Cooking the numbers: a half-baked idea

Many have commented on the Republican conspiracy theories about BLS cooking the September employment numbers to help Obama's reelection. Enough has been said about how preposterous those theories are and how irresponsible the media have been in covering them as if they deserved some respect. I have nothing to add to that, but I am puzzled that I haven't seen anyone ask an obvious question:

How do those people think embellished statistics would help a candidate in an election?

Seriously. Do you know any person who would decide for whom to vote, or whether to vote, based on the published unemployment rate? Can you imagine somebody thinking, "I wasn't going to vote for Obama when the unemployment rate was 8.1%. But wow, the unemployment rate is 7.8% now, and by golly, I'm voting to reelect the President!"?

OK, there may be two or three Aspergians nationwide who are so much into economic data that  oher shethey set a voting decision rule based on such statistics. But, if such people exist, I bet they live in DC anyway (working for the government or some think tank), and DC's electoral votes are already spoken for.

Other than that hypothetical and vanishingly small demographic, no voter thinks that way. Most voters don't even know what the unemployment rate is, and those who know couldn't care less.

Of course, people care about the economy. But they don't care about abstractions. They care about how they are doing, and how their friends and relatives are doing. They react to what they experience. If you are unemployed, and the unemployment rate declines from 8% to 6%, did you somehow become less unemployed? Nothing has changed for you. Things did change for a lot of other people, and they are now more likely to reelect the current officials. That's why job numbers matter for the election. But they don't matter to any individual voter.

That's why you don't see a bounce for Obama in the polls after the good economic news, unlike the day before, when his advantage diminished because of the poor debate performance. But the debate—however irrelevant in substance—was news to every individual. By contrast, the employment numbers were news in the aggregate, but not to individuals. Nobody's experience of the economy changed with the publication of those numbers.

That is not to say that those job numbers are not reflected in the polls. Of course they are—in polls conducted in September. Think about it: if more people became employed in September, more people liked the current administration in September. Incidentally, Obama's poll numbers did improve in September, although we'll never know how much of that was due to the improving economy, how much  to the energized Democratic Convention, and how much to Romney's gaffes and the "47%" recording. But, whatever the effect of the economy, it was there in the polls long before it was in the published data.

And this is generally true: economic statistics are lagging indicators of voters' opinions, because they are published with a delay and voters form opinions based on their experiences immediately. To put it another way: the polls predict the economic data, not the other way around.

So how would making up job numbers help reelect the President? In the immortal words of South Park's underpants gnomes:

1. Cook the numbers
2. ?????
3. Second term!!!

The idea that this could work is crazy—even more crazy than the blind, rabid partisanship that transforms otherwise sane people into paranoid conspiracy theorists.