People love to peddle conspiracy theories about economic data.
The most popular realm is in the area of inflation: Tons of people think the government purposely distorts inflation data to make it seem cooler than it is.
This is a very complicated subject, because trying to gauge prices in a single index is tough, but the most serious economic minds think the government does a good job of measuring inflation and that the Ron Pauls and Shadowstats of the world are off the mark.
Another popular area: Folks these days think there’s a conspiracy regarding initial jobless claims, and the reason they think there’s a conspiracy is that every single week, without fail, the previous number gets revised higher by a few thousand.
Today ZeroHedge wrote a post called Propaganda 101 implying that the reason for these slight upward revisions to the previous week is that it makes it easier to say that claims “fell” in the subsequent week.
So for example, last Thursday jobless claims jumped to an unexpected and disappointing 380K (they had been hovering around 350K before that). For this week, this morning, claims came in at 386K (even worse!), but since last week was revised higher to 388K, technically this week was a decline. And yes, we’d be willing to bet that this week ends up getting revised even higher, so that there was no decline.
ZeroHedge posted this chart and table to show that in 2012 so far, there have been nearly 90K initial claims added over the 14 periods, and that all 14 periods have been revised higher.
So anyway, is this proof of a conspiracy? Hardly.
See, with most economic reports, the government has to take a sample survey. So for example with the Non-Farm Payrolls report, the government surveys a lot of businesses, and then extrapolates what the “real” number is for the overall economy based on the sample and a model. So the number can get revised a lot and in all different directions. That’s the case with a lot of government data: You can’t get everything, so in subsequent months you end up with more accurate data.
But initial claims are different. Actual jobless claims are events that happen at the state level every week as people file for jobless benefits, and then these numbers are sent in to the Department of labour. It’s not a sample. It’s literally a tabulation of every event. And then each week a few thousand are missed and get sent in later, and are added back to the previous week. Here’s a WSJ story on the details (via Betsey Stevenson).
Again, the key thing is that this tabulation process is not about getting a small sample and then extrapolating. It’s about collecting all the data, and then gathering a few more as it flows in later.
But what’s particularly stupid about thinking this is some conspiracy is that there would be no point to it.
The revisions didn’t change the fact that throughout much of 2011 and early 2012, claims did trend steadily lower each week. Nor do the revisions change the fact that there’s been a worrisome spike over the last two weeks.
So not only is there no basis to the conspiracy (there’s a clear explanation for what happens each week), there’s also no point to it, since it doesn’t change the underlying trend at all.
This is why the conspiracy theory is especially dumb. There are other dumb conspiracy theories (like that Obama isn’t an American), but at least the implications would be significant if found to be true. Here the implications aren’t even significant.
An annoying aspect of all this is that U.S. economic data is the world’s gold standard.
Some governments clearly have data quality issues (hello: China) and most governments aren’t even in the same ballpark in terms of U.S. data transparency and availability.
Revisions aside, the latest spike is cause for concern, since this is one of the highest quality high-frequency economic indicators we have.
All that being said, Nomura has argued that the recent spike is NOT about a weakening labour market, but rather an issue with seasonality.
See, to seasonally adjust data, you look over trends from past years to see what happened in the same week or month, to see what patterns you can find. But during winter of 2008-2009, claims were exceptionally elevated due to the Great Recession. Thus winter data these days may be understated because compared to that period, things are so much better. Note that in the chart above, we had a similar spike at the exact same time, but then things quickly resumed their downward trend.
So, bottom line: This is a worrisome trend, but not yet cause for total alarm. And on the other hand, the idea that there’s a propaganda conspiracy is total non-starter.