How We Were Wrong About the Jobless Trend [View article]
OilFinder: Thank you for your continuing comments. We greatly appreciate your ongoing dedication in seeking to assign a specific cause to the weekly variations in the seasonally-adjusted initial unemployment insurance claim filing data. Most serious analysts would describe that variation as being the result of common causes, which we recognize in this case because the effects of the seasonal adjustment process, like the effects of weather, are always present in the data, as you have repeatedly acknowledged.
Fortunately, our analytical methods work in the absence of the absolutely perfect knowledge of the exact specific cause of such small and frequent variations in the data. We commend you for your efforts in seeking to fully account for these small and frequent variations by attributing them to the role of the seasonal adjustment process and wish you luck in doing so.
To clear up another misunderstanding you appear to have, while we might add notes regarding outliers or other interesting shifts in the data to our charts, what we're really tracking are the changes in trend over time, which is largely unaffected by such outliers. We apologize for any confusion that our chart notations may have unnecessarily caused you.
Finally, we are very disappointed that you would interpret our sincere appreciation for your extensive comments as being sarcastic in any way. As a matter of practice, we only respond to a select few comments where the commenter has offered some potentially relevant substantive criticism that we find useful. We would certainly hope you would receive substantive criticism regarding your analyses as constructively.
To that end, we find your suggestion that we either suppress or censor our analysis to be less than useful (seekingalpha.com/artic...):
"If you aren't aware of really basic things like this about initial claims, you should not be wasting your time churning numbers and writing commentary about them, because you really don't know what you're talking about."
No apology is necessary. For our part, we're happy to continue "wasting our time and writing commentary."
The real question is how much does that really bother you? And should it? Really?
Wouldn't it be more constructive on your part to write up and publish the results of your studies that clearly, definitively, and irrefutably establish your view of the matter as being the one and only correct view so that it becomes widely accepted by the community of statisticians, analysts, economists and others? Including us?
How We Were Wrong About the Jobless Trend [View article]
OilFinder: Thank you for your multiple comments - we greatly appreciate your willingness to continue sharing your extensive and absolutely perfect knowledge of all the factors that affect initial unemployment insurance claim filings at such length with such churlish intensity.
For our part, we're happy to continue "wasting our time" (thanks for your touching and heartfelt concern!) and as always, we're willing to be wrong.
Zero Deficit Line: The Link Between Median Household Income and Federal Deficits [View article]
golddragon: Thank you for your comments.
Seeing as the top 20% of households in the U.S. pay 69.3% of all taxes, while the lowest 60% (which includes those earning the median household income) pay just 14.0%, that total federal tax collections so closely tracks changes in the median household income is an unexpected result.
OilFinder: Thank you for your extensive comments - we hope you found the exercise as useful as we have in evaluating what lies behind your analysis. Since you've been kind enough to recommend a number of sites for us to review, and given what you've written, we'd like to return the favor with just the following two suggestions:
For more insight into what's behind our analysis, replace the word "control" with "equilbrium" where you see it in the article above. To the best of our knowledge, this kind of analysis has only rarely been done in economics, so we're going to see where it takes us.
Given all that you've written, we strongly recommend the following article:
We challenge our own views continually and are happy to receive constructive critiques. We're also willing to be wrong, even if we rarely are, because we learn more about how things really work that way.
Just one quick point - you allowed "It could be true that a quick change in layoff trend can be triggered largely by layoffs in small businesses, but you'd have to prove that first before continuing to hold your assumption."
We would only have to demonstrate that the capability exists. In 2004, the Census reports there were 5,885,784 firms with payroll employees in the U.S., of which 86,538 had 100 employees or more. There were 41,839,701 employees at the 5,799,246 smaller firms, accounting for 36% of all employees in the U.S. Since these firms are not handcuffed by the federal 60-day notice requirement for layoffs, they are more than capable of acting to accommodate significant changes in the business outlook in sufficient numbers to affect the overall new jobless claims data in very short order.
I'm afraid you've missed the point of the exercise - to determine when a new trend might be beginning with respect to U.S. layoffs and then to identify what event might have triggered the change in trend. The method used in the analysis is only capable of projecting the future level of new jobless claims while an established trend exists and continues - its real value lies in its ability to verify that a new trend has been established which happens after the forward-looking projections stop working. The potential change in trend is what we find to be interesting. Whether such a change in trend will be seen with the arrival of new data remains to be seen - at present, we've only noted the possibility of that being the case and identified what may be the proximate trigger for such a change.
But since you brought forecasting track records up, here's the link to what we forecast would be the trend in new jobless claims back on 2 September 2010:
was the first we offered revising the forecast, which we extended only for the week ending 30 October 2010, the data for which was not yet available. But frankly, we could have projected where new jobless claims would be on 20 November using just the data we had available to us back when we first introduced our analytical method on 23 June 2010:
As for your method, I'm sure you recognize that it would not have captured the rising trend in new jobless claims that took place from 6 March 2010 through 21 August 2010, so your timing in making your projections was fortuitous.
Oh, and you'll find that your method for projecting the future level of new jobless claims would have been mostly useless in 2007 or 2008 or 2009. The seasonal patterns you recognize will only work in forecasting when an overall trend has already been well established and is continuing - which is the case for the period over which you made your predictions.
Visualizing the U.S. Higher Education Bubble [View article]
The limiting factor is the data for average tuition - we have U.S. median income data going back into the 1960s, but only found the national average tuition data going back to the 1976-77 school year.
Divining the Future According to Dividends [View article]
We don't write the headlines that appear here at Seeking Alpha - although at this writing, it does appear that it's been changed from "Diving" to "Divining". It just goes to show that you can't rely on spell check!...
U.S. Per Capita Oil Consumption Plummets [View article]
That particular sentence is the result of a bad edit that didn't get corrected until after the Seeking Alpha folks accessed the original post's RSS feed. Here's how the sentence should read (which fits right into the context of the post):
"That drop has occurred even as the resident population of the United States steadily increased throughout this period, which means that most of the decline may largely be attributed to falling aggregate economic productivity during these years."
A Look at Who Really Owns U.S. Government Debt [View article]
How We Were Wrong About the Jobless Trend [View article]
Fortunately, our analytical methods work in the absence of the absolutely perfect knowledge of the exact specific cause of such small and frequent variations in the data. We commend you for your efforts in seeking to fully account for these small and frequent variations by attributing them to the role of the seasonal adjustment process and wish you luck in doing so.
To clear up another misunderstanding you appear to have, while we might add notes regarding outliers or other interesting shifts in the data to our charts, what we're really tracking are the changes in trend over time, which is largely unaffected by such outliers. We apologize for any confusion that our chart notations may have unnecessarily caused you.
Finally, we are very disappointed that you would interpret our sincere appreciation for your extensive comments as being sarcastic in any way. As a matter of practice, we only respond to a select few comments where the commenter has offered some potentially relevant substantive criticism that we find useful. We would certainly hope you would receive substantive criticism regarding your analyses as constructively.
To that end, we find your suggestion that we either suppress or censor our analysis to be less than useful (seekingalpha.com/artic...):
"If you aren't aware of really basic things like this about initial claims, you should not be wasting your time churning numbers and writing commentary about them, because you really don't know what you're talking about."
No apology is necessary. For our part, we're happy to continue "wasting our time and writing commentary."
The real question is how much does that really bother you? And should it? Really?
Wouldn't it be more constructive on your part to write up and publish the results of your studies that clearly, definitively, and irrefutably establish your view of the matter as being the one and only correct view so that it becomes widely accepted by the community of statisticians, analysts, economists and others? Including us?
We look forward to your published work.
How We Were Wrong About the Jobless Trend [View article]
For our part, we're happy to continue "wasting our time" (thanks for your touching and heartfelt concern!) and as always, we're willing to be wrong.
Zero Deficit Line: The Link Between Median Household Income and Federal Deficits [View article]
Seeing as the top 20% of households in the U.S. pay 69.3% of all taxes, while the lowest 60% (which includes those earning the median household income) pay just 14.0%, that total federal tax collections so closely tracks changes in the median household income is an unexpected result.
www.cbo.gov/publicatio...
And the last I checked, no "policy suggestions" were included in the article. Being imaginary, it's no wonder that you found them to be "100% wrong."
A Breaking Trend in U.S. Layoffs? [View article]
en.wikipedia.org/wiki/...
For more insight into what's behind our analysis, replace the word "control" with "equilbrium" where you see it in the article above. To the best of our knowledge, this kind of analysis has only rarely been done in economics, so we're going to see where it takes us.
Given all that you've written, we strongly recommend the following article:
en.wikipedia.org/wiki/...
We challenge our own views continually and are happy to receive constructive critiques. We're also willing to be wrong, even if we rarely are, because we learn more about how things really work that way.
Just one quick point - you allowed "It could be true that a quick change in layoff trend can be triggered largely by layoffs in small businesses, but you'd have to prove that first before continuing to hold your assumption."
We would only have to demonstrate that the capability exists. In 2004, the Census reports there were 5,885,784 firms with payroll employees in the U.S., of which 86,538 had 100 employees or more. There were 41,839,701 employees at the 5,799,246 smaller firms, accounting for 36% of all employees in the U.S. Since these firms are not handcuffed by the federal 60-day notice requirement for layoffs, they are more than capable of acting to accommodate significant changes in the business outlook in sufficient numbers to affect the overall new jobless claims data in very short order.
The source data is available here:
www.census.gov/epcd/ww...
Cheers!
A Breaking Trend in U.S. Layoffs? [View article]
But since you brought forecasting track records up, here's the link to what we forecast would be the trend in new jobless claims back on 2 September 2010:
politicalcalculations....
Our 2 November post:
politicalcalculations....
was the first we offered revising the forecast, which we extended only for the week ending 30 October 2010, the data for which was not yet available. But frankly, we could have projected where new jobless claims would be on 20 November using just the data we had available to us back when we first introduced our analytical method on 23 June 2010:
politicalcalculations....
As for your method, I'm sure you recognize that it would not have captured the rising trend in new jobless claims that took place from 6 March 2010 through 21 August 2010, so your timing in making your projections was fortuitous.
Oh, and you'll find that your method for projecting the future level of new jobless claims would have been mostly useless in 2007 or 2008 or 2009. The seasonal patterns you recognize will only work in forecasting when an overall trend has already been well established and is continuing - which is the case for the period over which you made your predictions.
Visualizing the U.S. Higher Education Bubble [View article]
Divining the Future According to Dividends [View article]
U.S. Per Capita Oil Consumption Plummets [View article]
"That drop has occurred even as the resident population of the United States steadily increased throughout this period, which means that most of the decline may largely be attributed to falling aggregate economic productivity during these years."