The geography of intergenerational mobility in Canada and the United States

This is Lecture 8 of the course ECON 85600, “Inequality, Economic Opportunity, and Public Policy,” that my class and I are now conducting online. You are welcome to participate, and can review all the course materials at .

Warning: this is likely to interest social scientists in sociology, economics, or other fields, interested in developing a specialized knowledge of the subject!

This presentation summarizes research on comparing social mobility in Canada and the United States that finds intergenerational income mobility is lower in the United States than in Canada, but varies significantly within each country.

The full reading list and access to other papers are on the page devoted to this lecture at /

The Canada-US border divides central and eastern Canada from the US Great Lakes and northeastern regions. At the same time some Canadian regions have more in common with the low mobility southern parts of the United States than with the rest of Canada. The fact that these areas represent a much larger fraction of the US population also explains why mobility is lower in the United States.

View this lecture in conjunction with your reading and of the Quaterly Journal of Economics paper published in 2014 by Raj Chetty, Nathaniel Hendren, Patrick Kline, and Emmanuel Saez called “Where is the Land of Opportunity? The Geography of Intergenerational Mobility in the United States.

Your reading is supported by the associated student presentation.

Be certain to leave a comment, question, or concern in the “What do you think?” box at the very bottom of this post. Frame your feedback in a way that is of benefit to the learning environment for all students, and don’t hesitate to raise a question of clarification if you don’t understand an issue.

As a bonus, and for your information here are links to a series of posts that summarize the findings of the Canada – US comparisons, and place them in a policy relevant context:

If there is such a thing as the “Canadian Dream,” it would look very much like what Americans say is the “American Dream”

Intergenerational mobility between and within Canada and the United States

The “middle class” is within easier reach for low income Canadian children, than it is for low income Americans

Equality of Opportunity is a Choice


17 thoughts on “The geography of intergenerational mobility in Canada and the United States

  1. I’m curious about the variation in intergenerational mobility that we see within the US (motivating picture three), but specifically within states. Montana, for example, has two counties on the northern border of the state with especially low levels of mobility. Is there research that explores within-state/cross-county variation? On a state to state level, I can imagine policy being an influential factor in differences, but I imagine that’s not the case on a cross-county level. Can the differences within states be accounted for by population density, main economic activity, or are there other factors at play?

    1. Thank you for this. As you can imagine the Chetty, Hendren et al paper has spurred a whole host of new studies. In part this is because they made their data—not the individual-level tax data that they were restricted to use inside a secure government facility, but data on the tabulations of mobility statistics by geographic area—publicly available. They and their colleagues have set up a nice website called . The data for the paper we read is available at: . My co-authors and I used some of this information to research and write our Canada-US paper, and many other researchers have and currently are using the information for the kind of more detailed studies you are highlighting.

      There are other more detailed analyses on this website for geographies more narrow than the Commuting Zones that were originally used, so you may uncover even more insights that raise questions for research and policy. I can’t speak to the specifics that you highlight, but I have seen studies that focus more closely on some of the correlates outlined toward the end of the Quarterly Journal of Economics paper. There are papers on the role of education and access to education in explaining the differences across the country, and at the other extreme, papers that look at historical demographic trends, like variations in the incidence of slavery in some of the southern states and how they relate to current rates of upward mobility.

      I hope you have the opportunity to explore the Opportunity Insights website, and it may trigger some research ideas, but I’m not really in a position to give a specific answer to your last question. I hope nonetheless that this helps a little bit. But certain the tone of your comment is exactly where we want to go … to start understanding some of the causal forces. In fact, this is a topic we will discuss next day when the student presentation examines the causal influence of nieghborhoods.

      Feel free to let me and the other other students know what you think.

  2. Can one infer from the data in the table showing correlation between community characteristics and mobility that you are recommending that public policy (1) focus on school retention, (2) provide more support (child care?) to single mothers, and (3) stimulate jobs that can be held by those who would otherwise hold manufacturing jobs, and that (4) these be relatively more focused on the black population? Also, the is negative correlation of the teenager in labor force suggesting that public policies that promote summer jobs are misguided or rather that teenage labor is an indicator that these teenagers are not in school or not as able to focus on school work because they are forced to work? If the latter, is there data about internships and summer jobs characteristics and if so how does it correlate to mobility?

    1. We have to be careful in making these inferences. There are still a couple of steps to take before we can move to causal inference, and then to policy recommendations.

      The table is only a starting point for this type of conversation, though certainly intended to inform such a conversation.

      The information refers to the one-to-one correlation between each of these variables and the variable representing mobility. So a lot of other things could be going on that could possibly lead to correlations being spurious. The information raises hints and offer some directions on next steps, but it does not offer supporting evidence for particular policy recommendations.

      As an illustration, consider the topic of next week’s student presentation on the role of neighborhoods. The interesting puzzle, from a scientific point of view, is whether the characteristics of neighborhoods themselves cause child outcomes, or it is because people with certain unobserved characteristics tend to live in the same neighborhood, with these other things being the real drivers. We need what economists call an “identification” strategy to be able to pull out the causal impact.

      An identification strategy refers to using plausibly exogenous variation in a potentially causal variable to discern its impact on the outcome, and there are a host of empirical methods available, but they all rely on us having detailed enough institutional knowledge to be able to discern changes that we might consider exogenous.

      Sibling differences are one of those methods, and they will figure prominently in the next paper we read, that I believe Max will present next week. This will give you a sense of the challenges and the opportunities, and a sense of why we should be cautious in making causal inferences from the kind of descriptive correlations in the table you reference.

      Does this help?

      1. It does help, and I appreciate the reminder to take an economist’s rigorous approach. Thank you Miles.

  3. It’s interesting to think about this study in light of the current pandemic. Countries the world over are facing economic shocks and are dealing with them very differently. It will be interesting to see if the border between Canada and the US becomes more relevant to intergenerational mobility as our relative governments step in (or don’t) to help individuals recover from the economic downturn associated with the outbreak of COVID-19. Additionally, a measure missing from this model is wealth. We know that wealth and income are highly correlated, however, wealth often allows individuals to withstand short-term economic crises. How will wealth affect intergenerational mobility, both in light of the current financial crisis, and also in times of relative economic security on a national level?

    1. Interesting, and good point about wealth … something we should always keep in mind. I know that Chetty et al tried to get at social capital in their analysis of the correlates, but it was sort of rough and ready. But maybe that factor will come more to the fore in these challenging times.

  4. As we look at the distribution of mobility across space in a number of measures both in Prof. Corak’s paper and the work by Chetty, it is clear that there is a high mobility cluster in the center of the United States. The region included in that rough cluster has some cities but it also home to much rural and sparsely populated land. I wonder if some of this mobility might be driven by the fact that in smaller communities there may just intrinsically be less space for stratification as in more densely populated areas. A small town may only have one high paid doctor or real estate magnate, but they also may have few low-paid service workers because there aren’t as many people to serve. It seems certainly preferable to have more mobility than less, but I wonder if the implication of being able to rise to the top 20% in a dense and competitive city carries a different qualitative meaning than doing so in a more egalitarian rural community.

    1. That is a cool idea, and may be one of the factors underlying the Great Gatsby Curve drawn with the sub national data. If your hypothesis is correct, then it would be the case that inequality is correlated with population density, and that in certain communities of lower density cluster in the lower part of the Great Gatsby Curve cluster in the picture I presented. That is something a researcher could check out with the data, and maybe build something around.

  5. Since I was responsible for making presentation slides on “Geography of Intergenerational Mobility”, I wondered as to why it is important to study this issue at all. I believe if we are studying an issue, it is imperative that we are convinced that the issue is worth studying in the first place. We want to live in a society where each person is given an equal chance of making it big. I hope this video convinces us all to at least move in that direction. I hope you enjoy the video and use it in your class wherever you are teaching. Here is the link to the video:

  6. Chetty et al 2014 assign children to CZs (Community Zones) based on where they lived at age 16 i.e,
    where they grew up- irrespective of whether they left that CZ afterward. So, my question is if a child grew up in Utah and went on to work in New York having moved upward on the “social ladder”, wouldn’t it contribute to the overestimation of their (Authors) measure of IGM in Utah?

    1. I think you have to interpret more in the spirit that I think they intend. That whatever it is about, say Utah, it is that children are raised to seize opportunity, wherever that opportunity arises. And in this sense we should appreciate that geographic mobility is a part of human capital that fosters mobility. So there is another research agenda here on the importance of geographic mobility in promoting mobility.

      Does this help?

  7. Hello professor, I hope you are doing fine in Canada. I do not have a question but a comment I may say:
    I really found very interesting that in the public opinion poll you have, race and gender have really small roles as perceived drivers of mobility. I am curios about the participants of the survey( well, you said you have the details in your page so I will go look there). And the second interesting thing is both Americans and Canadians have similar perspectives about these drivers and also the definition of American dream, yet the results about mobility and inequality are really different in those countries. Such as Americans have less mobility comparing to Canadians.
    I am starting to think if people on average may not have a good understanding about the policy results (or what those policies may bring in the future) in their countries (as I said, not all of them but on average).

    My regards


    1. Yes, I think your intuition is leading us in the right direction. Certainly, a subtext of our Canada-US paper is that race is probably an important reason why the degree of social mobility is lower in the United States. Other research backs this up. And this is to say nothing about gender. So it is interesting that there is such a disconnect between public opinion, and some of the underlying forces that we know impact on mobility. This would be an important area of research, and I do know that there is a literature developing (and even developed) on perceptions of inequality and actual inequality, though I don’t know if has extended to inequality of opportunity.

  8. Hello Professor,

    While you were discussing the relationship between mobility indicators and community characteristics I started to wonder the relationship between mobility indicators and industry characteristics of the regions identified as having similar mobility characteristics. Can we observe some relationship between mobility characteristics and the prevalent industry of the region?

  9. Hi, hope you are all well.
    There is an apparent correlation between population density and income mobility in the US, i.e., the coastal areas in the US are more densely populated and are, with the exception of south California, also areas with lower intergenerational mobility, while inland low densely populated areas have higher mobility.
    That correlation is, apparently, reversed in Canada, i.e. the more densely populated areas, the ones closer to the US border, are also areas with higher intergenerational mobility.
    I find that surprising, and wonder what could justify that inversion of the correlation. Maybe an answer can be found along social geography. I mean, coastal areas in the US are racially more diverse than inland areas, and maybe they also changed as a reaction to the Great Migration, as Ellora Derenoncourt explained us about. However, that explanation seems to not work for Canada, since southern regions of Canada are more ethnically diverse.
    Another hypothesis may have to do with the economic history of settlements and indigenous persecution. In the sense that inland lower densely populated areas of the US are mainly populated by white people that descend from settlers, while lower densely populated areas in Canada are mainly populated by indigenous people.
    (I am not searching for exact numbers to support these statements, so pardon me for any inaccuracy)
    Do you have any thoughts on this demographic/geographic dimension of the issue?
    Thank you,

What do you think?

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.