Written evidence submitted by Professor
Patrick Dunleavy (PVSCB 14)
EXECUTIVE SUMMARY
1. There is a strong, non-partisan case
for changing how MPs are elected so as to restore to all MPs the
clear support of a majority of their constituents which they enjoyed
in earlier periods. Britain is now a multi-party system like others
in western Europe, a trend that is highly unlikely to reverse.
Sticking with first past the post elections will lead only to
fewer and fewer MPs enjoying the legitimacy of local majority
support, further damaging the already poor standing of Parliament
in the public's eyes.
2(a) There are four highly relevant variants
of the Alternative Vote, and the government will need to make
clear to citizens which variant is to be voted on in the referendum
and why. The different systems each have things they do well but
also some limitations and things they do badly.
2(b) Australian AV seems to be the government's
chosen variant but it may allow candidates ranked 3rd or 4th in
voters' first preferences to none the less end up winning seats.
In UK conditions it is also likely that it will end up not counting
millions of second preference votes, reducing the legitimacy boost
from point 1.
3. Versions of AV using numerical ranking
of preferences are incompatible with most existing British voting
systems, creating large-scale problems in holding Westminster
elections on the same day as other elections.
1. The non-partisan case for changing how
MPs are elected
1.1 From one election to the next, more
than two thirds of MPs in Great Britain now no longer have the
support of a majority of voters in their constituency. The forthcoming
referendum on introducing the Alternative Vote provides an opportunity
for everyone, regardless of their party or views on proportional
representation, to recognize the case for a minimum change of
voting system, to at least restore local majority support to all
MPs. That could be a crucial basis upon which, slowly at first,
MPs could begin to rebuild some of the legitimacy of Parliament
that has been so dramatically imperilled in recent decades, by
careless legislation, a public perception of broken election promises,
and the expenses scandal of 2009.
1.2 How have MPs so extensively lost local majority
support without themselves or most media commentators really noticing
it? The change has happened gradually, election by election since
the 1960s, as the number of parties contesting seats in the UK
has gradually increased and voters have shifted to back them.
Today we have a six or seven party system in every region of the
UK, with third party (Liberal Democrat) ministers in government
for the first time since 1945, and many parties now scoring salient
vote shares at different elections. In both 2009 and 2004 parties
to the right of the Conservatives (namely UKIP and the BNP, both
with MEPs and London Assembly members) gained just under a quarter
of all votes nationwide in European Parliament elections. The
Greens have their first MP in Westminster to add to their MEPs
and London Assembly members. This is a trend that has been operating
for many decades and that is not now going to retreat or mysteriously
go away. British voters want to vote for more parties and more
viewpoints than before.
1.3 If we go back before this trend happened
to the 1955 general election, a low point for Liberal support
and the heyday of the two-party Conservative-Labour system, the
vast majority of MPs drew on majority support in their constituency,
as Figure 1 below shows. Here each black blob is a single constituency
outcome. The bottom axis shows the Conservative vote share minus
the Labour vote share. So the further from the centre a constituency
is to the right, the greater the Tory lead. And the more a constituency
is on the left of the centre-line, the more solidly Labour it
is.
Figure 1
THE CONSTITUENCY OUTCOMES IN THE 1955 GENERAL
ELECTION, IN GREAT BRITAIN
The vertical axis here shows the combined share
of votes going to all other parties, in 1955 pretty much only
the Liberals and a few independents. Taken with the horizontal
axis, this means that all feasible outcomes must lie inside the
overall double-triangle shape outlined in the green borders here.
There cannot be any outcomes outside this overall area.
1.4 In 1955, Figure 1 shows that in the
vast majority of constituencies there were no other candidates
except the Conservatives and Labour. Hence in all those hundreds
and hundreds of seats the outcome lies on the bottom axis itself—where
seats after seat lies piled on top of each other, so many that
the chart cannot possibly show them all. In the 110 seat that
the Liberals still contended, and a few others, the "third
party" vote held up, and these are the scatter of blobs above
the bottom axis, mostly concentrated in Conservative areas.
1.5 The shaded triangles inside the overall
feasible space in Figure 1 allow us to see that the vast majority
of MPs could draw on majority support in their constituencies
in 1955. Every black blob in the blue-shaded area is a Tory MP
with majority backing, and similarly Labour MPs with majority
support fall inside the pink-shaded triangle. (There are just
a handful of seats won with majority support by the Liberals).
The constituencies where MPs lacked majority support are the black
blobs on a white background—there are only a few tens of
constituencies, out of 650 seats in all. In other words for an
MP not to have a local majority was a rare exception.
1.6 Flip forward to 2010 shown in Figure
2 below and a huge difference is apparent. Two-party contests
have completely disappeared, and support for parties other than
the Conservatives and Labour is rarely less than a fifth of total
votes, occurring in only a few handfuls of seats. As a result
the whole set of black blobs showing constituency outcomes has
shifted radically upwards. Across the whole bottom third of the
feasible area there are only a scattering of seats with total
`other party' votes below 20%. The core band of seats has moved
up the chart in Figure 2, but it still shows a marked Conservative
versus Labour patterning—yet with much higher levels
of voting for third, fourth and subsequent parties. Above all
the advent of the coalition government reflects the number of
seats where the Liberal Democrats, SNP, Plaid Cymru and other
parties now regularly win around 90 constituencies out of the
628 in Great Britain, shown mainly by the "curling over"
of seats on the sides of the distribution and extending in an
upper swathe across the middle of the diagram. Where the total
other party vote is above 33%, many seats are still won by the
Conservatives or Labour, because remember that the "other"
vote is split across several different parties. But the higher
up blobs occur on the chart, the less likely they are to be held
by one of the top two parties.
1.7 Figure 2 clearly shows how few MPs now
win local majority support in the vast majority of seats, those
appearing as black blobs on a white background. Because the Conservatives
did relatively well in 2010 they have rather more MPs with majority
backing than Labour (who did badly). But looking back at the 2005
election would show an almost reversed distribution of majority
MPs. In both elections it is completely clear that only the fringes
of either major party's seats now fall into the shaded triangles
showing majority support, whereas in 1955 almost all did so. (The
seats where `other' parties win also tend to be the most multi-party
ones, so there has been little compensating growth of MPs from
third or fourth parties with majority support).
Figure 2
THE CONSTITUENCY OUTCOMES IN THE 2010 GENERAL
ELECTION, IN GREAT BRITAIN
1.8 Finally on this point Figure 3 below
shows in parts (a), (b) and (c) a simplified picture of that development
that has brought us to where we are today. The trend for more
British voters to support parties other than the Conservatives
or Labour has not been absolutely continuous over time, but it
has been ineluctable, long-lived and in one direction for a long
time now. There is no reason whatever to suppose that the pattern
of change across the 1955 and 2010 charts above is somehow going
to reverse. Hence unless we change the voting system to acknowledge
it we will perforce have to live with a situation where fewer
and fewer MPs have majority backing amongst voters in their local
areas.
1.9 You also do not need to be a far-sighted
prophet to predict the long-run endpoint of the UK political systems'
evolution, shown in part (d) above. The UK is increasingly becoming
a standard European liberal democracy with a full range of parties,
running from the Greens on the left to well-supported anti-foreigner
parties on the right. The end point of this development will essentially
be a multi-party competition situation where virtually no MPs
have majority support.
Figure 3
THE PAST EVOLUTION AND PREDICTED FUTURE DEVELOPMENT
OF THE BRITISH PARTY SYSTEM
2. What does the "Alternative Vote"
mean?
2.1 Constitutional changes last a long time,
and election system arrangements can be designed and fine-tuned
in many different ways, each of which may have far-reaching consequences
for the political system. Hence voters need to consider in great
detail what choice they are being offered in the 2011 referendum.
2.2 The government's proposed referendum question
asks UK voters if they prefer "the Alternative Vote"
(AV) to "first past the post". Yet what AV means here
is not clear and will require careful specification because:
— In political science, the label Alternative
Vote is widely used to describe a class of voting systems,
all of which:
— elect a single office holder;
— in an "instant run-off" fashion;
and
— by counting multiple (ie one or more)
preferences;
thus effectively replicating either "exhaustive
balloting"or "dual ballot" elections but in just
one round of voting.
— The four main variants of AV relevant
for the UK are:
— "classic AV", where
voters must number all candidates to cast a valid ballot (see
Figure 1 below). This is how AV operated in Australia for many
decades until recently;
— "Australian AV" in its
current form, where voters need number only one or more candidates
(see Figure 1 below). This is how AV operates in Australia now;
— London AV (also called the "supplementary
vote"), where voters use X voting to indicate first and second
preferences only (see Figure 2 below), and only one of the top
two candidates on first preferences can win. This system has been
used very successfully since 2000 to elect the London Mayor, and
to elect Mayors in 11 other towns and cities in England; and
— London AV with numerical ordering
(LAVno), where voters number preferences as with Australian
AV (see Figure 1 below) and again only one of the top two candidates
on first preferences can win. (This system is not yet in use).
— The similarities and differences of these
systems are set out in Table 1 below. At the counting stage where
no one has a majority on first preferences, classic and Australian
AV both eliminate candidates from the bottom one by one. By contrast,
in one step London AV (in both versions) eliminates from the count
of second preference votes all but the top two candidates.
Figure 4
BALLOT PAPER FOR THE AUSTRALIAN, CLASSIC
AND "LAVno" FORMS OF ALTERNATIVE VOTE
Figure 5
BALLOT PAPER FOR THE LONDON FORM OF THE ALTERNATIVE
VOTE (ALSO CALLED THE SUPPLEMENTARY VOTE)
Table 1
HOW THE FOUR MAIN FORMS OF ALTERNATIVE VOTE
OPERATE
|
Classic AV
|
Australian AV | London AV with numerical ordering
|
London AN |
The system essentially is an instant run-off form of?
| Exhaustive balloting |
| Dual ballot |
Voters' task in
the polling
station is to?
| Number all candidates 1, 2, 3 etc up to N in order of preference
| Number candidates 1, 2, 3 in order of preference, expressing as many or as few preferences as they wish
| Vote X in the first preference column, and if they wish vote X in the second preference column
|
If no candidate has a majority of first preferences
| | | |
|
how are
candidates
eliminated?
| Candidates are eliminated in order
from the bottom, until someone left
in the race has a majority of votes
| All candidates placed 3rd or lower in the first preferences count are eliminated at one go. Any of their voters' second preference votes cast for the remaining top two candidates are added to their piles. Whichever one has more votes now wins
|
What is the winning post
when second
or subsequent preferences
are counted?
| 50% +1 of initial votes. All voters shape the result.
| 50% +1 of initial votes, or 50%+1 of votes remaining when only 2 candidates left in the race. Most voters shape the result.
| 50%+1 of votes remaining when only 2 candidates left in the race—we pass through later preferences for eliminated candidates to reach all those for top two candidates. Most voters shape the result.
| 50%+1 of votes remaining when only 2 candidates left in the race. Second preferences for eliminated candidates are discarded as ineligible and hence those voters who did not choose a top two candidate do not influence the result.
|
Which candidates can win? |
In multi-party contests, the 3rd or 4th placed candidate on first preferences count may still win the seat. The more parties there are, and the closer they are in vote shares, the more likely this outcome becomes
| Only one of the top two candidates from the first preferences count can win
|
Which second preference votes are likely to be left uncounted?
| The second preferences of supporters for 3rd and 4th placed parties are not counted when eliminating a lower-ranked candidate gets one of the top few candidates past the winning post
| None. These approaches always deliver a full count of second preferences.
|
| |
| | |
— Under the Australian or classic AV variants a candidate
initially placed 3rd or 4th in the first preference count may
none the less win the seat, so long as they were fairly close
to the front-runners to start with and can gain more preferences
from 5th, 6th or lower preference candidates who are eliminated.
For example, consider Table 2.
Table 2
A HYPOTHETICAL EXAMPLE OF HOW LOW-PLACED PARTIES ON FIRST
PREFERENCES MAY WIN SEATS, UNDER THE AUSTRALIAN VERSION OF AV
(OR CLASSIC AV)
| 1st preference votes
|
2nd count |
3rd count
|
4th count |
5th count
|
Conservative | 13,000 |
13,000 | 13,000 | Eliminated
| |
Labour | 12,500 | 15,000
| 15,000 | 15,500 | Eliminated
|
Liberal Democrat | 11,900
| 13,200 | 13,200 | 18,500
| 29,500 wins |
UK Independence Party | 11,500
| 11,500 | 15,500 | 20,500
| 24,500 |
BNP | 4,100 | 4,100
| Eliminated | |
|
Green | 3,800 | Eliminated
| | | |
Total votes | 56,800
| 56,800 | 56,700
| 54,500 | 54,000
|
Assumptions | | Green vote splits across Labour and Liberal Democrats
| Almost all BNP vote goes to UKIP | Conservative vote goes 500 to Labour, 5,000 to UKIP and rest to Lib Dems
| Labour vote goes 4,000 to UKIP and rest to Liberal Democrats
|
| |
| | | |
In every region of the UK there are now six or seven viable
parties with significant vote shares, wit especially balanced
situations in many Scottish and Welsh constituencies. Hence scenarios
like Table 2 could well occur in the UK far more frequently than
they have done in Australia (where the top two parties are more
dominant and such cases are not common).
— Under Australian and classic AV losing candidates
are eliminated one by one, and the counting of second preference
votes stops as soon as one candidate reaches 50%+1 of votes. In
current UK multi-party conditions, this will mean that in many
constituencies the second preferences of 3rd and 4th candidates
will rarely if ever be counted, affecting Liberal Democrat voters
especially. MPs will be pronounced elected with 50%+1 support,
whereas in many cases they may have far more backing from voters
whose second preferences are never inspected. The London AV system
by contrast counts and published the second preferences of all
voters.
2.3 The government appears to mean by "Alternative
Vote" only one particular variant of this class of systems,
namely Australian AV. One option open to the government and Parliament
would be to let voters express a preference not just on shifting
from FPTP to AV, but also in deciding what specific type of AV
they would like, especially as between Australian and London AV.
2.4 Alternatively, because the choice of system for Westminster
is likely to affect all other voting systems used in the UK, it
might be wisest for a small commission to deliberate on what is
the best form of AV to offer voters in the referendum; or for
the government to discuss and agree this issue in detail with
the Political and Constitutional Reform Committee and with Parliament.
3. Changing the Westminster election system will almost
certainly lead to further changes in other UK voting systems
3.1 In recent years general elections have often held
on the same day as other elections in the UK, especially the local
government elections (normally held on the first Thursday of May)
and the Elections for the European Parliament (always held in
early June on a fixed five year cycle). In addition, in Great
Britain it is conceivable that the general election day may on
occasion coincide in the future with one or more of the following
list—the London Mayor and Assembly elections; the Scottish
Parliament elections; elections for the Welsh National Assembly.
3.2 In January 2011 the government will also publish its proposals
for reforming the House of Lords, bringing in elections for all
or most of its members. The data for these elections is likely
to coincide by law with those for general elections. There are
three front-runner proportional representation systems for electing
members of a reformed upper chamber:
(a) a regional top-up additional member system, with some
large constituencies (counties?) and some top-up seats at regional
level. Like other British AMS systems in Scotland, Wales and London
this would use a ballot paper with two X votes, one for the constituency
and one for the top-up members;
(b) a regional list PR system, similar to that used for electing
MEPs, where voters cast a single X vote and candidates are elected
off party lists. (For the Lords, this might involve a capacity
for voters to change the order in which candidates are elected
off party lists, but this will still use X voting); and
(c) a single transferable vote (STV) system with regional
or sub-regional constituencies, electing multiple members using
a numerical preference ordering ballot paper.
3.2 It will be very important to plan ahead for consistency
in the ballot paper designs that confront voters. Otherwise it
may create some acute comprehension and familiarity difficulties
for voters if they are asked to handle different ballot papers
in different fashions on the same day. In 2007 a Single Transferable
Vote (STV) system was introduced for Scottish local government
elections, using numerical preference ordering. The first STV
elections were on the same day as elections for the Scottish Parliament,
using a system called the Additional Member System (AMS), which
uses a two-vote X voting system. A great many additional spoilt
ballots and voter confusion resulted, something of a fiasco for
Scottish democracy.
3.3 Hence it is important to consider the compatibility
of the Alternative Vote version being proposed with other elections.
Table 3 shows the situation for the four versions of AV considered
above. The London AV system would be the most compatible with
other UK elections. Its adoption would entail the fewest knock-on
changes in other voting systems.
3.4 If Australian AV (or another version using numerical
preferences on ballot papers) is adopted for the referendum and
wins popular endorsement, we could expect to see:
— a greater likelihood of STV being adopted for future
elections for the House of Lords;
— perhaps a greater likelihood of STV being introduced
for local government elections in England and Wales; and
— perhaps a need to change the London Mayor and other
English mayoral elections to use numerical preferences also, such
as Australian AV or the `LAVno' variant of London AV.
Table 3
THE COMPATIBILITY OF DIFFERENT VERSIONS OF THE ALTERNATIVE
VOTE WITH OTHER VOTING SYSTEMS IN USE OR IN PROSPECT IN THE UK
|
Classic AV
|
Australian AV | London AV with numerical ordering
|
London AN |
Compatible with existing elections for
| Scottish local government (STV)
Local government and Stormont elections in Northern Ireland (both STV)
| |
— European Parliament (cannot be changed in UK alone: this is a List PR election)
— London Mayor and Assembly (AMS)
— Scottish Parliament (AMS)
— Welsh National Assembly (AMS)
— Local government in England and
— Wales (FPTP)
| | | |
|
Incompatible with existing elections for
| — European Parliament (cannot be changed in UK alone)
— London Mayor and Assembly
— Scottish Parliament
— Welsh National Assembly
— Local government in England and Wales
| |
Scottish local government
Local government and Stormont elections in Northern Ireland
| | | |
|
Compatible systems for electing House of Lords members would be
| — Single transferable vote (STV)
| |
— Additional member system (AMS) on a regional top-up basis
— regional List PR version with X voting
| | | |
|
25 July 2010
| | | |
|
|
