Myth of the Ghetto? European Immigrant Groups’ Residential Segregation and Socioeconomic Achievement: The Early Twentieth-Century United States ()
1. Introduction
The classic Chicago School sociologists (e.g., Burgess, 1925) have been criticized for promoting the idea that European immigrant groups in the early twentieth-century United States lived in ghettos that impeded the groups’ socioeconomic achievement. While this idea accords with earlier Progressive Era reformers’ graphic narratives of immigrant neighborhoods as squalid and isolated (e.g., Riis, 1890), critics of the Chicago School nevertheless declare that the idea is unfounded and constitutes a “myth of the immigrant ghetto” (Massey & Denton, 1993: p. 32; citing Philpott, 1978). Moreover, the critics assert that, although European immigrant groups at the time did tend to live in ethnic neighborhoods that were, in many cases, poor and spatially isolated from the U.S.-born population, these neighborhoods offered amenities, such as ethnicity-based voluntary associations, that helped the groups adjust to American society and subsequently advance, both economically and socially (Chudacoff & Smith, 2005). Critics of the Chicago School have thus encouraged the belief that European immigrant groups in the early twentieth-century U.S. never resided in ghettos in which residential segregation hurt the groups’ socioeconomic achievement.
The present study, however, argues that this belief has been too readily accepted by historians and social scientists and has not been subjected to the careful empirical scrutiny that it deserves. It deserves such scrutiny, this study contends, because, as will be explained below, the belief is a key part of a widely-accepted explanation of European immigrant groups’ early twentieth-century settlement experience. This explanation assumes that, while these groups did endure severe prejudice and discrimination, their socioeconomic achievement never suffered from ghettoization, one of the worst forms of disadvantage.
The study will focus on 1910, the high point of the volume of European immigration to the U.S. and a time when anti-immigrant sentiments in American society reached a fever pitch, as evidenced by, among other things, the deliberations of the U.S. Congress’ Immigration Commission (Parrillo, 2019). Established in 1907 due to popular outcry against immigration, this Commission—better known as the Dillingham Commission—arrived at the conclusion, published in its 1911 report, that Southern, Central, and Eastern (SCE) European immigrant groups, unlike their counterparts from Northern and Western (NW) Europe, were unprepared for success in their new environment, with profoundly deleterious consequences for the nation as a whole (Parrillo, 2019). The publication of this report was a turning point in U.S. immigration history. Motivated by unprecedented nativism and xenophobia in the U.S.-born population (Higham, 1994), it laid the groundwork for anti-immigrant public policies that culminated in legal restrictions on foreigners’ entry to the U.S., and these restrictions effectively ended the Great Immigration Stream from Europe by drastically reducing the flow of arrivals from SCE Europe (McLemore & Romo, 2005; Parrillo, 2019).
2. The Immigrant Ghetto: A Myth?
The term “ghetto” has historically been applied to predominantly minority neighborhoods that arise from a minority group’s involuntary or forced residential concentration and are visibly beset by social ills, such as restricted opportunities, accumulated disadvantages, and exclusion from the larger society (e.g., Small, 2008). It is not surprising, then, that the neighborhoods of the early twentieth-century European immigrant groups—groups viewed by many U.S.-born persons at the time as socially undesirable (Bogardus, 1928)—were often characterized as ghettos. Late nineteenth-century Progressive Era reformers (e.g., Riis, 1890) described these neighborhoods in vividly negative terms, calling them tenement slums that were unsanitary, hazardous, disease-ridden, overcrowded and, therefore, desperately in need of humanitarian intervention (Chudacoff & Smith, 2005).
Against this backdrop, it is understandable why the Chicago School sociologists, notably, Ernest W. Burgess (1925: pp. 55-56), used the term “ghetto” in maps that, in the 1920s, delineated European immigrant groups’ residential concentrations. On Burgess’s famous concentric zonal map of Chicago, the respective labels of two of these concentrations may have inadvertently linked “immigrant” with “ghetto” in the sociological lexicon. Both residential concentrations were immigrant neighborhoods near the “Slum” inside the “Zone in Transition,” an inner-city area of dilapidated, high-density housing and a common receiving point for immigrants due to its low rents. One neighborhood was an Eastern European Jewish neighborhood labeled the “Ghetto” apropos of the term’s original reference to tightly regulated Jewish quarters in Medieval Europe (Haynes & Hutchison, 2008). The other was an Italian immigrant neighborhood labeled “Little Sicily” after the economically depressed sub-region that sent large numbers of emigrants from Italy to the U.S. (McLemore & Romo, 2005).
But the map that, according to the Chicago School’s critics (Massey & Denton, 1993; Philpott, 1978), most strongly conflated the European immigrant groups’ neighborhoods with ghettos in the early twentieth-century U.S. was the one that Burgess prepared with 1930 Census data (Burgess & Newcomb, 1933, cited by Philpott, 1978). The map explicitly labeled as “ghettos” the respective residential concentrations of Chicago’s most prominent European immigrant groups, namely, the Czechs, Germans, Irish, Italians, Poles, Russians, and Swedes. This map, the critics further argued, also misleadingly implied that, in the so-called ghettos, the immigrants and their U.S.-born children were just as spatially isolated as were residents of the “Negro Ghetto” on the city’s South Side (Massey & Denton, 1993: p. 32; Philpott, 1978: p. 141).
The fundamental flaw of Burgess’s 1930 Chicago map, the critics assert, was that the European immigrant groups’ residential concentrations were not, as the word “ghetto” indicates, neighborhoods in which the groups were socially or spatially constrained. On the contrary, as the lead critic, Thomas Lee Philpott (1978), showed, these groups were scattered across the city, and their respective neighborhoods were remarkably heterogeneous, often containing numerous European ethnic ancestry groups.
So by loosely applying the term “ghetto” to immigrant neighborhoods that, in the critics’ view, bore no resemblance to the term’s historical association with a minority group’s restricted social and spatial mobility, the Chicago School sociologists created the “myth of the immigrant ghetto” (Massey & Denton, 1993: p. 32). However, the present study maintains that the critics’ exposé of this supposed myth with 1930 Census data overlooks a crucial detail, to wit, the social and spatial boundaries separating European immigrant groups from the U.S.-born population—boundaries that were palpably solid at the Great Immigration Stream’s height (circa 1910)—were noticeably blurring by the 1920s (Fox & Guglielmo, 2012: p. 334). For this reason, the present study suggests, the above critiques cannot dismiss the possibility that immigrant ghettos existed before that time, for example, at the peak of European groups’ immigration to the U.S.
Unfortunately, this possibility has yet to be considered in research on residential segregation. Scholars have explored the association between residential segregation and socioeconomic achievement for recent immigrant groups, finding that the association is positive (Cutler, Glaeser, & Vigdor, 2008a). However, they have not investigated this association for the European immigrant groups of the early twentieth-century. Indeed, two prominent scholars of U.S. immigration history have remarked, “To our knowledge, economic historians have not studied how residential segregation affected the labor-market outcomes of immigrants in the past” (Abramitzky & Boustan, 2017: p. 1331).
One possible reason for this oversight is that the data needed for such studies—in particular, residential segregation measures for a broad number of European immigrant groups—have not been available until recently (see Eriksson & Ward, 2018, 2019). Another possible reason is that historians and social scientists have merely assumed that, since these immigrant groups were steadily incorporated into the nation’s economic and social mainstreams in the early and mid-twentieth century, the conditions of their settlement, including their residential segregation from the U.S.-born population, must have been advantageous. The following two examples, one from history and the other from sociology, illustrate this point. These examples, the present study contends, reflect what is now, in history and social science, the most widely accepted view—hereafter called the received view—of European immigrant groups’ settlement in the early twentieth-century U.S.
First, a respected urban history textbook declares that, for the European groups of the Great Immigration Stream, “The ethnic neighborhood was one of the strongest institutions of inner-city life. The urge for familiarity and cultural identity drove many immigrants to seek out their own kind. …There were Little Italys, Bohemiantowns, Jewish sections, Greek districts, and other places identified with a particular ethnic group. Such communities softened the shock of migration and prepared immigrants for merger into American culture” (Chudacoff & Smith, 2005: p. 141). The book further opines, “If the term ghetto is defined as a place of enforced residence from which escape is at best difficult, only nonwhites and Latinos in this era had a true ghetto experience” (Chudacoff & Smith, 2005: pp. 141-142).
Second, a definitive sociological treatise on U.S. residential segregation makes a similar pronouncement about the ethnic neighborhoods of the European groups of the Great Immigration Stream, calling these neighborhoods “enclaves” and asserting, “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). The treatise additionally observers that, for the immigrant groups, “…ethnic enclaves proved to be a fleeting, transitory stage in the process of immigrant assimilation” (Massey & Denton, 1993: p. 33).
To be sure, this received view, which rests heavily on case-study anecdotes (cited in the aforementioned sources), is not unanimously accepted. In fact, it has been recently challenged by research suggesting that members of some European immigrant groups in the early twentieth-century U.S. were better off economically outside of immigrant communities. Studies of Norwegian immigrants (Eriksson, 2020) and Jewish immigrants (Abramitzky, Boustan, & Connor, 2024) show that, during this time, foreign-born men who left their respective ethnic neighborhoods had higher earnings than did their counterparts who stayed behind. Nevertheless, the extent to which these studies’ findings can be generalized across the larger society and to the full range of European immigrant groups in the early twentieth-century U.S. is unknown due to a dearth of research, and, for this reason, a gap exists in historical and social-scientific knowledge of immigrant communities. Simply put, scholars still do not know if these groups actually lived in ghettos. In particular, scholars do not know if the association between residential segregation and socioeconomic achievement is positive for these groups, as it seems to be for the immigrant groups of the late twentieth- and early twenty-first century U.S. (Cutler, Glaeser, & Vigdor, 2008a), or if it is negative, as the Chicago School sociologists and the Progressive Era reformers have implied.
3. Hypothesis
The ghetto hypothesis tested in the present study is derived from the above accounts of the Chicago School sociologists and Progressive Era reformers. The hypothesis predicts a negative association between (X) the groups’ residential segregation from the U.S.-born population and (Y) the groups’ socioeconomic achievement at a high point of U.S. nativism and xenophobia (circa 1910). That is, the hypothesis predicts that those groups that are the most residentially segregated will, on the average, have the lowest socioeconomic achievement.
4. Data, Variables, and Method
4.1. Data and Variables
The hypothesis is tested with data, shown in Table 1, for the entire U.S. The dependent variables are measures of European immigrant groups’ socioeconomic achievement, calculated by Borjas (1994: p. 558) with 1910 U.S. Census data for the European country-of-origin groups in Table 1’s first column. These national origin groups are the present study’s analytical units (N = 20). The dependent variables measure, respectively, literacy, occupational status, and earnings. Percent literate, 1910, is the percentage of immigrant (i.e., foreign-born) men who can read or write any language. Percent laborers, 1910, is the percentage of immigrant men working as laborers, an occupation at the lowest skill-level of the workforce (Borjas, 1994: p. 556). Adjusted log wage, 1910, is calculated for immigrant men by Borjas (1994: p. 558) with a regression that includes age, age-squared, region, metropolitan residence, and fixed-effects for each national origin group. Its values, expressed as logarithms, are easily interpreted; for example, Scottish immigrant men, who have the highest adjusted log wage, earn 30% more than do Portuguese immigrant men, who have the lowest adjusted log wage, i.e., (6.455 – 6.155) × 100 = 30.0.
The main explanatory variable in the regression analyses to follow is the next-door-neighbor-based segregation measure that Eriksson & Ward (2018: p. 43) computed for 18 European country-of-origin groups, hereafter, called immigrant groups, for simplicity. These groups are “Austria/Hungary,” Belgium, Denmark, England, Finland, France, Germany, Greece, Ireland, Italy, Netherlands, Norway, “Poland/Russia,” Portugal, Scotland, Spain, Sweden, and Switzerland. The European immigrant groups in Table 1 are those from Eriksson & Ward’s (2018: p. 43) list that can be matched to Borjas’ (1994: p. 558) data, with the segregation value for “Austria/Hungry” matched to Austria and Hungary separately and the segregation value for “Poland/Russia” matched to Poland and Russia separately.
Eriksson & Ward (2018, 2019) computed their segregation measure with U.S. Census data, using publicly-available decennial census enumeration schedules and applying the Logan & Parman (2017) method. This method exploits census-takers’ official practice of enumerating households linearly and sequentially “such that households listed next to each other on a census [taker’s enumeration schedule] page are reasonable proxies for next-door neighbors” (Eriksson & Ward, 2019: p. 995). Following this method, a household head’s country of birth defines the household’s nativity, and values are computed using European country of birth as the “in-group” and U.S.-born persons with two U.S.-born parents (i.e., third-plus generation U.S. persons) as the “out group” (Eriksson & Ward, 2018: p. 9). Hereafter, for simplicity, third-plus generation U.S. persons are referred to as U.S.-born persons or collectively as the U.S.-born population.
The next-door-neighbor-based segregation measure is an improvement over those residential segregation measures, such as the well-known index of dissimilarity, that are based on relatively large spatial units—for example, city wards, census tracts, or enumeration districts. These units can be “problematical” for the measurement of residential segregation because they are “not always consistently sized across time and space,” and thus, they can “hide segregation” (Eriksson & Ward, 2019: pp. 990, 993). In contrast, the next-door-neighbor-based segregation measure “is consistent across time and space” (Eriksson & Ward, 2019: p. 991).
The formula for the next-door-neighbor-based segregation measure is “one minus the ratio of immigrant households with a [U.S.-born] … neighbor over [i.e., divided by] the expected number under random assignment” (Eriksson & Ward, 2019: p. 997). The values generated by this formula are conveniently interpreted, as two examples from Table 1 illustrate. The value for Russian immigrants, the European immigrant group most residentially segregated from the U.S.-born population in 1900 (along with Polish immigrants), indicates that the actual number of Russian immigrants with a U.S.-born neighbor is only 19.6% of the expected number under random assignment, i.e., (1 – 0.804) × 100 = 19.6. Note that, at this time, the vast majority of Russian immigrants in the U.S. were Jewish (Lieberson & Waters, 1988: pp. 10-11). The value for English immigrants, the European immigrant group least residentially segregated from the U.S.-born population in 1900, indicates that the actual number of English immigrants with a U.S.-born neighbor is 67.6% of the expected number under random assignment, i.e., (1 – 0.324) × 100 = 67.6. The present study uses the next-door-neighbor-based segregation measure’s 1900 values to avoid the possible endogeneity of a European immigrant group’s 1910 residential segregation measure with its 1910 socioeconomic achievement measures.
A binary independent variable differentiating SCE European immigrant groups (identified with an asterisk in Table 1) from their NW European counterparts is included in the regression analyses to take account of well-documented inequalities between these two segments of the European immigrant population of the early twentieth-century U.S. This binary variable is coded as SCE European immigrant group = 1 and NW European immigrant group = 0. The SCE European immigrant groups were mostly from Europe’s least industrialized and least economically developed societies, and they were seen by many U.S.-born persons as poorer and more culturally distinctive than the NW European immigrant groups. The SCE European immigrant groups also tended to be more recent arrivals than the NW European immigrant groups. Most SCE European immigrants entered the U.S. after 1870, whereas most NW European immigrants entered before this time (McLemore & Romo, 2005; Parrillo, 2019). Further, SCE European immigrant groups suffered greater exposure to U.S. nativism and xenophobia than did NW European immigrant groups, largely because the former groups were, owing to their cultural distinctiveness, perceived by many U.S.-born persons as politically and socially threatening.
This binary variable’s coding, however, cannot parse-out two important sources of variation, namely, 1) variation due to an immigrant group’s national origin, and 2) variation due to an immigrant group’s average length of residence in the U.S. Hence, the variable is a relatively simple measure of an immigrant group’s time of arrival in American society.
Finally, to consider group-size differences, the regression analyses include an independent variable for the number of cases that Borjas (1994: p. 558) used to calculate European immigrant groups’ respective values of percent literate, percent laborers, and adjusted log wage for 1910. These sample Ns approximate group-size differences in the volume of 1900-1910 immigration to the U.S., which unfortunately cannot be computed for all European immigrant groups in Table 1. Data for 1900-1910 immigration volume are unavailable for Finland and Poland (which from 1899 to 1919 was included with Austria-Hungary, Germany, and Russia) and for England and Scotland separately, according to Parrillo’s (2019: pp. 543-546) collation of data from the U.S. Office of Immigration Statistics. The data that are available for 1900-1910 immigration volume correlate closely (r = 0.881) with Borjas’ (1994: p. 558) sample-size data, so it is reasonable to surmise that the latter data accurately reflect group-size differences. The 1910 sample-size values are logarithmically transformed in the present study to amend their skewed distribution.
4.2. Method
The dependent variables percent literate, 1910, percent laborers, 1910, and adjusted log wage, 1910, are, in separate multivariate equations, regressed on the next-door-neighbor segregation measure, 1900, the binary variable for SCE European immigrant group, and sample-size, 1910, logged. Preliminary estimations of the Ordinary Least Squares (OLS) equations are performed to ascertain the best-fitting functional form specifications of the associations between the dependent variables and the next-door-neighbor segregation measure. These estimations (not shown) indicate that a linear-log specification provides the best fit for the association between percent literate and the next-door-neighbor segregation measure (Equation (1)) and that a log-log specification provides the best fit for the association between percent laborers and the next-door-segregation measure (Equation (2)) and for the association between adjusted log wage and the next-door-neighbor segregation measure (Equation (3)). Lastly, the estimates’ statistical robustness is evaluated by an outlier/influential case analysis. Outlying observations, defined as those more than two standard deviations above or below any variable’s mean, are identified and excluded from the analyses. Then the equations that were estimated with the full N (Equations (1) (2) and (3) are re-estimated with Ns in which all observations are within two standard deviations of the variables’ respective means (Equations (4) (5) and (6)). Comparing the two sets of estimates will show if any outliers are unduly influencing the hypothesis testing results.
Table 1. Data and variables.
National origin (birthplace) |
Percent literate, 1910a |
Percent laborers, 1910a |
Adjusted log wage, 1910a |
Next-door-neighbor segregation, 1900b |
Sample size, 1910, loggeda |
Austria* |
77.6 |
32.4 |
6.254 |
0.753 |
7.45 |
Belgium |
90.1 |
26.4 |
6.321 |
0.589 |
4.26 |
Denmark |
99.1 |
15.4 |
6.374 |
0.488 |
5.85 |
England |
99.0 |
10.4 |
6.416 |
0.324 |
7.23 |
Finland |
92.4 |
22.8 |
6.267 |
0.774 |
5.47 |
France |
97.3 |
7.9 |
6.348 |
0.403 |
5.00 |
Germany |
96.1 |
14.9 |
6.370 |
0.549 |
8.13 |
Continued
Greece* |
80.5 |
42.3 |
6.317 |
0.498 |
5.40 |
Hungary* |
87.7 |
41.7 |
6.210 |
0.753 |
6.71 |
Ireland |
96.9 |
24.3 |
6.287 |
0.502 |
7.40 |
Italy* |
63.0 |
42.3 |
6.218 |
0.748 |
7.70 |
Netherlands |
97.7 |
20.9 |
6.280 |
0.615 |
5.14 |
Norway |
97.8 |
18.3 |
6.322 |
0.674 |
6.36 |
Poland* |
75.0 |
36.7 |
6.270 |
0.804 |
3.87 |
Portugal* |
58.2 |
31.3 |
6.155 |
0.586 |
4.58 |
Russia* |
79.2 |
21.4 |
6.367 |
0.804 |
7.72 |
Scotland |
99.7 |
7.3 |
6.455 |
0.330 |
5.95 |
Spain* |
83.3 |
22.2 |
6.227 |
0.427 |
3.58 |
Sweden |
98.0 |
16.9 |
6.349 |
0.559 |
6.98 |
Switzerland |
97.3 |
21.8 |
6.252 |
0.451 |
5.23 |
Medians |
94.3 |
22.0 |
6.302 |
0.573 |
5.90 |
Note: national origin groups are listed alphabetically; * = SCE European country. aFrom Borjas (1994: p. 558). The sample-size data for calculating adjusted log wage, 1910, are virtually identical (r = 0.997) to the data in the last column, which Borjas (1994: p. 558) used to calculate percent literate, 1910, and percent laborers, 1910. bFrom Eriksson & Ward (2018: p. 43).
5. Regression Analyses
The ghetto hypothesis is partly supported by the regressions in Table 2. A one standard deviation increase of a European immigrant group’s next-door-neighbor segregation from the U.S.-born population is associated with a 0.517 standard deviation increase of the percentage of group men working as laborers (two-tailed p < 0.05) and with a 0.388 standard deviation decrease of group men’s adjusted log wage (one-tailed p < 0.05). A one standard deviation increase of a European immigrant group’s next-door-neighbor segregation from the U.S.-born population is also, consistent with the ghetto hypothesis, associated with a 0.142 standard deviation decrease of the percentage of group men who are literate, but this association is not statistically significant. The ghetto hypothesis’s prediction, then, is partially borne out by these results, which suggest that, at a high point of U.S. nativism and xenophobia (circa 1910), those European immigrant groups that were the most residentially segregated from the U.S.-born population had, on the average, the lowest occupational status and the lowest earnings.
However, in the regressions in Table 3, the ghetto hypothesis is supported only for the literacy measure. In contrast to the previous analyses, a European immigrant group’s next-door-neighbor segregation from the U.S.-born population is not significantly associated with the percentage of group men working as laborers or with group men’s adjusted log wage. The regressions in this table are estimated with samples that exclude outliers, that is, all observations for estimating the equations are within two standard deviations of all the variables’ respective means. A one standard deviation increase of a European immigrant group’s next-door-neighbor segregation from the U.S.-born population is associated with a 0.220 standard deviation decrease of the percentage of group men who are literate (two-tailed p < 0.05). In line with the ghetto hypothesis, at a high point of U.S. nativism and xenophobia, those European immigrant groups that were the most residentially segregated from the U.S.-born population had, on the average, the lowest literacy rates (sans immigrant men from Italy and Portugal). Evidently, this significant association was suppressed in Table 2 by the outlying cases of immigrant men from Italy and from Portugal. The undue influence of these cases on the estimates is not surprising, however, given the cases’ extreme values. Immigrant men from these two countries had the lowest literacy rates of all immigrant men; moreover, immigrant men from Italy were one of the largest segments of the Great Immigration Stream, whereas immigrant men from Portugal were one of the smallest segments (see Table 1).
The ghetto hypothesis’s overall support, then, is relatively weak. However, the hypothesis cannot be unequivocally rejected. Thus, it is plausible that, at a high point of U.S. nativism and xenophobia, those European immigrant groups that were the most residentially segregated from the U.S.-born population had the lowest socioeconomic achievement, on the average.
Table 2. Regression Equations (1), (2), and (3): full sample.
Independent variables |
b |
β |
t-ratio |
Equation (1) Dependent variable: Percent literate, 1910 |
|
|
|
Next-door-neighbor segregation, 1900, logged |
−6.241 |
−0.142 |
−0.982 |
SCE European immigrant group |
−19.458** |
−0.784 |
−5.470 |
Sample size, 1910, logged |
0.656 |
0.072 |
0.563 |
Intercept |
88.532 |
|
|
Adj R-squared |
0.701 |
|
|
F-ratio |
15.860‡‡ |
|
|
N |
20 |
|
|
Equation (2) Dependent variable: Percent laborers, 1910, logged |
|
|
|
Next-door-neighbor segregation, 1900, logged |
0.939** |
0.517 |
3.226 |
SCE European immigrant group |
0.461* |
0.450 |
2.831 |
Sample size, 1910, logged |
−0.043 |
−0.114 |
−0.802 |
Intercept |
3.675 |
|
|
Adj R-squared |
0.632 |
|
|
F-ratio |
11.862‡‡ |
|
|
N |
20 |
|
|
Equation (3) Dependent variable: Adjusted log wage, 1910 |
|
|
|
Next-door-neighbor segregation, 1900, logged |
−0.102† |
−0.388 |
−1.924 |
SCE European immigrant group |
−0.057† |
−0.381 |
−1.914 |
Continued
Sample size, 1910, logged |
0.018† |
0.324 |
1.812 |
Intercept |
6.163 |
|
|
Adj R-squared |
0.413 |
|
|
F-ratio |
5.461‡ |
|
|
N |
20 |
|
|
†Slope coefficient is significant, p < 0.05 (one-tailed test), *Slope coefficient is significant, p < 0.05 (two-tailed test), **Slope coefficient is significant, p < 0.01 (two-tailed test). ‡F-ratio for R-squared is significant, p < 0.01, ‡‡F-ratio for R-squared is significant, p < 0.001.
Table 3. Regression Equations (4), (5), and (6): samples with outliers excluded.
Independent variables |
b |
β |
t-ratio |
Equation (4) Dependent variable: Percent literate, 1910a |
|
|
|
Next-door-neighbor segregation, 1900, logged |
−6.464* |
−0.220 |
−2.282 |
SCE European immigrant group |
−14.334** |
−0.814 |
−8.438 |
Sample size, 1910, logged |
0.795 |
0.126 |
1.437 |
Intercept |
87.535 |
|
|
Adj R-squared |
0.875 |
|
|
F-ratio |
40.496‡‡ |
|
|
N |
18 |
|
|
Equation (5) Dependent variable: Percent laborers, 1910, loggedb |
|
|
|
Next-door-neighbor segregation, 1900, logged |
0.321 |
0.196 |
0.979 |
SCE European immigrant group |
0.449** |
0.665 |
3.455 |
Sample size, 1910, logged |
−0.035 |
−0.147 |
−0.779 |
Intercept |
3.383 |
|
|
Adj R-squared |
0.494 |
|
|
F-ratio |
6.202‡ |
|
|
N |
17 |
|
|
Equation (6) Dependent variable: Adjusted log wage, 1910c |
|
|
|
Next-door-neighbor segregation, 1900, logged |
−0.037 |
−0.161 |
−0.587 |
SCE European immigrant group |
−0.046 |
−0.429 |
−1.660 |
Sample size, 1910, logged |
0.009 |
0.246 |
0.980 |
Intercept |
6.240 |
|
|
Adj R-squared |
0.115 |
|
|
F-ratio |
1.691 |
|
|
N |
17 |
|
|
aItaly and Portugal excluded. bEngland, France, and Scotland excluded. cEngland, Portugal, and Scotland excluded. *Slope coefficient is significant, p < 0.05 (two-tailed test), **Slope coefficient is significant, p < 0.01 (two-tailed test). ‡F-ratio for R-squared is significant, p < 0.01, ‡‡F-ratio for R-squared is significant, p < 0.001.
6. Conclusions and Implications
The ghetto hypothesis cannot be completely rejected and, hence, more research is needed to ascertain whether, or to what extent, European immigrant groups’ socioeconomic achievement was inhibited by residential segregation in the early twentieth-century U.S. Such research should, like the present study, explore the peak years of the Great Immigration Stream (1900-1910), when the volume of immigration from Europe reached its zenith and when nativism and xenophobia in American society were at historic levels, most likely because of widespread perception among U.S.-born persons that immigration was an existential menace to the nation. This period has been overlooked by scholars who, in dismissing the idea that European immigrant groups ever lived in ghettos, focus on 1930 (Massey & Denton, 1993; citing Philpott, 1978), a time when anti-immigrant sentiments and perceptions of group-threat had been largely (though not completely) dissipated by the Great Immigration Stream’s end due to national origins restrictions, and by the broader incorporation of European immigrant groups into the U.S. white racial category. In fact, there is ample evidence of substantial blurring in the 1920s of the social and spatial boundaries that had, in the 1900-1910 era, separated European immigrant groups from the rest of American society (Fox & Guglielmo, 2012).
If future research discovers, for 1900-1910 (or some earlier period), a negative association between European immigrant groups’ residential segregation and the groups’ socioeconomic achievement that is more statistically robust than the one observed in the present study, then both the historical and social-scientific literatures would need to discuss the adverse consequences of European immigrant groups’ ghettoization. To be sure, these literatures do appropriately document the formidable obstacles encountered by these groups, such as prejudice and discrimination by the U.S.-born majority as well as cultural barriers and material deprivation (e.g., Lieberson, 1980). These literatures also recognize that some European immigrants fared better economically outside of their residentially segregated communities (Abramitzky, Boustan, & Connor, 2024; Eriksson, 2020). Yet, these literatures also emphatically deny that European immigrant groups suffered from ghettoization, a serious disadvantage described at the turn of the twentieth century by Progressive Era reformers (Riis, 1890).
Further tests of the ghetto hypothesis should strive to identify the causal mechanism(s) producing the negative association between European immigrant groups’ residential segregation and the groups’ socioeconomic achievement. Such mechanisms are suggested by research indicating that residential segregation 1) spatially isolates minority groups from schools, workplaces, and transportation hubs that offer socioeconomic advancement opportunities and 2) spatially concentrates joblessness, poverty, and other disamenities into minority groups’ communities (Massey & Denton, 1993). Unfortunately, these mechanism(s) cannot be explored with the present study’s cross-sectional data and methods, which can only investigate the association between the two variables of interest.
Further testing of the ghetto hypothesis should also take account of the different spatial distributions of European immigrant groups across cities and regions in the early twentieth-century U.S. Immigrant groups from SCE Europe tended to be concentrated in major cities of the northeast and midwest, whereas immigrant groups from NW Europe were often more scattered across the nation (Abramitzky & Boustan, 2017). These various patterns could have influenced differences in the association between the immigrant groups’ residential segregation and the groups’ socioeconomic achievement at the height of the Great Immigration Stream. If those groups that were spatially clustered and relatively large—for example, prominent groups, such as the Italians and Russian Jews—were perceived by U.S.-born persons as formidable competitors in urban labor markets, then such groups might have been high-profile targets for intense prejudice and discrimination in these places (Lieberson, 1980). It follows that such groups would have been at greater risk of ghettoization than their more dispersed and relatively smaller counterparts.
The ghetto hypothesis’s plausibility implies that historians and social scientists must amend the received view by toning down the idea, emphasized by studies cited above (Chudacoff & Smith, 2005; Massey & Denton, 1993), that the European immigrant groups of the early twentieth century benefited from ethnic communities that helped them to enter the U.S. economic and social mainstreams. However, revisions along these lines would not necessarily be sweeping. They would still acknowledge that these neighborhoods may have been economically auspicious settings for some co-ethnic professionals (e.g., physicians; see Lieberson, 1980) or for those co-ethnic entrepreneurs specializing in unique goods or services demanded exclusively by co-ethnic consumers (Aldrich et al., 1985; citing Light, 1972). A revision of the historical and social-scientific literatures would also recognize that the findings of aggregated data analyses, such as those of the present study, cannot refute claims that some ethnic neighborhoods did harbor authentic enclave economies that sustained a relatively large share of the local co-ethnic immigrant workforce. An example of such a neighborhood in the early twentieth-century U.S. would be the Jewish immigrant community of the lower east side of Manhattan, which housed a bustling garment manufacturing industry (Sowell, 1981: pp. 84-85). Yet, the present study’s results indicate that such ethnic neighborhoods were the exception rather than the rule.
The plausibility of the ghetto hypothesis, furthermore, does not overturn the received view’s suggestion that ethnic neighborhoods helped European immigrant groups by bolstering primary-group relations (e.g., family and/or friendship ties) in immigrant communities. For instance, the immigrant groups’ residential concentrations may have, as some historians and social scientists opine, generated non-pecuniary benefits, such as collective support networks that cushioned the psychological and cultural dislocations of migration and settlement and thereby socially stabilized immigrant communities (Chudacoff & Smith, 2005; Massey & Denton, 1993). In addition, such concentrations may have also, by encouraging group solidarity, strengthened kinship structures and peer-groups that enhanced the primary socialization of the U.S.-born children of European immigrant groups. Evidence shows that, for relatively advantaged immigrant groups, ethnic neighborhoods do expose U.S.-born children to older co-ethnics’ desirable behaviors (e.g., a positive work ethic) and, in doing so, facilitate intergenerational human-capital transfers that raise the group’s total productivity (Borjas, 1995).
The ghetto hypothesis’s plausibility also adds a new perspective on the rapid decline of European immigrant groups’ residential segregation in the decades after legal immigration restrictions. The received view emphasizes that such declines reflect the groups’ assimilation into the larger society’s workplaces and neighborhoods, as well as the propensity of numerous group members to return to their respective homelands (Abramitzky & Boustan, 2017; Eriksson & Ward, 2019; Xu, 2020). Given ethnic neighborhoods’ adverse socioeconomic consequences for European immigrant groups in the early twentieth-century U.S., the present study implies that another relevant factor in the groups’ residential segregation declines was the motivation of many group members to leave such neighborhoods in order to avoid the costs of residential segregation.
Finally, the plausibility of the ghetto hypothesis contributes a broad historical insight to the literature on U.S. immigrant enclaves in the late twentieth and early twenty-first centuries. Despite inconsistent findings in this literature, there is compelling evidence that post-1965 immigrant groups with impressive human- and financial-capital (e.g., Asian immigrant groups in ethnoburbs (Wen, Lauderdale, & Kandula, 2009)) do economically benefit from residential concentration in ethnic neighborhoods. Perhaps this is because such groups’ resources are superior to those of the white majority and, for this reason, the groups can realize a premium, rather than a penalty, from conducting economic transactions within their own segregated markets. Hence, among such immigrant groups, genuine ethnic enclaves do appear to exist. To the extent that this inference is true, the present study’s results indicate that the positive association (Cutler, Glaeser, & Vigdor, 2008a) between recent immigrant groups’ residential segregation from the U.S.-born population and the groups’ socioeconomic achievement is a historically unique phenomenon. It follows, moreover, that if a flight from ghettos contributed to the decline of residential segregation for immigrant groups of the past, then the availability of enclaves might be contributing to the observed rise of residential segregation for at least some immigrant groups of the present (Cutler, Glaeser, & Vigdor, 2008b). If this is indeed the case, then it would be an unprecedented development in the American experience and, as such, one certainly deserving of further investigation by historians and social scientists.
Acknowledgements
I appreciate the reviewer’s helpful comments.