When the numbers mask good news
WASHINGTON -- bmckenna@globeandmail.com
Seasonally adjusted numbers are an economic staple. They help analysts to distinguish between changes that occur every year, and ones that have broader significance.
Retail sales and employment spike every Christmas. Employment and housing starts go up in the warm months and down in winter. People travel more in the summer.
To smooth out these distracting seasonal blips, data collectors, such as Statistics Canada or the U.S. Bureau of Labor Statistics, apply historical adjustments to the raw data. The result allows economists to compare one month to another, apples to apples, and to spot trends that transcend the seasons.
But what if one year isn't like another, and the past isn't a good predictor of the future? What if the adjustments are flawed?
Vancouver economist Doug Smyth says he's seen enough aberrations in recent U.S. data, including jobs and housing starts, to convince him that seasonal adjustments aren't reflecting the strong turnaround in the U.S. economy.
"Both the U.S. Bureau of Labor Statistics and the Census Bureau have had serious problems with their seasonal adjustment factors this year," Mr. Smyth said in a recent report to clients.
He suspects that wacky winter weather across much of the United States this year and the resurgence of the economy is screwing up the government's numbers.
Consider this: Between February and May, the U.S. economy created 929,000 jobs. The seasonably adjusted number - a combination of the widely reported monthly headline numbers - is pretty good.
But if you tally up the actual payroll numbers employers report monthly to the BLS, the picture is even better: a whopping 3.1 million.
So, has the economy created 929,000 jobs or 3.1 million?
Mr. Smyth says that the answer is a no-brainer. The 3.1 million jobs is the real world number - real people making real money, and spending it.
Unfortunately, too many analysts are convinced the economy is stalling because they're fixated on the seasonally adjusted numbers. In May, for example, private sector employment was up just 41,000, compared to a 411,000 increase in temporary workers hired to do the U.S. Census - government jobs.
Financial markets interpreted that as a disaster.
Quite the opposite is true, according to Mr. Smyth. The unadjusted numbers show that 1.1 million jobs were created in May, including 700,000 in the private sector.
"Those new incomes are actually present," he said, and that's vitally important to sectors such as the depressed housing market.
"Given the strong underlying trends in the economy and the job markets, employment levels will continue to improve through the end of the calendar year," Mr. Smyth said.
Even the BLS is having trouble figuring that out for sure. Already, in the first three months of this year, it has revised its seasonally adjusted number by upwards of 150,000.
In May, officials at the U.S. Mortgage Bankers Association seemed to throw up their hands in frustration over the mortgage default figures they collect. The rate of seasonally adjusted delinquencies went up in the first quarter, suggesting the housing crisis is getting worse. But the unadjusted rates went down.
The problem is that the MBA's model assumes that delinquencies follow a typical seasonal pattern. They peak in the fourth quarter and then fall in the first three months of the year. So the adjusted figure could be too high, casting an overly gloomy picture of the market.
"Simply put, fundamental market factors may be having a greater influence on the delinquency rate than is normally the case," acknowledged Jay Brinkmann, the MBA's chief economist.
Love 'em or hate 'em, economists can't live without seasonally adjusted numbers, insisted Peter Morici, an economist and business professor at the University of Maryland.
"Seasonality is seasonality," he said. They're used because, in the end, they are more meaningful, he added.
Indeed, for a quick, and often dirty, impression of what's going on in the economy, seasonally adjusted housing starts or jobs figures are okay.
But they can be a very poor marker to hang a trend on, as economists (and reporters) are often tempted to do.
