From Crikey newsletter June 28 2005

7. Do incumbents really have it easier?

Crikey psephologist Charles Richardson writes:

Very interesting piece in Saturday's Financial Review by Mr Mumble, Peter Brent (available on his website) who argues that conventional wisdom about the next federal election is wrong and that "Labor will probably win the next federal election," because there's an "overriding rule" that "young governments find re-election easy, while old ones find it increasingly difficult."

To support his case, Brent produces figures which show that since 1970 the success rate for first-term state and federal governments seeking re-election has been 86%, but that it declines to 33% for four-term governments and zero for fifth-term. In other words, and contrary to what I argued in Crikey last week, incumbency is not such a great advantage – or at least, to borrow Michael Warby's term, it's a "wasting asset."

I'm not convinced, partly because I think Brent's view is coloured by his extreme distaste for Mark Latham. He is certain that Latham lost because of his own failings, not because incumbency made it easier for Howard. And conversely, he is compelled to believe that Beazley is a good thing for next time. He could be right of course, but we're entitled to be sceptical.

More importantly, I don't think the figures make as strong a case as Brent thinks. By confining himself to governments elected since 1970 (but excluding those still in office), he's keeping the sample very small. Extending it to the whole period since 1909 (when the two-party system was established) gives us the following record:

Re-election rate
One term served: 63%
Two terms served: 69%
Three terms served: 59%
Four terms served: 67%
Five terms served: 70%
Six terms or more served: 77%

Historically, governments have about two chances in three of getting re-elected, regardless of how long they've been in office. The question is whether the different experience of recent times is just an aberration or whether the rules have really changed.