It looks like the residuals may be associated with stock and bond market returns. Is that in your model? I would also be curious about the effects of mortgage rates and home prices -- the usefulness of the "home equity ATM" effect that we know influences consumer spending. Are these factors included in your sentiment model? It also seems possible that these factors would affect Republicans and Democrats differently, although unlikely to change your conclusions.
Have you ever come across Michael Niemira's 1992 Public Perspective article arguing that Michigan consumer confidence index (CCI) in October is a good predictor of November presidential elections? 7 out of 9 correct predictions from 1956 through 1992 according to Niemira's simple model; incumbent party's share of popular vote = -20% + (0.79 x UofM October CCI)%. Model based on 9 observations not likely to be very useful. I doubt that Niemira's finding still holds up with 1996 to 2020 elections factored in. And, I don't know whether UofM CCI results can be read as consistent series from 1956 to today. I have heard pundits close to my advanced age tout UofM consumer confidence as an election predictor. Probably another case of Keynes' dictum: “Practical men who believe themselves to be quite exempt from any intellectual influence, are usually the slaves of some defunct economist."
Really interesting research.
It looks like the residuals may be associated with stock and bond market returns. Is that in your model? I would also be curious about the effects of mortgage rates and home prices -- the usefulness of the "home equity ATM" effect that we know influences consumer spending. Are these factors included in your sentiment model? It also seems possible that these factors would affect Republicans and Democrats differently, although unlikely to change your conclusions.
Have you ever come across Michael Niemira's 1992 Public Perspective article arguing that Michigan consumer confidence index (CCI) in October is a good predictor of November presidential elections? 7 out of 9 correct predictions from 1956 through 1992 according to Niemira's simple model; incumbent party's share of popular vote = -20% + (0.79 x UofM October CCI)%. Model based on 9 observations not likely to be very useful. I doubt that Niemira's finding still holds up with 1996 to 2020 elections factored in. And, I don't know whether UofM CCI results can be read as consistent series from 1956 to today. I have heard pundits close to my advanced age tout UofM consumer confidence as an election predictor. Probably another case of Keynes' dictum: “Practical men who believe themselves to be quite exempt from any intellectual influence, are usually the slaves of some defunct economist."