Are you suggesting that the final unemployment figure is the one reported by these thousands of statistical experts? These thousands that you refer to are only employed for groundwork. They are scattered and fragmented and not a united whole. They report it to their superiors, who perform further analysis and then report it again to their superiors.
I agree also with your conclusion as well: it does logically follow from what you just said that a collusion may still happen in principle:
In the process of filtering through such layers of reporting, the thousands who actually collected and worked on the data are never in the know of what the actual figure is.
Indeed, all people, all those thousands do not have common knowledge of how aggregation was done. However, at each level where aggregation is performed, there are some people that perform that aggregation. Moreover, people providing input into a given level of aggregation do have a feel for the aggregate (there are series that historically covary, etc.) There is nothing unusual here: an engineer does not look over the shoulders of workers who implement his design, but he would catch a deviation from his design in a heartbeat. That is all the more true if such deviations were done repeatedly. Those that provide inputs into level 3 aggregation do not have a feel for the level 6 aggregate number, but they do have a feel for the aggregate at level 3.
In sum, (i) a collusion would still be necessary at any given level where stepwise aggregation takes place, and (ii) it is quite detectable by those that provide inputs.
I would actually go further: I cannot claim to know specifics here since I never worked for the BLS, but I would be very surprised if various multi-level checks were not in place to avoid bias (even inadvertent bias). The reason for my confidence is that it is a standard procedure in data gathering and aggregation.
A well-known example is from surveys and census procedures. As you probably know, interviewers are often part-time workers that are completely disinterested in the integrity of the data and very much interested in doing less work. They are always tempted to fill out a dozen questionnaires at the kitchen table rather than actually walk to a dozen homes and conduct interviews at each. There are numerous checks and measures to detect this agency problem. Basic courses on survey methods emphasaize this problem and suggect specific measures for detection of such agency problems.
BLS procedures are nothing but surveys; I would be very suprisied if all those Ph.Ds at BLS, which is in focus of entire nation, would not know or fail to implement what every student of survey methods is taught. I cannot attest to which specific methods are implemented, but I have no doubt that some methods to prevern even an inadvertent bias --- let alone a collution -- are in place.
Finally, your argument, as well-constructed as it is, is different from mine. It shows specifically how a collusiuon may occur. But I never argued that it cannot occur. I only said that, whenever someone claims that it has occurred --- or as some on this trread claim that it has occurred continually for decades --- the burdern of proof is on the one making a claim. That is what I ask people: if you have proof, let's hear it; and, please reported to the FBI --- I want to see the perps in jail (no, I am not saying this sarcastically). In sum, I am not asking the accusers to do anything other than what our traditional values and law always asked: put up or shut up. Please see #151.
To me, this is different from what you have argued and with which I completely agree: collusion may happen in principle, despite measures against it that are put in place.
Thank you very much for your post.
Thank you for clarifying. However, some points in this reply, I observe, are in contrast to your previous posts.
What I understand from your previous posts is that you (a)consider that the unemployment figure reported is correct or not grossly misrepresented (b) hold the opinion that the top management of the banks are blameless and has nothing to do with the failure of these banks e.g. GS, F&F, AIG (c) hold the opinion that the govenment is responsible for the failure of the banks (This is not an attempt to put words into your mouth, do correct me if I am wrong)
TopQuark wrote “It shows specifically how a collusiuon may occur. But I never argued that it cannot occur”
First through numerous posts you attempted to convince the members that there is no collusion on part of the BLS top level (in this case) or the top management in case of F&F, AIG, GS etc then you said you are open to the idea that, in the general environment, collusion can occur (may I add that this general environment includes BLS, F&F, AIG, etc)
I explained to you how collusion takes place in order to suggest that the BLS is not immune to any sort of collusion taking place (I will be glad to debate this)
Sticking to the unemployment bit, you say “ Indeed, all people, all those thousands do not have common knowledge of how aggregation was done. However, at each level where aggregation is performed, there are some people that perform that aggregation. Moreover, people providing input into a given level of aggregation do have a feel for the aggregate (there are series that historically covary, etc.)”
As far as unemploment, inflation (and even census!) figures are concerned the thousands providing input, I am sorry to say, do not have a feel for the whole. Why so? you may ask, because, they are as I said earlier, fragmented. Calculations are done on a regional basis. Those calculating unemployment for the state of California have no clue (officially) of the rate of unemployment in Alabama or the US as a whole entity. The link is broken here itself and hence no question arises of them having a feel of the aggregate.
You further wrote “In sum, (i) a collusion would still be necessary at any given level where stepwise aggregation takes place, and (ii) it is quite detectable by those that provide inputs”
Only if it was so simple. It is as good as saying that the accounts head (statistician reporting for California) at a branch of a conglomerate (California) has the idea of the profit at the group level (United States). Rates of every state are reported by the statistical head of that state to a central location and this central location is where distortions are permeated. Nobody is daft enough to do it in a way that invites attention. There are ‘politically correct’ tricks. For example, a higher level of confidence (confidence interval), statistical significance, t-values ... almost anything can be manipulated to get the results that the BLS wants the public to believe.
This cannot be reported to the Fed can it? What would the poor officer say...
Officer to Fed: The rate of unemploment is 14% not 8%. I am here to report Mr. X who has deliberately taken a higher confidence level.
Fed to Officer: Get lost!
As you say, in practice the statistical experts have to ‘put up or shut up’
My only request to you is to refrain from attempting to convince people that the figure of unemployment at 8.9% is correct. In doing so, I have also provided reasonable explanation why I do not believe it is the correct figure.
P.S: I trust you are also aware of the numerous political motivations behind distorting critical figures like unemployment rate.
TopQuark “Indeed, all people, all those thousands do not have common knowledge of how aggregation was done. However, at each level where aggregation is performed, there are some people that perform that aggregation. Moreover, people providing input into a given level of aggregation do have a feel for the aggregate”
I am sorry to say, they do not have a feel for the aggregate figure as far as enemployment rate is concerned. Statistical analysts in California are (officially) not aware of the figure of unemployment in Alabama. They can therefore not know the figure for the country as a whole.
You further wrote “In sum, (i) a collusion would still be necessary at any given level where stepwise aggregation takes place, and (ii) it is quite detectable by those that provide inputs”
This can be true of pure sciences and not for statistical inferences which require human estimation e.g. confidence interval, t-values, statistical significance. I trust you are acquainted what purpose they serve in statistical calculations. Manipulating these figures to supress the level of unemployment is easy and cannot be reported even by those honest statisticians you refer to. They merely put up and shut up.
I made an attempt to convince you how collusion takes place in order to display that the figure of unemployment is distorted.
It is naïve to believe that the BLS figure is untainted and exact and there is no collusion within the top level of BLS to manipulate the figure especially for a crucial figure like level of unemployment