January Gymnastics: How Worrying?

??

‘Tis the Seasonal

Seasonally adjustment is essential for spotting turns in the economy. Year-over-year growth rates can help remove seasonal effects, but by mushing together 12 months of data, they lag properly adjustment monthly or quarterly growth numbers. The hardest part of the year to separate out seasonal effects is in the winter, particularly in January. This is because bad weather is bad for the economy and because of the massive swing in holiday-related activity. One month does not make a trend, particularly if that month is January.

?

The seasonal effects are much bigger for real data than for price indexes. Typically, the CPI tends to be about 0.7% higher than normal in December and 0.4% above normal in January. That is a small seasonal compared to housing starts that typically are about 15% below average both months. Employment and retail sales are also much more seasonal than prices.

?

Even more important, unusually bad weather has a huge impact on seasonally-adjusted real activity and a very small impact on seasonally-adjusted prices. If the weather is cold and stormy, home foundations aren’t dug, people forgo discretionary shopping, and mining activity stops. However, bad weather does not cause businesses to revamp their prices.

?

This is why January growth indicators should be buried in a bucket full of de-icing salt. In January housing starts plunged 15%, retail sales slipped 0.9%, and industrial production dipped 0.1%, all much weaker than expected. However, bad weather likely played a significant role. Let’s see how the three perform over the whole season—say from November to February.

?

The industrial production data illustrates the impact of winter weather. On the one hand, bad weather likely caused the 5.8% surge in utilities; on the other hand, it probably caused the 0.5% drop in manufacturing output and the 2.3% decline in mining activity.

?

By contrast employment was surprisingly strong in January. I’m inclined to put more weight on that. It is possible that there is something weird about the number, but it follows similarly strong reading in December. Other parts of the report appear less reliable—the work week fell to 34.1, breaking out of 34.3 to 34.4 range of recent months. This likely contributed to the large 0.6% increase in hourly earnings—if hours were down and some of the workers still got the same paycheck that boosts the hourly average. Again, let’s see what happens this month.

?

The inflation data is also erratic from one month to the next but is much less sensitive to weather. What is worrisome about the inflation data is that the strong January numbers—for both the CPI and PPI-- are not just a one-month story. Since early last fall there have been signs that underlying inflation is getting stuck above target.

?

The Fed, and hence the financial press, are putting a lot of focus on annualized six-month growth rates as a relatively timely indicator of current inflation. On that basis, the trimmed mean CPI bottomed at 3.1% in August and is now running at 3.8%. The median CPI bottomed at 4.1% in October and is now 4.8%. January data confirmed a pre-existing problem.

?

Looking ahead, the PCE is released on February 29th. The good news is that underlying inflation has been better behaved in the PCE. Six-month growth has tended to level off rather than re-accelerate. The bad news is that the components of the CPI and the PPI deflator that go into the PCE were strong in January. Stephen Juneau at BofA does a nice job of mimicking the construction of the PCE deflator and he argues for a 0.4% increase in the traditional core. My guess is that the trimmed mean and median will be as strong or stronger.

?

When in doubt, wait.

Powell’s decision to largely rule out a March cut is looking even smarter in hindsight. The Fed will not cut until it gets clear confirmation of both weaker growth and a resumed drop in underlying inflation.

?

?