E-piphany

Here is a summary of my post-Employment tweets (@inflation_guy on Twitter):

• Before any comments about unemployment, a note: it isn’t the NUMBERS that affect the political race, it’s the reality of unemployment.
• Unemployment rate 7.796% from 8.111% and vs expectations of 8.2%. 114k new jobs, tho weak in private payrolls, +86k net revisions though.
• According to household survey, 873k more people are employed this month. Uh huh. can you say ‘seasonal adjustment problem?’
• I doubt there is manipulation in these numbers, but they’re the kind of numbers that make people suspect manipulation.
• avg hourly earnings and avg weekly hours both a tick higher than expected.
• In sum, this isn’t a robust report by any stretch (114k new jobs), but it’s better than expected. I expect the unemp drop will retrace.
• Series I like, “Not in Labor Force, Want a Job Now,” fell to 6.727mm people: still 2nd highest since series started in 1994.
• As a reminder: people don’t vote the numbers they see on TV. They vote the numbers they see and feel.

The Employment report was good, but not terrific. Any way you slice it, 114k new jobs isn’t going to get anyone dancing in the streets, even with revisions. A quick word to the conspiracy theorists out there: if I was going to manipulate this number, I would do a better job. By monkeying with seasonal adjustments and secondary assumptions, you could produce a much better “new jobs” number. It would require a whole lot of people to be “in on it,” so you might as well get some value out of it, if you’re that kind of sleazebag. So I don’t think there’s any manipulation here, just bad seasonal adjustment.

But more importantly, it isn’t the number that affects the election. People think that because the unemployment rate and the re-election success of the incumbent are related, the former causes the latter. It’s not so. It’s that both are related to a third factor, the actual condition of the labor market. Remember that these are experiments, imperfect estimates of real world conditions. As it turns out, individuals are really good at assessing the state of the job market without any help from the statisticians – they just look at how many of their pals are out of work. So even if this was a completely made up number (or, more likely, just a fortuitous wiggle in the Obama direction), it wouldn’t affect the polls. Thought experiment: if the BLS today had reported 4% unemployment, what would the effect on the election be? The correct answer is zero, because everyone would know that’s not the real unemployment rate.

It’s the same thing here. If the unemployment rate really dropped 0.3% in one month, then it will affect the polls. If it didn’t, it won’t.

Judging from the total of the various employment and activity series we have, I think we have an economy that is still growing slowly, but decelerating and risking a relapse into recession. But we’re not there yet.