Self-Selection and 19th Century European Emigration: The Case of Border-Crossing Hessians, 1832-1857

Simone A. Wegge
Northwestern University


Note: Due to space constraints, the text has been shortened, and all tables and most footnotes have been removed. Complete papers are available from the author.

Economic studies on migration examine such questions as who migrates, what are the returns to migration, and how does migration affect the source and host populations. This paper will tackle the first question in regards to 19th century international migration and will specifically address how migrants are self-selected out of the home or source population. Using a new micro dataset that concerns 50,000 emigrants from the German region of Hesse, I will address how these emigrants represented a nonrandom sample of the general population.

Studies of modern day migration movements focus to a large part on addressing how the earnings of immigrants compare to those of native Americans or on the effects of immigration on the United States. Work by George Borjas (AER, 1987; J. of Labor Economics, 1985), Barry Chiswick (JPE, 1978), and Michael Greenwood and J. McDowell (Review of Economics and Statistics, 1991; J. of Economic Literature, 1986) is typical in this regard. In contrast, I compare a specific emigrant population to its source population, something not addressed in detail in previous work1. This orientation allows me to study more precisely how an emigrant population is self-selected.

The examination of 19th century data, in addition, affords the great advantage over studies of contemporary immigration movements that the U.S. did not regulate the influx of migrants in the last century. Since 1926 the United States has screened potential migrants with numerical quotas or proof of family connections in the U.S. If one wants to properly characterize self-selection in a migrant population, one must use 19th century data to avoid these pre-filtering mechanisms.

This study contributes to the existing literature by examining data that describe better the heterogeneity of the emigrant and source populations. Previous studies from the labor economics literature use primarily U.S. census data and thus lack detailed information regarding the migrants' own economic status at the time of migration. My data address this, and because I analyze census data broken down by village on the source population along with the emigrant dataset which is by village as well, I capture better the heterogeneity of both the emigrant and source populations. The data are from local government sources in the native country, and are recorded in the vernacular of the emigrants. They may thus give finer descriptions of such variables as occupation than the American shipping records used for 19th-century migration research do.

Other researchers have used large datasets to examine self-selection amongst 19th century European emigrants, including Cormac O'Grada's study of Irish emigrants, Kristian Hvidt's study of Danish emigrants, and Robert Swierenga's study of Dutch emigrants. None of these studies, either because the researcher focused on other questions or, more likely, because the data did not consistently record village of origin, used detailed datasets on the respective emigrants' villages of origin, as I will.

In the first section of this paper I give a short summary of mid-19th century Hessian economic conditions. I then describe my data and compare the emigrant distribution to the source population over various dimensions. In the last section I draw my conclusions and describe how I extend the analysis in the complete paper.

In 1842, the number of immigrants to the United States passed the 100,000 mark. During this period, immigrants from Great Britain, Germany, and Ireland comprised around 85% of the total number of immigrants to the United States. The Irish made up the largest group in the latter half of the 1840's, while the Germans replaced them as the largest group in the 1850's.

My dataset concerns people who emigrated between 1832 and 1857 from the principality of Kurhesse, which consisted of the northern and eastern regions of the present state of Hesse. It became a part of Prussia in 1866. Relative to other German principalities Kurhesse was a very rural and economically backward country. After Napoleon's defeat, for instance, Kurhesse reinstated artisan laws that basically kept intact the centuries-old guild system. These laws did not change in any substantial manner until 1869 under Prussian rule.

In large part the population of Kurhesse lived in small towns and villages and secured its existence directly and indirectly through agriculture. Approximately 74% of the total population was considered as "rural": they lived in villages and earned their living directly in or indirectly through the agricultural sector.2 Partible inheritance and primogeniture co-existed as methods of inheritance traditions in Kurhesse, although primogeniture was the more common one.

I have acquired a dataset from the Hessian State Archive in Marburg.3 The dataset that has so far been assembled concerns approximately 50,000 emigrants. Prior to 1852, Kurhessian authorities kept records only on those who had formally applied for and been granted permission to leave. In 1852 they started keeping track of "illegal" emigrants as well.

For each emigrant leaving, the following information is given: (i) Name (First and last); (ii) Village and county of origin; (iii) Country (and sometimes city) of destination; (iv) Number of family members accompanying, or whether traveling with family at all; and (v) Date of application. The dataset, however, is richest with regards to the men that are listed, as only men served in the military and most of the taxpayers were men. For family men or single persons traveling alone, this extra information includes: (vi) Age; (vii) Money being exported; and (viii) Occupation. While the first five variables are very complete, the last three are not, especially for the last five years of the dataset. For the 50,000 different emigrants, I have age information for approximately 25% of them.

Data on 19th century emigrants was collected at various points of their journey. The data from the Marburg archive is particularly valuable in that it lists for emigrants what village and county he or she came from, something that cannot be found to this degree of precision in any of the ship lists in the United States. It affords a comparison of the home population through census statistics by village or county with the emigrant dataset.

In Marburg, Germany, I prepared similarly detailed census datasets that describe the distribution of artisans for Kurhesse and village economic and demographic characteristics to match the geographical origins of the emigrants. This census and general survey data, gathered for each of the 1,200 Kurhessian villages, serves to describe the heterogeneity of the source population.

In the tradition of A.D. Roy (Oxford Economic Papers, 1951), who explained how nonrandom selection processes sort people into occupations, recent migration studies have used this idea to emphasize that migrants are nonrandomly selected from the native population. These studies, however, have not shown in detail how a multivariate density of an emigrant population might look compared to the source population. I describe below how a distribution of emigrants, based on occupation or age differed from the general population. I examine each of these dimensions below.

To compare the occupational distribution of the emigrants with that of the source population I have relied on a detailed and wide-ranging classification system used in the various German censuses from 1852.4 The breakdown for the emigrant data and both census datasets is quite fine. I exclude farmers, because the census datasets do not count them.5 Male and female farm hands, however, are included in both populations. I also focus here on the years 1852 to 1857, as these years include information about the illegal emigrants as well.

Hypothesis tests for comparison of proportions show that the proportion of day workers and farm workers is remarkably different in the emigrant population from that of the source population. The proportion in the home population is actually 30% higher. In contrast, many artisan groups emigrated in higher proportions than their respective shares of the general population. Textile workers (primarily weavers), bakers, shoemakers, butchers, construction workers and shopkeepers (including traders) all emigrated in greater proportions than their respective shares in the census data.

In looking for reasons as to why various emigrants departed, we must recognize various push factors. Some were affected by the pervasive structural changes in the West European economy, a chief one being the industrialization that many European regions were undergoing. New and efficient production processes in the textile industry, for instance, displaced many linen weavers in Kurhesse in the first half of the 19th century.

In the villages, exclusive of the communities classified as cities, guild regulations from 1816 tolerated the practice of only certain artisanal occupations. These occupations included blacksmiths, nailsmiths, wheel-makers, construction workers, roofers, potters, shoemakers and weavers. Such regulations protected the city artisans, but left few opportunities for the villager with little or no land except to work in one of these trades or as a laborer. As stated above, emigration rates were very high for some of these "village" occupations.

Assessing the quality of emigrants and how this may change over different cohorts is another important component towards answering who the emigrants were and why they emigrated. A hypothesis from the brain-drain literature states that the "best and the brightest" of the home population have high emigration rates. I test for this by examining the degree of age-heaping in the emigrant data. This is the circumstance that poor and illiterate people tend to round off their ages to the nearest zero or five (Joel Mokyr, Why Ireland Starved, 1983, pg. 244-46).

The focus in the archival records is on the family men and on the men traveling alone, which means that the ages of the wives and children were often ignored. Because the information happens to be more complete for 1832-51, I focus on these years. The number of people at the age of 21 is only 30% of the 20 year olds, and the 18 and 19 year old categories are very large as well. I examined the age breakdown across sexes and found that the clustering around 18-20 does not occur with the female emigrants. Records in the dataset until 1851 concern only the `legal' emigrants, and hence I attribute this bubble in the age distribution to the matter that the military recruited heavily amongst the men in their early twenties and that men in this age group emigrated under illegal circumstances. Results from a new study on Hessian emigrants 1855-1866 based on ship records from Hamburg show a much smoother age distribution.6 Most likely, many men of eligible military service age left their homeland without going through the proper channels.

Hence I restrict the calculations of age-heaping to those between the ages of 24 and 39. I use the following measure of age-heaping, [[Sigma]] [(ni - n*i) 2/n*i], whereby ni and n*i are the reported and smoothed distributions respectively. Nevertheless, in using the Graybill's weighted moving average of Sprague coefficients (Shyrock and Siegel, 1971) to first smooth the age distribution, subtotals from younger ages still affect the smoothed distribution.

In the full paper I will explore other methods of smoothing to decrease this affect. Preliminary results show that the null-hypothesis of no age-heaping can be rejected at the 99% level. If I restrict the age group to the ages of 26 to 39, I can only reject this at the 97.5% level. Evidence of the presence of age-heaping is thus not quite as convincing as that found on Ireland (Mokyr, 1983). Further, if positive selection occurred across the occupational dimension, we can infer that these emigrants may have been more highly educated or skilled, and thus more literate than their Irish counterparts.7

The evidence shows that day workers, those of the lowest socio-economic classes, did not emigrate in any degree to their proportion in Kurhessian society. This result differs with Walter Kamphoefner's finding that the emigrant population from various Westphalian regions was lower class in nature. He mentions, however, that other scholars have found the German emigrant population to originate primarily from the middle-classes. ("At the Crossroads of Economic Development...", in Glazier & DeRosa, Migration Across Time and Nations, 1986).

Information networks may have been poorer for many impoverished peasants, but the cost of transportation relative to their earnings may have created the greatest barrier. Permanent settlement in other German regions was not a viable alternative for everyone, as other regions had preferences for certain types of immigrants. Between 1832 and 1857 the ship fare fluctuated anywhere between 20 and 50 Thaler, with the low thirties as the mode, whereby the cash portion of a day worker's annual wage ranged anywhere from 5 and 30 Thaler. In comparison, in the pre-famine years in Ireland, the annual wage for a laborer ranged from [[sterling]]10 to [[sterling]]15, while the cost of passage varied between [[sterling]]4 and [[sterling]]11. (O'Grada, C. "Across the Briny Ocean", in Glazier & DeRosa, 1986) This is particularly noteworthy, as the occupational distribution of the Irish emigrants has been found to be heavily weighted towards the day laborers. Age heaping analyses of the Irish emigrant and total Irish population have also shown that the emigrants tended to be more illiterate than their fellow countrymen in general. (Mokyr, 1983) It may be that educational practices differed greatly between Germany and Ireland, but it may also be that emigration to America was more accessible to the Irish laborer.

Annual emigration rates for the individual villages range from zero to 20%. In the complete paper I use regression analysis to examine the cross-sectional variation of emigration rates across village characteristics such as inheritance tradition practiced, day worker wages, the artisanal distribution, number of factories, and the distribution of land amongst the village populace. In this analysis I compare villages with high rates to those with low emigration rates.

Endnotes

1Borjas (1987) and Greenwood, et al (1991 have both used national descriptives of income inequality and political stability. These are aggregate measures, however, and do not give information on inter-regional differences within the source country.

2Hildebrand, Bruno, in Statistischen Mittheilungen über die Volswirtschaftlichen Zustände Kurhessens, pp. 53, 99.

3This dataset is the fruit of more than a decade-long effort on t he part of the Hessian State Archive and historian Inge Auerbach to catalog the extensive records on Kurhessian emigrants.

4Köllman, Wolfgang. Quellen zur Bevölkerungs-, Sozial-, und Wirtschaftsstatistik Deutschlands, pp, 399-411.

5Farmers will be eventually added in from the data collected on villages.

6Assion, Peter. Über Hamburg nach Amerika, pp. 69.

7Unfortunately, the lack of suitable census data for any part of Germany from this time period prevents a comparison of these results with those from a German population.