Following the mainstream media reports on economic data is a waste of time for investors. It may have some value to traders, because the headlines give them an idea of what to fade. Most of the time any widely reported and widely accepted economic “fact” is almost certain to be wrong. Wizened traders have learned to do the opposite of what the headlines encourage the majority to do.
The incorrect fact of the week this week is that retail sales were weak in June. That misconception was a result of the silly Wall Street expectations game and the use of arbitrarily fabricated, fictitious numbers in the monthly report.
These fictitious numbers are known as “seasonally adjusted” (SA) data. SA data is an abstract impressionist view of reality which may or may not represent reality. It is designed to smooth the actual data for viewing by the unwashed masses. The process is completely haphazard because seasonal adjustment seeks to remove seasonal variation based on past patterns, which change and may include non-seasonal elements that appear seasonal. You may have known that, but you probably did not know that SA data also uses future data, which has yet to be reported.
The government agencies that report the SA data subsequently adjust the first release, what they call the “advance estimate,” multiple times in the months and years to follow as additional actual data becomes available. They fit the each month’s number to the actual trend for 5 years following the first release. Nobody reports those later revisions. They only report the advance release, with minor mention of the revision to the previous month.
The media consigns subsequent revisions to the dustbin of history, never reporting them. However, they publish charts of seasonally adjusted (SA) data which DO reflect these subsequent adjustments. It therefore appears that the SA graph reasonably represents the trend. This is only true of the older data, because it has been fit to the actual data over time. It is not true of the current release, where the reported SA data is little more than a wild guess based on an arbitrary factor derived from the past several years.
I’m a technical analyst. The mainstream practice of ignoring actual data to look solely at a highly idealized, complex moving average that purports to, but may or may not, represent reality, mystifies me. What possible justification is there for doing that? Put the actual data on a chart and draw a few trendlines, and voila, you see EXACTLY what is actually going on. It is not some abstract impressionist view that in the short run may be completely distorted and unreal, and which the artist will repaint 7 times over 5 years before completing the work and storing it away in the attic where nobody sees it.
There are easy ways to see exactly what retail sales are really doing. First, look at the actual, not seasonally adjusted data. Second, back out inflation in order to see the real trend of the volume of goods sold, not just price increases. Third, back out gas sales. They’re 10% or so of total retail sales and large changes in gas prices will send the total sales number in the opposite direction of sales volume, and that can skew total sales a bit.
For example, as gas prices collapsed in the second half of 2014, total dollar sales of gas fell sharply, reducing total retail sales. But in truth, as gas prices fell, consumers actually bought MORE gas and more stuff in general. The falling gas price suppressed the reported gains. When gas prices rebounded for a few months, the opposite was true. Lately gas price changes have been muted so that their effect on the top line has been minimal, but for the sake of the consistency of the measure and to view the effect of gasoline on total sales, I continue to run one chart including them and one excluding them.
Finally, I want to know whether retail sales are increasing relative to population growth. If population is growing by 1% a year, then 1% retail sales growth only reflects population growth. I want to know if there’s growth over and above that, so I divide the total sales by total population. This shows how fast sales are growing on a per capita basis. That does not mean that the growth is shared equally across the the population, just that we can see if sales are growing faster than population.
The reported fiction this month was that retail sales fell 0.3% in June against Wall Street economists consensus expectations of a 0.3% gain. What no one on Wall Street or in the media told you was that this was based on dividing actual, not seasonally adjusted (NSA) nominal retail sales by the seasonal adjustment factor of 1.008, thereby reducing the reported print below the actual sales. For June 2014, they divided by .994, thus increasing the print for June 2014. They used .996 in June 2013, increasing the print for that month. Why would they increase the number for the 2 prior years, then reduce it this year? Shouldn’t the seasonal adjustment be the same for the same month each year? Had the BEA applied the same seasonal factor this year as it applies to the NSA for 2014, the SA monthly change would have been +0.1%, not -0.3%. That’s still less than the guessing game expectation, but it’s not negative as the headlines reported.
Here are the facts. Nominal, not seasonally adjusted retail sales are virtually always down in June. The issue is whether the current month is stronger or weaker than the trend in a way that would indicate that the current trend is changing direction. This year the decline was $16 billion from May. That’s significantly stronger than the $25 billion drop in June of last year, and the $22 billion drop in June 2013. As Junes go, this one was pretty good. There was no deviation from the trend.
On a year to year basis, sales rose by 2.9%. This was the fastest growth rate since January. Contrary to the implication of the headlines, retail sales are on the same slow growth trend they have been on since 2012, a fact that is clearly visible when the actual data is plotted on a chart with monthly trend lines.[Click here to view chart if reading in email.]
Getting down to the nitty gritty of whether there’s growth beyond population growth and without gas sales the message is similar. Sales were down in June versus May as they virtually always are, but they fell less than in June 2014 and June 2013. The year to year growth rate was +4%. That’s in the upper half of the growth rate range of the past year. Here again there’s no sign of deviation from the trend.[Click here to view chart if reading in email.]
This is not to say that all is well in retail sales land. Jeff Snider at Alhambra Partners showed that the growth rate of retail sales ex autos has turned negative. That means that the growth in total sales has been driven by surging auto sales as other retail sales have slowed. People who can get credit, or have the marginal income or wealth to buy cars are spending more, a lot more. Some of those sales are due to the growth of questionable non-prime credit growth.
Likewise, we know that top line sales growth per capita is not due to the majority having growing incomes. Wage growth is slowing. Real median incomes are stagnant. Sales growth has been driven purely by the growing wealth at the top of the income spectrum. They are spending their stock market gains. Bernanke was right about that. It was probably the only thing he was right about. Where he was wrong is that there would be trickle down effects. Instead the wealth pooled at the top. The trend of rising retail sales can persist only as long as the stock market keeps rising. When that stops, the dam will break and the fictitious wealth created by central bank driven stock market gains will disappear. Retail spending will collapse, without any of the gains having reached the majority of people.[Click here to view chart if reading in email.]
Obviously, neither of these trends is good for the economy or sustainable retail sales growth. But the Fed will look at the top line numbers and see that the purported slowdown in retail sales is “transitory.” It will not be deterred from attempting to raise interest rates sooner rather than later.