Europe’s Tired, Poor, Huddled Masses: Self-Selection and Economic Outcomes in the Age of Mass Migration

Abstract

During the age of mass migration (1850–1913), one of the largest migration episodes in history, the United States maintained a nearly open border, allowing the study of migrant decisions unhindered by entry restrictions. We estimate the return to migration while accounting for migrant selection by comparing Norway-to-US migrants with their brothers who stayed in Norway in the late nineteenth century. We also compare fathers of migrants and nonmigrants by wealth and occupation. We find that the return to migration was relatively low (70 percent) and that migrants from urban areas were negatively selected from the sending population.

“Keep, ancient lands, your storied pomp!” cries she With silent lips. “Give me your tired, your poor, Your huddled masses yearning to breathe free, The wretched refuse of your teeming shore. Send these, the homeless, tempest-tost to me, I lift my lamp beside the golden door!”

——Emma Lazarus (1883)¹

The age of mass migration from Europe to the New World was one of the largest migration episodes in human history. Between 1850 and 1913, the United States absorbed nearly 30 million European immigrants. This paper asks two related questions about this migrant flow. What was the economic return to migrating from Europe to the United States in the late nineteenth century? And, were migrants positively or negatively selected from the European population? We study whether the United States acquired wealthier and higher-skilled European migrants who were able to finance the voyage, or whether it absorbed Europe’s “tired, poor, huddled masses” who migrated to the United States in search of opportunity.

Understanding migration in this era is of particular importance. First, the United States maintained a nearly open border in the late nineteenth century, allowing us to study migrant self-selection and the economic return to migration without interference from the legal factors that govern migrant selection today. In contrast, in the current period, the immigrant flow is a product not only of individual migration decisions, but also of complicated entry rules and restrictions, which obscure the underlying economic forces. Thus, comparing our findings with contemporary studies can illuminate the effect of modern immigration policy on migrant selection. Second, given the large magnitude of the migration flow, the skill composition of departing migrants had potentially large implications for economic growth in Europe and the United States.²

Our empirical methods are also of interest to labor economics and the economics of migration. Because migrants are not selected randomly from the sending population, it is challenging to separately identify the return to migration and the selection into migration. Measuring the return to migration with a naïve comparison of migrants and stayers can be confounded by migrant selection. For example, migrants may have earned more than men who remained in Europe in part because the brightest people—those who would have earned more regardless of location—were the most likely to move to the United States. Therefore, in the presence of positive selection, a naïve ordinary least squares (OLS) estimate of the return to migration will be biased upward, and, similarly, in the presence of negative selection it will be biased downward.

We compare the earnings of migrants to the earnings of their brothers who remained in Europe. The resulting estimate of the return to migration eliminates the across-household component of migrant selection, which can result from differing propensities to migrate from households that are financially constrained or that face poor economic opportunities in Europe. Furthermore, this within-household estimate eliminates the component of unobserved individual ability that is shared between brothers.

Beyond providing a more accurate estimate of the return to migration, this method allows us to infer the nature and extent of migrant selection across households. Specifically, a comparison of the within-household estimate and the naïve OLS estimate reveals the direction of the across-household component of migrant selection. For example, if the naïve OLS estimate of the return to migration were smaller than the within-household estimate, this contrast would suggest that the naïve OLS estimate was biased downward by negative selection of migrant households.

We focus on Norwegian migrants to the United States for two reasons. First, Norway had one of the highest out-migration rates among European sending countries, with over a quarter of its population eventually migrating to the United States. Second, Norway has completely digitized two censuses from the period (1865 and 1900), allowing us to follow large samples of men over time. We supplement the Norwegian data with a newly assembled dataset of all Norwegian-born men living in the United States in 1900 using US census records from the genealogy website Ancestry.com. We then match men by name and age from their birth families in Norway in 1865 to the labor market in either Norway or the US in 1900.

The outcome we observe for each individual is an occupation either in the United States or in the Norwegian labor market.³ We then assign individuals the mean earnings for their occupation in either Norway or the United States (in real purchasing power parity (PPP)–adjusted 1900 US dollars). For simplicity, we often refer to this occupation-based earnings measure as “earnings,” but it should be considered an occupational ranking. Although this measure captures two components of the return to migration, namely the potential for higher mean earnings in the United States for each occupation and the potential for occupational upgrading, it cannot account for the possibility of a higher return to skill within occupation in the United States.⁴ Despite this drawback, the historical data have an important advantage over their modern counterparts. Due to privacy restrictions, the individual names that we use to match migrants to their birth families are released only 70 or more years after the initial census was taken, rendering historical census data the only large dataset available for sibling comparisons of migrants and nonmigrants.

We estimate a return to migration of around 70 percent after accounting for this occupation-based selection into migration across households. Such returns are lower than contemporary estimates for the return to migration from Mexico to the United States (200–400 percent; see Hanson 2006). This method also reveals evidence of negative occupational selection for migrants leaving urban areas. The difference between the within-household estimates and the naïve OLS estimate in the urban sample suggests that negative occupational selection leads the naïve OLS estimate of the return to migration to understate the true return by 20 to 30 percent for this group. According to this method, rural migrants exhibit positive selection.⁵

Additional data we gathered on the childhood households of migrants and nonmigrants provides evidence of negative selection from both urban and rural areas. Specifically, we compare the occupation-based earnings, asset holdings, and property tax revenues of the fathers of migrants and nonmigrants in Norway. We find that the fathers of migrants are poorer than those of nonmigrants in both rural and urban areas, suggesting that men with poorer economic prospects in Norway were more likely to move to the United States in the late nineteenth century. Overall, we find consistent evidence of negative selection of migrants hailing from urban areas, and mixed evidence on selection for rural migrants. Our finding that urban migrants were negatively selected by occupation is consistent with Borjas’ (1987, 1991) work on migrant selection in the Roy model. Borjas predicts negative selection when migrants move from origins with more unequal income distribution to destinations with less unequal income distribution. Unlike today, Norway had a more unequal income distribution in the nineteenth century than did the United States (see Figure 1).

Figure 1.

Cumulative Income Distribution Functions in the United States and Norway in 1900

Notes: US and Norwegian distributions contain all men aged 38 to 50 in the respective censuses of 1900. The x-axis is scaled in 1900 US dollars. Individuals are assigned the mean earnings for their occupation and are arrayed from lowest- to highest-paid occupations. The Norwegian distribution is rescaled to have the same mean as the US distribution (the actual Norwegian and US means are US$(1900)350 and US$(1900)643, respectively).

The remainder of the paper proceeds as follows. Section I discusses the historical context and related literature on the age of mass migration and migrant selection. Section II describes the data and the procedures we used to match migrants to their birth families in Norway. Section III presents our estimates of the return to migration, while Section IV contains additional direct evidence of migration selection. Section V concludes.

I. Contemporary and Historical Literature on Migrant Selection

A. Migrant Selection in the Roy Model

The Borjas model of migrant selection is both well-known and much-disputed in the migration literature. Borjas (1987, 1991) modified the Roy model (Roy 1951) of occupational choice to generate predictions about the nature of migrant selection.⁶ In this framework, migrant selection is determined by the relative return to skill in the sending and destination economies. If the destination country exhibits higher return to skill than the source country, and therefore greater levels of income inequality, migrants will be drawn disproportionately from the top end of the source country’s skill distribution. If, instead, the destination country offers lower return to skill and is therefore more equal than the source, migrants will be drawn disproportionately from the lower tail of the source country’s skill distribution.

Work on contemporary immigrant flows has found only mixed support for the Borjas model of migrant selection. Gould and Moav (2010) show that Israeli migrants to the United States are positively selected, at least on observable skills, as would be predicted by the more compressed distribution of Israeli wages.⁷ Chiquiar and Hanson (2005), however, observe that Mexican migrants to the United States are drawn from the middle, rather than the low end, of the Mexican skill distribution, despite the fact that income inequality is higher in Mexico than in the United States.⁸ Moreover, Feliciano (2005) and Grogger and Hanson (2008) find that migrants are selected positively on educational attainment from almost every sending country in the world, even those countries with very high levels of income inequality.

Scholars have attempted to reconcile the Borjas model with the facts about positive selection in a variety of ways. A new generation of models incorporates borrowing constraints and shows that, as the cost of migration increases, the poorest residents of sending countries can no longer afford to move (Borger 2010; McKenzie and Rapoport 2010).⁹ Alternatively, Grogger and Hanson (2008) demonstrate that a classic Roy model with a linear, rather than a logarithmic, utility function generates predictions of positive selection whenever the skill-related differences in wage levels, rather than the relative return to skill in percentage terms, are high. In this framework, positive selection is a likely outcome in the contemporary world given the dramatic difference in wage levels between developed and developing countries.

B. The Age of Mass Migration

Between 1850 and 1913, more than 40 million Europeans moved to the New World, nearly two-thirds of whom settled in the United States. Initially, migrants from the British Isles and Germany constituted the majority of the migrant flow to the United States. These early migrants were joined by Scandinavians and other northern Europeans in the 1870s and by southern and eastern Europeans in the 1880s. Norway experienced one of the highest out-migration rates in the 1880s, during which time 1 of every 100 Norwegians left the country on an annual basis.¹⁰

With the shift from sail to steam technology on the Atlantic, the cost of migration fell dramatically over the nineteenth century (Keeling 1999). The declining cost of migration, coupled with rising real incomes in the newly industrializing European periphery, relaxed the financial constraints on households that had previously been too poor to pay for passage to the New World. As migration became affordable to a greater share of the European population, the migrant flow shifted from the richer countries of western Europe to the poorer countries in the south and east (Hatton and Williamson 1998, 2006; O’Rourke and Williamson 1999, 2004).

Beyond these broad “macro” patterns, we know very little about the characteristics of individuals who chose to leave Europe and move to the New World in the 19th century. For example, within a country, was the migrant flow drawn from the top or bottom end of the occupation or skill distribution? What was the economic return to this migration, after accounting for migrant self-selection? To our knowledge, Wegge (1999, 2002, 2010) are the only papers to provide individuallevel evidence on migrant selection in the nineteenth century.¹¹ Wegge documents intermediate selection for the emigration flow leaving Germany in the 1850s: members of the highest- and lowest-skill occupations were less likely to migrate than were workers in the mid-skill range, such as machinists, metal workers, and brewers. She concludes that the poorest migrants may have lacked the resources necessary to finance their trips. This result may be specific to the 1850s, when the cost of passage to the New World exceeded the total annual earnings of the average German laborer. Following the logic of Hatton, O’Rourke, and Williamson, we would expect the later Norwegian migration to be more negatively selected than the earlier German flow.

An application of the Roy model to our historical context also generates a prediction of negative selection. In the modern context, the Scandinavian countries are more equal than the United States. In the late nineteenth and early twentieth century, however, the opposite was true. Figure 1 compares the occupation-based cumulative earnings distribution functions in the United States and Norway in 1900. We array individuals from lowest- to highest-paid with earnings represented in US dollars and the Norwegian distribution rescaled to share the US mean (the earnings data are described in more detail in Section II). United States workers below the fiftieth percentile of the earnings distribution out-earned similar Norwegians, while Norwegians above the ninetieth percentile commanded higher earnings than their US counterparts. These occupation-based earnings distributions suggest that Norway offered a higher return to skill than did the United States circa 1900, which is consistent with the historical evidence on 90–50 ratios in the two countries (Soltow 1965; Goldin and Katz 1999).¹²

Not only was the occupation-based earnings distribution in the United States more compressed at a point in time, but the US economy also offered the opportunity for substantial occupational upgrading over the life cycle. Long and Ferrie (forthcoming) document that only 18 percent of men in the United States who held an unskilled, blue-collar job in 1850 remained unskilled workers by 1880. By comparison, 47 percent of men in unskilled, blue-collar occupations in Norway in 1875 remained unskilled workers in 1900.¹³ Men who started their careers in unskilled occupations were twice as likely to move up the occupational ladder in the United States than in Norway over their lifetimes; much of this mobility was accomplished by moving into owner-occupier farming.

In historical terms, the costs of migration were relatively low in the late nineteenth century. We estimate that the total cost of migration, including forgone earnings during the voyage, represents around 18 percent of the annual earnings of a Norwegian farm laborer.¹⁴ Migrant networks also helped to defray the cost of passage for new arrivals; 40 percent of Norwegian migrants during this period travelled on prepaid steamship tickets financed by friends or relatives (Hvidt 1975).

II. Data and Matching

A. Matching Norwegian-Born Men to Their Birth Families

Our goal is to create a dataset of Norwegian migrants and nonmigrants whom we can observe both when living in their childhood households and when participating in the labor market later in life. We rely on three census sources: the complete digitized Norwegian censuses of 1865 and 1900 (North Atlantic Population Project and Minnesota Population Center 2008), and a newly assembled dataset containing the full population of Norwegian-born men in the United States in 1900 (Ancestry.com). We use a standard iterative matching technique to match the population of Norwegian-born men in 1900 to their childhood households in 1865. This procedure (“Match 1”), which is described in more detail in Sections I and II of the online Appendix, generates a sample of 2,613 migrants and 17,833 nonmigrants.

Our baseline matching procedure uses only name, age, and country of birth to compare individuals across censuses. We can double our sample size by adding province of birth as a third match criterion for men who remain in Norway in 1900. This approach (“Match 2”), however, may introduce a bias by using different matching procedures for migrants and stayers. As a robustness exercise, we also match a restricted sample of men who are unique by name within a five-year age band in both censuses (two years around the reported age in each direction). This procedure (“Match 3”) limits the potential for false matches in 1900 but also reduces the sample size to 9,423.

B. Occupation and Earnings Data in Norway and the United States

We observe labor market outcomes in the year 1900, when the men in our sample were in their 30s and 40s. Neither the US nor the Norwegian census of 1900 contains individual information on wages or income in that year. Instead, we assign men the mean (PPP-adjusted) income earned by members of their occupation based either on the US 1901 Cost of Living survey (Preston and Haines 1991) or on tabulations published by Statistics Norway (Statistik Centralbureau 1905). Section III of the online Appendix describes these sources in more detail, while Section IV of the online Appendix presents our earnings estimates for farmers and fishermen, two occupations that are not included in the primary sources.

Our unavoidable reliance on mean earnings by occupation prevents us from measuring the full return to migration. Conceptually, the return to migration derives from three channels: (i) the presence of higher wages in the United States in the typical occupation; (ii) the possibility that migrants are able to switch from low-paying to high-paying occupations upon arriving in the United States; and (iii) the potential for higher within-occupation return to ability in the United States. Our estimate of the return to migration cannot capture the third aspect of the total return because we do not observe variation in earnings within an occupation.

We face a related limitation in our ability to describe the extent of migrant selection. Positive selection, for instance, could be generated either by high migration rates among men from occupations with high mean earnings or by high migration rates among men at the 80th or 90th percentiles of the wage distribution within their occupation. The reverse is true, of course, for negative selection. With our data, we can document the fact that more (fewer) common laborers moved to the United States, but we will not be able to observe whether the best (worst) among the laborers made the journey. Throughout the paper, we refer to this aspect of migration selection as “occupational selection.”

C. Comparing Matched Samples with the Full Population

Our matched samples may not be fully representative of the population of Norwegian-born men, either in 1865 or in 1900. In particular, matched samples are selected for having uncommon first and last names, which may have been associated with higher socioeconomic status. Section V of the online Appendix compares our matched sample to the population in 1865 and 1900. Men in the matched sample are demographically similar to the population in terms of age, number of siblings, and birth order. Perhaps because of the selection on having an uncommon name, however, matched men are more likely to live in urban areas, both in childhood and as adults, and to hail from households of somewhat higher socioeconomic status than the population average. By adulthood, this privilege translates into slightly higher labor market earnings.

Although our matched sample is not fully representative of the population from which it is drawn, three things are worth noting. First, the direction and extent of this bias is nearly identical both for migrants and stayers. Therefore, the distinctive features of our matched sample are not likely to affect our estimates of the economic return to migration or our conclusions about migrant selection, which depend on a comparison of matched migrants to matched stayers rather than a comparison of the matched sample to the population. Second, the main difference between our matched sample and the population is on urban status; therefore, throughout the paper we conduct our analysis separately for men hailing from urban and rural areas. Finally, to further reduce the differences between our matched sample and the general population, we consider specifications that reweight our matched sample to resemble the frequencies of the following characteristics in the general population: urban residence (full sample only), asset holdings, and above-median occupation of the household head.

We address a few additional limitations of our matching procedure here. First, our matching procedure will not capture migrants who anglicize their names upon arrival in the United States, which could be a concern if changing one’s name is correlated with economic success. Following Fryer and Levitt (2004), we use the complete 1880 US census to construct indices of a name’s distinctively Norwegian content.¹⁵ By this metric, we find that men in our matched sample are no more likely than the typical migrant to have a distinctively Norwegian name (see online Appendix Table 7, row 7).¹⁶ In addition, we find no evidence that the “Norwegianness” of a man’s name is related to our occupation-based earnings measure.¹⁷

Second, our sample of matched migrants will not include temporary movers who returned to Norway before 1900. According to the aggregate statistics, 25 percent of the Norwegian migration flow eventually returned to Norway (Semmingsen 1978).¹⁸ Return migrants may have been drawn disproportionately from the upper or lower end of the occupational distribution, either because unsuccessful migrants return home to lean on their familial support network or because the most successful migrants are able to build up a certain level of savings most quickly in order to return home. The availability of an intermediate US census in 1880 allows us to address this point. We identify over 25,000 Norwegian-born men in the relevant age range in the 1880 census. We are able to locate 14 percent of these men in either the US or the Norwegian censuses of 1900; one-third of these had returned to Norway. We compare the economic outcomes of migrants who eventually returned to Norway and those who remained in the United States in 1880, when both sets of migrants were still living in the United States. Figure 2 reveals few discernible differences in the occupational distributions of these two groups.¹⁹ Men who eventually returned to Norway are slightly overrepresented at the bottom end of the occupational distribution, but the mean occupation scores of returners and persisters are statistically indistinguishable.

Figure 2.

Comparing the Occupational Distributions of Migrants Who Stay in the United States or Return to Norway between 1880 and 1900

	N	Mean occupation score (1880)	Percentage with occupation score < 12
Unmatched	21,949	14.68	35.20
Matched	3,597	14.65	36.50
Matched (US)	2,392	14.77	35.50
Matched (Norway)	1,205	14.39	38.40*

Notes: Return migrants are defined as Norwegian-born men observed in the 1880 US census who are matched to the 1900 Norwegian census (N = 1,205). Persistent migrants are Norwegian-born men in the US census of 1880 who are matched to the 1900 US census (N = 2,392). For comparison, unmatched men are Norwegian-born men in the 1880 US census who cannot be matched to either Norway or the United States in 1900. The occupation score measure, which is taken from the 1880 IPUMS sample, is constructed by ordering occupations according to their median earnings in 1950. The mean occupation score and share of the sample with an occupation score in the bottom quartile (score < 12) are reported in the accompanying table. On both measures, the differences between matched and unmatched men are not statistically significant. We mark differences between return and persistent migrants that are statistically different at the 10 percent level with an *.

D. The Occupational Distribution of Migrants versus Stayers

This section compares the occupational distributions of Norwegian migrants to the United States and men who remained in Norway in 1900. Table 1 reports the ten most common occupations for our sample of matched migrants in the United States and matched stayers in Norway. Although 40 percent of both groups report working in farm occupations, migrants to the United States were much more likely to be owner-occupier farmers (36 percent versus 22 percent). Migrants were also more likely to report being general laborers (8 percent versus 1.4 percent). Other common occupations in both countries include carpenters, fishermen, and sailors.²⁰

Table 1.

Common Occupations Held by Norwegian-Born Men in the United States and Norway

Panel A. Top ten occupations in matched sample, Norwegian-born men living in the United States in 1900
Rank	Occupation	Frequency	Percentage	Earnings
1	Farmers and planters	1,012	35.81	691
2	Laborers (general)	256	9.05	373
3	Carpenters and joiners	174	6.15	630
4	Farm laborers	101	3.57	255
5	Painters, glaziers, and varnishers	66	2.33	624
6	Sailors	60	2.12	467
7	Saw and planing mill workers	42	1.49	572
8	Machinists	39	1.38	736
9	Railroad laborers	36	1.27	460
10	Salesmen	32	1.13	680
Total		1,818	64.33

Panel B. Top ten occupations in matched sample, Norwegian-born men living in Norway in 1900
Rank	Occupation	Frequency	Percentage	Earnings
1	General farmers	4,189	22.26	393
2	Farmer and fisherman	1,522	8.09	321
3	Merchants and dealers	722	3.84	837
4	Fisherman	709	3.77	248
5	Husbandmen or cottars	658	3.50	114
6	Farm workers	597	3.17	175
7	Carpenters	505	2.68	312
8	Shipmasters and captains	459	2.44	298
9	Cottar and fisherman	412	2.19	321
10	Seamen	351	1.87	182
Total		10,124	53.79

Notes: N = 2,826. Occupation data collected by hand from census manuscripts on Ancestry.com. Annual earnings by occupation data from the 1901 Cost of Living Survey reported in Preston and Haines (1991) in year 1900 dollars. Average income of owner-occupier farmers is estimated using data from the US census of agriculture.

Notes: N = 18,820. Historical International Standard Classification of Occupations (HISCO) occupation categories. Annual earnings by occupation data from Statistik Centralbureau (1905) and Grytten (2007). Values reported in year 1900 dollars. Average incomes of owner-occupier farmers and fishermen are estimated using data from the Norwegian census of agriculture.

Figure 3 presents the occupational distributions of these migrants and stayers, with occupations arrayed from lowest- to highest-paid according to the average US earnings in that occupation.²¹ We omit farmers, the largest occupational category, for reasons of scale, but results are qualitatively similar when farmers are included. For men born in urban areas, migrants are more likely to hold low-paying jobs such as day laborer or servant, while the men remaining in Norway exhibit an occupation distribution that is skewed toward higher-paying jobs (for example, merchants). Men born in rural areas are employed in similar jobs in both countries.

Figure 3.

Comparing the Occupational Distributions of Norwegian-Born Men in the United States and Norway In 1900

Notes: Each figure presents the relative frequency of 144 earning categories (representing 189 distinct occupations) for Norwegian-born men in the United States and in Norway. All men are assigned the mean US earnings in their occupation. Men are divided by rural or urban place of birth. Farmers, the largest occupational category, is excluded from the figure for reasons of scale. We report coefficients and standard errors from OLS regressions of ln(earnings) on a dummy for living in the United States controlling for a quadratic in age.

When we impose the same mean earnings by occupation in Norway and the United States, we find that, on average, migrants from urban areas hold occupations that pay 19 log points less than those held by the typical man from an urban background in Norway. Rural migrants also hold lower paying occupations although this gap is not as large (five log points). This negative “return to migration” is consistent with either initial negative occupational selection or occupational downgrading in the United States or both.²² We find similarly negative “returns to migration” for migrants who have been in the United States for 20 years or more, however, which suggests against the possibility of temporary occupational downgrading. We would expect that the initial disadvantages that encourage migrants to take lower-skilled jobs upon first arrival in the country would erode over time as these migrants learn English and invest in other US-specific skills.

III. Estimating the Return to Migration in the Presence of Selection

A. Naïve Estimate of the Return to Migration: Mean Earnings of Migrants versus Stayers

One naïve approach to estimating the return to migration is to compare the occupation-based earnings of all Norwegian-born men living in the United States to all men in Norway in 1900:

$$\text{ln}({\mathit{\text{Earnings}}}_i)=\alpha +{\beta }_1({\mathit{\text{Migrant}}}_i)+{\beta }_2({\mathit{\text{Age}}}_i)+{\beta }_3({\mathit{\text{Age}}}_i^2)+{\epsilon }_i,$$ (1)

where Earnings_i denotes the mean earnings of members of individual i’s occupation in 1900 in his country of residence, Migrant_i is a dummy variable equal to one if individual i lives in the United States in 1900, and Age_i and Age_i² are individual i’s age and age-squared in 1900.²³ The US census data are taken from IPUMS.²⁴ For now, we measure the return to migration with β₁, which measures the difference in the earnings of migrants and nonmigrants, adjusted for differences in the age profile.

We estimate equation (1) from a sample combining all Norwegian-born men between the ages of 38 and 50 from the 100 percent 1900 Norwegian census and the 1 percent sample of the 1900 US census.²⁵ The first column of Table 2 shows that Norwegian migrants to the United States earned 61 log points (84 percent) more than men living in Norway in 1900. Columns 2 through 4 reproduce the OLS estimates from equation (1) in our three matched samples. The implied return to migration in our matched samples ranges from 57 to 64 log points (76 to 89 percent). The population estimate represents the midpoint of this range. The fifth column reweights our matched sample to reflect the urban status, asset holdings, and occupational distribution of fathers in the full population, with little qualitative effect on the results.

Table 2.

OLS Regressions of the Return to Migration from Norway to the United States

	Dependent variable = ln(earnings)
	Population (1)	Match 1 (2)	Match 2 (3)	Match 3 (4)	Weighted (5)	Match 1 Iowa data (6)	Add penalty (7)
In US	0.609 (0.017)	0.606 (0.009)	0.644 (0.009)	0.572 (0.015)	0.641 (0.024)	0.554 (0.010)	0.466 (0.009)
N	122,620	17,501	33,641	7,596	14,647	17,352	17,501

Notes: Standard errors are reported in parentheses. All regressions control a quadratic in age. The first column contains a representative sample of the population of Norwegian-born men between the ages of 38–50 in 1900 from the 100 percent 1900 Norwegian census and 1 percent 1900 US census sample (IPUMS). Column 2 reports estimates from the first matched sample, which is based on an iterative matching strategy that searches first for an exact match and then for matches in a one- or two-year age band. Column 3 uses the second matched sample, which allows men to match in Norway by name, age, and province of birth. Column 4 reports estimates from the third matched sample, which instead requires that matched observations be unique within a five-year age band. Columns 5 through 7 return to the first matched sample. In column 5, US migrants are assigned earnings from the 1915 Iowa census (appropriately adjusted for inflation). We lose 157 observations whose occupations do not match categories in the Iowa census. In column 6, we reduce the Cost of Living earnings by 13 log points in each occupation based on the earnings penalty for Scandinavian migrants reported in Hatton and Williamson (1994). Column 7 weights the matched sample to reflect the urban status, asset holdings and occupational distribution of fathers in the full population. We lose 2,905 observations because of missing information (primarily missing data on fathers’ occupations).

In column 6, we assign US migrants the average earnings for their occupation from the 1915 Iowa census (appropriately deflated), which is more representative of the urban/rural composition of Norwegian migrants, resulting in a lower return to migration of 55 log points (73 percent). The data from the Iowa census is described in more detail in the online Appendix, Section III. The seventh column builds in the 13 log point wage penalty experienced by Scandinavian migrants in the US labor market. We calculate the wage gap between Scandinavian migrants and native-born workers in the United States circa 1900 using data from the Immigration Commission and the census.²⁶ As expected in this case, the return to migration falls to 47 log points (60 percent). Taken together, these adjustments suggest that the baseline estimates may be overstated due to the native-born and urban bias of the earnings data.

Note also that we capture the return to migration at a specific point, after three decades of United States-to-Norway migration. Ultimately, one could expect wages in the two countries to converge as out-migration reduced the labor supply in the sending country (O’Rourke and Williamson 1995, 1999, 2004). As a result, the return to migration would likely fall over time as the two countries experienced wage convergence.

B. Comparing Migrant and Nonmigrant Brothers within Households

The return to migration estimated in equation (1), β₁, would be the true return if migrants were selected randomly from the Norwegian population. If, however, migrants are (positively or negatively) self-selected, then β₁ will be biased. We next compare the occupation-based earnings of migrants and their nonmigrant brothers to eliminate selection across households. Such selection occurs if men from richer or poorer households are more likely migrate to the United States.

We consider the following equation in which the individual error term is decomposed into two components:

$$\text{ln}({\mathit{\text{Earnings}}}_{ij})={\beta }_1^{\prime }({\mathit{\text{Migrant}}}_{ij})+{\beta }_2^{\prime }({\mathit{\text{Age}}}_{ij})+{\beta }_3^{\prime }({\mathit{\text{Age}}}_{ij}^2)+{\alpha }_j+{\nu }_{ij},$$ (2)

where α_j is the component of the error that is shared between brothers in the same household j and ν_ij is the component that is idiosyncratic to individuals.

Running an OLS regression of equation (2) with household fixed effects will absorb the fixed household portion of the error term (α_j). Such within-household estimation will eliminate bias due to aspects of family background that are correlated both with the probability of migration and with labor market outcomes later in life.²⁷ In this case, the coefficient ${\beta }_1^{\prime }$ measures the return to migration, free from selection across households.

Table 3 compares the naïve OLS and within-household estimates of the return to migration. In order to contribute to the within-household estimation, a household must contain at least two members who match between 1865 and 1900. We begin in the first row of each panel by conducting OLS on the subsample of households with two or more matched members (including migrant-stayer, migrant-migrant, and stayer-stayer pairs). The unweighted return to migration is 55 log points (73 percent), which is a weighted average of 61 log points (84 percent) for men born in rural areas and only 39 log points (48 percent) for men born in urban areas.²⁸ The second row in each panel adds household fixed effects. In both weighted and unweighted specifications, we find that adding household fixed effects reduces the estimated return to migration (relative to OLS) in the rural sample and increases the estimated return to migration in the urban sample. The differences between the coefficients are statistically significant in both regions.

Table 3.

Ols and Within-Household Estimates of the Return to Migration. Households with Two or More Members in the Matched Sample

Dependent variable = ln(earnings); Coefficient on = 1 if migrant
	Full sample, 1865	Rural, 1865	Urban, 1865
Panel A. Unweighted
OLS	0.545 (0.027)	0.607 (0.034)	0.384 (0.044)
Within household	0.511 (0.035)	0.508 (0.045)	0.508 (0.057)
Chi-squared	1.49	7.47	8.31
p-value	0.2218	0.0063	0.0039
N	2,655	1,823	832
Number of migrant-stayer pairs	326	167	159
Panel B. Weighted
OLS	0.586 (0.029)	0.609 (0.033)	0.443 (0.067)
Within household	0.542 (0.039)	0.529 (0.042)	0.561 (0.049)
Chi-squared	2.13	4.60	5.65
p-value	0.1441	0.0320	0.0175
N	2,241	1,666	306
Number of migrant-stayer pairs	269	140	129

Notes: Each cell contains coefficient estimates and standard errors from regressions of ln(earnings) on a dummy variable equal to one for individuals living in the United States in 1900. Regressions also include controls for age and age squared. In each panel, the first row conducts an OLS regression for the restricted sample of households that have at least two matched members in the dataset and the second row adds household fixed effects. Panel B contains results from regressions weighted to reflect the urban status (full sample only), asset holdings, and occupational distribution of fathers in the full population. We conduct chi-squared tests of the null hypothesis that the OLS and within-household coefficients are equal.

By comparing the naïve OLS estimate of the return to migration (β₁) and the within-household OLS estimate $({\beta }_1^{\prime })$, we can infer the direction and magnitude of occupational selection across households. Specifically, if ${\beta }_1^{\prime }>{\beta }_1$, it would suggest that β₁ was biased downward by negative selection of migrant households. That is, men from the types of households that send migrants to the United States would have had low earnings even if they stayed in Norway. In contrast, if ${\beta }_1^{\prime }<{\beta }_1$, it would suggest that β₁ was biased upward by positive selection of migrant households.

By this method, we find evidence of negative occupational selection across households for migrants leaving urban areas (that is, ${\beta }_1^{\prime }>{\beta }_1$). In particular, the return to migration estimates increase by 22 to 30 percent with the inclusion of household fixed effects. In contrast, the within-household estimates are smaller than their OLS counterparts (13 to 15 percent) in the rural sample. Overall, this method provides evidence that the migrant flow from Norwegian cities and towns was drawn from households from a lower occupational stratum and that migrants from rural areas were positively selected.

C. Individual-level Instruments for Migration

Even within households, brothers can differ in unmeasured personal attributes (denoted as ν_ij in equation (2)). Appendix A provides complementary evidence on the return to migration and migrant selection using the gender composition of a man’s siblings and his place in the household birth order to instrument for migration. Both of these factors influence a man’s expectation of inheriting farmland in Norway and therefore his probability of migrating to the United States. The exclusion restrictions are that these two factors do not affect our measure of occupation-based earnings directly, and the Appendix provides supporting evidence that this was likely the case in our context.

We focus on the subsample of men born in rural areas whose childhood household held some assets in 1865. Conditional on the number of siblings in the household, the presence of an additional brother increases an individual’s probability of migrating to the United States by 1.6 percentage points (relative to the sample migration rate of 11.9 percent). Men who rank third or higher in the son order are around 5 percentage points more likely to migrate than their older brothers.²⁹ The OLS estimate of the return to migration for this selected sample is 64 log points (90 percent). Our IV estimates range from 67 to 70 log points (95 to 101 percent). The larger IV coefficient suggests that the simple earnings comparison may be biased downward, a pattern that is consistent with negative occupational selection in this rural subsample.

IV. Additional Evidence of Migrant Selection: Fathers’ Earnings and Wealth

Thus far, we have inferred the direction of migrant selection by comparing naïve estimates of the return to migration with estimates that eliminate across-household selection. In this section, we provide additional evidence on migrant selection by comparing the economic outcomes of the fathers of migrants and stayers in the 1865 Norwegian census and the 1886 Land Registers (Norwegian Historical Data Centre 2010). Focusing on fathers’ outcomes has two advantages. First, fathers’ outcomes are all measured in the same economic environment. Second, fathers’ outcomes are predetermined and not affected by the migration decision.

Table 4 compares the occupations, asset holdings, and property tax values of the heads of migrant and nonmigrant households in 1865. In particular, we regress each of these economic outcomes (Y_jr) on an indicator for the migrant status of the household (MigrantHousehold_j), controlling for the age of the head of the household (Age_j, and ${\mathit{\text{Age}}}_j^2$) and a series of province dummies (δ_r):

$$Y_{jr}={\gamma }_1{\mathit{\text{MigrantHousehold}}}_j+{\gamma }_2{\mathit{\text{Age}}}_j+{\gamma }_3{\mathit{\text{Age}}}_j^2+{\delta }_r+{\epsilon }_{jr}.$$ (3)

Households are placed in the “migrant” category if at least one son lived in the United States in 1900 and in the “nonmigrant” category otherwise. The first panel includes all households with at least one matched son. In the full sample, however, we may erroneously place true migrant households into the nonmigrant category because we cannot determine the migration status of unmatched sons. Therefore, in the second panel, we restrict our attention to households in which all sons are matched either to the United States or to Norway in 1900. In both cases, we find evidence of negative selection from urban and rural areas. That is, the heads of households that sent migrants to the United States were less likely to own assets and more likely to hold occupations with below-median average earnings.

Table 4.

Economic Outcomes of Household Heads with Migrant and Nonmigrant Sons, 1865

	All households		Households with matched sons
Dependent variables	Mean	Coefficient on migrant HH	Mean	Coefficient on migrant HH
Panel A. Urban
Occ. > median	0.593	−0.001 (0.022)	0.583	−0.030 (0.042)
Occupational income	428.27	−27.501 (10.216)	440.47	−26.555 (20.798)
Assets	0.260	−0.030 (0.018)	0.252	−0.058 (0.035)
N		4,038		1,074
Panel B. Rural
Occ. > median	0.608	0.008 (0.014)	0.577	−0.054 (0.032)
Occupational income	321.21	6.092 (3.847)	315.30	−9.077 (9.072)
Assets	0.665	−0.032 (0.012)	0.613	−0.035 (0.028)
Match tax records	0.130	−0.037 (0.009)	0.134	−0.040 (0.021)
Property tax bill N = 1,410; 300	2.759	−0.372 (0.307)	2.821	0.044 (0.887)
N		12,177		2,499

Notes: Results for Match 1. The left-hand panel includes all households while the right-hand panel considers only households in which all sons can be matched to 1900. In each panel, sample means are reported in Column 1 and the coefficient and standard error from a regression of each dependent variable on an indicator for being the head of a migrant household is reported in Column 2. Regressions also control for a quadratic in head’s age and a series of province dummies. We assign income levels to household heads using mean Norwegian earnings by occupation in 1900. Above-median occupations are those earning more than US$(1900)311 (urban) and US$(1900)393 (rural) per year. Assets is an indicator variable equal to 1 for men who own a business, own land, or are master craftsmen in an artisanal workshop. Tax records refers to the 1886 Land Registers. For fathers who match to the tax records, we report the value of the property tax bill in 1900 US dollars. The data is provided in speciedalers and marks. In 1875, one speciedaler was equal to four Norwegian kroner. The number of households that can be matched to the property tax records is reported below that variable.

As expected, the patterns are stronger when we focus on households in which the migration status of all sons is known. In the urban portion of this subsample, we find that the heads of migrant households are 5.8 percentage points less likely to own assets than the heads of nonmigrant households. In rural areas, heads of migrant households are 5.4 percentage points less likely to hold an occupation with above-median earnings. All of these contrasts are statistically significant at the 10 percent level. In addition, rural heads of migrant households are 3.5 percentage points less likely to own assets and 4.0 percentage points less likely to match to the Land Register of 1886. Failure to match to the register is a strong indication that the household did not have any taxable assets.³⁰ Conditional on matching to these tax records, we find little difference in the property tax bill between migrant and nonmigrant households. A comparison of these results to the within-household estimates in Section IIIB implies that rural migrants are selected negatively on observable dimensions of family background—specifically, land ownership and father’s occupation—but that, within these categories, they are positively selected on some unobservable characteristic captured by the household fixed effect.³¹

V. Conclusion

We know surprisingly little about how migrants during the Age of Mass Migration were selected from the European population and about the economic return from their journey. In this paper, we construct a new dataset of Norwegian-to-United States migration to estimate the return to migration in the presence of selection into migration. We compare the occupation-based earnings of men who moved to the United States and their brothers who stayed behind in Norway. This approach eliminates the component of migrant selection that takes place across households. We gather further evidence about the nature of migrant selection by comparing the economic outcomes of fathers of migrants and nonmigrants.

We estimate a return to migration from Norway to the United States of around 70 percent, which is substantially smaller than the 200–400 percent return for migration from Mexico to the United States today (Hanson 2006). The contemporary return to migration may be higher than in the past because of the sizeable cost of migration—both the bureaucratic costs of legal immigration and the cost of evading detection for the undocumented—which together reduce the supply of immigrants to the country. In the late nineteenth century, the border was open to almost all prospective migrants and, therefore, the return to migration was relatively low. We note, however, that the decision to migrate in the nineteenth century (as today) may have entailed other nonpecuniary considerations that would have increased the total return to migration (Bertocchi and Strozzi 2008). For example, at the time, Norway was under Swedish control and limited its franchise to men with wealth, while the United States offered the opportunity (for white men) to participate in the democratic system even to new migrants.

We find mixed evidence on selection for rural migrants, with some methods suggesting positive selection and others suggesting negative selection. In contrast, we consistently find that men from urban areas who faced poor economic prospects in Norway, as measured by occupation, were more likely to migrate to the United States. One result of this negative occupational selection is that the naïve return to migration underestimates the true return by 20 to 30 percent for the urban sample. The fact that migrants to the United States appear to have been drawn from the lower end of the occupational distribution is consistent with a standard Roy model of migration, as in Borjas (1987), which predicts that men at the lower end of the occupational distribution would have more to gain by moving from relatively unequal European countries to the New World. Furthermore, the fact that European migrants from urban areas, when unhindered by entry restrictions, were negatively selected from the sending population may explain why the United States eventually closed its border to the free flow of labor in 1924 (Hatton and Williamson 2006) and why some countries explicitly select for more skilled applicants in their immigration policies today.

Appendix A

IV Estimates

Comparing the within-household and naïve OLS estimates of the return to migration reveals selection in the type of households that sent migrants to the United States. Even within households, however, brothers differ in unmeasured personal attributes (denoted as ν_ij in equation (2)). In this section, we turn to a complementary instrumental variables estimation that can address both across- and within-household forms of selection. In particular, we aim to find individual characteristics that are correlated with the propensity to migrate but not otherwise associated with labor market potential.

One factor that may have influenced the decision to migrate was the expectation of inheriting farmland in Norway. Historically, property was only passed to sons. Some regions of Norway also relied on a primogeniture system of inheritance wherein the eldest brother stood to inherit the family assets and the corresponding obligation to care for his aging parents. In this social context, younger brothers, who had to “make their own way” in the world, may have been more likely to migrate to the United States.³² In accordance with these historical inheritance practices, we consider two instruments for migration status: the gender composition of a man’s siblings and his place in the household birth order. Conditional on the number of siblings, men with fewer brothers received a larger share of the total inheritance. In areas that practiced primogeniture, elder brothers were more likely than younger brothers to inherit land (again, conditional on total number of siblings).

For the instrumental variable (IV) analysis, we focus on the subsample of men who were most likely to receive an inheritance in land, namely those who were born in rural areas and whose childhood household held some assets in 1865. Furthermore, to minimize measurement error in household composition, we limit our attention to men whose mothers were young enough for a (near-)complete household structure to be observed in 1865.³³ Using this sample, we estimate the following first-stage equation:

$${\mathit{\text{Migrant}}}_{ij}=\alpha +{\gamma }_1({\mathit{\text{Brothers}}}_j)+{\mathrm{\Gamma }}_1({\mathit{\text{Siblings}}}_j)+{\mathrm{\Gamma }}_2({\mathit{\text{Rank}}}_{ij})+{\gamma }_2({\mathit{\text{Age}}}_i)+{\gamma }_3({\mathit{\text{Age}}}_i^2)+{\epsilon }_{ij},$$ (A1)

where Migrant_ij is equal to 1 for individual i from household j living in the United States in 1900. The variables of interest are the number of brothers in the household (Brothers_j) and an individual’s rank among these brothers (Rank_ij). We include both specifications that include both instruments and ones that use them separately. We also include dummy variables for the number of siblings in the household (Siblings_j) and control for a quadratic in age. We note that conditional on the number of siblings in the household, both the number of brothers and an individual’s place in the birth order of sons are random.

The second stage of the IV specification is

$${\text{ln}(\mathit{\text{Earnings}})}_{ij}=\alpha +{\gamma }_1({\mathit{\text{Migrant}}}_{ij})+{\mathrm{\Gamma }}_1({\mathit{\text{Siblings}}}_j)+{\gamma }_2({\mathit{\text{Age}}}_i)+{\gamma }_3({\mathit{\text{Age}}}_i^2)+{\mu }_{ij}.$$ (A2)

The independent variable of interest is Migrant_ij, which is instrumented with the number of brothers in the household Brothers_j and/or with an individual’s rank among these brothers Rank_ij.

The identifying assumption underlying these instruments is that household composition has no influence on labor market outcomes beyond its effect on the probability of migration. Men who have few brothers or who are eldest among their brothers, however, may not only benefit directly from receiving an inheritance but may also be the beneficiaries of complementary human capital investments in childhood.³⁴ This possibility is not borne out in our data: among men who remain in Norway, we find no relationship between the composition of a man’s childhood household and his labor market outcomes in 1900. We estimated regressions of ln(earnings) on either number of brothers or birth order for the subsample of men who lived in Norway in 1900, controlling for number of siblings, age, and province of birth. We find no relationship between number of brothers and earnings (coefficient = 0.003, standard error = 0.008) and a statistically insignificant and positive relationship between being third or higher in the birth order of sons and earnings (coefficient = 0.021, standard error = 0.023).³⁵

Our IV results are presented in Appendix Table A1. Panel A reports estimates from the first-stage equations. Conditional on the number of siblings in the household, the presence of an additional brother increases an individual’s probability of migrating to the United States by 1.6 percentage points (relative to the sample migration rate of 11.9 percent). In the second column, we find that men who rank third, fourth, or higher in the son order are 5 to 8 percentage points more likely to migrate than their older brothers. We do not observe a difference in migration rates between the first and second born sons, however, suggesting either that inheritance practices may not have followed a strict form of primogeniture in this period or that second sons regularly replaced first sons who died or were otherwise incapacitated before the family property could change hands. Interestingly, in rural households without assets, men with more brothers or who are themselves further down the birth order are somewhat less likely to migrate (not shown), which we take as suggestive evidence that the effects of household composition estimated here operate through an inheritance channel.

Panels B and C, respectively, contain OLS and IV estimates of the return to migration for the rural subsample (equation (4) above). The OLS estimate of the return to migration for this selected sample is 64 log points (90 percent). Our IV estimates in panel C range from 67 to 70 log points (95 to 101 percent). The larger IV coefficient suggests that the simple earnings comparison may be biased downward by a small amount, a pattern that is again consistent with mild negative selection of migrants from rural areas.³⁶

Table A1.

Birth Order and Number of Brothers as Instruments for Migration to the United States

	(1)	(2)	(3)
Panel A. First stage	Dependent variable = In US in 1900
Number of brothers	0.016 (0.006)		0.011 (0.006)
2nd brother		−0.000 (0.012)	—
3rd brother		0.047 (0.019)	0.037 (0.019)
4th or higher brother		0.076 (0.035)	0.058 (0.036)
Panel B. OLS	Dependent variable = ln(earnings in 1900)
In US in 1900		0.642 (0.019)
Panel C. IV	Dependent variable = ln(earnings in 1900)
In US in 1900	0.669 (0.436)	0.696 (0.381)	0.668 (0.338)
Over-ID test (p-value)			0.869
N	4031	4031	4031

Notes: Standard errors are reported in parentheses. The sample includes men in Match 1 who lived in a rural household that had some assets in 1865 and whose mother is 42 years old or younger in 1865. The regressions also include a quadratic in age and dummy variables for total number of siblings in the household (see equation (3) in the text). In column 3, we report the p-value from a Sargan (chi-squared) test of overidentification.