The problem with federalism

Accountability doesn’t function so well in an ideological setting.  Here is a Princeton political science job market paper from Steven Rogers:

Theories of political accountability suggest that governing parties and their members should be electorally punished when they perform poorly in office. However, I find little evidence of this type of accountability in state legislatures. State legislative elections are not referendums on state legislators’ own performance but are instead dominated by national politics. Presidential evaluations and the national economy matter much more for state legislators’ elections than state-level economic conditions, state policy outcomes, or voters’ assessments of the legislature. Previous analyses of state legislative elections fail to consider which party controls the state legislature and whether voters know this information. When accounting for these factors, I discover that even when the legislature performs well, misinformed voters mistakenly reward the minority party. Thus, while state legislatures wield considerable policy-making power, elections are ineffective in holding state legislative parties accountable for their own performance and lawmaking.

Hat tip goes to Matt Yglesias on Twitter.


State and local elections have become (or always were) partisan affairs that are built upon a foundation of the national parties' reputations. This is disconcerting when you consider that party affiliation is more closely aligned with geography at the local level than at the regional level. Thus, the norm becomes (or always has been) one-party rule in local and some state elections even where, nationally, or in larger, poltically diverse states, the two parties are competitive with each other.

I find an interesting comparison with India where the typical complaint is the converse i.e. National Elections are won or lost on local / regional issues or performance. Wonder what causes the contrast.

Cultural differences, perhaps?

India has many different peoples with their own languages and histories and identity, so someone might consider himself Gujarati first and Indian second. In America, on the other hand, tribal affiliations are less strong, and not strongly associated with specific areas--side effect of an immigrant's nation. Even the proudest Texan would be hard-pressed to consider himself Texan first and American second. And given Texas's history, a Texan-first attitude would be more possible than, say, Iowan-first or Arizonan-first.

I'm less knowledgeable about culture between groups in India, though, so let me know if I'm mistaken.

I think that's a good explanation. I also suspect that low levels of literacy / education might translate into a more regional outlook.

So also, poverty: You'd care more about cheap potatoes than foreign policy positions.

Also, high mobility in America.

I was born and raised in California, but I've lived and worked in Pennsylvania, Michigan, and New Jersey as well. Although I do have clear Californian cultural markers that others are quick to notice, I vote as I live, so I've officially been Pennsylvanian, Michigander, and New Jerseyan at different times, as well.

I wonder if it has something to do with access to and ability to digest information at the local level. Roughly 3/4 of the population lives in rural areas. Does national news filter through all those areas? Also, while there is a literacy rate over 70 percent, there is no language that's native to the majority of the population. Though I believe differences between some languages are minimal, could that still be an inhibiting factor?

"Even the proudest Texan would be hard-pressed to consider himself Texan first and American second"

It seems that you've never lived in Texas, you can in fact find people who are Texan first and American's second. Granted there are not a lot of them, but they exist. The larger population is where being Texan is only barely 2nd to being an american.

Local elections are massively susceptible to the "small groups Who Care" problem. In a school board election, you don't have a clue who these people are unless you work closely with the school district or pay lots of attention. So, school boards end up being dominated by teacher unions in many areas, or sometimes by so-con groups. Even elections that should be more high-profile, such as mayoral elections or state legislative elections and up dominated by local machines in my experience - this is definitely the case here in California, where the government-union machine runs all.

Mr. Im-moderate-so-please-keep-giving-me-MSM-scraps says that a study showing voters throw out Democrats from the state legislatures because of a socialist President means federalism does not work. Another profile in courage from the eunuch sell out.

Do you realize this isn't even a complete sentence?

you pedantic asshole.

How dare you call someone pedantic.

That explains why Colorado just went Democratic in both houses as well as governor?

"Thus, while state legislatures wield considerable policy-making power, elections are ineff ective in holding state legislative parties accountable for their own performance and lawmaking."

Solution: reduce state legislatures' policy-making power.

Another solution, if one wishes to retain a role, would be to move to more of a parliamentary system. The paper does suggest that elections are effective in holding the party of the governor accountable. People just have a better idea of who is governor than who controls the state Senate or state House.

I am skeptical as to how the performance of the governor and the legislators of the same party can be separated like this. The legislature works in concert with/opposition to the governor. It is perfectly logical to try to increase or decrease the governor's power through changing the composition of the legislature.

Nor do I see how one can separate the performance of the two. Unless there is a scandal or a major breakdown in government, the executive and the legislators in his/her party are working together on a single agenda. The 2010 Congressional elections were largely seen as a referendum on Obama's performance, for example, and I see nothing illogical about that.


One data point from Wisconsin - I reached out to multiple candidates for state office in advance of the election and despite lots of questions, I could not get a policy position beyond: "I agree with Scott Walker," or "I do not agree with Scott Walker."

Though the study does note that people do recognize the governor's party and often do vote on the basis of it, rather than just national approval of the President (or their approval of the legislature.)

That makes me wonder whether or not a similar study for legislative results nationally would also be dependent upon approval ratings for the President instead of the approval ratings for Congress. People do actually show greater confusion about who controls Congress-- I remember reading polls showing in 2006-2007 a significant percentage of people thought that Republicans still controlled Congress, presumably because they knew that George W. Bush was President. (And even more striking results in 2001-2002, with the split legislative control.)

If that is true, then perhaps the implications are not so much for federalism as they are for separation of powers (and bipartisanship) and whether or not it's worth it for the legislature to work with the governor when the governor is from a different party.

If good policies are rewarded on the national stage, but not necessarily rewarded on the state level
if bad policies are punished on the national stage, but not necessarily punished on the state level

If your policies are good, which level of gov't would you prefer to enact them? If bad?

This highlights what is a major charm of unitary (no-federal) parliamentary systems, that the voters know clearly who to blame and punish. Along the same lines, I think one of the EconLog guys proposed an elected for long terms but re-callable executive as the sole elected official of local government (perhaps with one other elected guy with a job to watch him). That said, the within group variation seems much larger than the between as there are plenty of basket case unitary parliamentary governments and relatively well run federal and presidential systems. Even the unitary executive of local office has some extant examples. The WSJ reported recently that Georgia (USA) has nine counties with sole commissioners who are the one elected official that calls all the shots. Not exactly a paragon of good governance but some of the parliamentary city states like Singapore are effectively this, with the prime minister as elected dictator and the parliament's primary job to fire him if things get out of hand.

The paper strongly suggests that the voters know clearly whom to blame and punish in a federal parliamentary system, like those of Germany, Australia, or Canada. The paper indicates that people know who the governor is, just not who controls the state legislature.

And that there is a variable indicating that voters have a separate effect for punishing based on governor's party and state conditions from the effect based on president's party and national conditions.

We mix ideology with performance. And then vote for everybody on one day. I suppose it used to be cheaper to do it this way, though perhaps today it is more expensive. With the increased emphasis on ahead of time voting we seem to be bumbling and stumbling towards improvement. We won't just come out and say "we don't have to vote on one day, so let's not" but we seem to be moving that way.

A governor (w/ legislature) or mayor (with city council) is more analogous to a president (with congress) than a parliament and so the responsibility is relatively defuse. Voters may blame and punish executive in presidential systems but officials seem to deserve their punishments and blame more in parliamentary ones.

Rings true. I barely know who my representative is, though I know whether he has a D or R after his name.

So, the problem with federalism is democracy? That's what we said, right?

So move the voting days?

In Canada the federal and provincial elections follow different schedules. Maybe we can say this reduces voter conflation of accountability between the two levels, although eg there is some relationship between support for the BC and federal Liberal parties despite the fact they do not share an affiliation and some BC Liberals are federal Conservatives.

A day or two ago it changed so only post titles come through my RSS reader, I have to click through to read the post. Was that intentional?


I don't know if it was intentional in this case, but the point of not putting the full articles on RSS is usually to get you to come to the site (and thereby increase the view count of the ads).

Sure, but federalism at least allows people to vote with their feet, albeit on a shrinking set of policies that are under state/local control.

Also in favor of federalism in America: our long history of state semi-autonomy, the vast size of the county and its large population.

+1. The OP does not understand how 'ordinary people' actually vote on issues.

Matt overlooks the power of gerrymandering in this country.

I'm curious as to whether this has been/can be tracked longitudinally. I wonder if the strength of the effect specified in the paper has increased with the growth in the size of the federal government (measured however you want.) It seems to me that as the federal government has gotten larger, it requires more focus, and therefore people are less likely to spend their scarce time investigating local politics.

An argument against this would be that, in actuallity, state and local law making makes a much larger difference on people's lives, so they should be paying closer attention to legistlative outcomes on that level. But either they've been misled, or my original proposition is competely wrong.

Ahh... the RSS feed on todays posts don't include the usual descriptions/full post. Please tell me you guys are not going the way of Krugman and others to raise site hits. Scanning massive amounts of blog generated info can be exhausting if only titles appear in the RSS feeds. Please correct this issue if possible.

Referendums? Agendums?

" I fi nd little evidence of this type of accountability in state legislatures ... instead dominated by national politics". How else to explain how Democrats in Illinois managed to keep easy control of the legislature after kicking the fiscal problems down the road for the last several years?

Mood affiliation.

Federalism was enacted with the objective of splintering power among local, state and Federal nodes, and splintering those nodes themselves among legislative, executive and judicial, and then splintering the electorate itself- permitting popular votes, legislative votes (for Senators) and executive appointments. The parties have overrun that system so that each part now works for the political party, not its intended (by the founders) constituents. Federalism has made it harder on the parties, but party sustenance and expansion is now the primary goal of every one of those nodes of power.

I did a very quick skim of the paper and I noticed that there was zero discussion of the impact of primaries. Yes, we have places where there are effectively one party states, but bad state legislators may be evaluated and replaced in the primary rather than the general. Certainly, my hometown was a one party affair that had the real election in August (Republican primary) and then dutifully voted in November to confirm the winners. With term limits, part-time state legislatures, promotion (to run for Congress or state-wide office), and other means of ensuring higher turnover, primaries may well be the biggest performance check on legislators with various faction within the party (e.g. fiscal vs social conservatives for Replublicans, environmental liberals vs organized labor Democrats). Even if the entire partisan decision is decided by the president, who is on partisan ballot line may still be a determined by performance.

The other shot is, I'm not sure how you can accurately measure the effects of federalism in the current era. Back in the day there were a wide variety of issues - education, basic crime, roads, etc. that were issues the federal government didn't touch; these days the federal government clearly acts in all of them. Every time decisions are arrogated to Washington DC, voters have less useful information to weigh state legislators. If you like the new bridge built next door as a voter, how are you able to tell if this is because your state government put up funds, the federal government coerced the state and paid for it, or some hybrid action took place? Given the heavy interplay between the feds and state governments, it may well be a rational choice to vote for or against national parties - a Republican state legislature would be better able to resist unpopular national Democratic intrusion while a Democratic one would be better able to advance Washington's agenda in the the state.

Yes, I think primaries are the best shot at performance accountability as long as you want to keep the party ideology. But incumbents don't always have strong primary opposition. It's just hard to hold incumbents accountable in general.

This seems entirely wrong to me.

Republicans did poorly in the recent presidential election, less well than expected in the sentate but as expected in the house, and well in state elections where they are gaining control of more and more states.

That strongly indicates that voters are willing and able to make individual choices regarding candidates.

I am from Pennsylvania - a pretty solidly blue state - with one senator from each party, and a history of mostly republican governors.
The state chambers have been evenly divided and are rarely controlled by the same party.

Within the past several years we voted out a large number of incumbents over midnight pay raises.
The backlash was so great that even those state court justices who approved the measure were defeated in retention elections.

It is my experience that voters in my state make their decisions in elections independent of what might be going on nationally.

I think the implication of this research is that local poltiicians are increasingly defined by their poltical party and that poltical parties are increasingly defined by national politics rather than local poltics. Thus, it is difficult for relatively liberal Republicans to compete effectively as anything but members of a minority party in the northeast and conservative Democrats as anything but members of a minority party in the southeast, for example.

You may not see this in Pennsylvania becauser it is not "soldily blue" in any meaningful sense. This past election was 52-47 for Obama in 2012, 54-44 for Obama in 2008, 51-48 for Kerry in 2004, and 51-46 in 200 for Gore. The Dems won all of these, but the Republicans were competitive in every year but 2008, and in no year did you see the sort of one-sided result that is common in a majority of states during these same elections.

In the states that are one-sided in the presidential election it has become increasingly difficult for the minority party to deviate from its national party and be competitive in local and state legislative races. In state-wide elections for governor or senator it is a bit easier, because any party can have a bad or good candidate or a special situation that leads to an otherwise unlikely outcome.

"Republicans did poorly in the recent presidential election, less well than expected in the sentate but as expected in the house, and well in state elections where they are gaining control of more and more states."

Some of that has to do with district boundaries. I think it was PA where Democratic house candidates state wide received more votes that their Republican opponents, but only 5 of 18 were elected.

Canada works very differently. Provincial politics and federal politics are completely unrelated. Even when federal and provincial parties have the same names, they rarely have any closer connection than, say, the Republican Party has with the UK Conservatives. The standing of national parties basically doesn't matter in provincial politics in Canada, and vice versa.

I've always wondered why that is. I wonder why, for example, some challenger to the California Democratic Party not named the Republican Party does not emerge, or why someone doesn't start a Conservative Party in Utah.

They do; California has a number of parties not represented nationally, such as the American Independent Party, or the Peace And Freedom Party. New Jersey had a presidential candidate from a party called "NSA Caused 9/11." However, these parties receive little attention, typically less than even the Libertarian and Green parties. Starting a Conservative Party won't do enough.

I realize there is a colourful tradition of fringe American third parties and of course nowhere more so than California. I am talking about something different. In Canada, provincial parties are essentially completely distinct from federal parties. In British Columbia, where I live, for example, most of the supporters of the "Liberal Party" provincially over the last four electoral cycles probably voted Conservative federally. BC was ruled for all but three years between 1952 and 1991 by a party that had no federal counterpart. BC is unusual in various ways, but the same pattern can be shown in most provinces and has become more marked over time.

Related to this, when people are mad at (or like) the incumbents in a province, this has been demonstrated to have no effect on the fortunes of their ideological bretheren in federal politics, and vice versa.

There are good structural reasons for this, at least in the Canadian context, because what it takes to put together a majority coalition in a region is different from what it takes to put together a national coalition. In addition, the issues and cleavages are different. (Quebec politics has always been dominated by independence.)

In contrast, there is no American state I am aware of where the two major parties are not the Republicans and the Democrats. Of course, the Nebraska Democratic Party is different from the one in the District of Columbia. But still, a Republican federally is a Republican in state politics, and the same is true of Democrats. This remains true, even though (for example) the California Republicans have zero chance of winning legislative elections. It's a bit of a mystery from my perspective.

Perhaps, the answer is related to the fact that no post-Confederation premier has ever become prime minister of Canada, while pretty much ever American governor sees themselves as a future President.

Charles Tupper. I'm not Canadian, but I recently read some Canadian refute that "fact".

Insert Captain America pun here.

1. Maybe what the author regards as poor performance in office, the voters regarded as strong performance. Which is really the entire point of federalism - let the people of that locality decide what they consider to be good policy, not people of other localities.

2. Even assuming that state legislatures were actually rewarded for what the voters considered to be bad behavior, this is not a problem with state level governance v.s. federal level governance. This suggests a problem with democracy, and low levels of political information among the populace as a whole. That doesn't mean that people outside the locality will be any better.

3. The paper underestimates the degree to which party loyalty matters in local elections. A state may have a crappy republican in office, but the citizens still may want a republican again because they perceive that to be better than the alternative given their political preferences.

4. The abstract/paper says nothing of the moral argument underlying federalism - that outsiders shouldn't tell people of a locality what is good or bad for them - that people have a right to political self-determination on issues that only affect them locally.

The problem is the media. CNN and Fox News only cover national politics; and local newspapers are in decline. The intellectuals' radio station of choice, NPR, can only cover national news. Decisions made in Washington dominate not only the news, but the cultural sphere too. When Lady Gaga endorses Obama, it's a big deal to those people who follow popular music; but she is unlikely to endorse Joe Sixpack for small-town sanitation officer.

Media coverage is essential to good accountability. For a decent amount of coverage and diversity of opinion, you need a large population area with dedicated media outlets.

Interestingly, in Europe the problem is reversed. Each country's media only covers its own elections; but coverage of the European elections is minimal.

One solution might be to scrap political parties. If voters can't just instinctively choose Democrat or Republican, but actually have to choose individuals on merit, they might pay more attention to their track record. (This theory remains untested.)

This is similar to my feeling. Johnson & Johnson and P&G benefited from mass marketing.

And that's my charitable theory on why NPR is left-leaning. They cover basically two things, national politics which leans left by its nature and goofy nonsense. Where would I be without knowing that a brothel owner was elected from Nevada or what it's like to be a transvestite from This American Life? It's just a cost thing.

"Theories of political accountability suggest that governing parties and their members should be electorally punished when they perform poorly in office. However, I fi nd little evidence of this type of accountability in state legislatures."

It is more than theory, it is reality.

The state legislatures draw the congressional districts following the taking of the census each ten years. The lack of accountability in those legislatures then spills over into the people's House via gerrymandering.

Only the presidential election is not based on the popular vote, but rather is based on the electoral college dynamics.

More people voted for Democrats in the House election than for Republicans by ~500,000 votes, nevertheless the Republicans retained the majority of the House.

Studies done well before the election show that it was gerrymandering, which was a no-brainer.

Alternate reading: this result shows not that federalism is weak but that the US is inadequately federal. National politics dominate local because the national government has arrogated to itself functions properly left to the states, and has vastly encroached on the people's liberties. Were the federalist principle reinvigorated and the federal government returned to its intended limits, state issues would dominate in state elections.

Hey I am so grateful I found your web site, I relaly found you by accident, while I was searching on Aol for something else, Nonetheless I am here now and would just like to say thanks a lot for a tremendous post and a all round thrilling blog (I also love the theme/design), I don’t have time to look over it all at the moment but I have saved it and also added in your RSS feeds, so when I have time I will be back to read a lot more, Please do keep up the great job.

As with everything else, this just proves that my political ideology is correct and we need to abolish the national government.

Yeah, I'm not sure what their baseline for accountability is when vote tallies is the baseline for accountability.

It's a bit like complaining that the best football team didn't beat the spread.

This is a problem with the two party system, not federalism.

Comments for this post are closed