Wednesday, September 14, 2016

Good economic news, too late to affect the election

Yesterday, a pretty good economic report came out on rising incomes.  Will it affect the election?  No.  It's too late.

Let's call it the 1992 problem.  In 1992, George H.W. Bush was running for reelection.  The economy stunk, and that's why he lost, right?  The informal campaign slogan for the Clinton campaign was, "it's the economy, stupid."  Um, here's GDP growth during the Bush the Elder Presidency (source: Federal Reserve Economic Data).



See that shaded region?  That's the 1990-1 recession.  It was over in 1992 when the election was held.  But, people didn't know it.  There is always a lag period before people start to notice changes in economic conditions.  Changes that occur too close to an election, then, won't register.

The 1992 example was an extreme case.  Usually, we go with around a six-month lag in empirical models, but the Bush recession ended more than 6 months before his reelection (or rather, failed reelection bid).  There is no hard rule.  However, we are two months from the election.  There is no time for recent trends to register.  The economic report suggests that some of the trends go back a bit, but nothing big goes that far back.  Sorry, Hillary, hope that cough gets better....

8 comments:

  1. I think there may be a difference between good and bad economic news.

    The economy had rebounded in 1992 (and in 2012), but nobody seemed to notice.
    The economy crashed right before 2008, and everyone noticed.

    Good news travels slower than bad. Probably due to a combination of basic human psychology, media preferring negative stories, and campaigns "informing" people of the state of the economy.

    ReplyDelete
    Replies
    1. Actually, the "great recession" started in December of 2007, according to Federal Reserve Economic Data (same source as above), and I'd argue that the normal lag followed. People started to feel it a little in early to mid 2008, with the normal lag. Interestingly, the Abramowitz model, using second quarter data, and other similar models did just fine for 2008 even though they didn't take into account late-breaking data, when the economy collapsed in mid-to-late summer.

      The puzzle for 2008 is why the models worked so well even though the total collapse didn't occur until after the models stopped incorporating new data. The models were treating it like a normal but bad recession, and so were the voters, even though it was objectively worse. That's actually pretty weird, and in my opinion, still unresolved.

      Delete
    2. Obama runs behind those models all through the summer.

      It took Palin + DJIA saying "fuck this shit!" for it to correct.

      Ask me, and we've figured out the electoral penalty for being dark is about equivalent to the difference between "meh" economy and "holy shit, sell the kids!"

      Delete
    3. The problem with that explanation is 2012. Obama was still black, and the models still worked just fine. There was no summer economic collapse, no Palin, no stock market crash, no nothin'. If all of the stuff working against the GOP in 2008 NOT captured by the models just counteracted the race penalty, then take them away, and Obama should have underperformed in 2012. He didn't. So, no, I'm not satisfied that it's the race penalty.

      Delete
    4. Which models?
      If they include any polling data whatsoever, including presidential approval, it's baked in.

      Delete
    5. When I was giving talks in 2012, I aggregated all of the models from the PS October issue, and they collectively put Obama at 51%. He got 51%. Back before the action got started, Sides just plugged GDP into a raw model and got roughly that. Basically, 2012 was a normal election.

      Delete
  2. What's the causal model, then?

    The brain-dead sheeple in the voting booth ask themselves "was I better off 3 months ago than I was 15 months ago?"

    ReplyDelete
    Replies
    1. Changes in "the economy" don't directly affect most people. In most downturns, very few people actually lose their jobs, and due to downward nominal wage rigidity, nobody takes a pay cut. Pay raises are rare all around. So, economic changes aren't directly experienced by most people. It takes a while for information to circulate. And, sociotropic voting is more prominent than pocketbook voting anyway.

      Delete