Jeff Madrick recently wrote a New York Times article boldly claiming “there is no evidence” that “low-tax countries grow faster than high-tax countries.”
“One of the most interesting research papers on the subject,” he wrote, was done “a few years ago” by Sergio Rebelo and Nancy Stokey.” That paper first appeared in 1993 — the same year Rebelo co-authored a more ambitious study with William Easterly, who Madrick quotes approvingly. That study, “Fiscal Policy and Economic Growth,” found that “marginal tax rates … tend to be highly correlated with the level of income.” Easterly and Rebelo also found “a negative association between growth and one tax variable: the marginal tax rate.”
But Madrick, of course, only mentions Rebelo’s other 1993 paper. And he also prudently avoids mentioning that Stokey’s more prominent co-author, Nobel Laureate Robert Lucas, has estimated that cutting taxes on capital to zero could raise the stock of capital by 30 percent to 50 percent.
According to Madrick, Rebelo and Stokey “noted that income tax revenue in the United States rose to 15 percent of gross domestic product in 1942, from about 2 percent in 1913, when the tax was introduced. … But, they concluded, ‘This large rise in income tax rates produced no noticeable effect on the average growth rate of the economy.’”
The first problem is that averaging growth rates from World War I through the Great Depression obscures what happened when. From 1913 to 1921, the highest income tax rate soared from 7 percent to 73 percent and average economic growth was minus 0.3 percent. The top tax rate was then cut from 56 percent in 1922 to 25 percent from 1925 to 1928, and economic growth averaged 6 percent from 1921 to 1929.
In 1930, there was a huge increase in taxes on trade (tariffs). In 1932, the Hoover administration nearly tripled all tax rates, putting the highest rate at 63 percent. From 1929 to 1938, economic growth again averaged minus 0.3 percent. But lumping it all together, as Madrick does, makes it technically correct to say that growth of real GDP averaged 2.9 percent from 1913 to 1942, down modestly from a 4.3 percent pace from 1870 to 1913.
The second problem is that Rebelo and Stokey confuse tax revenues with tax rates. Revenues collapsed after the huge increases in tax rates and tariffs. Individual income tax receipts in 1939 were still 10.3 percent smaller than in 1930. And revenues from the steep tariffs on imports fell 38.4 percent from 1929 to 1936. In 1942, by contrast, individual income tax revenues rose by 130 percent in a single year because of war mobilization. To refer to the tax receipts of 1942 as if that was part of a long-term trend was deceptive.
Madrick also misquotes Joel Slemrod and Jon Bakija’s 1996 book, “Taxing Ourselves,” by saying, “Relatively low-tax nations like the United States and Japan did well, they found, but so did high tax nations in Scandinavia.” What Slemrod and Bakija actually wrote was that “high tax countries like Sweden … did relatively poorly.”
Aside from that subtle distinction between doing well and doing poorly, it is not even clear what it means to boast that Sweden has merely a third less GDP per person than the United States when well over half of Sweden’s GDP consists of government spending. Numerous studies, including Easterly and Rebelo, show that government consumption spending is bad for growth. My recent column “Supply Side Goes to Harvard” cited two of the latest studies showing that big government and high tax rates hold economies down.
Madrick instead looks backward, claiming Slemrod and Bakija “contradict earlier findings that purported to show that high taxes reduced growth rates.” One of those “earlier findings” was the Easterly-Rebelo study. But Easterly and Rebelo were estimating the effect of marginal tax rates — the income tax on each extra dollar earned. Slemrod and Bakija, by contrast, were talking about average tax receipts from 1970 to 1990 — the sum of all taxes divided by GDP. That is a useless measure of tax distortions because taxes that do the most damage usually yield the least revenue.
Countries with the most progressive income tax rates, with top rates of 50 percent or more, collect very little revenue from that tax. That includes Japan and Turkey, which Slemrod and Bakija mislabeled as “low-tax” countries. France, with income tax rates as high as 57 percent, collects only about 6 percent of GDP from that tax. The United States, with federal tax rates below 40 percent, collects about 11 percent of GDP. When New Zealand’s top tax was still only 33 percent, their individual income tax brought in nearly 17 percent of GDP. And since Russia adopted a 13 percent flat tax last year, revenues have been rising dramatically.
Countries trying to collect punitive taxes from the rich, like France and Sweden, end up short of rich people to tax. As a result, they mainly rely on flat-rate payroll and sales taxes (VAT). That is another reason we cannot combine revenues from all taxes to gauge the economic damage from high income tax rates. A 1999 study in the Journal of Public Economics by Kneller, Bleaney and Gemmell found that progressive income taxes were clearly harmful to growth, but found far less damage from flat-rate taxes on what people earn or spend.
Contrary to Madrick, the evidence from 1913 to 1942 shows that increased taxes and tariffs were accompanied by long periods of zero economic growth, and that sharply lower tax rates from 1922 to 1929 had the same invigorating effect that lower tax rates had in1964-69, 1983-89 and 1997-99. Ironically, even Madrick’s handpicked sources, Easterly and Rebelo, found that marginal tax rates affected both the level of income and the rate of economic growth. Far from proving “there is no evidence” that lower tax rates are conducive to faster economic growth, Madrick has inadvertently demonstrated the extreme degree of trickery required to defend such an indefensible remark.