Residential Segregation and Housing Attainment: European Immigrant Groups in the Early Twentieth-Century United States ()
1. Introduction
Sociohistorical research on European immigrant groups in the early twentieth-century United States disagrees about the consequences of residential segregation for these groups. On the one hand, the literature suggests that the residential segregation of these groups from the U.S.-born population created “enclaves” (as opposed to “ghettos”) that helped recently-arrived members of such groups to establish themselves in their new surroundings (Massey & Denton, 1993: p. 33). On the other hand, the literature also acknowledges that these enclaves, also called ethnic neighborhoods, had potential drawbacks. Notably, these neighborhoods tended to be socially and spatially isolated from the city’s most desirable housing stock and most highly accessible locations (Burgess, 1925), and such neighborhoods’ support of ethnicity-based institutions (e.g., businesses and voluntary associations) is suspected to have offered cultural and social amenities (e.g., linguistically receptive communities) that slowed or discouraged some immigrants’ assimilation (Alba et al., 1997: p. 885). Nonetheless, a definitive study of U.S. residential segregation optimistically concluded that, “For European immigrants, [these] enclaves were places of absorption, adaptation, and adjustment to American society. They served as springboards for broader mobility in society ...” (Massey & Denton, 1993: p. 33).
This sanguine conclusion is understandable for a simple reason: by nearly all indications, European immigrant groups were steadily incorporated into the U.S. economic and social mainstreams in the early twentieth century, despite formidable obstacles, not the least of which were intense nativism and xenophobia at the time when the Great Immigration Stream from Europe peaked circa 1910 (Fox & Guglielmo, 2012; Parrillo, 2019). Given such incorporation, and the subsequent advancement of the groups’ later generations (Borjas, 1994; Abramitzky et al., 2021), it is easy to assume that the early-twentieth-century European immigrant groups benefited from the conditions of their settlement, including, perhaps, their residential segregation from the U.S.-born population.
However, the present investigation contends this assumption, while plausible, deserves more scrutiny than it has heretofore received. The assumption deserves scrutiny, the investigation suggests, because its tacit acceptance—as evidenced by, for example, the above quote from Massey & Denton (1993)—may have created the impression that European immigrant groups’ socioeconomic achievement and incorporation into American society were aided by an enclave advantage. A reconsideration of this assumption, moreover, seems timely in light of the recent emergence of two lines of sociohistorical research on early twentieth-century European immigrant groups: one on the groups’ participation in the U.S. economy (e.g., Abramitzky & Boustan, 2017) and another on their residential segregation from the U.S.-born population (e.g., Eriksson & Ward, 2019). Unfortunately, with some exceptions (discussed below), these lines of inquiry are only loosely connected to each other, much to the detriment of the social-scientific and historical literatures on European immigrant groups in the U.S. The present investigation, accordingly, seeks to further unify the two lines of scholarship by addressing the question: did residential segregation from the U.S.-born population (hereafter, “residential segregation,” for simplicity) help or hurt the socioeconomic achievement of these groups in the early twentieth-century?
The study will analyze, for 25 European immigrant groups in 1930, the relationship between residential segregation, assessed by the next-door-neighbor method (discussed later), and two measures of housing attainment: home ownership and home value. Housing attainment is the study’s focus because home purchase and asset accumulation via home-equity growth are widely viewed, in both academic and popular discourse, as vital pathways to realizing the American Dream. Hence, both variables are primary indicators of socioeconomic achievement. Of course, other measures of socioeconomic achievement, such as business ownership or entry into well-paying jobs, might also be examined. Yet, housing attainment warrants special attention, not only because it reflects the pursuit of the American Dream, but also because surprisingly little research on early twentieth-century European immigrant groups has investigated the topic, despite an abiding social-scientific interest in patterns of post-1970 immigrant groups’ home ownership (Borjas, 2002) and housing prices (Cutler et al., 2008a).
The year 1930 is analyzed in the present study because it is the earliest year for which Census data are available for the foreign-born population’s housing tenure (i.e., owner vs. renter) and home values (U.S. Bureau of the Census, 1933). Fortunately, this year provides a useful vantage point for examining European immigrant groups’ housing attainment. It comes almost a decade after the Great Immigration Stream was virtually halted by the 1921 and 1924 national-origins restrictions on foreigners’ admittance to the U.S. Perhaps not coincidentally, there are indications that, in the years immediately following these restrictions, European immigrant groups’ residential segregation declined (Xu, 2020) and, in spite of lingering nativism and xenophobia (Bogardus, 1928, 1968), their members were gradually incorporated into the U.S. economic and social mainstreams, particularly in the workforce and the emerging labor union movement (Roediger, 2005).
These indications of incorporation were evident not only for the Northern and Western (NW) European immigrant groups (e.g., the English, Irish, and Germans), but also for the Southern, Central, and Eastern (SCE) European immigrant groups (e.g., Russian Jews, Poles, and Italians). Such indications are notable. The SCE European groups were perceived as uniquely exotic or alien by many U.S.-born persons during the Great Immigration Stream’s apex; hence, these groups were often regarded by the latter as particularly threatening to American society, both culturally and politically (Parrillo, 2019). But over the 1920s, there was, as will be described below, a palpable blurring of the formerly solid or “bright” social boundaries separating SCE European immigrant groups from U.S.-born persons, many of whom descended from earlier-arrived NW European immigrant groups (Fox & Guglielmo, 2012: p. 334). It is fair to assume, then, that heading into 1930, European immigrant groups from a variety of origins were entering housing markets with both the desire and capacity to own homes and build wealth through home-value appreciation. However, it is unknown whether, or to what extent, residential segregation’s spatial boundary aided or impeded the groups’ realization of such goals at the time.
2. Literature and Hypotheses
To address the above question, the study will test three competing hypotheses about the slope of the relationship (β) between European immigrant groups’ residential segregation (X) and their housing attainment outcomes (Y). The first hypothesis, derived from the ethnic enclave literature and related literatures, is grounded in the proposition that residential segregation elevates an immigrant group’s housing attainment by yielding benefits that raise the group’s economic standing (see exemplary reviews of these literatures by Cutler et al., 2008a; Edin et al., 2003). The presumed benefits of residential segregation for ethnic minority groups discussed in these references are many and varied. They include: group solidarity that can be exploited to mobilize collective action on the group’s behalf (Fischer, 1995); co-ethnic social networks that furnish group members with exclusive access to information on employment and business opportunities in the immigrant community and beyond (Waldinger, 1996); an ethnic economy of group-owned and/or -managed enterprises sustained by the presence of exploitable co-ethnic labor and co-ethnic consumer markets (Light & Gold, 2000); and enhanced exposure of younger group members to older co-ethnics’ desirable work habits, skills, and other positive behavioral attributes, which, by facilitating intergenerational human-capital transfers, increases the group’s overall productivity (Borjas, 1995). Based on these supposed benefits, the first hypothesis, called the enclave hypothesis, is that, for European immigrant groups, residential segregation is positively associated with home ownership and home values (that is, H1: β > 0).
The second hypothesis is motivated by literature that describes and explains residential segregation’s consequences in terms of ghetto formation or, as it is sometimes called, ghettoization. In this literature, “ghetto” is used “nonpejoratively” to refer to the majority group’s strict enforcement of majority-minority group boundaries in a manner resembling controls on “Jewish quarters” in Medieval Europe (Cutler & Glaeser, 1997: p. 827). This literature suggests that residential segregation constrains an immigrant group’s housing attainment by creating liabilities that restrict the group’s economic standing. These well-known liabilities, documented most extensively in a long line of research on Black Americans, are theoretically applicable to studies of immigrant groups in the U.S. (Cutler et al., 2008a, 2008b). They include: isolation of groups from workplaces, schools, and transportation hubs that offer upward mobility opportunities; concentration of groups’ joblessness, poverty, and related disamenities into co-ethnic communities; and limitation of groups’ constructive involvement in civic affairs, cultural exchanges, and political alliances with other racial/ethnic groups (see reviews of this seminal literature by Farley & Allen, 1987; Massey & Denton, 1993). In addition, as noted above, immigrant groups’ residential segregation has been suspected of generating culturally and socially insular ethnic communities that delay or deter assimilation and thus moderate or impede such groups’ socioeconomic attainment (Chiswick & Miller, 2005). Consistent with these broadly recognized penalties, Norwegian immigrants (Eriksson, 2020) and Jewish immigrants (Abramitzky et al., 2024) who moved out of their respective ethnic neighborhoods in the early twentieth-century U.S. had higher average earnings than did their counterparts who stayed behind. However, the extent to which the suspected penalties of residential segregation might be evident for a broad range of European immigrant groups at the time is unknown. Hence, the second hypothesis, called the ghetto hypothesis, is that, for European immigrant groups, residential segregation is negatively associated with home ownership and home values (that is, H2: β < 0).
The third hypothesis follows from literature inferring that, in the early twentieth-century U.S., many of the social boundaries separating European immigrant groups from U.S.-born persons were blurring to the point of being almost inconsequential. This literature, alluded to above, implies that, by the 1920s, if not earlier, the social and spatial segregation of European immigrant groups was neither severe enough nor pervasive enough to affect group members’ socioeconomic prospects in any substantial way, either positively or negatively. The literature suggesting this argument shows that, at the turn of the twentieth century, NW European ancestry groups, including those that were mostly Roman Catholic (e.g., the Irish), as well as those that were mostly Protestant (e.g., the Dutch and Scandinavians), were incorporated into a single white racial category (Feagin & Feagin, 2003; Roediger, 2005). This literature demonstrates, too, that, by the early twentieth century, and especially after the Great Immigration Stream subsided, SCE European ancestry groups were also becoming incorporated into this category, as evidenced by, among other things, rising intermarriage rates with NW European ancestry groups and nationwide legal and conventional recognition of SCE European ancestry persons as white (Fox & Guglielmo, 2012).
Other literature, moreover, implies that spatial boundaries were blurring and becoming less salient for European ancestry groups during this time as well. The urban history literature’s critique of the Chicago School sociologists’ pioneering research on immigrant communities is a noteworthy case in point (Philpott, 1978; cited by Massey & Denton, 1993: pp. 32-33). The critique declares that these scholars, namely, Park et al. (1925) and Wirth (1928), misleadingly characterized the identifiable neighborhoods of early twentieth-century European immigrant groups by dubbing such neighborhoods as “ghettos” (Philpott, 1978: p. 141; citing Burgess & Newcomb, 1933). The characterization is misleading, according to the critique, because these so-called ghettos were typically heterogeneous neighborhoods that contained several European ancestry populations at any given time. For example, Chicago’s Polish ghetto was “ethnically mixed,” having noticeable representations of “other Slavic groups,” such as Lithuanians (Philpott, 1978: p. 140). The critique further shows that Chicago’s European immigrant groups were residentially dispersed. Only in the case of Poles did over half of any of these groups’ foreign-born members (with their U.S.-born children) live in one of the city’s alleged ghettos in 1930. Specifically, 61.0% of Polish immigrants lived in the neighborhood designated as the Polish ghetto, while other percentages ranged from 2.9% of Irish immigrants residing in the Irish ghetto to 49.7% of Italian immigrants residing in the Italian ghetto (Philpott, 1978: p. 141).
The spatial boundary-blurring thesis is additionally consistent with literature anchored in trailblazing Census-data analyses of ward- and tract-based residential segregation measures for cities (Duncan & Lieberson, 1959; Lieberson, 1963, 1980; cited by Massey & Denton, 1993: p. 33). These analyses reveal that European immigrant groups became less spatially isolated from other groups after 1910, indicating that the immigrant groups’ residential concentrations “proved to be a fleeting, transitory stage in the process of immigrant assimilation” (Massey & Denton, 1993: p. 33). More recent and innovative Census-data analyses with a next-door-neighbor-based segregation measure largely affirm these pivotal studies’ results, showing steep declines in European immigrant groups’ spatial isolation after the 1921 and 1924 immigration restrictions (Xu, 2020) and from 1900 to 1940, more generally (Eriksson & Ward, 2019).
In sum, the boundary-blurring literature and related literatures suggest that, in the early twentieth-century U.S., European immigrant groups had ample prospects for achieving the American Dream outside of their respective co-ethnic communities. In particular, these literatures imply that the groups 1) never faced long-term barriers that restricted their social or spatial mobility to demarcated areas that could be accurately labeled as ghettos and 2) had no incentive to create and/or maintain ethnic enclaves (through group solidarity, social networks, ethnic economies, etc.) for the goal of pursuing socioeconomic advancement. Thus, the third and last hypothesis, called the blurred-boundary hypothesis, is that, for European immigrant groups, residential segregation is unassociated with home ownership and home values (that is, H3: β = 0).
3. Data, Variables, and Method
3.1. Data and Variables
Data for testing the hypotheses, shown in Table 1, are for the entire U.S. The dependent variables are European immigrant groups’ respective home ownership rates (Yr) and median home values (Yv), calculated for nonfarm homes occupied by foreign-born white families in 1930. Nonfarm homes are examined to omit the small percentage of European immigrants (<20%) living on farms in the early twentieth-century U.S. (Borjas, 1994: p. 555). Hereafter, nonfarm homes are called “homes,” for simplicity. A group’s national origin is identified in the data source (U.S. Bureau of the Census, 1933) by the family head’s country-of-birth, and all European countries-of-birth available in this source are presented in Table 1. These national-origin groups—the study’s analytical units (N = 25)—are hereafter called “immigrant groups,” for simplicity. The choice of these groups as the analytical units comports with descriptive-data presentations in past research on immigrant groups’ socioeconomic attainment (Borjas, 1994: p. 558, 560) and residential segregation (Eriksson & Ward, 2018: p. 43). However, as will be discussed below, the relatively small sample derived from this choice requires parsimonious statistical model specifications and raises the possibility that the estimates might be unduly affected by cases with extreme values or outliers, that is, immigrant groups with variable values that are two standard deviations above or below the variables’ mean values (Agresti & Finlay, 2009). Such cases are identified in Table 1.
Table 1. Data and variables.
National origin (birthplace) |
Home ownership rate (percent), 1930 |
Median home value (dollars), 1930 |
Next-door-neighbor segregation, 1920 |
Housing
population-size, 1930, logged |
Austria* |
42.80 |
5928 |
67.60 |
11.92 |
Belgium |
56.37 |
4734 |
47.80 |
10.05 |
Czechoslovakia* |
60.69 |
4991 |
67.60 |
12.19 |
Denmark |
60.83 |
4772 |
40.90 |
11.10 |
England |
50.09 |
5559 |
29.90 |
12.67 |
Finland |
54.98 |
3086† |
69.30 |
10.63 |
France |
52.82 |
5421 |
38.10 |
10.79 |
Germany |
62.08 |
5460 |
43.80 |
13.32 |
Greece* |
28.58† |
5929 |
51.80 |
11.12 |
Hungary* |
47.62 |
5800 |
67.60 |
11.59 |
Ireland |
49.00 |
6203 |
43.00 |
12.71 |
Italy* |
47.89 |
5752 |
71.80 |
13.58 |
Lithuania* |
54.02 |
5820 |
74.40 |
11.33 |
Netherlands |
62.60 |
4978 |
52.30 |
10.75 |
Norway |
59.84 |
4229 |
56.30 |
11.57 |
Poland* |
50.26 |
5703 |
74.40 |
13.20 |
Portugal* |
40.54 |
3995 |
66.90 |
10.10 |
Romania* |
38.03 |
6353 |
67.60 |
11.00 |
Russia* |
34.96 |
7543‡ |
74.40 |
13.08 |
Scotland |
43.33 |
5933 |
29.70 |
11.70 |
Spain* |
29.31† |
3884 |
54.50 |
9.82 |
Sweden |
62.36 |
5025 |
47.70 |
12.27 |
Switzerland |
59.95 |
5010 |
36.10 |
10.56 |
Wales |
56.36 |
5152 |
29.90 |
10.12 |
Yugoslavia* |
56.65 |
4132 |
67.60 |
11.35 |
Mean |
50.48 |
5255.68 |
54.84 |
11.54 |
SD |
10.22 |
939.69 |
15.39 |
1.10 |
Note: National origin groups are listed alphabetically. * = SCE European country. † = value is two standard deviations or more below the mean. ‡ = value is two standard deviations or more above the mean.
An immigrant group’s home ownership rate is the percentage of all group-occupied homes that are owner-occupied, that is, the number of the group’s homes that are owner-occupied divided by the group’s total number of homes (i.e., owner-occupied plus renter-occupied), multiplied by 100. A home “is counted as owned if it is owned wholly or in part by any related member of the family” (U.S. Bureau of the Census, 1933: p. 7). An immigrant group’s median home value (in 1930 dollars) is the value reported for each group in the data source. Home values in this source are based on Census enumerators’ determinations of “the approximate market value” of owner-occupied homes (U.S. Bureau of the Census, 1933: p. 7).
These dependent variables are accurate yet imperfect indicators of housing attainment. To be sure, home ownership, in large part, reflects achievement of the American Dream. However, home ownership alone does not capture the possession of such housing-related assets as rental property, undeveloped residential land, or other types of real estate. Additionally, while median home value takes account of market value (and by implication, home equity), its use as a dependent variable may not fully capture housing attainment for an entire immigrant group because ownership is a condition for the inclusion of observations in the analysis of median home values and thus the housing attainment of renters (see Lieberson & Waters, 1988: pp. 140-143) is excluded from consideration.
The main independent variable (X1) in the multiple regression analysis is a next-door-neighbor-based measure of a European immigrant group’s residential segregation from the population of U.S.-born persons of U.S.-born parents. This measure is calculated by Eriksson & Ward (2018: pp. 15-16, 43) with Census data, using the publicly-available enumeration schedules from the decennial Censuses and applying the Logan & Parman (2017) method. The method leverages Census enumerators’ official practice of linear, sequential visitation of households, such that households listed next to each other closely approximate next-door neighbors (Eriksson & Ward, 2019: p. 995). Values of this variable are computed using country of birth (defined by the household head’s nativity) as the “in-group” and U.S.-born persons with two U.S.-born parents—that is, third-plus generation persons—as the “out-group.” Values in Table 1 are those computed by Eriksson & Ward (2018: p. 43) for 1920, multiplied by 100 to facilitate interpretation. The present study analyzes 1920 values to avoid the possible endogeneity of a group’s 1930 residential segregation and its 1930 housing attainment.
The values of the next-door-neighbor-based segregation measure have straightforward interpretations (see Eriksson & Ward, 2019: p. 997), as the following examples from Table 1 illustrate. The value of 29.9 for English immigrants indicates that the actual number of English-born householders with a third-plus generation householder next door is 70.1% of the expected number under random assignment. The value of 74.4 for Russian immigrants indicates that the actual number of Russian-born householders with a third-plus generation householder next door is 25.6% of the expected number under random assignment. Thus, as one might expect, in 1920, English immigrants are much more likely to have third-plus generation next-door neighbors than are their Russian counterparts, reflecting the latter’s markedly greater residential segregation from the U.S.-born population, specifically, the population of U.S.-born persons of U.S.-born parents.
Eriksson & Ward (2018: p. 43) calculated their next-door-neighbor-based segregation measure for 18 European country-of-origin groups, namely, “Austria/Hungary,” Belgium, Denmark, England, Finland, France, Germany, Greece, Ireland, Italy, Netherlands, Norway, “Poland/Russia,” Portugal, Scotland, Spain, Sweden, and Switzerland. Thus, seven of the 25 countries of birth for which housing attainment data are available (Table 1) are not directly incorporated into Eriksson & Ward’s (2018: p. 43) calculations. To include those seven countries—and hence, to fully utilize the housing attainment data—the present study imputes some of Eriksson & Ward’s (2018) values as follows. Their value for England is imputed to nearby Wales. Their value for “Poland/Russia” (based on persons born in “Russia, Poland, Estonia, Latvia, and Lithuania” [2018: p. 14, note 19]) is imputed to Russia, Poland, and Lithuania. Their value for “Austria/Hungary” (based on persons born in “Austria, Hungary, and Czechoslovakia” [2018: p. 14, note 19]) is imputed to Austria, Hungary, and Czechoslovakia, and to the adjoining countries of Romania and Yugoslavia. The present study’s rationale for this last imputation is that Romania and Yugoslavia had mostly Slavic ancestry populations that were, at the time, roughly similar to the populations of Austria, Hungary, and Czechoslovakia (Hoffman, 1983: p. 12), countries which together had comprised the Austro-Hungarian Empire that was dismantled after World War I.
A binary independent variable (X2) differentiates SCE European immigrant groups from NW European immigrant groups. This variable (1 = SCE European country-of-birth, 0 = NW European country-of-birth) reflects a common distinction made in the early twentieth-century U.S. between the new immigrant groups from the former region and old immigrant groups from the latter one. Most SCE European immigrants arrived in the U.S. after 1870, whereas most NW European immigrants arrived before that time (McLemore & Romo, 2005; Parrillo, 2019). This variable thus takes account of the SCE European immigrant groups’ more-recent arrival to, and shorter settlement duration in, the U.S. It also takes account of the SCE European immigrant groups’ relatively greater exposure to nativism and xenophobia in the early twentieth-century U.S., which, as discussed above, stemmed from American society’s widespread perception of these groups as uniquely alien or exotic and hence culturally and politically threatening during this time. SCE European countries-of-birth are identified in Table 1 with an asterisk.
Lastly, the housing population-size of a group (X3) is included in the regression analysis as an independent variable that controls for group-size variation. Such variation can substantially influence housing markets, because larger groups generate greater demand for housing than do smaller groups. This variable is an immigrant group’s total number of homes (owner-occupied plus renter-occupied) in 1930 (U.S. Bureau of the Census, 1933). It is logarithmically transformed in Table 1 to amend the raw values’ skewed distribution, following standard practice. The raw values (not shown) range dramatically, from 13,355 for Spain to 787,725 for Italy, with a median of 85,294 for Yugoslavia.
3.2. Method
In separate ordinary least-squares (OLS) multiple regression equations, European immigrant groups’ home ownership rates (Yr) and median home values (Yv) are regressed on the next-door-neighbor-based segregation measure (X1), national-origin region binary variable (X2), and group housing population-size, logged (X3). To check the findings’ robustness, supplemental analyses will explore whether the parsimonious equations’ estimates are unduly affected by the outlying values flagged in Table 1, namely, the home ownership rates of Greek immigrants (28.58%) and Spanish immigrants (29.31%) and the median home values of Russian immigrants ($7543), the vast majority of whom are of Jewish ancestry (Lieberson & Waters, 1988), and Finnish immigrants ($3086). The supplemental analyses will omit these values and then re-estimate the equations, ascertaining whether the values’ inclusion in the primary set of regression equations alters the hypothesis-testing results. As a final robustness check, cases with imputed values of the next-door-neighbor segregation measure—namely, Czechoslovakia, Lithuania, and Wales—will be omitted and the supplemental equations will then be re-estimated to determine whether the hypothesis-testing results are affected by the three immigrant groups whose residential segregation was not directly measured by Eriksson & Ward (2018).
4. Regression Analyses and Discussion
The enclave hypothesis (H1: β > 0) is supported by the home ownership rate regression in Panel A of Table 2. A 10% increase in a European immigrant group’s next-door-neighbor segregation from the third-plus generation population is, on the average, associated with a 3.74% increase in the immigrant group’s home ownership rate (b = 0.374). This finding accords with the proposition, derived from the enclave literature and related literatures, that an ethnic minority group’s residential segregation can yield benefits (e.g., group solidarity, social networks, and/or ethnic economies) that raise the group’s housing attainment. Moreover, the finding is consistent with the implicit assumption, found in many conventional views of the early twentieth-century U.S. (Massey & Denton, 1993), that European immigrant groups’ ethnic neighborhoods aided the groups’ socioeconomic achievement.
The ghetto hypothesis (H2: β < 0), however, is supported by the median home value regression in Panel A of Table 2. A 10% increase in a European immigrant group’s next-door-neighbor segregation from the third-plus generation population is, on the average, associated with a $330.60 decrease in the immigrant group’s median home value (b = −33.060). This finding tallies with the proposition, derived from the ghettoization literature, that an ethnic minority group’s residential segregation creates liabilities (e.g., isolation from economic opportunity, spatial concentration of social problems, and/or limitations on constructive civic involvement) that restrict the group’s housing attainment and, particularly, its wealth-building through home-equity appreciation. The finding also aligns with the notion, suggested by the Chicago School sociologists’ contemporary studies (Burgess & Newcomb, 1933) and historical economists’ recent research (Abramitzky et al., 2024; Eriksson, 2020), that in the early twentieth-century U.S., European immigrant groups’ ethnic neighborhoods were ghettos that constrained the groups’ participation in the larger society’s economic and social mainstreams.
Table 2. Primary and supplemental regression analyses.
Independent variables: |
Dependent variable: home ownership rate, 1930 |
Dependent variable: median home value, 1930 |
Slope |
t-ratio |
Slope |
t-ratio |
Panel A: primary regressions |
|
|
|
|
Next-door-neighbor segregation, 1920 |
0.374** |
2.209 |
−33.060* |
−2.043 |
SCE European country of origin |
−21.040*** |
−4.173 |
1082.782** |
2.243 |
Housing population-size, 1930, logged |
0.869 |
0.591 |
505.210*** |
3.588 |
Intercept |
30.064 |
|
718.414 |
|
R-squared |
0.494 |
|
0.451 |
|
F-ratio |
6.828† |
|
5.758† |
|
N |
25 |
|
25 |
|
Panel B: supplemental regressionsa |
|
|
|
|
Next-door-neighbor segregation, 1920 |
0.194 |
1.106 |
−22.741 |
−1.273 |
SCE European country of origin |
−13.905** |
−2.478 |
631.145 |
1.210 |
Housing population-size, 1930, logged |
−0.191 |
−0.130 |
368.769** |
2.819 |
Intercept |
49.954 |
|
1916.265 |
|
R-squared |
0.339 |
|
0.305 |
|
F-ratio |
3.247‡ |
|
2.783†† |
|
N |
23 |
|
23 |
|
Note: OLS estimation; slopes are metric (unstandardized) slope coefficients. *p < 0.05 (one-tailed test), **p < 0.05 (two-tailed test), ***p < 0.005 (two-tailed test). † F-ratio for R-squared is significant, p < 0.005. ‡ F-ratio for R-squared is significant, p < 0.05. †† F-ratio for R-squared is significant, p < 0.10. a The supplemental regression for home ownership excludes the outliers of Greece and Spain; the supplemental regression for median home value excludes the outliers of Russia and Finland.
The above results, taken together, infer that residential segregation had mixed associations with European immigrant groups’ housing attainment in the early twentieth-century U.S. To be sure, these associations alone are insufficient evidence of cause-and-effect relationships. However, they do accord with the theoretically-based predictions. On the one hand, residential segregation may have promoted the groups’ home ownership rates, helping the groups to realize the American Dream of home purchase. And on the other hand, residential segregation also may have degraded the groups’ home values, hurting the groups’ full enjoyment of the American Dream of financial asset accumulation through home ownership.
Yet, these inferences are undermined by the robustness checks, which show that the regression estimates are sensitive to the outliers flagged earlier. The enclave hypothesis’s support disappears in the supplemental regression of the home ownership rate in Panel B of Table 2. A European immigrant group’s next-door-neighbor segregation from the third-plus generation is still positively associated with the group’s home ownership rate (b = 0.194), but it is no longer statistically significant. The ghetto hypothesis’s support likewise vanishes in the supplemental regression of median home value in Panel B of Table 2. A European immigrant group’s next-door-neighbor segregation from the third-plus generation remains negatively associated with the group’s median home value (b = −22.741), but it is no longer statistically significant. Based on these results, the enclave hypothesis and the ghetto hypothesis are both rejected. These decisions are unchanged by the final robustness check’s omission of the cases with imputed values of the next-door-neighbor segregation measure (in fact, the estimates are so similar to those in Panel B of Table 2 that they are not presented here in order to conserve space and are available from the author on request).
The blurred-boundary hypothesis (H3: β = 0) thus emerges from the regression analyses as the most viable hypothesis, according to the logic of falsification. Without the undue influence of outlying values, European immigrant groups’ residential segregation is unassociated with the groups’ home ownership rates and median home values. This finding agrees with literature indicating that, by the early twentieth century, social boundaries (Fox & Guglielmo, 2012) and spatial boundaries (Eriksson & Ward, 2019; Xu, 2020) separating European immigrant groups from U.S.-born persons had blurred, perhaps to the point of being virtually inconsequential for the groups’ housing attainment. The finding is, furthermore, in line with the literature’s arguments that by the 1920s, European immigrant groups were pursuing the American Dream outside of their respective co-ethnic communities (Abramitzky et al., 2024; Eriksson, 2020) and 1) had no incentive to create and sustain within their communities ethnic economies that could be called “enclaves” and 2) did not face unyielding residential barriers that made their communities into places that could be called “ghettos.”
5. Summary and Implications
Did residential segregation help or hurt European immigrant groups’ socioeconomic achievement in the early twentieth-century U.S.? The question arises because of the disagreement over whether or not (in spite of residential segregation’s well-established drawbacks) these groups’ social and spatial distance from the U.S.-born population produced, at the time, ethnic neighborhoods that assisted their residents’ steady incorporation into American society. The present study addresses this question by proposing competing hypotheses, drawn from scholarship on enclaves, ghettos, and ethnic boundaries, about the relationship between European immigrant groups’ next-door-neighbor segregation from third-plus generation persons and the groups’ housing attainment, measured by their respective home ownership rates and median home values. These housing attainment outcomes are key indicators of socioeconomic achievement because they quantify group members’ home purchase and asset accumulation through home-equity appreciation, both of which are widely regarded as the American Dream’s hallmarks. Tests of these hypotheses in regression analyses of 1920-1930 Census data have several implications for theory and research on European immigrant groups’ socioeconomic achievement in the early twentieth-century U.S.
First, the enclave hypothesis’s rejection implies that social scientists and historians should reconsider the assumption that, in the early twentieth century, European immigrant groups’ residential segregation created ethnic neighborhoods that aided the groups’ socioeconomic achievement. The tacit acceptance of this assumption by some analyses of European immigrant groups (e.g., Massey & Denton, 1993: p. 33) is unsurprising for two reasons: 1) there is little doubt that these groups readily entered the U.S. economic and social mainstreams after the 1920s’ immigration restrictions, despite lingering nativism and xenophobia; and 2) enclave theory persuasively argues that the groups’ residentially-segregated communities generated benefits for their members, including group solidarity, social networks, and ethnic economies. Yet, the view that these or any other presumed advantages of residential segregation furthered European immigrant groups’ housing attainment, a prime reflection of the groups’ overall socioeconomic achievement, in the early twentieth-century U.S. has no support in the present study.
It follows that social scientists and historians should develop a theory that more accurately depicts the relationship between European immigrant groups’ residential segregation and the groups’ socioeconomic achievement in the early twentieth-century U.S. In formulating a new theory, scholars might consider that, although ethnic neighborhoods failed to advance European immigrant groups’ overall socioeconomic achievement during this time, such neighborhoods may still have benefited specific group members or group activities (Boyd, 2025). For example, ethnic entrepreneurs in certain fields, such as the retail trade, may have profited from exploiting protected markets generated by the social and spatial distance of co-ethnic consumers from the U.S.-born population (Aldrich et al., 1985). This possibility cannot be ruled out by the present study’s relatively broad-brush approach to examining residential segregation and socioeconomic achievement. Scholars might also consider that, while ethnic neighborhoods did not measurably improve foreign-born group members’ socioeconomic achievement, such neighborhoods may still have boosted the prospects of U.S.-born group members, i.e., the second generation. For instance, the latter may have benefited from intergenerational human-capital transfers that occur when co-ethnics live together in close proximity. Such “human-capital externalities,” observed for children raised in ethnic neighborhoods (Borjas, 1995), could not be ascertained in the present study, since only the foreign-born were analyzed. With these considerations in mind, revised explanations of European immigrant groups’ socioeconomic achievement in the early twentieth-century U.S. must recognize that any general or specific advantages created by the groups’ residential segregation were far more limited than past interpretations have suggested, if indeed they existed at all (Boyd, 2025).
A second implication is that the ghetto hypothesis’s rejection both affirms and expands critiques of the Chicago School sociologists’ depictions of European immigrant groups’ ethnic neighborhoods. Descriptive statistics show that the sociologists’ portrayal of these neighborhoods as ghettos is empirically shaky. As noted, Census data (Lieberson, 1980; Philpott, 1978) reveal that most European immigrant groups’ populations were fairly dispersed in the early twentieth-century U.S. and became even more so in the decade after the national-origins restrictions on entry. But critiques relying on these descriptive statistics (e.g., Massey & Denton, 1993) are incomplete, for they do not investigate the possibility—explored in the present study—that even moderate spatial distance from the U.S.-born population may have hampered the groups’ socioeconomic achievement, perhaps by slowing or discouraging their members’ assimilation into American society. Hence, dismissal of the ghetto hypothesis in the present study extends the critiques by confirming that early twentieth-century European immigrant groups were not ghettoized, either in terms of facing rigid boundaries on their residential choices or in terms of suffering a socioeconomic penalty because of their spatial distance from U.S.-born persons.
However, the ghetto hypothesis’s rejection does not imply that European immigrant groups in the early twentieth-century were unrestricted by housing market discrimination or by U.S.-born persons’ prejudicial desires to avoid them in residential space. The extent to which these groups’ spatial distance from the U.S.-born population was voluntary or involuntary during this time is an open question. Some evidence suggests that the distance largely reflects European immigrant groups’ preferences to live in close proximity to co-ethnics (Boyd, 2023). Yet, other evidence suggests that the distance at least partly reflects U.S.-born persons’ inclinations to steer clear of the groups, particularly those of SCE European origins (e.g., Russian Jews), whose members were regarded by many U.S.-born persons as unsuitable for intimate social relations, such as friendship and marriage, even into the middle twentieth century (Boyd, 2023; citing Bogardus, 1928, 1968).
Lastly, the blurred-boundary hypothesis’s confirmation implies that, anti-minority prejudice notwithstanding, European immigrant groups’ residential segregation was neither severe enough nor pervasive enough to help or hurt the groups’ socioeconomic achievement in the early twentieth-century U.S. This confirmation accords with indications (e.g., rising intermarriage) that, by this time, the various European ancestry groups of American society were starting to amalgamate into a single, legally- and conventionally-recognized white racial category, and thus, social boundaries that had separated the groups in prior decades had noticeably diminished in significance (Fox & Guglielmo, 2012). The confirmation aligns, too, with the waning tendency, noted above, of European immigrant groups to live in spaces that could be called ethnic neighborhoods (Eriksson & Ward, 2019; Xu, 2020). The present study infers, then, that social or spatial barriers that had constrained European immigrant groups at the Great Immigration Stream’s height were, by 1930, lowered to the point of being relatively inconsequential for the groups’ socioeconomic achievement, as measured by the housing attainment variables. In short, because of ethnic boundary blurring, these groups were not compelled by external obstacles to reside in actual ghettos, and owing to opportunities to participate in the nation’s economic and social mainstreams, they had no incentive to build and operate residentially-based enclaves to pursue the American Dream (Abramitzky et al., 2024; Eriksson, 2020).
In closing, this study finds no grounds for assuming that European immigrant groups in the early twentieth-century U.S. benefited or suffered in any substantial way, either socially or economically, from their residential segregation. This conclusion is, of course, tentative because of the study’s limitations. The statistical analyses were fairly sweeping, for they were based on aggregated data. Moreover, while robustness checks were performed for outlier influence, these checks were somewhat restricted. They did not consider whether the estimates are sensitive to the choice of OLS estimation (versus, say, Weighted Least Squares estimation, which is sometimes recommended for the analysis of aggregated data with widely varying values across cases, such as those for the housing population). The robustness checks also did explore sources of undue influence that might be detected with more sophisticated diagnostic techniques, such as Cook’s distance. In addition, the enclave hypothesis’s test was relatively narrow, for it did not consider outcomes, such as business ownership or entry into high-wage jobs, that are often examined in enclave research. Nevertheless, the study provides a starting point for future inquiries into the question of whether or not European immigrant groups gained any social or economic advantages from living in ethnic neighborhoods in the early twentieth-century U.S. Given the importance of this question to social scientists and historians, such inquiries would seem to be indispensable to furthering the emerging lines of scholarship (e.g., Abramitzky & Boustan, 2017; Eriksson & Ward, 2019) that promise to advance knowledge about these groups’ incorporation into American society.
Acknowledgements
I gratefully acknowledge the reviewer’s constructive comments on the manuscript.