Initial weekly unemployment claims for the week ending on February 18th, 2012 were 351,000. The DOL reports this as unchanged from last week. The previous week was revised, from 348,000 to 351,000, an increase of 3,000.

Every week initial unemployment claims are also revised and the above graph shows this report is statistically noisy. One simply cannot compare the reported numbers on a week to week basis due to the lag in States reporting claims data and revisions, plus the fact this is a 1 week time window, versus a monthly one. One needs to at minimum look at the 4-week moving average. The 4 week moving average is now 359,000 and decreased 7,000 from last week's revised from 365,250 to 366,000 4 week average. This is the lowest moving four week average since March 2008.

Below is the 4 week moving average, set to a logarithmic scale to remove even more statistical noise, for the last year. We have a clear downward trend, which is good news. The initial unemployment report is indicating moderate to modest job growth.

The magic number to show job creation is at minimum, below 400,000 initial unemployment claims, per week. Most Economists will quote 375,000 as the magic number to indicate job growth. Houston, we have a pattern. While UI claims have not returned to pre-recession levels, they are clearly trending down and indicating job growth, finally.

Below is the mathematical log of initial weekly unemployment claims, so one can get a better sense of the rise and fall of the numbers. A log helps remove some statistical noise, it's kind of an averaging. As we can see we have a step rise during the height of the recession, but then a leveling where every week initial unemployment insurance claims hover around 400,000, refusing to really drop for over two years, a never ending labor malaise for most of the time after the recession ended in July 2009. We seem to be finally seeing a steep decline.

Below is the 4 week moving average, set to a log scale, from January 1st, 2000. Here we can see we still are not at pre-recession initial weekly unemployment claims levels. Clearly UI numbers have greatly improved from the above trend graph. Of course everyone in America should have been fired at least once by now.

Weekly initial unemployment claims are statistically noisy, the numbers are always revised, there is large variance because it's a weekly data report instead of monthly, and only through a long term pattern can one say anything about the unemployment situation. This report is great news, and what we want to see is a strong trend, dropping down to 2007 (well, 2000 would be better), levels, where the labor market still wasn't so great.

The Financial Times blog Alphaville has been examining seasonal adjustments for bias. They have identified a dip in initial claims for January and February for the last three years.

As the chart above shows, over the past three years claims have fallen from yearly highs marked in August-September to lows in February, only to stagnate in the following months

But is the above graph really valid? First, we just went through the above exercise to show when one has weekly claims data using seasonal adjustments, there will be statistical noise. Secondly, while analysis is still ongoing, it does appear the seasonal adjustment algorithms are having difficulty with years 2008-2010. The layoffs were so extreme, the seasonality of Economic Armageddon became more difficult to identify. That said, below is a graph of the four week moving average on a year to year comparison, from years 2005 to 2012. We could go back to 2000 but the graph becomes even more impossible to read. Bottom line is one should look at the four week moving average, not the weekly claims. Additionally, when looking at the 4 week moving average, the claim there is a true pattern to a decline at the beginning of the year disappears.

Continuing unemployment claims dropped but bear in mind people can plain being running out of benefits.

The advance number for seasonally adjusted insured unemployment during the week ending February 11, was 3,392,000, a decrease of 52,000 from the preceding week's revised level of 3,444,000. The 4-week moving average was 3,453,250, a decrease of 43,750 from the preceding week's revised average of 3,497,000.

In the week ending February 4th, not seasonally adjusted, the raw number was 7,502,791 official people obtaining some sort of unemployment insurance benefit. Officially, there are 12.758 million unemployed and if one takes all of the people who want a job, including those stuck in part-time because they can't get anything else, the number is over 28 million. Even at the January's job growth rate, it will take close to a decade to recover America's jobs.