Raj Chetty’s empirical restructuring of Harvard’s undergraduate economics

Here is good coverage from Dylan Matthews, here is one excerpt:

[Chetty’s] Ec 1152 is an introduction to that kind of economics. There’s little discussion of supply and demand curves, of producer or consumer surplus, or other elementary concepts introduced in classes like Ec 10. There is no textbook, only a set of empirical papers. The material is relatively cutting-edge. Of the 12 papers students are required to read, 11 were released in 2010 or after. Half of the assigned papers were released in 2017 or 2018. Chetty co-authored a third of them.

Why not excerpt the cameo of me?:

…fellow traditionalist Tyler Cowen…told me he’s excited about the class. “I am for experimentation, and more of it in academia, and for that reason I approve,” he writes. “Of course it was not what I do, which is more traditional micro, more theory, less overlap with sociology. If the instructor is great, that is really what matters.”

There is much more at the link.  And here is Daniel Simonsen, Norwegian comic.


Chetty is good, but 1/3rd of the reading list authored by the instructor is a titanic red flag.

Or makes the course self-recommending.

Wonder what Mankiws reading list looks like.

So yeah he uses his journal articles instead of Mankiws $280 textbook.

Hi, mouse!

So you think there's only one social liberal who posts here. Fine. Hi Ed!

You are obviously posting under many pseudonyms. So again,

Hi, mouse!

'Why not excerpt the cameo of me?'

And to think that some people accuse economists of lacking humor.

However, it is a bit strange to hear one of the founders of MRU (with its mythical origin story involving two professors, youtube, and a $4 app) as 'fellow traditionalist Tyler Cowen,' Particularly as it seems to have been years (decades?) since Prof. Cowen has taught an undergraduate introductory econ class.

And honestly, reading current research in a class is not exactly ground breaking, though distinctly unusual in general introductory classes (which this may not be - often, there are general introductory survey classes that are considered unsuitable as introductory for students in that major).

I find Chetty's data-gathering quite extraordinary, although the insights he pulls from all those numbers tend to be dull.

Here's my analysis of Chetty's 2015 dataset on transgenerational income mobility by county, which I think many people would find more interesting:


I actually read your article.
Prof. Chetty might do well to collaborate with a kibitzer rather than rely too extensively on his own cross-cultural insights.
i'm probably racist for suggesting this.

Chetty's amazingly good at getting access to data, such as tax returns, that the government promised would be kept confidential. I love his databases and have written many thousands of words analyzing his data. I find his data pretty reliable. For example, in his study of 2015 intergenerational income mobility, the worst county in America for working class parents to raise their children in is the tragic Pine Ridge Sioux reservation, which sounds awfully plausible.

The best county, according to Chetty's database, is Sioux County, Iowa, which has the highest raw farmland prices in the Midwest, is the most Dutch county in America, and one of the most Protestant. Sioux County really is very close to the best farm county in the Midwest. So that also sounds plausible.

On the other hand, Chetty's not very good at noticing interesting patterns in his data. He tends to come up with dull platitudes, which I guess is safe for his career, but kind of shame when you consider how rich his databases are in insights.

I encourage anybody who likes to analyze social statistics to check out his databases at:


Thanks, I followed up on your links

It's interesting to read this post in conjunction with Alex's about Malibu. Sailer takes aim at liberals in Malibu, but seems reluctant reach similar conclusions about Sioux Co., IA. There's no Martin Sheen, but good luck buying farm land, especially if you're not Dutch Reform. Even Devin Nunes's family had to settle for the next county over: https://www.esquire.com/news-politics/a23471864/devin-nunes-family-farm-iowa-california/

"but seems reluctant reach similar conclusions about Sioux Co., IA"

Umm, why would he. There's no shortage of land to buy in Sioux Co and it's far, far cheaper than Malibu.

"In late 2011, some farmland in Sioux County sold for a Midwestern record of $20,000 per acre"

Now reflect on the incongruity between record prices and your claim that "[t]here's no shortage of land to buy."

Feel free to scour online and see what you can find in terms of sales in Sioux Co. the last 3-5 years. Especially note the size of the parcels transacted.

There's no incongruity. You certainly won't find land in Malibu for $20K an acre. It's much more expensive than that.

Are you seriously trying to say that it's harder to find land in Sioux County IA than in Malibu, CA? Because frankly that just makes you look stupid.

Here's a house and 18 acres for $335K:

"Feel free to scour online and see what you can find in terms of sales in Sioux Co. the last 3-5 years. Especially note the size of the parcels transacted."

Next time, you should probably take you own advice before you post it. It would save you some embarrassment.

You clearly don't understand agriculture. You provide a listing with 10 acres of farmable land. Your link is to an acreage, not a farm. How many half sections, even quarter sections, get sold publicly? If you need more than both hands for the last three years, I'll be stunned. I don't live in that area anymore, but pick a county in that part of the state and look over the plat books from every 10 years for the last 50 years or so. The only time you'll see property broken into smaller holdings in when multiple children with their separate trusts inherit part of a farm.

My point is that starting a new farming operation in Sioux Co. is about as difficult as opening the Whole Foods in Malibu that was referenced in Sailer's article (that Alex linked in his post). You can find that difficult to believe, but an 18-acre homestead for sale is not evidence to the contrary.

It would be trivial to buy a farm in Sioux County. There's no legal impediment to starting a farm. Whereas, you can't open a Whole Foods in Malibu even if you already bought the land.

"Feel free to scour online and see what you can find in terms of sales in Sioux Co. the last 3-5 years. Especially note the size of the parcels transacted."

Here is a list containing an 80 acre farm that sold in February, a 71 acre farm that sold in March and a 122 acre farm that sold in April.

"You clearly don't understand agriculture. "

No, this isn't about my knowledge of agriculture. You are just making a bad argument. There's nothing wrong with just admitting your original argument was wrong.

I don't deny the legal/regulatory differences. However they're both form of opportunity hoarding.

If you want a more exact comparison, the rural version of Malibu liberals is Monsanto. If the neighbor uses Roundup Ready soybeans and you don't, well, good luck getting any compensation for the effects of dicamba drift.

Or, it's large-scale hog operators getting the county to loosen requirements on where the large confinement buildings can be located so that neighboring homes are now regularly subjected to a very unpleasant smell.

Should we just apply the Coase theorem? Isn't Rob Reiner just applying the Coase theorem by other means?


As for the listings you provided, none are even a quarter section, but I appreciate you being seriously engaged enough to put in the effort.

Thanks. This sounds like an interesting topic that you clearly know a lot about. Can you summarize your ideas about what is going on with land sales in Iowa and why?

By the way, here's my 2012 column about The Edge of U2's attempt to build 5 houses on 156 acres he bought in Malibu in 2005:


The battle was still ongoing in 2016, but I don't know what has happened since then.

Thanks. Informal restrictions on land sales in Sioux County sounds like an interesting topic for an economist to research. Off hand, I would think that they would hold down land prices, but instead Sioux County was the first in the Midwest to have land sell for $20k per acre.

Maybe it's like Oak Park, IL, where a covert and almost certainly illegal racial quota imposed on real estate agents in the 1970s (only one black per block allowed), saved the city's property values from the disaster that overtook the next door neighborhood of Austin on Chicago's westside?

But I don't know anything about farming, so I've never thought about the topic.

No fair! I'm calling the Office of Fair Housing or whatever it's called. Seriously, Oak Park is still a really nice place where black and white middle class people can live comfortably together. Austin is godawful. It worked.

I prefer the Sumner approach of teaching students what they don't want to hear by relying on time-tested economics. Chetty seems to be doing the opposite using newish papers that haven't had enough time to generate the validation that comes with time.

This is a good point. In psychology, especially social psychology, a lot of the major discoveries in the field from the late 80s, 90s, and 2000s that became trendy insights, even for policymakers and corporations, have shown to be mostly bunk incapable of replication. I worry that Chetty's work, because of popular concerns about inequality post-Occupy, is being too eagerly lapped up.

Keep in mind that Chetty's databases have unbelievably colossal sample sizes: they are often fairly close to being a universe rather than just a sample of the relevant data.

Agree. The new papers are not Lindy, and therefore likely are giving false info to the students.

This is just another example of "Average is over". Chetty can really excel at teaching his own work (and collecting data about it) and style of Econ. If students can't get a micro 101 via youtube or downloading a textbook then something is wrong. We should end up at the point where the best deliverer of Micro 101 in the world delivers it to the world, and in person econ instructors either offer their own take or focus more on interactive teaching (identifying misunderstandings etc).

Good article except he leaves out one salient fact. Harvard does not have an undergraduate business program. So most of those aspiring business types do Econ as their undergrad major. No wonder a 100 level big data Econ course pulls in a lot of students. Still, an interesting idea to offer it.

Right. A lot of what Chetty does is closer to my old career, market research, than it is to economics. For example, Matthews writes:

"big chunk of Chetty’s lecture focused on his paper with Adam Looney and Kory Kroft on sales tax salience, the one for which they plastered new sales tax-inclusive price tags on hairbrushes and makeup at a number of California grocery stores. ... But most of Chetty’s discussion of the paper was about his methodology, what’s known in economics as a “differences in differences” approach. The key was to compare how sales of unaffected products in the stores changed from the start (26.48 sold per week) to the end (27.32 sold per week) of the experiment to how sales of affected products with the new label changed: from 25.17 per week to 23.87 per week."

That's exactly what I was doing for a living at a marketing research firm in the early 1980s: running test markets with a test cell and a control cell and reporting results like that. (We actually had a far more sophisticated system for running test markets in 1982 than Chetty has today.)

Probably why I'm so much better than Chetty at making sense of Chetty's own data is my long experience in American market research gave me a toolkit for thinking about his data superior to what economics provided Chetty with.

There is reason to think a representative agent may be doing market research, to some extent. Getting a job is a lot like market research, even traditional economists have acknowledged that. Find the best schools is a market research. In neither case do traditional supply and demand apply, instead we get representative agents fitting their portfolios to a potential new city. And that search problem boils down to solving a queuing problem, avoid long waits in line.


I have this vague hunch that the various social sciences tend to turn into market research in the long run, perhaps because corporations will pay for market research. For example, a lot of Gladwell Era psychology, such as "priming" research, is more or less marketing research -- Can we induce a desired behavior in people by clever propaganda?

Of course, priming research is notoriously bad at replicating. That seems to be the main difference between Science and Market Research: the former is supposed to replicate now through forever, while the latter isn't.

The great thing about market/marketing research is you don't market it as Science which therefore will always replicate.

Instead, you just market your results as being accurate right now for the moment, but the findings could change at anytime, so you'd better keep paying us money to do more research from now on. If you can get a monopoly on tracking some product, the way Nielsen has had a monopoly on TV ratings, it can be a long-term gold mine.

The standard intro economics course provides the most useful and durable tools to analyze economic and policy issues, so his course should serve as a complement rather than a substitute for the intro course.

Right, Econ 101 Microeconomics is, if my undergrad experience was typical, a great course.

I'm all in favor of smart Harvard kids taking Chetty's course in, what is in effect, Quantitative Methods for Market Research, but it's not a substitute for microeconomic theory, just as it's not a substitute for Philosophy 101 or for English 201: Shakespeare.

That's kinda what I thought, too.

Chetty should have me come to Harvard and give one lecture to his students on my tips and tricks for analyzing Chetty's databases. For example, I always look closely at whatever Chetty has ranked as the best few dozen and worst few dozen examples in America, and read up on them in Wikipedia, looking for patterns. For example, in 2015, I
walked through the top and bottom 50 of his rankings of 2,478 counties. When thinking about Big Data, long found it extremely useful to look at the highest and lowest examples in detail to see what kind of patterns leap out. It’s extremely easy these days to look up facts about outliers, so more people should do it. This doesn't seem to be a common practice among academic data analysts, however, who evidently fear contamination by bias and stereotypes. But instead they wind up suffering from ignorance, which is worse.

For example, Chetty's list of best counties for working class parents to have lived in in 1996-2000 so that their kids would be well off in 2011-2012 was dominated by rural counties in the Great Plains and Great Basin. I hadn't realized how well these parts of the country have done in this century, in part due to fracking, but heavily due to China's demand for resources (just as Australia has similarly prospered). But once I read up on Sioux County, Iowa and several dozen similar counties at the top of Chetty's ranking, the pattern was obvious.

The bottom counties for income mobility were dominated by counties full of Indians, Eskimos, and blacks. Why? Because, while everybody regresses toward the mean, poorer groups regress toward poorer means.

Looking at the worst white county, Horry in South Carolina, was revealing too. Horry, home of the giant Myrtle Beach golf resort, was booming in the 1996-2000 control period. But then the Golf Recession arrived in 2001 and persisted through Chetty's 2011-2012 test period.

In other words, Chetty's data advises working class parents to try to anticipate long-term swings in supply and demand that if they could predict them accurately, they'd have been working for the Renaissance hedge fund rather than driving a bulldozer building yet another Myrtle Beach golf course that would soon be empty.

Teaching economic lessons to this generation has very broad political, and ideological, implications. The concern is that millennials and this newest generation are being drawn to socialism as the result of the great recession, excessive inequality, and the lack of opportunities, especially for millennials. How does one opposed to socialism combat this trend? Blame baby boomers. That's what Joseph Sternberg does in his new book. How is it the fault of baby boomers? Because they accepted a middle course between markets (i.e., libertarianism) and government regulation, producing the worst of both. The cure is to get rid of regulation and rely on markets. Of course, the tune of markets (libertarianism) is music to our hosts' ears. Sternberg's trick is to pick a target to blame, his target being the always unpopular baby boomers. Clever. As Trump so ably demonstrates, picking one's enemies is the first step to winning over a crowd. https://www.vox.com/2019/5/22/18617686/baby-boomers-millennials-capitalism-joseph-sternberg

Good. May be the next generation of economists can do better than the failures that made up the last batch.

That is doing economics with our abstract algebra tree. Big data is making the tree trunk rounder, the hologram more accurate.

Talk about Chetty's 'family' mobility study. The task is to find a representative agent virtual balance sheet, which finite, and small, set of concerns match the collected IRS data. Jobs, schools, access to city, housing costs, weather. There is a much larger set of concerns Chetty has to check, and then select the finite 'family' as a virtual balance sheet.

Chetty, the agent, and the market researchers are all doing the same, finding a tree trunk that is round and creates the hologram effect, and that becomes a queuing problem, keep the flows in and out of the virtual balance sheet at optimum congestion.

What is the hologram? Velocity equations work. The 'family' can schedule events and transactions necessary to life with low probability of being stuck third in line.

The big data solution has two outcomes, there is a small set of portfolio concerns that match the data well, or the null hypothesis applies.

Perhaps if students (and everyone else) had better means to control the supply and demand of their time management priorities in relation to each other, supply and demand would still be central in these discussions.

"He also gives Mankiw credit for moving the curve of Ec 10 to match the curves of other large Harvard classes, based on research showing that unnecessarily tough grading of economics classes disproportionately discourages women from taking them."

Great, let's increase grade inflation for equality's sake

Comments for this post are closed