Showing posts with label forecasting. Show all posts
Showing posts with label forecasting. Show all posts

The Number of New Jobless Benefit Claims for the First Week of 2012

How many people filed for new unemployment insurance benefits in the first week of January 2012?



We'll find out the answer to that question later this morning, but in the meantime, we can make a pretty good guess using our updated forecast chart for the number of seasonally-adjusted initial unemployment insurance claims filed each week!



Residual Distribution for Seasonally-Adjusted Initial Unemployment Insurance Claims, 26 March 2011 - 31 December 2011

Assuming the trend established since 9 April 2011 remains intact, and that the variation of individual data points with respect to that mean trend line follows a normal, bell-curve kind of distribution, we can give the following odds that the number of seasonally-adjusted new jobless claims for the week ending 7 January 2012 will fall within the indicated ranges:




  • There is a 50% chance the number will be above 384,593. Likewise, there is a 50% chance the number will be below 384,593.

  • There is a 68.3% chance that the number will be between 372,552 and 396,634.


  • There is a 95.4% chance that the number will be between 360,510 and 408,676.


  • There is a 99.7% chance that the number will be between 348,469 and 420,717.




Using our statistical control chart-inspired methodology, should the number of seasonally-adjusted initital unemployment insurance claim filings for the week ending 7 January 2012 come in either below 348,469 or above 420,717, it would be an indication that the trend established since 9 April 2011 is potentially breaking down.



As you can see though, it's been going pretty strongly for nearly eight full months now. If it continues through the end of March 2012, it will be very unlikely that we'll see the number of new jobless claims climb above the 400,000 mark, where it spent much of 2011.



In looking deeper at the chart, we see some indications that the established trend may indeed be breaking down. Here, in going from 17 September 2011 to 24 September 2011 and then again from 26 November 2011 to 10 December 2011, we see that the volatility in the number of new unemployment benefit claim filings being recorded each week has increased.



If that continues, the big unknown for us right now is that we don't know yet which way the numbers will break!




Update (12 January 2012, 9:36 AM: Here's the updated chart, incorporating the data from today's report:



Residual Distribution for Seasonally-Adjusted Initial Unemployment Insurance Claims, 26 March 2011 - 7 January 2012

Today's new jobless claim value of 399,000 was just outside our 68.3% probability range. Given the BLS' track record in under-reporting the actual number of new jobless claims each week, we can reasonably expect that the figure for the week ending 7 January 2012 will be revised upward in the next report to be 400,000 or higher.



Today's number also underscores the increasing level of volatility in the data - when the current trend was establishing itself, it was characterized by relatively small changes in the number of new jobless claims being filed from week to week.



Today new data marks the fourth time in the last six weeks in which the size of the change in the reported numbers from week to week has exceeded one standard deviation. That's specifically what we're looking at when we suggest that the established trend may be beginning to break down.


S&P 500: In the Zone!

Nobody can predict where stock prices will go next, can they? Especially given the volatility of stock prices, especially in today's market, where the market can swing by more than 3% in any given day, right?



It's just not possible, is it?



To really find out, we ran a two-year long experiment, from April 2009 through April 2011, to see if we could forecast the average value of stock prices for a month at the end of the previous month. Here were our final results:



S&P 500 Average Monthly Index Value, April 2009 to April 2011

As you can see, we offered a split final forecast option for April 2011. Here's what we believed would happen instead:


What we believe is likely is that stock prices will track upward from the average level of 1304 they recorded in March 2011 toward the 1393-1429 level our primary method would forecast as the noise currently in the market subsides.




And that's what happened. In April 2011, stock prices did indeed track upward, rising to an average level of 1331 for the month, with the S&P closing the month at 1363.61.



We took the next several months off from offering public forecasts of where the S&P 500 would head next, but by 26 September 2011, we couldn't resist any more, and posted the following chart, which presents a graphical prediction that happens to cover the period through the end of 2011:



S&P 500 Average Monthly Index Value vs Trailing Year Dividends per Share, December 1991 through 23 September 2011

And here's what the updated chart looks like, through the end of 16 December 2011:



S&P 500 Average Monthly Index Value vs Trailing Year Dividends per Share, December 1991 through 16 December 2011

What can we say? We're still in the zone! And that concludes, for real this time, our public experiment in forecasting the future for the S&P 500!

New Jobless Claims: Still On Track

Today's news that the number of initial claims for unemployment insurance benefits being filed for the week ending 26 November 2011 ticked back up over the 400,000 mark is right on track with the prediction we published over a month ago.



Here's that prediction:




... since that slowly downward trending line is currently projected to stay above the 400,000 level through the end of 2011, we can therefore expect that there is over a 50% probability that the number of new, seasonally adjusted initial unemployment claims will be above the 400,000 mark through the end of the year.



In fact, what we can expect as we go forward in time is that we'll see an increasing number of times in the weeks ahead where the number of new jobless benefit claim filings will fall below the 400,000 mark, as the number of layoffs from U.S. employers each week continues to decline gradually.




Lo and behold, that's pretty much exactly what has happened so far in the time between 21 October 2011 and today, as shown in our chart below, which adjusts the trend line slightly:



Residual Distribution for Seasonally-Adjusted Initial Unemployment Insurance Claims, 26 March 2011 - 26 November 2011

For the six most recent observations of the number of new jobless claims, which cover the period of time since we made our prediction, three have been at or above the 400,000 mark, while three have been under.



We also see that the mean trend line has shifted to be slightly steeper, which suggests that the number of initial unemployment insurance claim filings each week will be more likely to fall under the 400,000 mark in the weeks ahead.



Meanwhile, Bloomberg reports that the uptick in new jobless claims is "unexpected":




Jobless claims climbed by 6,000 to 402,000 in the week ended Nov. 26 that included the Thanksgiving holiday, Labor Department figures showed today in Washington. The median forecast of 43 economists in a Bloomberg News survey called for a drop to 390,000. The number of people on unemployment benefit rolls and those getting extended payments increased.




Whoops! There's 43 economists whose forecasting ability is now in doubt, which is a shame because the number of new jobless claims filed each week is perhaps the easiest of all economic data to forecast while its basic trend is intact!



And at present, it appears that the current trend for weekly new jobless claims remains well in force.