This is the final installment of my three-part review of Alan Reynolds's excellent new book, Income and Wealth. In the first installment, I laid out some of Reynolds's criticisms of the view that the vast majority of families' incomes have not increased much in the last few decades. Reynolds's evidence showed that they have, in fact, increased. In the second installment, I laid out Reynolds's main evidence against what I called "the real-wage pessimists," those who believe that the vast majority of U.S. workers have had little or no growth in real wages in the last few decades. This installment highlights his evidence on the claim that the top 1% has garnered the lion's share of income gains in the last two decades and that CEO pay is about 500 to 1,000 times the pay of the average worker.
The Top 1%
The seminal study showing the percent of income going to the top 1% is a National Bureau of Economic Research (NBER) study by economists Emmanual Saez of the University of California, Berkeley and Thomas Piketty of École normale supérieure in Paris. A version was later published as "Income Inequality in the United States" [pdf] in Harvard's Quarterly Journal of Economics. Their data showed that the share of income going to the top 1% of taxpayers (including capital gains), after being as low as 8 to 9% in the 1960s and 1970s, rose from 9.3% in 1980 to 16.4% in 1998.While many commentators in the popular press have reported the Piketty/Saez results uncritically, Reynolds dug into their data and discovered some huge problems, not mainly with the data themselves, but with other people's and, to some exent, Piketty's and Saez's interpretation of the data.
First, notes Reynolds, the Piketty/Saez data are on "tax units," not on individuals or families, a fact that Piketty and Saez admit. As I noted in my first installment, if a married couple filing jointly has an older child living in the couple's household but earning income, as my daughter did while in high school, that family will count as two tax units, not one. This, write Piketty and Saez, is not a problem because: "The average number of individuals per tax unit decreased over the century but this decrease was roughly uniform across income groups." (p. 4) Yet, after reading through their whole article, I could find neither data on this nor a citation to back up their assertion.
Second, notes Reynolds, a large part of the increase in the share of the top 1% of "tax units" occurred in 1986. This was understandable, Reynolds writes, because the 1986 Tax Reform Act raised the top tax rate on capital gains from 20% to 28%, effective in 1987, giving people with substantial unrealized capital gains a strong incentive to realize those gains and pay taxes on them in 1986. Because the changes in the tax law became widely known in August 1986, people had more than 3 months to cash in their capital gains. And, of course, a disproportionately high share of capital gains is earned by people in the top income brackets. Sure enough, Pikkety's and Saez's own data support this interpretation. They show that between 1985 and 1986, the income share of people in the top 1%, excluding capital gains, barely wiggled, rising from 9.09% to 9.13%. But the share of the top 1%, including capital gains, rose from 11.24% to 13.40%, an increase exceeding 2 percentage points.
Of course, Reynolds's interpretation explains why the share of the top 1%, including capital gains, rose in 1986 and fell in 1987 (from 13.40% in 1986 to 11.84% in 1987.) But why would the share of the top 1% rise after 1987 and stay high? Pikkety and Saez attribute it to a relatively sudden increase in earned income of what they call the "working rich." Reynolds has an alternate explanation that goes back to a separate provision of the 1986 Tax Reform Act—the dramatic changes in the top tax rates on corporate income and individual income. Before 1986, the top tax rate on corporate income was 46 percent, but the top income tax rate was 50% (and had been 70% as recently as 1980.) The 1986 law reduced the top tax rate on corporate income to 34%, but reduced the top income tax rate even more dramatically, from 50% to 28%. This, notes Reynolds, gave people who owned corporations an incentive to shift from C corporations, which paid corporate income tax, to S corporations, which paid individual income taxes. The shift in tax rates under the 1986 law was complete by 1988. With less income being taxed by the corporate income tax system and more falling under individual income taxes—and with a disproportionately high share of the people who had shifted to S corporations being in the highest-income groups—this could account for a few-percentage-point shift in the share of the top 1%. But, as Reynolds writes, this "was simply a bookkeeping change, having nothing to do with the actual distribution of income."
But, if the 1986 tax law is the main explanation for the increasing share of the top 1%, why would that share continue to rise well after 1988, when the new tax law was completely in force, instead of leveling off at a higher level? Reynolds points to other changes in the tax law that could have caused this shift to S corporations to continue over a number of years. He writes, "Rules governing Subchapter-S originally allowed only up to thirty-five stockholders but their number was expanded in stages to 100." Moreover, he notes, a 1996 law allowed banks to file as Subchapter-S corporations and, by December 2003, there were over 2000 such banks, with the largest at $9 billion (he doesn't specify whether he means income or assets, a big difference.) So income of "tax units" after 1987 included more and more "units" that had previously filed under the corporate tax system.
Reynolds describes a number of other factors that can help account for the increasing share of the top 1% of tax units. I'll highlight two. One is that before 1987, there was no requirement to report income from municipal bonds on tax forms and, from 1987 on, this income had to be reported. Because such income is not taxed at the federal level, municipal bonds, which pay low interest rates to reflect this preferential tax treatment, are attractive only to people paying the highest marginal tax rates—that is, the highest-income people. So, some of the increase in the share of the top 1% simply reflects a change in the law on reporting income, not a change in income. The second is that the income share of people in the lower income categories is understated because much of our income is 401(k) and 403(b) income and, therefore, does not show up on our tax forms.
What about the claim—one that has become widespread—that the incomes of the top CEOs are 500 to 1,000 times the incomes of average workers? Reynolds is at his most devastating in taking this claim apart. Here he is at his best:
On January 21, 2006, a Wall Street Journal feature story said, "In 2004, the aggregate compensation of the chief executive officers of the S&P 500 companies . . . totaled about $5 billion, up 39% from 2003, according to Paul Hodgson of the Corporate Library." If you divide $5 billion among 500 CEOs, that is $10 million apiece. But the alleged 39 percent increase in 2004 implies that the 500 CEOs received just $7.2 million apiece in 2003. Meanwhile, a graph in the same Wall Street Journal story shows average CEO compensation as only $2.16 million in 2005. At that point, most readers must have been as befuddled as the writer. But that is not the end of the confusion.
A sidebar to the same report claimed that "the average CEO's salary in the U.S. is 475 times greater than the average worker's salary." If average CEO pay was $2.16 million, then a ratio of 475:1 implies average workers earn only $4,547 a year. If average CEO pay was $7.2 million in 2003 then a ratio of 475:1 implies "the average worker's salary" was only $15,158 (about $7.29 an hour), which is much less than half of any credible average. If the wildly different estimates on CEO pay were even close to being "average," then the ratio of 475:1 makes no sense.
New York Times columnist and Princeton economist Paul Krugman wrote an article in which he claimed that the 100 highest-paid CEOs in the Fortune 500 in 1999 made 1,000 times the pay of ordinary workers. Where did Krugman get the $37.5 million average CEO pay that led him to this conclusion? He credulously accepted Fortune magazine's estimates, even though, according to Reynolds, at least three quarters of this $37.5 million estimate was for stock options granted in 1999. Fortune valued these options by assuming that the prices of the stocks would rise by one third. Of course, the March 2000 stock market crash belied this assumption. Yet Krugman wrote his piece in October 2002, well after this crash and well before the stock market recovery, and he didn't bother revaluing the options.
Krugman was not the only one who ignored the stock market crash. Ben Stein, writing in 2006, asserted, "the graph for the pay of CEOs is a vertical line in the last five years." In fact, notes Reynolds, CEO pay for the S&P 500 fell by almost 48 percent from 2000 to 2003. Imagine that: CEO pay falling when a huge proportion of it is in stocks and stock options, and the stock market has fallen. Who'd have thunk it?
There's so much more in Reynolds's chapter on CEO pay and in the other chapters, such as the one on mobility across income groups, that I don't have room for. My best advice: read the book, and mark it up while doing so.
David R. Henderson, is a research fellow with the Hoover Institution and an economics professor at the Graduate School of Business and Public Policy, Naval Postgraduate School in Monterey, Calif. He is author of The Joy of Freedom: An Economist's Odyssey and co-author, with Charles L. Hooper, of Making Great Decisions in Business and Life (Chicago Park Press, 2006).