The U.S. Capitol Reflecting Pool. (Mark Wilson/Getty Images)

A core problem in American politics is ideological polarization, which makes compromise impossible, not only in Congress but also in the states. Polarization has many causes, most of which can’t be altered. But over the past few years, a growing crowd of scholars has explored the intriguing possibility that stronger parties could reduce polarization.

The logic of this argument goes something like this: Since political parties above all else want to win elections, they should support moderate candidates, who are more likely to win in general elections (as compared to extreme candidates). This is often referred to as the median voter theorem. One way to strengthen parties would be to allow them to raise and control more of the money in the process.­ Leading thinkers like Jonathan Rauch and Rick Pildes, as well as others skeptical of traditional reform approaches, have offered more detailed elaborations of this general point.

But it’s not just a theoretical question. It’s also an empirical question. Because different states have different rules on how much money parties can raise, we can empirically test the proposition that making it easier for parties to raise more money reduces polarization.

So far, we have one influential test of this proposition. Political scientists Ray LaRaja and Brian Schaffner have compared legislators in states with limits on contributions to parties against those in states without limits. They conclude that legislators in states that limited contributions to parties are more ideologically extreme. Their finding has since become a key part of the argument for strengthening parties by allowing them to raise more money.

This is too quick a conclusion to draw on the basis of a single analysis. Looked at another way, the same data suggest the very opposite result: that states that limit contributions to parties are actually less polarized than states without limits. Before building a reform program around strengthening parties by opening up some of the few remaining limits on money, we ought to be careful and modest about what we think we know.

A different look at the data

LaRaja and Schaffner combine all the legislators into a single distribution, thereby overweighting states with more elected officials and ignoring the differences across states in the same limits category (are Alaska Democrats and California Democrats really equivalent?).

Another analytical approach, the one I explore here, would be to measure the difference between the ideological position of the median Democrat and the median Republican in each chamber, and then compare across states. This is the measure that Boris Shor and Nolan McCarty, who produced the state polarization data, describe in their code book as their “preferred measure of polarization.”

Below, I compare the differences in party medians across states, looking separately at upper and lower chambers. Following LaRaja and Schaffner, I’ll limit this analysis to the 20 most “professional legislatures.” The list of states with and without limits on contributions to parties comes from the National Conference of State Legislatures.

The simple conclusion is that states with limits are less polarized. Six of the seven least polarized lower chambers have limits on contributions to parties, and four of the five least polarized upper chambers have limits. In their lower chambers, states with limits on contributions to parties have, since 2004, averaged a median difference in party medians of 1.31, as compared to a median difference of 1.68 for the states with no limits. In the upper chambers, states with limits also are slightly less polarized (they average difference in party medians since 2004 was 1.50, as compared to 1.59 for states with no limits).

Polarization in states, limits vs. no limits on contributions to parties

Looking more closely at the distribution of states up and down the y-axis, the first thing that should be clear is that there is considerable variation in polarization across states. So, while we do observe differences, it’s hard to put too much emphasis on party limits as a key explanatory factor. A lot else is probably going on.

A further complication is that several of the states with limits have important exemptions. California, New York, Ohio and South Carolina allow unlimited contributions for party building and administrative expenses. Washington allows unlimited contributions in the 21 days before elections. If we reclassify these five states into the no limits category in the figure below, we see a more sizable gap between states with strict limits as compared to states with limits/major exceptions. Among the lower chambers, 10 of the 11 most polarized legislatures either have no limits or major loopholes. Among upper chambers, seven of the eight most polarized upper chambers have no limits or major exceptions.

Polarization in state legislatures, states with no limits on contributions to parties and major  exceptions grouped together

So what’s going on here? One possibility is that the parties are not independent actors trying to move to the middle, and therefore strengthening them does not make them more likely to follow the median voter logic. Rather, parties are coalitions of their most active and generous supporters, their “intense policy demanders” who tend to be highly polarized. The bigger the checks these partisan policy demanders can write, the more active a role they will have in determining party priorities. And the more money parties can raise from unlimited sources, the more they can use the money to obfuscate how far they have moved to the policy extremes by aggressively shaping the public narrative. They can also use aggressive negative advertising to make elections more about character than about policy. Jacob Hacker and Paul Pierson provide a more detailed critique of the median voter theorem here, from which I am drawing.

Still, even if we accept this critique, one could still argue that in a post-Citizens United world, there’s no stopping the money. Either it goes to the parties, where there’s some possibility of accountability, or it goes to the outside groups like super PACs and 501c4s, which have no accountability mechanisms, and are likely to be even more polarizing.

The first problem with this argument is that the dichotomy is not so clear. Though candidates and parties technically can’t coordinate with super PACs, they sure come close, and they tend to all be staffed by the same larger network of operatives. Note that, for example, in the 2014 congressional elections, that 9 of the top 10 “independent expenditure” organizations were party-affiliated organizations. The one exemption was the Chamber of Commerce, though it might as well be considered a party organization, since it supports Republicans pretty much exclusively. In other words, if you actually examine the data, the outside groups sure look a lot like the parties.

Framing the choice as parties vs. outside groups also ignores another alternative: supporting small donors through a public-matching system. It is true that, in the current system, small donors do tend to be highly polarized. But a small donor matching system offers candidates a different way to run for office. Rather than go from fundraiser to fundraiser, pledging fealty to the wealthy partisans who attend, potential candidates would have incentives and opportunities to develop their own base. They’d have reason to appeal directly to voters for matched contributions that make it worth their while to get to know their constituents, who are generally more moderate than the existing set of donors. Such a system might draw in a different type of candidate, who could bring in some fresh ideas and raise new issues outside the traditional partisan battle lines. While such candidates could still face opponents funded by large outside groups, they would be more likely to have the money to be a competitive candidate. And above a certain point, money has a diminishing marginal impact on election outcomes.

There’s also something to be said for candidate-centered politics. Where Pildes and others might see “fragmentation,” I see potential for variation in ideas and coalitions, which could reduce polarization. I also see potential for candidates more focused on local and regional interests, which could bring back some of the horse-trading that many long for in our legislatures. These are possible interpretations of the patterns presented here, though they’d require more analysis. It’s also worth noting that during the candidate-centered election period of the 1960s and 1970s, polarization in Congress was relatively low and productivity was relatively high.

Like most factors in political life, polarization is complicated. The likelihood that polarization can be reduced by a single institutional change is doubtful. We ought to be especially careful when that institutional change – allowing unlimited contributions to parties – comes with the danger of further empowering a very small, unrepresentative and highly polarized slice of the population that accounts for the vast majority of high-dollar donations.

Lee Drutman is the author of “The Business of America is Lobbying: How Corporations Became Politicized and Politics Became More Corporate” (Oxford UP 2015) and a senior fellow at New America. A longer essay about this research is available.