But it's interesting that the London School of Economics blog points out that the results of the 2010 Australian elections mean that first-past-the-post electoral systems, the Alternative Vote (aka instant-runoff voting), and other systems of proportional representation have led to a hung Parliament in every major country using the Westminster system.
Every key ‘Westminster model’ country now has a hung Parliament, following Australia’s ‘dead heat’ election (LSE):
Thanks largely to the success of the Greens in attracting one in every nine votes, Australians now have a lower house (called the House of Representatives) which is completely hung, for the first time since 1940.
...For the first time in history, the Australian outcome means that every key ‘Westminster model’ country in the world now has a hung Parliament. These are the former British empire countries that according to decades of political science orthodoxy are supposed to produce strong, single party government. Following Duverger’s Law their allegedly ‘majoritarian’ electoral systems (first past the post and AV) will typically produce reinforced majorities for one of the top two parties.
But now the table below shows that four of the five key countries have coalition governments in balanced parliaments where no party has a majority. The one exception is Canada, where the Parliament has been hung since 2004, across three general elections.
The linked paper has some interesing conclusions about Duverger's law, the premise that first-past-the-post voting systems lead inexorably to two-party systems (while conversely, proportional representation leads to a more diverse system of parties):
Few propositions in political science are as well known as Duverger’s association of plurality rule systems with two (or few) party competition and an accompanying ‘hypothesis’ (more tentatively) linking proportional representation systems to multi-party systems.
...While the DL literature undeniably opened the way for us to explore some fundamental underlying regularities in the operation of all elections, we have argued here that it extensively misattributes competition space effects to electoral system differences. Using the key concepts of the number of observable parties and 'effectve competition space', we developed a series of precisely measurable tests for Duvergerian two-party drift, and also contrasting predictions of how district-level performance should be structured if results are random or reflect only equi-probability influences (that is, assuming that all logically possible vote combinations across parties are elikely to occur).
When the ‘number of observable parties’ is just 2 or 3 we can predict the district-level results by assuming equi-probability With only two observable parties, outcomes should followbi-nomial probability distribution, defined essentially by the mean vote for the largest party (V1). For three observable parties, outcomes outcomes should reflect the multi-nomial probability distribution, defined by the mean support for the two largest parties (V1 and V2). This approach means that most of the ‘strong’ cases conventionally cited as evidence of Duverger’s Law (sucperfect 2 party district-level outcomes in many US Congressional districts) can be fully and more parsimoniously explained in ECS terms, rendering them inadmissible as support for DL effects.
Both the theoretical argument and the preliminary data from India, the UK and USA reviewed here, suggest that as more and more parties enter competition so strong automatic pressures come into play that make it less and less feasible for two-party drift to occur. Again it is not clear that any reference to electoral system effects is needed to explain multi-party vote outcomes – the driving force here is simply the increase in the number of parties getting a tiny 1 percent of the vote each and the impacts of such changes on the shaping of competition space. we might turn around Sartori’s famous (and loaded) question (2005, p. 107): ‘How much feebleness makes a party irrelevant?’ A clear implication of our approach (and of the data reviewed here) is that every small party getting enough votes to enlarge the competition space can be very relevant indeed for the evolution of party systems.
See also the LSE guide to voting systems.