Hooray: The Second Derivative of the Unemployment Rate Improved | Hacker News Reader

Hooray: The Second Derivative of the Unemployment Rate Improved

fivethirtyeight.com

38 pointssoundsop16 comments

I never thought I would be a fan of writing about statistics. But Nate Silver has become a must-read.

Uh,

Sure, it sounds clever but taking the second derivative of noisy data is poor statistics as far I'm concerned.

Keep in mind that unemployment figures are not a simple polynomial. Specifically, an _improvement_ in the second derivative (not even necessarily positive) doesn't directly imply that we're going to improve. The article mentions this too:

>----------------------------------------------------

The $787 billion question, of course, is whether a decrease in the rate of job losses indeed portends a recovery, or whether such data is subject to false starts.

>----------------------------------------------------

I'm all for optimism, but there's enough short-term fluctuation in economic statistics that this is like getting the faintest whiff of baked goods and wondering whether we're all getting donuts next week.

You can also quote an article by using asterisks:

From the help http://news.ycombinator.com/formatdoc :

Text surrounded by asterisks is italicized, if the character after the first asterisk isn't whitespace.

Contrarian view: http://www.shadowstats.com/alternate_data

"actual" unemployment if computed the way it was in 1992: 19% for "shadowstats version of U6" versus "official U6"

Also note that the "birth-death" model fudge-factor they used for April was 226,000 which was about double what it was for March.

Watch out for revisions as well: "February was revised down from -651k to -681k (-30,000) and March was revised from -663k to -699k (-36,000). See a pattern there? That's 66,000 jobs that were reported as existing but revised out later."

http://market-ticker.org/archives/1026-Unemployment-Fried-Da...

Don't forget the phantom ('seasonal') 60,000 jobs created by temporary hiring for the census...

Using the term second derivative obscures the meaning here somewhat. 'Deceleration' would be a more accessible concept to use to describe what's happening. It doesn't mean we're going backwards yet (i.e. net employment gains), but the rate of loss is slowing. At least that's what they told me in Physics class.

I don't think there's any obscurity; this is high school calculus. If the rate of change (first derivative, negative in this case) is slowing, that means the second derivative is positive here. I'm not yet convinced this is a sign of things to come, however; there are five inflection points in his table where the rate of loss improves.

I'm not saying the terminology is wrong or anything like that. I'm just saying 'deceleration' is a more concise and comprehensible way of talking about changes in the second derivative. You might be surprised how little high school calculus most people have retained.

Or how little high school calculus people have taken. At my high school not even precalculus was a requirement. Of a graduating class of around 300, I'd say 30-45 students had taken a calculus course.

I agree that acceleration would be a much clearer term.

I understood what people meant by "the second derivative" before I'd taken any calculus. It's intuitive; you hear it in context once and you get it.

I think 'deceleration' is more ambiguous.

I tend to think of derivatives taken with respect to time in terms of high school physics.

Let x = postion

x' = velocity

x'' = acceleration

x''' = jerk

(and, incidentally, the 4th through 6th derivatives of position are known, respectively, as 'snap', 'crackle', and 'pop' ;-))

you sir, is a genious.

It might be kind of obscure, but it emphasizes how far we'll go to find good information in a bad enconomy. It's like when people ask you if you know somebody at a party and say "sure, the place belongs to my girlfriend's 2nd best friend's 3rd cousin twice removed" Nobody knows what a 3rd cousin twice removed is, but they know its a long way from being relevant.

Not that this isn't relevant, but it's not an improvement in employment percentage, which is what people who don't need to speculate about the economy really care about.

Just a slightly associated random data point. Across the 10 yard border Canada announced their own surprise - 35k net job increase in April: http://www.thestar.com/business/article/631083

Reply on news.ycombinator.com