The Saving Behaviour of Natives and Migrants in the
Netherlands
Master Thesis written by: Romy Janssens Student number: 10335560
MSc. Business Economics - Managerial Economics and Strategy University of Amsterdam
Supervisor: Bas ter Weel ECTS: 15
2 Statement of Originality
This document is written by Romy Janssens who declares to take full responsibility for the contents of this document. I declare that the text and work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it. The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
3 Abstract
This thesis examines the saving behaviour among migrants and natives in the Netherlands. The increasing number of migrants in the population has raised the need for their
participation in the Dutch economy and, in particular, in our financial markets. The saving patterns of natives and migrants are analysed using panel data from respondents in the Longitudinal Internet Studies for the Social Sciences panel. The results suggest that migrants are less likely to have positive saving amounts. Additionally, if they do save, they save 43% less compared to native-born individuals. Among the migrant population, first-generation migrants save, on average, 67% less than migrants from the second generation. Non-Western migrants significantly save less than Western migrants. Finally, there is evidence that years of residence appear to influence migrants’ saving intentions. However, there is no evidence that cultural differences affect their saving intentions.
4 Table of contents
1. Introduction ...5
2. Theoretical framework and literature review...7
2.1 Saving...7
2.1.1 Definition of saving...7
2.1.2 Saving motives...7
2.1.3 Determinants of saving...9
2.2 Migration...9
2.2.1 Brief history of migration in the Netherlands...9
2.2.2 Definition of migrants...10
2.2.3Actual number of migrants...11
2.2.4 Saving behaviour of migrants...12
2.3 Literature review...12
3. Data...13
3.1 Data set...13
3.2 Variables...15
3.2.1 Dependent variable...15
3.2.2 Independent and control variables...16
3.3 Sample characteristics...18
4. Methodology...20
4.1 Hypotheses...20
4.2 Model...21
4.2.1. The saving behaviour among migrants and natives………....22
4.2.2. The saving behaviour among first- and second-generation migrants….23 4.2.3. The saving behaviour among Western and non-Western migrants……24
5. Results...24
5.1 The saving behaviour among migrants and natives………....24
5.2 The saving behaviour among first-and second-generation migrants………..27
5.3. The saving behaviour among Western and non-Western migrants………...28
6. Discussion...29
7. Conclusion...30
References...32
5 1. Introduction
Most European countries are confronted with an ageing population, high
unemployment and slow economic growth. At the same time, Europe remains one of the prime destinations for international migration (Münz et al., 2006). This large-scale migration, in many European countries, is a more recent phenomenon compared to the United States, Canada or Australia (Dustmann, 1997). European countries first started attracting migrants from less-developed regions—in particular, the ones linked to them through a colonial past. However, the establishment of the European Economic Community, which granted its citizens the right to free movement of capital, goods and people within the Member States, led to a major migration wave to Western and Northern Europe. With the enlargement of the European Union during the 2000s, these countries have seen another wave of migrants coming from other European countries (Williams & Hall, 2000).
In the Netherlands, most non-Western migrants have come from Turkey, Surinam, Morocco and the Netherlands Antilles (Van Amersfoort & Van Niekerk, 2006). Currently, people with a foreign background account for almost a fourth of the total population of the Netherlands (see Figure 1). Because these migrants represent a substantial share of the population in the Netherlands, their participation in the Dutch economy in general and financial markets in particular is of considerable interest to both economists and policy makers. Further, because of the increasing scale of international migration, the economic and social integration of this foreign-born population into our country’s society has become more and more important (Bauer & Sinning, 2011).
Figure 1. The Netherlands: Population by country of origin, 2016
Source: Central Bureau for Statistics (2016)
According to Bauböck (1995), saving intentions are regarded as one of the indicators of the integration of foreign-born citizens.The active participation of migrants in the
financial market is vital for ensuring cohesion in the host country and the ability of migrants to function as autonomous, productive and self-realised citizens (OECD, 2015). Aggregate personal and household savings also directly affect the economy as a whole (Hira, 1987). The change in the Dutch population has prompted much economic research on the migrant
10% 12% 78% Western migrant Non-Western migrant Native-born
6
population. However, despite the great deal of attention that has been paid to the labour market progress of migrants, little is known about their saving patterns. It may be the case that migrants behave differently and have different motives, opportunities and needs with respect to their saving behaviour.
This thesis examines whether there are observable differences in the saving behaviour among migrants and natives in the Netherlands. The main research question is as follows: Does saving behaviour differ between migrants and natives in the Netherlands? To reveal the determinants of saving intentions of migrants, this thesis investigates how socioeconomic and household characteristics stimulate their saving. Moreover, it analyses the ethnic background of migrants because the diversity among the migrant sub-population itself, in terms of
country of origin and year of arrival, is likely to exhibit a broad range of propensities with regard to economic decision-making.
This thesis attempts to gain insight into saving behaviour by comparing saving patterns of migrants and natives using data from the LISS. Based on empirical analyses, the hypothesis is proposed that migrants have different saving outcomes than natives. Results from the first analysis suggest that migrants are less likely to have positive saving amounts. Further, even if migrants have positive saving amounts, they save 43% less compared to native-born citizens. The second analysis strengthens the notion that educational level affects the saving intentions of migrants. First-generation migrants save 67% less than migrants from the second generation. Savings are affected by the country’s level of development. Western migrants have significantly higher saving outcomes compared to migrants from non-Western countries.
A number of correlational studies have described differences in the saving behaviour among migrants and natives. According to Kee (1995), differences in wages between
migrants and native-born workers suggest that the latter have more opportunity to save
money. Dustmann (1997) has suggested that migrants will carry out more saving compared to native workers, because migrants will be subject to greater income risk in the home country. However, this thesis found evidence that migrants save significantly less than native
individuals even after controlling for employment status. Further, besides wages, several other psychological, social and economic determinants of saving intentions have been found to be seemingly important (Lunt & Linvingstone, 1991). In this study, various socioeconomic and household characteristics appear to significantly affect the saving intentions of
individuals. Finally, a study has proposed that cultural background primarily affects the saving intentions of migrants (Al-Awad & Elhiraika, 2003). However, the results of this
7
thesis are in line with the findings of Piracha and Zhu (2012), which suggest that many of the differences in terms of saving behaviour among migrants are driven by uncertainties about future income and permanent residence rather than cultural differences.
This thesis is structured as follows: Section 2 provides a theoretical framework and explores the existing literature on saving and migration. Section 3 gives an overview of the data and variables used in the study. Section 4 discusses the methodology, and Section 5 presents the results. Section 6 discusses the implications of the results, and finally, Section 7 presents the conclusion.
2. Theoretical framework and literature review 2.1 Saving
2.1.1 Definition of saving
Generally, a saving is defined as ‘an amount of something that is not spent’ (Merriam-Webster’s Collegiate Dictionary, 2017). Moreover, saving is a decision made by people to postpone their consumption and avoid spending their disposable income. Mostly, saving takes the form of a deposit account held at a retail bank, which offers a certain interest rate. The accounts let customers set aside a portion of their liquid assets while earning a monetary return. According to Keynesian economics, saving consists of the amount left over if the cost of consumer expenditure is subtracted from the amount of disposable income over a given period. Furthermore, saving can be turned into further income by investing in different investment vehicles (Keynes, 1937). In the following sections, all kinds of saving motives will be discussed.
2.1.2 Saving motives
Individuals are motivated by various reasons to make certain economic decisions. Regarding saving behaviour, many motives are complementary, and it is unlikely that one person’s motive to save will be sufficient for all members of the population at a given time or for the same person over a long period (Yuh & Hanna, 2010). The concept of saving has received much theoretical and empirical consideration in the economic literature, and slightly different views of saving have been found in different disciplines.
One of the most important economic theories regarding saving is the life cycle hypothesis proposed by Modigliani and Brumberg (1954). This theory is concerned with optimal consumption over one’s lifespan According to this theory, individuals save because of the desire to retire with enough money. Retirement requires people to have enough money
8
saved to fund their lifestyle as they will no longer be earning an income. By themselves, most retirement provisions, such as social security contributions, are not always enough to pay for retirement expenses.
Second, precautionary saving is necessary for people to have an emergency fund that covers unexpected expenses (Xiao & Noring, 1994). This so-called precautionary motive includes the desire to accumulate assets through saving to meet possible emergencies, whose occurrence, nature and timing cannot be perfectly foreseen. Such emergencies might take the form of a temporary fall in income below the planned level or of temporary consumption requirements over and above the anticipated level (Modigliani & Brumberg, 1954). A third reason to save arises from the desire to invest in the future. For example, renters may desire to buy their own house or homeowners may desire to buy a bigger place. People can obtain a mortgage towards the purchase of a house, but they can also make a down payment, which is a fraction of the value of the house. In this case, the saving reduces the excessive risk that potential buyers are dealing with (Hayashi et al., 1988).
Furthermore, individuals can be motivated to save for expensive purchases such as luxury goods, vacations, or children’s education. However, these motives are strongly
associated with individuals’ financial resources. A number of previous studies have examined the effect of financial resources on these saving motives. Xiao and Anderson (1997) have suggested that an increase in income expands the saving motive priority of families from saving for daily necessities to saving for emergencies, children, retirement and holidays. Further, low-income consumers are more likely to report saving for daily expenses, middle-income people are more likely to report saving for emergencies, and high-middle-income consumers are more likely to report saving for growth (Xiao & Noring, 1994).
Additionally, according to Horioka and Watanabe (1997), saving motives can be divided into three different groups. First, life-cycle motives are defined as motives that arise from temporary imbalances between income and expenditures at various stages in one’s life cycle. This type includes saving for leisure, retirement expenses, consumer durables, housing purchases and children’s education. Next, precautionary saving includes saving for income fluctuations, unemployment, illnesses and accidents. Finally, the bequest motive arises from the desire to leave assets behind to one’s children and others in the form of transfers or bequests.
9
2.1.3 Determinants of saving
Individuals’ saving patterns are determined by multiple internal and external
characteristics. Previous empirical research has found age, gender, education and income to be the most important discriminators of personal saving habits (Furnham, 1985).
Further, personality aspects may also play a decisive role in saving intentions.
Attributes such as risk aversion, emotion, or complexity have always counted in the decision-making of an individual. A study by Nyhus and Webley (2001) has found personality factors such as emotional stability, self-control and extraversion to be predictors of individual saving. For instance, happier people are found to save more because they expect to live a longer life and seem more concerned about the future than the present.
Furthermore, culture appears to influence saving intentions. Xiao and Fan (2002) compared the saving motives of Chinese and American workers and found that Chinese workers were more likely to report saving for daily expenses, emergencies, children and investment, whereas Americans were more likely to report saving for major purchases and retirement.
In addition to internal factors for personal saving, economic performance and facilities provided by one’s country can be other major potential determinants. One of these factors is the country’s security system, because saving tends to decline as benefits available from this system increase (Evans, 1983). Further, empirical analysis demonstrates that foreign shock to the domestic economy affects saving. For instance, during the past years, the saving
behaviour among Dutch households has been constantly fluctuating. Clearly, saving intentions are affected by uncertainty in financial markets, such as the 2008 financial crisis (see Figure 2). For this reason, not only individual characteristics but also external factors such as periods of financial unrest determine saving behaviour among the population.
Figure 2. Netherlands: Net household saving 2004–2016 (values in billions of Euros)
Source: OECD Economic Outlook No. 100, Statistics and Projections
2.2 Migration
2.2.1 Brief history of migration in the Netherlands
According to Statistics Netherlands, the number of migrants to the country has
0 10000 20000 30000
10
increased sharply from the second half of the twentieth century. In the Netherlands, from the 1960s onwards, more people have been entering the country than leaving it. At first, from 1950, citizens from the newly independent Indonesian republic migrated to the Netherlands. After World War II, there was a period of construction and industrialization, and the labour shortages in the industrial sector led to an increasing demand for low-skilled workers (Van Amersfoort, 1986). As a temporary solution, low-skilled male workers were recruited from Southern European countries, such as Italy, Spain and Portugal, and later on from Turkey and Morocco. These so-called ‘guest workers’ were expected to return to their home country, and so they left their families behind in the home country (Vermeulen & Penninx, 2000).
In 1973, the situation changed because of the oil crisis and the subsequent economic downfall. Most of the Spanish and Italian guest workers returned to their home countries, but most of the Turks and Moroccans stayed in the Netherlands (Rusinovic, 2006). Then, as a result of family reunification and formation, migration from Turkey and Morocco increased again (Garssen & Nicolaas et al., 2008). Further, in the 1970s and 1980s, another wave of migrants arrived from the newly independent Suriname and Dutch Antilles. These migrants still held a Dutch passport and were hoping for a better future in the Netherlands.
From the end of the twentieth century, the demand for labour started to grow again, particularly in agriculture, horticulture, industry and services. This attracted a lot of labour migrants from new EU member states in Eastern Europe like Bulgaria, Romania and Poland. However, skilled workers from India, Japan and the United States also migrated to the Netherlands and made great contributions towards strengthening the current Dutch economy (Lucassen & Lucassen, 2011).
2.2.2 Definition of migrants
In research on international migration, many terms such as immigrants, strangers, foreigners, asylum seekers and refugees are used. In 1999, Statistics Netherlands, in
consultation with the Ministry of Interior and Kingdom Relations, proposed a new definition of the concept of migrants. For the purposes of this thesis, in accordance with Statistics Netherlands’ definition, migrants are defined as people who have at least one parent born outside the Netherlands (Statistics Netherlands, 2004).
Statistics Netherlands defines the population with a foreign background in two steps by specifying the so-called first and second generations. The first generation is defined as persons who were born abroad and who have at least one parent who was also born abroad. The second generation is defined as people born in the Netherlands who have at least one
11
parent who belongs to the first generation. Additionally, migrants are, besides their generation, defined according to their country of origin. Non-Western countries include Africa, Latin America, Asia and Turkey. Meanwhile, Western countries refer to all the other countries in Europe, North America, Oceania, Indonesia or Japan (Alders, 2001). Table A in the Appendix shows which countries are included in Western countries and which in non-Western countries.
2.2.3 Actual number of migrants
The actual number of migrants in the Netherlands in the period after 2000, first and second generation, are shown in Table 1 below. As of the start of January 2016, of the 16,979,120 people in the Netherlands, 3,752,291 people are considered to be migrants. Therefore, the share of migrants is approximately 22.1% of the total population. Migration has contributed most to population growth because the number of people with a Dutch background has been decreasing (Statistics Netherlands, 2016). Further, the increase among Turks and Moroccans in the Netherlands is mostly due to second-generation migrants. The percentage of first-generation migrants from Turkey, Surinam, Dutch Antilles and Aruba is declining, partly as a result of emigration and death.
Next, the number of Western migrants in the Netherlands has increased sharply over the last decade. In 2008, a record number of 140,000 migrants came to live in the
Netherlands, and the number of labour migrants from the European Union was particularly high (Statistics Netherlands, 2016).
Table 1. Demographic development of population in the Netherlands, 2000–2016
________________________________________________________________________________________________________________ 2000 2004 2008 2012 2016 ________________________________________________________________________________________________________________
Total population 15,863,950 16,258,032 16,405,399 16,730,348 16,979,120
Migrants (absolute numbers) 2,775,302 3,088,152 3,215,416 3,494,193 3,752,291 Migrants (percentage) 17.5 19.0 19.6 20.9 22.1
Total first generation 1,431,122 1,602,730 1,619,314 1,772,204 1,920,877
Western 544,890 581,890 602,130 690,203 772,428 Non-Western, including 886,232 1,021,074 1,017,184 1,082,001 1,148,449 Morocco 152,540 166,464 167,063 168,214 168,336 Dutch Antilles and Aruba 66,266 84,024 78,968 82,693 82,462 Surinam 183,249 187,990 185,284 183,752 177,720 Turkey 177,754 194,319 194,556 197,107 190,621 Other non-Western 306,423 388,277 391,313 450,235 529,310
Total second generation 1,344,180 1,485,422 1,596,102 1,721,989 1,831,414
Western 821,645 838,199 847,556 866,339 883,271 Non-Western, including; 552,535 647,223 748,546 855,650 948,143 Morocco 109,681 139,775 168,064 194,740 217,425 Dutch Antilles and Aruba 37,931 46,698 52,873 61,299 68,519 Surinam 119,265 137,291 150,515 163,045 171,302 Turkey 131,136 157,329 178,158 195,816 206,850 other non-Western 154,516 166,130 198,936 240,750 284,04 ________________________________________________________________________________________________________________ SSource: Statistics Netherlands, 2016
12
2.2.4 Saving behaviours of migrants
In general, there exist four different types of migration. First is labour migration in the European Union, where individuals move from one country to another with the purpose of seeking work. Next is internal migration, which occurs inside a particular country from economically poor areas to major cities. Third is international retirement migration, which is characterized by the residential mobility of retired people who have the economic power to buy properties abroad (Rodriguez et al., 1998). Finally, the fourth type is refugee migration. This type involves people fleeing wars on political repression or natural disasters (Castles & Van Hear, 2005). These people leave their countries for reasons such as low incomes, high unemployment rates or internal conflicts. It is reasonable to assume that saving is not always one of their biggest concerns. However, it is easy to come up with reasons that migrants save money instead of spending it.
One of the main reasons for migrants to save more is that some migrants are highly skilled people. These people face the prospect of better job opportunities and higher income in a particular host country compared to their home country (De Arcangelis & Joxhe, 2015). Second, some of the migrants’ cultures encourage higher saving rates. Additionally, risk aversion and uncertainties about future income can result in migrants saving money instead of spending it (Nyhus & Webley, 2001). Migrants are also more likely to transfer money to their home country. There are various motives for migrants to remit sums to families in their country of origin, such as altruism, portfolio diversification, insurance or anticipation of future return migration (Faini, 1994). These remittances result in lower observed saving rates among migrants in the host country.
2.3. Literature review
Previous studies have examined the disparity between migrants and native saving behaviour. First, according to Dustmann (1997), it is ambiguous in general whether the savings of migrants are more or less than those of natives. His findings suggest that this depends on the risk in host and home country labour markets and on the correlation of labour market shocks.
However, Galor and Stark (1990) have provided a theoretical model that shows that based on the possibility of return migration, migrants save more than native-born citizens. In accordance with the life-cycle theory of consumption, those who expect a future income that is lower than their current income save more to smooth consumption over the life cycle. However, migrants who anticipate a positive return possibility may transfer some of their
13
savings as remittances to household members who have stayed behind in their home country. Evidence from Sinning (2010) reveals that these return intentions positively affect
remittances of migrants to their home country.
Further, Amuedo-Dorantes and Pozo (2002) have found that migrants save less than natives because of their different risk preferences and opportunities to diversify away from economic risk. Their results are also consistent with Carroll et al. (1994), who have found that recent migrants to Canada save less than Canadian-born individuals. Further, they have found that natives appear to carry out more precautionary saving than similar migrants. However, migrants’ apparently lower saving can be driven by the fact that they, unlike natives, engage in precautionary saving by remitting some amount of what they earn to their home countries.
Finally, the saving patterns of migrants are significantly different across countries of origin. Al-Awad and Elhiraika (2003) have found that saving rates are remarkably different across households from different countries or regions. Migrants from developing countries appear to be uniform in terms of average household size, age, education and occupation. However, migrants from Pakistan and India have been found to have higher average savings than those from Arab countries, although they have relatively lower incomes. This finding could suggest the strong effects of culture on saving, but it remains possible that households with different cultural origins will have similar saving behaviours. Finally, significant variations by country of origin in migrant rates of holding stock, mutual funds, U.S. saving bonds, and other fixed income securities have been observed (Seto & Bogan, 2013).
3. Data 3.1 Data set
The database used in this study was the LISS panel. This panel, administered by CentERdata, consists of almost 8,000 participating individuals from around 5,000
households. The panel members complete online questionnaires every month and are paid for each completed questionnaire (Scherpenzeel, 2011). Furthermore, one member of each household provides additional household information every month to account for any changes that may have occurred. The LISS panel is a representative sample of households drawn from the population register by Statistics Netherlands. The sample also includes individuals without computer or Internet access who are loaned this equipment
(Scherpenzeel, 2011).
14
living conditions of panel members. The survey consists of a large number of questions concerning subjects such as work, education, income, political views, values and personality. There is also room to collect data for different research purposes. The data contain
background variables, multiple core studies and some assembled studies. After signing a statement, academic researchers can freely download the data or can even make use of the panel to collect additional data themselves. The data are made available through the LISS data archive from October 2007 until December 2016.
In this study, the data were utilized because they provided detailed information on saving behaviour. Furthermore, the database included information on migrant status, country of origin and year of arrival in the Netherlands. The variables used in this study were selected from the following surveys in the LISS panel: Background Variables, Religion and Ethnicity,
Work and Schooling and Economic situation (which was divided into Assets, Income and Housing).
First, Background Variables includes information about gender, age, marital status, position, dwelling, income, children and education. The contact person of each household presents the so-called household box. The household box needs to be completed before the household can start completing other questionnaires.
Next, Religion and Ethnicity contains data on religiosity, ethnicity, nationality and country of origin of the participants. The country of origin and year of arrival are used to learn about the behaviour of migrants in the Netherlands. Work and Schooling contains information on participants’ employment status, wages, pensions and education. Both surveys have been implemented in the panel every year from the start of the LISS data archive.
Finally, Economic Situation is divided into three broad fields. Data on Assets are gathered from four periods. The questions in this survey always refer to participants’ situation at the end of the previous year. For example, Wave 1 presents data collected in 2008
concerning the situation at the end of December 2007. This increases the reliability of the data because end-of-year savings have to be submitted to the tax authorities. This implies that people are more likely to give a precise estimate of their savings compared to an estimate of their current savings. This study will mainly focus on the years 2008 and 2014, that is, the periods before and after the global financial crisis.
The data sets of the above-mentioned years were merged is such a way that the missing values and coded observations for each individual were updated with the
observations available in another year. Then, the merged data set was formatted for use in this study, and the participants who were missing data for all the saving methods were
15
dropped from the analysis. Analysis of the background characteristics of these households showed that it was unlikely that their savings were equal to zero. An alternative way to deal with the missing variables was to assign them values based on the observed characteristics of households who did report savings. Because the remaining number of observations was sufficiently large, the missing observations were dropped. The same decision was made for a number of outliers, that is, a few people with savings above 1 billion euros. It is unclear whether the reported savings were reliable, and so they were dropped from the analysis. Taking into account the fact that children and participants younger than 18 years are unlikely to take saving decisions within the household, a data set was composed based on the head of the household or his or her partner. The household head is the person whose name appears on the rent contract or who is the homeowner. If more than one name appears, the household head is defined as the person with the highest income in the household
(Scherpenzeel, 2011).
3.2. Variables
3.2.1 Dependent variable
In this study, the main variable of interest was the saving behaviour of people living in the Netherlands. The respondents were asked, in 2008, 2010, 2012 and 2014, about the total balance of their saving accounts. The following measures of saving were used in the
empirical analysis:
First is the personal saving account, which contains liquid assets that have cash value or that can be easily converted into cash. This total balance of participants’ current accounts, saving accounts, deposit accounts, saving bonds and saving certificates was directly
measured. The absolute change in personal saving can be measured by taking the difference in the total balance during the periods 2008 to 2010 and 2012 to 2014.
Second is the investment saving account, which contains stocks and bonds that are considered to be important intermediary forms of saving as they get transformed into a capital investment that produces value. The absolute change in investment saving can be measured by taking the difference in the total value of the growth funds, share funds, bonds,
debentures, stocks, options and warrants during the periods 2008 to 2010 and 2012 to 2014. Third is the insurance saving account, which includes saving for a long-term target such as retirement provision. The absolute change in insurance saving can be measured by taking the difference in the total value of the single-premium insurance policy, life annuity insurance and endowment insurance over the periods 2008 to 2010 and 2012 to 2014.
16
Finally, loaned money to family, friends or acquisitions is regarded as saving too. The value of loaned money is particularly important for the migrant population because, for instance, an above-average amount of money loaned out by a migrant could be a remittance to his or her home country. The total value of money loaned out is directly measured. The absolute change in loaned money can be measured by taking the difference in the total amount of money loaned out over the years 2008 to 2010 and 2012 to 2014.
For each saving account mentioned above, respondents were asked about the total value or sum of each particular account. Respondents who said they did not know the value were asked to say to what category the saving amount belonged. To obtain sufficient observations in these particular cases, the midpoint value of each category was regarded as the saving amount. The values of all the saving options together represented the total amount the respondent was saving over a particular period. In case this value was below or equal to zero, the value was not regarded as saving.
3.2.2 Independent and control variables
Multiple independent and control variables were used; these were divided into migrant characteristics, socioeconomic characteristics and household composition. Table 2 provides an overview of the definition and measurement of all the variables used in this study. The main explanatory variable was the possibility of being a migrant. Further, some interesting background characteristics for the migrant population were taken into account. For instance, the generation to which the migrant belonged, the country of origin and duration of residence in the Netherlands were likely to affect migrants’ saving intentions.
Second, socioeconomic characteristics such as age, gender, income, educational level and employment status were defined. Fernandez et al. (2009) have found that age has a positive impact on saving. Their results showed that the probability to save is rising with age, but at a progressively lower rate. According to the life-cycle hypothesis, younger individuals have a greater propensity to consume while middle-aged people have a greater propensity to save that is enhanced by a typically higher income (Ando & Modigliani, 1963). Further, married persons are more likely to be more interested about their wealth and saving because of their offspring (Fernandez et al., 2009). It is also generally assumed that people with higher education will earn a greater income, thereby leading to higher absolute and relative savings (Revoredo & Morisset, 1999). In particular, the positive effect of education on saving appears to be present in more industrialised countries. However, the relationship between education and income can be negative at first because education expenses initially increase
17 consumption and reduce current disposable income.
Finally, household composition contained the number of household members, civil status, and number of children and stated whether or not the participant was a homeowner. Attempts to introduce the household size effects on the life-cycle model have revealed that a larger family size reduces saving rate (Davies, 1988). Orbeta (2006) has shown that an increase in household size has a negative impact on saving. Further, the status of the household head has been found to influence saving behaviour. Married persons are more likely to be more interested in wealth and saving (Fernandez et al., 2009). The same goes for households with a large number of children. Finally, Engelhardt (1996) has suggested that an increase in house prices has a negative impact on savings because people seem to invest more and/or borrow more in periods of boom. Apart from these house price fluctuations,
homeowners are also influenced by their mortgages, and their savings decrease with the value of the mortgages.
Table B, shown in the Appendix, presents the correlation matrix with the variables included in this study. High correlations between the independent variables are interesting but are also potentially problematic for the empirical analysis. Multicollinearity occurs when the effect of one predictor cannot be separated from the effect of another. Although this does not influence the total predictive power of the model, the individual effects are then not precisely estimated. This also increases the chance of a type-II error, where the null hypothesis is falsely accepted because of high standard errors of the coefficients. According to Stock and Watson (2011), Pearson correlations above 0.7 are considered to be high, and using them as independent variables within the same regression should be avoided. Pearson is useful for variables on an interval scale, while Spearman’s correlation is used for variables taken from an ordinal scale (Stock & Watson, 2011). The correlation table suggests that saving is increasing with age and income in the LISS data. Further, the correlation between migrant status and the amount of savings can provide an initial sign that migrants save differently compared to natives. The one-on-one correlations between migrant status and the several forms of saving seem quite low (r<0.3).
Table 2. Definition and measurement of the variables used in the study
____________________________________________________________________________________________________
Variable Description
____________________________________________________________________________________________________ Dependent variable
Saving Total value of differences in personal, investment and insurance
saving plus the differences in total value of money loaned out over the time periods 2008 to 2010 and 2012 to 2014.
18
____________________________________________________________________________________________________ Positive saving 1 if respondent has positive value of saving (defined above) over a particular time period; 0 otherwise.
Independent and control variables Socioeconomic characteristics
Migrant 1 if respondent is defined as migrant; 0 otherwise.
Age Age of the respondent in years.
Male 1 if respondent is male; 0 otherwise.
Income Respondent personal net monthly income.
Education Respondent highest level of education with diploma (1= primary school, 2= intermediate secondary education; 3= higher secondary education, 4= intermediate vocational education, 5=higher vocational
education, 6= university).
Employment 1 if respondent is currently employed; 0 otherwise. Household composition
Members Number of persons living in the household of the respondent. Married 1 if respondent is married; 0 otherwise.
Children Number of children living in household of respondent. Homeowner 1 if respondent is homeowner; 0 otherwise.
Migrant characteristics
Origin Background of migrant respondent (1= respondent is first-generation migrant with a Western background, 2= respondent is first-generation migrant with a non-Western background, 3= respondent is second -generation migrant with a Western background, 4= respondent is second-
generation migrant with a non-Western background). Generation 1 if respondent is first-generation migrant; 0 otherwise. Western 1 is respondent is Western migrant; 0 otherwise.
Country Country of origin of respondent (1= Turkey, 2= Morocco, 3= Dutch Antilles, 4= Suriname, 5= Indonesia, 6= other Western origin, 7= other
non-Western origin).
Residence Migrant respondents’ residential stay in the Netherlands in years
____________________________________________________________________________________________________
3.3 Sample characteristics
Tables 3 and 4 below present descriptive statistics for the socioeconomic variables and household composition. In 2008, the average age of the respondents was 48 years. They had, on average, medium levels of education, and the majority of the respondents were female. The biggest difference between natives and migrants was regarding their saving and income. On average, the native-born people saved €30,136.36, compared to an average amount of €11,558.77 for migrants. Further, on average, migrant earnings were substantially lower than the earnings of natives: €1,693.64 vs. €2,118.16.
19
The number of natives and migrants in the sample differed somewhat from the number of migrants in the Dutch population. The sample of households used in the empirical analysis consisted of about 3,200 households. Of these, about 10 per cent identified as migrant households. Additionally, the number of migrants in the Dutch population was found to be more than 15 per cent.
Table 3. Descriptive statistics: Socioeconomic characteristics
2008 N Savings
(x€1,000)
Age Male Income
(x€1,000) Education Employment Total sample 3,241 Mean 28.25 48.63 0.46 2.07 3.64 0.67 Standard deviation 30.01 0.24 0.01 15.31 0.02 0.01 Natives 2,912 Mean 30.14 48.79 0.46 2.12 3.63 0.68 Standard deviation 13.33 0.25 0.01 16.63 0.03 0.01 Migrants 329 Mean 11.56 47.12 0.44 1.69 3.67 0.61 Standard deviation 18.39 0.77 0.03 32.63 0.08 0.03
Next, on average, the Dutch households consisted of 2.66 members, most of whom were married. They had, on average, 0.84 children, and more than three-quarters owned a house. Further, migrants had, on average, more children and were owners of their own houses much less often.
Table 4. Descriptive statistics: Household composition
2008 N Members Married Children Homeowner
Total sample 3,241 Mean 2.66 0.69 0.85 0.74 Standard deviation 0.02 0.01 0.02 0.01 Natives 2,912 Mean 2.66 0.69 0.84 0.76 Standard deviation 0.02 0.01 0.02 0.01 Migrants 379 Mean 2.66 0.67 0.90 0.60 Standard deviation 0.07 0.03 0.06 0.03
Table 5 presents the summary statistics of the migrant respondents in the sample. The majority of the migrants were second-generation migrants with a Western background. Further, on average, the migrants arrived in the Netherlands 27 years ago.
20
Table 5. Summary statistics: Deviation of migrants in the sample
________________________________________________________________________________________________________________ 2008 N ________________________________________________________________________________________________________________ Natives 3,252 89.56% Migrants 379 10.44% Migrant respondents
Per cent first generation, Western 91 24.01% Per cent first generation, non-Western 103 27.18% Per cent second generation, Western 159 41.95% Per cent second generation, non-Western 26 6.86%
Average number of years resident in the Netherlands 379 27.59 years
____________________________________________________________________________________________________
4. Methodology
This section explains the study’s hypotheses and the empirical approach used to answer the question of whether differences exist in the savings of migrants compared to natives and to what extent the differences can be assigned to different motives. This way, it can be shown whether or not migrants are equally likely to save as natives and which characteristics affect their saving behaviour.
4.1 Hypotheses
First, this thesis investigates whether saving behaviour is different between migrants and natives. I expect that migrants save less compared to the native population. As
documented by various studies, including Miller and Neo (2003), most migrants have lower wages compared to native-born people. Native-born people are more likely to have better job opportunities, resulting in higher wages and, subsequently, a higher income. For this reason, I expect lower incomes to result in lower saving amounts for migrants. Therefore, the first hypotheses is
H1: Migrants in the Netherlands are equally likely to save compared to natives.
Second, this paper considers several characteristics of the migrant population to analyse the effects of their saving intentions. As mentioned before, migrants are divided into two generations. I expect first-generation migrants to save less than second-generation migrants. One possible explanation for this is that the level of educational attainment of people born abroad lags behind that of their Dutch counterparts. Second-generation migrants born in the Netherlands have educational levels that are much higher, and according to the life-cycle
21
hypothesis, education positively influences saving (Morisset & Revoredo, 1995). Therefore, the second hypotheses is
H2: First-generation migrants are equally likely to save compared to second-generation migrants.
Next, I expect migrants’ country of origin to have an effect on their saving intentions. As mentioned before, education and income are important determinants of the saving patterns of individuals (Furnham, 1985). Education and income levels are much higher in developed Western countries than non-Western countries. Further, the findings by Al-wad and Elhiraika (2003) suggest that saving rates are remarkably different across countries, and migrants from developing countries appear to be uniform to native-born citizens. Therefore, the third hypothesis is
H3: Migrants from non-Western countries are equally likely to save compared to Western migrants.
4.2 Model
The goal of this paper is to gain a better understanding of how saving behaviour differs between migrants and natives in the Netherlands. The quantitative analysis consisted of three main regressions, described in more detail in the following paragraphs. Each model described in this section was tested to find the appropriate regression model. One problem with measuring saving is that savings seem to have a long-tailed distribution. Figure 3 presents the distribution of the total savings during the period 2008 to 2010.
Figure 3. Distribution of total savings
0 1 0 0 2 0 0 3 0 0 4 0 0 F re q u e n cy 0 20000 40000 60000 80000 100000 Saving
22
The histogram shows that the right tail is longer and that the mass of the distribution is concentrated on the left of the figure. In other words, the distribution is skewed to the right. Log transformations were used to improve the model fit. If residuals are not normally distributed, then taking the logarithm of the skewed variable may improve the fit by altering the scale and making the variable more ‘normally’ distributed. Figure 4 below presents the distribution of total savings after log transformation. The plotted histogram shows that the outcomes are normally distributed. This also applies to respondents’ income, whose logarithm is normally distributed but whose untransformed scale is skewed.
Figure 4. Distribution of total savings after log transformation
4.2.1. Saving behaviour among migrants and natives
To test the first hypothesis, first, a dummy variable was used as the dependent
variable. The dummy variable took a value of one if the respondent had a positive total saving and zero otherwise. In this case, the probit regression model was used because the dependent variable was binary; that is, it only had two possible outcomes. The main explanatory variable was whether the individual was a migrant or not. I controlled for age, income, education and employment status. The outcomes are represented by a binary
indicator variable Yi as follows:
Yi = 1 if Yi* > 0 and probability Pr(Yi=1) = Pr(Yi*> 0) = Pr(xiTβ+εi > 0)
Yi = 0 if Yi* ≤ 0 and probability Pr(Yi=0) = Pr(Yi*≤ 0) = Pr(xiTβ+εi ≤ 0)
Probit was based on a latent model: = Pr(Yi= 1| x) = P(Yi*> 0 | x) = Pr(xiβ + εi > 0 | x) = Pr(εi > - xiβ | x) = 1- F (-xiβ) 0 50 1 0 0 1 5 0 F re q u e n cy 0 5 10 15 Saving
23
The error terms were assumed to be independent and normally distributed. By symmetry of the standard normal distribution, the model took the following form:
PR(Yi=1 | x) = Pr(Yi* > 0) = Φ(xiTβ) and Pr(Yi=0 | x) = Pr(Yi* ≤ 0) = 1- Φ(xiTβ)
Where Φ indicated the cumulative distribution function of the standard normal distribution. This model could not be consistently estimated using ordinary least squares (OLS); instead parameter β was estimated by the maximum likelihood theory according to Stock and Watson (2011).
Second, to explore the differences in saving behaviour among migrants and natives in
more detail, the OLS regression model was used. The OLS estimator chose regression
coefficients so that the estimated regression line was as close as possible to the observed data, where closeness measured by the sum of squared mistakes was made in predicting y given x. The total savings after log transformation were used as the dependent variable. The main explanatory variable was whether the individual was a migrant or not. I controlled both for the socioeconomic and household characteristics. To estimate the unknown parameters, the following linear regression model was used:
log Savingit =β0 + β1Migrant+ β2Age + β3Male + β4logIncome + β5Education
+ β6Employment + β7Members + β8Married + β9Children
+ β10Homeowner + uit
Where,
- log Savingit denoted the total saving of individual i after log transformation in period t;
- Migrant represented whether the individual was a migrant or not;
- Age, male, income, education and employment represented socioeconomic characteristics;
- Members, married, children and homeowner represented the household characteristics; - uit denoted the error term.
4.2.2. Saving behaviour among first and second-generation migrants
To test the second hypothesis, the OLS regression model was used. The total savings
after log transformation were used as the dependent variable. The main explanatory variable was whether the individual was a first- or second-generation migrant. I controlled for age, income, education, employment and country of origin. To estimate the unknown parameters, the following linear regression model was used:
log Savingit = β0 + β1Generation + β2Aget + β3Male+ β4logIncomet + β5Educationt
24
+ β10Homeownert + uit
Where,
- log Savingit denoted the total saving of individual i after log transformation in period t;
- Generation represented whether individual i was a first- or second-generation migrant; - Age, income, education and employment represented the socioeconomic characteristics; - Members, married, children and homeowner represented the household characteristics; - uit denoted the error term.
4.2.3. Saving behaviour among Western and non-Western migrants
To test the third hypothesis, the OLS regression model was used. The total savings after log transformation were used as the dependent variable. The main explanatory variable was whether the individual was a Western or non-Western migrant. I controlled for age, income, education, employment status and years of residence in the Netherlands. The linear regression model took the following form:
log Savingit = β0 + β1Western+ β2Age + β3Male + β4logIncome + β5Education
+ β6Employment + β7Residence + β7Country + uit
Where,
- log Savingit denoted the total saving of individual i after log transformation in period t;
- Western represented whether individual i was a Western or non-Western migrant; - Age, income, education and employment represented the socioeconomic characteristics; - Turkey, Morocco, Dutch Antilles, Surinam, Indonesia and other Western and non- Western represented the countries of origin of migrants;
- Residence represented individual’s years of residence in the Netherlands; - uit denoted the error term.
5. Results
This chapter will present the study’s results and interpretations, which will be divided into three parts. First, I will present and discuss the regression of saving behaviour of both migrants and natives in the Netherlands. Thereafter, the results of the migrant sub-population will be discussed and divided in terms of generation and country of origin.
5.1 The saving behaviour among migrants and natives
First, the dummy variable positive saving was analysed to see whether migrants were equally likely to have positive saving outcomes compared to natives. Table 6 presents
25
the results from the probit regression of positive saving. The variable migrant was significant at a confidence level of 1%, even after controlling for various socioeconomics and
demographic variables. The size of the coefficient varied between -0.17 and -0.19 depending on the model specification. The migrant coefficients were negative, which implies that on average, migrants are less likely to have positive saving amounts. However, the results from Table 6 can only be interpreted by the sign. Therefore, I present the marginal effects in table 7.
Table 6. Probit regression of positive saving over the period 2008 to 2010
Positive saving (1) (2) (3) (4) (5) Migrant -0.1913*** (0.0524) -0.1878*** (0.0525) -0.1815*** (0.0573) -0.1880*** (0.0589) -0.1788*** (0.0611) Age 0.0031** (0.0014) 0.0033** (0.0016) 0.0058*** (0.0017) 0.0062*** (0.0022) Income 0.0340*** (0.0116) 0.0142 (0.0122) 0.0148 (0.0136) Education 0.0918*** (0.0176) 0.0830*** (0.0180) Employment 0.0052 (0.0672) Constant -0.6639 -0.8241 -0.9844 -1.3120 -1.3062 Observations 4,145 4,145 3,331 3,202 3,032 Chi-2 Test 13.52 18.20 24.75 48.47 41.47 Prob>chi2 0.0002 0.0001 0.0000 0.0000 0.0000 Pseudo-R2 0.0030 0.0040 0.0064 0.0130 0.0117
* Significant at confidence level of 10%, ** Significant at confidence level of 5%, *** Significant at confidence level of 1%
Table 7 indicates the precise marginal probability of migrants having positive saving amounts. The variables age and income hold at their means, and the categorical variables education and employment are hold at the fourth and first categories, which include respondents that had intermediate vocational education and were employed.
Table 7. Marginal effects of positive saving over the period 2008 to 2010
Variable Positive saving
Migrant -0.0600***
Age: 51.73 (mean)
Income: €1,612.75 (mean) Employment: Employed
Education: Intermediate vocational education
26
The results can be interpreted as follows: an employed individual at the age of 52 years with a monthly income of €1,612.75 is 6.0% less likely to have positive saving amounts when she/he is a migrant compared to a native-born individual. This result implies that the first hypothesis can be rejected. However, the saving behaviour of both migrants and natives will now be analysed in more detail. Table 8 presents the results of the OLS regression of saving behaviour among migrants and natives over the period 2008 to 2010.
Table 8. OLS regression of total savings over the period 2008 to 2010
Total saving (1) (2) (3) (4) Migrant -0.7659*** (0.1481) -0.6431*** (0.1590) -0.6432*** (0.0917) -0.5628*** (0.1538) Age 0.0351*** (0.0033) 0.0249*** (0.0061) Male 0.2279* (0.1365) 0.2407* (0.1337) Income 0.3813*** (0.1158) 0.2972*** (0.1155) Education 0.2383*** (0.0465) 0.1988*** (0.0453) Employment 0.4420*** (0.1703) 0.2755 (0.17154) Members 0.0390 (0.1405) 0.0414 (0.1771) Married 0.3013* (0.1593) 0.2936* (0.1793) Children -0.1780 (0.1772) -0.1694 (0.1881) Homeowner 1.1275*** (0.1333) 0.9997*** (0.1535) Constant 8.7095 2.8312 7.5908 3.2886 Observations 996 758 996 758 F-Test 26.76 21.93 30.78 20.37 Prob>F 0.0000 0.0000 0.0000 0.0000 R-squared 0.0262 0.1491 0.1345 0.2143
* Significant at confidence level of 10%, ** Significant at confidence level of 5%, *** Significant at confidence level of 1%
The variable migrant was significant at a confidence level of 1%, even after
controlling for various socioeconomic and demographic variables. The size of the coefficient varied between -0.56 and -0.77 depending on the model specification. The migrant
coefficients were negative, implying that on average, migrants have lower saving amounts. The interpretation of the coefficient in the last column was that migrants save
less. Furthermore, the betas of the socioeconomic variables age, income, education and employment were statistically significant, which implied that these have an influence on saving behaviour as well. Older people have more savings, and the same goes
27
for male household heads. This is consistent with the lifecycle theory. Income, education and employment are also positively correlated with savings, which is quite logical because the higher the income, the more room there is for savings. In addition to Dustmann (1997), these results suggest that migrants are more likely than natives to have lower saving outcomes behaviour.
5.2 The saving behaviour among first- and second-generation migrants
This thesis highlighted the determinants of the saving behaviour of migrants in particular. The differences in migrant generation were analysed. Table 9 presents the results of the OLS regression of saving behaviour among first- and second-generation migrants.
Table 9.
OLS regression of total savings among first- and second-generation migrants
Total saving 2008–2010 2010–2012 2012–2014 Generation -0.4477 (0.3597) -0.1536 (0.3782) -1.1042** (0.4326) -0.3115 (0.3641) -0.5939* (0.3470) -0.5206 (0.4139) Age 0.0482** (0.0199) 0.0156 (0.0171) 0.0229 (0.0180) Male 0.1961 (0.4178) 0.4217 (0.4619) 0.6286 (0.4678) Income 0.0805 (0.4346) 0.6575 (0.4523) -0.2062 (0.4076) Education 0.2196 (0.1408) 0.3291*** (0.1229) 0.1134 (0.1473) Employment 0.5134 (0.5603) 0.3059 (0.5065) 0.3574 (0.5955) Members 0.5152 (0.5789) -0.0517 (0.5128) 0.1654 (0.5647) Married 0.0419 (0.5784) 0.6391 (0.5468) 0.0545 (0.5641) Children -0.7004 (0.6363) -0.0281 (0.5612) -0.1363 (0.5712) Homeowner 1.0564** (0.4747) 1.4484*** (0.4445) 0.9594* (0.5208) Constant 7.9277 2.1558 8.1891 3.3461 8.2444 6.7710 Observations 114 81 116 85 130 88 F-test 1.55 8.73 6.51 5.67 2.93 1.43 Prob>F 0.2159 0.0000 0.0120 0.0000 0.0894 0.1847 R-squared 0.0136 0.1172 0.0541 0.4337 0.0147 0.1563
* Significant at confidence level of 10%, ** Significant at confidence level of 5%, *** Significant at confidence level of 1%
The variable generation was negative but not significant for the period 2008 to 2010. After controlling for various socioeconomic and household characteristics, the estimates remain insignificant. The variables age and homeowner were both positive and
28
significant, which implied that they both have an influence on saving behaviour of migrants. Additionally, over the years 2010 to 2012 and 2012 to 2014, the variable generation was negative and significant. The interpretation of the coefficient in the third column was that first-generation migrants save less, . After controlling for the
other characteristics, the estimates were still negative but not significant anymore. The variables age and education were positive and significant, which was consistent with the life-cycle hypothesis, as education positively influences saving (Morisset & Revoredo, 1995). Furthermore, the variable homeowner was positive and significant. In addition to Engelhardt (1996), this result suggests that homeowners have higher saving outcomes behaviour.
5.3 The saving behaviour among Western and non-Western migrants
Finally, migrant characteristics in terms of country of origin were analysed. Table 10 presents the results of OLS regression of saving behaviour among Western and non-Western migrants.
Table 10. OLS regression of total savings among Western and non-Western migrants from 2008 to 2010
Total savings (1) (2) (3) (4) Western 1.5200*** (0.4192) 1.5282*** (0.4722) 1.2066** (0.4803) 1.4865** (0.6163) Age 0.0411** (0.0198) Male -0.0077 (0.4019) Income 0.5299 (0.4042) Education 0.2196 (0.1408) Employment 0.5134 (0.5603) Residence 0.0295** (0.0129) Country -0.1399 (0.1879) Constant 6.4881 0.3087 8.1891 3.3461 Observations 114 81 59 114 F-test 13.15 5.11 6.77 6.52 Prob>F 0.0004 0.0002 0.0023 0.0021 R-squared 0.1050 0.2931 0.1948 0.1051
29
The variable western was significant even after controlling for various socioeconomic and migrant variables. The size of the coefficient varied between 1.49 and 1.53, depending on the model specification. The western coefficients were positive, implying that on average, Western migrants have higher saving amounts. The interpretation of the coefficient in the first column was that Western migrants save more.
Furthermore, the betas of age and residence were statistically significant, implying that these have an influence on saving behaviour of migrants as well. This is consistent with the
findings from Galor and Stark (1990). The savings of migrants are increasing with their years of residence in the Netherlands. If migrants are expecting a permanent residence in the Netherlands, they are more likely to save for long-term purposes. Further, the variable of country was not found to be significant. In addition to Al-Awad and Elhiraika (2003), this result suggests that the country of origin does not influence the saving behaviour of migrants.
6. Discussion
This thesis used data from the LISS, and the respondents were randomly chosen from different areas in the Netherlands. Therefore, the sample can be considered characterize the population of the entire country. However, the results for saving behaviour found in this thesis cannot be generalized to countries like the United States. Findings from Al-Awad and Elhiraika (2003) have suggested that cultural differences across countries affect the saving behaviour of their population. Further, migration processes are not homogenous, and migration streams differ across countries. Additionally, all countries have different social security programs that influence retirement saving. For example, workers are significantly less likely to save for retirement in countries that offer higher benefits to retired employees and their families.
Next, self-reports are an effective tool in behaviour research because of their utility. However, they might not be entirely valid (Austin et al., 1998). For example, participants may not be honest about the balance in their saving accounts or may be too embarrassed to reveal other private details. Self-reports studies are inherently biased by the person’s feelings at the time of filling out the questionnaire. For instance, if a person is feeling depressed at the time of filling out the questionnaire, his or her answers could be more negative.
Further, respondents in the database were asked to fill in the total balance in their saving accounts. As mentioned before, respondents who reported that they did not know the balance were asked to say to what category the amount belonged. In this thesis, the midpoint value of each category was regarded as the saving amount. This may have led to some
30
measurement bias as this outcome was not directly measured.
Another limitation of the data was that only a part of the responses in all the surveys were taken into consideration in this study. In other words, when considering various surveys which accounted for different variables (socioeconomic, demographic and migration
background), only a part of the responses was made available for all surveys. Moreover, the number of observations for migrants was quite low. However, the data were still very relevant as they included information on saving patterns for migrants in the Netherlands that was difficult to find in previous research.
Finally, another potential limitation was the definition of migrants. In this study, both first- and second-generation migrants were defined as migrants and treated the same way. However, generally, the second generation might not have migrated; only one or both of the parents might have migrated. Consequently, the results may not be generalizable for the entire migration sub-population. For instance, if a mother moved to the Netherlands when she was a one-year-old, her child would be labelled as a second-generation migrant as they have both been living in the Netherlands (almost) their whole lives. Therefore, it is reasonable to assume that these migrants behave similarly to other non-migrant citizens.
7. Conclusion
The results suggest that migrants are less likely to have positive saving amounts. Furthermore, if migrants do save, they save less than native-born individuals. Therefore, the first hypothesis can thus be rejected. This result is in line with my expectation that migrants have lower saving outcomes. Further, socioeconomic and household characteristics appear to influence the saving behaviours of individuals as well.
First-generation migrants have significantly lower saving outcomes compared to migrants from the second generation. However, this result is not significant for the years 2008 to 2010. Therefore, the second hypothesis can be rejected. Additionally, consistent with my expectation, educational level appears to influence the saving behaviour of migrants as well.
Western migrants have significantly higher saving outcomes compared to migrants from non-Western countries. The third hypothesis can thus be rejected. The permanent residence of migrants in the Netherlands appears to influence their saving behaviour. However, there is no evidence that migrants’ specific country of origin influences their saving behaviour.
31
differ between migrants and natives in the Netherlands? The results imply that migrants do have lower saving outcomes compared to the native-born population. Further, first-generation and non-Western migrants are more likely to have lower outcomes in terms of their saving intentions.
In this thesis, saving was regarded as an increase in assets over a period of two years. If some individual saved over the last two years, for instance from 2008 to 2010, but spent everything the third year, he saved only over the first two years; overall, the individual did not save anything. To cover the saving behaviour of individuals properly, the economic literature agrees that this behaviour should be examined over a large period. It would be interesting to see what the effect of migration background would be over a longer period in a future study. Finally, different types of migrants are expected to have different saving
outcomes. If more data on this become available, it will be useful to examine the outcomes for migrants in other non-European countries as these outcomes are likely to differ from those for migrants in the Netherlands.
32 References
Alders, M. (2001). Classification of the population with a foreign background in the Netherlands. The measure and mismeasure of populations’ conference. Al-Awad, M., & Elhiraika, A. (2003). Cultural effects and savings: Evidence from
immigrants to the United Arab Emirates. The Journal of Development Studies, 39(5), 139-151.
Amuedo-Dorantes, C., & Pozo, S. (2002). Precautionary saving by young immigrants and young natives. Southern Economic Journal, 48-71.
Ando, A., & Modigliani, F. (1963). The" life cycle" hypothesis of saving: Aggregate implications and tests. The American economic review, 53(1), 55-84.
Bauböck, R. (1995). The integration of immigrants, report for the Council of Europe. Bauer, T. K., & Sinning, M. G. (2011). The savings behaviour of temporary and permanent migrants in Germany. Journal of Population Economics, 24(2), 421-449.
Canova, L., Rattazzi, A. M. M. & Webley, P. (2005) The hierarchical structure of saving motives. Journal of Economic Psychology, 26, 21–34.
Carroll, C. D., Overland, J., & Weil, D. N. (2000). Saving and growth with habit formation. American Economic Review, 341-355.
Carroll, C. D., Rhee, B. K., & Rhee, C. (1994). Are there cultural effects on saving? Some cross-sectional evidence. The Quarterly Journal of Economics, 109(3), 685-699.
Castles, S., & Van Hear, N. (2005). Developing DFiD’s policy approach to refugees and internally displaced persons. Report to the Conflict and Humanitarian Affairs
Department, Refugee Studies Centre, University of Oxford.
Central Bureau for Statistics. (2004). Migrants in the Netherlands. Voorburg.
Central Bureau for Statistics. (2016, October 3). Statistics Netherlands: Population by origin.
Retrieved from: http://statline.cbs.nl/Statweb/publication/?VW=T&DM=SLEN &PA=37296eng&D1=2551&D2=0,10,20,30,0,50,65-66&HD=1705152256&LA =EN&HDR= G1&STB=T
Davies, J. B. (1988). Family size, household production, and life cycle saving. Annals of
Economics and Statistics, 141-165.
De Arcangelis, G., & Joxhe, M. (2015). How do migrants save? Evidence from the British Household Panel Survey on temporary and permanent migrants versus natives. IZA
