In November, voters will be asked to decide 6000 state legislative elections, with 87 of the nation’s 99 state legislative houses holding their regularly scheduled elections. But, on the day after Election Day, all of these elections will be understood by the mass media as having asked essentially only one question: Did voters support Donald Trump? The nationalization of state politics poses serious legal and constitutional problems for our federal system.
Consider how the New York Times, in both news and opinion pieces, covered state elections in Virginia in 2017: “Virginia’s Election and Trump’s Whupping”; “In Obscure Virginia Races, a Test of Anger at the President”; “How the Virginia Governor’s Race Will Be a Test of Trump’s Impact”;“Virginia Rejects Your Hateful Politics, Mr. Trump.”
There is a logic to this focus. An increasingly large literature in political science has made clear that voters in fact respond almost entirely to national politics when voting for state legislature.
Look at this graph, created by Steve Rogers:
There is a very tight correlation between Congressional voting and state legislative voting. Further, events that have nothing to do with state policy drive election results, whether it was the war in Iraq or shifts in interest rate policy at Federal Reserve. When people vote for their state legislature, they are most often expressing their preferences about national politics.
In his new book, The Increasingly United States, Daniel J. Hopkins shows that this is a product of changes in party politics and in the mass media. Declining local media makes it less likely voters will learn anything about state and local officials. Political parties have nationalized and polarized, leading voters with little knowledge of state officials to rely increasingly on national party label when voting in state elections.
In the parlance of European political scientists, state legislative elections are increasingly “second-order,” or dependent on what happens at another level of government. Voters know little about individual candidates or state parties, and for both rational and group-based affective reasons, end up voting for members of the same party across levels of government.
Let that sink in: What state legislatures actually do has little effect on whether they are reelected in general elections. How well they fund infrastructure, pensions, or education doesn’t matter much. How they shape tort, contract, and property law doesn’t really matter. The way they shape labor, occupational licensing, land use, and environmental regulation doesn’t matter. More than any other factor, how voters feel about President Trump will determine state legislative elections this cycle.
The 2018 election will cause huge shifts in public policy in states and localities in areas like health care, housing, labor, and education, as Amy Liu of Brookings shows. But what the candidates for the workhorses of state policymaking say for themselves or actually have done will hardly affect the results of those elections.
Notably, state legislative races are far worse in this respect than Governors races. In high-profile chief executive races, candidates can develop their own brands, allowing voters to judge them on their promise and performance and not national politics (at least to some degree). Incumbents from parties not popular in their states — like Governors Charlie Baker (R-MA) and Larry Hogan (R-MD) — look likely to get reelected this year despite their co-partisans having no chance of winning down-ballot.
While political scientists are increasingly writing about this subject, the failure of state elections is a legal and constitutional story as much as it is a political or media story.
It is no particular problem if voters do not split their tickets in national races – Democratic candidates for Senate are quite like Democratic candidate for the House of Representatives or for President. But a failure to split tickets across levels of governments has a very different effect. Second-order state elections means that state government lacks meaningful accountability or representation.
As I show in the Texas Law Review, any justification for federalism in a democracy — whether it is the greater fit between preferences and policies, sorting, laboratories of democracy, or protection of local identities — will rely heavily on state elections working to create representative, accountable, and locally-differentiated politics. If elections simply produce either “red” or “blue” outcomes, there will be fewer choices for mobile residents, worse matching of preferences to policy, less representation of diverse cultures across states, and less policy experimentation.
If state elections don’t work to provide locally-representative and accountable outcomes, the case for decentralization becomes much weaker. Federalism needs effective state politics.
Some, like Jessica Bulman-Pozen in this terrific piece in the Harvard Law Review, put a positive spin on second-order state elections by arguing that when state politics follows national politics, it can produce benefits in serving as a minor league for national politics. But it’s not clear that having second-order elections leads to state government providing a better breeding ground for national politicians or ideas than a differentiated one would produce. Many effective national politicians — think Bill Clinton — rose up in state political systems that were quite different from national politics, and thus were able to build personal brands that were distinct from unpopular national party brands. And the creation of new coalitions or ideas that might shake up national politics may be inhibited by national partisan voting at the state level.
But, more significantly, state politics is extremely important in its own right. States and localities employ a huge majority of government employees and regulate lots of economic activity. Many states — Illinois, New Jersey, and Connecticut foremost among them — are in horrific fiscal shape and the absence of political competition and accountability may help explain why. Further, many states and localities are causing a huge amount of damage to the national economy through their regulation of housing and labor markets. The lack of state and local-specific party competition allows well-organized interests — homeowners, regulated professions, etc. — to dominate politics in these areas. The lack of state-specific party competition in state politics is very, very costly.
One might think that second-order state elections aren’t a problem because Democratic states get Democratic state legislatures and vice versa for Republican states. But preferences on state issues are not necessarily correlated with national issues, as the federal government deals with issues that are extremely distinct from the issues state governments do, including war and peace and macroeconomic stabilization. The absence of state-specific party competition links issues in ways that a more differentiated politics might not.
But more basically, we should want political competition even in states that are dominated by one party in national elections, so that the interests of the state median voter get listened to. In overwhelming Democratic or Republican states, we don’t get competition across the real differences of opinion that exist among state voters. And primaries, where one might assume such competition occurs, are both low turnout and low information. Primary voters, after all, do not not get any useful information on the ballot about which faction of a party a candidate comes from, making it difficult for voters to hold groups of legislators accountable for outcomes. And, of course, primary voters are not representative of the broader electorate.
People who like federalism think allowing states and localities to make decisions will produce outcomes that match the wild diversity of American life. But second-order elections cannot deliver outcomes that match the differences we see when we look across the cultures and economies of our fifty states.
Broad cultural trends contribute to the nationalization of state elections, but law plays a central role as well. The forces discussed by Hopkins and Rogers are matched by state election laws that do little to encourage state elections to serve the goals of federalism. As Chris Elmendorf and I have argued, state election laws limit the ability of losing state parties to rebrand themselves along state issues. Laws governing eligibility to vote in primaries, ballot access, and the ability of parties to participate across levels of government make it very difficult for parties to develop level-specific coalitions and voter bases. And primary ballots do not allow voters to see which faction an official belongs to, making them less than useful for generating level-specific group conflict among partisans.
These laws can be reformed to help disentangle the state and national elections.
Election laws could make an effort to differentiate state parties from their national co-partisans in the minds of voters, allowing state parties and officials to be held accountable. For instance, the ballot itself could identify which party controls the state legislature (most voters don’t know) and identify state officials as members of the state party, separating them in voters’ minds from national officials. Primary ballots could include information about factional or interest group endorsements. State constitutions could also – and to be fair, increasingly do, as excellent work by Miriam Seifter has shown – favor Governors over legislatures, as Governors have more individual brands and thus are easier to hold accountable for their actual behavior in office.
On both the left and right, scholars and public intellectuals have renewed their interest in federalism as a way of dealing with the divisions in opinion in American political life. But these advocates should consider whether state politics deserves their faith and what can be done to improve it.