Confidence and Sentiment

FRED has two datasets that I want to look at today: Consumer Confidence, and Consumer Sentiment. Here's the default view:

This Graph at FRED: https://fred.stlouisfed.org/graph/?id=CSCICP03USM665S,UMCSENT,

It doesn't make a pretty picture.

I want to compare the two datasets. Some of the early years' data is intermittent. I can make the red line start at an earlier date by changing the data from monthly to annual. Since I want to compare them, I will change both datasets to annual.

Mimicking a project I worked on, egad, eleven years ago, I want to center each dataset on the zero axis, by subtracting the series average from each value in the series. Centering both datasets at the zero level centers them on each other, and makes them easier to compare.

You can see that the red line has much greater up-and-down spread than the blue line. That up-and-down spread is measured by the standard deviation. I will take the datasets (after the subtraction) and divide each one by its standard deviation. This makes the up-and-down variation the same for the two datasets. So now they're centered, the one on the other, and they are the same height. That should make them easy to compare visually.

Next, I want to smooth out some of the jiggies. Changing the monthly data to annual helped with that, but not enough. So I'm figuring each dataset as a 5-year moving average, with the data plotted at the midpoints of the 5-year periods.

That's it for data manipulation. Here is the result:

The two datasets are now so similar that they almost look identical. This seems odd to me, given that the data measures people's feelings. I will evaluate the result anyway.

For both datasets, the high points occur in the early 1960s, the mid-1980s, the late 1990s, and the Trump years. The low points occur in the 1970s, the early 1990s, the years around the financial crisis, and, well, the Biden years.

So now we know why the Trump years are so fondly remembered.

Now that I know what to look for, I can see it on the first graph, too. There is a big increase in the red line (consumer sentiment) in the latter part of 2014: From barely above the 80 level, it rose to near 100. That was followed by a second jump, a rebound from just below the 90 level in October 2016 to just below 100 in January 2017. That gives us the high in the Trump years.

There was a similar sharp rise in 1983 on the first graph, from near the 70 level at the end of 1982 to above the 100 level in early 1984. That increase arose, I think, from the boost Reagan's charm gave to the recovery after the 1982 recession.

The high of the latter 1990s was different. There was gradual increase from the early-90s low: gradual increase, rather than sudden increase. Consumer sentiment in the 1990s rose along with the improving economy. 

The economy doesn't improve in a sudden blast. If consumer sentiment does, it has more to do with "charm" than with economic performance.