Thursday, May 5, 2022

It's not (just) the economy, stupid, part 2

 In my last post, I tried to explain why satisfaction with the state of the country declined from about 1986 to 1992, despite decent economic conditions and very positive developments in foreign affairs.  Now I'll turn to the rising satisfaction from 1992 to the late 1990s.  The obvious answer would be that economic conditions were good, but as I noted in my last post, there's still a substantial positive trend after you control for the unemployment rate.  Of course, there are other economic conditions that might matter, but  the three that generally have the most impact on public opinion are unemployment, economic growth, and inflation.  I controlled for recent change in unemployment, which is strongly correlated with growth, and that didn't add anything; inflation wasn't much of a factor in the 1990s.  So I'll turn my attention to other things.  I proposed that events in national politics might explain the declining satisfaction in 1986-92, but I don't think that they could explain rising satisfaction in 1992-9.  The big events were conflict between Congress and the President, including the government shutdowns of 1995-6 and the endless investigations of Bill Clinton culminating in the impeachment effort of 1998-9.  So overall, the politics of the 1990s weren't inspiring and nobody came out of it looking good.  

Then what could explain the positive public mood?  In their analysis of the data, Andrew Gelman and Yair Ghitza noted that their was a correlation between satisfaction with the direction of the country and satisfaction with one's own life.  It seems plausible that at least some of this is causal going "upwards"--if things around you are going well, you'll feel good about the direction of the country.  One of the things that changed in the early 1990s was crime, which began a long decline.  I regressed satisfaction on unemployment, the murder rate, and a time trend, and all were statistically significant.*  The observed and predicted values:

Adding the murder rate doesn't fully account for the rise in the 1990s, but it helps.  More generally, I think that in the 1990s people began to adjust to the social changes of the 1960s and 1970s.  Their seemed to be less tendency to regard single parenthood as a moral failing and as crack use declined, less alarm about drugs.  The Aids epidemic started to come under control, with effective treatments starting to appear and the number of new cases peaking in the early 1990s.  As a result, people became less alarmed about the overall direction of social change. 

Of course, this is an ad-hoc analysis--I've proposed two different things to explain different periods and I'm still in the 20th century.  But while parsimonious explanations are intellectually satisfying, sometimes things really are complicated.

*I decided that I should make some effort to take account of serial correlation.  To my surprise, I couldn't find any software for dealing with serial correlation in irregularly spaced time series, or even much literature.  So I began by assuming a first-order autoregression, used non-linear regression to estimate the autoregressive parameter, and then computed generalized differences, with the degree of differencing depending on the gap from the previous observation. 

 

 




1 comment:

  1. As I recall, the 1990s were economically good for lower income people, and the data at Realtime Inequality confirm this: post tax average income in 2021 dollars for the bottom 50% went from ~$24k to ~$28k.

    ReplyDelete