Like everyone else, I was expecting a hung Parliament in 2015. The day before the election, the odds of a Conservative majority, even on the best-informed pollsters’ websites, were about 10%, and those of a hung Parliament 90%.

This is not a hung Parliament. But nor does it mark the return of two-party politics, as is obvious from the vote-share figures. It is certainly possible that the Parliament will become hung in the next five years through by-election losses and perhaps defections from the Conservatives to UKIP.

My main claim is while this Parliament is not hung, the outcome of the election – the Conservative majority – is a direct consequence of the fact that the last Parliament was a hung Parliament. That implies a counter-factual: that if it had not been – if the Conservatives had won a slim majority in 2010 – they would probably be out of power now.

This was a deliberate campaigning choice. These Liberal Democrat seats – you can find them in the clickable map above – were not those with the smallest majorities. There were many Labour seats with smaller majorities. But the Labour vote was (rightly) believed to be fairly solid. This was why the Conservatives abandoned the London marginals to Labour: even if the Labour vote would not grow it would not shrink. But almost all the English Liberal Democrat seats were vulnerable to the Conservatives because the swing against the Liberal Democrats had been so strong. The Conservatives did not even need much of a direct swing to them as a party. The British Election Study, about six weeks before the election, predicted that only around a sixth of the Liberal Democrat vote would go to the Conservatives. Much more, up to half according to some recent estimates, went to Labour. This didn’t help Labour. Labour only took 12 seats from the Liberal Democrats; the Conservatives took 27. But anything that eroded the Liberal Democrat vote in England benefitted the Conservatives.

WHAT EVERYONE KNOWS

Let’s start with something everyone knows about British politics.

“There have been few exceptions to the British norm of one-party majority Cabinets.”

This is from Arend Lijphart’s Patterns of Democracy. It is broadly true of the period since 1945. But if we look over a longer period, it seems quite wrong. If we go back to the 1884-85 Third Reform Act, we can see single party majority government is far from the norm. Minority governments and coalitions are more common.

The table below shows British Governments since 1885. Before 1945, single party majority government was only the case for the ‘New Liberal’ Government before the First World War (and only up until 1910 when it became a minority government) and when the Conservatives governed in the 1920s (and these were bordered by a coalition and two minority Labour governments).

The norm at other times has been for minority and coalition government, albeit of a peculiar kind. The Liberals in 1886, in 1892, and twice in 1910, relied on eighty or so Irish Home Rulers for their majorities. On the other side, the Liberal Unionists – the opponents of Home Rule for Ireland who split from the Liberals in 1886 – gave the Conservatives their majority between 1886 and 1892. When the Conservatives won elections again in 1895 and 1900 they took power as a coalition with the Liberal Unionists, not as a single party majority government. The first two Labour Governments were also minority governments, in 1924 and in 1929. The first was sustained by Liberal fears of a fresh election and of being seen not to ‘give Labour its chance’, and the second by an informal agreement which the Liberals hoped would deliver electoral reform. When the second one collapsed in 1931, it was replaced not by a Conservative Government, but a National Government which included – indeed was led by – the Labour leader Ramsay MacDonald. There were also coalition governments in the two world wars.

Even in the period since 1945, there have been three minority governments – the two everyone knows about – February to October 1974, and again when Labour lost its majority in 1976 – and the one people always forget – when the Conservatives under John Major lost their majority at the end of 1996.

So the first point to note is:

(1) Single-party majority government has not always been the norm.

The tables below show the minority and coalition governments in more detail.

But there are two further points that these charts show. The first is this:

(2) Hung Parliaments hardly ever lead to coalitions.

The hung Parliaments are indicated in green in the left-hand column of the first table, and they almost always lead to minority government and not to coalition. The exceptions are, of course, 2010, and also the coalitions of 1915 and 1916. 1915 and 1916 are better regarded as wartime coalitions. In 1915, Parliament had been hung for over five years sustaining a Liberal minority government. It was the need for wartime unity that prompted coalition, as it did again in 1940 when Parliament was not hung. So 2010 is a unique case: a peacetime hung Parliament that produced a coalition rather than a minority government.

How should we explain this strong historic preference for minority government? First, thanks to the electoral system, when Parliaments are hung they tend to be only just hung. A large party faces the choice between a bare minority government and a coalition with a small party. The large party tends to prefer minority government, because it does not have to share policy, posts and patronage with a partner, but only make such compromises as are needed to secure legislation. Sometimes no concessions have to be made because the opposition parties fail to combine. Furthermore, because the timing of the next election has until recently been the Prime Minister’s prerogative power, he or she can also use it, as head of a minority government, to improve his or her electoral chances. Equally, small parties are deterred by the obligation, in a coalition, to compromise in advance, rather than offer or withhold support as suits their changing interests. Above all, they worry that being a permanent minority in a coalition will leave them unable to maintain the distinctiveness of their party appeal. Cabinet responsibility in Britain makes it very hard for dissent within the executive to be heard. Hence – at least until 2010 – a hung Parliament has always meant minority government, not coalition.

So when do coalitions form? In wartime, when national unity demands it, but at other times too. Here is the third point:

(3) The Conservatives have repeatedly formed peacetime coalitions when they didn’t have to do so.

They have done this five times. In 1895 and 1900 they did so with the Liberal Unionists. In 1918, they had a clear majority on their own, but maintained the wartime coalition for a further four years. In 1931, they were even more clearly the dominant party. They needed 308 MPs and had 473. But they governed as a coalition then, and again in 1935, until May 1940. Indeed, on these two occasions, they served under a Prime Minister from a different party: the Liberal Lloyd George in 1918 and the Labour leader Ramsay MacDonald in 1931.

Why do they do this? The Conservatives, I think, exist in a state of doubt, or anxiety, about being popular. Before the 1930s that took the form of a fear of democracy. More recently, since the 1970s, it has taken the form of a fear that people will not vote for what is good for them: that is, free market economics and competition. Coalition on equal terms is no solution to this anxiety because it obliges compromise. Hence, like other parties, the Conservatives prefer minority government rather than coalition in hung parliaments. But coalition on unequal terms is highly advantageous. The capture of other parties, and especially fragments of other split parties, has enabled the Conservatives to obtain votes they could not otherwise obtain, to recruit leaders they couldn’t otherwise recruit, and to prevent natural majorities forming against them. In the 1890s they acquired the political talents and radical supporters of Joseph Chamberlain, and kept the Liberals split over Ireland and tariffs. In 1918, they acquired Lloyd George, the ‘man who won the war’, with his ability to reach the new voters of the expanded electorate. In 1931, they managed to split both parties at once, acquiring the Labour PM Ramsay MacDonald, and a fragment of the Liberal leadership, which enabled them to claim to be a national, and not party or class, government.

Why do the party fragments join the Conservatives? If the government is popular, they gain by involvement in its success. They also gain by being strengthened against the other fragment which remains outside the government. Coalition is the only immediate route to power. They are also well rewarded. Their own projects, provided they do not offend the Conservatives too much, can get adopted as government measures. They get a generous share of Cabinet posts. The Liberal Unionist leaders were offered the Foreign Office and any other post they wanted. They got 5 Cabinet places out of 19. Lloyd George’s Liberals in 1919 got 8 places out of 20. National Labour and Liberal Nationals were similarly well rewarded.

There is a price: the fragments either have to get out of the coalition or lose their identity. The Liberals got out in 1932 when free trade was abandoned. If the fragments stay, they are eventually absorbed into the Conservative Party. But it is startling how long this can take. The Liberal Unionists, born in 1886, did not fuse with the Conservatives until 1912. The National Liberals, created in 1931, managed to preserve an independent existence until 1966.

DECISIONS IN 2010

These three findings are helpful in understanding what happened in 2015. In 2010, there were three choices for the Liberal Democrats, all of them quite bad, but not all as bad as each other. They were as follows:

1. Supporting minority government

The first choice was minority government: not to enter the government, but to give and withhold support according to the political demands of the moment. This was what the Liberals did in 1924. The Liberal Democrats could have offered a ‘confidence-and-supply’ agreement to maintain Labour in office in return for a say over other legislation. Or they could have done what they did in 1929: stayed outside the government, but helped informally to develop government proposals. In 1929, this was quite successful. They gained significant concessions on coal, India, and unemployment. They came very close to a deal over electoral reform (AV not PR) which would probably have been secured had the government not fallen in 1931. Or they could have gone even further, as they did in the 1970s, making government policy on joint committees, while still not taking up government posts. In the period of the Lib-Lab Pact (Mar 1977-Jul 1978) they vetoed things they disliked, got some small measures of their own, and secured routine co-operation through joint committees, even forming their own Shadow Cabinet. This was impressive: there were only 13 of them, every one a Shadow minister.

These sorts of arrangements are not certain to work. In a hung parliament, the large party does not need to give much. It can always form a minority government and call fresh elections when it is ready. Labour offered nothing in 1924, little in 1929, and little more in February 1974. To extract the maximum, the small party has to be pivotal (it has to be credible that it could put either side in). The weakness of the Irish parties after 1886 was that they were never pivotal, one reason why Home Rule was so hard to achieve. The small party also risks offending its voters by whom it puts in and whom it keeps out. The Liberals in 1924 first alienated their middle-class support by putting Labour into power, and then their working-class support by putting it out. So the small party has to think carefully about when to offer and withhold support, and communicate a public, principled or vote-maximizing rationale for each decision. But being outside the government helps it to explain.

The Liberals in 2010 had a strong position. They were pivotal: they had the credibility to go in either direction. Their leader Nick Clegg had been the breakthrough figure of the 2010 campaign where the other party leaders had been falling over themselves to agree with him. Neither of the large parties was keen to face fresh elections until an economic recovery was underway.

2. Coalition with conditions

Another choice for the Liberal Democrats was to follow the model of the Liberal Unionists and the Liberal Nationals. They could have entered a coalition, but insisted on a higher price for doing so. What might these conditions have been? The Liberal Unionists, National Liberals and National Labour maintained their independent organisation in Parliament, just as the Liberal Democrats did. But their support for the Conservative-dominated coalition had three further features that the 2010 arrangement did not have.

(i) They were renegotiable.

This has been explicitly or implicitly a feature of every previous coalition arrangement, even one of the wartime ones. In the First World War, the party truce agreed in August 1914 was renewed at roughly three monthly intervals until December 1916. All the peacetime coalitions had scope for renegotiation, even if it was not used. The Lib-Lab Pact was renewed conditionally on Labour performance, and it was not even a coalition.

(ii) There was provision for the small party to make its dissent plain and audible to the public.

When the Liberals disagreed with the Conservatives over protection in 1932, cabinet collective responsibility was suspended. This was possible, of course, because the Conservatives had a majority for protection. This, incidentally, raises the intriguing possibility – quite contrary to coalition theory – that a small party can get as much, or even more, when it is not essential to the coalition of which it is part, as when it is pivotal.

(iii) There was an electoral pact.

The precise terms of such agreements were flexible: they could be tight – an agreement not to run against each other – or loose – allowing local contests so that party activists opposed to coalition do not leave the party or run independents. The important thing is that they be negotiated between the parties as equals so that neither is damaged.

In 2010, none of these was done.

(i) The coalition was insufficiently renegotiable.

There was a mid-term review of progress, but the coalition was, from the start, a five year lock-in.

(ii) There was too little provision for the small party to make its dissent plain and audible to the public.

The coalition agreement allowed for ‘agreements to differ’ on certain named questions. But the doctrine of collective responsibility was formally and explicitly maintained.

(iii) There was no electoral pact.

Indeed, worse than this, there was no electoral co-operation whatsoever. The Liberal Democrats were lucky that they only had to defend one seat at a by-election – Chris Huhne’s seat in Eastleigh in 2013. They were even luckier to hold it. The Conservatives ran a candidate, the Liberal Democrat vote halved, and, had there not been a UKIP intervention, the Conservatives would have taken the seat, as indeed they did in 2015 with a majority of 9000. But the Conservatives were even luckier. They only had to defend three of their seats in by-elections and none of them to Liberal Democrat challengers. And in the 2015 election, far from helping the Liberal Democrats, the Conservatives targeted them even more than Labour.

3. Coalition without serious conditions

There is a third choice, which is the one the Liberal Democrats made: to join a coalition without serious guarantees. The first and second choices are not guaranteed to succeed, but this third choice is almost certain to fail.

Why this choice was made is the crucial question future political historians will ask. There are various circumstantial explanations of why it was done. But despite these, it remains genuinely puzzling. Coalition theory tells us that the power of a party in a hung Parliament is not much a function of its size, but of its strategic position. In 2010, the Liberal Democrats were the kingmakers. Their bargaining position was as strong as it is ever likely to be in a hung Parliament. And yet they made almost nothing out of this strategic position.

Why the electorate turned so sharply against the Liberal Democrats is therefore one of the key questions concerning this election. One view is that they got too little for their participation. But that is not persuasive. In the initial coalition agreement, they got an impressive share of the policy commitments. The same was arguably true in posts, with the significant reservation that they occupied no major office of state. Furthermore, there are many issues on which the Liberals could credibly argue that they had moderated coalition policy. They include blocking police powers to monitor electronic communications, the amendment of the Human Rights Act, climate change, green energy subsidies, the BBC, worker protection, and health service reform. It is also certain that their manifesto policies were much more popular than their vote suggests.

Furthermore, the economy was improving. Many embarrassed pollsters, in attempting not to be surprised by the outcome, have suggested that it’s predictable that the Conservatives, as the governing party, presiding over a rising curve of economic optimism, won. But this merely compounds the puzzle. There were two governing parties. Why didn’t the Liberal Democrats gain from the rising curve? It seems clear that they didn’t. Their poll rating slumped very quickly as soon as soon as they joined the coalition. But unlike that of the Conservatives, it never recovered. According to the preliminary BES data from March 2015, the Liberal Democrats were not much blamed for unpopular coalition policies, but nor were they much credited for coalition successes. 44% of voters agreed that the economy was getting better; but of these only 19% attributed this to the Liberal Democrats, compared to 73% who credited the Conservatives.

This suggests to me a possible answer: that the Liberal Democrats’ mistake in coalition-making was to focus more on policy content and on posts than on structures, rules and procedures. Policy and posts is what the theorists tell us coalition formation rests on. Perhaps it does in systems used to coalition. But structures, rules and procedures also matter: the length of the agreement, the break-clauses, and the rules governing the expression of dissent.

The failure to secure these certainly worked against the Liberal Democrats. Their achievements were very hard to hear. This mattered especially because much of the Liberals’ historic strength has been achieved as a protest vote against the two main parties. If it cannot be heard protesting it is bound to do badly.

Three specific errors stand out:

(1) They should have insisted on ‘voice’ within the coalition.

The rules of the British system – collective responsibility in the executive, party discipline in the legislature – do not make this easy. But it is startling that the Liberal Democrats failed to insist that these rules were modified. In the executive, the priority was stability, which was purchased at the expense of flexibility. The Liberal Democrats were silenced just when they had a lot to explain. This was especially damaging because another of the Liberal Democrats’ historical assets has always been that it is trustworthy and a party of principle, less tarnished than the other parties with a reputation for unreliability. The about-turn on student tuition fees damaged them not only because it was unpopular, but because it was a broken pledge.

Lower down the party, the Liberal Democrats insisted on high levels of party discipline. It is striking that, compared to earlier coalitions, there were so few revolts from below, in Parliament or the parties. The discipline of the Liberal Democrats made them reliable coalition partners, actually more so than the Conservatives. But its price was the destruction of the incumbent advantage they generally enjoy through being effective constituency representatives. Lord Ashcroft’s polls predicted that Liberal Democrat incumbents might survive, but this seems not to have been the case. By my calculation, the Liberal Democrat vote fell by 15 points in those seats that one of their current MPs was trying to defend, barely different to the decline the party suffered across the country as a whole.

(2) They should also have insisted on a less binding agreement.

The coalition agreement bound the Liberal Democrats into a five-year programme, with little flexibility to renegotiate, or exit early, and with no guarantee of electoral co-operation either during the coalition or at the end. The problem was not new issues because these were not covered by the coalition agreement. The Liberal Democrats were able to take their own line in response to the Leveson enquiry on media regulation, and to Syria. But in Scotland, their binding commitment to austerity, probably more than their opposition to independence, lost them ten seats to the SNP. Outside the coalition, securing exemptions for Scottish constituencies as the price of their support, they would unquestionably have done better.

(3) They should have got out earlier.

This was what the Liberals did in the 1970s. They abandoned the pact in July 1978 to give themselves time to prepare for the coming election. This time, far from getting out, the Liberal Democrats sought praise for having kept the coalition going for the full five years.

In the light of this longer history, the Conservative position is also an interesting one. In 2010, they abandoned their historic preference for minority government, but not their historic preference for acquiring weak coalition partners. Now, as in 1922, they are back on their own. They no longer have to make the compromises of coalition. But it may be that, paradoxically, they will regret what they have done. They now have nowhere to go when they seek to widen their support in the way that Conservative governments of the past have needed to do.

It is worth a quick comment on another historical precedent: the Irish Home Rule Party and the analogy with the SNP. In the late nineteenth century, the Irish Nationalists held the balance of power at Westminster, but much good it did them. Once the Conservatives had rejected Home Rule, and the Liberals had embraced it, there was only one parliamentary route to Home Rule for the Irish, which was to keep supporting the Liberals in the hope that the Liberals would give it to them. In other words, the Irish were never truly pivotal. They could put the Liberals in, and they could keep the Conservatives out. But they could not put the Conservatives in, and therefore they could not put the Liberals out. But while the Liberals had every reason to support Home Rule, they also had every reason not to grant it, if in doing so they deprived themselves of this useful bloc of votes. They therefore kept promising Home Rule but relying on the House of Lords to block it.

The SNP have ruled out being pivotal by promising not to support the Conservatives. But the Conservatives have been careful not to rule out working with the SNP. In 1885, Salisbury, the Conservative leader, could have contemplated a deal with the Irish: a reduction in the number of Irish MPs and fiscal autonomy for Ireland. This would have suited them both: Salisbury in reducing Liberal representation, and the Irish in achieving Home Rule. Salisbury ruled this out because he thought it would lead to the disintegration of the Union, which he believed in. But the Union is much less meaningful to today’s Conservatives, and this opens up the possibility of a similar deal over Scotland. If so, then there are again two games in town, and therefore much scope, in a future hung Parliament, for the SNP to exploit them as a pivotal party.