Bryan Caplan  

Trust and Diversity: Not a Bang But a Whimper

Larry Summers, market monetari... Henderson Review of Schuck...
Robert Putnam, famed author of Bowling Alone, has spent much of his career regretfully publicizing the dangers of diversity.  His most famous claim, of course, is that "social capital" - usually operationalized as "trust" - is vital for a good society.  And though he's a liberal in good standing, he urges us to face facts: diversity - especially ethnic diversity - is very bad for trust.  As Putnam's 2006 Johan Skytte Prize Lecture summarizes:
Diversity does not produce 'bad race relations' or ethnically-defined group hostility, our findings suggest. Rather, inhabitants of diverse communities tend to withdraw from collective life, to distrust their neighbours, regardless of the colour of their skin, to withdraw even from close friends, to expect the worst from their community and its leaders, to volunteer less, give less to charity and work on community projects less often, to register to vote less, to agitate for social reform more , but have less faith that they can actually make a difference, and to huddle unhappily in front of the television. Note that this pattern encompasses attitudes and behavior, bridging and bonding social capital, public and private connections. Diversity, at least in the short run, seems to bring out the turtle in all of us.
What's impressive about Putnam's piece, however, is that he takes the danger of spurious correlation so seriously:
[T]he diverse communities in our study are clearly distinctive in many other ways apart from their ethnic composition. Diverse communities tend to be larger, more mobile, less egalitarian, more crime-ridden and so on. Moreover, individuals who live in ethnically diverse places are different in many ways from people who live in homogeneous areas. They tend to be poorer, less educated, less likely to own their home, less likely to speak English and so on. In order to exclude the possibility that the seeming 'effect' of diversity is spurious, we must control, statistically speaking, for many other factors.
Indeed.  And here is what Putnam finds when he uses multiple regression to predict individuals' trust on a 4-point scale.

putnam.jpgPutnam's measure of diversity - the Census tract Herfindahl Index of Ethnic Homogeneity - is in bold.  It's the sum of the squares of each group's population shares, so it theoretically ranges - note the reverse coding! - from 0 (infinite diversity) to 1 (zero diversity).  But since Putnam only sub-divides the population into four groups - Hispanic, non-Hispanic white, non-Hispanic black, and Asian - his diversity measure can't fall below 4*.25^2=.25. 

Now imagine we move from a world with zero diversity to the maximum diversity.  According to Putnam's results, how much will this reduce trust?  .18*.75=.14.  Is that a lot?  No way.  Remember, he's using a 4-point scale.  And since the current national Herfindahl Index of Ethnic Homogeneity is about .46, moving from the diversity of today to maximum diversity reduces predicted trust by a microscopic .18*.21=.04.  "Diverse" communities have low trust, but the reason isn't that diversity hurts trust; it's that non-whites - especially blacks and Hispanics - have low trust.

What's especially striking, though, is that Putnam finds several variables that strongly predict trust that almost no one discusses.  Look at the effect of home-ownership.  Not only do home-owners average .25 higher trust; there's also a -.14 coefficient on "Census Tract Percent Renters."  Net effect of moving from 0% to 100% home ownership: .39.  Holding all else constant, citizenship is good for trust: a mere .06 for the individual, but a solid .21 for the community.  Net effect of moving from 0% to 100% citizenship: .27.  There are also big effects of crime, population density, and commuting time.  Geographic mobility, strangely, seems to reduce individual trust but raise social trust.

The latter variables don't just matter more than trust; they're also much more policy-relevant.  Reducing diversity is very hard; indeed, without massive human rights violations, it's almost impossible.  Home ownership, in contrast, can be fostered with not only tax incentives (the standard way), but housing deregulation (the wise way).  Population density, similarly, can be reduced by deregulating development of surburban and rural land.  Commuting time can be slashed with better public transit (the standard way) or congestion pricing (the wise way).  Better policing and law enforcement - not to mention thoughtful decriminalization - all reduce crime rates.  And if we take the estimated benefits of citizenship at face value, it can be raised to 100% with the stroke of an amnesty pen.

Putnam's numerous alt-right fans seem to relish his anti-diversity claims because he lends an air of respectability to their misanthropy.  But if you read Putnam's whole article, it's hard to detect any ill will.  Why then would he so grossly overstate the dangers of diversity?  My best story is just confirmation bias.  Early on in his research, Putnam found strong univariate links between diversity and trust.  When better methods show that this relationship matters slightly, Putnam - like most human beings - treats this as a vindication of his initial claims.  But if all you knew were Putnam's final results from Table 3, you'd never reach Putnam's anti-diversity conclusion.  Diversity's so unimportant for trust, you might not mention it at all.

Comments and Sharing

COMMENTS (17 to date)
BH writes:

I'm a gonna state the bleeding obvious: just because current home owners have higher trust does not mean increasing home ownership will increase trust.

LK Beland writes:

Quick question: what steps were taken by Putnam to ensure that his independent variables were not multicollinear?

Glenn writes:


No kidding!

Not to mention entirely glossing over the fact that high diversity areas are correlated with all kinds of objective bads, like higher crime. The so-called controls here just dilute the problem, deliberately deflating the trust impact. I have less trust in a high crime society; diverse societies are high crime; ergo 'control' for crime?

Really bad econometrics here.

Glenn writes:

To be clear, the bad econometrics are in the blog post and not, to my knowledge, the subject pallet.

Dangerman writes:

"'Diverse' communities have low trust, but the reason isn't that diversity hurts trust; it's that non-whites - especially blacks and Hispanics - have low trust."

Oh boy, you've done it now Prof. Caplan.

You see, the reason Putnam is "a liberal in good standing" is *because* he doesn't come out and admit this.

Of course, all those "numerous alt-right fans" *already know* that "diversity is bad" is code for "non-white hispanics and blacks are bad."

You're not supposed to come right out and admit that.

Nathan Smith writes:

The correlation/causation objection can be made to any of the variables involved here, including diversity. Maybe it's not that diversity exogenously reduces social trust, but that people who highly value social trust seek out homogeneous neighborhoods. Maybe diversity doesn't bring out the turtle in us, but attracts the turtles.

That doesn't change the basic message that forced integration is bad. Blacks are a special case because we have the national sin of slavery to live down. For the rest, it would be better to take the libertarian, freedom of association course and permit voluntary residential segregation. Let there be gated communities for Muslims only, Christians only, college grads only, whatever people want. If there's demand for it, allow supply.

Thaomas writes:

Actually we should remove obstacles to density in both rural-suburban areas as well as cities. Congestion taxation would also promote more density. I'm not persuaded that subsidizing home ownership is a good idea, but if it is, we should do it with partial tax credits rather than deductions.

Samuel writes:

"'Diverse' communities have low trust, but the reason isn't that diversity hurts trust; it's that non-whites - especially blacks and Hispanics - have low trust."

This implies that if you have a group of exclusively black people, and then increase the level of diversity by adding a handful of white people, then trust should go up, as whites have higher trust than non-whites.

However, the model implies that if you have a group of exclusively black people, and then increase diversity by adding a handful of white people, trust should go down, as, controlling for race, there is a negative effect of diversity on trust. Nothing in the model would support the conclusion quoted above, so I have no idea why that sentence was included in the article.

PedroS writes:

The table seems to show that the non-violent crime rate decreases trust AND that the violent crime rate dramatically INCREASES trust. I cannot understand how this happens...

Hazel Meade writes:

Moreover, individuals who live in ethnically diverse places are different in many ways from people who live in homogeneous areas. They tend to be poorer, less educated, less likely to own their home, less likely to speak English and so on.

To what extent is this kind of ethnic diversity due to exclusion from white communities? When we talk about diverse communities, are we really just talking about non-white communities composed largely of immigrants? In which case, would this result carry over to mixed race communities of non-immigrants? It should be obvious that immigrants are poorer, less educated, and less likely to speak English, but that cause of that has nothing to do with living in a diverse community.

Hazel Meade writes:

@Dangerman, maybe blacks and hispanics don't trust eachother in the same way that whites don't trust blacks or Hispanics and vice versa. But white people are more able to isolate themselves into all-white enclaves, while blacks and Hispanics often have to live in mixed race communities. So the low-trust variables are partly a consequence of white racism and social exclusion of non-whites.

HoodlumDoodlum writes:

The trust benefits of citizenship can be increased by 100% by simply making everyone a citizen. That's brilliant, Prof. Caplan.
Why stop with the current population physically present here, though? There are a few billion people on the planet who aren't US citizens--surely making everyone in the world a US citizen would cause worldwide trust levels to explode!
Why bother with housing policy, though--just declare everyone a homeowner. Boom, solved.

So, so smart.

Glenn writes:


You have to read the table under an "all else equal" condition, meaning holding all the other control variables - which are positively correlated with violent crime AND 'low trust' - higher violent crime raises trust.

As a casual event in real life this is of course non sensical. That is exactly why caplans underlying blog post is also nonsense. The results do not show what he claims.

Lorenzo from Oz writes:

Hazel Meade: how is removing your children from high crime environments racism?

Hazel Meade writes:

white people have lots of different reasons for isolating themselves in white enclaves. Sometimes it's to escape crime. Sometimes it's because of racial bias.

Beor the Old writes:

It might be hard to reduce diversity, but perhaps we might stop encouraging it.

benis69 writes:

Am I grossly misreading this or is there a typo/sign error here?
Non-violent Crimes per Capita B=-2.57
Violent Crimes per Capita B=6.59
"How do you plead?"
"I was only trying to drastically increase social trust in my community"


Return to top