Presidents and the Unemployment Rate

By Morgan Housel

Elections tend to have themes. If the 2000 presidential election was about education, the 2004 election about national security, the 2008 election about Wall Street and health care reform, then the 2012 election will surely be about jobs, jobs, jobs.

How does unemployment affect the odds of being re-elected? There are all kinds of theories. Most point back to the classic line: "Are you better off than you were four years ago?"

Harvard historian Niall Ferguson wrote in his book The Cash Nexus:

A good illustration of this new economic determinism was the widespread explanation of the failure to impeach President Clinton for perjury and the obstruction of justice in connection with his numerous sexual misdemeanors. By February 1999 a majority of Americans believed Clinton was guilty of the charges against him, but only a small minority wanted him to resign as president. According to Senator Robert Byrd – and many other commentators – the explanation was simple: "No president will ever be removed when the economy is at record highs. People are voting with their wallets in answering polls."

On the contrary, Ferguson writes, in the year before Richard Nixon resigned in 1974 with approval ratings under 30%, "unemployment rose by almost 1 million and the inflation rate doubled ... on Wall Street the stock market fell by a third." Indeed, by 1974, the Dow Jones (INDEX - ^DJI) was in one of its worst slumps since the Great Depression. Clinton, on the other hand, enjoyed the greatest bull market in history.

Anyways, here's how the unemployment rate has worked out over the last nine presidential terms:

Source: Bureau of Labor Statistics. *Through April 2012.

This doesn't look good for the current president. As The New York Times once noted, "No American president since Franklin Delano Roosevelt has won a second term in office when the unemployment rate on Election Day topped 7.2 percent." (Despite the numbers in the table, the quote still applies to Ronald Reagan in 1984, as the unemployment rate rose between Election Day and Inauguration Day.)

But statistician Nate Silver looked closer at the data going back to William Howard Taft in 1912 and found something surprising: There's virtually no correlation between the unemployment rate on Election Day and an incumbent's chance of winning. Even looking at the change in unemployment throughout an incumbent's term, the correlation was weak. Same if you narrowed it down to just the post-World War II period. Even if you are "more or less deliberately cherry-picking" the data, Silver wrote, the correlation between unemployment and presidential elections is elusive.

But here's what I find interesting: On Election Day 2008, the nationwide unemployment rate was 6.8%, yet according to Census Bureau data, just 3.5% of those who actually voted were unemployed at the time. And more than a fifth of those who voted came from a household with an annual income of more than $75,000 – a significantly higher portion than the national average. It was virtually the same in the 2004 election. The nationwide unemployment rate was 5.4% on Election Day 2004, yet just 2.7% of those who voted were unemployed at the time.

The average voter in the last two elections, then, has not been representative of the broader economy. They've been in much better financial shape than the average American. That could help explain why there's such a low correlation between the unemployment rate and the odds of being re-elected. And if that pattern repeats during this November's election, then what seems obvious today – that the election will be about jobs, jobs, jobs – might not be quite right.