Now I've found one:
The horizontal axis is year of birth, and the vertical axis is the percentage of people born in that year whose income exceeds their parents' income. The graph says, basically, that when viewed on a generation-by-generation basis, American income has flat-lined; and if the overall probability is 50%, a substantial number of young people are in fact making less than their parents did.
I suspect this has much do to with the loss of high-paying blue collar jobs — many of them in the manufacturing sector — and their replacement by low-paying jobs in the retail sector. The push for a higher minimum wage is a strong indicator of this replacement; in the past, the minimum wage was not a factor in as many households as it is now.
The phenomenon is worse in some states than others, and in particular it's bad in the so-called Rust Belt states:
Bear that in mind as you look at this somewhat-truncated map from TIME magazine:
In much of the country, like the South and the far West, there wasn't much difference between 2016 and 2012. But wow, look at the region from Minnesota-Iowa-Missouri eastward to Pennsylvania and upper New York state. To my eye there's a strong correlation between those counties in dark red and the regional variations in the study that Brookings describes. Correlation is not causality, but if I had to place my bet on why Trump won, I'd choose this explanation over any other that I've seen.
My hypothesis is simple: people who are anxious or angry about the economy will voice that anger at the polls, regardless of any other factors. It happened to George H.W. Bush who fell from an 89% approval rating in February 1991 to 22% in July 1992 — opening the door for Ross Perot, the forerunner of Trump, to have his 15 minutes.
What can Trump actually do to improve the economy? Wait for Part II.