Notes on Non-Poll Based Models: (Moody, Fair, Luskin, Dale, Vinod B. and Litchman's Forecasts)

Each of these models are different. Vinod B. has not updated his model since the coronavirus hit. Initially, he felt Trump would win. The other models predict a victory for Biden. Dr. Fair does not have an official forecast.

There is a commonality in these models, as the incumbent individual is more likely to win a second term, if all goes well. Thus, the presidency is really for the incumbent to either hold on to or lose. Don Luskin examines electoral vote margin (EVM). If Trump wins 300 EV's and Biden wins 238 EV's, then the EVM is 62. In his model, just being an incumbent, is worth 81 margin votes. Allan Litchman awards 3 "true" answers for Trump because he is the unopposed incumbent.

However, one party can "wear out their welcome" by winning more than 2 four year terms. Hillary Clinton was at a disadvantage in 2016 because her party was deemed "shopworn" by Luskin and the electorate would look to the opposing party. Luskin penalizes the candidate by 127 EVM. Moody's analytical model includes a "fatigue" variable (0 or 1 values) in their models to account for this effect. Since Republicans have held the presidency for only one term, the shopworn/fatigue variable is not applicable.

I would suggest anyone interested in the election predictions, should go beyond the results, and think a bit of what is being modeled. At the core, it is human nature to want new leaders at a time when the news is bad and very evident to the electorate. If there is a war in Somalia, for example, perhaps it is not a real effect. However, if there is crisis closer to home, particularly an economic crisis, such as stagnant job growth or long term inflation, it is human nature to vote against the sitting president.

On the other hand, there is a party allegiance factor at play. As Moody's summary points out, the election in 2020 on a state by state basis will likely be similar to 2016, particularly in those states won by a popular vote margin in excess of about 6%. Somewhere between 6 to 15%, states become labeled as "safe" Republican or Democratic states. Litchman uses the 2018 midterm election as an indicator if the incumbent is gaining strength. Democrats increased their numbers in the House, regaining control, and this cost Trump, one "False" point.

Deteriorating economic conditions such as rising unemployment lowers the chance of an incumbent being re-election. But, these are extraordinary circumstances due to Covid-19. The real father of economic modeling, Dr. Fair, correctly states that his model doesn't cover pandemics right before an election. As I write, we are in the midst of a Democratic Convention, and it is clear that every effort is being made to connect the terrible consequences of the epidemic to the federal government's response. The Republican Convention is up begins August 24. We'll see their response. To many forecasters, like Allan Litchman and others who believe in quantitative analysis, the die is already cast. We shall see.

Short summaries of each these models are provided below.

---

Moody's model was updated on July 15 and now predicts Biden wins in a tight race. The webpage contains an online webinar, presentation slides, and a full documentation of their forecast. Very valuable information is provided within the documentation's appendices on all swing states. Because they extend the predictive equations to each state, their evaluation is comparative to the poll based predictions. It is noted that their three predictions are dependent on the party supporters turnout that can not be estimated until after the election. For the Sept 2019 model documentation (pre Covid-19) see: Moody's Presidential Election Prediction Model. Moody's Analytics get an A+ on documentation. July 2020 update: Moody's Model

---

Professor Ray Fair's model uses % GDP changes in the last 3 quarters before an election. He writes "the model has nothing to say about the effects of pandemics" and puts a question mark in the column for a calculated % popular vote result. Fair says he'll let users make their own predictions. If one assumes a 5% increase in Q3 GDP, annualized, Biden wins the popular vote with 53%. Not really his forecast, just a numerical example, so I felt best to exclude it from the table of results. Dr. Fair receives an A+ on documentation. Website: https://fairmodel.econ.yale.edu/vote2020/index2.htm

---

Don Luskin forecast has gone from the most optimistic for Trump in March 2019 to the most pessimistic one today. His model is strongly affected by changes in current unemployment which has nosedived due to the Covid-19 lockdown. His model now shows that Biden will win by a margin of 361 EV's. Based on this margin, we calculate Biden wins 450 EV's to Trump's 88 EV's. Luskin states that it would take a gain of 8 million jobs (~ 6% unemployment) to make this a tight race. I am disappointed at Don Luskin's documentation, as it does not reveal the coefficients used in his model variables. \

Website: Don Luskin Model

---

Skyler Dale notes that his forecast shows a very close election. If the Democrats win Wisconsin, his model shows the election would be tied 269 to 269 EV's, a rather scary event. The discussion is very good as he shows how little the voter margins change from one election varies to the next election. Thus the states which were narrowly decided in 2016, have in general become the 2020 swing states. I am disappointed Dale has not updated his forecast. It is well documented. Website: Skyler Dale

---

Vinod B. model predicted the Republican candidate would receive 64% of the vote, if held today (Oct 2019) or 54% of the vote at election time. Since the pandemic, I would say that all bets are off the table unless Vinod updates his prediction. It is a well documented model.

It is worth reading as fully one year before the election, Vinod was predicting a diminishing support for the incumbent. A difficulty in the use of economic variables, the necessity to project what the values of variables might be months ahead. It is noted that Professor Ray Fair uses an econometric model to forecast the input variables to his predictive equation. Vinod Bakthavachalam

---

Professor Allan Lichtman in July 2020, updated his model, and now predicts Joe Biden will win the election. I like the quote from Lichtman that "the presidency is for the incument to lose." The only key that Democrats can take credit for is gaining members in the House of Representatives. Trump gains one "true" because Biden is not charismatic. It is likely that most incumbents will have true's for #2, #3 and #4, so Trump has 4 "true's" without doing anything except being an incumbent. Trump gains 2 more true's (#7, #10), but it's not enough to push him over the threshold.

He has a short clip which explains his prediction:

New York Times, He Predicted Trump’s Win in 2016. Now He’s Ready to Call 2020. Most historians just study the past. But Allan Lichtman has successfully predicted the future.

Lichtman in the New York Times brief video clip explains his ratings. Certainly the Trump "charisma" is open to discussion. Litchman defends his false rating, in saying that Trump is an excellent showman, but his charisma seems narrowly focused.

	Definition	Rating
1: Party Mandate	After the midterm elections, the incumbent party holds more seats in the U.S. House of Representatives than after the previous midterm elections.	False
2. Contest	There is no serious contest for the incumbent party nomination.	True
3. Incumbency	The incumbent party candidate is the sitting president.	True
4. Third Party	There is no significant third party or independent campaign (more than 5%).	True
5. Short-term economy	The economy is not in recession during the election campaign.	False
6. Long-term economy	Real per capita economic growth during the term equals or exceeds mean growth during the previous two terms.	False
7. Policy change	The incumbent administration effects major changes in national policy.	True
8. Social unrest	There is no sustained social unrest during the term.	False
9. Scandal	The incumbent administration is untainted by major scandal.	False
10. Foreign/ military failures	The incumbent administration suffers no major failure in foreign or military affairs.	True
11. Foreign/ military successes	The incumbent administration achieves a major success in foreign or military affairs.	False
12. Incumbent charisma	The incumbent party candidate is charismatic or a national hero.	False
13. Challenger charisma	The challenging party candidate is not charismatic or a national hero.	True

 

True: 6 keys, False 7 keys. Trump loses because there are 7 false keys. Anytime there are 6 or more "false" answers, the incumbent will lose according to Lichtman. So, Trump exceeded the false count by 2. The collapse of the economy due to the Covid-19 lockdown has reduced Trump's score by 2 keys. The protests following the death of George Floyd qualifies as social unrest. The most subjective key are #11 and #12 in my opinion.

David Lord

August 18, 2020