Americans tend to assume that polarization is bad for democracy. It is supposed to undermine compromise and contribute to gridlock. It’s furthermore thought to be linked to growing inequality.

But when we talk about polarization, we are usually talking about Congress and political elites. Ordinary citizens show very little polarization.

In a new article, we argue that this lack of polarization among the public isn’t necessarily a good thing. Indeed, it might be a sign of serious democratic failure.

Why does polarization decrease when there’s more economic inequality?

First, take a look at this puzzling graph. If we look at advanced democracies, we find that income inequality has a negative relationship with polarization (which we measure as the share of the population who think of themselves as being non-centrist on a scale of left to right). In general, as inequality increases, polarization decreases (and vice-versa).

This is surprising. When income inequality is higher, we would expect people to disagree more about issues such as government spending and redistribution (which we know are closely associated with whether people view themselves as being on the left or on the right).

Looking at the data more closely shows that the median member of the public tends to be further to the right when income is more unequal. Again, this is unexpected. Because the distribution of income favors the rich everywhere, the mean income is higher than the median (it is pulled up by those with very high incomes). If the majority are well-informed and self-interested, you would expect them to want more redistribution when inequality rises. Just the opposite is true. Why?

We think it’s because most people don’t examine public policies closely — and so drift to the center by default

We argue that the answer to these puzzles involves information. People are often not well-informed about politics, and they few have reason to be. Public policies affect large numbers of people, but as individuals we don’t have much power to change these policies, so why spend valuable time acquiring information about them?

As it turns out, this simple fact (sometimes called “rational ignorance”) has implications for polarization. The reason is that uninformed voters tend to place themselves in the center of the political space, which they see as “safe” compared to more extreme options.

Imagine that a voter can pick a left, center or right party, but doesn’t have enough information to know which has the best economic policy given her interests. She just assumes that there’s an equal chance that each of these parties has the best policy. The best strategy (if she doesn’t have much information) is to pick the center party, since it may actually have the best policy, and even it doesn’t, its policy is more likely to be closer to the best policy than either of the other two, because it’s located between them (for a step-by-step derivation see the original paper). This logic also applies to ideological self-placement, and we refer to it as the centrist bias.

The more people know, the less centrist they become

The centrist bias dissipates as people acquire information. Those whose interests would dispose them to identify with the center will still do so, but those who we might expect to identify with the left or the right will move away from the center. Information breeds polarization. So one way to think about why some countries have more non-centrist voters is to ask how and why voters acquire political information.

Most people get information through groups and networks — like unions, families or co-workers

We argue that most people get information through the groups and networks they belong to. One such group is unions, although unions, like other formal groups, have declined in importance in most societies. Unions often expose their members to political messages, and they also sometimes engage their members in political discussion.

When discussion is involved – say, when a union official brings up political issues around the workplace lunch table – it pushes people to look for information from papers, TV, the Internet and so on.

This logic also applies to political discussion in social networks of family, friends, neighbors and co-workers. People are more likely to acquire political information when others around them care about such information and are evaluating them in part based on whether they seem well informed. Needless to say, it is easier for well-educated people to acquire political information, so educated people tend to respond more effectively to such social incentives.

So who tends to be well-informed?

This has partisan implications. Those with higher incomes tend to have better education and be better integrated into social networks with a lot of political discussion. These people are the natural constituents of center-right parties, which means that the centrist bias is less pronounced for the right.

Yet the extent to which this is true varies across time and space. Many different factors — union membership, membership in social networks with political discussion, and education — are associated with higher levels of information and therefore also with ideological polarization.

In countries with strong unions, the centrist and right biases are less pronounced. The same is true in countries with good public education systems and more pervasive informal discussion networks. The figure below shows the relationship between the frequency of political discussion in social networks and ideological polarization. The political discussion taking place at Parisian street cafes really does matter, and so does a blog like this! We should not be surprised.

Some public policies can lead to more or less general knowledge and public discussion — and therefore, whether there’s an informed left 

This implies that countries with certain institutions and policies — strong unions, heavy investment in public education, and extensive social networks — end up with more politically informed electorates which are also more ideologically polarized and left-leaning than countries where these institutions are less well developed.

At the same time, unions, public education and network integration promote income equality for reasons that are well understood in economics.

Together this provides a plausible explanation for the pattern showed in our first graph, where there is a negative relationship between inequality and polarization.

Because governments of the left and right affect the income distribution though public policies — most obviously through redistribution and public education — these relationships may be self-reinforcing, producing distinct “varieties” of democracies. From this perspective, polarization may in fact be a sign of a well-functioning democracy — one that has well-informed electorates and governments that support the majority’s interests in pushing back against rising inequality.

Torben Iversen is the Harold Hitchings Burbank Professor of Political Economy in the department of government at Harvard University’.

David Soskice is a professor in the department of government at the London School of Economics.