Think the Economy Is Hard to Predict? Try the Midterms.

Based on the economy alone, Democrats face a big problem in the midterm elections.

Inflation has been extremely high and economic growth has been weak or even negative. That is a toxic political combination — bad enough for the Democrats to lose the House of Representatives by a substantial margin.

That, at least, is the forecast of an econometric model run by Ray Fair, a Yale economist. He has used purely economic variables to track and predict elections in real time since 1978, with fairly good results, which he shares with his students and which are available on his website for anyone who wants to examine the work.

The party in power always starts off with a handicap in midterm elections, and a bad economy makes matters worse, Professor Fair said in an interview. “At the moment, the Democrats definitely have an uphill climb.”

Yet Professor Fair acknowledges that his model can’t capture everything that is going on in the country.

While his analysis shows that the Democrats have fallen into an increasingly deep hole as the year has gone on, prediction markets and public opinion polls are more upbeat for the Democrats right now, and show a surge that began in late June.

Eric Zitzewitz, a Dartmouth professor who has studied prediction markets extensively, says the improved odds for the Democrats may be linked to an important development beyond the economy: the Supreme Court’s decision to overturn Roe v. Wade.

The power of the economy

No one would question whether economic conditions have a major influence on politics.

But Professor Fair’s work goes further than that. In his book “Predicting Presidential Elections and Other Things,” and in a series of papers and online demonstrations, he has shown that the economy is so powerful that it explains the broad outcome of most national elections since 1916. His relentlessly economic approach does not include any consideration whatsoever of the staples of conventional political analysis: the transcendent issues of the day, the personalities of the candidates or the tactics employed by their campaigns.

This year, as high inflation has persisted and economic growth has slowed, he finds that the electoral prospects for the Democrats have worsened. Based on data through July, he estimates that Democrats will get only 46.70 percent of the raw national vote for congressional candidates in November.

How this projection translates into results for individual congressional seats is beyond the scope of Professor Fair’s grand experiment.

“That’s not what this model is built to do,” he said. “I leave that to the political scientists. But I think the model is showing that, because of the economy, the odds aren’t good for the Democrats holding the House of Representatives.”

Something missing

Yet, as Professor Fair readily acknowledges, his model’s single-minded exclusion of noneconomic factors inevitably misses some important things.

In the 2016 presidential election, for example, it projected that Hillary Clinton would lose the popular vote to Donald J. Trump. She won the popular vote but lost the presidency in the Electoral College.

“It’s possible,” he said, “that some of that was Trump’s personality, and that the model couldn’t pick that up.”

Something similar may have happened in 2020. The model estimated that Joseph R. Biden Jr. would receive only 47.9 percent of the popular vote but he actually got 52.27 percent. In both cases, Professor Fair said, “Trump did not do as well as he was predicted to do by the model.”

The model’s singular focus may be unable to adequately account for what Professor Fair calls “the Trump effect.” That shorthand encompasses the array of norm-shattering behaviors and issues associated with President Trump and his adherents, including the Jan. 6 insurrection; Mr. Trump’s denial of President Biden’s election win in 2020; and the decision of the Supreme Court, with three Trump appointees, to overturn Roe v. Wade, which had been the law of the land for 50 years.

What prediction markets say

In theory, the prices in perfectly efficient markets synthesize the knowledge of each participant, making them better at assessing complicated issues than any individual can. But perfect conditions don’t exist for any market on earth and certainly not for prediction markets. Still, Professor Zitzewitz says these markets are highly informative.

He pointed out that as recently as June 23, PredictIt, a leading prediction market, gave the Republican Party a 76 percent chance of taking the House of Representatives and the Senate from the Democrats in November.

But the next day, June 24, the Supreme Court overturned Roe v. Wade. By June 30, the probability of a Republican sweep, calculated from bets placed on the PredictIt site, dropped to 60 percent. They were down to 39 percent on Thursday, with a higher probability, 47 percent, given to a different outcome: Democratic control of the Senate and a Republican victory in the House.

Public opinion polls appear to have moved in a similar direction. The average of the polls tracked by Real Clear Markets has shifted from total Republican dominance to a virtual dead heat in the generic congressional ballot. On the other hand, Mr. Biden’s job rating in those polls is still awful, with nearly 16 percent more people disapproving of his performance than approving of it.

Did the Supreme Court ruling shift the odds for the midterms? Did the Jan. 6 hearings swing public opinion? Has a string of legislative victories added luster to the Biden aura and moved some voters toward Democrats?

It’s impossible to prove cause and effect for any of these things.

Economic anomalies

It is conceivable that the unique economic situation is muddling the projections in Professor Fair’s model.

Gross domestic product and the inflation rate are the only economic factors the model uses, and may not be adequate for analyzing the state of the economy now, with the pandemic and Russia’s war in Ukraine causing disruptions around the globe. Both the G.D.P. and inflation numbers for the United States are bleak and the Federal Reserve is raising interest rates, giving rise to speculation that the country is heading into a recession or is already in one.

But other metrics, like gross domestic income and the unemployment rate, have been more positive. If the economy turns out to be in better shape than the core G.D.P. and inflation data indicate, the vote projections for the incumbent Democrats would improve, and they would worsen for the Republicans.

Then again, Professor Fair said, “The economy and the political situation are always unique.”

Reality is recalcitrant. Human behavior never fits entirely into any model or market yet invented.

It’s worth knowing as much as you can about the underlying factors, but they come down to people. I find that reassuring.

In the end, elections depend on the voters coming out and the public as a whole respecting the results. Astonishingly, in 2022, that basic civics lesson needs reinforcement. The legitimacy of the 2020 election is still under attack.

So, remember, whatever the models, the markets, the polls, the pundits or the candidates say, the future is in your hands. When Election Day comes around, it’s more important than ever to get out and vote, and to make sure your vote counts.