Global: The Job-Quality Debate
______________________________
Stephen S. Roach (New York)
Morgan Stanley
Jul 22, 2004
Not surprisingly, the jobs debate is heating up in the United States as we move into the heart of the political season. My interest in this issue bears more on the economic and financial market implications of underlying trends in the US labor market. A jobless recovery puts pressure on consumer purchasing power and challenges the sustainability of an economic upturn. It also forces income-short consumers to rely on riskier sources of support—namely, outsize tax cuts, which blow up the budget deficit, and equity extraction from homes, which pushes debt loads into uncharted territory. By contrast, in a hiring-led recovery, the economy draws support from internal, or organic, growth—thereby avoiding the imbalances and other stresses and strains that have long concerned me.
Most have drawn great comfort from recently improved trends on the US hiring front. With jobs on nonfarm payrolls up 1.024 million over the past four months (March to June 2004), it is tempting to conclude that the long drought of the jobless recovery is over. I have argued to the contrary—maintaining that the recent improvement on the hiring front is skewed decidedly to the low end of the quality spectrum (see my July 9 dispatch in the Global Economic Forum, “America’s Job-Quality Trap,” and my Op-Ed piece in the July 22 edition of the New York Times, “More Jobs, Worse Work”). If I am correct, that means the recovery remains in precarious shape.
Several have challenged this conclusion, slicing and dicing the employment data with a different set of tools. The first such effort showed up in the form of a July 9 article posted on www.factcheck.org, a Website sponsored by the non-partisan Annenberg Public Policy Center of the University of Pennsylvania. In the interest of full disclosure, I should note that several senior government officials have brought this article to my attention—implicitly urging me to rethink my conclusions. A recent article in Business Week has elevated the debate over job quality to the national stage. The following letter to the editor of that publication is a direct response to this critique:
The Editor
Business Week
Dear Sir:
In “Another Look at Those Jobs Numbers” (Business Week, July 26, 2004), Peter Coy takes issue with my conclusion that the recent upturn in hiring has been concentrated at the low end of the quality spectrum. However, there is a serious shortcoming to his analysis: It is based on unpublished data taken from the government’s smaller and less reliable survey of US households—the so-called Current Population Survey. My conclusions are based on the far more comprehensive and accurate survey of workers on payrolls in business establishments—the Current Employment Statistics Survey.
There is really no comparison in the sampling accuracy of these two surveys. According to the US Bureau of Labor Statistics, the “active sample” of some 400,000 establishments in the payroll data covers about one-third of the total universe of such workers. By contrast, the monthly sample of only 60,000 households covers just 0.06% of the universe of more than 106 million households in the United States. There is no doubt in my mind as to which of these two surveys should be trusted more. Other experts, including Federal Reserve Chairman Alan Greenspan, have come to similar conclusions.
Moreover, it is a real stretch to glean such finely calibrated insights—by industry as well as by occupation—from this tiny survey of households. The data are unpublished for good reason—much of the granular detail in the 154 job categories that Mr. Coy has analyzed is simply not statistically significant. If you divide 60,000 households into 154 cells, then each cell contains an average of only slightly less than 400 households—a tiny sample for an economy the size of the US. Moreover, to the extent that some categories are populated by a greater-than-average number of households, that leaves a miniscule representation in the remaining cells. In my view, that seriously compromises the integrity of the statistical analysis. That's why I gave up years ago using the household survey to do industry analysis—I was getting nonsensical results. I am afraid that is still the case. In other words, the household data should not be trusted as the ultimate arbiter of job quality.
Nor is this finely calibrated detail of the household survey reliable enough to seasonally adjust. That further complicates an accurate analysis of underlying trends in the US labor market. This latter shortcoming should not be minimized. Lacking in seasonally adjusted data, Mr. Coy examines trends on a year-over-year or 12-month trailing basis. That misses a key aspect of this debate—a decomposition of job trends over the past four months, March to June 2004.
Overall, the main problem with using the household survey for detailed industry analysis is that it was never designed for that purpose. It was mainly intended to get a good sample of unemployment. The industry and occupational detail that can be gleaned from the household survey is largely an after-thought—a statistical by-product of how individuals view their role in the business sector. That stands in sharp contrast with how companies may see it. While nothing is ever perfect in the realm of statistical analysis, my strong recommendation in attempting to measure job quality is to stay with the establishment survey.
For those reasons, I stand by my own detailed analysis based on the far more reliable survey of business establishments. By industry, restaurants, temporary hiring agencies, and building services were the leading sources of hiring over the past four months. Accounting for only 9.7% of total nonfarm payrolls, these three low-quality segments of the US work force contributed fully 25% to the cumulative growth in overall hiring from March to June 2004. Disproprtionaely large hiring contributions also occurred in other industries at the low end of the job hierarchy—namely, clothing stores, couriers, hotels, grocery stores, trucking, hospitals, social work, business support, and personal and laundry services. All in all, lower-end industries, which employ 22% of the work force, have accounted for 44% of new hiring—or twice their fair share—over the past four months.
The US Bureau of Labor Statistics does an excellent job in providing a rich body of data that can be used to analyze trends in America’s labor market. But like all statistics, the data must be used with care. In contrast to the headline in the Business Week article on jobs creation, the BLS has not put any spin on newly revealed employemt detail by recommending one conclusion over another. That task, according to Mr. Coy’s reporting, apparently has fallen to the White House’s Office of Management and Budget. Now why would they be interested in painting a rosy picture of one of America’s toughest problems?
Sincerely,
Stephen S. Roach
Chief Economist
Morgan Stanley
morganstanley.com |
