A Brief History of Pre-1914 German Emigration

German emigration before World War I had three main characteristics. First, most departures from Germany took place before the worldwide mass migration. Along with Ireland, which was struck by famine, and Great Britain, which aimed to settle its colonies, Germany was thus one of Europe's main labor exporters before 1860. Between 1851 and 1860, more than one million Germans left their country. The second characteristic was that most German migrants (90.6 percent on average between 1871 and 1913) chose the United States as their destination country. As a result, Germany became the biggest source of migration to the United States before World War I. Between 1850 and 1914, 5 million Germans crossed the Atlantic to work in the United States, that is, more than Italians (4 million), Austro-Hungarians (4 million), Irishmen (3.5 million), Britons (3.5 million), or Russians (3.2 million). The third characteristic is that Germany was one of the few European countries that experienced a structural drop in the emigration rate before 1914.

Figure 1 shows the evolution of the German emigration rate during the century that preceded World War I and the most important events that characterized the period. It is possible to distinguish three phases. The first one runs from 1820 to 1854 and is marked by a continuous increase in the number of German departures. Despite its strong development, German industry could not absorb the growing population (24.9 million inhabitants in 1820; 33.7 millions in 1850), and overseas emigration represented an important safety valve for the labor force. The peak years of German emigration were reached at the beginning of the 1850s, motivated by the twin events of an agricultural depression in Germany and the gold rush in California. It is also important to note the importance of chain migration in this process.Footnote ²

The second phase ran from 1854 to 1882 and was marked by a strong volatility in the emigration rate. The American Civil War (1861–1865) cut the German emigration rate (0.9 per thousand in 1862), but after 1865 emigration quickly returned to earlier levels (3.2 per thousand in 1866 and 3.6 per thousand in 1867). The Franco-Prussian War temporarily interrupted the transatlantic flow (76,000 emigrants in 1871 versus more than 120,000 in 1870 and 1872), although not in the same proportion as that of the Panic of 1873, which initiated a five year depression that brought about a sharp drop in the emigration rate (2.76 per thousand in 1873; 0.54 per thousand in 1877). Emigration from the German States surged in the late 1870s and early 1880s, in a small part due to Bismarck's Antisocialist Laws, and in a large part as a consequence of the American economic boom. As a result, between 1880 and 1882, about 540,000 Germans landed on American soil.

Figure 1

EVOLUTION OF THE GERMAN EMIGRATION RATE: 1820–1914

Notes: The emigration rate corresponds to the annual number of emigrants per thousand resident population. Due to the lack of information on immigration and return migration, the figures refer to gross emigration. Explanations in bold refer to German causes, and the ones in italics are related to American factors.Sources: Emigration: Ferenczi and Willcox, International Migrations; Population: Maddison World Economy.

Finally, the years 1881–1882 were the last with high emigration rates (more than four per thousand) and were followed by a rapid decline in outflows. This third period continued to World War I, leaving Germany one of the few European countries with a diminishing emigration rate. Whereas most countries (in particular the southern and eastern European nations) experienced booming emigration in the 1890s and 1900s, Germany had one of the lowest emigration rates (one per thousand on average between 1885 and 1913) and the most rapid fall in migration between the periods 1865–1884 and 1885–1913 (Table 1).

The Limits of Traditional Explanations for the Decrease in German Emigration

Traditional explanations for declining emigration emphasize the international wage convergence process.Footnote ³ As labor leaves a country, real wages should rise, while in receiving countries they should fall. The result should be a narrower wage gap between sending and receiving countries, which should reduce migration. Yet this analysis does not explain the German decline. As Figures 2 and 3 show, the wage ratio between the United States and Germany was remarkably stable before World War I. During the period 1867–1913, the variance of the wage ratio was equal to 0.005, and the wage ratio was exactly the same in 1913 as in 1867 (1.84). It is therefore unlikely that the decrease in German emigration between 1882 and 1914 was the result of wage convergence.

Table 1

COMPARISON BETWEEN THE EUROPEAN EMIGRATION RATES: 1865–1913

Note:

The emigration rate corresponds to the annual number of emigrants per thousand resident population.

Source:

Author's calculations based on Ferenczi and Willcox, International Migrations.

Another explanation for the fall in emigration rates lies in the rapid industrialization of the German economy in the second half of the nineteenth century. The growth of the German GDP was indeed impressive, and this new prosperity surely influenced the migration pattern of Germany. The need to go abroad to find a job was less pressing, and Germany became a labor-importing country at the end of the nineteenth century. Yet, industrialization cannot account for the drop in German emigration either. The GDP growth was much higher in the United States than in Germany (Figure 4), and the wage gap between both economies remained high, which means that there were still more opportunities to improve living conditions in America than in Germany.

A complementary argument comes from demography: reduced migration could have stemmed from the decline in Germany's fertility at the end of the nineteenth century. However, as Figure 5 shows, the German demographic transition only began to make its mark at the turn of the twentieth century, when the natural growth rate of the population began to ebb, having peaked in 1902 at 15.7 per thousand. It is hence very unlikely that there was a direct link between the slowdown in the natural growth rate and the decrease in emigration, all the more so since demographic changes affect emigration movements with roughly a 20 year lag.Footnote ⁴

One final possibility is that German emigration slowed because of competition with migrants from other nations. According to Oliver Grant, more arrivals from eastern and southern European countries at the end of the nineteenth century heightened competition between “old” and “new” migrants in the U.S. labor market.Footnote ⁵ Grant argues that German workers were particularly affected because their skill endowments were roughly the same as those of immigrants from eastern and southern Europe, and because they tended to settle in the same states as the new migrants. Yet this argument overlooks the impact of chain migration. Unlike eastern and southern candidates for emigration, German nationals benefited from the advice and financing assistance from the relatives who already lived on the other side of the Atlantic, which should have given them a comparative advantage. International competition should have favored them instead of new migrants, who could not rely on the chain migration process.

Figure 2

AMERICAN AND GERMAN REAL WAGES: 1850–1913

Note: Wages correspond to international real wages (100 = U.S. wages in 1871).Source: Williamson, “Global Markets.”

Figure 3

REAL WAGE GAP: 1850–1913

Notes: Wage gap=U.S. wages/German wages. 2.4 = maximum; 1.22 = minimum; 1.9 = mean.Source: Author's calculations based on Williamson, “Global Markets.”

Figure 4

COMPARISON BETWEEN GERMAN AND AMERICAN GDP PER CAPITA GROWTH: 1870–1913

Note: 1870 = 100.Source: Author's calculations based on Maddison, World Economy.

It therefore appears that the traditional explanations (decrease in the wage gap, industrialization, fall in fertility, or international competition) may represent part of the reason why German migration declined, but they are not a complete answer. We have to look elsewhere, in particular at the social legislation of the 1880s. At the time, the combined effects of industrialization and the war against France had heightened a new sensitivity to the consequences of migration, both in economic and military terms:

As a result, the attitudes of public authorities began to change. Social legislation, alongside other laws such as the broadening of land ownership, constituted part of their response.

Meanwhile, in the United States

Whereas European countries were inspired by Bismarck's scheme to set up their own social insurance system, the implementation of social programs in the United States was tardy. American authorities, in the name of both economic liberalism and social Darwinism, considered that their country did not need any social intervention because domestic and immigrant workers could benefit from a large number of jobs, as well as high wages. It was only when the Great Depression began that opinions changed and first social measures could be adopted. Before that it was the family and friends network or private charity institutions that took care of the poor.Footnote ¹³

For insurance, American workers had to organize, and from the 1870s they created mutual associations, which provided their members with health or life insurance. Although employers sometimes contributed to these mutual associations, the bulk of the costs fell on the employees, who generally lost their rights to benefits when they left a firm. Some trade unions, in particular railway unions, did pay sick benefits to their members, but they were a minority. In any case, benefits were low (on average, five dollars a week) and of limited duration (13 weeks at the most).Footnote ¹⁴ The voluntary system in America did not allow a significant expansion until the 1930s.

In fact, before the enactment of the Social Security Act in 1935, the only real organized insurance mechanisms were the workmen's compensation systems, which covered workers against industrial accidents. Yet their scope too was limited, particularly before World War I. The most significant measure was adopted in 1910 by the State of New York. It required employers in 12 dangerous occupations to pay compensation, at a rate set in law, in case of accidents. Twenty-two other states enacted similar workmen's compensation laws between 1911 and 1914, but it was only in the 1920s and 1930s that the system really expanded and the employers' “social responsibility” was widely recognized. Nonetheless the compensation laws did not immediately bring about a development of universal insurance against industrial accidents. Most employers could still chose not to insure and most did not adopt workmen's compensation insurance, which only became compulsory in the 1930s.

Old age insurance was another problem before the New Deal: people too old to work had no option except charity, either private or public. Wages were usually too low to permit workers to save in the long run, whereas pensions and insurance were costly or simply unavailable to industrial workers. Private pensions really only developed in the 1920s, and the only other private insurance mechanism before World War I was life insurance, which represented a small share of total personal savings—8.2 percent, on average, between 1897 and 1913.Footnote ¹⁵ As for public pensions, the first bill to create them was introduced in 1903 in Massachusetts, but it failed to pass, and other proposals for old age pension laws at the time met a similar fate.Footnote ¹⁶ In 1935 the Social Security Act eventually implemented a federal scheme of old age pensions.

In total, the level of protection of American workers against the main threats to workers was very low before the Great Depression and virtually nonexistent before World War I. By contrast, most German workers were covered by social insurance mechanisms by 1913. The difference made America less attractive, and that can explain why German emigration dramatically decreased.

Table 2

CONTRIBUTIONS TO THE SOCIAL INSURANCE SYSTEM

Note:

Contributions correspond to the share that the main agents had to pay on average during the whole period.

Source:

Author's calculations based on Statistisches Jahrbuch für das Deutsche Reich.

Was the German Social Legislation a “Free Lunch”?

Transfer mechanisms are supposed to reduce productivity and hence deter economic growth, yet as Peter Lindert has argued, it seems that the welfare state has had no major impact on long-term growth.Footnote ¹⁷ In that sense, it is possible to consider social spending as a “free lunch,” because social benefits do not have significant economic costs. But was that the case in late-nineteenth-century Germany? Did Bismarck's social legislation carry a cost in terms of slower economic growth that might have induced migrants to leave?

The main costs of the 1880s' social legislation were the contributions to the insurance system. Employers and employees were the main contributors (Table 2). Between 1885 and 1913 employers paid almost half the contributions (44.7 percent), the employees paid 39.9 percent, and the State only 5.4 percent. The remainder of the contributions (10 percent) corresponds to interest earned on previous years' surpluses. Employers paid for the industrial accident system, whereas the employees had to pay for most of the health insurance. The state shouldered little of the expense for any of the programs.

The state's minor financial contribution does not mean that its role was unimportant. It contributed to implementing the system and coordinating involved agents' actions:

Figure 6

GDP PER CAPITA AND REAL WAGES IN GERMANY: 1880–1913

Note: 1880 = 100.Source: GDP/capita: Maddison, World Economy; and Real wages: Williamson, “Global Markets.”

As a result, the impact of contributions on the German economy was limited because all the firms had to adopt the system at the same time. The risk of free riding was very low, and the repercussions in terms of competitiveness between German firms were virtually nil.

In any case, Bismarck's social legislation does not seem to have restrained economic activity, because the GDP grew at a rapid 3.1 percent annual pace between 1885 and 1913. That the social insurance did not hobble economic growth probably stemmed from its positive impact on workers' health, that is, on human capital and hence on productivity. As Figure 6 shows, GDP per capita rose faster than real wages between 1880 and 1913, which implies that firms kept the main part of value added. In a sense, there was an implicit contract between employers and employees, with firms agreeing to pay contributions to the insurance system knowing that their employees would be healthy and therefore more productive. Workers, for their part, accepted lower wage increases in return for social benefits. The upshot was that the insurance system's losses were relatively low, and as a result Bismarck's measures did not have a significant impact on long-term growth and the attractiveness of the German economy.

How Do Social Benefits Affect Emigration?

Most studies of international migration emphasize the wage gap between sending and receiving countries. As Jeffrey Williamson has shown, the wage convergence that accompanied mass migration phenomenon slowed migration from Europe to the New World.Footnote ¹⁹ Migration was also affected by the demographic structure and industrialization in Europe, by differences in employment opportunities, and by the chain migration process. Migration policies are also important, above all to explain why migration plummeted after World War I, and why it is lower than one would expect today given the difference in incomes between developing and industrialized countries.

But the literature on international migration rarely mentions the effect of social benefits in the emigration process. Dudley Kirk, whose study on the behavior patterns of the European population during the interwar period includes a chapter on the slowdown of labor mobility, does refer to social insurance.Footnote ²⁰ In the same way, Henry Gemery tries to establish how social insurance in sending countries could have affected labor flows in the interwar years.Footnote ²¹ But Kirk and Gemery are an exception. And even for them, the question of social insurance is not a central issue.

One would certainly expect the welfare state to reduce the incentive to migrate, for the existence of social insurance raises the opportunity cost of moving. With a pension system, for example, workers do not need to save money for their old age, and it is therefore rational for them to agree to lower wages during their working lives in anticipation of the pensions they will receive when they retire, so long as they remain in their country. Programs such as health or accident insurance have a similar effect. Workers who are sick or injured will receive financial compensation, but if they move to a country where such measures do not exist, they run the risk of losing everything following an accident.

In that sense, social insurance constitutes a kind of indirect wage that candidates for migration will take into account in their income expectations: the higher the domestic indirect wage, the lower the emigration rate. What really matters in the wage gap is the sum of direct and indirect wages, that is, the combination of labor earnings and social benefits. Whether the social benefits are the product of private insurance or public subsidies is not really an issue, for in both cases the results for the workers are the same: their net wage earnings are lower than in the countries without the social programs (because they have to pay private insurance premiums or additional taxes), but they in turn benefit from the insurance mechanisms (either public or private) that workers in other countries do not have. However, it is important to know whether the system is compulsory or not. Indeed, compulsory insurance helps to avoid the adverse selection phenomena that affect virtually all insurance mechanisms. In general terms, insurance is more expensive when optional, and fewer people obtain insurance.

Eventually, the development of welfare benefits contributes to a reduction in emigration because it increases the opportunity cost of moving abroad. The underlying assumption is that potential migrants are risk adverse. They would therefore prefer to receive lower direct wages at home, along with broad insurance against the major risks, rather than earn higher wages in another country, but without the guarantee of a job and with less social insurance.

Calculation of German Indirect Wages

The calculation of the indirect wages that the German social insurance provided hinges on the way direct wages are estimated. My source for wages is Ashok Desai.Footnote ²² Between 1871 and 1886, Desai's annual money wages are a weighted average for six industries: printing, building, coal-mining, machines, cotton textiles, and steel. The employment in these industries in 1882 is used for weighting. From 1887 onwards, calculations are based on average earnings in industry, transport, and distribution. In total, 27 sectors are taken into account. Desai focuses on the wages of the workers who had accident insurance, but their benefits are not included in his wage measures. Insurance contributions were proportional to employees' earnings, and Desai derives average wages from 1887 to 1913 by dividing total earnings in each industry by the number of insured persons. Then, he calculates a weighted average by using the number of insured workers (Appendix Table 5, column 2).

In the same way, Desai estimates the cost of living before 1871 and 1913. Because nominal wages refer to industry, the cost of living is basically based on urban prices. It consists of an index of the average cost of consumer goods. The index is based on estimates of incomes and spending compiled from 852 families with incomes below 3,000 marks and with three to five children (Appendix Table 5, column 3).

Real wages are calculated from money earnings and cost of living and are given in 1871 prices (Appendix Table 5, column 4). In order to make a comparison, German real wages are then calculated in terms of American wages (100 = 1871), using Williamson's international benchmark (Appendix Table 5, column 5).Footnote ²³ These can be compared with German international real wages, as calculated by Williamson (Appendix Table 5, column 6). The correlation between columns 5 and 6 is 0.984.

Appendix Table 6 shows the impact of workers' contributions on direct wages, with German remuneration measured relative to pay in the United States. Nominal contributions are calculated by dividing total contributions by the German economically active population. In that sense, total contributions represent the weighted average of contributions for health, disability, and old age insurance (again, employees did not contribute to the accident insurance system). Net direct wages are the difference between direct wages and contributions.

It is true that the value an individual places on social benefits will depend on personal circumstances, which will make it difficult to determine what the actual level of indirect wages is. Working in a coal mine, for instance, increases the probability of suffering an injury, which raises the benefits of accident insurance. But at the same time the likelihood of reaching retirement age diminishes. An office worker, by contrast, is less likely to have an accident, which lessens the benefits of accident insurance, but his probability of receiving an old age pension is higher than that of a miner. Indirect wages thus vary not only with the level of direct wages, but also with the occupation and location. In order to cope with this difficulty, my strategy was to set indirect wages equal to the average level of expenditures for insured workers. One could argue that value was zero for uninsured workers, but the new social insurance scheme probably influenced their decisions. Indeed, uninsured candidates for emigration, such as agricultural laborers, had the option to move either abroad or inside Germany, mainly to the industrial cities, where they would be eligible for insurance. Therefore, Bismarck's social legislation also had an effect on noninsured workers.

To be precise, I calculated German indirect wages Wi_ger as the sum of the ratio of expenditures by category of insurance to the number of insured workers in each category

where Exp_health, Exp_acc, and Exp_old.dis are the expenditures for health insurance, accident insurance, and the combined expenditures for old age and disability insurance, respectively, and Ins_health, Ins_acc, and Ins_old.dis the number of insured workers for each class of insurance. The results of these calculations are reported in Appendix Table 7. Appendix Table 8 presents German total wages, that is, the sum of direct and indirect wages.

Was the social insurance in Germany before World War I enough to keep German workers from moving abroad? The overall weight of social benefits was admittedly relatively low. On average the share of indirect wages was only 4.3 percent of total wages and it reached a peak of only 5.9 percent in 1913. But beyond the figures of indirect wages, the fact remains that there was a strong and fast reaction to the new social legislation. Table 3 shows the evolution of German emigration rates after the adoption of the three 1880s social insurance laws. The strong drop in emigration that followed each law is striking, in particular following the health and industrial accident programs, which had direct effects on German workers' living conditions. Old age and disability insurance had less of an impact, because there were fewer potential beneficiaries. It is noteworthy that the 1880s were a period of relative prosperity in the United States, which reinforces the idea that social changes in Germany had an impact on labor flows.

Risk aversion can help explain why the legislation could have had such an impact, even though the indirect benefits were only a fraction of wages. Potential migrants, after all, could move to the United States, where the economy grew faster and direct wages were higher than in Germany. However, they had no guarantee of finding a job there, they faced tougher competition from other migrants and could lose everything in the event of sickness or accident. On the other hand, they could stay at home, where direct wages were lower but where they would receive social benefits that would give them insurance against sickness and accident. In addition, they would not have to cope with the psychological costs of moving to a foreign country. For risk-adverse individuals, the utility of this insurance would exceed the relatively small monetary value of the indirect wages, and that could explain why the German social legislation had such an impact on emigration patterns.Footnote ²⁴

Table 3

SOCIAL INSURANCE AND GERMAN EMIGRATION

Source:

Author's calculations based on Ferenczi and Willcox, International Migrations.

Social Legislation and German Emigration: The Econometric Evidence

In order to establish the role of Bismarck's social legislation on German emigration, I have estimated the legislation's impact using the Harris-Todaro model, which relies on the assumption that migration is the result of differences in expected rather than actual earnings. In the model, potential migrants consider the different labor market opportunities available and select the one that maximizes their expected gains from migration. Expected gains are measured by difference in real incomes between countries and by the probability that a new migrant finds a job abroad. The contribution of the current analysis is that it takes into account social insurance mechanisms, by introducing a distinction between direct and indirect wages.

The results are reported in Table 4. The estimation method is ordinary least squares (OLS) using the German time series from 1872 to 1913. The dependent variable is the annual gross emigration rate. Emigration figures comes from Imre Ferenczi and Walter Willcox, and population from Angus Maddison.Footnote ²⁵ There are no data on return migration, which explains why the figures are for gross migration. Around 90 percent of German migrants went to the United States between 1871 and 1913. Therefore, the analysis focuses on migration from Germany to America. The direct wage gap is the logarithm of the ratio of American to German real wages. American real wages are given by Williamson, and German real wages are derived from Desai.Footnote ²⁶ For Germany, real wages are net direct wages, that is, the difference between direct wages and workers' contributions (Appendix Table 6, column 7). The total wage gap is the logarithm of the ratio of total wages in the United States to total wages in Germany, where total wages equal direct plus indirect wages. Given the lack of social insurance in the United States during the studied period, American indirect wages are considered to be equal to zero. German total wages are taken from Appendix Table 8, column 5. The total wage gap is then the logarithm of the ratio of direct wages in the United States to the sum of direct and indirect wages in Germany. Finally, employment opportunities are given by the evolution of economic activity, which is measured by the deviation of the logarithm of real GDP from a linear trend. GDP calculations are based on Christina Romer for the United States and Maddison for Germany.Footnote ²⁷

Table 4

REGRESSION RESULTS (dependent variable: German emigration rate)

* = Significant at the 10 percent level.

** = Significant at the 5 percent level.

*** = Significant at the 1 percent level

Notes:

Number of observations: 42. The t-statistics (in parentheses) are corrected for heteroskedasticity using the White procedure. Indeed, Breusch-Pagan tests indicated the presence of heteroscedasticity in the equations. Due to the inclusion of a lagged dependant variable in the regression, there is a risk that least squares estimates may be inconsistent if the error terms are correlated. Nevertheless, tests for autocorrelation rejected the hypothesis of autocorrelation in the residuals. In order to explore the issue of the stationarity of the variables, which is inherent in any time series analysis, unit root tests were performed on the dependent variables, using the augmented Dickey-Fuller test (ADF). The results show that there are no serious stationarity problems.

Source:

See the text.

All the equations show the importance of economic conditions in the decision to migrate: an improvement in the domestic economic activity was accompanied by a decreased in labor outflows, while a rise in the U.S. GDP meant more emigration.Footnote ²⁸ Moreover, the coefficient of the lagged dependent variable (one year) is positive and significant at the 1 percent level in all six equations, which confirms the role of chain migration. But what is really interesting is the impact of the wage differential on migration.

Equation 1 measures the effect of the direct wage gap on German emigration. Although the coefficient has the expected sign (an increase in the American/German wage gap brings about a rise in the emigration rate), it is not significant, which is consistent with my argument that the gap in direct wages alone does not explain the decline in German emigration. Equation 2 takes into account indirect wages, by measuring the impact of the total wage gap. In this case, the coefficient of the wage ratio not only has the expected sign, but it also is significant at the 5 percent level, which confirms the role of social insurance in explaining German emigration. The increase in German indirect wages reduced the total wage gap between the United States and Germany, and that in turn led to less emigration.

Because indirect wages tended to rise over time, there is a risk that the total wage gap variable might pick up other time-related effects. I have added a time trend to all of the remaining equations to see if this is the case. This additional variable does not affect the previous result. In particular, the coefficient in regressions 2 and 3 are similar, and the level of significance is the same. The t-statistic for the total wage gap variable is higher, which strengthens the robustness of the results.

Bismarck's social legislation meant not only the adoption of a complete scheme of social insurance, but also an improvement in working conditions, including a reduction in working hours. To see if the shorter hours also affected emigration, I added the differential between the average number of working hours in the United States and in Germany to regression 4. Figures on average weekly working hours are taken from Brian Mitchell.Footnote ²⁹ The coefficient of the differential in working hours is negative and significant at the 1 percent level, implying that an increase in the working-hour differential entailed a fall in German emigration. Thus, while German workers had to work 72 hours a week, on average, in 1872, American workers worked “only” 64 hours. But, at the end of the period (1913), the Germans worked 57 hours a week and American workers 55.5 hours. Even though average working hours in Germany were still higher than in the United States, the gap between the countries significantly decreased. It is important to note that although in equation 4 the coefficient of the total wage gap is lower than in equations 2 and 3, it is still significant at the 5 percent level. Therefore, not only indirect wages but also improvement in working conditions, proxied by the differential in working hours, played an important role in the decision of German workers to migrate or not.Footnote ³⁰

To be sure that indirect wages were the determining factor in the migration decision, I removed the total wage gap from regression 4 and replaced it with the direct wage gap in regression 5. As in regression 1, the coefficient is positive but insignificant. Finally, equation 6 tests the demographic explanation for the drop in German emigration, by adding as an explanatory variable the German rate of natural increase, lagged 20 years to allow for entry into the adult labor force. The coefficient of this new variable has the expected sign (an increase in population brings about a rise in emigration), but the coefficient is not significant. Therefore the hypothesis that slower German population growth explains the decrease in emigration can be rejected.

The econometric results support my argument that social insurance reduced German emigration, but how could small variations in indirect wages have such a pronounced effect on the decisions of potential migrants? To begin with, the rise in indirect wages in Germany reflected not just an improvement in social benefits but also (and even more so) a rise in the number of insured workers. Therefore, the impact on emigration could result not from the increase in expenditures per worker but from the growing number of workers enrolled in the system. Another possible way for the social reforms to have affected German emigration is via their redistributive effect. The burden of the system was mainly on the employers' shoulders, in particular in the case of the industrial accident system, which covered most German workers (94 percent of the economically active population in 1913). In addition, the German State, that is, taxpayers, financed much of the old age and disability system. German workers (and notably the poorest among them, who did not pay taxes) were thus the net beneficiaries of the social system. It is in any case telling here that the German emigration rate did not pick up when the American economy revived in the second half of the 1890s, in contrast to what happened elsewhere in Europe. The conclusion that social insurance mechanisms mattered for potential migrants seems therefore inescapable.