Memorandum by Professor Michael Gallagher,
Professor of Comparative Politics, Department of Political Science,
Trinity College Dublin
In this submission I offer some reflections,
having conducted some research in this area, on some of the issues
that the Committee is considering.
In summary, I suggest:
(a) in principle one could make a case either
that the referendum is an essential component of a truly democratic
state, or, conversely, that the referendum is inherently destructive
to coherent government and perhaps to democracy itself. The evidence
does not support either of these extreme claims;
(b) while there would be major difficulties in
accommodating the citizens' initiative within the UK's system
of parliamentary democracy, the referendum could much more easily
(c) it is certainly possible to argue reasonably
that certain issues should, and others should not, be the subjects
of referendums. Complications, though not insuperable ones, arise
in defining these issues in the UK given the absence of a codified
(d) threshold and turnout requirements are in
most cases unnecessary and undesirable, though a case can be made
for some combination of the two;
(e) there is little to be said for a referendum
that is designated as merely "advisory"; and
(f) the history of the referendum experience
of one particular country, Ireland, is on the whole supportive
of the claim that the referendum can enhance representative democracy.
I will address a number of the specific questions
posed in the Call for Evidence.
What are the strengths and weaknesses of the referendum
as a democratic and constitutional tool?
These are fairly well rehearsed so I will refrain
from relaying conventional wisdom at length.
On the positive side, the referendum can be
seen to enhance democracy by increasing opportunities to participate,
educating the voters by enabling them to decide on issues directly,
and enhancing the legitimacy of decisions made in this manner
compared with those made by Parliament. On the negative side,
it is accused of weakening representative institutions of government,
disrupting the process of government, and allowing majorities
to override the rights of minorities.
There is a certain amount of rhetoric and hyperbole
in both of these cases, and the most sweeping statements about
the supposed transcendent benefits or disastrous consequences
of the referendum are difficult to sustain on the basis of the
evidence. Democratic societies can function perfectly well with
the referendum or without it, and within the USA some states,
especially in the south-west, use the referendum extensively while
others do not use it at all, without any obvious difference in
the "level of democracy" enjoyed by their inhabitants.
Such countries as Belgium, Finland, Germany, India, Japan, the
Netherlands, the USA and indeed the United Kingdom are fully functioning
democracies despite making little or no use of the referendum.
On the other hand, so are Europe's most extensive users of the
referendum, such as Switzerland, Italy, Ireland and France, and
we cannot point to any case where the referendum has led to the
collapse of a democratic system.
If used sparingly (that is, as a supplement
to representative democracy, rather than so extensively that we
could speak of a "referendum democracy" or of "direct
democracy", a straw man erected by some of those who oppose
referendums in principle) and in a regulated fashion, the referendum
can enhance rather than subvert representative democracy.
How does, and how should, the referendum relate
to the UK's system of parliamentary democracy?
The distinction between two states of affairs
(i) a referendum can take place only after Parliament
has passed a bill or motion to that effect, specifying the precise
proposal to be voted on; and
(ii) extra-parliamentary actors, such as a specified
number of electors, can promote an initiative on which the people
Broadly speaking, where the situation is as
in (i), there is no reason to doubt that parliamentary democracy
and the referendum can co-exist and the referendum can reasonably
be seen as enhancing representative democracy rather than challenging
or threatening it. Where the initiative exists (situation (ii))
the implications for parliament are much greater, and the potential
for conflict and for a fundamental alteration to the existing
practices of representative democracy is much higher. In short,
introducing provision for the initiative is more "high risk",
and few countries have done this, Italy and Switzerland being
the best known examples. I imagine that the Committee is unlikely
to recommend the introduction of the initiative, either concerning
the introduction of new proposals or the rejection of recently-passed
legislation, as in Switzerland, or on the abnegation of existing
legislation no matter how old, as in Italy. Consequently, I shall
confine most of my comments to the first situation outlined above.
In theory, even the existence of referendums
in such regulated circumstances might be thought to pose a threat
to representative democracy. The essence of the argument is that
voters are less informed, and less cognisant of the need to respect
minority rights, than legislators ("legislators may be ignorant,
but not so ignorant as the masses", as Lord Bryce summed
up this line of thought over a century ago). In reality, there
is no evidence that the quality of democracy is weakened in the
extensive list of democracies that hold referendums under these
circumstances. Electorates can make ill-considered decisions,
but so can legislatures, and if the arguments highlighting the
supposed incompetence of voters to decide on specific policy issues
are pushed too far they can become an argument against allowing
people to vote at elections.
QUESTIONS 4, 5
Is it possible or desirable to define which issues
should be subject to a referendum?
Should "constitutional issues" be subject
to a referendum? If so, how should "constitutional issues"
Analytically it certainly seems that some issues
are generally deemed more suitable than others for resolution
by referendum. There is a clear logic in putting to a referendum:
(a) fundamental questions concerning sovereignty
or a major constitutional settlement, especially if they concern
steps that would be completely or virtually irreversible once
(b) important issues that cut across party lines
and upon which the electorate cannot reasonably be said to have
delivered a verdict at a parliamentary election.
Conversely, it does not make sense to put to
a referendum either:
(a) questions that relate closely to the left-right
spectrum on which the party system is based, since it is reasonable
to take the position that a general election broadly indicates
the electorate's decision here; and
(b) issues of secondary importance.
While this makes sense in theory, defining the
relevant terms is not straightforward. This applies a fortiori
in the UK given the absence of a single document entitled
"The Constitution" which, if it existed, could specify
the issues on which a referendum could (or must) be held. The
absence of any such document constraining the parliamentary majority
(i.e. the government) of the day might well mean that in practice
the decision as to whether a particular issue is to be put to
a referendum is likely to be influenced by political considerations.
This is something that has bedevilled the history of referendums
in France, for example, resulting in a perception that the government
of the day there chooses to hold or not to hold a referendum on
a particular issue depending mainly on where it sees its own partisan
advantage to lie.
Perhaps with sufficient thought a legal framework
could be devised that would give strong guidelines as to the issues
on which a referendum is ruled out, is possible, or is mandatory.
The government of the day would still inevitably have freedom
through its parliamentary majority to interpret this legislation
to its own advantage in any particular case, or to amend the legislation,
giving something of an ad hoc character to every referendum,
but the risk of losing support through being seen to violate the
"spirit" of the rules might constrain governments to
some extent. The Danish rule that delegation of sovereignty to
international authorities requires a referendum unless there is
a five-sixths majority in Parliament, or some variant of this
rule, would also be worth considering.
What comment would you make on key components
of a referendum campaign, such as:
(a) Whether or not there should be any threshold
requirements, for instance in terms of the percentage of the vote
required, or the level of turnout required, for a vote to be carried;
I do not see any justification for a requirement
that a super-majority of more than 50% should be required for
a proposal to be deemed to have received the support of the people.
A turnout requirement is, in my opinion, definitely
a bad idea, because this creates a dilemma for opponents of a
proposal, who may be unsure whether to turn out and vote "No"
or not to vote at all. Such a requirement means that a voter who
takes the trouble to vote "No" to a proposal that receives
majority support from voters may inadvertently be helping to validate
the poll by raising the turnout level.
Such turnout requirements do exist in some countriesfor
example, in Italy a "Yes" vote is not valid unless turnout
exceeds 50%. This leads to the anomalous situation where if 49%
of electors turn out and vote "Yes"while absolutely
no-one votes "No", the proposal is deemed not to be
approved, yet if 26% of the electorate turn out to vote "Yes"
and 25% vote "No", it is carried, even though quite
evidently support is much greater in the first case than in the
A requirement that avoids these difficulties
and is employed in Hungary is that the proposal for change must
receive a majority of the votes cast, and the Yes vote must amount
to at least 25% of the electorate. This ensures that proposals
cannot be passed unless they generate significant support, but
at the same time removes from opponents of a proposal both the
uncertainty as to whether a negative vote could have the unintended
effect of validating the result, and the power to thwart the majority's
preferences by simply not voting.
(b) the wording of the referendum question
(including the appropriateness of multi-option questions);
A referendum question should be framed in such
a way that the answer is unambiguous. Multi-option referendums
carry the risk that none of the options receives a majority, leaving
it open to debate and interpretation as to what the electorate
has said. If no one option receives a majority, some might interpret
this as a choice of the one that received most votes, yet it is
possible that that option is the "Condorcet loser",
i.e. the option that would be defeated by any other in a head-to-head
contest. Hence a choice between two optionsin essence,
asking the electorate to give a "Yes" or "No"
answer to a question as to whether it supports a specific proposed
This also implies that the questions that should
be put to voters are precisely those that can be answered with
a clear "Yes" or "No". It makes more sense
to put to the electorate a proposal put together by political
actors which is then referred to the people for their verdict,
rather than to bring in the electorate by referendum at an earlier
stage in an attempt to define the options.
(c) the design of the ballot paper;
(d) whether there should be formal, constitutional
triggers for referendums;
That would be the ideal, i.e. a referendum could
not be triggered simply by a possibly opportunistic decision made
by the government of the day. As discussed re Qns 4 and 5 above,
while the problems of institutional design involved in achieving
this are not insuperable, they would require some thought.
(e) whether a referendum should be indicative
Binding, in my opinion. An indicative referendum
is little more than an expensive opinion poll. If the electorate
knows that the result is not binding, then there is an obvious
temptation either not to vote or to use the referendum to express
a view on something other than the issue at stake, such as the
current performance of the incumbent government.
Given the UK's absence of a codified constitution,
it may be impossible to treat a referendum as being legally or
constitutionally binding, but it could certainly be politically
binding. That was the case with the UK's only national referendum,
for example; the 1975 referendum on whether the UK should
leave the EC could not be formally binding, in that Parliament
could not have been prevented from disregarding the decision of
the people had it chosen to do so, but there was never any question
of that happening. A number of other countries, including Denmark,
Finland, Italy, Norway and Sweden have held referendums that,
while formally only advisory, have been effectively binding.
(f) whether a referendum should ask broad
questions of principle, or refer to specific legislation;
As indicated at 8(b) above, a referendum makes
sense only if it carries a specific question to which voters may
respond with a "Yes" or a "No". Asking about
broad principles ("Should the membership of the upper chamber
be broadened?", "Should the Scottish parliament have
greater powers?") has the effect of turning the referendum
into an expensive opinion poll, and leaves huge scope for argument
as to what specific proposals are required by, or are compatible
with, the broad principles that the electorate has said that it
(g) whether a referendum should precede or
follow statutory enactment;
In principle it would be more logical for Parliament
to receive the approval of the electorate before enacting a piece
of legislation. That said, there is provision in some countries
(such as Switzerland) for a bill passed by Parliament to enter
a short state of suspension, typically three months or so, pending
the outcome of a referendum. Provided the people are asked to
vote on something specific, as opposed to a broad principle, the
precise sequencing does not seem to be crucial.
(h) campaigning organisations and the funding
(i) Public information campaigns and media
(j) Party political activity;
(k) whether referendums should coincide with
other elections or not;
There are advantages and disadvantages in holding
referendums on the same day as other elections, or in holding
more than one referendum on the same day. On the plus side, this
is likely to increase turnout compared with the result of holding
each contest on a separate day. On the minus side, there is a
risk that the, or a, referendum issue will be submerged by what
voters see as a more important contest.
In practice, many referendums around the world
have taken place on the same day as an election or another referendum,
and voters seem able to distinguish the two contests.
(l) the strengths and weaknesses of in-person,
postal or electronic forms of voting.
How does the referendum relate to other tools
such as citizens' initiatives? Should citizens be able to trigger
As noted above in answer to Q3, provision for
the initiative is a step furtherprobably several steps
furtherthan introducing provision for the referendum. The
initiative could be presented as a device that empowers citizens,
but critics would be more likely to see it as empowering well
organised lobbies who are able, if necessary by employing commercial
signature-gathering organisations as in California, to obtain
the requisite number of signatures.
Retrospective referendums would be another adventurous
option; even Switzerland does not employ this device. They exist
in Italy, where a specified number of citizens can launch an initiative
to strike down or amend any piece of legislation, no matter how
old, which enables citizens or interest groups to launch "random
shocks" against the political system. I very much doubt whether
this option is under serious consideration in the UK, and nor,
do I imagine, would many people recommend that it should be.
How would you assess the experience of other countries
in relation to the use of the referendum? What positive or negative
aspects of international experience would you highlight?
It is difficult to summarise worldwide experience
of the referendum in a paragraph or two, but I believe that the
evidence supports the view that the referendum (as opposed to
the initiative) is perfectly compatible with, and indeed enhances,
Since the Constitution Committee may be receiving
many submissions that evaluate worldwide experience or experience
in particular countries, I will briefly discuss Ireland's experience
with the referendum.
Ireland has a written constitution, dating from
1937, and this can be changed only by referendum. The referendum,
then, is intrinsically tied to constitutional change. If a particular
step requires an amendment to the constitution, then there must
be a referendum; if it does not require an amendment to the constitution,
then there is no referendum. The question of whether there should
be a referendum is thus taken out of the hands of political actors;
while they have discretion as to whether to try to bring about
some change (e.g. the legalisation of divorce in 1995), they have
no discretion as to whether there should be a referendum on the
issue once it has been decided to go ahead. As already noted re
QQ 4 and 5 above, there is an obvious contrast here
with the UK given that the UK does not possess a document called
There is surprisingly little debate in Ireland
about the merits of the referendum as a decision-making device,
which could be taken as an indication that it does not give dissatisfaction,
or that it is taken for granted as an unmoveable part of the institutional
furniture. The last time there was a sustained debate on the subject
appears to have been in 1928, when the government of the day used
its parliamentary majority to remove provision for the initiative
from the 1922 constitution, despite vehement protests from
the then main opposition party (which took a rather different
view once it came to government).
Ireland held 31 referendums in the period
1937-2009, all tied to constitutional change. The direct involvement
of the people in deciding issues such as abortion, divorce, and
constitutional recognition of Northern Ireland may have resulted
in a slower pace of change than would have occurred had parliament
alone been able to make these decisions, but it has conferred
much greater legitimacy and finality upon these decisions once
they were taken. A good example is the legalisation of divorce,
a long-running issue in Irish politics from the 1960s to the 1990s,
which was approved very narrowly in a 1995 referendum and
which then ceased overnight to be a political issue; opponents
immediately folded their tents following this decision in a way
that they would have been very unlikely to do had the decision
been made by Parliament alone.
In recent years the issue most commonly put
to the people by referendum has been Ireland's relationship with
the European Union, an issue that has been responsible for five
of the most recent eleven referendums to take place (1998-2009).
As is well known, Ireland was one of the few EU member states
to decide by referendum on the Nice Treaty, and was the only EU
member state to hold a referendum on the Lisbon Treaty. The initial
No vote on both treaties led to major problems for Irish governments
within the EU, and after an initial period of uncertainty on each
occasion the government eventually re-ran the referendums, having
achieved clarifications on some points that seemed to underlie
the opposition to the treaty, and on the second occasion each
time the treaty was ratified by a comfortable majority, with the
"Yes" vote exceeding 60%.
There is undoubtedly a feeling among some commentators
in Ireland that treaties of this nature are too complex to be
suitable objects of referendums, and that the requirement that
there should be a referendum on each EU treaty (a requirement
resulting from a 1987 Supreme Court judgmentsthat raised
eyebrows at the time) is an encumbrance to Irish governments when
it comes to taking part in the European project. Others maintain
that when questions of sovereignty arise it is right that the
people should have the final say, though the fact that the other
26 member states did not require a referendum on the Lisbon
Treaty raises the question of whether every such delegation of
sovereignty, no matter how minor, should require a referendum.
However, there are no proposals to change the requirement that
a referendum be needed for constitutional change, if only for
the simple reason that any such change would itself require approval
from the people in a referendum.
Overall, an evaluation of Irish experience of
the referendum would be more supportive of the institution than
critical of it. Partly because the question of whether or not
it should be used to resolve a particular question is determined
by the constitution and not by political actors, the institution
has not become discredited by partisan deployment. Even though
the number of referendums is high by European standards, referendums
are still sufficiently rare (fewer than one every two years since
1937) that the electorate does not appear to be jaded or resentful.
While turnout is low at what are seen as unimportant referendums
(for example, in a 1996 referendum on changing the bail laws,
turnout was only 29%), when the issue is seen as important it
is not much less than at a general election (it was 58% in the
second referendum on the Lisbon Treaty, held in October 2009).
Every country is different and has its own conventions, practices
and traditions, but comparative experience strengthens the basis
for assessments about how an institution will operate in a new
22 December 2009