Just How Big Is the Service Sector?

Figuring out the size and growth of services is trickier than it looks
Photograph by Matt Mawson/Getty Images

The U.S. government thoroughly measures the output of American farms. On July 18, for example, the Agriculture Department announced that milk production per cow averaged 1,888 pounds in June. The government collects less information about services, which range from $10 manicures to $1,000-an-hour legal advice. That’s a problem, because services account for 77 percent of private-sector output in the U.S. There are only six principal economic indicators for services (including ones for construction and retail trade), vs. 17 for agriculture, construction, manufacturing, mining, and energy. “The economic data collected by U.S. statistical agencies remain highly skewed toward more traditional industries,” writes economist Kris Dawsey, author of a new Goldman Sachs report.

The lack of information about the service sector mainly affects the government’s first two stabs at calculating gross domestic product each quarter. By the third try the government’s calculation is more accurate, because it has more data to work with—in particular the Quarterly Services Survey, which is prepared by the Census Bureau. Those squishy early estimates matter because they can influence public opinion and potentially even affect the decisions of the Federal Reserve and Congress. This year the Bureau of Economic Analysis, a branch of the Commerce Department, initially estimated that health-care usage rose at a 9.9 percent annual rate in the first three months of 2014. It later concluded that it declined at a 1.4 percent annual rate—a huge swing by any measure.

While waiting for the hard data from the Quarterly Services Survey to arrive, the Bureau of Economic Analysis might do a better job at its early estimates if it tapped into data it now ignores. Goldman’s Dawsey wrote that surveys of business executives are an accurate—and early—indicator of service sector activity. He took six existing monthly business surveys, all of which are conducted by other organizations, and synthesized them into a single “service sector survey tracker.” Dawsey found that the tracker is a better predictor of what the government eventually will report than the government’s own first and second estimates.

The six monthly surveys that Goldman rolled into one are the Institute for Supply Management Non-manufacturing Index, the Chicago Purchasing Managers Index, the Texas Service Sector Outlook Survey, the Markit Purchasing Managers Index, the Federal Reserve Bank of Richmond Survey of Service Sector Activity, and the New York Fed Business Leaders Survey. Right now those surveys indicate that the economy is stronger than the latest report on GDP shows.

In an interview, Brent Moulton, the BEA’s associate director for national economic accounts, said that “it is interesting that there’s some data that might potentially be helpful to us that we’re not currently using.” But he added that his first impression from looking at the Goldman report is that incorporating the business surveys would produce “relatively small improvements relative to what we’re doing.”

Moulton and his team have some other ideas about how to improve their early readings on the service economy. The BEA is talking to the Census Bureau about getting earlier delivery of the Quarterly Services Survey, says Moulton. There have also been conversations inside the BEA about incorporating Big Data, such as credit card receipts, says Nicole Mayerhauser, chief of the BEA’s National Income and Wealth Division. Meanwhile, you can get a pretty good early read on the service sector by looking at those business executive surveys.

Before it's here, it's on the Bloomberg Terminal.