Introduction

With the 2010 British political campaign fresh in the history books, Britons will again face another national election in 2011. The 2011 election, however, will not be another three-way battle royale to determine who leads the country for the next few years. Rather, it will be a referendum election to determine how parliament members will be selected in the future. The United Kingdom currently uses a system known as “first past the post” (plurality) for its 650 single member districts to elect their members. In the 2011 election[1], the referendum will pit alternative vote (AV) in single member districts against the current “first past the vote” whereby voters will get to choose between the two. In AV, voters rank the candidates in their constituency from first to last. If a candidate receives a majority of first place votes, (s)he wins. If no candidate receives 50% of the first place votes, the last place candidate is eliminated, and ballots are counted again, but “each ballot cast for the defeated candidate counts for the next choice candidate listed on the ballot”. This process of counting and reallocating continues until one candidate receives a majority of votes. The group NO2AV opposes AV partly because they fear it will lead to less productive “hung parliaments”, in which no party holds a majority of seats. For the purposes of this paper, I will employ the term “divided government”[2] to describe hung parliaments. Majority or minority coalitions can rule these governments (see bottom of page 5 for further discussion). If AV had been used in the 2010 election, the largest party (Conservatives) would have actually only won 281 seats (not 307), while Labour would have won 262 (not 258) and the Liberal Democrats 79 (not 57). A BBC study of British election returns from 1983-2005 actually illustrates that AV would have resulted in the majority holding more seats than they would have under “first past the post”. Regardless of whether AV or plurality in single member districts leads to more hung parliaments[3] (and I do not know the answer), NO2AV is assuming that hung parliaments are bad.

        Interestingly, no one has set out to determine if 2011 referendum voters should be concerned over the prospect over more divided government. That is, no one has determined whether prime ministers leading a hung parliament are any more or less competent (i.e. do a better job at leading the United Kingdom into the future). Using past election returns from Britain and another plurality single member district country, Canada, I will demonstrate that prime ministers who ranked highly in historical rankings do not tend to have gained greater majorities in the parliament, or even have a majority government. Therefore, I believe that while a switch to AV may or may not give fairer representation, it will not actually lead to more, or less effective government, in the United Kingdom and other countries that use this voting method.

Arguments for and against Divided Government

        When a divided government seemed probable during the 2010 campaign, the public and academics put forth different beliefs on what it would mean for the quality of government. Current Prime Minister David Cameron warned that all of the promises that the Conservative Party made during the 2010 campaign "could die" in a hung parliament. Why? Oxford University Professor Vernon Bogdanor believed that parliaments lacking a mandate in seats were not very stable and often times lacked any policy initiative2. In the case of a hung parliament, constitutional crises can also arise whereby the normally apolitical Queen (as per British rule) is forced to choose a party leader to form a government who might not actually be the best one to do so. Therefore, the Queen's foray into politics would only further destabilize the country.

The Guardian columnist Simon Jenkins is equally as pessimistic about hung parliaments writing that they not only are insecure, but they also fluctuate in makeup (with parties entering and leaving possible coalitions)[4]. Because of this coalition instability, it would be nearly impossible for the public to actually know whom to vote out of office to end the hung parliament. Jenkins also remarks that hung parliaments give a disproportionate amount of power to smaller parties who are part of the coalition, and thus try to pass legislation that is against the public will. Such happened during the 1974 divided government, when, according to British historian Dominic Sandbrook, the Labour Minority government was  "hobbled by the need to appease its own militant backbenchers, and so obsessed by the complicated parliamentary arithmetic". The result, Sandbrook believes, was a government unable to deal with the financial crisis of its time (???). Opposite Jenkins stands Philip Cowley who believes that when parliament is closely divided, smaller parties do not actually dilute power of the government, but actually strengthen it in a bad way. Instead of parliament being able to form a coherent and vocal opposition to possible overreaching by the prime minister and his cabinet, the entire parliament is in such disarray that prime ministers can practically do whatever they want. Anthony McGann, John Ensch, and Theresa Moran’s study of German and Israeli politics using Monte Carlo simulations seems to support this viewpoint. They found that the party with the greatest number of seats in a coalition holds a disproportionately high amount of power as the amount of parties in the coalition becomes greater[5].

        Of course, many people argue that divided government can actually be productive. The Independent's Steve Richards contends that parliaments with large minorities can actually put a check on the power of the cabinet. He considers the large Tony Blair led Labour majority would have been "subjected to more internal scrutiny [in it's Iraq War position]. Its indiscriminate approach to civil liberties, although nowhere near as sinister or threatening as some believe, would have been challenged even more effectively than it has been". Torsten Persson, Gerard Roland, and Guido Tabellini’s study of American politics finds that when there is a conflict of interest between the different branches of the government[6], internal information that a government may wish to keep hidden, but is good for the public to know, is revealed to the general population. Another argument for divided government is that minority party voters, whose voices are infrequently heard, could actually see legislation address their concerns. After the 2010, Australian elections which left a hung parliament, farmers in the rural outback were able to become part of a coalition in which farming issues became key.

On the issue of the 1974 minority government and the financial emergency, Chris Rogers disagrees with the view put forth by Sandbrook. Rogers writes "the major lesson that should be drawn from the experience of February 1974 is that a hung Parliament does not necessary equate to uncertainty and indecision that will inevitably precipitate economic crisis, but could equally allow for measured compromise and response that reflects more appropriately both the political and economic constraints faced by state managers". He goes on to note that financial crises affect different segments of the public in different ways, and it is in these circumstances when a coalition among different parties is needed most. Labour Austin Mitchell adds that the Labour government won re-election (without a majority) in a second 1974 election and was able to rescue the United Kingdom from almost certain economic downfall. Finally, Liberal Democrat Simon Hughes agrees, pointing out that strong single party governments do not necessarily lead to more secure governments, and that majority governments that are smaller or forced to compromise, balance the opinions of the country.

        It should be pointed out that none of the modern arguments in Britain for or against divided government notes the difference between a majority coalition and minority government. A majority coalition government is one that forms when no one party holds a majority of seats, and two or more parties together combine to hold a majority. A minority government, which can be made up of one or more parties, occurs under a similar circumstance, except the party(ies) leading the government does (do) not hold a majority of seats. In both the AV and plurality electoral system, voters can only determine the makeup of parliament, not the votes members cast, such as whether in the case of a hung parliament to form a majority coalition or minority government. This paper is solely interested in the difference in the amount of divided governments (as it relates to prime minister effectiveness) that AV vs. plurality elections could create; therefore, the distinction between the two types of hung parliaments should not and will not be made in this paper.

How do we answer the question?

What method would be best to answer objectively the question of this paper, should 2011 referendum voters be concerned over whether AV or plurality leads to more hung parliaments (as it relates to prime minister effectiveness)? As the last section demonstrates, proponents for and against strong one-party majority government in the United Kingdom are convinced of their position. Yet, none of these academics, columnists, or politicians presents anything more than one or two case studies to support it. Trying to combine the many factors that lead to a successful government into one variable is daunting. For example, a government may deal with the economy well, but lack any progress in social reform. In the United States, the answer to this question has been to ask a panel of experts many questions in order to rank their leaders on individual issues that sum up to a final score. In recent years, British historians have noted the lack of scholarship in this area. University of Leeds Kevin Theakston and Mark Gill remark "it is disappointing that in the British context we have only 'coffee table' type books by prime ministers on this subject, like Harold Wilson's A prime minister on prime ministers and Harold Macmillan's The Past Masters". Fortunately, the scholarship in this area has expanded over the last decade. There is now one survey, the 2010 Sunday Times survey, which ranks every single prime minister since the Great Reform Act of 1832[7].

The key for our study will be to compare these rankings to the election results. The issue for this paper is that only five British elections have ever lead to a hung parliament. Considering that anti-AV forces want the public to believe that divided government is a big problem with AV, I believe it wise to expand the dataset to include more of them. Doing so, also allows us to understand whether British elections are flukes in and of themselves, or whether the results from my exercise can be expanded to other countries struggling with the AV vs. “first past the post” debate. I have decided to include results from the Canadian parliamentary elections since 1867. Like the United Kingdom, Canada's House of Common elections are plurality and in single member districts. Also, similar to Britain, Canada had a two party system until the 20th century. Unlike the United Kingdom, however, eleven (11) Canadian elections have ended in hung parliaments. Finally, Canada provides the only other single member district “first past the post” parliamentary system where we have rankings available for prime ministers. University Professors Norman Hillmer and J. L. Granatstein's 1997 rankings provide the most up-to-date scoring for Canadian prime ministers. I must note my dismay that neither the British nor Canadian study had more historians or journalists partaking in them instead of the five (5) and two (2) respectively, but unlike in America, there simply are not many studies on ranking country leaders.

Methods

        A prior paper I wrote on the subject of American U.S. presidential elections and historical rankings provides the outline for how I will compare election results and prime minister rankings. First, I generate four figures to test two different theories. Figures 1 (Britain) and 2 (Canada) are scatterplots in which historians’ rankings are compared to the percentage of seats held by governing party. If we see that the percentage of seats held by a party has a high negative correlation with the rankings, it can be interpreted to mean that prime ministers whose parties secured a high percentage of seats in parliament had more successful administrations. In this case, the eventual system of government agreed upon by the British should try to insure that the party in control of the government has a large majority in parliament. Figures 3 (Britain) and 4 (Canada) are boxplots with one box representing the prime ministers that had a one party majority in parliament, and the other where prime ministers controlled either a majority coalition or minority government. If in a given country’s boxplot the one-party majority plot has a small inter-quartile range and a significantly lower median (that is higher ranked prime ministers) than the hung plot, then we can conclude that hung parliaments lead to less effective government. If, on the other hand, the hung plot is higher, then the opposite is probably true. If no recognizable pattern appears in any figure, then governments in Britain and other countries facing a similar dilemma of turning away from “first past the post” should not be convinced (worried) that either system produces more (less) productive government. 

        This election analysis looks at all prime ministers in Britain and Canada who faced an election since 1832 and 1867 respectively. Prime ministers who were NOT elected to their post immediately following a general election are not included. These non-elected prime ministers include Vincent Lambs, Arthur Wellesley, and Neville Chamberlain in the U.K, and John Turner and Kim Campbell in Canada.

The reason these prime ministers are discounted is that I want to be consistent. British studies, as summarized by Brian Gaines, show that voters use their vote for parliament not to register support for their local member, but to support a particular candidate for prime minister. That is, voters see local candidates as “mere ciphers” to back the voters’ choice for leader of parliament. Those prime ministers who do not go through election process could be weaker than the size of their coalition suggests (due to a lack of a mandate from the voters).  Citing this rationale (among others), David Cameron has argued for a reform to force an early vote on these unelected prime ministers. I also want to keep the votes-to-seats-to-prime minister connection intact, as the “votes” portion of this connection is what AV would reform[8]. If voters want to take ownership of these “substitute” prime ministers, they can do so by voting their coalitions into office. Therefore, those prime ministers who were "substitutes" (e.g. Winston Churchill), but were later elected are included here. The rankings available to us do not discriminate between multiple terms (if an administration served through more than one election). For administrations that were re-elected, I count each re-election term as its own observation.  Averaging the results from multiple elections for one prime minister does impact my results.

Results/Analysis

        A look at the scatterplots and boxplots from both Britain and Canada demonstrates that there does not seem to be a statistically significant relationship between seat share and prime minister rankings.  Some of these figures indicate that a prime minister who has more seats in the parliament is more likely to lead a successful administration, while others show the opposite. None of the figures reveals any information that is statistically significant.

        Figure 1 (British seat share against prime minister ranking) at first glance may seem to show a relationship, but it is not statistically significant. Most of the highest ranked prime ministers are tucked in the lower right corner indicating that the best prime ministers (ranked closest to 1) worked with large one-party majorities, but there are exceptions to this rule. The highest ranked prime minister (Churchill) was not the head of his party when it gathered a majority that allowed him to serve as prime minister, but when he finally was elected in his own right the Conservatives only controlled 51.36% of the seats. Likewise, H. H. Asquith ranks a high 11th (out of a possible 48) in the rankings, despite his party, when he led it, never actually winning more than 40.6% of the seats in parliament. Opposite Asquith and Churchill is Anthony Eden who managed to have 54.8% of the seats in parliament at his disposal, but was 47th in the rankings. Thus, it is not surprising that while more British prime ministers ranked highly have a higher percentage of the seats in Parliament under their party’s control, the result is not statistically significant.

In Canada, we also see that higher ranked prime ministers’ parties tend to control more seats, but it is statistically insignificant. Some prime ministers, such as John Diefenbaker, were able to win huge majorities (78.5% of Parliament's seats) and yet only scored a 13 out of 20 in the expert scorings. Other lower ranked prime ministers, such as Arthur Meighen scored a 14 out of 20, but controlled a much smaller percentage of parliament (47.4%). The same non-trend holds true for highly ranked prime ministers. Lester Pearson served two terms and in both terms he led a minority government, but he ranks 6th in the historian rankings, while Wilfrid Laurier reaches 3rd place in the historian rankings and never had a majority smaller than 55.4% of parliament. Figures 3 and 4 confirm the findings of Figures 1 and 2. In the British set, prime ministers forced to employ either a minority or majority coalition government following an election during their terms actually scored slightly higher in the historical rankings than those that did not. In the Canadian set, the opposite is revealed. Prime ministers who only led one party majority governments were rated slightly higher. The results were not statistically significant in either case. In both figures, the inter-quartile of each plot engulfs the median in the corresponding plot in the same figures.

        An important warning to keep in mind about these results is the limitation imposed upon them due to prime ministers whose party won a large percentage of seats one term and a smaller percentage of seats the next. High ranking (3rd) William Ewart Gladstone managed to form a cabinet after elections where his party won as little as 40.9% of the seats to as large as 58.8% of the seats, while middle ranking John Russell formed governments after elections with as small as 44.5% and as large as 56.1% parliament backing him. The same sort of split occurred in Canada where the highest ranked prime minister, Mackenzie King, and 13th ranked John Diefenbaker took control of Parliament after elections where their party held, and did not hold, a majority. As we have mentioned, the historians in the surveys used in this paper and all surveys perused, ranked the tenure of each prime minister rather than each individual term a prime minister lead. What type of impact this phenomenon had on our results is unknown. Research that can break down the effectiveness of each prime minister by individual term is needed. As it is, there is no clear evidence that prime ministers whose parties win majority of seats after an election lead to a more effective government in either Britain or Canada.

Conclusion

        The results of this paper reveal that prime ministers in Britain and Canada who head one party majority governments were no more successful in historians’ eyes at leading their countries than those prime ministers who led a majority coalition or minority government. Therefore, I believe that when voters in Britain think about replacing the “first past the post” method in single member districts with the alternative vote system, they should not concern themselves with a system that may lead to prime ministers facing smaller one-party majorities or even majority-coalition/minority government. These findings verify neither those that argued that hung parliaments were bad (Cameron, Bogdanor, Jenkins, Sandbrook, and Cowley), or those that argued they were good (Richards, Rogers, Mitchell, or Hughes). The use of both British and Canadian prime minister rankings does illustrate that these results are potentially transferable to other parliamentary countries debating whether hung parliaments are productive or unproductive. This paper does not answer the question as to why one-party majority governments perform as well as coalition majority or minority governments. Further analysis in which historians and political journalist rankings’ are broken down publicly by category (such as economic management, social programs, wartime leadership, etc.) is needed to do that. This paper also does not address other rationales as to why countries would want to change from plurality single member districts to alternative vote. For instance, a voter may prefer alternative vote because they want the freedom to vote for a third party without the fear of wasting his or her vote. The “fairness” of electoral system may be as important as the results it obtains. It is hard to imagine Britons arguing in favor of parliament turning its power over to the Queen of England even if she promised better government for all.  The right balance of effectiveness with other issues such as representation of all of a country’s citizens in an electoral system is something that only each country can decide for itself. The findings of this paper indicate that a parliamentary country can afford to focus on issues other than determining which electoral system will lead to productive governments because the size and type of a majority (or minority) does not seem to matter.


[1] In Britain, however, the two party system has collapsed in recent years with each of three parties (Conservative, Labour, and Liberal Democrat) regularly gaining at least 20% of the vote nationwide. This trend has trickled down the constituency level. In 2005 for instance, 66% of winning candidates, received less than 50% of the vote. Due to this issue, the Liberal Democrats often found their share of seats won in parliament was only a fraction of the vote their vote nationwide. To deal with the votes-to-seats incongruity, the Liberal Democrats have argued for a system known as single-transferable-vote (STV). STV elections occur in multimember districts where the number of seats a party wins is more closely associated with the number of votes they win than under the current plurality rule in single member districts. After the 2010 election, the Liberal Democrats, finally a pivotal part of the majority coalition in parliament, pushed for STV. The largest part of the coalition, the Conservatives, balked at the proposal. Instead, they agreed upon a referendum placing AV against plurality.

[2] In political science literature, divided government often refers to presidential systems where one party controls one branch of government and a different party controls a different branch.  

[3] It could be that under certain conditions AV will lead to more hung parliaments while under others plurality will. Research on this subject would be interesting.

[4] Hung parliaments are not necessarily unstable. It is merely the argument Bogdanor & Jenkins is making.

[5] This study does not compare majority governments to divided governments as none have ever been elected in either Germany or Israel. This study also demonstrates that small parties in ruling coalitions in hung parliaments are most likely to hold a disproportionate amount of power when the number of parties in the coalition is small. In Britain, coalition governments, historically, have had a small amount of parties (never greater than 4 parties).

[6] The United States has a presidential system government (with a separate legislative branch) in comparison to the British parliamentary system (i.e. one body that acts as executive and legislative). Thus, the relevance of this literature to the British debate is questionable; however, the overall finding that governments with more competing interests, as one would find in a hung parliament is important, to note.

[7] Allowed all freehold property owning men to vote in elections.

[8] Whether or not an AV majority built on a large number of members of parliament elected with “second”, “third”, “fourth”, etc. choice votes holds the same mandate as one elected with first choice votes is an important question, but beyond the scope of this paper. A review of Australia’s House of Representatives, a major parliamentary body that has employed AV since 1918, elections through 1996 finds that 3 majorities (1961, 1969, and 1990) would have actually lost if only first place votes counted. A 2004 Sky Magazine survey that asked Australian historians to rank the 11 non-caretaker prime ministers that served from 1939-2004 determined that the 3 prime ministers whose government would have lost if only first place votes counted ranked 2, 3, and 9. This difference in prime minister ranking between losers and winners under first place votes is not significant. Considering this fact and for the sake of simplicity, we will assume for this paper that the type (first, second, etc.) of votes that lands a government in power is inconsequential as to whether they hold a mandate.