I can see what reverence people on this site have for data. Government data are not good. If it is seasonally adjusted it has been contaminated. Government surveys are crap. And private data are better.
Well are they really?
And if you reject so much government data how you know anything about anything? Or do we just make it up from our own neighborhood experience like we are Mr. Rogers or something?
Today Tyler put up a video of the revered Mr. Biderman of Trim Tabs fame. Mr. B was making an absolute fool of himself while trying to be critical of government reports. The video is embarrassing to watch.
It’s his right to do that of course. Just as it’s your right ( I guess) to reply snarkily to what I write using profanity. But as sacred cows go this guy is probably better off on the grill than being worshipped. I’d like mine well done, please.
Where to start?
Advance release: Retail sales every month comes out as an advance release… I know how stupid that is; they should release the final report first then the prelim then the advance! But they don’t you actually have to wait for the final report for two more months. Then they are revising data in each of the next two months. Of course we analyze the report for what it is when it’s released. Economists are well aware that for retail sales sometimes the revisions bring a bigger change than the original release. It’s why the retail sales report is not just about the headline. And no Mr. B did not discover this; no he did not discover America either.
Maybe we economists and those in the press do not stress that this is the advance report enough? But the report write up usually features comments about how previous data were revised. I’d assume the ‘we have a brain in our head crowd’ would realize that next month this number, too, could be revised. It’s not final. Admittedly that’s s a subtlety.
Auto sales; Biderman does his worst when he takes a pot shot at auto sales. The BLS reported that the advance report’s value of retail auto sales rose. But Biderman insists that - and everyone knows that - this last month unit sales fell in March from Feb. Well, yes but…
Production is not consumption: I just went through this subtlety with something I wrote on industrial production. Apparently some people do not make a distinction between production and consumption or producer good and consumer goods. . We economists do. Biderman is so confused in this rant I hope he is embarrassed to have gone on so smugly on and on about something he was just dead wrong about. The auto figures reported as unit sales by the industry are about all units produced and imported, they are not about consumption at the retail level.
Dollars are not units- The retail sales report is a dollar-value report not a volume report. So that is the first disconnect for Biderman. When you look at car production in the US you get unit sales the percentage change is volume percentage. Retail sales is a dollar-value report.
Investment goods are not consumer goods - You might be aware that businesses use vehicles. In the GDP accounts we keep separate data for investment and for consumption. So all the vehicles that are produced each month plus imported do not go into retail sales…big mistake here Mr B. When businesses buy cars or trucks the amount of domestic production allocated to consumption will differ from the headline on unit sales. One big factor here is the auto rental companies who periodically need to upgrade their fleet. When they buy to renew their product it makes a big subtraction relative to the headline reported that month.
Used cars sales are in retails sales- Retail sales have yet another fly in the ointment. Retail sales are about all sales, not just the sale of new production. Used car sales are in the BLS mix but are not in Biderman’s. Yet another reason why monthly retail car sales expressed in dollars are different from the monthly production and import numbers.
And you guys LIKE his data? And he is going to save us?
He is so confused I don’t know what you can learn by listening to him. You cannot even compare the monthly unit sales numbers to car sales in retailing. It makes no sense. Yet, go to the tape, there he is spouting off and smug.
BLS data collection is archaic but….Now this is always a problem because economists have lots of rules. Data are carefully collected to ensure things get counted up correctly. And if they are not careful then they can undercount or double count so data in the economics system are put though very specific processes. I’m sure that there are ways to make things more up to date. The BLS system is old and it an old technology system. But there are issues in collecting data if you want to report out data that really make sense.
USING FICA tax receipts as a jobs proxy – Next, the idea of using the FICA taxes data as a proxy for job growth may seem like a good idea – a no brainer. Economists do look at it as a gauge but it is a very flawed gauge. There are some major assumptions in doing that which is to say several ways in which there is actual slippage so that the gauge may actually be wrong. .
Wage/salary differences are an issue- The employment tax is a flat rate tax but it’s assessed on the income you make that week until you reach the max. So there is a big zone in between a zero pay check and the max. There is no way of knowing when workers are added if they are higher or lower paid than the average for workers as a whole last week or last month.
Hours or length of the work week - Another little wrinkle is that hours-worked change. As that happens comparisons wills change week to week or month to month even if the number of workers does not change.
Overtime- If you get paid more for working overtime that could be another factor causing this tax data to shift without any link to new employment.
So you have several disconnects that make this approach only a very raw proxy for what Mr B uses it. He acts like its gospel and it is only a ballpark sort of guess-timate. Yet he wants you to think it is superior to government data? Actually it is government data and it is being tortured when put to that use.
Interesting.
So thumb your nose at me, at economists, whatever. But at least after posting this I can say I told ya so or that I explained it to you.
You can ignore this or read it and learn something the choice is yours. But in my opinion the government data are not cooked. They have not been bent for any election cycle, though that allegation is often made. And private-sector data are rarely are offered up with the kind of attention that would allow them to compete in any way with the government’s reports. The only exception to this is the ADP which is often different from the government report it tries to track and is derided mercilessly despite the great effort put into that report.
I just think some people want data that go the other way because they want to be contrary. So be contrary and don’t change your opinion. But I will continue to try and point out how things are shifting and some of you will appreciate it and a few of you will continue to make fools of yourselves.
Your choice.
Bidermans’ data are flawed. He has no insight on job growth or on retail sales. He does not even understand the concepts.