -0.255** -0.202* -0.277** (-1.09) (-1.10) (-0.88) (-0.60) (-0.49) (-2.89) (-2.31) (-2.08) (-1.66) (-2.21) 48 Table B1: Probabilities to Aspire to Return or Move Onwards versus Stay(after dropping countries with few observations) Panel A. Aspire to Return versus Stay Panel B. Aspire to Move Onwards versus Stay Demograph ics, Human & Spiritual Capital (1) Networks (2) Individual Economic Indicators (3) Community Wellbeing (4) Macroeco nomics, Environm ent, Polity (5) Demograph ics, Human & Spiritual Capital (1) Networks (2) Individual Economic Indicators (3) Community Wellbeing (4) Macroeco nomics, Environm ent, Polity (5) Judaism -0.415 -0.874 -0.792 -0.828 -0.889 -0.751* -0.791 -0.719 -0.774 -0.855 (-0.57) (-1.53) (-1.32) (-1.31) (-1.44) (-1.68) (-1.01) (-0.87) (-0.91) (-1.11) Other religions -0.250 -0.256 -0.282 -0.276 -0.360 0.023 -0.057 -0.056 -0.016 -0.052 (-1.30) (-1.21) (-1.32) (-1.25) (-1.52) (0.24) (-0.41) (-0.39) (-0.11) (-0.34) No religion - - - - - - - - - - Network abroad 1.095*** 1.122*** 1.094*** 1.092*** 0.332*** 0.394*** 0.384*** 0.404*** (11.25) (11.60) (11.12) (10.49) (5.62) (6.67) (6.38) (6.19) Local Network Index -0.663*** -0.571*** -0.388*** -0.385*** -0.603*** -0.479*** -0.317*** -0.325*** (-5.20) (-4.38) (-2.88) (-2.71) (-5.87) (-4.53) (-2.88) (-2.86) Employed -0.127* -0.112 -0.112 -0.016 0.002 -0.008 (-1.72) (-1.49) (-1.46) (-0.29) (0.04) (-0.14) Basic Wealth Index -0.091 -0.081 -0.085 0.052 0.050 0.029 (-0.60) (-0.56) (-0.57) (0.40) (0.36) (0.21) Living comfortably on HH income 0.040 0.093 0.121 -0.096 -0.077 -0.068 (0.43) (1.01) (1.29) (-1.27) (-1.00) (-0.87) Satisfaction w. standard of living -0.301*** -0.237*** -0.249*** -0.457*** -0.384*** -0.398*** (-3.87) (-3.14) (-3.07) (-7.29) (-5.87) (-5.86) Community Diversity Index -0.003** -0.003** -0.003*** -0.002*** (-2.27) (-2.18) (-3.10) (-2.79) Community Basics Index -0.008*** -0.008*** -0.008*** -0.008*** (-4.54) (-4.35) (-7.54) (-6.93) GDP growth (Host) -0.009 0.009 (-0.51) (0.91) GDP growth (Birth) -0.018** 0.014*** (-1.99) (3.18) Disaster frequency (Host) -0.017 -0.022 (-0.44) (-0.83) Disaster frequency (Birth) 0.027*** -0.015* (3.29) (-1.70) 49 Table B1: Probabilities to Aspire to Return or Move Onwards versus Stay(after dropping countries with few observations) Panel A. Aspire to Return versus Stay Panel B. Aspire to Move Onwards versus Stay Demograph ics, Human & Spiritual Capital (1) Networks (2) Individual Economic Indicators (3) Community Wellbeing (4) Macroeco nomics, Environm ent, Polity (5) Demograph ics, Human & Spiritual Capital (1) Networks (2) Individual Economic Indicators (3) Community Wellbeing (4) Macroeco nomics, Environm ent, Polity (5) Political instability (Host) 0.138 0.223* (0.63) (1.68) Political instability (Birth) -0.262 -0.039 (-1.36) (-0.38) Constant -2.121*** -2.407*** -2.337*** -2.127*** -2.166*** -0.633*** -0.455** -0.533** -0.276 -0.263 (-5.98) (-6.58) (-6.75) (-6.24) (-4.86) (-2.85) (-1.98) (-2.33) (-1.19) (-0.86) Log likelihood -18093.548 - 10334.707 -9680.394 -9434.011 -8791.051 -18093.548 -10334.707 -9680.394 -9434.011 -8791.051 Observations 28,005 16,131 15,237 14,942 14,051 28,005 16,131 15,237 14,942 14,051 Countries 123 123 122 120 111 123 123 122 120 111 Notes: The table displays estimated coefficients when we restrict the sample to countries with over 16 observations. Each model includes host country and year fixed effects. t-statistics are in parentheses. Standard errors are robust to heteroskedasticity and serial correlation and clustered by host country. *, **, *** denote significance at the 1, 5 and 10% confidence level, respectively. 50 Table B2: Probabilities on Aspirations to Return versus Migrating Onwards Demographics, Human & Spiritual Capital Networks Individual Economic Indicators Community Wellbeing Macroeco nomics, Environment, Polity Age 35-49 0.182*** 0.227** 0.232** 0.226** 0.207* (2.60) (2.31) (2.22) (2.15) (1.87) Age 50-64 0.424*** 0.462*** 0.464*** 0.459*** 0.460*** (4.32) (3.33) (3.17) (3.13) (3.01) Age 65-75 0.874*** 0.901*** 0.838*** 0.838*** 0.841*** (6.07) (4.97) (4.56) (4.50) (4.30) Age 18-34 - - - - - Male -0.347*** -0.313*** -0.275*** -0.268*** -0.267*** (-6.10) (-3.91) (-3.31) (-3.20) (-2.94) Married or with domestic partner 0.230*** 0.174** 0.166* 0.156* 0.177** (3.46) (2.09) (1.95) (1.81) (2.00) Household size -0.023** -0.016 -0.015 -0.014 -0.018 (-2.01) (-1.17) (-1.03) (-0.98) (-1.20) Secondary education -0.445*** -0.487*** -0.437*** -0.433*** -0.445*** (-6.10) (-5.17) (-4.05) (-3.95) (-3.93) Tertiary education -0.600*** -0.530*** -0.477*** -0.459*** -0.477*** (-4.84) (-3.89) (-3.33) (-3.25) (-3.44) Elementary education - - - - - Residence in host ≤ 5 years 0.213*** 0.209** 0.200** 0.175* 0.141 (2.80) (2.01) (2.02) (1.71) (1.31) Personal Health Index 0.000 -0.001 -0.002 -0.002 -0.002 (0.25) (-0.76) (-1.48) (-1.61) (-1.41) Christianity 0.082 0.053 0.041 0.041 0.044 (0.63) (0.34) (0.24) (0.24) (0.25) Islam 0.124 0.151 0.139 0.132 0.208 (0.59) (0.66) (0.59) (0.55) (0.84) Judaism -0.052 -0.519 -0.485 -0.473 -0.479 (-0.17) (-1.27) (-1.15) (-1.11) (-1.16) Other religions -0.297 -0.213 -0.270 -0.302 -0.379 (-1.33) (-0.90) (-1.04) (-1.13) (-1.24) No religion - - - - - Network abroad 0.735*** 0.696*** 0.694*** 0.659*** (7.50) (6.93) (6.96) (6.33) Local Network Index 0.064 0.040 0.036 0.028 (0.45) (0.28) (0.25) (0.18) Employed -0.163* -0.158 -0.154 (-1.70) (-1.64) (-1.54) Basic Wealth Index -0.114 -0.086 -0.065 (-0.48) (-0.36) (-0.27) Living comfortably on HH income 0.086 0.094 0.132 (0.74) (0.79) (1.06) Satisfaction with standard of living 0.159 0.146 0.156 (1.43) (1.34) (1.33) Community Diversity Index -0.000 -0.000 (-0.06) (-0.13) Community Basics Index 0.001 0.001 (0.70) (0.57) GDP growth (Host) -0.014 (-0.62) 51 Table B2: Probabilities on Aspirations to Return versus Migrating Onwards Demographics, Human & Spiritual Capital Networks Individual Economic Indicators Community Wellbeing Macroeco nomics, Environment, Polity GDP growth (Birth) -0.035*** (-2.93) Disaster frequency (Host) -0.001 (-0.02) Disaster frequency (Birth) 0.048*** (3.39) Political instability (Host) -0.174 (-0.84) Political instability (Birth) -0.086 (-0.38) Constant -1.598*** -2.105*** -1.908*** -1.924*** -1.964*** (-5.13) (-6.47) (-5.86) (-5.83) (-4.36) Log likelihood -3869.386 -2220.710 -2097.256 -2062.368 -1908.096 Observations 6,875 3,999 3,762 3,691 3,418 Countries 117 111 110 107 99 Notes: The table displays estimated coefficients for the logit model for aspirations to return versus migrating to another country. Each model includes host country and year fixed effects. t-statistics in parentheses. Standard errors are robust to heteroskedasticity and serial correlation and clustered by host country. *, **, *** denote significance at the 1, 5 and 10% confidence level, respectively. TheUNU‐MERITWORKINGPaperSeries  2021-01 Transformationtowardssustainaledevelopmentgoals:Roleofinnovationby MichikoIizukaandGeraldHane 2021-02 Accesstosocialprotectionforplatformandothernon‐standardworkers:A literaturereviewbyTamaraA.Kool,GiulioBordonandFranziskaGassmann 2021-03 Labour‐augmentingtechnicalchangedataforalternativeelasticitiesof substitution,growth,slowdown,anddistributiondynamicsbyThomasZiesemer 2021-04 DemocracyandCOVID‐19OutcomesbyGokhanKarabulut,KlausF.Zimmermann, MehmetHuseyinBilginandAsliCansinDoker 2021-05 Patentlandscapingusing'green'technologicaltrajectoriesbyÖnderNomaler& BartVerspagen 2021-06 Macroeconomicstimuluspackagesandincomeinequalityindevelopingcountries: Lessonsfromthe2007‐9GreatRecessionfortheCovid‐19crisisinSouthAfricaby AlexisHabiyaremye,PeterJacobs,OlebogengMolewaandPelontleLekomanyane 2021-07 Deepeningordelinking?Innovativecapacityandglobalvaluechainparticipationin theICTsectorsbyRasmusLema,CarloPietrobelli,RobertaRabellottiandAntonio Vezzani 2021-08 COVID‐19policyresponses,mobility,andfoodprices:Evidencefromlocalmarkets in47lowtomiddleincomecountriesbyStephanDietrich,ValerioGiuffrida,Bruno MartoranoandGeorgSchmerzeck 2021-09 AttitudestowardsinequalityinBrazil:Ananalysisofahighlyunequalcountryby CintiaDeniseGranjaandAnaMariaCarneiro 2021-10 MobilephonesandHIVtesting:Multi‐countryevidencefromsub‐SaharanAfricaby FrancescoIacoellaandNyashaTirivayi 2021-11 WhatmakesaproductivePh.D.student?byAlbertoCorsini,MichelePezzoniand FabianaVisentin 2021-12 DoinstitutionsandideologymatterforeconomicgrowthinLatinAmericainthe firsttwodecadesofthe21stcentury?byPamelaL.NavarreteGalloandJoRitzen 2021-13 HowsocialassistanceaffectssubjectiveWell‐being:LessonsfromKyrgyzstanby FranziskaGassmann,BrunoMartoranoandJenniferWaidler 2021-14 Dopandemicsleadtorebellion?PolicyresponsestoCOVID‐19,inequality,and protestsintheUSAbyFrancescoIacoella,PatriciaJustinoandBrunoMartorano 2021-15 ForeignR&DspilloverstotheUSAandstrategicreactionsbyThomasH.W.Ziesemer 2021-16 WilltheAIrevolutionbelabour‐friendly?Somemicroevidencefromthesupplyside byG.Damioli,V.VanRoy,D.VertesyandM.Vivarelli 2021-17 Theinfluenceofvalue‐chaingovernanceoninnovationperformance:Astudyof ItaliansuppliersbyEmanueleBrancati,CarloPietrobelliandCaioTorresMazzi 2021-18 Doesentrepreneurshipincreasethechancesofthepoor?byOmarRodríguezTorres 2021-19 Howdifferentarenecessityandopportunityfirms?Evidencefromaquantile analysisoftheColombianmicroenterprisesectorbyOmarRodríguezTorres 2021-20 AtaxonomyofEuropeaninnovationclubsbyArielL.Wirkierman,TommasoCiarli andMariaSavona 2021-21 Supportingacademicadvisingthroughself‐directed,blendedlearningbyMindel vandeLaar,KaterinaN.TriantosandLutzF.Krebs 2021-22 PoweringstructuraltransformationandproductivitygainsinAfrica:Theroleof globalvaluechainsandresourceendowmentsbySolomonOwusu 2021-23 Amapofthefractalstructureofhigh‐techdynamicsacrossEUregionsbyArielL. Wirkierman,TommasoCiarliandMariaSavona 2021-24 Howdoesmarketcompetitionaffectfirminnovationincentivesinemerging countries?EvidencefromLatinAmericanfirmsbyJoseMiguelBenaventeandPluvia Zuniga 2021-25 Theeffectivenessofinnovationpolicyandthemoderatingroleofmarket competition:EvidencefromLatinAmericanfirmsbyJoseMiguelBenaventeand PluviaZuniga 2021-26 KnowledgetransferactivitiesandconditionsforimpactinBulgarianpublicresearch institutions:Asurvey‐baseddiagnosticbyAnwarAridi,DanielQuerejazuandPluvia Zuniga 2021-27 Religiosity,smokingandotheraddictivebehavioursbyMonicaRoman,KlausF. ZimmermannandAurelian‐PetrusPlopeanu 2021-28 Women'sempowermentinColombia:AmultidimensionalapproachbyEsmeralda RodríguezGuerrero,ZinaNimehandAndreaFrancoCorrea 2021-29 SimulatingtheimpactofaraiseineducationsalariesoneconomicgrowthinPeru byLucianoParedesChervelliniandJoRitzen 2021-30 Socialnetworksandagriculturalperformance:Amultiplexanalysisofinteractions amongIndianricefarmersbyBruhanKonda,MarioGonzález‐Sauri,RobinCowan, YashodhaYashodhaandPrakashanChellattanVeettill 2021-31 PositioningfirmsalongthecapabilitiesladderbyAlexCoad,NandithaMathewand EmanuelePugliese 2021-32 PoliticalpolarizationandtheimpactofinternetandsocialmediauseinBrazilby LorenaGiubertiCoutinho 2021-33 Women'smobilephoneaccessanduse:AsnapshotofsixstatesinIndiabyLina Sonne 2021-34 Economicdevelopment,weathershocksandchildmarriageinSouthAsia:A machinelearningapproachbyStephanDietrich,AlineMeysonnat,Francisco Rosales,VictorCebotariandFranziskaGassmann 2021-35 Longingforwhichhome:Evidencefromglobalaspirationstostay,returnormigrate onwardsbyElsBekaert,AmelieF.Constant,KillianFoubertandIlseRuyssen