Socioeconomic success of Asian immigrants in the United States

ABSTRACT It was the aim of this study to explore the effects of social, cultural, and transnational factors on the socioeconomic success (SES) of Asian immigrants in the United States. The participants in this study were 1371 Chinese, Vietnamese, Filipinos, and other Asian immigrants who were interviewed with computer-assisted software in Mandarin, Cantonese, Tagalog, Vietnamese, and English. The subcategory ‘other Asians’ consisted of Koreans, Japanese, Asian Indians, and individuals of other Asian backgrounds. Results showed that Chinese had a 56% higher probability of SES than other Asians. Men had an approximately 49% higher probability of SES than women due to gender hierarchies and disparities. SES increases for every unit increase in English language proficiency, native language proficiency, social networks, and parental education. Asians who migrated to the United States between the ages of 18 and 34 have an approximately 102% higher chance of SES than a person who migrated after the age of 35.


Introduction
In recent years, there has been a significant increase in immigration from Asian countries to the United States. Between 2000 and 2015, Asian immigrant populations in the United States grew faster than any other migrant group (Zong and Batalova 2016). This increase has stimulated a growing number of researchers to focus on Asian immigrants although research on Latino migrant populations is still dominating the field of migration studies in the United States, especially when it comes to socioeconomic success (SES). Sakamoto, Goyette, and Kim (2009) defined SES as ranked values of societal rewards including constructs and indicators such as income, wealth, educational attainment, and occupational status. They further discussed the impact of the majority-minority paradigm, which asserts that racial and ethnic minorities have lower rates of SES due to discrimination. They contrast this with the finding that Asian immigrants and Asian Americans not only have high educational achievements but also tend to have approximate parity with the white majority in most areas of the labour market (Sakamoto, Goyette, and Kim 2009;Sakamoto et al. 2016). They relate these outcomes to improved economic and social opportunities for Asian Americans and Asian immigrants and to a more multicultural ethos within society. However, they also note the rising significance of national origin and socioeconomic variability within the racial group of Asian. Yet, transnational and sociocultural differences with regard to SES have hardly received any attention in previous research. Indeed, research studies were usually based on generalisations of Asian immigrants and their high achievements, which are not only misleading but also damaging to groups that are extremely disadvantaged (Kao and Thompson 2003). As pointed out by Kim and Mar (2007, 181), 'socioeconomic outcomes vary tremendously by ancestry'.
Furthermore, previous quantitative studies did not focus on native language proficiency and transnational networks as predictors of SES of Asian immigrants despite the fact that the majority of Asian immigrants in the United States is foreign-born. Therefore, it is the aim of the study to analyse the impact of major predictors of SES of Asian immigrants that have not been addressed in nationally representative studies in the United States, such as the role of the native language and transnational factors as well as predictors that have received some attention in previous research such as the role of gender, English language proficiency, and parental education.
I first review relevant literature on SES. I then describe the settings of the current study, the participants, and the methods of data collection and analysis. SES is coded and measured by taking into consideration education, assets, and finances in the country of origin and the United States, profession and placement within the segmented labour market, and poverty levels. Factors that impact SES are analysed and discussed.

Predictors of SES
Empirically, it is the aim of this study to estimate the effects of transnational, linguistic, generational, social, and gender-related factors on SES of Asian immigrants in the United States. Much of the current research about socioeconomic attainments relates to Asian immigrants and Asian Americans as a broadly defined racial group, thereby potentially concealing important variations. However, immigrants' characteristics and SES tend to significantly differ by region of origin (Camarota 2005;Sakamoto, Goyette, and Kim 2009). Hence, studies of SES of immigrants ought to attend to variations at the national but also transnational level. Migrants of the first generation often maintain social networks with people of common ancestry in their country of origin and they also establish networks in their new country (Glick Schiller 2013). Furthermore, migrants may 'build transnational social fields and these fields may extend across generations. These networks may or may not draw on common ancestry as a domain of interconnection' (Glick Schiller 2013, 26).
Besides national origin and transnational issues, generational factors seem to have an important impact on SES. Segmented Assimilation Theory provides the most suitable framework for understanding the impact of generational factors. According to Zhou (1997Zhou ( , 2014, segmented assimilation provides three possible outcomes for immigrants and their offspring: The first one replicates the time-honored portrayal of growing acculturation and parallel integration into the white middle class; a second leads straight into the opposite direction to permanent poverty and assimilation into the underclass; still a third associates rapid economic advancements with deliberate preservation of the immigrant community's values and tight solidarity. (Zhou 1997, 975) Using the Segmented Assimilation Hypothesis, many researchers (Portes and Zhou 1993;Rumbaut and Portes 2001;Portes, Fernandez-Kelly, and Haller 2005) argue that the children of immigrants are at high risk for downward mobility into a new rainbow underclass. However, this risk is highly linked to the social backgrounds of later generation immigrants and the background of their parents (Zhou 2014). Studies show that the offspring of middleand upper-class immigrants will move ahead, using their cultural capital and resources provided by their parents (Portes and Zhou 1993;Rumbaut and Portes 2001;Portes and Rumbaut 2006). However, children of low-skilled immigrants face more socioeconomic barriers such as high poverty rates and low levels of education and income, which are major risk factors for downward assimilation (Portes, Fernandez-Kelly, and Haller 2005). Borjas (1993) also found in his analysis of census data that the income potential of the second generation was strongly affected by the socioeconomic status of their parents in the country of origin and argued that immigration policy affects the economic potential not only of first generation immigrants but that of their children. In a later summary of research on immigration and social mobility, Borjas (2006) found that the increase in relative wages between first and second generation immigrants rises only about 5-10% and that the average age-adjusted relative wage for both has been falling in recent years. However, Pastor (2001) argues that poverty rates are significantly reduced by the age of immigration, with rates of recent immigrants three times that of those suffered by households headed by individuals who were born in the United States and these findings are also supported by other researchers (Lee 2014). But Pastor (2001) also refers to the misconception of the media that assumes a correlation between immigrant status and poverty, failing to recognise the strong work ethic, attachment to the labour market, and increasing success despite hardships, especially among the first generation. He further analysed data for public assistance of low-income households and found the lowest rates among long-term first generation immigrants and the highest rates for U.S.-born householders and most recent immigrants. Nevertheless, it is a fallacy to simply assume a correlation between a most recent migration status and low socioeconomic capital. Indeed, Waldinger and Gilbertson (1994) point out that while in the early twentieth century immigrant groups worked mostly in the low-skilled segments of the labour market, late twentieth century immigrants were highly diverse in the degree of human capital they possessed at the time of immigration. While a high proportion of immigrants entered with low skills and low education levels in former times, more recent immigrants also entered with higher skills and higher education levels than the majority of native-born whites. Feliciano (2006) analysed data from international studies, the United States Census, and the Children of Immigrants Longitudinal Study and found that both parental and group pre-migration educational status affect the educational and economic attainment of immigrants. Furthermore, a study by Zhou and Bankston (1998) showed that for Vietnamese immigrants in the Versailles neighbourhood in New Orleans adherence to family traditions, strong social networks and ethnic involvement all had positive effects on academic expectations and socioeconomic attainment. In particular, immigrant success can be explained by the fact that both parents and also members of the wider community required their children to go beyond academic demands at school to ensure economic upward mobility (Zhou 2014). In addition, Rumbaut (1990) observed that they also tried to delay acculturation in order to ensure success. Gans (1992) questions whether delayed acculturation works in all instances but presents major aspects that contribute to immigrant success, including the finding that migrants must be able to cope with countervailing pressures and must resist a one-way assimilation into the American culture.
A few studies focused on the impact of gender and English language proficiency on the SES of immigrants and ethnic minorities. Regarding gender, researchers (Segura 1989;Salaff and Greve 2003;Waldinger and Feliciano 2004;Kim and Zhao 2014) found that the labour market and also socioeconomic opportunities are highly segmented with only a very few promotional opportunities available for women. They criticised the fact that major work is mostly applied to men and that there is hardly any attention to gendered inequalities within migrant and ethnic groups. In fact, gender structures heavily impact adversarial relationships at work and in school. Kim and Zhao (2014) also provided evidence that Asian immigrant women are more likely to be unemployed and they are also less likely to obtain positions as supervisors. Especially, Asian American women who immigrated after high school are highly disadvantaged, even if they hold the highest degree from an American university. Regarding language, previous research found that immigrants who are proficient in English have higher rates of SES, including higher earnings (McManus, Gould, and Welch 1983;Grenier 1984;Borjas 1994;Chiswick and Miller 2012;Lee, Zhou, and Kim 2014). Grenier (1984) showed that immigrants who could not speak English faced a 17% wage penalty.

Methods of data collection and analysis
Data were derived from the National Latino and Asian American Study (NLAAS). The NLAAS core sampling procedure included a nationally representative sample of Asian immigrants and Asian Americans who resided in the United States. The NLAAS dataset had one of the most comprehensive designs ever developed (Alegría et al. 2004;Wilson 2010, 2011). Below is a description of the sample, procedures, and the measures of this study.

Sample and participants
Within the NLAAS dataset, there was a nationally representative sample of 1371 Asian immigrant household heads who were employed in the United States labour market. In particular, the sample consisted of 405 Chinese, 341 Filipinos, 318 Vietnamese, and 307 other Asians. The subcategory 'other Asians' consisted of Koreans, Japanese, Asian Indians, and individuals of other Asian backgrounds. Among the study participants there were 639 women and 732 men. Among them were 199 immigrants who were 35 years or older when they came to the United States, 611 immigrants were 18-34 years old, 91 immigrants were 13-17 years old and 164 immigrants were younger than 12 years when they came to the United States. The remaining 306 participants were second generation immigrants. All participants were recruited between May 2002 and November 2003 as part of the larger survey.

Procedures
The NLAAS instruments were available in Cantonese, Mandarin, Tagalog, Vietnamese, and English. They were translated using standard translation as well as back-translation techniques. All participants received an introductory letter and the study brochure in their preferred language. Those who gave their consent to take part in the study were screened and interviewed by professionals who had cultural and linguistic backgrounds similar to those of the sample population. Interviews were conducted with computer-assisted interviewing software in the preferred language of the participants. Face-to-face interviews with the participants were administered. Exceptions were made when respondents specifically requested a telephone interview or when face-to-face interviewing was prohibitive. As a measure of quality control in this study, a randomly selected sample of participants with completed interviews was contacted to validate the data.
Written consent was obtained for all study participants, protocols, and procedures. Human subject approval was given by Harvard University, the University of Michigan, and the University of Washington.

Socioeconomic success
The construct SES was defined by taking into consideration education, occupation and placement within the segmented labour market, financial assets in the country of origin and the United States, and poverty index for all participants. It is presented in Table 1.

Predictors
As predictors of SES, several single variables of interest were selected as well as indices or constructs as weighted sums of ordinal survey questions. The single variable predictors included national origin (NAT), current age (AGE), gender (GEN), and age at immigration (AGEIMM). The multivariable predictor constructs included native language proficiency (NLP), English language proficiency (ELP), social networks (SNT), and parental education (ParEd). Cronbach alphas for all multivariable predictors were determined. They ranged between 0.9613 and 0.7827. These Cronbach's alphas are a measure of internal consistency, that is, how closely related a set of items is as a construct. It is considered to be a measure of scale reliability. Cronbach alphas which are higher than or equal 0.8 indicate a good internal consistency and Cronbach alphas which are higher or equal 0.7 indicate an acceptable internal consistency. The single variables and the multivariable constructs are presented in Tables 2 and 3. Regarding some of the survey constructs, differences had to be applied prior to data analysis in order to give the construct the right direction. Multivariable predictor constructs were used to allow for a qualitatively and quantitatively most suitable predictor representation. This was necessary for comprehensive predictors such as social networks and language proficiency.

Statistical approaches and methods
Summary statistics are shown in Table 4. A logistic regression model for SES was fit. As predictor variables single variables and multivariable constructs were used. All composite indices were treated as continuous variables in the logistic regression. To take into consideration a possible nonlinear dependency with respect to age, the variable AGE was included in quadratic form (AGE2 = AGE × AGE). The SES construct included education, occupation and labour market placement, finances, and poverty index. SES was defined to be a SES score above average.

Multiple imputation by chained equations of the missing data
The dataset contained missing values that were applicable to this study and were missing at random. These missing data have been previously documented (cf. Wilson and Lueck 2014). These data were imputed by using Multiple Imputation by Chained Equations (MICE). Using multiple imputations rather than single imputations takes into consideration statistical uncertainty in the imputations and the chained equations approach is   also very flexible and can be applied to impute variables of varying types, such as binary variables and continuous variables, as well as complexities such as survey skip patterns (Azur et al. 2011;Romaniuk, Patton, and Carlin 2014). The approaches taken in this study were as follows: First, a simple imputation was conducted for every applicable value that was missing at random. Second, the mean imputations for one variable were set back to missing. Third, the observed values from the variable in Step 2 were regressed with regard to the other variables in the imputation model. Fourth, the missing values for x were replaced with imputations from the regression model. Fifth, these processes (as applicable in x) were repeated for each variable that had missing data and were repeated for a number of cycles, with the imputations being updated at each cycle. At the end of these cycles, the final imputations were retained, resulting in one imputed dataset. Overall, 10 cycles were performed.

Findings and discussion
Statistical findings The final model of the predictors of SES included national origin (NAT), gender (GEN), parental education (ParEd), social networks (SNT), age at immigration (AGEIMM), English language proficiency (ELP), and native language proficiency (NLP). The complete model with the estimates of the coefficients for each predictor variable is shown in Table 5.
The odds ratios are shown in Table 6. The results show that for every unit increase in English language proficiency SES increases by about 22% on average for Asian immigrants, increases by about 6.8% for every unit increase in the native language proficiency score, by 2.3% for a unit increase in the social network score, and by 5.5% for every unit increase in the parental education index. Asian men have an approximately 49% higher probability of success than Asian women. Vietnamese have only about 83% and Filipinos have only 89% the probability of success compared to other Asians. Chinese have about a 56% greater probability of success than other Asians. A person who immigrated between the ages of 18 and 34 has an approximately 102% higher chance of success than a person who immigrated after age 35.

The importance of national origin
Results show that Vietnamese have only about 83% and Filipinos have only about 89% the probability of success than other Asians, while Chinese have an approximately 56% higher probability of success than other Asians. There is a lack of nationally representative studies to support these findings because the majority of previous studies usually applied to racial groups (e.g. Asian Americans in general), were conducted in English only, and excluded non-English speakers. It is highly problematic to exclusively use English instruments despite the fact that the majority of Asian immigrants in the United States is foreignborn and many do not speak English well or have not acquired English at all. Furthermore, while the focus on Asian Americans in general can be important when it comes to pan-Asian collective action, it may be a problem when it comes to the needs of highly disadvantaged Asian immigrant groups in the United States. Indeed, current research by Basu (2016) shows significant levels of economic diversity between Asian immigrant groups, comparing the wage assimilation of Asian immigrants from new and old source countries. Her study also shows the significant impact of country-specific migration and countryspecific occupation clustering from Asian countries since 1965. Country-specific economic niches have become more segregated for Asian immigrants in recent years and there is also a significant decline in wages of more recent Asian immigrants compared to previous Asian migrant groups in the United States.

Transnationalism
Previous quantitative studies also did not consider transnationalism. This study shows that the invocation of social networks and diverse languages makes reference to variables that draw from the connection that migrants maintain across borders. They draw from a transnational framework (i.e. the country of origin and the United States) or a transnational social field, which extends over immigrant generations.

Native and English language proficiencies
Findings indicate that English language proficiency, native language proficiency, parental education, and social networks have a significant impact on SES. The results on English language proficiency are well-supported by previous research. However, native language proficiency has not received attention in previous studies due to the status of minority languages in the United States. There are still strong demands for linguistic and cultural assimilation in the United States, which is reflected in all areas of society. However, it is a major problem that the high cultural capital brought into this country in the form of linguistic diversity is often lost through these demands of linguistic assimilation and homogeneity. This study shows that proficiencies in both the native language and English have a positive impact on socioeconomic attainment of Asian immigrants. Indeed, high levels of English language proficiency and native language proficiency represent higher cultural capital, which leads to higher SES. The native language is especially of importance for generation 1.5 and the second generation because it allows immigrant parents to transfer education and skills to their children. As Wong Fillmore (1991) points out, if the younger generation does not speak the native language well, parents face difficulties in monitoring their children's achievement, they cannot directly impact their cultural capital development, and they cannot transmit educational and socioeconomic advantages. Usually, '[s]tudents from the cultured classes are those best prepared (…) to adapt themselves to a system of diffuse, implicit requirements' (Bourdieu and Passeron 1979, 75), but this is not the case when there are linguistic barriers in society that undermine these cultural advantages. Indeed, cultural participation is a key aspect in the reproduction of both educational and occupational advantages (Bourdieu and Passeron 1979;Bourdieu 1986).
The importance of the native language as a form of capital must be also viewed on the macro level. Many Asian economies show rapid developments and the knowledge of Asian languages in an age of globalisation and interdependence is crucial. Thus, linguistic diversity is not the problem here as it includes English in addition to other languages, but the limited notion of a single language may hinder the SES of immigrant populations and ethnic minorities in the United States. As Rumbaut, Massey, and Bean (2006, 459) state: language fluency is an asset and that knowledge of the foreign tongue represents a valuable resource in a global economy (…) [Hence] efforts to maintain this part of their cultural heritage and pass it on to their children should not be discouraged.
However, strong structural support is needed to promote the native language besides English in American society because of American policies and regulations that still have the aim to extinguish native languages (Portes and Rumbaut 2006;Rumbaut, Massey, and Bean 2006;Lueck and Wilson 2011).

The role of gender
The findings of this study show that men had a 49% higher probability of success than women. This is due to the fact that states (e.g. countries of origin and the United States as the receiving country) but also communities enact gender hierarchies and disparities.
Gender disparities in SES against migrant women have had a long history. Early European immigrants were already patriarchally organised, and women mostly worked at home and therefore were largely invisible (White and Mullen 2016). Today both men and women work but poor migrant women work predominantly in the lowest levels of the labour-intensive markets. They face gender hierarchies in society and in their ethnic groups (Gans 1992). Kim and Zhao (2014) found that even highly educated Asian American women were disadvantaged in the American labour market when it comes to employment, annual earnings, and the number of people supervised. They argued that in contrast to men, labour market success and socioeconomic attainments of Asian immigrant women in the United States were affected by strict cultural norms and by their immigration status (Kim and Zhao 2014). Furthermore, Asian women also face socioeconomic disadvantages in interracial relationships (Basu 2015;Wang 2015). Basu (2015) found that interracially married Asian women earn 8-13% less than intra-married Asian women in the United States. Interracial-marriage penalties even increase for Asian women with a higher level of the husband's education. These societal, ethnic, and interracial factors help to explain the lower probabilities of SES of Asian women as indicated in this study.

Age of migration and segmented assimilation
Findings indicate that a person who immigrated between the ages of 18 and 34 has an approximately 102% higher chance of success than a person who immigrated after age 35. Especially, the first generation is more likely to have a stronger work ethic and attachment to the labour market, which may contribute to increased success. However, there seems to be a tipping point within the first generation when it comes to success. This tipping point is associated with age. In particular, this study shows that younger first generation immigrants are significantly more successful than middle-aged and older first generation immigrants. Earlier research found that SES increases with age of immigration (Oropesa and Landale 1997;Farley and Alba 2002;Alba and Nee 2003;Sakamoto and Woo 2007;Sakamoto, Goyette, and Kim 2009;White and Mullen 2016) whereby American-born Asians are most successful. In this study, significant differences were only observed within the first generation whereby younger immigrants have a significantly higher probability of SES. This result partly supports Zhou's (1997Zhou's ( , 2014 acculturation theory and provides evidence for economic upward mobility rather than a downward assimilation for younger migrants within the first generation. This can be explained by the fact that younger first generation immigrants have the highest probability to acculturate into American society by facilitating both their ethnic and new culture.
However, not just age of migration, but the length of time spent in the United States is of importance for the first generation, especially given the centrality of English skills to occupational status and income levels.

Nationally representative study
This is the first nationally representative study on the effects of national, transnational, social, and linguistic factors on SES among Asian immigrants and Asian Americans in the United States. Results indicate that a focus on national origin and transnationalism is indeed of high importance when it comes to SES and adds significantly to the previous literature. Although structural and cultural shifts in the United States have contributed to a decrease in racial inequality (Sakamoto et al. 2016), there are still major differences in SES between the different Asian groups in the United States. Hence, rather than paying attention to race only, there also needs to be a focus on the country of origin and the receiving country (e.g. transnational issues and differences).
This study also went beyond typical racial binaries in migration research. There is usually a comparison between white populations (i.e. Hispanic and non-Hispanic whites), blacks (i.e. African Americans and African migrants), and Asian populations (i.e. Asian Americans and Asian migrant populations). These racial meta-identities do not allow for variations and a precise focus on differences and disparities within a racial group. This may in turn not allow for affirmative action for certain Asian groups that face higher levels of inequality.

Comprehensive methods
The methods of data collection and analysis further contributed to the strength of this study. Indeed, the NLAAS dataset had one of the most comprehensive designs ever developed (Alegría et al. 2004;Wilson 2010, 2011). Furthermore, the study participants were screened and interviewed by professionals who had cultural and linguistic backgrounds similar to those of the sample immigrant population, adding a cultural perspective and understanding. All interviews were conducted with computer-assisted interviewing software in the preferred language of the participants and as a measure of quality control in this study, a randomly selected sample of participants with completed interviews was contacted to validate the data. The high importance of data validation has been emphasised in previous research (Bryman 2012;Gravetter and Forzano 2015).

Future research
Future quantitative and qualitative research on ethnic and social factors that influence SES can build on this study. It is crucial to provide a better understanding of the current Asian immigrant population in the United States and their advancements in SES also in the light of current immigration policy.