That news about lower unemployment claims? Watch that reverse in the next 2 weeks

You may have read in the news today that unemployment claims are back on the decline.

The number of Americans who applied for unemployment benefits at the end of November fell to the lowest level in seven months and returned close to a 50-year low, but the sharp decline in jobless claims likely stems in part from the Thanksgiving holiday.

Initial jobless claims, a rough way to measure layoffs, dropped 10,000 to 203,000 in the seven days ended Nov. 30, the government said Thursday.

That’s the lowest level since mid-April, when new claims fell to a 50-year low of 193,000, and a sign that the U.S. labor market remains rock solid.

I have my suspicions about whether fewer people are really losing their jobs, and not just because ALEC states are making people jump through increasing hoops to get their unemployment benefits.

The words being left out of that story are “seasonally-adjusted, and they seem especially relevant here, if you dig into the numbers that the US Department of Labor has put out (which you can see if you run a search at this web address).

Let’s look at the week before Thanksgiving and the one including Turkey Day over the last 2 years, using the raw numbers to begin with.

New unemployment claims
Not Seasonally Adjusted
Week before Thanx 2018 226,576
Week of Thanx 2018 218,658
Week before Thanx 2019 252,413 (+25,837)
Week of Thanx 2019 216,737 (-1,921)

Given that this happened at the same time of year, we’d logically think that we’d see a seasonally-adjusted increase for 2019 in the week before Thanksgiving, and the two years being near the same for the week of Thanksgiving.

But we don’t.

New unemployment claims
Seasonally Adjusted
Week before Thanx 2018 224,000
Week of Thanx 2018 234,658
Week before Thanx 2019 213,000 (-11,658)
Week of Thanx 2019 203,000 (-31,658)

So what gives? It’s the seasonal adjustment factor that the DOL does for each month. For whatever reason, that factor did little to 2018’s figures (in fact, it raised the “seasonally adjusted” total on Thanksgiving week), while it deflated the numbers in both weeks in 2019.

That doesn’t necessarily mean that there is any kind of nefariousness going on in DC, because 2019 won’t be as deflated as 2018 for the next 3 weeks. In this chart, 100= no change, and a number over 100 deflates the non-seasonal totals to give the seasonally-adjusted totals that are commonly reported in the news.

As you can see, 2018 has more deflation for the 1st 2 weeks of December than 2019 does. The upshot of that is if we see the same amount of unemployment claims in real life for this week as we did for the 1st week of December in 2018 (aka, not seasonally adjusted), then we would see unemployment claims jump by 50,000 in the next report.

So stay tuned on that, because we could see the same economic media that is trumpeting a "strong jobs market" writing articles about "signs of recession" due to an alleged jump in claims over the next two weeks. And it might showcase nothing more than a seasonal adjustment that wrongly modified for a late Thanksgiving vs an early one.