He used two measures of income mobility. The first measured what percentage of children made more than their parents had at the same age. The other was a measure of what percentage of children born in the bottom 20% made it to the top 20%. Those seem to me like very different measures.
Doing Better Than Your Parents
90% of the people born in 1940 were making more at age 30 than their parents had at age 30. That is a really clear measure of progress: the next generation is more affluent than the last. But as you can see in this graph, that percentage drops sharply until about 1960 and then continued to drop, albeit more slowly, up to the point of people born in the 1980s. Roughly 90% of 30 year olds in 1970 were doing better than their parents had at 30, but by 2010 only about 50% were. That seems alarming but I don't think it's as bad as this first graph looks.
First, there are adjustments that Chetty himself makes.
Adjusting for inflation across generations is not trivial. How do you properly adjust for the price of a mid-sized sedan in 1970 and 2018? The first might reach 100,000 miles and the second might reach 200,000; the first has seat belts and the second has air bags. A TV in 1970 might have been 15" and offered 3 channels; a TV in 2018 might be 50" and offer 300 channels. We could contrast a list of products like this, nearly all of them showing a similar uptick in quality that makes price adjustments tough.
Even assuming that inflation adjustments let you accurately compare incomes from 1970 and 2010, families are smaller. If you make $28,000 with a family of four in 1970 or $27,000 with a family of three in 2010, is your family income really lower? In this example, family income has dropped by $1,000 but income per family member has actually gone up from $7,000 to $9,000.
If you adjust for inflation and family size, the downward trend is less severe. About 95% of 30 year olds in 1970 were making more than their parents at the same age and that dropped to roughly 70% by the 1980s (not a mere 50% as suggested before making these adjustments).
A drop from 95% to 70% of the next generation doing better is not great but even that is arguable. The average person in 2010 had a library of on-demand articles, books, songs, movies and TV shows that dwarfed the choices of even the richest people of 1970. House prices have gone up but so has the average square footage of homes and the quality of appliances within them. We have a wealth of choices at the grocery store and in 1970 not only did you have just a couple of tomato sauces to choose from in the store but it was tough to find good sushi, ramen or falafales in most of the country. People in 2010 had more choices about how to live their lives than people in 1970 and not all of that can be properly captured in income statistics.
One last thing? The Great Recession was awful. Between 2000 and 2009, the economy destroyed a million jobs. In the 1980s (and 1990s and probably 2010s), the economy created roughly 20 million jobs. Any comparison of how people in 2010 are doing with people in 1990 has to account for the terrible shock that was the Great Recession. All else being equal, we would expect to see a downturn in the percentage of people in 2010 who are doing better than their parents did at 30. Millennials - like the rest of us - had to learn how to swim. Unlike us older folks, they had to learn how to swim in a tsunami and because of that careers were slower to take off and that could not help but be reflected in these numbers. I suspect that as we get further from the recession, this measure of what percentage of 30 or 40 year olds are doing better than their parents will rise.
Doing Better Than Your Peers
What about Chetty's other - very different - measure of income mobility? What percentage of people born in the bottom 20% make it to the top 20%?
Let's explore this number a little. If parents made zero difference, we would expect that any kid born in any part of the distribution could land in any other part of the distribution by the time she's an adult. Maximum mobility means that there is no correlation between where you start and where you land. 20% of the kids born in the bottom 20% would make it to the top 20%. 20% of the kids born in the bottom would land in the middle. And 20% of the kids born in the bottom 20% would end in the bottom 20%.
This measure is zero sum, though. Every one percent of the kids who move out of the bottom 20% displace someone else. No matter how much your economy grows or stagnates, there will never be more (or less) than 20% in the top (or bottom) 20%.
It is true that perfect income mobility by this measure means that a kid born in the bottom 20% is just as likely to end up in the top 20% (or middle 20%) as she is the bottom 20%. It is also true that any kid born in the top 20% of income distribution is just as likely to end up in the top (or middle) 20% as he is to end up in the bottom 20%. Perfect mobility means that parents make no difference. That's certainly not the case now. Chetty shared a remarkable statistic: kids born into the top 1% of households (those with incomes of $650,000 or more), were 80X more likely to be admitted to Stanford than kids born into median income households.
What I find curious about this measure of income mobility is that if Chetty could convince CEOs, mayors, senators and tenured professors to pursue policies that would lead to perfect mobility, it means that the children of those policy makers would be just as likely to end up in the bottom 20% as in the top 20% where they started. I find it hard to imagine many of these leaders willing to adopt policies that allow for perfect income mobility by this measure. By both absolute and relative measures, affluent parents like the idea of their children doing well.
That said, it does seem like a healthy community would allow for children born in poverty to rise to the top and for children born rich to fall into the middle or bottom based on their own - and not their parents' - merit.
There are very real differences in a communities' ability to raise a child born in the bottom 20% up to the top 20%, from poor to affluent. Segregation is one big reason for this. Atlanta and Sacramento have the same percentage of blacks, whites, and Hispanics but Atlanta is much more segregated. (Whites live in one part of town, blacks another, etc.) A kid born in the bottom 20% has a 10% chance of reaching the top 20% in Sacramento; in Atlanta that kid's chances are just 4% and this seems illustrative of what Chetty sees across cities in the US. Segregating people by any grouping - education, race, income - seems to result in less income mobility. (And, as seen in other research, this ability of a community to expose its kids to a variety of other people seems to raise income and wealth for everyone.)
Entrepreneurship and innovation also seems to matter. If your community is creating new jobs and wealth, your kids have a better chance to rise. San Jose, San Francisco and Seattle are among the best communities for giving kids a chance to rise to the top; Cleveland, Detroit and Atlanta among the worst. (See one comparison of communities here.)
Finally, one of the most fascinating points Chetty made was merely implied. Some communities do a much better job of creating opportunities than others. A poor kid growing up in such an area has double or triple the odds of becoming affluent. I don't know enough about the data to conclude this but the impression I was left with is that spending money to get your kid into an innovative, integrated, affluent neighborhood will do more for her prospects than using those same dollars to get her into a better university. Geography is culture, and culture matters.
I'd be fascinated to know more about the differences in communities that are more effective at letting you do better than your parents but curiously, most of Chetty's research focused instead on the differences in communities that raised the probability of poor kids becoming rich. Given there will only ever be 20% of the population in the top 20% but 100% of us could be doing better than our parents, the latter seems like a goal that is easier to align around and more effective.