Post

Avatar
Quick thread about the labor market after continuing jobless claims rose much more than expected this morning. What are continuing jobless claims? These are the number of people who are currently receiving jobless benefits. This is hard, actual counts of people, not a sampled survey data set. 1/n
Avatar
This week we got data for the week ending June 22nd. Using Dept. of Labor seasonal adjustment, the number of people receiving UI benefits rose sharply to the highest level in almost three years after moving sideways for most of the past 12 months. 2/n
Avatar
One of the concerns in US macro data lately has been that the unemployment rate (based on large samples of the population each month) is rising. It's rising very slowly from extreme low levels, but historically rising unemployment tends to snowball and accelerate rapidly, leading to recession. 3/n
Avatar
If the continuing claims move (which is pretty rapid) is real, then this could be the labor market hitting a tipping point and the start of a recession. This data is very timely so it could be a leading indicator of the wheels coming off. Is it? A clue that it isn't can be found in seasonality. 4/n
Avatar
Seasonality refers to the regular annual pattern of a data set. In the case of UI benefits, the number of claimaints always peaks for the year right at the start of January and bottoms in the fall. Seasonal adjustment seeks to account for this pattern and make one period comparable to the next. 5/n
Avatar
The thing is seasonal adjustment can be misleading! If calendar-driven patterns change before the backward-looking seasonal adjustment algorithm can account for them, you'll get instances where data looks better / (worse) based on the part of the calendar not the actual underlying trend. 6/n
Avatar
The chart below attempts to show why this lag in SA might be creating some issues (description in alt text). As shown, 2022-2024 see a much steeper ramp in NSA continuing claims than was the norm pre-COVID. '21 (included in chart) and '20 (excluded) were very unusual due to COVID so ignore them 7/n
Avatar
If there is a "new" seasonal trend where continuing claims rise faster in this part of the calendar but that hasn't been captured in the seasonal adjustment algo, that would make seasonally adjusted claims look higher than they are. And we know what the SA factors are! 8/n