The chart below shows the increase in currency in circulation that occurred just before 9-11, which Bill Bergman discussed on the most recent Robert Wenzel Show. As a Fed employee at the Chicago Fed in the money laundering division, Bergman started investigating the odd withdrawals of currency before 9-11 and was fired.

(Chart prepared by Bob English)

In the interest of putting this in context, how does this stack up against previous spikes in history? (if that data is available)

ReplyDeleteAlso, can someone quantify the effect of the Argentine crisis Bergman mentioned?

That's the thing Capn Mike, if Bergman was allowed to continue his investigation and not fired, he would have been able to determine whether there was an innocent explanation for the withdrawals, but for whatever reason, after he started asking legitimate questions, as part of the money laundering division, he was first removed from the division and then a week later from the Fed, after working there 13 years.

DeleteAnd so, we don't have answers to the question about Argentinian withdrawals that you ask.

Capn Mike, at the time I was first looking at the issue, in late 2003, in the 'not seasonally adjusted' data, the increase in the currency component of M1 from June-August 2001 was easily the largest June-August % increase in the 50+ years since 1947, the first year the current data series started. There are other ways to look at it, too.

ReplyDeleteInteresting.... of course, there are 2 explanations. (a) the withdrawals prior to 9/11 are truly abnormal, (b) withdrawals prior to 9/11 are consistent with the average and the 1-standard (& 2-standard) deviations around the average.

ReplyDeleteIf Bill Bergman was on to something, he would have put the standard deviations around the averages... I am sure he has the numbers (if he had the numbers to compute the average...he had the numbers to compute the deviations).

Given that the standard deviations are not posted, my suspicion is that he never bothered to compute them, and that he will likely find that the withdrawals prior to 9/11 are consistent with the 5 year average.... and that his work is just junk science.

Yeah, it was just a coincidence that he got fired, right afterwards. Yeah, that's the ticket. A coincidence. Yeah.

DeleteWhat this obvious econometric wank fails to realize is that standard deviation tests aren't required here (even though Bergman conducted them). All that would have been required is for the Federal Reserve, instead of removing Bergman from his job, to allow him to conduct his investigation and he could have seen exactly who withdrew the money for what reason.

DeleteEconometric wanks don't even think in reality anymore. Amazing.

dude, if he's saying it was the MAX seen for that seasonal period since the data set started, then "almost by definition" [unless it's a flat data series, or there are only two data points] it's going to be more than sigma ...

DeleteThat said, if I was a betting man, given the two choices presented before me (argentina, and 9/11 conspiracy), I'd choose argentina. I'm also extremely naive (self-diagnosed), so you should really pay zero attention to my opinions (serious, not sarcastic).

The only problem with it "just being Argentina" is that they would have let him continue his investigation AND not password limit the data.

DeleteIn the nonseasonally adjusted data for June-August, Anonymous I, the increase in 2001 was over 2 standard deviations above the average growth rate for all the June/August intervals for 1947-2012

ReplyDeleteI've shared with Bob a chart showing a frequency distribution I prepared about 7 years ago for all the months (e.g. not just the June-August interval) since 1947.

Great, 2-sigma deviations happen fairly often (about 5% of the time if you assume & have a Gaussian distributions), specially when you have a finite number of observations with which you are estimating 2-sigma and really have no clue as to the distribution of these observations (if your distribution is log-normal or something, and you calculate your sigma with a gaussian assumption, then your estimates are even more off).

DeleteSo the implications of the 2-sigma values you cite are: (a) there is a decent chance that the "spike" that you see is well within the distribution of all the withdrawals, and nothing unusual happened in terms of withdrawals. (b) if there is anything unusual happening, it is impossible to see this from the data you present.

Remind me, what was the purpose of the chart presented in this post (I mean, other than inciting people with exciting suggestions).

And for Robert... only in a bizarro, crackpot world are "standard deviations tests not required". Most sane scientists would have said that "standard deviation tests are necessary, but not sufficient". But whatever... it's your life, you can waste it any way you want.

Why don't you update this posting with the histogram that Bill Bergman talks about, and point out the spike that you guys are so worked up about. Should be easy to do...

To see if a bank wire hit my bank, I check my account balance, not plot the odds of it not hitting my account. A standard deviation test is not necessary to see if the balance hit. Same thing with Bergman's investigation. If it would have been allowed to continue, he would have seen EXACTLY where the money went, no standard deviation test would have been as accurate, or required. You econometric wanks are something else.

Deletehmmm - I seem to recall a year 2000 bug liquidity spurt followed by a drain, followed by the steepest stock market crash in Nasdaq history. Perhaps some perspective would be helpful.

ReplyDelete" (a) there is a decent chance that the "spike" that you see is well within the distribution of all the withdrawals, and nothing unusual happened in terms of withdrawals. (b) if there is anything unusual happening, it is impossible to see this from the data you present."

ReplyDeleteFor all the technical terminology you seem unable to grasp what is at issue here. If a) holds true, then what is the Fed's issue?