Jared Bernstein, former chief economist to US Vice President Joe Biden, has an op-ed on the Washington Post website purporting to refute supply-side economics, the school of thought that believes that lower taxes (and better taxes) means higher economic output overall. He plots a few charts that show that the top marginal tax rates of the USA in a year is uncorrelated (or even positively correlated in some cases) with investment growth, employment growth, productivity growth, growth in GDP/capita, family income growth, and tax revenue growth.
He makes astonishingly strong claims on the back of this data, but he is deeply confused and mistaken about the evidence he'd need to make the case he wants to make. What's more, there is a lot of rigorous empirical evidence against his argument. This evidence suggests that lower taxes do lead to more output being created; although at the current rates, tax cuts are unlikely to create so much more output that they 'pay for themselves' like some previous cuts.
Before explaining why the bulk of the evidence goes against Bernstein, we should ask why an economic model of the economy predicts that lower taxes means higher output, employment, productivity and so on. Nearly all taxes distort incentives—that is reduce the incentive to do productive and socially beneficial things. Taxes on consumption and labour income make leisure cheaper compared to market goods, so people take more leisure than they otherwise would. They also make career paths with higher hours, more delayed consumption, less pleasant conditions, and less social prestige, less attractive compared to more pleasant but less well-paid careers. Taxes on transactions gum up efficient allocation, reducing the incentive for older people to downsize in the housing market, and reducing the incentive for traders to buy when they think prices are away from reality. Taxes on investment returns reduce saving and investment, and increase current consumption, so there are less tools, training, communication and less efficient organisation in the future.
Bernstein's argument assumes that people ignore these incentives—but economists believe there is strong evidence both anecdotally and in the empirical academic literature that people respond to incentives, even when it comes to very serious personal decisions. For example, when a US state bans affirmative action policies, multiracial Americans are 30% less likely to self-identify as their minority ethnicity. American divorcees who had been married for 10-years are eligible for spousal Social Security benefits—divorces rise 20% around the 10-year mark.
A swathe of papers show that financial incentives drive retirement decisions—when benefits are more generous, people retire earlier. Immigrants will typically not leave their country unless their expected lifetime earnings are at least $500,000 more in their new home. They often return to their native country if their earnings expecations fall—as did a third of Polish immigrants in the UK. And travellers in Sierra Leone even pick between different modes of transport between Freetown and the airport—ferry, helicopter, hovercraft, and water taxi—based on a trade-off between mortality risk and cost. It would be very surprising then if people weren't partly affected by financial incentives when interacting in the market sphere. In fact, there is a consensus among economists that taxes have very large costs in terms of distorted activity.
Aren't Bernstein's graphs evidence against this? No. When you test an economic theory, you need to try and control for "confounders"—factors that you haven't measured that could affect your results. If you find a strong correlation between breastfeeding and child IQ, but you haven't controlled for parental IQ, then you don't know whether the breastfeeding itself is driving higher child IQ, or if those children had high IQ mothers, and would likely have had high IQs whether or not they'd been breastfed.
In the same way, the top marginal income tax rate is not the only thing going on in a year—for one thing, there are lots of other taxes in the economy, all of which could be high when income taxes are low, and vice versa. We don't even know how many people are paying this top rate—this will change between years. Supply-side economists predict that lowering the overall tax burden will improve economic outcomes, not that the top rate of one particular tax is the key issue. But supply-side economists also think that there are other very important factors. For example, the entire world grew very quickly in the 1940s, 50s and 60s, as we rebuilt after the second world war, trade links expanded, and new technologies filtered through economies. The US also had high top tax rates then—but with lower rates, the US might have grown even faster.
So one more rigorous way academic economists test theories about the macroeconomic effects of tax is by looking at the economy before and after tax changes. For example, James Cloyne's paper "Discretionary Tax Changes and the Macroeconomy: New Narrative Evidence from the United Kingdom" in the top economics journal, the American Economic Review finds that "a 1 percent cut in taxes increases GDP by 0.6 percent on impact and 2.5 percent over three years". A 2012 literature review from the Tax Foundation found only three of 26 empirical studies where higher taxes did not mean lower growth.
It would be very surprising if Jared Bernstein was right that taxes do not affect growth and other economic variables, because our simplest, most intuitively appealing, most empirically verified models predict large effects when incentives are distorted. But Jared Bernstein is wrong: his own tests are extremely simplistic, and do not attempt to account for confounding factors. Once you do, the evidence is clear: higher taxes mean lower growth. Of course, there are still reasons why we might want to tax—but it's a tough trade-off: the more government programmes we fund, the poorer we are on average.