It's not (just) the economy, stupid

 A couple of months ago, Andrew Gelman had a post (based on work by Yair Ghitza) about a Gallup question on satisfaction with the direction of the country--the exact question is "In general, are you satisfied or dissatisfied with the way things are going in the United States at this time?"  The main point of the Gelman/Ghitza analysis is that there was a correlation between satisfaction with the country and satisfaction with one's personal life--satisfaction with one's own life was higher and less variable, but they tended to rise and fall together.  I'll return to that point, but I wanted to take a closer look at the "direction of country" question.  It was asked occasionally (usually a few times a year) starting in 1979, and then more frequently(almost every month) starting in the early 1990s.  Here is a plot of the percent saying that they were satisfied minus percent saying dissatisfied:

Satisfaction rose from the late 1970s to somewhere around 1986, declined until about 1992, then rose until the late 1990s and remained high for a a few years.  Then it declined until 2008, and kept rising until 2020, when it dropped.  The recent decline is presumably due to Covid, but what could account the ups and downs before that time?  The obvious possibility is economic conditions.  I regressed satisfaction on the unemployment rate, and got an estimate of -5.6 with a t-ratio of about 7--that is, satisfaction is lower when unemployment is higher.*  But there is some systematic variation that is not accounted for:

This figure shows actual and predicted satisfaction.  I indicate a couple of important historical events.  The two surveys in the early 1990s with unusually high levels of satisfaction coincided with the first Gulf War.  September 11, 2001 also had an impact--satisfaction had been declining, falling to 43% (vs. 55% dissatisfied) in a survey taken September 7-10, 2001.  It jumped in the wake of 9/11, but soon started to decline again, although it wasn't until early 2003 that it reached its 9/10/01 level.   But there's clearly a lot else going on.  Of course, there are other economic conditions that might matter--for example, inflation probably contributed to the low levels of satisfaction in the late 1970s.  But I think that any reasonable list of economic factors would still leave a lot unaccounted for.  So I'll offer some ideas based on general knowledge and my own memories.   I'll focus on the 1986-92 decline--I'll consider the 1992-8 increase in my next post and then maybe move on to the current century.

The period from 1986 to 1992 saw a lot of political scandals (most of which seem pretty trivial today) and popular discontent with Congress.  The only one that's widely remembered now is Iran-Contra (1986-7), but the scandal that led to the resignation of House Speaker Jim Wright (1989) and the House banking scandal (1992) got a lot of attention at the time.  There was also the strange saga of the 27th amendment, which says that congressional pay increases don't apply until the next congress (ie after an intervening election).  That had been proposed at the same time as the Bill of Rights and ratified by a number of states and then forgotten until someone rediscovered it in the early 1980s.  Then states started ratifying it again, and it was enacted in 1992.  More generally, I remember a lot of negative attention to Congressional pay increases at around that time.  So it seems like people were just mad at the government, particularly Congress.  That might reflect the long Democratic control of the House of Representatives (since 1955)--people felt like the same crowd had been running things forever.

Another important event was the Omnibus Budget Reconciliation Act of 1990, which was passed by a bipartisan majority (31 Democrats and 23 Republicans in the Senate) and signed by President Bush.   Bush had pledged "no new taxes," so a lot of Republican voters were mad, and Democrats weren't that happy either--although most of the tax increases applied to people with high incomes, some affected the middle class.  The point this illustrates is that although people like bipartisanship in general, it doesn't always produce popular results.   When politicians from both parties get together to pass an unpopular measure, people can feel like it doesn't matter who they vote for, politicians will just do what they want.   In this case, that feeling was aggravated because Bush had made a public declaration that he wouldn't raise taxes and would veto any Congressional attempt to do so.  Often public discontent passes, but this time it seemed to linger.  Perhaps that was for specific reasons (Bush foolishly making an explicit pledge) or perhaps it was because the public had become less inclined to defer to political leaders. 

*Technical appendix:  I started by including the change in unemployment over the last year as well, but it was nowhere near significant, so I dropped it.   There was clearly serial correlation in the residuals, but I didn't try to control for it.  However, because of the uneven spacing of observations, I weighted them by the time from the previous observation--e. g., if the previous survey had been a year ago, the observation counted 12 times as much as if it had been a month ago.