Partisan Bias in Fed Inflation Forecasts?
[This article was first published on Christopher Gandrud, and kindly contributed to R-bloggers]. (You can report issue about the content on this page here)
Want to share your content on R-bloggers? click here if you have a blog, or here if you don't.
Following on from my previous post about US Federal Reserve inflation forecast errors, I decided to put together a descriptive graph to see if there might be a partisan bias to these forecast erros. Also, given all of the work in the political economy on political business cycles, I wanted to see if forecast errors changed around elections. (See the previous post for what I mean by Fed inflation forecast error.) Want to share your content on R-bloggers? click here if you have a blog, or here if you don't.
So, I have two questions:
- Have Fed inflation forecast errors been different during Democratic and Republican presidencies?
- Are Fed inflation forecast errors different for election periods and non-election periods?
To answer these questions, I simply made a graph using the same inflation forecast error data as before, but arranged in terms of quarters before a US presidential election (quarters with elections are coded 0). I then coloured the inflation errors by the sitting president’s party. Finally, I used R‘s ggplot2 loess function to summarise and compare the errors made during Democratic and Republican presidencies.
Question 1: The Fed did tend to overestimate inflation during Democratic presidencies and underestimate it during Republican presidencies (an Error/Actual score of 0 means that the forecasters perfectly predicted actual inflation). Admittedly we have a pretty small sample of Democratic presidencies (only Carter and Clinton), but it is striking how all of the big underestimates were during Republican presidencies and almost all of the big overestimates were when Democrats had power.
Maybe, Federal Reserve staff anticipate–to an incorrect degree–that Democratic presidents will pursue expansionary policies and vice versa.
Question 2: It is not as clear that forecasts systematically differ in election periods as opposed to non-election periods. Though the spread of the errors across parties does shrink very close to the election. I wonder why this might be?
More to come . . .
The R code to reproduce the plot is:
To leave a comment for the author, please follow the link and comment on their blog: Christopher Gandrud.
R-bloggers.com offers daily e-mail updates about R news and tutorials about learning R and many other topics. Click here if you're looking to post or find an R/data-science job.
Want to share your content on R-bloggers? click here if you have a blog, or here if you don't.