Bryan Caplan has an interesting post on the recent Swiss referendum to restrict immigration from the European Union. Tyler Cowen also blogged on the same issue twice. Caplan’s point is that the Swiss imposed restrictions because there was insufficient immigration rather than too much. Areas of Switzerland that had fewer immigrants voted to restrict immigration while areas with many immigrants voted to keep the doors open.

A similar theory could explain why immigration quotas were first imposed in the United States after World War I. That war substantially reduced immigration from Europe. From 1904 through 1914, almost 1 million immigrants arrived annually in the United States – a total of 10.9 million. This large population, combined with their children, opposed numerous legislative efforts to restrict immigration from Europe.

World War I erupted in August 1914, slowing immigration and causing the percentage of immigrants to decline more than the increase in the second generation. During the four years of the war, slightly more than one million immigrants arrived. That minor decline, especially in the 1st generation, might be part of the reason why anti-immigration politicians succeeded in passing the first immigration quotas in 1921. During that time many non-citizens could vote and it was much easier to naturalize than it is today.

The post-war U.S. recession, the continuing blockade of Germany, and chaos in Europe prevented immigration from rebounding until 1921 when 805,228 people immigrated – the same year that numerical quotas restricted immigration for the first time. If the pre-war pace of immigration was uninterrupted by World War I, 4.6 million additional immigrants would have landed in America by that time – boosting the immigrant share of the population to somewhat less than 17.7 percent of the total population and the second generation by a smaller amount too. Combined, the first and second generations would have been equal to around 40 percent of the American population. Supporters of immigration restrictions might have understood this and known that immigration from Europe was about to rapidly accelerate, meaning that they only had a narrow window to approve restrictions before the changing nativity of the population made that more politically difficult.

Several reasons would have made it more difficult to achieve the 1921 vote to restrict immigration if there were that many more immigrants.

First, 66.9 percent of the House of Representatives voted for the bill. President Harding supported the law but a previous attempt to restrict immigration like this was vetoed, making the 2/3 threshold important. If there were 4.6 million more immigrants, it would have been more difficult to clear that threshold.

Second, redistricting in 1920 gerrymandered Congressional districts to reduce the political power of immigrants – which was aided by the slight decrease in the percentage of foreign born. Representatives who voted against the bill came from states that had, on average, 20.4 percent of their population as immigrants according to the 1920 census. Representatives who voted for the restriction came from states that had, on average, 10.7 percent of their population as foreign born. The 104 Representatives who did not vote came from states that had, on average, 15.2 percent of their populations who were foreign born.

Looking at Massachusetts offers a puzzle though. In 1917, Congress voted on an immigration restrictionist bill called the Literacy Act. In that year, only four of Massachusetts’ 15 Congressmen voted “yea” on the Literacy Act with 11 voting “nay.” For the 1921 Act, however, Massachusetts only had 13 seats. Of those 13, five voted “yea,” three voted “nay,” and five did not vote. Between 1910 and 1920, the immigrant population of the state increased by 2.5 percent and it voted for more immigration restrictions. Gerrymandering could explain this shift, although I do not have the data to show that, or something else might have changed.

Other factors contributed to the end of the first era of immigration in the United States. Southern politicians opposed it because it gave more electoral weight to the Northeast. Labor unions opposed immigration and some business interests began to turn against it in fear that immigrants would bring socialism with them. The growing state-based welfare programs might have contributed to the public turning against immigration, Italian immigrants who went on a wave of terrorism, the Eugenics movement, and numerous other factors likely contributed to ending native support for immigration.

If this hypothesis is true, U.S. voters will support a more liberalized immigration policy as the percentage of the population that is foreign born and the second generation continues to increase.

Different states have also produced more strict and more lenient immigration-related laws over the years despite large differences in the immigrant percentages across states. Here are partially complete lists of state laws:

Sources: American Community Survey, U.S. Census, Immigration Policy Center, and drea​mact​.org.

Sources: American Community Survey, U.S. Census, and Immigration Policy Center.

These lists do not include the states that didn’t pass laws or tried to pass them but failed, but they provide a starting point to analyze. States that pass pro-immigration laws typically have more immigrants as a percentage of their populations. Interestingly, the difference is not huge, although there is a great deal of variance. The average for anti-immigration states is 10.7 percent – pretty high. Compared to the Swiss example, anti-immigration American states have some immigrants – just enough to make some natives dislike them – and certainly aren’t devoid of them.

California and Arizona are odd cases because so much of their population was foreign born when they created their anti-immigration laws; but there is a good argument for excluding them from the anti-immigration list because they were the trend setters. California really pioneered the anti-immigration state law through Proposition 187 (it was certainly viewed that way by the public) in 1994. There was a higher fixed cost for Californians to develop the concept of an anti-immigration state law so it likely would have taken more of a public outrage to spur that bit of legislative innovation. The case is similar for Arizona and their anti-immigration laws in 2008 and 2010. After Californians and Arizonans developed the framework for opposing immigration on the state level, the marginal costs of other states copying anti-immigration laws were smaller, making it easier for others to adopt such laws – which is exactly what happened. Arizona also developed the anti-driver license idea for DACA recipients.

Excluding California and Arizona from that already short list lowers the average immigrant population of anti-immigrant states to 7.4 percent – almost half of the pro-immigration states – and the standard deviation to just 2.9 percent. Just eyeballing it, there might be a Kuznets curve for immigration restriction with the percent of a state’s population that is immigrant on the X‑axis and support for immigration restrictions on the Y‑axis. A state needs some immigrants to pass anti-immigrant laws, but after the immigrant population grows past a point, pro-immigration laws are instead passed. Getting to the far side of that curve makes further immigration restrictions very difficult to impossible.

This brings us back to Switzerland.  

Although 27.3 percent of Switzerland’s population is foreign born, far fewer than that can vote. Switzerland also doesn’t have birthright citizenship, so the number of second-generation Swiss who could vote against restrictions was likely small. This could also explain why despite Switzerland having a relatively high percentage of its population foreign born, it was able to pass anti-immigration laws like American states with low immigrant populations did.

This is not enough data to support my theory in the United States, but it is a starting point. To avoid some of the worst anti-immigration laws, it seems that the immigrant population only has to literally outgrow them.