Over the past three decades, America's state and local governments have experienced a large and underappreciated divergence. Some places, usually but not always led by Republicans, have become friendlier to free enterprise. Other places, usually but not always led by Democrats, have become friendlier to big government. Some political scientists call it the Big Sort. You can see the self-selection manifest itself in polls, in voting behavior, in migration patterns, and in policy outcomes.
Think of it as a vast natural experiment in economic policy. Because states have a lot otherwise in common-cultural values, economic integration, the institutions and actions of the federal government-testing the effects of different economic policies within America can be easier than testing them across countries. Governors, state legislators, and other policy makers have dutifully supplied the experimental data. And scholars have been studying the results.
I head the John Locke Foundation, a state policy think tank in North Carolina. A recent change in our state's political leadership prompted us to assemble a summary of all that diverse academic research for legislative leaders and our new governor, Pat McCrory. Setting aside studies published by think tanks, we limited ourselves to scholarly articles about economic policy published in academic or professional journals. At present our database contains 528 articles published between 1992 and 2013. On balance, their findings offer strong empirical support for the idea that limited government is good for economic progress.
Of the 112 academic studies we found on overall state or local tax burdens, for example, 72 of them-64 percent-showed a negative association with economic performance. Only two studies linked higher overall tax burdens with stronger growth, while the rest yielded mixed or statistically insignificant findings.
On smaller categories of taxation, the trend was similar: There was a negative association between economic growth and higher personal income taxes in 67 percent of the studies. The proportion rose to 74 percent for higher marginal tax rates or tax code progressivity, and 69 percent for higher business or corporate taxes.
Some of the strongest negative results appeared when scholars were able to isolate policy variables from background effects. For example, a 1996 study in the American Economic Review exploited the fact that some foreign countries gave domestic tax credits to companies that pay taxes in the United States, so those companies would be expected not to care much about state tax rates. In other countries, companies didn't receive such credits and would thus be subject to greater variation in state tax burdens. By looking at the behavior of firms based on their home country, author James Hines of the University of Michigan found that "state taxes significantly influence the pattern of foreign direct investment in the U.S." A 1 percent change in the tax rate was associated with an 8 percent change in the share of manufacturing investment from taxed investors.
Another study, published in Public Finance Review in 2004, zeroed in on counties that lie along state borders. Because territories in such close proximity often share characteristics that might not be captured in other measurements, this is a promising approach for isolating the effects of state policy. Studying 30 years of data, the authors concluded that states that raised their income tax rates more than their neighbors had significantly slower growth rates in per-capita income.
Similarly, economists Brian Goff, Alex Lebedinsky, and Stephen Lile of Western Kentucky University grouped pairs of states together based on common characteristics of geography and culture. This is the economic equivalent of studying identical twins to probe the relative importance of genetics and environment. Writing in the April 2011 issue of Contemporary Economic Policy, the authors found "strong support for the idea that lower tax burdens tend to lead to higher levels of economic growth."
Liberal analysts often argue that isolating variables such as tax rates misses the point. States would be better served, they suggest, adopting a growth strategy that maintains or increases current tax burdens to fund education, infrastructure, or other government programs. Whatever economic cost may come with higher taxes, they say, is more than offset by the economic benefits of the amenities that government spending provides.
This argument may seem plausible in theory, but it lacks empirical support. Of the 43 studies testing the relationship between total state or local spending and economic growth, only five concluded that it was positive. Sixteen studies found that higher state spending was associated with weaker economic growth; the other 22 were inconclusive.
Admittedly, economic benefits may not be evident for total state budgets, which combine three different categories of spending: protective (law enforcement and the courts), productive (education and infrastructure), and redistributive (health, welfare, and other transfers). Although a few Keynesian bitter-enders insist that transfer programs such as Medicaid boost the economy via multiplier effects, most proponents of government spending as an economic development tool prefer to emphasize the potential benefits of schools, roads, and other infrastructure. That's wise: Nearly three-quarters of the relevant studies found that welfare, health care subsidies, and other transfer spending are bad for economic growth.
States Don't Invest Effectively
That said, the empirical evidence isn't overwhelmingly friendly toward "investments" in education and infrastructure either. While most relevant studies found that measures of education outcomes, such as test scores or college attainment, are correlated with economic growth, the same can't be said for expenditures. Of 79 research findings on the relationship between education spending and economic growth, only 30 were positive, with 34 findings of mixed or insignificant results and 15 negative. (In the latter cases, any economic gains from additional education spending are more than offset by the economic losses from the taxes required.) Of the 84 studies examining infrastructure spending, 44 percent found positive results, but most studies still found either inconclusive or negative relationships.
One recent paper deserves special attention. In the Winter 2013 issue of the Journal of Education Finance, two political scientists, Norman Baldwin of the University of Alabama and William McCracken of the Ohio State University, published one of the most carefully designed studies of education spending I've seen. For example, they built in time lags to account for the fact that education spending is cumulative-its net effects show up years later, when pupils become workers or entrepreneurs, not in real time. So in evaluating overall higher-education spending, Baldwin and McCracken used a seven-year lag to allow for university students to obtain degrees, find employment, and adjust to their new jobs. They used a 13-year lag for K-12 spending. Neither variable demonstrated a consistent relationship with state economic growth. A separate measure of state spending on academic research did show a positive effect from 1988 to 1996, but the effect turned negative for the 1997-2008 period.