I have heard some people dismiss jobless claims as a data series for two distinct reasons that I want to flag given my view that claims data matter.
First, there is the view that the eligibility for jobless claims benefits have changed such that jobless claims numbers are lower now than they have been in decades past. So the 250 or 260,000 claims numbers we see today are misleading regarding the health of the job picture. That’s the first complaint.
Beyond that, there is a mistrust of seasonal adjustments to time series like jobless claims. And so, just because the claims numbers say one thing doesn’t mean that the real situation isn’t different.
I acknowledge both of these views has having validity but still believe we can use the claims data as an effective real time indicator. Here’s why.
On the first complaint, it’s not the level of claims that matters, it’s the change in claims over a discrete time period. So, what I am looking to see is whether claims data rises significantly enough over a short time horizon to negatively impact income or consumption. And unless jobless claim eligibility changes dramatically within that short window in time the complaint about eligibility is moot.
And when I look at the data, these days I always use year-over-year data to get rid of the impact of seasonality. And therefore, I could use seasonally-adjusted data or unadjusted data since the seasonality factors are almost identical from one year to the next, especially when using 4-week average data.
Why this matters: The reality then is that seasonality or eligibility have no impact on this analysis. What the numbers are telling us pretty unambiguously is that there has been no uptick in the number of people losing their jobs and filing for unemployment insurance. In fact, it’s just the opposite; claims were at record lows last year, but have fallen even lower this year. That speaks to a decent job picture, one that will keep the Fed raising rates, at least for now.