Interview with Josh Angrist

From the Richmond Fed Bulletin:

EF: You’ve looked at the question of how much peers matter. Many parents obviously seek schools where they believe their children will have higher-quality peers, whatever they may mean by that term. You and your co-authors have looked at Boston and New York City selective public schools, and you concluded that peer effects don’t seem to matter much. Why is that?

Angrist: I’m always beating that drum. I think people are easily fooled by peer effects. Parag, Atila Abdulkadiroglu, and I call it “the elite illusion.” We made that the title of a paper. I think it’s a pervasive phenomenon. You look at the Boston Latin School, or if you live in Northern Virginia, there’s Thomas Jefferson High School for Science and Technology. And in New York, you have Brooklyn Tech and Bronx Science and Stuyvesant.

And so people say, “Look at those awesome children, look how well they did.” Well, they wouldn’t get into the selective school if they weren’t awesome, but that’s distinct from the question of whether there’s a causal effect. When you actually drill down and do a credible comparison of students who are just above and just below the cutoff, you find out that elite performance is indeed illusory, an artifact of selection. The kids who go to those schools do well because they were already doing well when they got in, but there’s no peer effect from being exposed to higher-achieving peers.

We also have papers where we show that the elite illusion is not just a phenomenon relevant for marginal kids. This is in response to an objection that goes, “If you’re the last kid admitted to Stuyvesant, it’s not good for you because you’re not strong enough.” We can refute that with some of our research designs.

There are good stories and analyses throughout.


This explains why I've been a complete failure in spite of going to classy places. :-)

Just because you hang around Cowen's blog that doesn't make you a Cowen. So many comments here forget that.

Our peers are Prior and Thiago, not Tyler and Alex.

And we have our internet winner!!! Well played. I thank you as now I can safely walk away for the weekend knowing MR is in good hands.

Oooooooh yeah, Education stuff. Keep it coming.

I doubt his claim that there is no peer effect. But, if he is correct, it says something is very rotten in the school systems. I went to a college-prep private H.S. While I had middling grades before that, I graduated 3rd in my class. Competition is a thing. And my peers came to class prepared (homework done, readings read). Claiming that it doesn't matter who a kid is taking classes with makes no sense to me. Or are the lessons so regimented that Stuyvesant and PS 1-Oh-Dumb are tested on the same criteria? Are they both teach-to-the-test ? My 5 siblings and actually my wife went to one of the top public H.S's in the nation (at the time). I know I received a better education and part of that (at least) was the teachers could teach at their pace and not be hindered by disruptive and/or unprepared students. Peer effect, indeed.

"the teachers could teach at their pace and not be hindered by disruptive and/or unprepared students": that's a plausible reason why good schools are good for their pupils.

Of course it could be achieved in many schools by taking advantage of streaming the pupils. Presumably racial dogmas makes that hard to do in the US?

I don't know whether it's hard to do. It certainly is the norm around here, and we're a pretty left-liberal place. It starts to happen in the early teen years. You are still in the same school with everyone.

Hard to do? That is an understatement.

I have a relative at Stuvesant. The homework and pace are brutal. Much harder than most ivy-league school. There is no way a typical public school could do that at all. The only reason you can do it is becuase every kid is competing at a high level. If Angrist isn't finding peer effects, he isn't looking in the right way.

I suppose he is saying that in the next school down, almost "every kid" is at that high level, and so on, all the way down, small marginal differences in intake, with not much process difference.

"but there’s no peer effect from being exposed to higher-achieving peers."

well, that is probably true. If a kid is working at capacity, I doubt he or she would achieve more than that, just by being around smart kids.

But parents generally aren't looking for *higher* quality peers, just at-par peers or, more precisely, avoidance of low quality peers.

"Claiming that it doesn't matter who a kid is taking classes with makes no sense to me. "

I agree, but his research seems to be much more narrow than that.

Two real problems he doesn't seem to address:

1) huge range of ability in a classroom. Top kids might not learn as much as they are capable which year over year, wouldn't hurt their capacity but might limit their ability to prove it for college admissions. Also might keep them ignorant of their abilities, or unaware of how hard they have to work to really excel.

2) out of control classrooms, which would impact not just top achievers but all levels of achievement, as well as pulling some borderline kids out of the academic game entirely.

In general, researchers who try to prove parents wrong, whether it be about test prep or peer impact or small class sizes, never sound like they really understand the behavior and thus perhaps shouldn't be designing research.

"avoidance of low quality peers"

This. A thousand times. If you're smart, you'll do great even in a class with just average students. But if there are two or three disruptive students in the class - or just students who soak up the majority of the teacher's time - then you're in trouble. Selective schools like Stuyvesant have already filtered out the bad kids, so this research ignores that aspect of peer effects.

A good school and a good teacher can overcome bad peers - as per the no-excuses charter schools mentioned in the article - but if you're unable to judge the quality of the school, selecting good peers is a decent heuristic.

Besides, your kid might enjoy being in a school with others of similar ability.

The flip side is that you don't have to be in the top 1%, or even the top 10%; just avoiding the bottom 50% is enough.

Avoiding low quality peers often has more to do with social downward mobility than academic achievement. Parents prefer their kids not be exposed to drugs, bullied or impregnated. Parents also choose to send their children to elite private schools for the networking opportunities (for both their kids and themselves) with wealthy and powerful families.

Thank you for writing the comment I wanted to write as I was reading the blog post, but was worried I would muck up.

I'm afraid that if economists just went with "never question the old ways", we would have stopped in the 1700s and burned the Wealth of Nations at a kirk in Edinburgh.

I wonder if this is still true when comparing students in rural areas of similar capability to those in urban areas. I imagine there may be some value to being exposed to a greater number/diversity of students that can have a positive effect. There isn’t much “keeping up with the Jones’s” going on in rural areas where you’re hard pressed to find an SAT class given the limited demand

He mentions much longer hours and school years for charter teachers. Is that sustainable at a much larger scale if charter teacher pay is lower?

"The 2013 U.S. Department of Education’s Schools and Staffing Survey ( found that the average salary of a traditional public school teacher was $53,400, but charter school teachers earned an average of $44,500. This could be because charter school teachers have worked for fewer years at the school where they currently work. There is no information available from this survey about seniority level, though. Overall, charter school teachers earn about 10 to 15 percent less than they would at a traditional public school, no matter what their experience level is. For example, in 2013, Michigan charter school teachers ( earned an average of $42,864, but traditional public school teachers earned $63,094."

Don't get me wrong - would be happy to double teacher (and police, etc.) salaries and cut benefits, but if they're not being paid more and are expected to work many more hours vs. conventional public school teachers, that might be a natural limit to scaling the model.

This might be more of a labor force composition effect. Based only on observation (I had a district level job in a large school system for a bit) and not on data collection, charter teachers seemed younger than traditional teachers, so you would expect average salaries to be lower at charters.

There's a lot of burnout and attrition for teachers in public schools, regardless of how much you pay them. The better ones might be willing to take a pay cut to work at a charter school.

No. Other way round. Charter teachers burnout and go to publics.

Teacher supply is a huge problem for private schools and charters, less so for publics. They get the best pay.

I hadn't read the whole article yet. Kind of hilarious he praises charter schools while saying that peer effects don't exist. So he thinks exam schools are bad, but charters are good? Why the hell do parents want to have charters if not for peer effects?

He compares peer effects (where it is the kids who are the reason for higher/lower performance) to school effectiveness (where it is the teachers, principal, curriculum, etc., that explain perfomance). I think it's a bit of both and don't really think you can disentangle the effects as well as Angrist thinks he can.

Oh! Well, then, he's an idiot. The only thing effective about charters *is* their selectiveness, and the fact that no one has a right to go to a charter.

I recall commenting on your site with New Orleans as a counter-example, since charters took over the whole system and had better results. Also, even giving the same results at a lower cost would mean charters are more effective (specifically, cost effective).

One thought on why the almost-got-ins do almost as well as the just-barely-got-ins: they end up attending "next best" institutions, which may not be that much worse.

So what they may have shown is that the #1 place doesn't confer that much of an advantage over the various #2 places. What we don't know is how #1 (and #2) compare to the *average* campus.

Sure, it refutes a version of the "superstar" hypothesis in which people who attend the very best universities earn much more than those who don't, but doesn't say anything about whether there are marginal returns to education quality.

Anecdotally, Ivy League education isn't anything to rave about. The graduates I have met don't say anything to suggest the quality of instruction is better than at any respectable state university. You might find the most passionate and skilled teachers at small liberal arts colleges, which do actually seek out and promote professors primarily based on teaching skills and not research productivity. But again, it's not obvious that having smaller classes or better instructors offers a wage premium.

Well, our country is currently run by Ivy Leaguers - Trump, Kushner, Mnuchin, Barr, Pompeo (HLS).

Odd how administrations supposedly on three separate ideological axes - Bush, Obama and Trump - were all dominated by Ivy Leaguers.

Biden would be the first non-Ivy President since Reagan.

The tiny difference between #1 and #2 is why this concept works. If you compare #1 to #12600, too many other things differ, like most obviously parental background and connections and wealth.

The pier effect is that if you go to a school with fewer gang-members, there's less a change your kid will be a gang-member. If you got to a school with fewer pregnant teenagers, there's less a change your kid will be a pregnant teenager. If you go to a high school everyone can walk to, there's more of a change you'll be able to walk there and not have to take the bus.

At first glance, this would appear to support Arnold Kling's Null Hypothesis. However, as many previous commenters have said, perhaps the peer effect is important in (avoiding) the low end rather than accessing the high end. It could be important to avoid schools with gang members as peers even if being grouped with smart kids isn't that helpful.

On charter schools, Angrist finds: "Charters of a particular type known as 'no-excuses charters' are very effective....They serve low-income students...But...suburban charters that are not in this no-excuses paradigm tend to take middle-class and upper middle-class children and reduce their achievement."

So, again, maybe avoiding the left tail of bad schools is more effective than trying to access some super-school on the right tail. (This charter school effect is about the schools, not peer selection, but the point is about avoiding left tail in either case rather than accessing right tail. Right tail seems to be a property of the individual student rather than peers or schools.)

Also, maybe these charter school effects explain why minority inner-city voters favor expanding charter schools even though white, suburban voters don't (at least in MA). The conflict between teachers unions and minority schoolkids seems to be the segregation battle of the 21st Century.

As you just pointed out, it's not teachers unions that stopped charters, but the voters generally.

"This charter school effect is about the schools, not peer selection"

Also as you just pointed, out, this is bulllshit. Charter school effect is also about selection. And the reason that blacks support charters is because they provide peer selection that they can't otherwise afford, not better teachers. (Ask them, they'll tell you). Teachers--and not just unions--oppose charters for the same reason that the public generally opposes charters: they are expensive, take away from publics, and don't provide the same rights that publics do.

BC didn't "point out" the charter effect was also about selection, the comment took for granted that Angrist was right about the school effect.

Are charters more expensive than other schools? I was under the impression they were cheaper.

"if you got into MIT or Harvard, it actually doesn't matter where you go....There isn't any earnings advantage from going to a more selective school once you control for the selection bias....People who get into selective colleges and universities are picked because they're people who are likely to be smart and successful."

Are they saying that there is no elite-college premium to explain, either through signalling or otherwise? In other words, doesn't this say that Harvard-admitted students somehow manage to signal their abilities through other means even if they don't actually end up going to Harvard? Plausible, although I guess this would suggest that it's actually not all that important for Harvard to expand its student body through online technology or otherwise.

Assuming elite signalling has a positive effect, there is a must be compensation. Perhaps, the education is less practical than at Flagship State?

When I was reading Greg Clarks work on heritability, one thing he said was that income often varied quite a bit between generations, but not status. So a person who gets a liberal arts degree from Harvard may not make more money than someone with an accounting degree from State U, but its pretty obvious they have more status. They choose a career with lower earnings, but could have done one with higher earnings if they wanted to. Meanwhile State U couldn't have been a professor if he wanted to, and took the boring but paying job.

It's obvious, isn't it, that people from Harvard seem to be more successful than people from State U when you consider status and not just income. I don't know the incomes of Supreme Court Justices, but they certainly have more status than many people with the same incomes.

That is a good point, and one that economists tend to miss. Ivy League graduates can “afford” to give up income for positions with social status. Most people can’t. This probably partially explains why so many Ivy Leaguers end up in lower paying but high status government positions like State Department or CIA or end up working as journalists. This of course becomes mutually reinforcing and those occupations are perceived as higher status because smart Ivy League grads are working in them.

If we use the word status to mean our respect for people based on their alma mater, sure.

There are contrarians among us. I recently advised a retired physician and his wife on their estate plan, which for them meant trusts for their (now very young) grandchildren. But get this: they wanted no distributions until the grandchild was 18, and from age 18 to 24 distributions of no more than one-third of the grandchild's original trust fund. I pleaded with them to give the trustee greater discretion, in case the grandchild is smart and would qualify for the best, and expensive, private prep school in the area (whose graduates go on to elite colleges) or is qualified for an elite, expensive college like Harvard. Nope, not with our money was the response: they believed expensive schools are a waste of money and their money won't be wasted on them. I was convinced that these two had early onset dementia.

Yes, and Triple A ball players are just as good as major league ones.

But except for players on the way down (who have contracts from their time in the elites), the pay is not what you might think. For those who aren't familiar with baseball, Triple A may seem like a notch below the majors for a player on his way up, but it's mostly for players on the way down. Double A is where one will find players on the way up, but even there the pay is low and the security non-existent. I didn't attend an elite college or law school (I attend my state university for both), but times are different now. Back then I was recruited by the best law firms in the state, even some some out of state. Ain't happening today. My godson will start Harvard Law School in the Fall after graduating from an elite college. His family will be making sacrifices to send him to Harvard - it is really expensive. He was offered not only a full tuition waiver but a stipend to attend another elite law school. His family believed their son, my godson, had worked hard and deserved to attend Harvard. He will excel there and in life for the benefit of all of us.

"He will excel there and in life for the benefit of all of us."

Except that he's going to become a lawyer. Worse, an American lawyer.

Adams, Jefferson, Madison, and many others, the list is long, the accomplishments many. We lawyers like to compare our antecedents to the antecedents of doctors: while our antecedents were writing the declaration of independence and constitution, their antecedents were using leeches to suck out the blood of their patients. As Jefferson quipped. when two or more physicians gather, buzzards can't be far.

The Great Stagnation hit the law, but not medicine.

Well, unless he thinks that he's entitled to success because he went to Harvard Law School. As someone who was involved in Big Law hiring for a long time, Harvard was most often the source of the problem summer associate who either was a complete jerk to the staff or thought they didn't have to work at all, which at least back then was the best way not to get a permanent offer.

How the heck do you reconcile his finding of no peer effects with Judith Harris, as highlighted in Stephen Pinker's The Blank Slate ?!?

Main findings were genes account for 50% of differences between individuals, peers a huge 40%+.

It was shared family environment that had no effect.

I think you misread those books - or at least the Pinker; I never read Harris. I remember Pinker's real conclusion being that it was 50% genes, and the other 50% were ???. He was careful to say that we did not rule out peer effects, whereas shared household was more or less ruled out as contributing more that 10% to outcomes. But you shouldn't see "did not rule out aliens" as "but it was aliens."

I think you do it by concluding peer effects are more random or local than we think. It doesn't really matter what school you go to. it matters which smaller, randomly determined group of peers you interact with while you're there.

Harris assumed peer effects were important rather than proving it. A lot of variation could be random chance we just have no way of explaining.

So if there are no peer effects from classmates, does this mean that we will be ending all efforts at having "diverse" student bodies? As I recall things, the only reason we are supposed to allow affirmative action in university admissions in order for there to be a "diverse environment" that allows students to learn more/better. This sounds quite like some sort of peer effect so if we buy this story, is it time to go to a wholly race/gender/sexuality/etc. blind admissions process?

Impeccable logic is a Thought Crime, comrade.

You're missing an important point if you're mad about "diversity accepts" at ivy leagues. If peer effects really don't matter then the whole injustice of unfair acceptance standards is silly - like fighting about who receives the fancy gelcap placebo pills and who must make due with ordinary placebo pills.

I like Angrist's work but he seems to be over-confident in drawing conclusions from his research. No peer effects? I think it was Bowen and Bok who found that at the very highest levels of students and schools, it didn't matter where the white students went, they were going to excel anyway. But the minority students (I don't recall if this includes=d Asians or not) gained significantly by attending an Ivy League or other elite school instead of a less selective school.

He appears to believe his empirically-founded beliefs too much.

Of course, this also suggests that all those "college graduates make $X more than high school graduates over their lifetime" advertisements also rely on "artifacts".

Bryan Caplan would say that graduates are more intelligent and conscientious, along with having a greater willingness to "play the game and follow the rules".

As someone that went to one of these elite schools, I think this totally misses the mark.

If I hadn't gotten into my magnet schools, its possible that my SAT score would have been the same and maybe even my lifetime earnings. Who knows. I do think there was a greater chance of "dropping out" in my local high school due to non school reasons.

What he doesn't seem to capture is that life at my magnet school was awesome. It was fun, it was challenging, I had great friends, and I learned a lot. None of that was true at my local high school. I got Straight As with little effort and spend most of the day hoping not to get beat up. It was miserable.

So how do Angrist and Chetty reconcile their work or think of each other’s work? Maybe I’m misreading their overall findings but Chetty’s work seems to be all about the importance of neighborhood effects which I see as also being about peer effects but maybe I’m jumping to particular mechanisms. Is the answer that there are neighborhood effects but not due to peer effects? I’d really love to see these two research agendas compared and critiqued in light of one another.

One possibility is that Chetty's work is simply wrong, correctly reporting change but assuming causation that isn't there.

Yup, this is why Sailer should be asked to comment on this type of research. He nailed it that the effects Chetty found seem mostly explainable by local economic growth. It really shows how inbred Economics is that Chetty could publish without dealing with this objection.

The irony is that Sailer would see "inbred" as a compliment, for a certain value of endogamy.

Thanks for responding.

Had you not heard about Heckman laying into Chetty? Chetty finds certain areas correlated with certain things, but that's insufficient to prove causation. Of course, Heckman himself leans heavily on a study with fewer participants than the number of papers he's published on it to support his preferred theory.

I haven’t followed Chetty’s work that much beyond what’s been reported here over the years so thank you for pointing this out.

Growing up within a strongly blue color community, I intentionally under-performed to avoid the mockery and bullying that came with good grades. My impression is that black-communities have something very similar.

Maybe good peers don't make people smarter, but bad peers definitely make them dumber...... IMHO.

Perhaps the students who benefit are those further above the cut-off.

You may well be right. How to test that?

The problem is that the marginal student -- however defined may not be the prime beneficiary. Sometimes elite schools are a force multiplier for the top of the top. To oversimplifiy -- it's the difference between a top student getting into an Ivy, vs. a top student getting into all the Ivies. Same later in life, with graduating from a top PhD program increasing the probability of tenure track, promotion, publication in the top journals, etc.

This is like saying Lennon and McCartney had no impact on each other's ability, and that both of them didn't influence Harrison.

View from an Unknown Commenter about K-12.

"Your child is not being held back by the idiots he has to sit with by government decree, and the trouble makers who disrupt class. The teachers have to develop their lesson plans around those low IQ dolts, which is why your kid finishes the work in half the time.

A benefit most are not considering is that parents are now getting to see the material their kids are being taught at the granular, day to day level. They can see the crazy math curriculum, the rabid anti-America and ant-white history, all the indoctrination that normally is hidden from view.

Public education as it existed in Jan 2020 is now dead, and just like universities, they don't know it yet."

Comments for this post are closed