The jobless claims number was a big surprise today. Consensus was for 380k initial jobless claim submissions. Actual claims came in at 353k, well under last month’s 386k claims and under the consensus estimate. While the volatility is fairly high in this number during the summer season, there is one major point to make here. Going back as far as the 60s, we have never had a recession unless the initial jobless claims number rose more than 15%, year-over-year. The long-term trend continues to be lower claims.
So why is this an important data point? I'll discuss. A citizen files for unemployment benefits and within two to three weeks they get their first check. That is to say, that counting each filing is a fairly easy task. The data is measurable, recordable, and reportable. While the number from the Department of Labor is seasonally adjusted (like most economic indicators), it doesn’t get revised like GDP or payrolls. Unadjusted for seasonality, the total initial claims were 337,059 last week, a decrease of 118,201 from the previous week and 32,148 less than a year ago in the same comparable week. There are no adjustments using models or voodoo economics. One thing it is not, is a good measure of persistent unemployment. People drop off unemployment insurance after their term is over. This has been extended for the past couple of years by Congress, but eventually, if a eligible worker stops looking and their term ends, they stop receiving benefits and are removed from the continuous claims.
Going back to my claim about never having a recession unless the initial jobless claims number rose more than 15%, year-over-year, here is the proof below (see chart). The red vertical bars represent recessions based on NBER’s official start and end dates. These are built into Bloomberg’s data here and they were not created by a secondary graphics program (to establish accuracy). Every time NBER made a recession official, the rate of change (ROC) for the initial jobless claims smoothed out 4-week average was always above 15% from a year ago. Currently, this rate of change is in the negative territory, down 11% from a year ago. The ROC has been above 15% before without calling a recession. I’m not saying this indicator calls recessions, I’m saying we’ve never had a recession that wasn’t confirmed by this indicator.
Another point I want to make is that the some of the most followed economic composite models use this data point. Some have a larger weighting than others, but many encompass the data into their economic models. Like I said, it has statistical integrity.
One of the things we do as economists is study the correlation between each economic indicator and GDP or stocks. There exists a high correlation between the 4-week initial jobless claims report (inverted) and the rate of change in Gross Domestic Product, year-over-year. While many economists have noted a deceleration in the year-over-year change in GDP from 2010 to now, we may be seeing an inflection point as of the March quarter. Simply put, the correlation between initial jobless claims and GDP suggests that the turn in the March quarter of 2012 from a year ago may be an inflection point as shown on the chart below. GDP can be revised down, but initial jobless claims are not revised and they point towards better GDP numbers ahead until claims reverse their downward trend.
We are still in a “growth recession”. That is to say that we are growing at a sluggish pace as we did in 1993, 1995, and the 2004-2006 periods. Slow growth shouldn’t cause a bear market in stocks. We’ve had corrections since the market bottomed in 2009 that have been scary. Each correction was a call for the return of the bear market. Risk management concerns since the market dropped 55% in 2008 have exasperated each correction in 2010, 2011, and in May of this year. However, the trend in jobless claims has signaled that the last three market corrections were temporary in nature.
While this was a great data point for the bulls, they don’t get to push the “iWin” button just yet. The prior week showed a large jump in claims which some interpreted as a sign of deteriorating conditions. Just as that week’s results shouldn’t be interpreted alone, we can’t take this week’s drop out of context. Instead, follow the trends and the correlations.
