Statistical Politics Update

My recent column predicted that "the closer we get to elections, the worse economic reporting becomes." To demonstrate how well that theory is working, we need only look at two samples of "news" from one newspaper on a single day, Oct. 15.

On that day, New York Times columnist Eduardo Porter asked "After Years of Growth, What About Workers’ Share?" A table showed changes in wages and benefits as a share of GDP from 2000 to 2005 in nine countries. The author concludes, predictably, that "workers in the United States appeared to get a break for a few years in the second half of the 1990s," but "the share of the economy devoted to wages and benefits has eroded in the United States over the last five years."

In my original column, critiquing another New York Times writer for using this same statistical trick, I explained that "serious economists never compare labor’s income shares to gross domestic product. GDP includes big items that are not any American’s income — notably, depreciation for wear and tear on everything from computers to highways (which rose from 11.9 percent of GDP in 1999 to 12.9 percent in 2005). The sensible practice is to examine labor compensation as a share of national income."

When wages and benefits are properly compared to national income, Porter’s conclusions evaporate. Wages and benefits amounted to a slightly subpar 65 percent of national income in 2005, yet that was identical to the figure for 1999, when workers supposedly got a big break. Wages and benefits were an unusually low 64.2 percent of national income in 1995, which is the only reason Porter is able to show an increase between 1995 and 2000. In fact, real hourly compensation fell in 1993, 1994 and 1995, but rose by 1.1 percent a year from 2001 to 2005.

Leaving aside self-employment (as Porter does), labor compensation was 65.56 percent of national income from 1990 to 1999, 65.75 percent in 2000 and 65.55 percent from 2001 to 2005. The New York Times cannot possibly make political news out of statistics that barely change, so they simply replaced good sense with bad numbers.

Porter even echoed the literally unbelievable hoax that, "in real terms, the wages of non-management employees are now 10 percent below their level in the early 1970s." I called this the "wage stagnation thesis" in my new book, "Income and Wealth," and devoted a chapter to burying it.

The statistic Porter refers to, "real average gross hourly earnings," never purported to measure typical wage rates. He neglects to mention that this dubious figure was a bit higher in 2005 than in 2000, despite the oil price spike, since that contradicts his insinuations about wages falling since 2000.

The fanciful claim that everyone except management is now paid less than in the early 1970s is contradicted by every other measure of living standards. The index of real hourly compensation, for example, has increased from 83.8 in 1974 to 123.1 — a gain of 47 percent.

The flip side of Porter’s strained story about capitalist exploitation of the proletariat was dutifully covered on the same day by another New York Times columnist, Gretchen Morgenstern. The latest in her endless series about the stock market gains of corporate executives featured a foot-wide graph — "The Unstoppable March of Executive Pay." Actually, that graph shows top executive pay marching downward for three years after 2000, before partly recovering along with the stock market in 2004.

Estimates by other critics of CEO pay show even steeper declines. Thomas Piketty and Emmanuel Saez gathered CEO pay from the top 100 in Forbes, and that figure fell by 54 percent from 2000 to 2003. Lucian Bebchuck and Yaniv Grinstein estimated that among the S&P 500 firms, average CEO pay fell 48 percent from 2000 to 2003. An unstoppable march?

Morgenstern tries to blame the increases (but not the declines) on compensation consultant Frederic Cook, who has been in the business since 1973. If that is to be believed, nobody listened to Cook until stocks began to rise in the Reagan years. After that, he must have advised corporations to increase executive pay when stocks soared, and to cut pay when stocks fell. Yet that is exactly what happens automatically when pay consists mainly of stock or options, rather than salary and perks.

Using the same small sample of executive salaries Morgenstern relied on, Xavier Gabaix of MIT and Augustin Landier of New York University found that "the six-fold increase of CEO pay between 1980 and 2003 can be fully attributed to the six-fold increase in market capitalization of large U.S. companies during that period." CEO pay rises and falls with the global value of U.S. companies. That is what "pay for performance" means. Stockholders foot the bill for stock-based CEO pay, not workers or consumers, and stockholders generally like it when the CEO makes money only if they do, too.

Every day from now until Nov. 7 is bound to bring additional evidence to support my thesis that the quality of economic reporting worsens with each passing day as we get closer to the election. If journalistic partisans must resort to statistical cheating to make their case, how can we trust the political party they seem so eager to represent?