On Thursday, the Census Bureau's report on May retail sales said that seasonally adjusted sales came in 1.2 percent higher than April. The press almost universally cited that result as demonstrating that the economy's rough patch earlier this year is likely over.
Yours truly and the contrarians at Zero Hedge both noted that the result is highly suspect, and doesn't adequately reflect the raw data behind it. The business press won't question it, because it hardly ever bothers to look at the raw data.
At the Associated Press, Josh Boak called the result "a sign that strong job growth has begun to boost store sales." He also cited the Census Bureau's other questionable statistic, that "over the past 12 months, sales have risen a solid 2.7 percent."
Bloomberg's Victoria Stilwell was pleased to report that "The American consumer, missing in action for much of 2015, showed up in May," and took the result as "a sign that employment and income gains are allowing households to buy more than just automobiles."
At Reuters, Lucia Mutikani said that it was "the latest sign economic growth is finally gathering steam."
A look at the past several years of sales will demonstrate why the seasonally adjusted result should be taken with a large measure of salt:
As I wrote at my home blog on Thursday afternoon, it's more than a little difficult to understand how "the smallest actual April to May percentage increase in the past four years somehow generated the BEST seasonally adjusted increase — by miles." An absolutely flat seasonally adjusted result would have been perfectly defensible.
The out-of-the-box seasonally adjusted result also distorts the year-over-comparison. As the AP's Boak reported, it's 2.7 percent as adjusted ($444.926 billion divided by $433.421 billion), even though the raw increase was just under 1 percent ($463.124 divided by $458.705).
A graph posted at Zero Hedge on Friday demonstrates how much of an outlier this year's seasonal conversion was:
As I've writen several times over the years, the press shouldn't be treating the seasonally adjusted numbers as the gospel truth without also looking under the hood at the raw data to see if there's anything unusual. But they never do. In this case (though they're just as negligent when things go in the other direction), it's making the situation appear much better than it really is.
Cross-posted at BizzyBlog.com.
