I remarked yesterday on the way pundits keep trying to interpret the polls either without any historical context, or at best trying to introduce a historical context that they proceed to get wrong. All the more refreshing, then, to read a commentator who knows what he’s talking about. Step forward Antony Green, with a fascinating piece yesterday on the cycles of state and federal politics.

Green’s graphs, going back to Labor’s low point in 1969, clearly show the cyclical nature of party fortunes. The same pattern keeps repeating: a party wins government federally at the high point of its fortunes, then the tide starts to run out again and it loses seats at state level until eventually it loses federally as well and the cycle starts again.

There’s nothing new about what’s happened over the past four years, it’s only that it’s been unusually sudden. Green’s second graph, just showing the proportion of state seats held, makes vivid, as he puts it, “how Labor’s representation has fallen off a cliff at the last two state elections.”

That certainly suggests that the Gillard government is doomed; whenever a party has previously reached that low level of fortunes in the states (Labor in 1975, the Coalition in 2006) it has lost the next federal election decisively.

But the cycle also provides comfort for Labor supporters who may be concerned that their party is beyond recovery; there are plenty of precedents for parties coming back from disaster. Green has valuable things to say about the changing nature of our political parties and their class basis, but ultimately he remains unconvinced by the argument that Labor’s predicament today is radically different to what parties have faced before. For what it’s worth, I think he’s probably right.

That’s not to say the trend always moves uniformly. Labor recovered somewhat from a trough in the late 1980s, winning the 1989 Queensland election and almost winning in 1991 in New South Wales. In an earlier period, the Coalition recovered from its low point in the early 1950s (when it was out of government in every state bar South Australia), going on to win a further five federal elections — it was saved by the great Labor split.

The historical record certainly warns against explaining election results by short-term factors — the personalities of the leaders, election campaigns, the headlines of the day. No doubt those things have some influence, but clearly there are longer-term causal influences at work.

It’s not at all clear, however, what they are. Green suggests (as have several others) that being in federal government is itself causally responsible for poor state performance: “the party in office federally clearly has a deleterious impact on that party’s representation across the country.” But I’m not sure that’s the best explanation.

I’d suggest that what’s happening is that over recent decades winning government federally seems for some reason to lag a bit behind a party’s performance at state level, so that the federal win tends to represent the high point of a party’s achievement. (This seems to be connected with the fact that close federal elections always go to the incumbent, whereas there’s no such regularity at state level.)

From that point they lose ground both state and federally (federal governments never seem to get big swings in their favour, but they almost always start with a bigger cushion) until they reach a new low point, which tends to coincide with a change of federal government, starting the pattern afresh. It’s not that people are consciously “balancing” state and federal governments, just that the two cycles are not perfectly aligned.

Ten years ago, for example, the Coalition was clearly on the nose at state and federal levels, but the “stickier” nature of federal elections meant that dissatisfaction showed up mostly at state elections. When Labor finally won federally in 2007 that was its peak, and the subsequent decline has been precipitous.

The question now is whether Labor can reverse that decline before it claims Julia Gillard. History is against it, but of course there is no precedent for anything until it’s done a first time.