You must hate Trump, part 2

 It occurred to me that I didn't show the average gap in Democratic and Republican perceptions of economic conditions by administration.  I'll do that, and show the gap in expectations too.  The gap is defined as the difference between supporters of the President's party and supporters of the other party, so it's always expected to be positive.

                            Current            Expected        Observations

Carter                        5.5                    3.1                        1

Reagan                    16.0                23.3                          5

GW Bush                16.6                18.4                        10

Obama                        7.6               23.3                        25

Trump                        21.7               52.0                        46

Biden                            4.4                46.6                        7

 

There is no clear tendency for an increase in partisan differences in ratings of current conditions.  It is larger under the three Republican administrations than in the three Democratic ones, although the numbers of observations in some of the administrations is small.  There is a clear increase in partisan differences in expectations--it was more than twice as large under Trump as under Obama, and has been almost as large under Biden.  Taking a closer look:

 

 

 Republicans grew more pessimistic from December to January (maybe in December some still believed that Trump would get a second term) and have stayed about the same since then.  Democrats have pretty steadily grown more optimistic.

A large gap from the beginning, with little change until Covid hit.   The very last observation was November 2020--presumably most of the survey was done after the election, explaining the big drop in Republican expectations.  Democrats became more optimistic in September and October, maybe because of increasing confidence that Biden would win.

Obama

Partisan differences were small at the beginning, then grew over most of the administration and seemed to decline towards the end.  The biggest gap was in Nov 2012, just after Obama was re-elected.  I

And GW Bush:

 

  Partisan differences declined as the recession deepened.  

It seems like there's a tendency for partisan differences in expectations to be smaller when economic conditions are bad.  That is, supporters of the president's party seem to be more affected by economic declines.  But the big story is the increased partisan gap since Trump.

Returning to the issue that started this series of posts, partisan differences in perceptions of reality have definitely increased on some things.  But some of those, like the extent of election fraud, are things about which the average person has no real information.  On economic conditions, people do have relevant experience, and there are also standard figures which are reported by all news outlets.  So if partisan differences on perceived economic conditions haven't increased, that's understandable.  But that is an "if"--they might have increased, but the evidence isn't clear enough to be confident.  

 

You must hate Trump

 Last week Paul Krugman had a column about perceptions of economic conditions.  Of course, they differ by partisanship--people rate them more favorably if the president if from their party--but Krugman says that (i) the difference is bigger than it used to be and (ii) it's not symmetrical--Republicans are more affected by the president's party than Democrats.   "Democrats did mark down their economic views after the 2016 presidential election, but not that much. The real question about the 2016 election aftermath is why Republican assessments became so much more favorable, even though not much had changed."  He drew on data from the Survey of Consumer Finances.  Unfortunately, his links didn't work, but the SCF has a page giving means by party at a number of times between June 1980 and June 2021.  I downloaded it and did some more analysis.  

First, the means under Democratic and Republican presidents:

                        President

                    Dem        Rep

Dem            90.6        94.8

Rep              94.8        115.1

Democrats rated conditions slightly better under a Republican president, Republicans rated them as a lot better.  Of course, actual economic conditions were not the same under Democratic and Republican presidents, so I did a regression with the unemployment rate as a control.  Controlling for unemployment the effect of having a president of your own party was +6.3 for Democrats and +22.0 for Republicans.*

 There were three times when there were surveys around a presidential transition--2008-9, 2016-7, and 2020-21.  I looked at the change from October to February for each one:

                    Dems    Reps

2008-9        +16        -0.5

2016-7        -0.6        +13.8

2020-21      +6.7        -12.9

In 2008-9 and 2020-1, economic conditions were changing rapidly, but in 2016-7 things were very steady (unemployment fell from 4.9% to 4.6% between Oct 2016 and Feb 2017).  So the large Republican shift supports Krugman's view.  

However, examination of the means during Trump's administration shows that Republicans became more positive until Covid came, while Democrats became more negative.  Just going by the statistics, the Republican response is more consistent with the facts (hence the title of this post).

 

 Here are the figures in the Obama administration (unfortunately the partisanship question was not asked as often).

 Again, the partisan divide grew during the presidency.  

There's not much data for earlier presidencies, but here is what they have for George Bush:

The partisan divide seems to have grown, although it's not clear whether Republicans or Democrats are more affected by partisanship.  There's also the unexpected fact that partisan differences were small at the beginning of both the Obama and Trump administrations.

The SCF also asks about expectations, which I'll consider in a later post. 

*I don't give standard errors because there was a lot of serial correlation, and observations were very unevenly spaced, making it difficult to model it.