Investors are hoping for yet another round of unconventional economic stimulus from the U.S. Federal Reserve. But odds are they will have to wait until late this fall for policy makers to make a decision.
It will take that long for the Fed – and the public – to have a clearer view of the state of the recovery. Why so long? The highly irregular pattern of the economy of the past few years, and the unusual weather of the past few months, are raising questions about the standard measurement of U.S. GDP.
You can see the problem reflected in the many false starts that the U.S. economy has appeared to experience over the past three years. According to the seasonally adjusted GDP data, growth in the fall and winter has been followed by a spring and summer lull. It is a strange pattern that needs to be explained before the Fed can decide on any further action.
I think that this odd up-and-down pattern in the economic figures is most likely related to quirks in the way that the GDP figures are adjusted for seasonal fluctuations. Statisticians typically use various mathematical techniques to remove the winter-to-summer variations that arise from holidays, weather and other recurring factors; the goal is to allow us to strip away recurring seasonal fluctuations and focus on the direction of the underlying economy. But the past few years have been so unusual that they have created problems in working with the data.
Statistics Iceland reached exactly this conclusion. In a report published in September, 2010, it states: "Massive and unexpected changes, e.g., an economic crash, can alter seasonally adjusted data. It is Statistics Iceland's view that the method for the seasonally adjusted GDP overestimated economic growth between third and fourth quarters of 2009 and thereby underestimated it between other quarters of the year."
To be sure, seasonal adjustments do not alter annual numbers, but they do affect how growth appears to be distributed through the year. Anyone watching quarterly GDP numbers can be misled by seasonal adjustments that skew the figures on the basis of highly unusual events in the recent past.
Consider what the economy has endured over the past four years. In 2008-09, nominal U.S. GDP declined for four consecutive quarters. One has to go back to the 1920s to witness a similar plunge.
And it doesn't end there. U.S. industrial production on an annual basis fell almost 16 per cent in the fourth quarter of 2008 and almost 20 per cent in the first quarter of 2009 – declines unheard of since the 1930s.
Seasonal adjustments have difficulty dealing with such extreme volatility and the significant disruptions to consumption and production that such dramatic events introduce to the economy.
The unusual weakness of late 2008 and early 2009 could be interpreted by standard statistical techniques as a change in seasonal patterns. That would result in adjustments for subsequent years that would push up economic statistics in the fourth quarter and the first quarter relative to historical norms while pushing down data for the second and third quarters.
The unusually warm weather of this past winter may have also caused additional problems with the seasonal adjustment, exaggerating the overshooting of the fourth and first quarters and the undershooting of the second and third.
If this conjecture is correct, as we enter this year's fall and winter months, the seasonally adjusted economic numbers may start coming in stronger than expected, rewarding the Fed's patience.
This is only a hypothesis, of course. But given the nature of the times, statistics departments may want to emphasize unadjusted data as well as the adjusted numbers. Economists should pay attention to both unadjusted and seasonally adjusted data and be increasingly skeptical about the inherent strength or weakness of any single indicator.
George Athanassakos is a professor of finance and holds the Ben Graham Chair in Value Investing at the Richard Ivey School of Business, Western University. Read more of his Globe and Mail articles here.