The Labor Department publishes Initial Jobless Claims each week, which tells us how many people filed for unemployment the week before who were not on the unemployment rolls the week before that.
That number always struck me as badly out of context - you'll sometimes read something like "...a number under 350,000 is consistent with a strong labor market...", but that doesn't tell me what I want to know, either. What I want to know is this: what proportion of the people employed last week are unemployed this week, and how does that look in historical context?
Thanks to economagic.com (highly recommended site for raw data, by the way), I have what I wanted.
I've taken the average weekly number of initial jobless claims reported each month since 1967 and divided that by the total number of people employed in that month (nonfarm payrolls). Below I've charted what I think I'll call the "Job Insecurity Index", or JII. Probably already trademarked, but it'll do until someone threatens a lawsuit.
Three things are clear. First, the seventies and early eighties were a lousy time for a lot of people. The JII peaked in September of 1982 at .00741 - 3.5 times what it is today, and from 1974 to 1984 things were consistently worse than they were at the height of the most recent recession. That's right, every single month, for ten years, was worse than October of 2001, the worst month of the 2000-2001 downswing.
Second, recent recessions have been milder, from a "change in JII" perspective, than earlier ones. The "spike" in the JII is much smaller.
Finally, things are pretty darn good right now. In fact, the JII in January 2006 (.00211) indicates that about 1 out of 500 employed people loses his or her job each week. There have only been two better months since 1967 - March and April of 2000 - the height of the tech bubble. This is a boom, people, like it or not. You can hardly lose a job right now.