Asian American ethnic enclaves as a predictor of financial success in New York

Evidence of enclave associated earnings amongst the Asian American immigrant populations within the United States is limited due to an inconsistent definition of an Asian ethnic enclave. By using the Dissimilarity Index (DI) and Isolation Index to identify if areas within New York City can be defined as Asian enclaves, an understanding of the effect of living within an enclave community for certain ethnicities affects earnings for Asian Americans living in the city of New York can be gleaned. The ethnicities studied in this paper are as follows: Chinese, Indian, Filipino, Vietnamese, Korean, and Japanese.

ABSTRACT Evidence of enclave associated earnings amongst the Asian American immigrant populations within the United States is limited due to an inconsistent definition of an Asian ethnic enclave. By using the Dissimilarity Index (DI) and Isolation Index to identify if areas within New York City can be defined as Asian enclaves, an understanding of the effect of living within an enclave community for certain ethnicities affects earnings for Asian Americans living in the city of New York can be gleaned. The ethnicities studied in this paper are as follows: Chinese, Indian, Filipino, Vietnamese, Korean, and Japanese. ii TABLE OF CONTENTS ABSTRACT ii INTRODUCTION 1 LITERATURE REVIEW 4 DATA 11 METHODOLOGY 18 ANALYSIS 30 DISCUSSION 41 APPENDIX 43 REFERENCES 50 iii 1 INTRODUCTION Researchers have studied a phenomenon they refer to as the enclave thesis, where they studied the effects of living in an ethnic enclave for immigrants. The effects can vary from some studying the health of the immigrants, others the assimilation of the individuals, and the focus of this study, earnings. The enclave thesis is one that speculates that immigrants benefit from living in these ethnic enclaves, but the results are mixed – perhaps due to the varying definition of an ethnic enclave. Inconsistencies across research papers make it difficult to find a universal definition for an ethnic enclave, in turn, making the research difficult to compare. Each study has their own defining characteristics and boundaries. The definition tends to be similar across papers for qualitative traits and features, but measuring an ethnic enclave quantitatively is where the issue occurs. Depending on the data source that is used, the quantitative definition is limited, and there are constantly new methods to define an enclave. In this study, an ethnic enclave will be defined qualitatively as a group of individuals who live in a distinct area and hail from the same or similar ethnic background(s). Quantitatively, an ethnic enclave will have dissimilarity and isolation indices higher than 0.6, consistent with the cutoff point previously used to measure high segregation for blacks and Hispanics (Lim et al., 2015). The dissimilarity index is a measure of segregation amongst neighborhoods that “measures the percentage of one group that would have to move across neighborhoods to be distributed the same way as the second group” and this measure is symmetrical so it can to either group (About Dissimilarity Indices, n.d.). It is measured on a scale of 0 to 1, with 0 representing total integration where both groups are proportionally distributed evenly across all neighborhoods and with 1 2 representing total segregation for both groups, such that “the members of one group are located in completely different neighborhoods than the second group” (About Dissimilarity Indices, n.d.). In this study, the dissimilarity index of 0.6 or higher will be one of the two indicators of living in an ethnic enclave. The isolation is a measure of spatial clustering that evaluates “the extent to which minority members are exposed only to one another” (Bureau, 2021). This index was also measured on a scale of 0 to 1, with 0 representing completely dispersed groups (ethnic groups completely mixed with other groups in the area) and 1 representing groups that were completely spatially clustered (ethnic group completely surrounded by the same ethnic groups). Due to this measure representing exposure members of the minority have to other members of the same minority, it can define “the degree of potential contact, or possibility of interaction,” which can indicate an enclave (Bureau, 2021). As there are two measures used in this study that define the ethnic concentration or spatial clustering and the segregation (which is assumed to be purposeful) across the population, the study has firm boundaries around how an ethnic enclave is measured. These two measure will later be used as regressors, so the results can show how segregation and spatial clustering affect earnings for the six ethnic groups studied. A study conducted by Lim, Yi, Cruz and Trinh-Shevrin used the dissimilarity and isolation indices to identify ethnic enclaves for Asian Americans to study health outcomes among those adults in New York City and were able to successfully identify enclaves throughout New York City. While that study also used data from the New York City Community Health Survey to access zip codes to identify neighborhoods, this study does not do so, as it covers the broader area of New York state. However, as the census tractlevel data is gleaned from the same American Community Survey data from IPUMS, it is 3 deemed to be a consistent enough with the previously mentioned study and thus the same quantitative definitions can still be implemented in the data without the zip codes. New York City will be the test subject of ethnic enclaves in this paper, as it is one of the top states where immigrants live; New York is home to 4.5 million immigrants, which is 10% of the nation’s immigrant population. Having this many immigrants and having established possible enclave areas in New York City like Chinatown, Koreatown, Little India, etc. make the state an attractive test subject. As the populations of different ethnic groups vary throughout the city, the state is chosen to enlarge the sample size and expand the possible enclaves located not just within New York City, but to analyze the state of New York itself. 4 LITERATURE REVIEW Enclaves can play a large role in an immigrant’s experience, possibly providing mentors or access to opportunities that others wouldn’t have. A previous study notes that there are two main types of settlement patterns of immigrants who came to New York and Los Angeles after 1965 (Logan, Zhang, and Alba, 2002). There is the “immigrant enclave” model, where ethnic neighborhoods have high populations of a specific ethnicity, and can serve as transitional spaces for new immigrants, and the “ethnic community” model, where these neighborhoods are there due to the choice of the immigrants and not due to economic necessity. This indicates that these communities are able to exist not purely out of necessity, but because there are benefits that are built into the communities that immigrants would not find elsewhere. As these communities don’t exist only out of necessity, it may go to show that there is some other benefit that they can provide – possibly one that is financial and not just emotional – and thus make it so that individuals prefer to live in enclaves rather than not. The phrase “it takes a village” is often used when referring to a community that can be of aid or use when people become parents. The phrase may also be applied to an ethnic enclave, as living in that area may automatically create somewhat of a village around an individual, contributing to emotional or physical aid. It can also potentially translate into professional and financial success later on in life as well. Being surrounded by people that can create access to opportunities or provide valuable information is often how individuals are able to capture and create professional or educational success. Having a network of supporters built into one’s location and community creates an advantage that one did not have to seek out or create. 5 There are arguments supporting the other side, that enclaves are actually damaging to economic success and mobility, or there is little to no benefit, with research from Xie and Gough arguing that that for some immigrant groups, there is minimal support for the thesis that immigrants benefit from working in ethnic enclaves. Their research actually further indicates that for some immigrant groups, ethnic enclave participation actually has a negative effect on economic outcomes. The study uses data from the restricted version of the baseline round of the NIS, collected in 2003 and 2004, supplemented with contextual data from the 2000 census. They then create regression equations for log earnings regressed on education (pre- and post-immigration), job experience (pre- and post-immigration), and a vector of other covariates. They test a set of hypotheses regarding the effect of the different regressors on the outcome. Xie and Gough argue that “workers in ethnic enclaves are disadvantaged because owing to residential segregation, they are limited to receiving ‘undesirable jobs and poor wages’” (Xie & Gough, 2011). Stating that this disadvantage comes from business owners weaponizing ethnic solidarity to exploit their workers and pay them less than their counterparts who do not live in these enclaves. They assert that living in an enclave can be an isolating experience (whether by preference or not), and while it functions primarily as a cultural vehicle for immigrants, it can have “real consequences for the economic assimilation of immigrant workers” (Xie & Gough, 2011). Another study conducted by Sanders and Nee support the arguments of Xie and Gough, claiming that immigrants may often not meet the educational requirements that are valued or required by U.S. employers and thus their “foreign-earned human capital… is not highly valued” (Sanders & Nee, 1996). By coming to a new country with education and work experience from the immigrant’s sending country, many individuals may face 6 challenges like the one Sanders and Nee describe. Finding work is often difficult for those who are already assimilated and pre-evaluated by social norms and culture, and that can become exacerbated when the individual travels to a country like the United States, one that is often not prepared to evaluate education and previous work experience of foreignborn individuals. Sanders and Nee cite this exact reason as to why “many immigrants view self-employment as a route to upward mobility,” also stating that owning their own business allows them to utilize their human capital or class resources in the ways they value, not as “foreign capital” by American employers or business owners (Sanders & Nee, 1996). This can explain the desire to turn to self-employment opportunities, especially for those who are less proficient in English, as that relates directly into an individual’s ability to translate their “foreign capital” into something valued by employers. Other locations in the United States also have high volumes of immigrants, like California. One study, Noneconomic Effects of Ethnic Entrepreneurship: A Focused Look at the Chinese and Korean Enclave Economies in Los Angeles by Min Zhou and Myungduk Cho, looks more deeply into understanding the noneconomic effects of ethnic entrepreneurship, especially focused on the linkage between entrepreneurship and community building. This paper also argues that it is the social embeddedness of entrepreneurship, rather than individual entrepreneurs per se, that creates a unique social environment conducive to upward social mobility, and that ethnic entrepreneurship plays a pivotal role in immigrant adaptation beyond observable economic gains. Oftentimes, they also have a built-in consumer base, as these entrepreneurs are selling products or experiences that have cultural ties, something that appeals to an immigrant audience being so far away from where they first lived. The study uses ethnographic data from their 7 comparative case studies of the Chinese and Korean enclave economies in Los Angeles, as well as U.S. Census data. The case studies compare the economic structures and social contexts of Chinese and Korean enclaves, such as clustering of businesses, presence of the middle-class, and the development of institutions. Language proficiency for immigrants is also an incredibly important factor when considering the effect on earnings. As mentioned earlier regarding the study done by Sanders and Nee, barriers to finding work in the United States can prevent immigrants from having their full human capital value recognized. Language proficiency (in this case, English) plays a large part in how immigrants navigate the job market. In a study conducted by Barry R. Chiswick and Paul W. Miller, they aim to understand the determinants and consequences of immigrant and linguistic concentrations (enclaves). The study argues that enclaves are expected to have an adverse effect on the destination language proficiency of immigrants, and greater proficiency is expected to result in higher earnings, and thus a larger enclave is expected to have a negative effect on nominal earnings. It describes how there are communication costs involved for immigrants who choose to learn the dominant language of the country they immigrate to, and how these costs can potentially be reduced by “living and/or working in a linguistic concentration area” (Chiswick & Miller, 2005). They state that this is due to not every individual who is part of the group needing proficiency in the dominant language of the destination country, as those who have arrived previously can aid the newcomers in navigating their new circumstances by serving “as either direct or indirect translators for communication between the enclave and the host society” (Chiswick & Miller, 2005). 8 Chiswick and Miller also examine how other factors influence the immigrant experience for those who live in an enclave. They research something they label as “ethnic goods”, relating to “the consumption characteristics of an immigrant/ethnic group not shared with the host population, broadly defined to include market and non-market goods and services, including social interactions for themselves and their children with the people of their same origin” (Chiswick & Miller, 2005). The economies of scale regarding ethnic goods may put immigrants at a disadvantage, due to small member sizes to use common resources, or the cost of importing certain culturally significant ethnic goods individually. The cost of living in a specified area depends distinctly on the cost of these ethnic goods, and as the size of the enclave grows, the cost of the goods decreases, with this being true but reversed when the immigrant does not live in an enclave (Chiswick & Miller, 2005). While it does not directly affect their earnings, it can have an effect on their financial success, as they may end up paying more for their ethnic goods than someone who did not immigrate and their goods do not have the same importing or resource costs as that of an immigrant. Immigration policy can also affect individuals who have immigrated at different times or from different sending countries. It has also been a hot topic all throughout United States history, especially as sentiments toward immigration change, likely due to a lack of understanding on how these individuals function within the U.S. economy. In The Economics of Immigration by George J. Borjas, he argues that certain types of immigration policy affect immigrants from different host countries in a way that creates winners and losers. He also argues that more recent immigrants have lower wages than those who arrived before them. Borjas states that the significance behind studying how immigrants 9 perform in their destination country is because “immigrants who have high levels of productivity and who adapt rapidly to conditions in the host country’s labor market can make a significant contribution to economic growth,” and contrarily, “if immigrants lack the skills that employers demand and find it difficult to adapt, immigration may significantly increase the costs associated with income maintenance programs as well as exacerbate the ethnic wage differentials already existing in the host country” (Borjas, 1994). Education can also affect immigrants and those living in enclaves differently, with a study from Zeng and Xie stating that “past research has reported that Asian Americans, and Asian immigrants in particular, have lower earnings than do Whites within the same levels of education” (Zeng & Xie, 2004). The study conducted aims to understand why the earning disadvantage exists between the two groups. An interesting observation that Zeng and Xie report is that within the same levels of educational attainment, Asian Americans earn less than Whites, but if the comparison is narrowed to Asian Americans that are U.S.born and Whites, they are not generally disadvantaged (Zeng & Xie, 2004). As mentioned previously, this disparity in earnings can be attributed to the difficulty of getting one’s human capital value recognized as an immigrant. This is especially common for those who possess advanced degrees as they will need to undergo further testing and examination to prove their degree holds up to the same standards as one obtained in the destination country. Zeng and Xie delve further into why the location of the education obtained is important for immigrant earnings, stating reasons like the difference in the quality of education (especially higher education) being lower in sending countries that are still developing compared to the United States, certain fields of study being limited due to the scope of the material, in areas like law and politics, thus the time spent and knowledge acquired is not 10 easily transferable to the destination country, and that education in the United States has another benefit in the job market that is independent of the intrinsic value of education (Zeng & Xie, 2004). Such benefits include networking and internship opportunities, increased proficiency in English, and a broad understanding of American culture. The study goes on to describe exactly how such education in the United States and foreign countries differ, with the U.S. educated groups returning higher status observations in earnings, education, and occupation. They also found that foreign-educated Asian immigrants have the lowest income and occupational attainment, proving that even if these groups were to obtain the same level of educational attainment, their location of attainment influences their earnings at an extreme level (Zeng & Xie, 2004). 11 DATA The primary data source used in this study was the IPUMS USA Census Data from the 2018 American Community Survey (ACS) 1% sample. This data source is a weighted sample that is a 1-in-100 national random sample of the population. All summary statistics are conducted for individuals ages 25-64 in the state of New York. The secondary source that was used was the data containing census tract information, as the tracts were not available on IPUMS. The tract-level data was obtained from the Missouri Census Data Center (MCDC) using the Geographic Correspondence Engine. This data was then cleaned to contain only adults ages 25-64 in the state of New York. Due to the lack of female data analysis in existing ethnic enclave studies, it is important to include these individuals. The particular ethnicities studied in this paper are Chinese, Indian, Filipino, Vietnamese, Korean, and Japanese, as they are the top six ethnicities that make up the total U.S. population, from 1.6% to 0.5% as of 2022. After that, the 7th ethnic group drops to 0.2% of the total U.S. population (Wikipedia). Using these groups allows for a comparison that examines ethnic group participation across the most prominent ethnicities and how their differences across immigration patterns or cultural histories may or may not affect their financial success in terms of earnings. Randomization is also an important factor to consider when analyzing data. While one’s personal reasons for choosing or landing in New York may not be random, as the state is popular destination not just for immigrants but for U.S. born individuals as well, data collection methods can ensure a random sampling of the population. AS IPUMS data collection methods provide a weighted variable, after the data has been randomly collected to represent a 1-in-100 sample, it can be reinstated to accurately represent the true demographics of the population as a whole. 12 The dissimilarity and isolation indices were calculated by matching the IPUMS 2018 ACS 1% sample with their respective census tracts and using the tracts to identify neighborhoods in the state of New York. Census tracts contain between 2,500 and 8,000 residents and are created with an effort to make them as homogeneous as possible with respect to certain conditions like population characteristics, economic status, and living conditions in mind (Bureau, 2021). These tracts were summed across PUMAs (Public Use Microdata Areas), which contain no fewer than 100,000 people each and are statistical geographic areas that do not overlap (Bureau, n.d.). The equation used to calculate the dissimilarity index is as follows: 𝑛 1 𝑃1𝑖 𝑃2𝑖 𝐷 = ∑| − | 2 𝑃1 𝑃2 𝑖=1 where D is the Dissimilarity Index, P1i is the population of the ethnic group in tract і, P1 is the population of the ethnic group in the puma, P2i is the population of the nonethnic group in tract і, and P2 is the population of the nonethnic group in the puma. The equation used to calculate the isolation index is as follows: 𝑛 𝐼 = ∑| 𝑖=1 𝐴𝑖 𝐴𝑖 ∗ | 𝐴 𝑇𝑖 where I is the isolation index, Ai is the population of the ethnic group in tract і, A is the total population of the ethnic group in the puma, and Ti is the total population including the ethnic group in tract і. 13 The indices are calculated by collapsing the data by puma and tract to get one observation per tract so that they could be summed across pumas. This was done individually for the dissimilarity and isolation indices as well as for each of the six ethnicities that were being analyzed. There was also a separate dissimilarity index calculated for a group labeled as “other” containing Whites and the remaining ethnicities that were not chosen as study subjects. The isolation index was calculated for these individuals but not implemented back into the main dataset as their values would skew the dataset. As the isolation index calculates the spatial clustering of one such group, all isolation indices for the “other” group were between 0.5 and 1, which would heavily influence the regression model when merged back into the original data, so it was left unmerged. All other collapsed data files for the dissimilarity and isolation indices were merged back into the original dataset that was created by matching the IPUMS 2018 ACS with the MCDC census tracts. To ensure the dissimilarity indices were kept for those in each group; after merging each collapsed data file, the standing dissimilarity index was replaced with the dissimilarity indices of the collapsed data file only for the ethnic group the data file represented. Not all values were replaced each time, as it only rewrote the indices when the code for the ethnic group equaled 1. All values were eventually replaced, as the “other” data file ensured that those who did not have a dissimilarity index from one of the six ethnic group files received one from that file. By doing so, all values were eventually replaced to represent only their ethnic group, thus being able to correspond to a dissimilarity and isolation index only relevant to their ethnicity or subgroup. 14 Socio-demographic characteristics for adults living in New York City (2018) Chinese Indian Filipino Japanese Korean Vietnamese NH White Total Weighted N 403,825 230,193 63,661 25,412 80,711 16,688 5,734,267 Mean age in years 43.81 42.26 44.75 42.97 42.23 43.41 45.14 Mean earnings in 2017 $48,879 $66,766 $59,676 $73,972 $57,459 $51,819 $62,448 Sex Male 54.3% 50.3% 43.1% 36.7% 41.8% 47.1% 49.8% Female 45.7% 49.7% 56.9% 63.3% 58.2% 52.9% 50.2% Speaking English only at home 12.3% 31.5% 31.6% 16.9% 28.0% 19.3% 87.7% Speaking mother tongue at home 87.7% 68.5% 68.4% 83.1% 72.0% 80.8% 12.3% English Proficiency 62.8% 91.2% 98.6% 80.3% 85.7% 80.2% 98.8% Bilingual Ability 50.5% 59.7% 67.1% 63.3% 57.7% 61.0% 11.0% Marital Status Married 67.9% 72.3% 60.9% 72.4% 56.7% 57.1% 57.9% Divorced 5.8% 3.7% 6.2% 6.2% 4.4% 7.6% 10.4% Widowed 1.0% 1.1% 1.3% 1.1% 0.9% 1.5% 1.8% Separated 1.2% 0.9% 1.6% 2.8% 1.1% 0.3% 2.0% Never married/single 24.1% 22.0% 30.1% 17.5% 36.9% 33.5% 28.1% Educational Attainment Less than high school 22.9% 12.6% 3.8% 3.3% 3.7% 25.5% 5.0% High school graduates 18.7% 15.1% 9.5% 9.4% 13.0% 13.1% 22.5% Some college 13.8% 14.2% 20.0% 15.0% 12.6% 16.7% 25.5% Bachelor's degree 26.3% 27.7% 49.2% 46.6% 41.8% 27.9% 26.1% More than BA 18.3% 30.4% 17.7% 25.8% 28.9% 16.9% 21.0% Employment Employed 79.6% 80.1% 85.3% 79.8% 81.2% 82.3% 80.7% Unemployed 20.4% 19.9% 14.7% 20.2% 18.8% 17.7% 19.3% Nativity US born 17.3% 14.6% 18.7% 13.6% 20.9% 14.6% 88.4% Foreign born 82.7% 85.4% 81.3% 86.4% 79.1% 85.5% 11.6% Number of years living in the US <5 years 9.2% 11.8% 9.0% 23.9% 6.7% 5.5% 1.5% 5-9 years 11.0% 12.2% 6.3% 9.9% 5.1% 4.1% 1.4% 10+ years 62.5% 61.4% 66.0% 52.7% 67.3% 75.8% 8.7% Not asked for US-born adults 17.3% 14.6% 18.6% 13.6% 20.9% 14.6% 88.4% Management occupation 7.9% 10.8% 10.0% 16.3% 13.5% 13.9% 11.5% Professional occupation 30.9% 36.1% 48.0% 35.3% 38.7% 23.3% 33.1% Self-employed 7.0% 9.6% 5.3% 14.6% 12.8% 9.2% 9.3% Source: U.S. Census of Population and Housing: 2018 IPUMS published by Ruggles et al. (2025.) IPUMS, incorporated public use microdata sample 15 Descriptive statistics describing factors that affect earnings were calculated across to understand how these factors may affect earnings when regressed. As expected, there are clear differences across all groups, but the statistics do not vary extremely across the chosen demographics. As expected, Non-Hispanic Whites (NH Whites) are the largest group, at 5.73 million, vastly outnumbering the Asian subgroups. From the data in this source, there were 403,825 Chinese individuals, 230,193 Indian individuals, 63,661 Filipino individuals, 16,688 Vietnamese individuals, 80,711 Korean individuals, and 25,412 Japanese individuals identified using the observations and weighting them. The mean age did not vary largely, staying between 42 and 45 for all groups. However, earnings across groups did vary considerably, with Japanese individuals making the most at $73,972 annually, on average, and Chinese individuals earning the least on average at $48,879 annually. Gender-wise, the groups are mostly balanced, with Filipino, Japanese, and Korean adults containing a few more females (56.9%, 63.3%, and 58.2%, respectively). Chinese, Japanese, and Vietnamese individuals largely speak their mother tongue at home (87.7%, 83.1% and 80.8%, respectively) and while they have the highest percentages, the other ethnic groups are right behind them, being in the high 60s and low 70s. Indians (91.2%) and Filipinos (98.6%) show a large percentage of English proficiency, while Chinese individuals have the least proportion of English proficiency, at 62.8%. All groups demonstrate some sort of bilingual ability, with over 50% of each ethnic subgroup being able to speak both English and their mother tongue. Marriage rates are similar across the groups, with exceptions for those who never married or were single being very low (17.5%) for Japanese individuals. 16 Educational attainment varied quite significantly, with Filipino, Japanese, and Korean individuals having less than 3.8% of their population only completing less than high school attainment (3.8%, 3.3%, 3.7%, respectively). These three groups also saw a high attainment of a bachelor’s degree with Filipinos having 49.2%, Japanese having 46.6% and Koreans having 41.8% of their populations in that attainment category. In turn for more than a bachelor’s degree attainment, Indians, Japanese, and Koreans led in that category with 30.4%, 25.8%, and 28.9%, of their populations completing more education after a BA, respectively. Then for employment, there was little variance across population proportions for those who were employed, with the lowest proportion being 79.6% for Chinese, and the highest being 85.3% for Filipinos. There were a large proportion of individuals who were foreign born for each category, with Koreans to be the only subgroup with less than 80% of their population born abroad. The Japanese subgroup had the highest proportion of those who had been living in the United States for less than 5 years at 23.9%, while other subgroups reported 9.2% (Chinese), 11.8% (Indian), 9,0% (Filipino), 6.7% (Korean), and 5.5% (Vietnamese). The Japanese group also had the lowest proportion for those who had been living in the United States longer for 10 years or longer, at 52.7% and Vietnamese had the highest proportion at 75.8%. All other ethnic subgroups fell into the 61-67% range. Occupation-wise, Japanese individuals again took the highest proportion of managerial occupations at 16.3% and then it was Vietnamese at 13.9%, Koreans at 13.5%, Indians at 10.8%, Filipinos at 10.0% and with the lowest proportion, Chinese at 7.9%. This pattern changed, however, for the professional occupations, with Filipinos leading at 48.0%, then Koreans at 38.7%, Indians at 36.1%, Japanese at 35.3%, Chinese at 30.9%, and Vietnamese at 23.3%. Self-employed 17 statistics also emerged, with the Japanese group having the highest proportion at 14.6%, followed by Koreans at 12.8%, then Indians and Vietnamese at 9.6% and 9.3%, respectively, and then Chinese and Filipinos at 7.0% and 5.3%. 18 METHODOLOGY This study uses an Ordinary Least Squares (OLS) regression method, which examines the relationship between one or more independent variable(s) (regressors) and a dependent variable. This regression is a multiple linear regression, rather than a single linear regression, as it has more than one independent variable. OLS is able to analyze the relationship between the variables by fitting a line to the observed data – called the best fit line. This line works to minimize the sum of squared residuals, with residuals being the differences between the actual observed data values and the values on the best fit line, indicating how far off the model was from the actual data points. Squaring these residuals ensures that the negative and positive values do not cancel each other out. The model also made use of population weights, applying iweight=perwt to ensure representation. The regression equation used to calculate the coefficients for each race is as follows: log(𝐸𝑎𝑟𝑛𝑖𝑛𝑔𝑠𝑖 ) = 𝛽0 + 𝛽1 𝐷𝐼𝑖 + 𝛽2 𝐼𝑠𝑜𝐼𝑖 + 𝛾𝑋𝑖 + 𝛿𝑂𝑖 + 𝛽3 𝐷𝐼𝑖 𝑅𝑖 + 𝛽4 𝐼𝑠𝑜𝐼𝑖 𝑅𝑖 + 𝜀𝑖 where log (𝐸𝑎𝑟𝑛𝑖𝑛𝑔𝑠𝑖 ) is the outcome variable and describes pre-tax income (either from self-employment or wages) for the previous year for individual 𝑖, the 𝐷𝐼𝑖 is the dissimilarity index matched for each individual 𝑖 , and 𝐼𝑠𝑜𝐼𝑖 is the isolation index matched for each individual 𝑖. In this sort of log-linear regression, the coefficients represent the approximate percent changes in the dependent variable for a one-unit change in the independent variable. For example, a coefficient of 0.5 implies a 64.87% increase in earnings, which can be calculated using (𝑒𝛽 − 1) ∗ 100 or (𝑒 0.5 − 1) ∗ 100 in this case. 𝑋𝑖 contains all of the individual controls: female (coded 1 if sex is female, 0 if male), age, age squared, English proficiency (coded 1 if the individual spoke English well or very well), bilingual ability 19 (coded 1 if the individual spoke English very well or well and spoke a language other than English), foreign born (coded 1 for born outside of the United States and 0 if born within the United States), martial status (coded 1 to 5, 1 for married, the reference, 2 for divorced, 3 for widowed, 4 for separated, and 5 for never married), ethnicity (coded 0 to 7, 0 for all other races that were not the other six subgroups, 1 for Non-Hispanic Whites which was used as the reference, 2 for Chinese, 3 for Japanese, 4 for Filipino, 5 for Indian (Asian), 6 for Korean, and 7 for Vietnamese), usual hours worked per week, educational attainment (coded 1 to 5, 1 if the individual’s highest level of education completed was 12th grade, no diploma and lower, 2 if the individual’s highest level of education completed was a high school diploma or GED or alternative credential, 3 if the individual’s highest level of education completed was some college but less than 1 year, 1 or more years of college credit, or an associate’s degree, 4 if the individual’s highest level of education completed was a bachelor’s degree (reference), and 5 if the individual’s highest level of education completed was a master’s degree, a professional degree beyond a bachelor’s, or a doctoral degree), employed (coded 1 if the individual was in the labor force and 0 if they were not), and 𝑂𝑖 contains the occupational controls: management (coded 1 if the individual’s occupation was labeled as a manager in IPUMS, 0 otherwise), professional (coded 1 if the individual’s occupation was labeled under “Financial Specialists”, “Computer and Mathematical”, “Architecture and Engineering”, “Technicians”, “Life, Physical, and Social Science”, “Community and Social Services”, “Legal”, “Education, Training, and Library”, “Arts, Design, Entertainment, Sports, and Media”, “Healthcare Practitioners and Technical”, “Healthcare Support” and “Protective Service”), and self-employed (coded 1 if self-employed and 0 if works for wages or n/a). All coefficients were statistically 20 significant at the 0.01% level and as many of my regressors were binary, logistic regression was used. Additionally, interaction terms were used to predict earnings, as the dissimilarity and isolation indices varied for each group that was studied. These interaction terms were coded into the regression and for each index, 0 was omitted as the reference group. The first interaction term is represented by 𝛽3 𝐷𝐼𝑖 𝑅𝑖 , serving as the interaction term between the dissimilarity index represented by 𝐷𝐼𝑖 and the ethnicity groups represented by 𝑅𝑖 . The second interaction term is represented by 𝛽4 𝐼𝑠𝑜𝐼𝑖 𝑅𝑖 , which serves as the interaction term between the isolation index, represented by 𝐼𝑠𝑜𝐼𝑖 , and the ethnicity groups represented by 𝑅𝑖 . Note that the regressor for ethnicity groups is not included as a separate regressor in the regression equation above, as it is first factored into the individual controls under 𝑋𝑖 . Several variations of the regression above were ran to build upon one another and truly understand the effects of the regressors. These variations consisted of the regression above, but only with the dissimilarity index regressor (removing the isolation index regressor and the interaction terms for both indices), then with the dissimilarity index regressor and only the interaction term for the dissimilarity index, then another with just the isolation index regressor (without the dissimilarity index or either of the interaction terms), then one with the isolation index regressor and the isolation index interaction term, and finally one regression with all of the indices and interaction terms.. By running these variations, it allows for a clearer look into how these regressors individually affect earnings. These regressions will be referred to as regressions 1 through 5, with regression 1 representing only the dissimilarity index, regression 2 representing the dissimilarity index and the dissimilarity index interaction term, regression 3 representing only the isolation 21 index, regression 4 representing the isolation index and the isolation index interaction term, and regression 5 representing the regression with all of the indices and interaction terms. The regression equations for each regression are as follows: Regression 1: Dissimilarity Index regressor only log(𝐸𝑎𝑟𝑛𝑖𝑛𝑔𝑠𝑖 ) = 𝛽0 + 𝛽1 𝐷𝐼𝑖 + 𝛾𝑋𝑖 + 𝛿𝑂𝑖 + 𝜀𝑖 Regression 2: Dissimilarity Index regressor and Dissimilarity Index interaction term log(𝐸𝑎𝑟𝑛𝑖𝑛𝑔𝑠𝑖 ) = 𝛽0 + 𝛽1 𝐷𝐼𝑖 + 𝛾𝑋𝑖 + 𝛿𝑂𝑖 + 𝛽3 𝐷𝐼𝑖 𝑅𝑖 + 𝜀𝑖 Regression 3: Isolation Index regressor only log(𝐸𝑎𝑟𝑛𝑖𝑛𝑔𝑠𝑖 ) = 𝛽0 + 𝛽2 𝐼𝑠𝑜𝐼𝑖 + 𝛾𝑋𝑖 + 𝛿𝑂𝑖 + 𝜀𝑖 Regression 4: Isolation Index regressor and Isolation Index interaction term log(𝐸𝑎𝑟𝑛𝑖𝑛𝑔𝑠𝑖 ) = 𝛽0 + 𝛽2 𝐼𝑠𝑜𝐼𝑖 + 𝛾𝑋𝑖 + 𝛿𝑂𝑖 + 𝛽4 𝐼𝑠𝑜𝐼𝑖 𝑅𝑖 + 𝜀𝑖 Regression 5: All regressors and interaction terms log(𝐸𝑎𝑟𝑛𝑖𝑛𝑔𝑠𝑖 ) = 𝛽0 + 𝛽1 𝐷𝐼𝑖 + 𝛽2 𝐼𝑠𝑜𝐼𝑖 + 𝛾𝑋𝑖 + 𝛿𝑂𝑖 + 𝛽3 𝐷𝐼𝑖 𝑅𝑖 + 𝛽4 𝐼𝑠𝑜𝐼𝑖 𝑅𝑖 + 𝜀𝑖 22 Weighted Regression Predicting Earnings For Chinese, Japanese, Filipino, Indian (Asian), Korean, and Vietnamese Adults Ages 25-64 in New York State (2018), Only DI Term, No Interaction Terms β SE Female -0.2053*** 0.00059 Age 0.5639*** 0.00023 Age Square -0.0005*** 0.00000 English Proficiency 0.2387*** 0.00143 Bilingual Ability -0.0314*** 0.00084 Foreign Born -0.0099*** 0.00085 Marital Status Married (reference) Divorced -0.7978*** 0.00101 Widowed -0.0864*** 0.00253 Separated -0.1145*** 0.00179 Never married -0.1258*** 0.00071 Ethnicity Other -0.0668*** 0.00072 Non-Hispanic White (reference) Chinese -0.0423*** 0.00161 Japanese 0.0548*** 0.00577 Filipino 0.0380*** 0.00355 Indian (Asian) 0.0096*** 0.00201 Korean -0.0609*** 0.00323 Vietnamese 0.0562*** 0.00698 Usual Hours Worked Per Week 0.0394*** 0.00003 Educational attainment Less than high school -0.6350*** 0.00125 High school graduates -0.4649*** 0.00088 Some college -0.3261*** 0.00082 Bachelors (reference) More than BA 0.1543*** 0.00088 Employed 1.0212*** 0.00163 Management Occupation 0.3750*** 0.00098 Professional Occupation 0.2062*** 0.00069 Self Employed -0.3136*** 0.00096 Dissimilarity Index -0.3132*** 0.00611 Constant 6.7708*** 0.00528 Number of Cases 8,348,612 Adjusted R Square 0.4334 Source: U.S. Census of Population and Housing: 2018 IPUMS published by Ruggles et al. (2025.) *p<0.1 **p<0.05 ***p <0.01 (two-tailed tests) IPUMS, incorporated public use microdata sample 23 Weighted Regression Predicting Earnings For Chinese, Japanese, Filipino, Indian (Asian), Korean, and Vietnamese Adults Ages 25-64 in New York State (2018), DI Term and DI Interaction Term, No Isolation Terms β SE Female -0.2054*** 0.00059 Age 0.5636*** 0.00023 Age Square -0.0005*** 0.00000 English Proficiency 0.2381*** 0.00143 Bilingual Ability -0.0312*** 0.00084 Foreign Born -0.0996*** 0.00085 Marital Status Married (reference) Divorced -0.7990*** 0.00101 Widowed -0.0863*** 0.00253 Separated -0.1146*** 0.00179 Never married -0.1258*** 0.00071 Ethnicity Other -0.0744*** 0.00118 Non-Hispanic White (reference) Chinese -0.0082 0.00330 Japanese 0.0557*** 0.00848 Filipino 0.0067 0.00480 Indian (Asian) -0.0576*** 0.00348 Korean -0.0428*** 0.00432 Vietnamese -0.0793*** 0.00834 Usual Hours Worked Per Week 0.0394*** 0.00003 Educational attainment Less than high school -0.6350*** 0.00125 High school graduates -0.4650*** 0.00088 Some college -0.3263*** 0.00082 Bachelors (reference) More than BA 0.1542*** 0.00088 Employed 1.0211*** 0.00163 Management Occupation 0.3749*** 0.00098 Professional Occupation 0.2062*** 0.00069 Self Employed -0.3138*** 0.00096 Dissimilarity Index -0.3148*** 0.00748 Dissimilarity Interaction Term Other 0.1192*** 0.01476 Non-Hispanic White (reference) Chinese -0.5207*** 0.04383 Japanese 0.1087 0.06839 Filipino -0.4218*** 0.04413 24 Indian (Asian) -0.7717*** 0.04577 Korean -0.2956*** 0.04645 Vietnamese -0.2246*** 0.04468 Constant 6.7722*** 0.00529 Number of Cases 8,348,612 Adjusted R Square 0.4334 Source: U.S. Census of Population and Housing: 2018 IPUMS published by Ruggles et al. (2025.) *p<0.1 **p<0.05 ***p <0.01 (two-tailed tests) IPUMS, incorporated public use microdata sample 25 Weighted Regression Predicting Earnings For Chinese, Japanese, Filipino, Indian (Asian), Korean, and Vietnamese Adults Ages 25-64 in New York State (2018), Only Isolation Term, No Interaction Terms β SE Female -0.2050*** 0.00059 Age 0.5652*** 0.00023 Age Square -0.0005*** 0.00000 English Proficiency 0.2438*** 0.00143 Bilingual Ability -0.0350*** 0.00084 Foreign Born -0.0125*** 0.00085 Marital Status Married (reference) Divorced -0.7930*** 0.00101 Widowed -0.0859*** 0.00253 Separated -0.1127*** 0.00179 Never married -0.1263*** 0.00071 Ethnicity Other -0.0640*** 0.00072 Non-Hispanic White (reference) Chinese -0.0723*** 0.00165 Japanese 0.0532*** 0.00577 Filipino 0.0432*** 0.00355 Indian (Asian) 0.0841*** 0.00201 Korean -0.0546*** 0.00323 Vietnamese 0.0500*** 0.00698 Usual Hours Worked Per Week 0.0394*** 0.00003 Educational attainment Less than high school -0.6331*** 0.00125 High school graduates -0.4630*** 0.00088 Some college -0.3249*** 0.00082 Bachelors (reference) More than BA 0.1551*** 0.00088 Employed 1.0203*** 0.00163 Management Occupation 0.3749*** 0.00098 Professional Occupation 0.2063*** 0.00069 Self Employed -0.3136*** 0.00096 Isolation Index 0.2293*** 0.00283 Constant 6.7259*** 0.00527 Number of Cases 8,348,612 Adjusted R Square 0.4345 Source: U.S. Census of Population and Housing: 2018 IPUMS published by Ruggles et al. (2025.) *p<0.1 **p<0.05 ***p <0.01 (two-tailed tests) IPUMS, incorporated public use microdata sample 26 Weighted Regression Predicting Earnings For Chinese, Japanese, Filipino, Indian (Asian), Korean, and Vietnamese Adults Ages 25-64 in New York State (2018), Isolation Term and Isolation Interaction Term, No DI Terms β SE Female -0.2051*** 0.00059 Age 0.5624*** 0.00023 Age Square -0.0005*** 0.00000 English Proficiency 0.2378*** 0.00143 Bilingual Ability -0.0365*** 0.00084 Foreign Born -0.0143*** 0.00085 Marital Status Married (reference) Divorced -0.7998*** 0.00101 Widowed -0.0853*** 0.00253 Separated -0.1133*** 0.00179 Never married -0.1285*** 0.00071 Ethnicity Other -0.0444*** 0.00087 Non-Hispanic White (reference) Chinese 0.0798*** 0.00274 Japanese 0.0998*** 0.00692 Filipino 0.0616*** 0.00418 Indian (Asian) 0.0694*** 0.00321 Korean 0.0502*** 0.00413 Vietnamese 0.1121*** 0.00753 Usual Hours Worked Per Week 0.0394*** 0.00003 Educational attainment Less than high school -0.6304*** 0.00125 High school graduates -0.4601*** 0.00088 Some college -0.3222*** 0.00082 Bachelors (reference) More than BA 0.1538*** 0.00088 Employed 1.0204*** 0.00163 Management Occupation 0.3739*** 0.00098 Professional Occupation 0.2061*** 0.00069 Self Employed -0.3145*** 0.00096 Isolation Index 0.4006*** 0.00380 Isolation Interaction Term Other -0.2581*** 0.00623 Non-Hispanic White (reference) Chinese -0.7897*** 0.01059 Japanese -0.6368*** 0.05685 27 Filipino -0.2197*** 0.03782 Indian (Asian) -0.6272*** 0.02495 Korean -1.4569*** 0.03687 Vietnamese -0.8600*** 0.04094 Constant 6.7259*** 0.00527 Number of Cases 8,348,612 Adjusted R Square 0.4342 Source: U.S. Census of Population and Housing: 2018 IPUMS published by Ruggles et al. (2025.) *p<0.1 **p<0.05 ***p <0.01 (two-tailed tests) IPUMS, incorporated public use microdata sample 28 Weighted Regression Predicting Earnings For Chinese, Japanese, Filipino, Indian (Asian), Korean, and Vietnamese Adults Ages 25-64 in New York State (2018), All Indices and Interaction Terms Included β SE Female -0.2051*** 0.00058 Age 0.5614*** 0.00022 Age Square -0.0005*** 0.00000 English Proficiency 0.2371*** 0.00143 Bilingual Ability -0.0362*** 0.00084 Foreign Born -0.0151*** 0.00085 Marital Status Married (reference) Divorced -0.7982*** 0.00101 Widowed -0.0853*** 0.00253 Separated -0.1141*** 0.00178 Never married -0.1289*** 0.00071 Ethnicity Other -0.0552*** 0.00124 Non-Hispanic White (reference) Chinese 0.0895*** 0.00377 Japanese 0.1479*** 0.01758 Filipino 0.0638*** 0.00562 Indian (Asian) -0.0466*** 0.00276 Korean -0.1624 0.00492 Vietnamese -0.0582*** 0.00964 Usual Hours Worked Per Week 0.0394*** 0.00003 Educational attainment Less than high school -0.6630*** 0.00125 High school graduates -0.4590*** 0.00088 Some college -0.3211*** 0.00082 Bachelors (reference) More than BA 0.1538*** 0.00088 Employed 1.0200*** 0.00163 Management Occupation 0.3738*** 0.00098 Professional Occupation 0.2063*** 0.00069 Self Employed -0.3144*** 0.00096 Dissimilarity Index -0.4464*** 0.00856 Isolation Index 0.4345*** 0.00380 Isolation Interaction Term Other -0.2791*** 0.00629 Non-Hispanic White (reference) Chinese -0.8053*** 0.01070 29 Japanese 1.0225* 0.37840 Filipino 0.6247*** 0.12643 Indian (Asian) -0.3148*** 0.02928 Korean -2.4140*** 0.06375 Vietnamese 3.1593*** 0.39986 Dissimilarity Interaction Term Other 0.1949*** 0.01488 Non-Hispanic White (reference) Chinese -0.1509* 0.04426 Japanese -1.6051*** 0.45317 Filipino -0.6171*** 0.14615 Indian (Asian) -0.7611*** 0.05366 Korean 1.8552*** 0.07987 Vietnamese -3.9528*** 0.43185 Constant 6.7543*** 0.00529 Number of Cases 8,348,612 Adjusted R Square 0.4345 Source: U.S. Census of Population and Housing: 2018 IPUMS published by Ruggles et al. (2025.) *p<0.1 **p<0.05 ***p <0.01 (two-tailed tests) IPUMS, incorporated public use microdata sample 30 ANALYSIS Each of the regressions has a differing R square value, although they are similar to one another. The regressions’ R square values are 0.4334, 0.4334, 0.4345, 0.4342, and 0.4345 for regression 1, regression 2, regression 3, regression 4, and regression 5, respectively. The R square value that the regression provides implies that the models can predict 43.34%, 43.34%, 43.45%, 43.42%, and 43.45%, respectively, of the variance in the dependent variable (log earnings). This suggests that each of the regression models have some prediction power regarding the earnings but does not completely predict the outcome. As for the indices, each regression provides new information regarding the effects of living in an ethnic enclave predicted by the dissimilarity and isolation indices. Before interpreting the regression results, it should be noted that the dissimilarity and isolation indices were measured on a scale of 0 to 1, and thus a one-unit increase in either index signifies moving from total integration to total segregation in the dissimilarity index and moving from completely dispersed (no spatial clustering) to complete spatial clustering. In regression 1, where only the dissimilarity index is used as a regressor alongside the other controls, the coefficient is -0.3132, meaning that for every one unit increase in the index, there is a 0.3132 log-point decrease or a 26.89% decrease in annual earnings, all else equal. This indicates that living in areas of high segregation negatively impacts earnings, regardless of race. When adding the interaction terms, it is to be noted that the R square value does not change, meaning regressions 1 and 2 have the same predictive power regarding the variance in annual earnings. The coefficient on the dissimilarity index does change to -0.3148, a 0.0016 log-point difference, now suggesting a 27.00% decrease in annual earnings, but only for the reference group, Non-Hispanic 31 Whites. Now that there is an interaction term, there is more information about how each ethnic group is affected by segregation. All ethnic groups other than “other” and Japanese have an extra negative affect associated with the dissimilarity index, with Chinese individuals having an extra -0.5207 log-points, Filipino with -0.4218, Indian with -0.7717, Korean with -0.2956, and Vietnamese with -0.2246. As these ethnic groups’ interaction term coefficients are negative, their annual earnings decrease by an extra 40.59%, 34.41%, 53.78%, and 25.59% from a one unit increase in the dissimilarity index compared to NH Whites, holding all else constant. The “other” group’s annual earnings increases by 0.1192 log-points or 12.66%, and Japanese individuals increase by 0.1087 log-points or 11.48%, holding all else constant. However, the Japanese coefficient is not statistically significant, thus these individuals’ do not have an extra effect associated with the dissimilarity index. Regressions 3 and 4 contain the isolation index and interaction term, indicating the effects of spatial clustering. Regression 3’s isolation index has a coefficient of 0.2293, indicating a 0.2293 log-point or 25.77% increase in annual earnings for those individuals who are exposed to the same minority group, holding all else equal. When examining the coefficient for regression 4, it is found to be a 0.4006 log-point or 49.27% increase in annual earnings for NH Whites, all else constant. The interaction term also provides interesting information, as all ethnic subgroups compared to NH Whites have a smaller effect on annual earnings, all else equal. The other group has a -0.2581 log-point or 22.75% decrease, with an overall isolation index coefficient of 0.1425, then Chinese had a -0.7897 log-point decrease, overall of -0.3891 thus a negative effect on annual earnings of 32.23%, Japanese had a -0.6368 log-point decrease, overall -0.2362 log-points and is negatively associated with earnings by 21.04%, Filipino had a -0.2197 log-point or 19.72% decrease, 32 overall still positive of 0.1809, Indian had a -0.6272 log-point decrease, overall of -0.2266 log-points or 20.28% decrease in annual earnings, Korean had a -1.4569 log-point decrease, with an overall effect of -1.0563 log-points or 65.23% decrease, and Vietnamese had a 0.8600 log-point decrease, overall of -0.4594 log-points or 36.83% decrease, holding all else equal. This signals that the NH White group benefits the most from spatial clustering, and all others benefit less, with it becoming a negative effect for some, all else equal. It could also be a sample size error, as there is significantly more possibility that the NH White group are exposed to other members of the NH White group, as there are 5,734,267 of those individuals compared to all of our other ethnic groups that have populations of 400,000 or less. It is notable that the overall effect is negative after factoring in log-point changes, for every ethnic subgroup besides Filipino and “other”. This suggests that spatial clustering actually does not benefit these ethnic minorities, and works against them. In regression 5, results suggest that there is a negative effect associated with the dissimilarity index, as the coefficient on the DI, β1, is -0.4464 and indicates that for every one unit increase in the DI, there is a 0.4464 log-point or 36.01% decrease in the log of annual earnings for the NH White group, holding all else constant. As the DI observes segregation amongst individuals, this outcome indicates that the more the tract (neighborhood measure) is segregated, the less an individual residing in that tract earns annually, holding all other variables constant. This coefficient is statistically significant at the 1% level, indicating that the change earnings due to the DI is significantly different from individuals in the NH White ethnic group for the individuals in the ethnic subgroups. This signals that ethnic segregation and low financial success correspond, suggesting that due to the cultural and social segregation individuals experience, their economic 33 opportunities are limited, perhaps due to lack of access to networks, certain industries, or mobility. This is also consistent with regressions 1 and 2, as both of those regression models had a negative coefficient on the DI. This, however, contrasts with the isolation index, another predictor of ethnic enclaves in the model, similar to the outcomes that were shown in regressions 3 and 4. The coefficient β2 on the isolation index is not only positive, but creates an almost 200% difference between the indices, as for every one unit increase in the isolation index, there is a 0.4345 log-point or 54.42% increase in the log of annual earnings for NH Whites, holding all else constant. The isolation index measures spatial clustering, so this indicates that living near or within an area that is spatially clustered with ethnicities similar to one’s own may yield higher annual earnings for NH Whites. However, when analyzing the interaction terms in regressions 4 and 5, most ethnic subgroups tend to be negatively affected by the isolation index, all else equal. The controls on the model correspond to previously existing statistical information, with the female regressor indicating that females earn annually about 0.2051 log-points or 22.76% less than males holding all else equal, age increases earnings by .5614 log-points or 75.31% per year, holding all else constant, but at a diminishing rate of 0.0005 log-points or 0.04%, holding all else constant, which is reflected in the age square regressor. English proficiency has a positive effect on annual earnings of 0.2371 log-points or 26.76%, holding all else constant, but bilingual ability has a negative effect of 0.0362 log-points or 3.56%, holding all else constant. This difference in language proficiency could be due to the fact that the individuals who were counted within the English proficiency included those who indicated that they could only speak English, which entertains Zeng and Xie’s 34 ideas from before about how foreign-educated immigrants earn less, as that group likely would be represented in the bilingual ability control, but not within the English proficiency control, as they are not likely to only speak English after obtaining a formal education in another country. Any marital status when compared to married has a negative effect on individual earnings, these being divorced with a -0.7982 log-points or -54.99% effect on annual earnings, widowed with a smaller negative effect of -0.0853 log-points or -8.18%, separated with a -0.1141 log-points or -10.78% effect, and never married with a -0.1289 log-points or -12.09% effect, holding all else constant. These numbers do not come as a surprise, as it is logistically sound that married individuals would make more than their divorced, widowed, separated, or single counterparts. However, it is surprising that there is such a large gap between married and divorced individuals. Results attached to the ethnic group regressor have varied effects on annual earnings. When compared to NH Whites, the “other” group, containing all of the ethnicities not included in the other subgroups, performs 0.0552 log-points or 5.37% less, holding all else constant. Chinese, Japanese, and Filipino individuals perform better, by 0.0895 logpoints or 9.36%, 0.1479 log-points or 15.94%, and 0.638 log-points or 89.26% better, respectively, holding all else constant. Indian (Asians), Korean, and Vietnamese individuals all perform worse than the NH White group, at 0.0466 log-points or 4.55%, 0.1624 log-points or 14.99%, and 0.0582 log-points or 5.65%, respectively, holding all else constant. It should be noted that the coefficient for Koreans is not statistically significant, so while the data observes that they perform 0.1624 log-points or 14.99% worse than their NH White counterparts, holding all else constant, it is not statistically significant at any of the percent levels, so in reality they do not perform any differently. The coefficient attached 35 to usual hours worked per week indicates that for every one unit increase in the usual hours worked per week, there is a 0.394 log-point or 48.29% increase in annual earnings, all else held constant. The educational attainment control also produces results that are expected, with the individuals who have educational attainments of less than high school, high school graduates, and some college performing worse than individuals who have completed a bachelor’s degree, at 0.6630 log-points or 94.06%, 0.4590 log-points or 58.25%, and 0.3211 log-points or 37.86% worse, holding all else constant. Then, an increase can be noted for those who complete more than a bachelor's degree, with a 0.1538 log-point or 16.63% increase in annual earnings from those who have completed a bachelor's degree, holding all else constant. The coefficient on the employment control shows a significant increase in earnings for individuals who are employed compared to those who are not employed, of 1.0225 log-points or 178.01%, holding all else constant. Occupational controls included an occupation in management, which resulted in a 0.3738 log-point or 45.32% increase in annual earnings for those who did have management occupations compared to those who did not, all else held constant. Individuals who had an occupation labeled as “professional” saw a 0.2063 log-point or 22.91% increase in annual earnings on average compared to those who did not have an occupation labeled as “professional”. Individuals who were self-employed saw a 0.3144 log-point or 26.98% decrease in annual earnings on average compared to those who worked for wages, holding all else constant. Returning to the interaction terms, in regression 5, both indices provide further valuable insight into how these indices affect earnings for each ethnic subgroup. For the interaction terms regarding both indices, the reference group is NH White, same as the 36 reference group in the ethnicity control. Thus, all resulting interaction coefficients are interpreting the effect of how each ethnicity’s dissimilarity or isolation index differs from the NH White group. Starting with the dissimilarity index interaction terms, if the individual is within the “other” group of ethnicities, a one-unit increase in dissimilarity index is associated with a 0.1949 log-point or 21.52% increase in earnings relative to NH Whites, the overall coefficient is -0.2515 or a 22.24% decrease. For individuals labeled as “Chinese,” a one-unit increase in the dissimilarity index is associated with a 0.1509 logpoint or 14.01% decrease in earnings relative to NH Whites, overall it is -0.5973 or a 44.97% decrease. This decrease can be interpreted as Chinese individuals who live in areas with high segregation between Chinese members and all other ethnicities earn less on average. When using the dissimilarity index to indicate an enclave, this combined with the 0.8053 log-point 55.30% decrease in earnings for every one-unit increase in the isolation index relative to NH Whites for Chinese individuals, shows that individuals who live in areas with higher dissimilarity and isolation indices (therefore enclaves) are affected negatively – in terms of annual earnings – by living in those areas. However, it can also be seen that the coefficient on the dissimilarity interaction term for Chinese individuals is only significant at the 5% level, meaning that there is 5% probability that Chinese individuals annual earnings are not different from their NH White counterparts. This is something to note, and although there is a probability, as it is so low it is unlikely that such a result would be observed. Simply put, Chinese individuals that live in Chinese ethnic enclaves earn less on average than their NH White counterparts, all else constant. This same analysis can be done for the remaining ethnicities, with Japanese individuals having a 1.6051 log-point or 79.91% decrease for every one unit increase in 37 the dissimilarity index, which overall comes to -2.0515 log-points or a 87.11% decrease, holding all else constant. This means that their annual earnings are heavily negatively affected by living in segregated areas compared to NH Whites. However, there is a large positive effect associated with the isolation index for Japanese individuals of 1.4570 logpoints, which suggests that Japanese subgroups who live in areas with high segregation are affected negatively, but if they are surrounded by other Japanese neighborhoods, they are affected positively, showing an opposing effect between the two indicators. This could be due to the small number of Japanese individuals throughout the state of New York, as the number of members needed to make an enclave could be considerably more than there are clustered in one area. Also to note is that the isolation interaction term on the Japanese individuals was only significant at the 5% level, so there is more probability that their earnings are not significantly different from the NH White group. Filipino individuals also experienced the opposing effects of the indices, similar to the Japanese individuals where the dissimilarity interaction term was negative, at a 0.6171 log-point or 46.05% decrease compared to NH Whites, and in total a 1.0635 log-point or 65.52% annual earnings decrease for every one-unit increase in the dissimilarity index, and the isolation interaction term was also largely positive, at a 0.6247 log-point or 86.77% compared to the NH Whites and overall 1.0592 log-point or 188.49% increase in annual earnings for every one-unit increase in the isolation index, holding all else constant. This suggests the same as the Japanese situation, where areas of high segregation return lower earnings compared to the NH White group, but areas with high spatial clustering returned higher annual earnings on average. This could be attributed to again the small population of the Filipino community, so while they may not have enough individuals to sway the 38 dissimilarity index, but there could still be areas where there was clustering of Filipinos but not enough of them to tip the segregation index. Indian (Asian - indicating that they are not Caribbean Indians or Indigenous peoples) individuals also had flipped effects for the indices with the dissimilarity index resulting in a negative effect on annual earnings with a 0.7611 log-point or 52.32% decrease from the NH Whites coefficients, shown in the dissimilarity interaction term and the isolation index resulting in a 0.3148 log-point or 27.01% decrease from the NH Whites, holding all else constant. Overall, the Indian coefficient for the dissimilarity index was -1.2075, indicating a negative log-points difference of 1.2075 and a 70.14% decrease in annual earnings, holding all else constant. For the isolation index, annual earnings increased by 0.1197 logpoints or 12.72%, holding all else constant. This shows that individuals who are Asian Indians do not experience a positive effect on annual earnings if they live in an area of high segregation from other ethnic groups, and experience a negative effect compared to their NH White counterparts. They do however receive the same positive effect from the spatial clustering as Filipino and Japanese individuals. The Korean subgroup displayed another mix in the interaction terms, with the dissimilarity index interaction term displaying a 1.8552 log-point increase (significantly large) and the isolation index interaction term displaying a 2.4140 log-point 91.05% decrease (also significantly large) in annual earnings for Koreans compared to the NH White group. The total coefficients were 1.4088 log-points or a 309.01% increase associated with the dissimilarity index and -1.9795 log-points or a 86.21% decrease in annual earnings associated with the isolation index. This indicates that there is an extremely strong effect of segregation and spatial clustering for Korean individuals, just in opposite 39 directions. The interpretation of these coefficients is that living in segregated areas has a large positive effect on annual earnings while living in spatially clustered areas has a large negative effect on annual earnings for Korean individuals compared to their NH White counterparts. This could be explained if Korean spatial clustering limited economic opportunity and mobility rather than improving or expanding it. Thus, living in a segregated area that would consist of less Korean individuals rather than being completely surrounded in terms of an enclave more positively affects earnings, and by a significant amount, holding all else constant. Lastly, the Vietnamese interaction terms also resulted in opposing coefficients, and by almost a large amount in the opposite direction, similar to the Korean coefficients. The dissimilarity interaction term for these individuals was -3.9528, indicating a log-point decrease of 3.9528 or 98.08% in annual earnings if the individual is Vietnamese compared to NH Whites, all else constant. The isolation interaction term was 3.1593, indicating a 3.1593 log-point increase in annual earnings for Vietnamese individuals compared to NH Whites, holding all else constant. Overall coefficients were -4.3992 log-points (98.77% decrease) for the dissimilarity index, and 3.5938 log-points (3536% increase – extremely high) for the isolation index. This giant disparity in the indices coefficients as well as the interaction terms could be due to outliers – seen with the large standard errors of 0.43185 (DI) and 0.39986 (Isolation Index) – or like the Japanese group, this disparity could be explained by the same N size, as there were only 16,688 individuals after weighting the population. Like many of the other ethnic subgroups, the segregation aspect proves to have a negative effect on earnings while the spatial clustering effect results in a positive effect on earnings. It could also be attributed to strong community based networks within the 40 community or lower barriers to employment, if the coefficients are truly representative of the real effect and not simply statistical noise. There could also be concentration of Vietnamese subgroups throughout the state that create place-based effects that are misattributed toward spatial clustering. 41 DISCUSSION Overall, the regression models provide valuable insights and a nuanced perspective into how segregation and spatial clustering affect earnings overall across ethnic subgroups. However, due to contradicting outputs, it produces an interesting outlook into enclave formation and definition. While the main effects of the interaction terms found in the regressions may show a general pattern, analyzing each part of the interaction term for both indices reveals that the economic consequences of segregation and spatial clustering are not uniform for the ethnic subgroups but rather deeply dependent on group-specific dynamics. The dissimilarity index across all five regressions displays a generally negative main effect, suggesting that segregation across neighborhoods – interpreted in this study as a separation from the majority population that does not include one’s ethnic group – largely impacts earnings negatively, all else held constant. However, the interaction terms uncover information regarding the specific ethnic subgroups that is not consistent with the main effect. Korean individuals are seen to benefit from a higher dissimilarity (i.e. more segregated communities), while it has a detrimental effect for Vietnamese individuals. This indicates that the dissimilarity index cannot be uniformly translated across ethnic groups and may reflect specific group characteristics or cultural values. As for the isolation index, the main effect is positive across the regressions, but when analyzing how the interaction terms affect each subgroup’s earnings, the effect does not stay positive. Opposite to how these individuals were affected by the dissimilarity index, Koreans face a large negative effect while Vietnamese individuals experience a significant 42 positive effect on earnings. These outcomes suggest that spatial clustering may create differing economic advantages and disadvantages for each subgroup. Such results emphasize the limitations of treating ethnic enclave effects as allencompassing, especially without considering socio-historical contexts or cultural backgrounds. Further quantitative definitions for an ethnic enclave may be necessary in a regression model to more accurately predict the effect of these regressors on annual earnings. While this model may not serve as an argument toward the binary framing of ethnic enclaves as beneficial or harmful, it exposes a complex underlying structure that depends on certain aspects across ethnic subgroups that work to shape economic outcomes in myriad ways. 43 APPENDIX APPENDIX A: 1.5 generation of immigrants Originally, this paper was to be based on the Generation and Earnings Patterns Among Chinese, Filipino, and Korean Americans in New York study conducted by Sookhee Oh and Pyong G. Min which studied another generation, the 1.5 generation, and the differences it has from first- and second-generation immigrants, specifically tied to earnings in the New York Metropolitan Area in 2000. While this paper served as a great starting point, the research conducted in this paper took a different route. However, it may still be valuable to see the research that was done to duplicate Oh and Min’s study, as well as how some of the summary statistics translated from the year 2000 to 2018. Oh and Min define the 1.5 generation as individuals as those who were born in their home countries and immigrated to the United States at age 12 or earlier. (Oh and Min, 2011). They define this generation as they believe that they have a unique experience as immigrants in the United States that differs from the experience of those in both the first and second generations. The difference is shown within a bilingual and bicultural ability, English proficiency, ethnic attachment, and ethnic identity. These differences could potentially translate into earnings differences across these generations. The IMR researchers only studied one location – one with an especially high Chinese population. They mentioned that this may have an effect on their results, due to the “effect of living in an enclave … [varies] considerably in economic attainments by ethnic group because ethnic institutionalization may be associated with alternative labor-market opportunities.” (Oh and Min, 2011). They also mention that “those in New York City may have an 44 advantage because of many professional and managerial occupations available in different Chinese enclaves there.” (Oh and Min, 2011). APPENDIX B: Summary Statistics for 2000 and 2018 data in New York-New Jersey and Los Angeles Metropolitan Areas The following figures contain the summary statistics for the New York-New Jersey and Los Angeles Metropolitan areas for the years 2000 and 2018. The tables were created based on the Generation and Earnings Patterns Among Chinese, Filipino, and Korean Americans in New York study conducted by Sookhee Oh and Pyong G. Min and thus are an attempt at duplicating their Table 2 and then transferring it to another area and year. Generation Total 31,729 100.0% 12.5 87.5 75.9 63.4 3.1 7.5 18.1 58.8 12.4 13.6 23.3 6.1 42 1st 31,094 85.9% 2.3 97.7 22.6 20.3 7.8 21.3 14.8 38.1 18.0 18.1 15.0 29.7 44 Korean 2nd 1.5 1,319 3,782 3.6% 10.4% 65.8 23.3 34.2 76.7 94.2 85.7 28.4 62.4 3.1 1.6 1.2 4.7 10.5 10.9 45.8 49.4 39.4 33.3 18.0 21.1 40.5 32.9 6.7 9.7 30 30 Total 60,962 100.0% 17.5 82.5 75.5 58.0 4.7 9.4 25.5 53.5 6.9 14.1 17.7 5.8 42 1st 42,932 83.1% 2.1 97.9 21.9 19.8 5.4 16.6 17.9 42.4 17.7 21.0 15.6 37.8 44 Korean 2nd 1.5 2,320 6,401 4.5% 12.4% 57.5 22.0 42.5 78.0 92.5 86.1 35.0 64.0 3.3 5.1 5.7 3.1 19.1 19.8 40.7 52.2 31.2 19.8 16.6 23.5 43.2 30.0 20.7 13.9 35 32 Descriptive Statistics: Males Age 25-64 in Labor Force in the Los Angeles Metropolitan Area (2000) Filipino Chinese 2nd 1.5 1st Total 2nd 1.5 1st 7,228 6,650 47,084 71,933 10,066 8,203 53,664 N 11.9% 10.9% 77.2% 100.0% 14.0% 11.4% 74.6% Percent 80.9 44.9 4.0 13.9 66.2 16.8 3.7 Speaking English only at home (%) 19.1 55.1 96.1 86.1 33.9 83.2 96.3 Speaking mother tongue at home (%) 98.5 92.2 69.6 49.6 93.8 81.4 36.5 Speaking English very well (%) 17.6 47.3 65.7 35.7 27.7 64.6 32.8 Bilingual ability (%) 5.1 4.1 4.7 15.3 2.5 4.4 19.4 Less than high school (%) 10.8 11.0 9.0 9.6 6.2 5.6 10.9 High school graduates (%) 28.1 32.4 24.2 12.4 12.8 20.1 11.2 Some college (%) 48.6 44.8 55.4 36.8 49.4 47.5 32.8 BA or associate degree (%) 7.4 7.8 6.7 25.8 29.2 22.5 25.7 More than BA (%) 15.1 15.4 13.8 21.5 24.4 22.7 20.7 Management occupation (%) 27.8 23.1 15.4 24.3 33.5 34.1 21.1 Professional occupation (%) 5.8 2.6 6.3 16.5 12.9 11.3 18.0 Self-employed (%) 35 33 44 42 39 33 44 Mean age (years) Source: U.S. Census of Population and Housing: 2000 IPUMS published by Ruggles et al. (2004.) IPUMS, incorporated public use microdata sample Generation Filipino Chinese 2nd 1.5 1st Total 2nd 1.5 1st 2,442 2,803 10,582 107,059 26,484 9,346 87,131 N 7.7% 8.8% 83.5% 100.0% 9.9% 8.7% 81.4% Percent 73.3 44.7 3.5 8.6 45.9 15.1 3.4 Speaking English only at home (%) 26.7 55.3 96.5 91.4 54.1 85.0 96.6 Speaking mother tongue at home (%) 94.4 93.3 72.4 39.3 94.0 71.1 29.2 Speaking English very well (%) 21.0 48.6 68.9 30.7 48.1 56.1 25.8 Bilingual ability (%) 3.5 2.0 3.2 29.0 3.1 9.4 34.2 Less than high school (%) 13.0 6.7 7.1 15.8 8.7 8.7 17.4 High school graduates (%) 13.6 21.3 18.2 10.0 16.1 17.9 8.4 Some college (%) 51.8 55.2 59.9 23.2 46.1 46.8 17.9 BA or associate degree (%) 18.1 14.8 11.6 22.0 26.0 17.3 22.1 More than BA (%) 12.8 25.4 12.4 13.9 23.4 20.0 12.1 Management occupation (%) 33.9 25.8 22.0 18.5 28.2 25.9 16.2 Professional occupation (%) 4.0 2.1 6.7 10.6 8.8 6.8 11.2 Self-employed (%) 34 31 44 41 36 34 43 Mean age (years) Source: U.S. Census of Population and Housing: 2000 IPUMS published by Ruggles et al. (2004.) IPUMS, incorporated public use microdata sample Descriptive Statistics: Males Age 25-64 in Labor Force in the New York-New Jersey Metropolitan Area (2000) Total 51,653 100.0% 7.0 93.0 33.0 26.0 5.3 14.4 18.2 43.5 18.6 21.1 18.7 34.1 42 Total 36,195 100.0% 6.8 93.2 31.8 25.0 7.0 18.8 14.2 39.6 20.4 18.4 17.8 26.7 42 US.-Born NH White Total 1,110,699 100.0% 86.7 13.3 95.9 9.2 5.6 14.4 25.3 36.7 18.0 23.2 23.3 19.2 43 US.-Born NH White Total 2,079,100 100.0% 82.9 17.1 93.7 10.7 6.7 21.4 17.9 33.2 20.8 21.5 19.9 15.6 42 45 Generation Total 44,981 100.0% 29.4 70.6 79.6 66.3 2.0 9.7 16.3 58.1 14.0 14.0 38.4 4.7 44 1st 28,852 61.6% 9.8 90.2 36.4 63.5 4.6 13.6 6.1 39.3 36.4 20.4 33.1 22.2 49 Korean 2nd 1.5 10,757 7,224 23.0% 15.4% 40.1 37.7 59.9 62.3 96.6 93.4 59.2 59.5 0.0 2.1 5.4 4.3 9.9 4.3 60.1 54.5 24.6 34.7 21.8 23.8 31.9 38.3 9.8 20.0 34 38 Total 97,337 100.0% 29.1 70.9 78.8 66.8 2.9 10.4 23.8 53.8 9.0 8.8 34.1 7.6 44 1st 42,789 56.5% 6.1 94.0 32.0 66.7 2.3 15.5 14.2 45.6 22.4 26.7 27.2 27.0 49 Korean 2nd 1.5 19,839 13,116 26.2% 17.3% 41.7 24.5 58.3 75.5 93.7 79.7 54.7 70.4 0.5 0.9 6.2 13.8 18.8 16.8 51.4 52.5 23.1 16.0 22.9 19.4 32.8 32.6 11.5 16.7 35 39 Descriptive Statistics: Males Age 25-64 in Labor Force in the Los Angeles Metropolitan Area (2018) Filipino Chinese 2nd 1.5 1st Total 2nd 1.5 1st 24,861 15,965 29,544 118,659 56,511 18,572 70,543 N 25.5% 16.4% 58.1% 100.0% 24.9% 15.7% 59.5% Percent 71.9 45.1 5.7 19.2 48.7 22.8 5.9 Speaking English only at home (%) 28.1 54.9 94.3 80.8 51.3 77.3 94.1 Speaking mother tongue at home (%) 96.1 90.7 67.8 58.3 94.9 88.2 35.0 Speaking English very well (%) 26.3 53.2 88.4 39.1 46.2 65.4 29.2 Bilingual ability (%) 2.3 1.6 3.6 8.9 0.6 2.7 14.0 Less than high school (%) 6.9 10.2 12.1 12.8 3.6 6.6 18.3 High school graduates (%) 21.2 28.7 23.5 10.8 13.1 14.7 8.8 Some college (%) 60.3 48.0 52.7 43.9 58.3 57.1 34.4 BA or associate degree (%) 9.4 11.5 8.1 23.6 24.3 18.9 24.5 More than BA (%) 12.1 15.6 5.4 16.8 15.3 20.3 16.6 Management occupation (%) 32.3 37.9 33.8 33.5 45.3 35.7 28.0 Professional occupation (%) 8.4 4.5 8.1 15.7 12.3 10.8 18.4 Self-employed (%) 37 40 48 43 38 40 46 Mean age (years) Source: U.S. Census of Population and Housing: 2018 IPUMS published by Ruggles et al. (2025.) IPUMS, incorporated public use microdata sample Generation Filipino Chinese 2nd 1.5 1st Total 2nd 1.5 1st 9,060 6,767 35,359 177,581 29,154 119,223 22,999 N 20.1% 15.0% 64.8% 100.0% 19.9% 13.0% 67.1% Percent 88.0 52.1 6.0 14.3 42.5 20.0 4.8 Speaking English only at home (%) 12.0 47.9 94.0 85.7 57.5 80.0 95.2 Speaking mother tongue at home (%) 98.2 94.3 70.5 47.4 95.5 66.9 29.4 Speaking English very well (%) 11.5 44.9 88.3 56.1 55.8 71.1 53.3 Bilingual ability (%) 1.9 1.6 2.1 17.6 1.6 5.8 24.6 Less than high school (%) 8.1 10.8 10.0 16.9 3.9 14.6 21.2 High school graduates (%) 14.6 12.3 17.7 6.7 6.3 8.9 6.4 Some college (%) 57.7 56.8 58.5 35.0 60.3 54.3 23.8 BA or associate degree (%) 17.7 18.5 11.8 23.8 27.9 16.5 24.0 More than BA (%) 27.5 15.7 9.5 11.7 15.8 18.5 9.2 Management occupation (%) 38.9 45.7 36.5 31.5 45.9 36.3 26.3 Professional occupation (%) 8.1 2.9 4.0 9.0 5.6 9.3 9.9 Self-employed (%) 37 38 48 43 36 39 46 Mean age (years) Source: U.S. Census of Population and Housing: 2018 IPUMS published by Ruggles et al. (2025.) IPUMS, incorporated public use microdata sample Descriptive Statistics: Males Age 25-64 in Labor Force in the New York-New Jersey Metropolitan Area (2018) Total 75,744 100.0% 18.6 81.4 56.4 64.2 1.6 12.8 15.9 48.3 21.4 24.5 29.6 21.2 43 Total 46,833 100.0% 21.1 78.9 59.0 61.9 3.2 10.3 6.7 46.4 33.4 21.2 33.6 19.0 44 US.-Born NH White Total 984,233 100.0% 83.4 16.6 95.9 15.4 2.3 13.5 20.9 42.8 20.5 19.5 34.0 20.4 44 US.-Born NH White Total 2,109,784 100.0% 82.9 17.1 95.0 15.5 3.2 19.2 14.3 40.6 22.7 17.8 29.9 13.1 44 46 47 APPENDIX C: Regression Tables for 2000 data in New York-New Jersey and Los Angeles Metropolitan Areas The following figures contain the regression tables for the New York-New Jersey and Los Angeles Metropolitan areas for the year 2000. The tables were created based on the Generation and Earnings Patterns Among Chinese, Filipino, and Korean Americans in New York study conducted by Sookhee Oh and Pyong G. Min and thus are an attempt at duplicating their Table 3 and then transferring it to another area. SE Chinese B Model 2 SE B SE 0.0639 0.0195 0.0002 0.003 0.0014 0.422 0.227*** 0.0835*** -0.0009** 0.0312 0.00757 0.162 7.157*** 0.0683 0.203*** 0.0153 0.0315** 0.0001 -0.0003** 0.0399*** 0.00395 0.0125*** 0.0026 0.358 10.743*** 0.0856 7.413*** 0.0235 0.332 31094 36195 0.0855 0.587 0.0605 -0.569*** -0.372*** -0.256*** -0.364*** -0.375*** -0.275*** -0.133 -0.0173 0.115 0.0872 0.0514 SE 0.124 0.12 Model 2 -0.2999** 0.0357 B 0.814 0.0725 SE 0.147 0.151 B Korean 0.164 -0.279* 0.0026 0.184 SE Model 1 0.0797 -0.452*** 0.202 -0.0323 B Model 2 0.0165 0.0993 Filipino 0.0896 0.119 Model 1 0.0485 0.0544 0.0599 -0.149*** -0.611*** First generation 0.18 0.0589 0.0252 0.0719 -0.110 1.5 generation Second generation (reference) -0.196*** 0.0600 Does not speak English very well 0.0567 0.0778 Bilingual ability Educational attainment 0.0422 -0.765*** Less than high school 0.0486 -0.637*** High school graduates 0.0458 -0.297*** Some college Bachelors (reference) 0.0394 0.368*** More than BA 0.0482*** 0.0103 Age 0.0001 -0.0005 Age Square 0.0362*** 0.0017 Weeks worked last year 0.0012 0.00959 Usual hours worked per week 0.244 10.398*** 0.05694 7.423*** 10.651*** Constant 0.001 0.0449 Adjusted R Square 107059 Number of cases Source: U.S. Census of Population and Housing: 2000 IPUMS published by Ruggles et al. (2004.) *p <0.05 **p <0.001 ***p <0.001 (two-tailed tests) IPUMS, incorporated public use microdata sample Generation B Model 1 Weighted Regression Equations Predicting Earnings For Chinese, Filipino, and Korean Males in the New York-New Jersey Metropolitan Area (2000) 48 SE Chinese B Model 2 SE B 0.0464 6.369*** 0.376 60962 0.223 10.402*** 0.00215 0.0489 0.226*** 0.0733*** 0.0106 -0.0008*** 0.0001 0.0368*** 0.00125 0.0179*** 0.00126 0.0954 6.465*** 0.263 51653 0.321 0.0469 0.164*** 0.0929*** 0.0149 -0.00102** 0.0002 0.0354*** 0.00165 0.0102*** 0.00121 0.0783 0.0517 0.0468 -0.379*** -0.296*** -0.143*** -0.455*** -0.237*** -0.212*** 0.0649 0.0444 0.0303 SE 0.0817 0.0786 Model 2 -0.231*** -0.058 B 0.0534 0.0459 0.0978 0.109 SE Korean 0.0999 0.0995 -0.119 0.0254 B Model 1 -0.0642 0.0429 SE -0.251*** -0.0845* B Model 2 0.0542 0.0553 SE Filipino 0.0497 -0.0439 0.0667 0.100* Model 1 0.0473 -0.0516 0.0482 -0.0813* -0.437*** First generation 0.0151 0.0538 0.0656 -0.00365 -0.264*** 1.5 generation Second generation (reference) 0.0492 -0.347*** Does not speak English very well 0.0459 -0.102** Bilingual ability Educational attainment 0.0395 -0.720*** Less than high school 0.0449 -0.468*** High school graduates 0.0397 -0.269 Some college Bachelors (reference) 0.0316 0.335*** More than BA 0.0829*** 0.0106 Age -0.0009*** 0.0001 Age Square 0.0383*** 0.00123 Weeks worked last year 0.0157*** 0.00108 Usual hours worked per week Percent ethnic population 0.228 10.323*** 10.724*** 0.0443 6.369*** Constant 0.000169 0.477 0.0236 Adjusted R Square 71933 Number of cases Source: U.S. Census of Population and Housing: 2000 IPUMS published by Ruggles et al. (2004.) *p <0.05 **p <0.001 ***p <0.001 (two-tailed tests) IPUMS, incorporated public use microdata sample Generation B Model 1 Weighted Regression Equations Predicting Earnings For Chinese, Filipino, and Korean Males in the Los Angeles Metropolitan Area (2000) 49 50 REFERENCES About Dissimilarity Indices. (n.d.). CensusScope. https://censusscope.org/print_about_dissimilarity.html “Asian Americans.” Wikipedia, Wikimedia Foundation, 13 Apr. 2025, en.wikipedia.org/wiki/Asian_Americans. Borjas, G. J. (1994). The Economics of Immigration. Journal of Economic Literature, 32, 1667–1717. Bureau, U. C. (n.d.). Public Use Microdata Areas (PUMAs). Census.gov. https://www.census.gov/programs-surveys/geography/guidance/geoareas/pumas.html Bureau, U. C. (2021, November 21). Appendix B: Measures of Residential Segregation. The United States Census Bureau. https://www.census.gov/topics/housing/housing-patterns/guidance/appendixb.html Chiswick, B. R., & Miller, P. W. (2005). Do Enclaves Matter in Immigrant Adjustment? City & Community, 4(1), 5–35. https://doi.org/10.1111/j.1535-6841.2005.00101.x Geocorr 2018 - MCDC. (2018). Missouri.edu. https://mcdc.missouri.edu/applications/geocorr2018.html Index of /geo. (2021). Census.gov. https://www2.census.gov/geo Lim, S., Yi, S. S., Lundy De La Cruz, N., & Trinh-Shevrin, C. (2015). Defining Ethnic Enclave and Its Associations with Self-Reported Health Outcomes Among Asian American Adults in New York City. Journal of Immigrant and Minority Health, 19(1), 138–146. https://doi.org/10.1007/s10903-015-0334-6 Moslimani, M., & Passel, J. S. (2024, September 27). What the Data Says about Immigrants in the U.S. Pew Research Center; Pew Research Center. https://www.pewresearch.org/short-reads/2024/09/27/key-findings-about-usimmigrants/ Oh, S., & Min, P. G. (2011). Generation and Earnings Patterns among Chinese, Filipino, and Korean Americans in New York. International Migration Review, 45(4), 852– 871. https://doi.org/10.1111/j.1747-7379.2011.00870.x Sanders, J. M., & Nee, V. (1996). Immigrant SelfEmployment: The Family as Social Capital and the Value of Human Capital. American Sociological Review, 61(2), 231–249. https://doi.org/10.2307/2096333 Xie, Y., & Gough, M. (2011). Ethnic Enclaves and the Earnings of Immigrants. Demography, 48(4), 1293–1315. https://doi.org/10.1007/s13524-011-0058-8 51 Zeng, Z., & Xie, Y. (2004). Asian‐Americans’ Earnings Disadvantage Reexamined: The Role of Place of Education. American Journal of Sociology, 109(5), 1075–1108. https://doi.org/10.1086/381914 Zhou, M., & Cho, M. (2010). Noneconomic effects of ethnic entrepreneurship: A focused look at the Chinese and Korean enclave economies in Los Angeles. Thunderbird International Business Review, 52(2), 83–96. https://doi.org/10.1002/tie.20316 52 Name of Candidate: [Youngjoo Hera Byol Yang] Date of Submission: [April 25, 2025]