The late President G.H.W. Bush famously reneged on his “no new taxes” pledge and signed the “Bush tax increase” on November 5, 1990, to take effect the following January. The new law was intended to raise more revenue from high‐​income households and unincorporated businesses. It was supposed to raise revenue partly by raising the top tax rate from 28% to 31% but more importantly by phasing‐​out deductions and personal exemptions as income on a joint return climbed above $150,00 (the phase‐​outs were called the PEP and Pease provisions). 

Treasury estimates expected revenues after the 1990 budget deal to be higher by a half‐​percent of GDP. What happened instead is that revenues fell from 17.8% of GDP in 1989 to 17.3% in 1991, and then to 17% in 1992 and 1993. Instead of rising from 17.8% of GDP to 18.3% as initial estimates assumed, revenues fell to 17%. In fact, revenues did not climb back to the 1989 level of 17.8% of GDP until 1995, despite much higher excise taxes since 1991.

Another way to gauge the 1990 and 1993 tax increase is to measure the revenue gains in real 2009 dollars, adjusted for inflation. According to Table 1.3 of the Historical Statistics in the U.S. Budget, real revenues (in 2009 dollars) soared from $1,308.8 billion in 1980 to $1,654.6 billion in 1990 (26.4%), as the top tax rate fell from 70% to 28%. After the Bush tax increases in 1991 and retroactive Clinton tax increases in 1993, by contrast, revenues were virtually no higher in 1993 than they had been before – $1,655.7 billion. GDP in 1993 was a bit larger than in 1990 but revenues fell as a percent of GDP despite higher excise taxes.

A recession began in October 1990, just as the intended tax increase was being enacted. To blame the weak revenues of 1991–93 entirely on that brief recession begs the obvious question: To what extent was a recession that began with a tax increase caused or at least worsened by that tax increase? 

Some describe the Bush tax increase of 1990 act of great political courage and bipartisan cooperation which supposedly helped shrink the budget deficit “by $492 billion … over just five years.” But that figure too was (1) just an estimate, (2) only 30% of it was ostensibly to come from higher taxes, and (3) most of the hoped‐​for added revenue was not from higher income tax on couples earning over $150,000 but from higher excise taxes on gasoline, alcohol, tobacco, telephones, etc. The gas tax went up a nickel; the beer tax was doubled. Nearly 10% of the revenue windfall was expected from a new luxury tax on cars, yachts, airplanes, furs, and jewelry which devastated those businesses (contributing to the recession) before being repealed in less than a year.

Journalists who look back at what happened to tax revenues after tax rates were raised or lowered, such as Washington Post fact checker Glenn Kessler, commonly rely on an updated version of a 1998 working paper by Treasury economist Jerry Tempalski. However, Tempalski only presented estimated effects on revenues, not actual effects. “Treasury estimates a bill when it is enacted… and sometimes reestimates a bill for several subsequent January budgets,” Tempalski explained, but some of “the first post‐​enactment estimates proved not very accurate.” Tax changes were often phased‐​in or phased‐​out, yet “the estimates… include no adjustment to capture the long‐​run, fully‐​phased‐​in effect of the tax bills.” Early estimates looked ahead only two years, later ones covered four.

These antiquated revenue estimates tell us nothing about what actually happened after tax laws were changed. They only tell us what notoriously erroneous revenue estimators expected. Yet the Tempalski estimates have been repeatedly cited as evidence that lower tax rates never even come close to “paying for themselves” by such leading journalists as Washington Post fact‐​checker Glenn Kessler and Lori Robertson of FactCheck​.org, and even by the chief economist for Tax Analysts, Martin A. Sullivan.

In the same vein, estimated revenue effects of the 1990 “tax increase” are still being cited as if they are facts rather than discredited old estimates. When discussing tax increases (or tax cuts), journalists and economists must take care to distinguish between intended effects on revenue and actual effects. Fact checkers can’t fact check the old estimates because they’re not facts. Estimates are just estimates.