top of page
  • Writer's pictureSeth Masket

Are the Forecast Models just plain Wrong?

I like to put together a little economic forecast model around this point in an election cycle. Not so much because I'm trying to predict what'll happen -- there are more sophisticated ways to do that -- but because I want to get a sense of a baseline. That is, what are the fundamentals of this year? Once we know that, we can get a bit of a sense of how other variables -- campaigning, candidate quality, etc. -- are performing.

In 2022, however, what traditional forecast models are telling is wildly different from what everything else is telling us.

I use a simple model for midterm elections that predicts the amount of House seats the president's party will losing with three basic variables: growth in per capita real disposable income, the president's approval rating, and the number of seats that party currently holds. This model came pretty close in 2018 and even in 2014.

As I mentioned a few months back, the model for 2022 predicts a very dire fall for Democrats. Using more recent models that take into account economic growth through the 2nd quarter of 2022 are even worse. Depending on which quarters of growth you focus on, you get a range of pretty dramatic forecasts for Republican pickups (data available here):

  • 2021Q2 - 2022Q2: 75 seats

  • 2021Q3 - 2022Q2: 66 seats

  • 2021Q4 - 2022Q2: 58 seats

And we get these crazy figures from the fact that RDI growth over last year is historically bad right now. In the forecasts above, RDI growth is -4.5, -3.4, and -2.2, respectively. (This is due to a combination of factors. For one, the federal government's expanded child tax credit was giving thousands of dollars to parents in 2021 and now it's not. People just have less spending money because that program ended. Also, inflation has reduced people's available spending money. All this makes this economic measure look unusually bad even while other economic indicators -- including GDP, unemployment rates, etc. -- look strong.)

By comparison, the latest FiveThirtyEight forecasts (based on a host of variables including the economy, polling, and other factors) has Republicans picking up around just 15 seats, with substantial variation around that estimate. There's even a 22% chance of Democrats holding the chamber. What explains this extraordinary disconnect? There are a few main factors:


There's just no doubt that the Supreme Court's ruling in Dobbs v. Jackson Women's Health Organization in the spring has dramatically shifted the political environment. Republicans had a much larger turnout advantage earlier in the year, and that's been pretty much erased. And as we've seen from a range of special elections this fall, Democrats have been out-performing their 2020 numbers.


One of the advantages that the out-party usually has in a midterm year is its ability to recruit strong candidates, while the president's party often has a harder time with that. This year we've seen the opposite. Just to focus on a few key Senate races in states that Biden won narrowly in 2020:

  • In Pennsylvania, Republican Mehmet Oz has been trailing Democrat John Fetterman substantially. Oz has proven to be a poor campaigner and is struggling with claims that he is not truly of the state, while Fetterman's campaign (especially his social media game, for whatever that's worth) looks like something for the history books.

  • In Georgia, Republican Herschel Walker has also proven a poor challenger to Democrat Raphael Warnock. Walker simply has a difficult time speaking coherently. Polls there are essentially tied.

  • In Arizona, Blake Masters has issued a number of statements indicating a general hostility to democracy. He is now trailing the Democratic incumbent, Mark Kelly.

Notably, all three of these Republican nominees were endorsed by Donald Trump. Yes, Trump was most active in encouraging Walker to run, but Oz won very narrowly with Trump's help, and Trump may have been influential in the Arizona contest, as well. To be sure, Trump isn't the only reason the GOP is recruiting celebrities, authoritarians, and conspiracy theorists lately, but he's surely the most influential person in the party pushing for that.

Fundamentals Improving

Inflation (especially gas prices) was pretty bad earlier this year but has actually been subsiding somewhat. Covid still exists, but infection, hospitalization, and death rates have been on a slow decline. Despite a lot of talk to the contrary, the US is not in a recession, and unemployment remains at an historic low. None of this is yet making Americans feel great, and Biden's approval ratings, while rising, are still pretty low -- but Americans' impressions seem to be slowly improving.

Trump Ain't Helping

I know many pundits are convinced that every time Trump gets in legal trouble that just helps him sell his persecution fantasies to the electorate, but he's really not a popular figure. The latest investigation of his theft of classified government documents is quite serious, and Trump staying in the news every day for criminal behavior is not really doing his party, which wants to keep the electorate focused on Biden's unpopularity, any favors.

The elections are still more than two months away, and things could move further in the Democrats' direction or they could snap back in favor of the economic forecast models. It's also possible the polls are understating Republican support, as they did in 2020. Generally, I would advise election predictors to bet on the simple forecast models. The times I've gotten burned in the past are when I've ignored those models in favor of news coverage and vibes. However, the models aren't sacred, and if anything could cause elections to defy them, it's probably the set of situations we're seeing this year.

4,003 views0 comments

Recent Posts

See All


bottom of page