The retail sales report for the month of May 2018 was generally positive on the headline, which is to say it was one of the better months of recent years. Total sales growth was 6.35% year-over-year (unadjusted). A big part of it was due to an 18% year-over-year gain in retail sales at gasoline stations.
The problem remains calibration, meaning that 6% shouldn’t be an outlier to the high side as it has been consistently going back to 2012. The last two months are illustrative of this deficiency. While the month of May was one of the better months, it follows retail sales in April where growth was less than 4% again. All we can count on is the unevenness.
It is still far too much like 2014. Even the hurricane boost hasn’t really changed matters. I wrote in August 2014 about the similar boost provided by the end of the Polar Vortex. Then, as now, it was more of an amplification in commentary than anything real in the underlying fundamentals. It is just as relevant now.
The bounce in April and beyond (becoming undone more so by June) simply moved retail sales (of whatever slicing) up closer to the 2013 trend – doing nothing like inaugurating a new trend that would be typical of recovery. A true recovery is something like 6% – 10% growth month after month after month. In a true growth period, 6% is actually the down months; now we are lucky to see 4% and only every once in a while.
If there is a major difference in retail sales then to now, it’s what is causing the tendency toward reluctant growth. Four years ago, auto sales were booming while gasoline sales were not, the latter about to become much, much worse as the overall economy skidded toward its 2015 downturn.
This year, auto sales are the drag while gasoline prices provide the upward emphasis as the hurricane effects fall further into the past (though still remain in the year-over-year numbers). It’s actually unsychronized growth.
The 6-month average for retail sales growth in autos is back down to 3.5%, matching the lows of 2016. That’s a big headwind for auto production (as the Industrial Production figures will show) as well as economic projections for a lasting acceleration and boom.
Historically speaking, rising gas prices in the absence of any offsetting increase in wages and worker pay tend to be highly depressive beyond some threshold. We can’t know exactly where the point of maximum pain might be for consumers, but it’s not a good combination.
Aggregate income estimates suggest a collision where income growth has stalled in the wake of the 2015-16 downturn but spending hasn’t been dampened by nearly as much. The result has been the several years now of languishing auto sales (and production) as well as the parallel decline in the Personal Savings Rate. In other words, where gasoline prices might trigger a negative reaction in further spending could be at a lower point. Consumers just don’t have that much room or discretion.
Even with this “good” month of May, retail sales are still rising at a less than 4% annual rate. Going back to last December, retail sales for a half year have increased by just 1.9% total (seasonally adjusted). Therefore, of that +6% yearly increase, more than one-third of it was produced during the hurricane months while another third is the last two months where gasoline spending has jumped on price appreciation.
The headline is good but what makes it seem good might not be.
The May report also brings up benchmark revisions. There wasn’t much to them this time except that the 2018 benchmark produced a downward bias in retail sales reversing the upward changes found in the 2017 version. Most of that emphasis was in the estimates for 2016.
For one example, the 2017 benchmark estimate for the month of December 2016 was $470.996 billion, practically the same as under the 2016 benchmark. The new figure for that same month is instead $467.928 billion.
These may seem like small numbers and in a narrow sense of individual months and individual benchmarks they are. A $3 billion difference out of nearly half a trillion in retail sales is a rounding error.
But it’s not just a single month nor is it a single benchmark. The difference is cumulative. Though revisions in retail sales don’t make for the dramatic alterations on the charts like they have for durable goods or especially industrial production, the fact of the matter is that the Census Bureau time and time again has to mark down economic progress. And there isn’t much for them to work with.
Not only does it speak to bias toward recovery, though a small one here in retail sales, it also raises more substantial issues about interpretations. The current retail sales report, for one, may seem good in its headline growth number today, but what will it look like in two or three years’ time?
The looming benchmark revisions next year will start to incorporate the 2017 Economic Census Results. If they are anything like the 2012 Economic Census, and I have to suspect they are (the difference between bias in the regular samples versus the correction to that bias by the broadest sample which back then showed much less economic recovery than all the statistics were projecting by their biases), we can’t have much confidence in the current figures especially when they may appear on the high side.
That’s where these benchmark changes are important in any of these data series. They continue to show a relatively consistent upward bias toward even minimal growth and recovery that is only revealed at a much later date. Though contemporary commentary never picks up on it, nor ever mentions the revisions and what they might mean, the net result is that no matter what is written and said the actual trend is further from recovery and health the more time passes.
It’s not quite the booming economy that is being written about yesterday and today. It does, however, speak to the seeds of hysteria (bias).