Ah, the day-that-isn't-Presidents-day is a good excuse to get around to a response I've been meaning to get to, to a typically provocative aside at the bottom of a recent Matt Yglesias post about nothing less than the future of the current American Constitution. His view is that "US-style constitutional setups are usually very unstable," and that given the current party structure, that inherent instability may well result in significant change. Dylan Matthews agreed, more or less, and talked about what it kinds of change would mean that "divided government is capable of producing real legislation again."
As regular readers might expect, I think they're wrong. On several counts.
First of all, what's the evidence for Matthews' proposition that divided government is no longer capable of producing "real" legislation? It's true that not all that much passed in the 110th Congress, during the waning days of the George W. Bush administration. But Clinton-era divided government produced plenty of significant legislation, including welfare reform, S-CHIP, a significant minimum wage increase, and balanced budgets -- and during Clinton's second term, they handled budget issues in general with a minimum of drama and brinkmanship. Yes, the 112th Congress promises to be unproductive, but it's also little more than month old; it's a little soon to conclude that it will wind up seeming dysfunctional.
Second, and more importantly, in my view the differences between the Constitutional system in the United States and the other "presidential" systems covered in the comparative studies Yglesias likes to cite are far more important than the similarities. Mainly, the United States isn't really a "presidential" system; it's a separated-powers system, or more accurately one of separated institutions sharing powers. Congress is generally thought of as by far the greatest example of a transformative legislature, which (along with federalism, and other decentralizing features of the American system) allows lots of interests to feel meaningfully represented in government. He's absolutely right that "high stakes" elections are problematic in a democracy. But the solution isn't necessarily parliamentary elections -- which can also produce single, all-powerful winners -- but to have multiple, overlapping elections.
Stepping back a bit...I agree with Madison (see Federalists 10 and 51, of course) that majority-rules systems tend to be unstable. Parliamentary systems tend to deal with that in a few different ways. One is, as Yglesias says, by inserting another round of elite bargaining in order to form a government after the voters have had their say. Most parliamentary systems also have a strong bureaucracy that narrows the importance of the elected government. In my view, these devises (as well as corporatist arrangements, which are not necessarily related to parliamentary forms of government but do tend to go with them), are all ways of avoiding the instability of majority rule by weakening democracy. That's especially true of bureaucratic rule; corporatism and elite-driven party rule can be democratic, depending in large part, I believe, in how permeable and internally democratic the parties and other institutions are.
The American system deals with it by weakening the influence of majorities, or (ideally) by preventing majorities from forming in the first place. For example, there is no "majority" on health care reform whether there was a plurality against passing ACA last year, or a plurality against simple repeal this time, so many people have different opinions or are entirely indifferent that a victory or loss is unlikely to break the system. And then when majorities do form, the various mechanisms of the system keeps them enacting things easily. That leads to plenty of short-term frustration for electoral winners, but also means that losing an election isn't that big a deal.
However, when I talk about the ways in which Madisonian democracy promotes stability, I always try to include three dangers. The first is that if everyone in the nation believes that a single issue is of overwhelming importance, the odds of producing a true majority which will either oppress the minority or be oppressed increase. The obvious example of that situation in American history was slavery. A second possibility is if everyone is strongly ideological; if that's the case, then losing on one issue probably means losing on every issue. My reading of the evidence (and I don't really keep up on new research in this area, but I'm pretty confident that this is still true) is that most Americans are not ideological -- not just in a strong sense of having identifiable broad ideas that drive issue positions, but even in the sense that issue positions are strongly correlated. In other words, knowing someone's position on abortion will not necessarily predict their position on gun control, or labor policy, or the Iraq War, or other issues.
The third danger is extreme partisanship: if regardless of issues we all believe that our side must be in office or else all is lost, it's a problem for stability. I think that's a fair description of some activists; it fits the Tea Party profile, in my view, a lot better than ideological coherence does. But still, only a relatively small percentage of Americans feel that way. Most of us, when our party loses an election, do not automatically feel tyrannized.
Now, in my view we're pretty far from either of the latter two conditions (and the first one isn't a factor at all; there's no single issue that even a strong plurality of Americans believe is the most important). Only our most politically active citizens are either so ideological, or so partisan, that they find our elections very high-stakes contests. Of course, those politically active citizens are also, more or less by definition, the ones most likely to bring us to a Constitutional crisis, so that's worth paying attention to, but I don't see it as a grave danger.
All that said...I think that strong parties are good for democracy, but that American parties are unique (as far as I know) in that they can be strong without being rigorously ideologically coherent and without being hierarchically organized. So I'm in favor of things that decentralize (but don't necessarily weaken) political parties, and I do think that Congress works better when Members reflect not only their partisan loyalties but also strong local influences. To the extent that American parties have become nationalized (which is certainly true) and more top-down (which might be true), it's a bad development for American democracy, at least as I view it. But we're nowhere near the point at which I'd join Yglesias and Matthews and worry about the stability of the system.