Why don't Americans trust each other anymore?

Or do they?

With pandemic UI ending, the CDC eviction moratorium finally set to expire, and a majority saying they don’t want Biden’s child allowance made permanent, Americans seem to be pivoting back to a scarcity mindset. The most egregious example I’ve seen was probably this quote from a New York Times story:

Americans seem to be slipping back toward their pre-pandemic mindset of constantly worrying about whether other Americans are getting a freebie. This attitude makes it hard to create a rational welfare state, instead biasing us toward a forest of means-tests and work-requirements that turns the whole system into a giant kludge. It also makes it hard to fund public goods. A country where people are suspicious that their countrymen are undeserving freeloaders is not likely to be a well-run country.

Which brings us to the issue of social trust. I’ve seen a number of people commenting lately about falling social trust in the U.S., so here’s the General Social Survey measurement that finds a long-term decline:

That’s a worrying trend! And the trend is probably real, since older cohorts tend to report much more social trust than younger cohorts.

Meanwhile, Pew finds that Americans are generally worried about falling trust as well:

So what’s the cause of the problem? Well, first we should figure out how much of a problem it really is.


Is U.S. trust really that low?

First of all, as anyone who has done this kind of survey research knows, how people answer this question depends heavily on how you ask the question. For example, Pew asked three trust-related questions and got pretty different percentages:

Looking at this, you might think that U.S. trust was at 52% (yay!) or 37% (oh no!).

Then Pew asked people specific questions about how they expect their countrymen to behave. The results look much more positive for American trust levels when the question is about anything other than politics:

We generally trust each other to obey the law, work together, and help those in need. That sounds like a pretty high-trust society to me!

OK, but what about compared to other countries? The World Values Survey asks people in various countries whether they agree with the statement that “most people can be trusted”. It turns out that the U.S. is pretty middle-of-the-pack here, pretty similar to Canada, Germany, Japan, and Singapore, and a bit higher than the UK, South Korea, Taiwan and France. Sweden and Norway really stand out here as the high-trust champions.

But of course, just because you ask people the same question doesn’t mean they think of the question the same way. Japanese people might be thinking of whether you can trust people to return a $1 bill they found on the ground, while Americans might be thinking of whether you can trust a stranger not to put an axe through your skull. It’s just hard to tell. (Actually, my subjective personal experience is that Japanese people and Americans are about the same in terms of trusting strangers; this was just a hypothetical).

So how do you measure trust across countries in a consistent way? You can do lab experiments, in which (for example) people give trust other people with their money. For example, this paper found pretty similar trust levels between the U.S., Japan, South Korea, and China (though the U.S. and Korea were slightly lower). But the real question is how much these experiments generalize — do they really predict national-level aggregate behavior outside the lab? I know of some large-scale real-world experiments on trustworthiness (the U.S. again is in the middle of the pack), but trustworthiness is a very different thing than trust.

So I’d say we should be very cautious in believing that the U.S. has become a low-trust society. I am definitely worried by the fall in reported trust levels over time, but it’s not clear that the level of trust is low enough yet that we should expect to see significant problems.

OK, but what’s causing the worrying drop? There are a number of hypotheses, but first let’s focus on the favorite hypothesis of the anti-immigrant Right: The idea that diversity decreases trust.

Trust and diversity

A common story on the political Right is that as the U.S. has become a more diverse place since the 1965 immigration reform, trust has fallen accordingly. Even moderate conservatives who recognize the economic benefits of immigration often put credence in this notion, and view it as a cost of letting in foreigners. The idea has a certain knee-jerk plausibility — people might not trust people who look different than themselves, or have different cultures and customs, etc. But it also has an intellectual heavyweight to back it up: Robert Putnam, the famous sociologist. In a 2007 lecture, Putnam asserted:

In the short run … immigration and ethnic diversity tend to reduce social solidarity and social capital … In ethnically diverse [American] neighbourhoods residents of all races tend to ‘hunker down.’ Trust (even of one’s own race) is lower, altruism and community cooperation rarer, friends fewer.

But in short, the evidence doesn’t really support the idea. I’ve written an old blog post and a Bloomberg post about this, so I’m going to crib heavily from those.

First of all, just look at the international trust measures above. Though they’re not as homogeneous as people think, Japan, South Korea, and Taiwan are more homogeneous than the U.S., and yet have lower trust levels. Meanwhile, Scandinavian countries Sweden and Norway, which have the highest trust, both have more immigrants as a percentage of population than the U.S. does!

Second of all, the finding that diversity is correlated with low trust at the neighborhood level doesn’t really replicate well. Here are a few papers that find either no correlation or a positive correlation:

  1. Does Ethnic Diversity Erode Trust? Putnam’s ‘Hunkering Down’ Thesis Reconsidered”, by Sturgis, Smith, Reed, and Allum (2010)

  2. Does Diversity Erode Social Cohesion? Social Capital and Race in British Neighborhoods”, by Letki (2008)

  3. Ethnic Diversity and Social Capital in Europe: Tests of Putnam’s Thesis in European Countries”, by Gesthuizen, van der Meer and Scheepers (2008)

  4. Unknown is unloved? Diversity and inter-population trust in Europe”, by Gerritsen and Lubbers (2010)

And here are three papers finding the same thing as Putnam:

  1. Trust in a Time of Increasing Diversity: On the Relationship between Ethnic Heterogeneity and Social Trust in Denmark from 1979 until Today”, by Dinnersen and Sønderskov (2012)

  2. Does Ethnic Diversity Have a Negative Effect on Attitudes towards the Community? A Longitudinal Analysis of the Causal Claims within the Ethnic Diversity and Social Cohesion Debate”, by Laurence and Bentley (2016)

  3. Ethnic Diversity, Economic and Cultural Contexts, and Social Trust: Cross-Sectional and Longitudinal Evidence from European Regions, 2002–2010”, by Ziller (2014)

How about literature reviews and meta-analyses? Here, results also differ. A 2020 review by Dinnesen, Schaeffer, and Sønderskov found a negative relationship between diversity and trust. But a 2014 review by van der Meer and Tolsma found that ethnic diversity was not related to less interethnic social cohesion.

Here we come to a subtle distinction. Again, it matters how you measure trust. When you ask people how connected they feel to their neighbors, you do tend to find a negative correlation between ethnic diversity and trust. But when you ask people how much they trust people of other ethnicities, you don’t find a negative correlation. This turns out to be a pretty consistent finding.

Personally, I suspect that there’s a pretty strong neighborhood selection effect at work here — people who trust people of other ethnicities move to diverse neighborhoods, while people who move around a lot and thus don’t know many of their neighbors also tend to move to diverse neighborhoods. So I suspect that this whole literature, including Putnam’s famous and oft-quoted original study, doesn’t tell us much.

(There’s also the literature suggesting that racism — specifically, White stereotypes of Black people as undeserving of government benefits — is a big reason the U.S. doesn’t have as good of a welfare state as most European countries. But note that Black people have been in America forever; “White-Black divisions decrease social cohesion” is a very different story than the “post-1965 immigration caused trust to decline” story!)

Finally, there’s the fact that the effect Putnam finds just isn’t that big. Bryan Caplan points out:

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.

Just not that big of a deal. I think we can safely dismiss the idea that growing diversity is the cause of falling trust

Now, as Caplan points out — and is visible in the Pew survey — Hispanic Americans do tend to report slightly lower interpersonal trust levels than White Americans. So the big increase in the Hispanic population percentage since the 80s, and the big drop in the White percentage, might be a reason that reported trust is falling. But looking at trust by income levels, it’s clear that the ethnic trust gap can easily be explained by the ethnic income gap.

So as Hispanic Americans advance up the income ladder, any negative effect of recent immigration on trust seems pretty likely to disappear.

Other explanations for low or falling trust

One popular idea is that suburbanization is responsible for making Americans isolated and lonely. A 2007 Pew survey found that people in suburbs are more trusting than people in cities, but that could easily be a selection effect.

A second idea is that inequality lowers trust. There is a ton of evidence showing that higher inequality is correlated with lower trust, but the causation could easily run both ways since low trust could lead to a crappy welfare state. I found an interesting paper relating inequality increases to later trust decreases in Chinese regions, but overall trust in China has stayed steady or risen, even as inequality has exploded.

Technology is another possible culprit. Cars allow people to move around, which means they know fewer of the people in their neighborhood. The internet has us sitting inside all day talking to people far away instead of learning to trust the people living near us, etc.

When you ask Americans the reason for the decline in trust, opinions are pretty split among a lot of these factors:

But if I had to make a guess as to what’s lowering trust in America, my guess would be politics.

Take a look again at the ways that Americans do and don’t trust each other:

Generally the things we do trust each other on are normal, everyday things, while the things we don’t trust each other on are all political. We don’t trust each other to stay informed about issues, to respect the rights of people who aren’t like us, to cast informed votes, or to have civil conversations with people with different views.

That suggests that when Americans say they don’t trust each other, what they’re actually distrustful of is the opposite political tribe. If you look at when the trust drop happened, it lines up pretty well with the rise of partisan polarization. That doesn’t prove causality, but it’s a hypothesis worth investigating. Hatred of political opponents has become one of the most defining and pervasive features of American life; I wouldn’t be surprised if the decrease in interpersonal trust is just another outgrowth of our all-consuming culture war.

Which means that, as in everything else involving the culture war, I don’t have much of a solution.