Funny Seasonal Adjustment Skews First Time Unemployment Claims

by Lee Adler, Wall Street Examiner

The Labor Department reported that in the week ending November 9 the advance figure for seasonally adjusted initial claims was 339,000, a decrease of 2,000 from the previous week’s revised figure of 341,000 (was 336,000). That’s the headline number, not the actual number, which I’ll get to. First, let’s look at the problem with the headline number.

The consensus estimate of economists of 330,000 for the SA headline number was too optimistic (see footnote 1). That follows two weeks in which they were close to the mark. Prior to that they were consistently unable to guess the impact of the government shout down, severely underestimating claims in those weeks. This looks like a return to that pattern, which is mystifying since the real-time hard data on Federal Withholding taxes, which I track weekly in the Treasury Update, was soft during the shutdown, and the subsequent rebound was also weak.

The ritual of the media reacting to fictional, seasonally adjusted headline numbers goes on. It’s especially nutty this week where the seasonal adjustment factor is even more out of whack than usual. The media made much of the fact that the prior week’s number was revised up by 5,000. The number is always revised up because the advance number reported the first week does not include all interstate claims. By the following week they are all counted. The usual upward revision is from 1,000 to 4,000. This one was larger than typical. But looking at the not seasonally adjusted actual count of claims which the 50 states submitted to the Labor Department, the upward revision was just 1,277. Let me emphasize that those are actual counts, not estimates, and not seasonally finagled hocus pocus.

So how did the seasonally adjusted upward revision turn into a big number of 5,000, when the actual change was 1,277? Without going into the arcana of the abstract impressionist work of seasonally adjusting real numbers into an idealized image, one factor stands out. In 2012, the comparable week saw a massive spike in claims in the immediate aftermath of Superstorm Sandy hitting the East Coast. That certainly has something to do with it, because the seasonal adjustment process attempts to idealize the trend by incorporating the average of the past 5 years. If one of those years has an unusually large change, that skews the average.

The headline seasonally adjusted data is the only data the media reports but the Department of Labor (DOL) also reports the actual data, not seasonally adjusted (NSA). The DOL said in the current press release, “The advance number of actual initial claims under state programs, unadjusted, totaled 357,836 in the week ending November 9, an increase of 26,485 from the previous week. There were 478,543 initial claims in the comparable week in 2012.” [Added emphasis mine] See footnote 2. The 2012 number was boosted sharply by Superstorm Sandy, which had swept across the northeast coast of the US the week before.

After jumping during the government shutdown, and excluding last year’s anomaly, claims have returned to trend. Actual filings last week were down 25.2% versus the corresponding week last year. That’s an outlier due to the post Hurricane Sandy spike that week of 2012. This comp should return to trend over the next week or two. The trend average is -8.1% per year over the past 104 weeks.

Initial Unemployment Claims - Click to enlargeInitial Unemployment Claims – Click to enlarge

There’s significant volatility in this number, with a usual range of zero to -20%. In the second and third quarters, claims as a percentage of the total employed were at levels last seen at the end of the housing bubble, just before the market and economy collapsed. Since then they’re returned to a somewhat higher level, settling in at about 0.23%-0.26% of total employed.

Initial Unemployment Claims Percentage of Total Employed - Click to enlarge

Initial Unemployment Claims Percentage of Total Employed – Click to enlarge

The current weekly change in the NSA initial claims number is an increase of 28,000 (rounded and adjusted for the usual undercount) from the previous week. That compares with an increase of +117,000 in the post Superstorm Sandy comparable week last year, but it’s very close to the 10 year average for this week of +26,000. The second November reading was up in 7 of the prior 10 years. The current number is entirely consistent with the trend.

Federal withholding tax data slumped sharply in the first half of October. It then began a gradual recovery after mid month but that recovery has stalled and it is below the trend growth rate of the past year. I report on this data in greater detail with graphs of the trends in the weekly Professional Edition Treasury update.

Current weekly claims would need to be greater than the comparable week last year to signal a weakening economy. That has not happened. The trend had previously been one of accelerating improvement in spite of the fact that the comparisons are now much tougher than in the early years of the 2009-13 rebound. The data has returned to trend over the past three weeks. With now much tougher comparisons versus the prior year, I would expect some slowing in the rate of improvement to be normal, and not an indication of a weakening economy.

Relative to the trends indicated by unemployment claims, stocks have been extended and vulnerable since May. QE has pushed stock prices higher but this short video shows that it has not helped employment. In fact, QE 3-4 has correlated with a stall in jobs growth.

Initial Unemployment Claims and Stock Prices - Click to enlarge

Initial Unemployment Claims and Stock Prices- Click to enlarge

I plot the claims trend on an inverse scale on this chart with stock prices on a normal scale. The acceleration of stock prices in the first half of 2013 suggested that bubble dynamics were at work in the equities market, thanks to the Fed’s money printing. Those dynamics appeared to have ended in July but the zombie has kept coming back to life. I address the specific potential outcomes in my proprietary technical research.

More charts below.

Stay up to date with the machinations of the Fed, Treasury, Primary Dealers and foreign central banks in the US market, in the Professional Edition, Money Liquidity, and Real Estate Package. Try it risk free for 30 days. Get the research and analysis you need to understand these critical forces and stay ahead of the herd. Click this link and begin your risk free trial NOW!

Footnote 1: Economists adjust their forecasts based on the previous week’s number, leading to them frequently getting whipsawed. Reporters frame it as the economy missing or beating the estimates, but it’s really the economic forecasters who are missing. The economy is what it is.

The market’s focus on whether the forecasters have made a good guess or not is nuts. Aside from the fact that economic forecasting is a combination of idolatrous religion and prostitution, the seasonally adjusted number, being made-up, is virtually impossible to consistently guess (see endnote). Even the actual numbers can’t be guessed to the degree of accuracy that the headline writers would have you believe is possible.

Footnote 2: There is no way to know whether the SA number is misleading or a reasonably accurate representation of the trend unless we are also looking at charts of the actual data. And if we look at the actual data using the tools of technical analysis to view the trend, then there’s no reason to be looking at a bunch of made up crap, which is what the seasonally adjusted data is. Seasonal adjustment just confuses the issue.

Seasonally adjusted numbers are fictional and are not finalized until 5 years after the fact. There are annual revisions that attempt to accurately reflect what actually happened this week. The weekly numbers are essentially worthless for comparative analytical purposes because they are so noisy. Seasonally adjusted noise is still noise. It’s just smoother. So economists are fishing in the dark for a fictitious number that is all but impossible to guess. But when they are persistently wrong in one direction, it shows that their models have a bias. Since the third quarter of 2012, with a few exceptions it has appeared that a pessimism bias was built in to their estimates.

To avoid the confusion inherent in the fictitious SA data, I work with only the actual, not seasonally adjusted (NSA) data. It is a simple matter to extract the trend from the actual data and compare the latest week’s actual performance to the trend, to last year, and to the average performance for the week over the prior 10 years. It’s easy to see graphically whether the trend is accelerating, decelerating, or about the same.

The advance number for the most recent week is normally a little short of the final number the week after the advance report, because the advance number does not include all interstate claims. The revisions are minor and consistent however, so it is easy to adjust for them. Unlike the SA data, after the second week, they are never subsequently revised.

Cliff-Note: Neither stopping nor starting rounds of QE seems to have had an impact on claims. Nor did the fecal cliff secastration. The US economy is so big that it develops a momentum of its own that policy tweaks do not impact. Policy makers and traders like to think that policy matters to the economy. The evidence suggests otherwise.

Monetary policy measures may have little impact on the economy, but they do matter to financial market performance. In some respects they’re all that matters. We must separate economic performance from market performance. The economy does not drive markets. Liquidity drives markets, and central banks control the flow of liquidity most of the time. The issue is what drives central bankers.

Some economic series correlate with stock prices well. Others don’t. I give little weight to economic indicators when analyzing the trend of stock prices, but economic indicators can tell us something about market context, in particular, likely central banker behavior. The economic data helps us to guess whether the Fed will continue printing or not. The printing is what drives the madness. The economic data helps to predict the central banker Pavlovian Response which is, when the bell rings -> PRINT! Weaker economic data is the bell.

Please share your comments below this post.

Initial Claims Seasonal Adjustment Off Track

Initial Claims Seasonal Adjustment Off Track – Click to enlarge

Initial Unemployment Claims Long View - Click to enlarge

Initial Unemployment Claims Long View – Click to enlarge

Four years into recovery, claims are at a level equal to the second year of the recovery from the 2002 recession. Four years into that recovery, claims were below 300,000 per week at this point in the year (note that 2005 had impacts from Hurricane Katrina).

Initial Unemployment Claims Seasonal Adjustment Factors - Click to enlarge

Initial Unemployment Claims Seasonal Adjustment Factors – Click to enlarge

The Labor Department, using the usual statistical hocus pocus, applies a seasonal adjustment factor to the actual data to derive the seasonally adjusted estimate. That factor varies widely for this week from year to year. The factor applied this week applied an increase to the actual data, rather than a decrease as in 9 of the 10 prior years.

Stay up to date with the machinations of the Fed, Treasury, Primary Dealers and foreign central banks in the US market, along with regular updates of the US housing market, in the Fed Report in the Professional Edition, Money Liquidity, and Real Estate Package. Try it risk free for 30 days. Don’t miss another day. Get the research and analysis you need to understand these critical forces. Be prepared. Stay ahead of the herd. Click this link and begin your risk free trial NOW! [I cover the technical side of the market in the Professional Edition Daily Market Updates.]

See Rick Santelli use one of my proprietary charts on CNBC to explain how the Fed impacts the stock market directly through its trades with the Primary Dealers. This is just one example of the dozens of proprietary charts that I build that will help you to clearly see and understand the market’s trend, and when that trend is beginning to change

Share this Econintersect Article:
  • Print
  • Digg
  • Facebook
  • Yahoo! Buzz
  • Twitter
  • Google Bookmarks
  • LinkedIn
  • Wikio
  • email
  • RSS
This entry was posted in Employment and tagged , , , , , . Bookmark the permalink.

Make a Comment

Econintersect wants your comments, data and opinion on the articles posted.  As the internet is a "war zone" of trolls, hackers and spammers - Econintersect must balance its defences against ease of commenting.  We have joined with Livefyre to manage our comment streams.

To comment, just click the "Sign In" button at the top-left corner of the comment box below. You can create a commenting account using your favorite social network such as Twitter, Facebook, Google+, LinkedIn or Open ID - or open a Livefyre account using your email address.