Jobless Rate Rounding: Apr. Increase Wasn't 0.20%
According to data in Summary Table A. Household Data in yesterday's employment report unemployment rates were 9.7% in March and 9.9% in April, rounded to one decimal place. dding a few more decimal places, the jobless rates were 9.749% in March (which got rounded down to 9.7%), and 9.863% in April (which got rounded up to 9.9%).
Therefore, because the March rate was rounded down to 9.7% and the April rate was rounded up to 9.9%, it made it look like the jobless rate increased a full 0.20%, when in fact the actual increase was only about half that amount: 0.11%, or a little more than a 0.10% increase.
This probably happens in a lot of months and mostly goes unnoticed, but this is a clear case where reporting the jobless rate to only one decimal place made the official increase in the April jobless rate (+0.20%) appear much worse than it actually was (+0.11%).
12 Comments:
I think reporting it with any more precision than a single decimal is scientistic - much like reporting the CPI to three decimals now instead of one.
Rather than that, we should report it to one with confidence intervals ...
Of course, could you imagine the news report on the meaning of "unemployment this month was 9.7% +/- 0.4%!
Normally I enjoy your optimistic take on data, for instance the Baltic Dry Index. But this post sort of misses the point that the jobless rate is still TERRIBLE.
Sometimes the data trees obscure the big picture forest. I think this is just such a case.
"Takes" on data ought to be realistic rather than optimistic or pessimistic, no?
Realist Theorist - Agreed, except that no take on data is ever really free from one's own perception. So 'realistic' is not easy.
Data itself doesn't provide meaning - what you do with the data adds value, or understanding. That's where perception interferes. That's not always a bad thing since it can lead to creativity and new ideas.
Then again some economics, like U6 (which is also up) is supposed to provide a snapshot (data) rather than insight (like say a predictive model).
Curious. I don't think I've ever seen a news story reporting to two decimal places, much less including the margin of error.
saron-
that's an excellent post.
i agree with you that the error rate in the unemployment figure (especially u3 which is affected by reclassification of "discouraged workers") is already too high for a .1 or .2% difference to even be meaningful.
the intensity of adjustment to a great many of the government statistics has reached a level where many are becoming suspect.
lots of investment professionals (myself among them)subscribe to alternate data services because they don't trust the government numbers.
i suspect that's going to be a growth industry.
The people who understand the variables of this data do not need the extra decimal point as they understand the big picture. The people who are clueless (the majority, including most of the news media) would not understand it and in the case of the media would not report it correctly because they hardly ever deal with decimals in the first place. The end result, the sophisticated clueless informing the totally clueless. The rest will go to alternative sources.
Just one person's opinion.
Grumbling about precision when accuracy is highly questionable?
OK, big deal. We all know the difference between accuracy and precision.
Sort of like Al Gore's $9 Million Villa?
$8,875,000
At least one person here doesn't know the difference between accuracy and precision.
some info i came across:
The BLS estimates a 90% confidence interval for a change in the unemployment rate of ±0.22%, and a 90% confidence interval for the monthly change in payrolls of ±108,000. The BLS, however, admits the payroll survey's confidence interval is not solid, given built in biases and the lack of randomness in the monthly sample.
this would seem to imply that adding a decimal place to unemployment would be pretty pointless.
the illusion of precision is far worse than the admission of uncertainly.
Post a Comment
<< Home