Economy no longer as accurate at predicting elections — but there's a catch: expert

It's longstanding conventional wisdom that voters decide elections based on the economy. But is that still true? According to Peter Coy in The New York Times, a recent study suggests it doesn't matter nearly as much as it used to.

Robert Gordon, an 84-year-old economist at Northwestern University, "has studied presidential elections going back to 1956, when he was 16 years old and Dwight Eisenhower, the incumbent Republican president, and his running mate, Richard Nixon, were squaring off against the Democrats Adlai Stevenson and Estes Kefauver," wrote Coy. "The economy is a big factor in elections, but it’s not as influential as it once was, according to a working paper released by Gordon just in time for this year’s election."

Specifically, he wrote, the paper shows that the economy indirectly sways the election by influencing the president's approval rating.

"The turning point was around 2000, when polarization increased and people became more likely to vote based on party affiliation, all other criteria (including economics) aside," wrote Coy. "Oddly, Gordon found that even in the period before 2000, consumer sentiment did not predict winners."

ALSO READ: 'People have had enough': Here are the 3 'big-picture' reasons why Kamala Harris will win

Rather, back in the times when the economy did predict winners, only two economic factors had a notable effect on a president's approval rating: economic growth adjusted for population, and the average inflation rate for the president's first three and a half years compared to the same period under their predecessor, known as the "excess" inflation rate.

And even this broke down starting this century, Coy noted: "While the equation correctly predicts 15 of the 17 elections since 1956, its two mistakes were in 2000 and 2016, when Al Gore and Hillary Clinton got fewer Electoral College votes than the equation predicted. The equation underpredicted how many electors Trump would win in 2020, but did get the result right."

What all this means, Gordon told Coy, is that his model would tend to predict a generic Republican cruising to victory in 2024, since "excess" inflation was high, with Vice President Kamala Harris predicted to win 183 electoral votes or possibly even less — but that the model can no longer predict this with any certainty, and she will "doubtless" outperform that, even if she loses.