This book meticulously documents common and widely circulated errors in reporting and interpreting economic data.
On 10 February 2005, the Economist reported that capitalists were enlarging their share of U.S. national income at the expense of workers. The highly regarded magazine based its conclusion on the fact that after-tax corporate profits stood at their highest level as a percentage of GDP in 75 years.
Upon closer examination, the economic data did not support the inference that corporations were fattening their profit margins by squeezing wages. Corporations’ profitability was not unusually high by historical standards. The corporate sector’s total output, however, had grown relative to the output of other sectors (i.e., households, government, not-for-profits, and unincorporated businesses). That factor made it mathematically possible for corporations’ profits to represent a record percentage of GDP , even though profits were only so-so as a percentage of corporate revenues.
Errors as elementary as this one are frighteningly common and widely circulated, as Gene Epstein meticulously documents in Econospinning: How to Read between the Lines When the Media Manipulate the Numbers. For example, in the best-selling book Nickel and Dimed, Barbara Ehrenreich claims that the officially defined poverty level is unrealistically low because it is crudely calculated by taking a bare-bones family food budget and multiplying by three.1Over time, Ehrenreich points out, adjusting the poverty threshold for inflation results in understatement because food prices rise more slowly than other household expenditures, such as rent. This would be a valid concern, says Epstein, except that the original “food cost times three” calculation was discarded in 1969 in favor of a method that reflects price rises in all goods and services. Ehrenreich would have learned of the change if she had read a source cited in her own book.
Errors in interpreting economic data arise for several reasons. One is the sheer underlying complexity of many statistical series. Journalists and even trained economists are prone to honest mistakes through unfamiliarity with quirks in compilation and computation of the numbers. Less innocent misrepresentations reflect politically inspired attempts to overstate or understate the vibrancy of prevailing economic conditions.
Other abuses of economic data result from inattention to questions of statistical significance. For example, market pundits frequently attribute importance to inconsequentially small monthly fluctuations in employment figures. Subsequent revisions routinely alter the magnitude or even direction of the initially reported changes. To reduce the noise in U.S. statistics, Epstein proposes that the U.S. Bureau of Labor Statistics place primary emphasis on the trailing-12-months series, which reliably reveals the true trend of employment.
Econospinning ranges beyond interim reporting of market-moving data. A determined debunker, Epstein details a statistical flaw in a research finding that gained wide attention through Steven Levitt and Stephen Dubner’s best-seller, Freakonomics.2 In a study published in 2001, John J. Donahue III and Steven Levitt claimed to have found evidence that legalization of abortion in 1973 contributed to reduced crime rates in the United States beginning in the 1990s.3 Epstein shows that, despite the decline in birth rates among women whose offspring would be highly prone to commit crimes, the crime-prone population actually increased as a share of the overall population. The women most likely to give birth to criminals had fewer babies, but that effect was offset by an increase in the number of such women.
Of particular interest to investors is a demonstration of the limits to accuracy in economic forecasting. In February of each year, the U.S. Federal Reserve Board publishes a forecast of the fourth-quarter year-over-year increase in GDP. Between 1994 and 2005, the board allowed itself a forecast range of 0.5–1.5 percentage points. The monetary chiefs potentially benefited from superior information about the economy, yet the actual GDP gain fell within their projected range only once during the 12-year span. In that case, moreover, the actual number was at the very bottom of the 1.25 percentage point estimated range.
Full disclosure compels me to note that Epstein provided a favorable comment for the dust jacket of my most recent book. He also praised the book in an article in Barron’s , for which he serves as economics editor. To allay any fears that I am consequently blinded to Econospinning’s imperfections, let me point out one: In a quoted passage in his preface, Epstein leaves uncorrected the misspelling “Council of Economic Advisors.” On a more material level, some practitioners might wish that instead of devoting 12 of his 17 chapters to the intricacies of employment data, the author had shed light on a few additional economic series. Epstein defends his allocation of space on the grounds that the monthly employment report is the biggest market mover among the ongoing economic statistical series.
In any case, trying to create a definitive guide to the pitfalls of relying on economic reports would have been quixotic. Over time, the reporting agencies revise their computation methods and thereby create new stumbling blocks. Instead of addressing every subtlety of inflation, industrial production, and interest rates, Econospinning heightens its readers’ awareness by intensely examining the misuse of selected data series. The book will make investors appropriately cautious about basing decisions on the headline numbers.