COMMENT : Losing the next election should tax even Labour

`There are reasons to think that a further rise in tactical voting is more likely than not. If it happens, the Conservatives will be in real trouble'

Some elements of the Labour Party were showing signs this weekend of embarking on the one route that could yet lead to electoral defeat: the route marked "higher taxes". A poll in the Sunday Times showed that a significant minority of Labour backbenchers, apparently forgetting the blindingly obvious lessons of the last four general elections, would like to levy a 50p top rate of tax at a starting income level of around pounds 40,000. As an object lesson on how to pluck defeat out of the jaws of certain victory, this takes some beating.

Nevertheless, a study I have been doing of the psephological state of the nation shows how hard it will be for the unreformed elements in the Labour Party to hand victory over to the Tories, however determined they may be. Psephology, it is true, took a nasty knock in the 1992 general election, and has barely since recovered. But I would offer the following five "facts" to consider, all of which suggest that the Tories face an even more difficult task than has been commonly realised if they are to be re-elected.

1. The finishing line is different for the two parties. Because of the way that the minor parties will presumably choose to vote after the election, Labour needs fewer of its own MPs than the Tories to form a government. Specifically, either side needs 326 votes in the Commons to form a government. The Conservatives can count on the support of 12 Ulster members, meaning that they need 314 seats of their own to stay in office.

Meanwhile, Labour will probably attract the support of at least 36 members of other minor parties (Liberal Democrats, Nationalists and the SDLP), implying that it needs only 290 seats of its own to form a government - 14 less than the Conservatives.

2. The electoral system is now skewed against the Tories. In 1992, the gap in seats between the two main parties (65) was far smaller than would have been expected, given the fact that the Tories led Labour by eight points in the popular vote across the country. Each winning Labour MP required fewer voters than each winning Tory, partly because Labour seats were smaller, and turnout was lower, and partly because the Tories piled up needlessly large majorities in many seats. This problem has been only partially redressed by the subsequent Boundary Commission changes, which will re-allocate only about six extra seats to the Tories (compared with the 20 generally expected beforehand).

If there are uniform swings in all seats across the country next time (admittedly a big if), then Labour requires a much smaller share of the popular vote to reach its finishing line target than the Tories.

In fact, if Labour obtains 36-37 per cent of the popular vote (compared with 34.4 per cent last time), it will probably be able to form a minority government. With 39 per cent of the vote, it could end up with an overall majority. Meanwhile the Conservatives need 40-41 per cent to form a minority government and 41-42 per cent to win an outright majority.

3. Regional swings seem to be favouring Labour. The arithmetic just quoted assumes uniform national swings, but this no longer happens in Britain. Nowadays, there is great variation between regions and between different types of seats within regions. This makes it harder to translate national shares of the vote into seats, but as far as we can tell, the Tories may end up doing worse than they would with uniform swings. The Gallup 9000 poll reveals that the smallest swings to Labour since 1992 are in its areas of greatest strength - Scotland and Wales - while the greatest swings are in areas where the marginals are concentrated - the North-west, the West Midlands and London. According to David Walton, of Goldman Sachs, these regional variations would add about 20 seats to the Labour total, compared with the operation of uniform national swings.

4. Tactical voting is becoming more important. Conventional wisdom at Annie's Bar in Westminster is that tactical voting is significant only in by-elections. Not so. According to an excellent analysis by John Curtice and Michael Steed in the Nuffield study of the 1992 election, there has been a creeping tendency for tactical voting to spread in key seats for at least two general elections. Last time, this particularly helped Labour in two-party marginals versus the Conservatives, since in these seats there was a disproportionate squeeze on the Liberal Democrat vote. The swing to Labour in such marginals was 1-2 per cent greater than in the rest of the country - which may not sound much, but which probably delivered a handful of otherwise unwinnable seats to Labour.

Nobody can say whether tactical voting will be even more significant next time. Certainly, the change in constituency boundaries will make tactical voting more problematic. But with the national relationship between Labour and Liberal Democrat parties much warmer than before, and with Labour breathing down the necks of Tory incumbents in many more seats, especially in the south of England, there are reasons to think that a further rise in tactical voting is more likely than not. If it happens, the Conservatives will be in real trouble.

5. The polls may be wrong, but they cannot be that wrong. Neither John Major nor Tony Blair is choosing to believe the present opinion poll results - the first to avoid despair, the second to avoid complacency. Both are inclined to believe that much of the Labour lead is illusory, and that there may still be some lying in responses to the pollsters. But how much of the Labour lead can these factors really account for? The final opinion polls before the 1992 election showed the likely result as a dead heat; the exit polls on the day showed a four-point Labour lead; and the actual result was an eight-point point win for the Government. According to Peter Kellner, three points of the eight-point polling error were due to a late swing and to differential turnout, which may go either way next time. A further two to three points were due to sampling error (uncovering too few Tory voters in remote rural areas, etc), a problem that should now have been eliminated. This leaves about two to three points for lying to the pollsters. Even if this is still happening, which is uncertain, it is a drop in the bucket compared with the 30-point Labour lead shown in recent opinion polls.

As the graph shows, the opposition lead is not only much larger than it was in previous parliaments, but it has persisted for much longer than ever before. And one final war game shows how difficult the Government's task now is. About a year before the 1992 election, the polls showed Labour roughly eight points ahead, while on election day the Tories won by the same amount. Hence the sum of polling error and the final year swing back to the Government was equivalent to 16 points.

It would need twice this swing back to the Government in the next 12 months for John Major to win again.

The arithmetic, therefore, looks daunting for the Tories. But the tax- raising minority of Labour backbenchers may still be sufficently obtuse to engineer a shock defeat for their party. With enemies like these, Mr Major must be thinking, who needs friends?