Why governments lose: UK elections since 1922

By Stephen Fisher, 12th April 2023.

Governments in the UK tend to win elections, but lose if there is an economic crisis. That pattern explains the outcomes of 19 out of the 27 elections since 1922. A further three elections can be accounted for by governments averting electoral disaster by changing prime minister after a crisis. The combination of economic crises and political changes at the top can explain 22 of the 27 elections, including all the elections since 1987.

That is the main conclusion from my recent working paper. A simplified version of the key table from the paper is shown below. The first row shows that governments won 10 of the 13 elections that were not preceded by an economic crisis. Governments tend to win these elections because governments have a lot of power in Britain. The three elections that were lost without a crisis were all extraordinarily early elections that should not have happened. The 1923 and 1951 elections were gambles, needlessly called by a majority government. To make things worse, on both occasions the government was offering an unpalatable economic policy (tariffs and rationing respectively). The third, 1924, only happened because 1923 produced a seriously hung parliament. None of the three are serious challenges to the idea that governments ordinarily win elections.

 Post-Crisis Political Change of PMGovernment wonGovernment lostTotal
No economic crisisNo10 (1935, 1955, 1966, Oct 1974, 1987, 2001, 2005, 2015, 2017, 2019)3 (1923, 1924, 1951) 13
Economic crisis since the last electionNo2 (1950, 1983)9 (1929, 1931, 1945, 1964, 1970, Feb 1974, 1979, 1997, 2010)11
 Yes3 (1922, 1959, 1992)03
Total 151227 
Table: Economic crises, post-crisis political changes of PM and government electoral fortunes

The bottom two rows of the table show what happened after economic crises. For pragmatic reasons, economic crises are identified by recessions and devaluations from fixed exchange rate systems (such as Wilson’s 1967 devaluation and the 1992 ERM crisis). Unemployment, inflation, strikes and other economic problems still matter, but recessions and devaluations are used because they are indicative of broader crises. 

Out of the 14 post crisis elections, the government lost 9. Using polls, and by-elections before there were polls, the paper sets out how, in each of the 9 cases, the economic crisis contributed to the eventual electoral defeat of the government.   

Even in the remaining five cases, there are signs that economic crises damage the government, even though they managed to cling on despite the damage. Labour only managed to survive in 1950 because they started with such a big majority. The Conservatives survived in 1983 because of the popularity boost they got from the Falklands war.

Attlee and Thatcher are the only two prime ministers since 1918 to have been re-elected after an economic crisis. More often the prime minister does not survive to fight the next election. The final row of the table shows that on three occasions governments have managed to avoid losing by changing prime minister for political reasons after the crisis: from Lloyd George to Law in 1922; from Eden to Macmillan after Suez; and from Thatcher to Major after the start of the 1990 recession. 

Political changes at the top have generally worked well for the Conservatives: they won the subsequent election five of the six times they tried the trick. That fact is key to explaining their greater electoral success overall. Since 1922, the Tories have won 16 elections to just 11 for Labour. That success is not because Conservative governments are better at avoiding economic crises. Actually, Conservative governments have sought re-election after an economic crisis 56% of the time, more often than Labour (45%). That makes greater Conservative success overall even more remarkable. The paper discusses how that greater success rate is largely explained by effective post-crisis changes at the top.

Interpretations and implications of the findings

The statistical associations and the case-by-case evidence supporting the theory are very strong, but that does not mean that the results should be interpreted as economic determinism. The findings are the products of a century of British politics with voters and politicians making choices. Choices to call an early election, especially in 1923 and 1951, were highly influential on the course of British politics. Had the 1950 Labour government kept going until 1954/5, the theory and findings in the paper suggest Labour would have won again and stayed in power until the 1960s. Furthermore, the results do not imply governments can act with impunity so long as they avoid economic crises. Rather the results may be conditional on a tendency for governments to try to please voters.

The findings support the folk law that governments lose elections rather than oppositions winning them. Indeed, the theory accords no agency to opposition parties. But that does not mean that their actions and choices do not affect election outcomes at all. Most obviously opposition strategy might affect the scale of a win or loss. For example, while the Conservatives lost in 1997 because of the ERM crisis, the scale of the Labour landslide was largely due to Blair’s centrist strategy.

Despite the decades of dealignment, increasing volatility, the rise of populism, Brexit and all the other dramatic developments in British politics there has been just two transitions of power since 1979: to New Labour in 1997 and to Cameron’s Conservatives in 2010.  Both were the consequences of economic crises. Despite all the political turmoil, it seems that the lower frequency of economic crises in more recent decades (perhaps as part of the Great Moderation) has had a powerful stabilising effect on party control of governments.

The theory in the paper is designed to help explain the past rather than forecast the future, but ordinarily it could be used to generate a crude forecast for the next election. Since 2019 there has been a recession and two political changes of prime minister. The 1922, 1959 and 1992 experience suggests the Conservatives should win the next election unless there is a recession under Sunak’s watch. However, this is an extraordinary electoral cycle. GDP is thought to have dropped by 11% in 2020, the largest since the Great Frost of 1709 led to an estimated 13% drop. When Sunak became prime minister GDP was still 0.6% lower than what it was at the time of the 2019 general election. Given such an extraordinary backdrop and poor starting point for his premiership, a new recession while Sunak is prime minister might be too high a benchmark as the sole indicator of a post-Sunak economic crisis. With inflation near 10%, real incomes falling, strikes, and public services in disarray, there is arguably a severe current economic crisis even without a technical recession. On that basis the government might, as polls also indicate, be expected to lose the next election.

There are more details about the points above, and other findings, in the working paper. Comments welcome.


I’m very grateful to Nelson Fernandes Serrao for research assistance, to participants the Elections, Public Opinion and Parties (EPOP) conference in September 2022 and the following people for comments on previous versions and helpful conversations while working on this paper: Lucy Barnes, Richard Coggins, Andrea Ferrero, Anil Gomes, Ken Okamura, Iain McLean, Rob McMahon, David Myatt, Dan Snow, James Tilley, and Matthew Williams. Errors are my responsibility.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s