San Diego had another pretty good month for job growth, according to data released last week by the California Employment Development Department.
Unfortunately, you may not want to believe the numbers.
Let’s see what they have to say, first.
Two of the three housing boom beneficiary sectors, those industries that I have long considered to be at risk due to their dependence on the robust real estate market, remained weak. Construction, the hardest-hit housing beneficiary sector, shrank by 4,100 jobs or 4.6 percent since last year, while the finance and real estate industry declined by 1,100 jobs or 1.3 percent. These year-over-year declines, while not exactly great news, are better than some of the declines these sectors were experiencing in mid-2007.
The retail sector, on the other hand, has surprised me by continuing to grow even though San Diego consumers can no longer so easily use their homes as ATMs. The EDD estimated that this industry grew by 2,600 jobs or 1.7 percent since last December. This strength is recent, incidentally — earlier in 2007, the retail sector was routinely showing year-over-year job losses in the thousands. (The spike in recent months represents the usual holiday-related rampup; I focus on year-to-year changes because they remove any seasonal considerations.)
Outside these three sectors, San Diego employment grew by 17,200 or 1.7 percent. This growth, in combination with the decreasing drag inflicted by the housing beneficiary sectors (which together fell by only .8 percent), led overall San Diego employment to increase by 14,600 jobs or 1.1 percent.
The size of the labor force grew slightly faster than that at 1.2 percent, meaning that the local unemployment rate ticked up a tenth of a percent to 4.9 percent. Still, employment growth was firmly positive.
That’s the good news. The bad news is that there are reasons to question the accuracy of these job statistics.
A recent North County Times article, for instance, offers up a raft of data that doesn’t sit well with the recent retail industry strength reported by the EDD. To cite one data point in particular, San Diego taxable sales for the second half of 2007 were 2.4 percent lower than they were for the same period in 2006. This indicates a pretty serious slowdown in retail activity.
The article also quotes George Whalin, a local retail expert with Retail Management Consultants in Carlsbad, as saying, “Retail employment has suffered badly over the last couple of months.”
I called Mr. Whalin to get his take on the latest EDD retail job growth figures. He was very skeptical, saying that he talks to many retail sector participants and that the anecdotal evidence presents little doubt that the industry is weak. He ascribed this frailty to the housing downturn, unsurprisingly, and summarized his take without equivocation: “Hiring in the retail sector has not been good this last fall, and it’s certainly not expected to be good for the foreseeable future.”
The EDD’s take, in contrast, can be seen on the graph to the right, which shows the number of retail jobs each month during 2007 compared to the same month in 2006. According to the EDD, the retail industry shook off early-2007 weakness to come back strong during the final months of the year, ending up 1.7 percent larger than it had been at the end of 2006. This conclusion seems very much in conflict to the impression given by either Mr. Whaling or the sales tax figures.
What could be causing the disconnect? It could be the case that more employment isn’t signaling greater strength — in other words, there could be more employees working fewer hours. But a more likely culprit is the birth-death model, a statistical method employed by the EDD and Bureau of Labor Statistics (which publishes the same data as the EDD) when putting their monthly employment data together.
I first described the birth-death model when reviewing last August’s numbers. To briefly recap, the agencies are only able to survey employment among the businesses that they know about, but much of the job-related action may be happening in the businesses that are not part of the survey sample because they only recently come into existence and are not yet known to the agencies. So the agencies employ a modeling approach that uses past business creation data to estimate how many new businesses are coming online. This technique, called the birth-death model, may indeed provide a better job growth estimate for as long as the employment trends are cruising along relatively unchanged. But because it rests on the assumption that the present is acting like the past, it is likely to miss the cyclical turning points in the job market.
The various conflicting data indicates that the EDD may be missing just such a turning point in the retail sector. And the same may be happening in other sectors as well.
According to the BLS, which only publishes birth-death model info on a nationwide basis, the birth-death adjustments were responsible for increasing reported national employment by 138,000 jobs in the construction industry and 114,000 jobs in the financial industry during 2007. We know that both sectors have suffered mightily in the aftermath of the housing bubble, and we also know that both have been losing jobs all year. Yet the birth-death model appears to insistently assume that a relatively stable number of new businesses are entering these faltering industries every month, even as the other data shows that the industries are shrinking.
No data is offered for the retail subsector, but the “trade, transportation, and utilities” supersector, of which retail accounts for 67 percent, shows a birth-death adjustment of 178,000 jobs nationwide for 2007.
The table to the right shows the effects of the birth-death model on reported 2007 employment growth in the housing boom sectors. In all three cases, the birth-death model assumed new job creation even as the estimates excluding the effects of the model (and even including them, in the case of finance and construction) would have shown the sectors as shrinking.
Just to be clear, I’m no expert on the matter of how the BLS and EDD calculate their job statistics. But we can all look at how the agencies’ estimates stack up against different data points, and we can apply common sense. The birth-death model appears to assume that the housing downturn hasn’t had much effect on new business creation within the sectors that benefited so richly as a result of the real estate boom. The different data points and common sense both say otherwise.
— RICH TOSCANO