Economy Doesn't Make Clinton a Favorite in 2016

Economy Doesn't Make Clinton a Favorite in 2016

By Sean Trende - December 3, 2014

I’d intended to write about the Democratic and Republican primaries this week, but an article from the New Republic’s very smart Danny Vinik caught my eye.  So, the primaries will have to wait. Referencing a piece by Ed Luce in the Financial Times, Vinik writes that Luce “overlooks a fundamental reason why [Hillary Clinton] is the early [2016] favorite: The economy is quickly improving under a Democratic president.”

Vinik then cites political science research suggesting that the economy is the crucial factor in predicting elections.  This is true.  But there are other factors to consider as well; no model of which I’m aware goes “full economic determinism.”

When we look at the overall political science view of elections, the picture is far less rosy for Clinton (or any other Democratic nominee). The 2016 election looks more like pure tossup; if anything she’s the underdog, not the favorite.

Before going any further, I’d invite readers to review some of my thoughts on economic forecasting models here and here.  I’ve since softened my view somewhat (as Keynes put it, “When the facts change, I change my mind. What do you do, Sir?”) and am probably now somewhere near John Sides’ view here or (particularly) here.

That is to say, I think fundamentals-based models provide good rules of thumb and, taken as a whole, can give a sense of how a “normal” election would look. They work well most of the time, and offer the best starting point for analysis. 

Fundamentals modeling tell us, for example, that we shouldn’t be that impressed when Ronald Reagan wins by 19 points in 1984: This is what happens when an incumbent runs with 7 percent economic growth at his back. Similarly, the models suggest that 2012 wasn’t an anomaly that requires some novel factor like demographics for explanation; the parsimonious explanation is that 2012 behaved like any other election where a president with tepid approval sought re-election amid slow growth.

With that said, one ought not be a fundamentals fundamentalist. These models all come with substantial error margins, which can be important in a close race. They also come laden with the (reasonable but not iron-clad) assumption that elections in the future will behave like previous elections. More importantly, there are elections that really don’t seem to fit in the “normal” election mold: Most models had Al Gore winning a substantial victory in the popular vote in 2000. No fundamentals model had Republicans picking up anywhere close to 63 seats in 2010.

But though we know we might encounter an outlying election in 2016, we don’t have any way of knowing in which direction it will be an outlier.  We should also keep in mind that these models will be re-estimated using 2012 data; the actual predictions might vary modestly from what follows.

What do these models say?  One of the more popular forecasting models is Alan Abramowitz’s “time for change” version.  It has two variants.  The first, which we might call “Time for Change Classic,” looks at net presidential approval, GDP growth in the second quarter, and how many terms the party in control has held the White House.  The second, which we’ll call “New Time for Change,” adds a polarization variable.

So what does the classic version tell us?  Let’s first use CBO’s estimate of 4 percent growth.  This is probably on the high side (CBO has been forecasting a surge in GDP just around the corner for five years now) so we’ll asterisk it as a high-end probability.  President Obama’s job approval rating is -10.8 percent today.  If we plug these two variables into Time for Change Classic, it suggests that Republicans should be favored to win by about three points: 51.7 percent to 48.3 percent.

But, you say, with 4 percent growth, Obama is unlikely to remain at -10.8 percent approval.  Fair enough.  But even if we move him up to a net-neutral job approval, the models forecast a narrow Democratic loss, 50.5 percent to 49.5 percent. Obama would have to reach a net job approval of +6 before the model would forecast the Democrat to win (narrowly).  Obama has accomplished this four times in his six years in office: During his two post-election “honeymoons,” after the shooting of Gabby Giffords, and after killing Osama bin Laden.  If he ties his post-2009 best of +12 percent net approval, the model would favor the Democrat by a point.

But what if the Fed forecast of 2.6-to-3 percent growth is more accurate? At 2.8 percent growth and using Obama’s current job approval, the model forecasts a Democratic loss of 4.6 points; at neutrality it forecasts a Democratic loss of 2.4 points, and even improving to a +12 percent net approval rating would suggest a very narrow Democratic win (.198 points, to be exact).

What about New Time for Change?  With 4 percent growth, it forecasts anywhere from a nine-point Democratic loss (Obama at -10.8 percent net job approval) to a seven-point Democratic win (Obama at +12 percent net job approval).  At 2.8 percent growth, it forecasts anywhere from an 11-point Democratic loss to a five-point Democratic win. 

Of course, “Time for Change” isn’t the only model out there.  There were many others in 2012, but they can be sorted into two basic buckets: models that use national economic indicators (e.g., GDP), like Time for Change, and those that use personal measures of economic health, such as Douglas Hibbs’ famous Bread and Peace model (which uses real disposable income).  The “national” models had a better go of it in 2012 (the “personal” models generally suggested a Romney win), while the personal models tended to be closer to the results in 2010. 

Without getting into the details of Bread & Peace’s computations, we’ll just note that real disposable income growth in Obama’s second term has been weaker than it was at the tail end of his first term. The model weights recent growth more heavily than it weights growth early in a term, so a late turnaround would have a disproportionate impact on the projected 2016 results.  For now, we suspect that the personal models won’t show a good result for Democrats, absent a substantial improvement.

We should recall that all of the caveats from earlier in the piece are real, and important.  These models aren’t the be-all/end-all of election analysis, and Republicans or Democrats probably could win under any of the scenarios outlined above.  We also should remember that we’re building estimates on top of estimates here: Personal or national measurements of the economy could surprise us, pleasantly or negatively, and this would alter our estimates for 2016.

Regardless, these models probably can give us a rough sense of what would happen under various fundamentals for 2016.  They point to a reasonably close election; they do not suggest that the Democratic nominee should be considered the favorite at this point.

Sean Trende is senior elections analyst for RealClearPolitics. He is a co-author of the 2014 Almanac of American Politics and author of The Lost Majority. He can be reached at Follow him on Twitter @SeanTrende.

Obama, Hillary "Catch Up" at the White House
Alexis Simendinger · December 4, 2014
Clinton's Stock Declining in Futures Market
Michael Barone · December 5, 2014
Hillary Clinton for Supreme Court Justice!
Mike McCurry · December 16, 2014
All Is Not Calm
David Shribman · December 14, 2014

Sean Trende

Author Archive

Follow Real Clear Politics

Latest On Twitter