Tabellini and others note that immigrant settlement patterns may reflect other local factors that are uncaptured by fixed effects. For instance, immigrants may settle in states that provide larger safety nets and that might appear to show that they grow the size of government while, in fact, they might just settle where it is already biggest. Similar to the approach used by Tabellini, we use a shift-share instrumental variable (IV) to partially address the endogeneity concerns. In particular, we construct an instrument that apportions national immigrant inflows into the United States to states according to the shares that settled in each state in 1960 using this method.
Government Spending and Immigration, 1970-2010
Figure 3 shows the correlation between changes in real state-level per capita total expenditures and changes in the state-level foreign-born population share. Figure 4 shows the correlation between changes in real state-level per capita direct expenditures and changes in state-level foreign-born population share. Both figures show a negative correlation between expenditure growth and growth in the foreign-born share of the population.
To test whether these negative relationships are statistically significant, we run a series of fixed effects IV regressions. In each regression we instrument a state’s foreign-born share using a shift-share instrument and find negative correlations between state government spending growth and the foreign-born share (Table 1). The first column of Table 1 shows the first stage of our IV regressions. As a measure of our instrument’s strength we report a first stage F-statistic of 29.8, which is a strong instrument. Columns 2 through 5 show correlations between state revenue growth, spending growth, and foreign-born shares.
The dependent variable in Columns 3 through 5 (Table 1) is the log difference of each per capita measure. The coefficient for each specification relates a percentage point increase in a state’s foreign-born share to a percentage change in the level of per capita revenues or expenditures. Our simple regressions show no significant correlation between both growth in general revenue and revenue from the state’s own sources, controlling for fixed effects. Only the estimate for total expenditure growth is significant at the 5 percent level. Column 4 shows a significant, negative correlation between a state’s foreign-born share and total government expenditure growth of about 1.3 percent. The estimate for direct expenditures in Column 5 is negative and only marginally significant.
Taken together, our regression results suggest that a larger share of immigrants at the state level is correlated with slower growth in state per-capita government expenditures. While these regressions attempt to control for endogeneity and other unobserved factors, the results are more suggestive than causal. Establishing a causal link between immigration and government spending or revenues would require additional empirical work outside of a blog post.
How immigrants affect the size and growth of government is a complex issue with many moving parts. Immigrants have tended to vote for the Democratic Party or its predecessor since the 1790s. When the Democratic Party was the laissez-faire party, immigrants voted for it. When it became the interventionist party, immigrants continued to vote for it. They did so because the Democratic Party has been more pro-immigration than its competitors during most of American history.
With some exceptions, people tend to choose a political party first and then change their opinions to match that party’s platform. Public opinion certainly has a huge influence on public policy as our system is designed to turn voter preference into policies, but the empirical work above indicates that more immigration is not related to growth in the size of state governments. The reactions of native-born voters to immigrants complicate the story significantly as immigrants don’t merely shift the median voter by being present but also by influencing the opinions of natives on myriad policy issues. Much of the work on this topic shows no relationship or a negative relationship between the two.
 Marx, K. and Engels, F. (2010). Collected Works, Volume 43, Letters 1868-70. London: Lawrence and Wishart, 474-6; Marx, K. and Engels, F. (1971). Marx and Engels on the Irish Question. Moscow, 354.
 Pierson K., Hand M., and Thompson F. (2015). The Government Finance Database: A Common Resource for Quantitative Research in Public Financial Analysis. PLoS ONE doi: 10.1371/journal.pone.0130119
 The first year of available for most states in the GFD is 1972. For simplicity, we backshift the 1972 data to represent 1970. Since the District of Columbia reports data directly for 1970, we use the provided 1970 data.
 Steven Ruggles, Sarah Flood, Ronald Goeken, Josiah Grover, Erin Meyer, Jose Pacas and Matthew Sobek. IPUMS USA: Version 10.0 [dataset]. Minneapolis, MN: IPUMS, 2020. https://doi.org/10.18128/D010.V10.0
 We pool the 2008-2012 ACS 5-year estimates for the year 2010.
 See Card (2001) or Tabellini (2020) among others.
 This F-statistic is the Kleibergen-Paap F-statistic, which adjusts for clustering. A value higher than 16.4 implies a relatively strong instrument according to the Stock and Yogo (2005) critical values.
This article by Alex Nowrasteh and Andrew C. Forrester first appeared at the Cato Institute.