Network News

X My Profile
View More Activity

It's always the economy, stupid

My Sunday column will be familiar to readers of the blog: When it comes to elections, the economy -- particularly real disposable income -- matters a lot more than, well, anything else. That goes for candidates, ads, gaffes, weather and, yes, even deficits. The graph below (and thanks to the Monkey Cage's John Sides, and the Washington Post's Pamela Tobey and Dylan Matthews, for making it) tells the story: Track election results against changes in the deficit and you get a mess. Track them against changes in the economy and you get a strong, clear line. Click on the image for a larger version:


One of the interesting features of the second graph is the presence of 1964, 1972 and 1984 at the top of it. Those years -- the Goldwater, McGovern, and Mondale years -- are, in the popular memory, the strongest possible argument that candidates matter. They were all historic routs that we blame on terrible, terrible candidates.

But placed on this graph, they become the strongest possible argument that candidates really don't matter. The three historic routs coincide with the three historic years of income growth. In other words, if you knew nothing about the candidates, you still could've predicted that the largest incumbent victories would be in those three elections.

That isn't to say that the candidates didn't matter at all. The incumbents somewhat overperformed, and it's possible that better candidates could've made them underperform. It's also possible that good economies attract bad challengers, as better candidates wait for a year when they have a better shot. But even so, most of the results are explained by the economy, even as we've explained them to ourselves in terms of the candidates. And those three elections are just particularly strong examples of an interpretive mistake we make all the time.

Update: For a variety of technical and space-related reasons, we ended up only graphing presidential elections in the piece. But over at the Monkey Cage, Sides runs the numbers for midterm elections set against changes in the deficit. The story remains the same:


"The line summarizing this relationship is nearly flat and the relationship itself is statistically insignificant. The story is also the same in midterm years only and in Senate elections," concludes Sides.

By Ezra Klein  |  July 12, 2010; 9:09 AM ET
Categories:  Articles , Political Science  
Save & Share:  Send E-mail   Facebook   Twitter   Digg   Yahoo Buzz   StumbleUpon   Technorati   Google Buzz   Previous: Wonkbook: BP replaces cap; Kagan against the left; White House pushes jobs tax credit
Next: Tom Toles is worth ten thousand words


"In other words, if you knew nothing about the candidates, you still could've predicted that the largest incumbent victories would be in those three elections."

This statement can't be justified unless you also show us the plot with those three points removed. Since all three were on the same side of the line, and since extreme results have a strong effect on regression lines, the trend might well be considerably weaker with those three points removed.

Posted by: Unwisdom | July 12, 2010 10:38 AM | Report abuse

Since so many of your posts have to do with correcting or better expressing the junk that people read about the economy in their newspapers, you could do a further service by linking regularly to Dean Baker's regular efforts in this area, in Beat The Press at CEPR

Posted by: bdballard | July 12, 2010 11:21 AM | Report abuse

The right-hand chart doesn't entirely confirm your thesis, though. What about the 1952, 1968, and 2000 elections? I'd say the determining factor in those elections was war hero, Vietnam, and press malfeasance, respectively, and not the economy.

Posted by: thehersch | July 12, 2010 12:28 PM | Report abuse

@thehersch: Ike was going to win the 1952 election before he even announced which party he was going to run for. But you could argue that the results in '52 and '68 show that wars can ruin a President's popularity (and by extension, that of his party). When a President eligible for re-election pulls out a race, that's not a good sign for his party.

As for 2000, it's worth noting that Gore won the popular vote. It's hard for a statistical model based on national popularity to account for the vagaries of the Electoral College.

Your approach to statistics would be, um...let's be charitable and call it "non-standard".

Posted by: rick_desper | July 13, 2010 12:53 AM | Report abuse

The comments to this entry are closed.

RSS Feed
Subscribe to The Post

© 2010 The Washington Post Company