I have a particular loathing for arguments that involve the words “at this stage in the cycle”. It implies that public support automatically moves in certain directions depending upon how distant or close an election appears, as opposed to people changing their opinions in response to events. The big daddy of this sort of argument is the belief that governments always recover as an election approaches, or that oppositions have to be X points ahead at “this stage in the cycle” to be Y points ahead come the election.

Now, there is some logic in the argument. After all, to some extent governments have the ability to time the announcement of policies and tax cuts so unpopular things happen mid term and popular things happen in the run up to elections (though this does rather depend upon there being ideas in the making and money in the kitty). I expect there is also a causal link in the other direction: governments see they have had a good run in the polls recently and take the opportunity to call an election. But anyway, theoretical arguments aside, lets look at what has actually happened at past elections.

We’ll start with 1992 – often a popular point of comparison, for we really did have a government that was trailing by over 25 points in the polls, yet came back to win. You can see on the graph that the Conservative deficit peaked at 24.5 points, and their average deficit in 1990 was around 12 points… yet they came back to win by 8 points. However, as we know, it wasn’t a gradual clawing back of support, it was a sudden transformation after the replacement of Margaret Thatcher with John Major. Moving into 1991 the Conservatives were on average roughly level with Labour in the polls. Normally this might suggest that the government recovered even further in the run up to the 1992 election, but of course, if one looks at the polls they didn’t. By the eve of the election the polls are still showing the parties roughly even – it wasn’t that Major’s government recovered in that last year, it’s just that the polls were wrong.

Moving onto 1997, in this graph I’ve taken only ICM polls, since they were the only company who immediately learnt the lessons of 1992 and corrected their methodology enough to get things right in 1997. Here the government’s deficit against the Labour opposition peaks at 29 points, and the average deficit across the worst year, 1995, is almost 20 points. The Conservatives certainly recover somewhat from that position, eventually losing to Labour by 13 points, but while that is a substantial recovery from 1995, it’s a more modest clamber back from the 16 or so point deficits they averaged in 1996.

Onwards to 2001 and a Labour incumbent. Labour’s worst point here is obvious – a brief 8 point deficit at the height of the 2000 fuel strikes. That is very brief however, and on average even in that, their worst year, they are on average 11 points ahead in the polls. They end up winning by 9 points. So there is no recovery here – apart from that brief blip Labour’s lead gradually falls throughout the Parliament. (Some of the pollsters are still making post-1992 adjustments to their methodology in this Parliament, particularly MORI who start filtering by likelihood to vote and Gallup who drop out alltogether to be replaced by YouGov. If you take just ICM, as I did for 1992-1997, the picture isn’t too different. Their average Labour lead in 2000 is 8.4, so there is a neglible government recovery).

For the 2001-2005 Parliament we really don’t see much of a recovery at all. The government’s lead over the Conservatives plummets as they lose support over the Iraq war. Their worst position in the polls (which they hit several times) is a 5 point deficit. In 2003 their average position is 3.8 points ahead of the Tories, in 2004 it is 2.4 points ahead. In reality they win by 2.9 points – so again, there is no significant improvement from their mid-term position.

So looking at past years, there is a big government recovery in 1992… but it is caused by changing a leader. In 2005 and 2001 there is no significant recovery at all. Only in 1997 do we see a government clamber back from its very worst ratings and stage something of a recovery and, as we know, it wasn’t nearly enough to win. What is true is that governments have always recovered from their very worst position – in every case its possible to cherry pick some awful moment of hideous unpopularity from which the government recovered, but that’s no great surprise (for starters, in most cases the most extreme outlying poll results were probably rogues anyway). What it doesn’t mean is that one can take any mid-term (or now late term) position and assume the government must do better.

So, let’s look at the same graph for the position now, and see what that looks like in comparison.

We’ve got a rather excitable graph here to begin with, with the big peak of the “Brown Bounce” in the middle of 2007. Labour’s worst point is a 28 point deficit in 2008 – they’d have to go to some effort not to do better than that at the election. Their average in 2008 and 2009 is a deficit of around 12 points. If they ended up with roughly the same sort of deficit as that at the next election it would fit the pattern of the last two elections. Then again, perhaps they could recover as the Tories did in 1997 – or perhaps they could drift down further.

That’s the point – there isn’t an automatic cycle where governments are guaranteed to recover X points at the Y part of the cycle. We’ve seem them recover, we’ve seen them fail to recover; if you take one date as your reference point they go up again, if you take another as your reference point they don’t. There is no guaranteed rallying of support – if the government wants to recover support it needs to do something to earn it (or hope the Conservative do something to drive it away).


30 Responses to “Do governments always recover?”

  1. “Great minds think alike” perhaps but lets not rule out coincidence…..

    So if we learn anything from this is it that Labours best chance is indeed as Andrew Neil suggested Darling to do a Howe and for labour to ditch Brown in the summer and go to the country in the autumn……

    Peter.

    Report comment

  2. An interesting exercise Anthony which effectively kicks the assumption that governments always recover into the long grass.
    I suspect that where they do recover has something to do with where the country is in terms of the economic cycle in the 18/24 months leading up to a general election. This government is at the wrong end of the cycle and whilst things may or may not have bottomed out in 12 months time-who can say?- there is zero chance that even a sniff of the old feel good factor will have reached the nostrils of the voters much before 2012 if then.
    Sometimes a government can appear to recover and still lose because a number of supporters who have apparantly returned to the fold -according to the polls -remain disenchanted and on the day sit on their hands whilst the opposition supporters especially in the marginal seats are fired up by a determination to get the other lot out and so turn out in greater numbers than perhaps some had anticipated. I believe this happened in both 1964 and 1970.
    But a horse race is a horse race as they say. Whatever the odds are, whatever the pundits say and the form book predicts, anything can happen in a horse race and if the Labour Party can get rid of their jockey some sort of recovery cannot be ruled out. Harriet for PM anyone?

    Report comment

  3. Very interesting analysis, Anthony. However, I don’t think that Labour can get any benefit from changing their leader. John Major, for all his faults, was a likeable individual. Most of the contenders for Labour’s crown are (as far as you can trust polls about personalities) disliked almost as much as Gordon is.

    Report comment

  4. Anthony,

    Very interesting, my only comment is that the reason 1992 is held up as an example is that as the election drew closer and there appeared to be a possibility that Labour could win their policies and Leader were found wanting by many voters. The Tory vote ended up being the higest ever in numerical terms.
    In 97 the tories were well spent and Smith/Blair had de-toxified New Labour.
    In other Elections of recent times the result has never seriously been in doubt so this claw back theory does not apply as indulgencies, like Lib Dem anti Iraq war votes, could be safely carried out.

    There is feeling amonst many Labour supporters that The Cons are not popular now but mainly protest voters (polster interviewee also) beneficiaries; that Osborne is out of his depth, Cameron vacuous etc and that despite GB’s errors the Gov’t is not detested as the Tories were pre-97.
    FWIW – I think this is more like ’79 but with a stronger 3rd party; expect some claw back but still Cons victory 40-60 seat majority

    Report comment

  5. 4 parties in Scoptland and Wales.

    Report comment

  6. A very interesting analysis as always!
    Going rather further back. the precedents of the 66 – 70 and 74 – 79 Parliaments tend to give Labour cause for greater optimism.
    From late 67 to Autumn 69 Labour lagged behind the Tories by margins as high as 28% , and suffered disastrous defeats at both byelections and Local elections – including the loss of Hackney and Lambeth to the Tories in May 68. Yet the surprise in June 1970 was that Labour lost – albeit by just 2% which represented a considerable recovery.
    Whilst the 74 – 79 Govt was quite heavily defeated in May 79, it still seems likely that Callaghan could have secured – at the very least – a Hung Parliament had he called an election in Autumn 78. To believe otherwise would imply that the Winter of Discontent was not worth a swing of 2% to the Tories! Over the period mid 77 to late 78 Labour staged a major recovery – only to throw it away again in Jan/Feb79.

    Report comment

  7. Graham,

    “Going rather further back. the precedents of the 66 – 70 and 74 – 79 Parliaments tend to give Labour cause for greater optimism.”

    Why stop there why not go back to WW1 or WW2.

    Peter.

    Report comment

  8. Thanks for that Anthony – very helpful indeed.

    Every News Organisation frequently comes out with the statement that in this stage of the electoral cycle the Tories need to be 20pts ahead at this stage .As this is so clearly a fallacy, any chance of you emailing this to Sky – BBC (Newsnight / Daily Politics and the Today Programme in particular) please?

    Report comment

  9. Peter,
    I cite those examples simply because they are the most recent precedents available of unpopular Labour Governments!

    Report comment

  10. I agree in general with what Anthony says.

    I would like to add an additional point. The idea of a cycle in political opinion contains the implicit assumption that voting intentions change smoothly over time, in such a way that they can be described mathematically in terms of a sine curve or of overlapping sine curves.

    However, I have come to think (as I have indicated on some previous posts on this site) that political behaviour may change “catastrophically” in the mathematical sense of the word. Indeed, Anthony priovides excellent examples. From Anthony’s data, Thatcher’s resignation and the 2000 fuel strikes are examples of “catastrophes” (in the latter case probably two – the beginning and end of the strike). One can also see from Anthony’s graph the sudden, never recovered before 1997, drop in Tory support following withdrawal from the ERM. And the Falklands War in 1982 may well have been an earlier instance. On a smaller scale, I think there were several catastrophic changes in voting intentions during the economic crisis last Autumn.

    It is quite possible, given current economic conditions, that there will be a further step change in voting intentions between now and the next General Election, perhaps for instance as a result of the G20 meeting next month.

    I think, I fear rather cynically, that there are reasons why people stick to the idea of a political cycle. Firstly, unlike catastrophe theory, a cyclical theory of voting behaviour is tractable for polling organisations who need to use mathematical modelling techniques that produce marketable results. Which leads to the other reason for the idea of a political cycle. Ordinary people who read the results of opinion polls in the papers just about understand the idea of voting intentions gradually changing as a reaction to other events. They don’t understand the idea, or at least have not yet been educated to understand the idea, that voting intentions may suddenly change, in the same way as the ground may suddenly change shape as a result of earthquakes.

    In the past, Prime Ministers may have been able to manufacture “catastrophes” before elections, e.g. as a result of tax give aways, thus giving voting intentions the appearance of being cyclical. It is doubtful how far, if at all, this is still posible because of European commitments, of international economic interdepence and becuase, even before recent events, the size of the National Debt has become so large as to cripple the ability of ministers to initiate attractive policy changes.

    Contrary to Peter, a 10% lead is not comfortable for the Tories at present as it does not provide sufficient cushion for a “catastophe” adverse to them.

    Report comment

  11. In 1979 there appear to have been two opposite trends during the campaign.

    Labour slowly clawed back the soft/didn’t really intend to vote Tory part of the Tory lead that was still left from the Winter of Discontent,

    but it also appears that the seriously considering Tory part of the electorate could have firmed up.

    Report comment

  12. It seems to me that what is important is to think of each term as a segment of history. Each term is the unfolding of another piece of the story of Britain. Those in power often have power to depict how this story unfolds. But not always, fortunately.

    An important feature of recent history is the inability of the Conservatives from the time of John Major to find an adequate leader with the serious potential to become Prime Minister. This changed when Cameron became leader.

    Before December 2005 when he was elected as leader the polls were typically showing Lab 37 to 40, Cons 30 to 32, Lib 18 to 21. But from the time of his election the polls began to change so that from April 2006 the situation became reversed typical results being Lab 30 to 32, Cons 37 to 40, Lib 18 to 21. And this situation continued for over a year until Brown took over from Blair. The situation became reversed again but only for a few months!

    It seems to me that previous Labour voters were hoping that Brown would prove to be better than Blair. I think it is fair to say that Blair had gained a reputation for heavy, cack-handedness. In a few months Brown was perceived to be as bad as Blair. And the abolishen of the 10p tax starting rated persuaded previous Labour voters that Brown was actually worse than Blair. Thus for a period of some months Labour typically polled 25 to 28%

    Then Brown used his power and experience to change the course of the story and convinced these floating voters that it was no time for a novice.

    But Labour never got properly above the mid 30s and a few months later we are back to where we have been for most of the time since April 2006. Many of the floating voters have already become dissillusioned about Brown’s ability to control the economy.

    As the economy declines so increasing numbers of the floating voters are likely to lose faith in Brown’s Labour. And I’m guessing Labour will back down to 25 to 28% in the course of several months.

    Regarding the Lib Dems I have noted that when they have polled below 12 to18% and Lab between 30 to 32% the Conservatives benefit by rising to 40 to 42%. And so no comfort for Labour there.

    I will say more if and when Labour fall to 25 to 28%

    Report comment

  13. I agree up to a point with Peter’s repport on Andrew Neil’s comment but when he made that comment on This Week he was talking about a June election following Darling doing a Howe.

    I have to say I think that is more likely than Labour actually getting rid of Brown and finding a new leader for an Autumn election.

    The other point on looking back is that there comes a point where the desire for a change can and does override anything a Government can do. In short it may no longer be worthy of trust for the future!

    Report comment

  14. Very interesting Anthony.

    Surely the most important factors at work in any Governments term of office “events”-as you have demonstrated.

    And these events will differ for each administration.

    The key event in Brown’s period of administration is the Recession.

    If he can clearly demonstrate that we are coming out of it before the GE, he can claim the credit -as you say Anthony he will have done something to earn support. Voters may not be inclined to give him the credit-but he can claim it.

    If the economy is still in recession with no end in sight – on what basis can Brown ask to be re-elected ?

    Report comment

  15. Those of us who grew up during the 1960s and 1970s well remember that goverments did always recover at least to some extent from a midterm slump. This was not just the case in polls, by the way, but just look at the byelection and local results, for example during the last two Labour governments, in 1967-69 and 1976-78 – both even worse than now.
    This continued in the 1980s. Mrs Thatcher also did vastly better in 1983 than it seemed in 1981, and in 1986 when I started polling professionally the polls were pointing at a Labour overall majority.
    What is more, if the argument that polls always used to overestimate Labour is correct, their true standing in those 1960 and 1970s slumps was even worse than it seemed at the time.
    There are other logical reasons why this happened. In a midterm voters give the government a ‘free-of-charge’ kicking, knowing they’re not actually choosing a new administration. There is clearly a small incumbency vote which does not kick in until the familiar MP’s name is seen on the ballot.
    Of course there would not be likely to be such a recovery in the 1997 and 2001 parliaments – as there was no slump!
    I would therefore say there is plenty of evidence for the theory being discussed.

    On the other hand, though:
    Labour did of course lose the 1970 and 1979 elections. What is more, I’d take a bet that at some point during 2009 the current government will be nearer 20% behind than 10%. I do not believe the recession/depression has really kicked home yet as it will. I don’t believe Gordon Brown will benefit from this as some kind of economic potential saviour.
    I therefore guess that Labour is likely on the limited current evidence to lose the next election, although whether to a Conservative largest party or overall majority is far from obvious yet, given the electoral arithmetic bias, to me – or I would suggest anyone other than a partisan.

    Report comment

  16. I think the 1979 model is quite likely to be correct, given some similar circumstances, unpopular and ineffective Labour government, difficult economic circumstances, refreshed Conservatives not entirely trusted so they don’t get a massive endorsement. Basically, they win by default since the last lot made a mess of things, lets try someone else.

    Report comment

  17. Colin

    “If the economy is still in recession with no end in sight – on what basis can Brown ask to be re-elected ?”

    Elected, not re-elected. The fact that Brown was never elected to be Prime Minister is in itself a sore point for many people.

    Report comment

  18. I think 1979 could be a good model except it’s difficult not to see the Conservatives polling about 5% less than they did then because of a whole range of factors, especially the fact that the LDs are undoubtedly stronger now than they were in 1979 – (there were still a few seats they didn’t even bother to contest then). Since the Tories polled 45% in 1979 (excl. NI) that would suggest a Tory vote of around 40% is possible.

    Report comment

  19. Anthiony would it be fair to say that if oyu are looking at local elections governemnts in the main recover some of the electoral postion by the election?

    Report comment

  20. Onthejob – mid term elections people use local elections to give the government a bloody nose, if the local elections are on the same day as a general election they are more likely to vote with the ticket.

    If a general election isn’t four years after the last one, I’d expect the outgoing government to always recover ground compared to the mid-term elections when the seats were last fought. If the election is four years after the last one, then the effect of having a GE on the same day would cancel out.

    Report comment

  21. I haven’t had time recently to read previous threads, so apols if someone has already made this point, but has anyone notced that three out of the four major polling companies have the same conservative lead in their most recent poll as in their previous one? Only populus is 12 points v 14 points.

    Does this suggest that whilst their methodologies are different, and therefore produce slightly different results, there is some intrenchment in people’s views?

    Report comment

  22. Agan a very useful analysis and explains why the Conservatives are still 6-1 on at the bookies. THos who feel Labour have a chance should pile in to Labour at 4-1. There aren’t many places you can get that sort of return if you have belief!

    Report comment

  23. Maybe there is some correlation with the economic cycle, but the value in terms of swing is only 3-5%, so is insufficient to reverse a large deficit, and other “events” are required ?

    This would correspond to the historic perception of a swing back to the government, especially as Governments can either choose to go to the country early if it suits, or hold out to the fifth year if needed.

    Given that the economic cycle had historically been around 5-7 years long up until the period after 1992 this enabled governments a fair leeway to steer the electoral cycle closer to the economic cycle.

    Brown thus has two problems – firstly he is trying to claw back a relatively large deficit. Secondly, and more importabtly, the clock is ticking and the economic cycle – which he claimed to have abolished – is clearly out of sync !

    Report comment

  24. Looking back again at 1979 and comparing it to the present, I think the obvious needs to be re-stated, because there is more to it.

    Although the Conservatives did often lead strongly in the polls and local/by-elections in 1976-78, it neeed the Winter of Discontent to clearly give people an issue to hang their coat on.

    One does rather feel that the Tories are still in that 1976-Sept 78 phase (before Jim Callaghan’s cancelled election). They haven’t yet really grabbed people so they think – yes, we want to go with this.

    But it looks like they will win a modest overall majority of 20-40. But with a mistake or two, it could be hung.
    Nor, do I rule out something completely unexpected, as the electorate will of course decide this.

    Report comment

  25. Andy Stidwill & the Lib Dems – comparison to 1979.

    The Liberals did poorly in 1979, although it was actually 14.8% adjusted for seats contested – a bit higher than the 14.1%.

    I think the Lib Dems probably (not definitely) will poll better than that in 2009/2010, but not necessarily significantly better. They only polled 17.2% in 1997 – at a time when the Conservatives were very unpopular indeed, and gaining many seats from them. So a share of the vote quite close to 1979 is not out of the question – unless they do better against Labour.

    Report comment

  26. This is an excellent point. I always thought this- there has beennoi government recovery in the polls since 1997 in the runup to an election- the model only works in either of two scenarios-

    1. Polling 1992 and earlier, with techniques which are now no longer used

    or

    2. When Conservatives are the incumbants and Labour are the opposition- people have always loved to say they hate the ‘nasty’ (right wing) party and then vote for them anyway.

    What you will see consistently is a TORY recvovery of support in the buildup to an election- thats happened each and every time.

    Report comment

  27. But Lukw, this time the Tories middle class support has the problem of a now well established Liberal Democrat competition, the Libs will reduce the con. comback signifcantly than they would have done in the days of the less organised or well established Alliance.

    Report comment

  28. I’m not sure if the Liberals currently poll more middle class support than they did even before the Alliance, when their support was (by estimates) composed of a high majority of middle class people. The expansion in Liberal support in recent times has clearly come amongst certain sectors- students, ehtnic minorities, people in Northern Cities where they now run councils- that are very seldom middle class.

    Report comment