Why Strike Ballots are Undemocratic

Since 2016, new legislation governing strike ballots has made it more dif ﬁ cult for trade unions to achieve a mandate for industrial action. Such a mandate now requires that a majority of members vote in the ballot. This article argues that these balloting processes are undemocratic. The turnout requirement means that a mandate for industrial action does not simply depend on its level of popular support amongst union members. This has surprising consequences. Sometimes opponents of action would be better advised to abstain, rather than to vote against it. Thus, it is not always clear how they should vote. Whatever they do, their actions may be counterproductive. Further, even when they do know how best to promote their desired outcome, there may be a con ﬂ ict between voting strategically and clearly expressing their true preferences. Consequently, there is no guarantee that the outcome of the ballot accurately re ﬂ ects what people really want.

THE TRADE UNION ACT 2016 restricts the right of unions to engage in strikes or other forms of industrial action, such as work-torule. A mandate for lawful industrial action requires a majority (over 50 per cent) of eligible members to have voted in the ballot. If this turnout threshold is not met, then there is no legal mandate for action, even if those who did vote overwhelmingly support industrial action.
While the act does not altogether remove the right to strike, it greatly constrains opportunities for workers to exercise this right. So significant are these constraints that the act was described by Oxford law professor, Alan Bogg, as 'a highly authoritarian strand of Conservative ideology which … is anti-liberal in its orientation'. 1 However, despite these additional obstacles, the UK has recently seen a wave of strikes arguably unparalleled since the 1970s. With so many unions balloting on industrial action, a re-evaluation of the Trade Union Act 2016 is particularly timely.
The 2016 act was intended to make it more difficult for unions to achieve a mandate for industrial action, which it does. However, I argue that the requirements for industrial action that it imposes are undemocratic. As a result of this act, the presence or absence of a legal mandate for industrial action need not accurately reflect the wishes of union members. Those who want to engage in action can reasonably complain that the odds are stacked against them. But these balloting rules can also frustrate opponents of action. It is sometimes better for them to abstain rather than to vote against action. This means that they may be unsure how best to promote their goals. In some cases, their attempts to block action can even be counterproductive.
Moreover, there is a further problem for opponents of industrial action, even if their best strategy is clear. They may be forced to choose between acting strategically, to promote their preferred outcome, and expressing their true opinions. For instance, even if they knew that abstention would be the best way to block a mandate for action, this does not signal opposition to action as clearly as a vote against it. This leaves opponents of action faced with a dilemma, giving them further reason to object to the balloting rules introduced by the act.
There is a similar problem for those who are neither in favour of nor against action. These people may wish to recuse themselves from the decision. However, since a ballot for action may be defeated by low turnout, abstention is no longer a neutral option. Thus, these people may also find that they cannot clearly express their attitudes. If people cannot unambiguously express their preferences, this undermines the democratic nature of the ballot.

Rights and democracy
One might argue that restrictions on the right to strike are undemocratic, because this right is 'one of the great keystones of democratic society'. 2 If the right to strike, like the rights to vote or to free speech, is itself part of being a democratic society, then any restrictions on this right would also be restrictions on democracy. However, even if the right to strike ought to be respected, the claim that this right is essential to a fully democratic society is contentious. Restrictions on this right may be unjust, but they do not obviously undermine democracy-in the way that restrictions on the rights to vote or to free speech do.
My argument concerns democratic processes, rather than whether certain outcomes better realise democratic values. I take it that a democratic process is one where the outcome depends on how people vote. While there are various voting systems that may be usedwhich can sometimes deliver different outcomes-these are all responsive to how people vote. In contrast, a constitution that imposes certain outcomes, regardless of the wishes of voters, is less democratic in process, whatever else might be said in favour of those outcomes.
The turnout requirements imposed by the 2016 act are undemocratic in this sense, because the legal mandate for industrial action no longer depends simply on how union members vote, whether for or against action. Rather, it also depends on how many union members vote. These twin requirements are intended to make it harder for unions to organise legal industrial action. Consequently, the procedure is not neutral between industrial action and no industrial action. There is an additional requirement that those seeking a mandate must satisfy, but that their opponents do not need to satisfy. This tips the scales against action, which might be regarded as objectionable by those who support the right to industrial action.
However, it is not only those who favour industrial action who have reason to object to the balloting procedures. These requirements also impose a very different kind of burden on those who oppose action (and also on some who are neutral towards action). Members of these groups may find themselves facing a dilemma, because it is not always obvious how they should vote-either to promote their preferred outcomes or to express their preferences. Indeed, they may even be forced to choose between expressing their true views and promoting their preferred outcomes.
While one might naturally assume that opponents of industrial action should vote against it, this is not necessarily their best strategy. In fact, voting against industrial action may even be counterproductive, since it could result in the mandate requirements being met when they would not have been otherwise. Hence, opponents of industrial action would sometimes be better off not voting at all (that is, abstaining) rather than voting against it.
Things would be bad enough if this was all. If that were the case, then opponents of industrial action would simply learn that abstaining was a better strategy than voting against action. But matters are not this simple. Though it is sometimes better for them to abstain than to vote against industrial action, there are other occasions where the turnout threshold is met anyway. In these cases, their abstention may lead to a mandate for industrial action that could have been avoided had more people voted against it.
The problem is not merely that it is better for opponents of action to abstain than to vote against it, but that their best strategy is not always obvious at the time they must vote. They may not know how to promote their desired outcome. Moreover, whatever they do, their actions may turn out to be counterproductive, in the sense that they could result in a mandate for industrial action that might not otherwise have existed, and be liable to misunderstanding by others. Consequently, these people cannot be sure how (or whether) they should vote. They are caught in a dilemma.
This problem threatens the democratic character of the ballot. If people cannot cast votes that unambiguously reflect their views, then some votes may be misleading regarding what people really want. If there are enough of these votes, then we cannot be sure whether the outcome of the ballot-whether it results in a mandate for industrial action or notaccurately reflects the wishes of union members either. It may be that the mandate is achieved, or not, because of votes that do not accurately reflect people's wishes.

Voting paradoxes
The turnout requirement introduced in the Trade Union Act 2016 is intended to make it harder for unions to achieve a mandate for industrial action. However, the turnout requirement imposed by the act does not do much to ensure that there is popular support for action. Further, it produces some rather surprising consequences.
Consider the hypothetical case of a union with 1,001 members. Any ballot for industrial action would need at least 501 members to participate, but it does not require that all of these voters support industrial action. Those who vote could be almost evenly split, with only 251 voting for action and the other 250 against. In this case, just over a quarter of members support industrial action, but they would nonetheless have a legal mandate because over half (501 out of 1,001) voted and a majority (251 out of 501) of those voted in favour of action.
In fact, even more extreme cases are possible. The turnout requirement includes not only valid votes, for or against action, but also spoilt ballots. So, in theory, the mandate could be achieved by only one member voting in favour of action and 500 more returning spoilt ballots. This would still satisfy the turnout requirement, because 501 members cast ballots, even though the vast majority of those were spoilt. Since these spoilt ballots would not affect the outcome either way, this would count as a vote in favour of strike action (by one vote to zero). So, a legal mandate could theoretically be achieved even when hardly anyone supports industrial action. However, for present purposes, I will ignore the possibility of spoilt ballots and assume that all votes cast are valid. Even so, a legal mandate for action can be achieved with little over a quarter of union members actively supporting action.
On the other hand, there may be other cases where there is more popular support for action, but the requirements for a mandate are not met. Suppose that 500 members voted for industrial action. This is almost twice as many as in the previous example, but this time assume that they are the only ones to vote and the other 501 all abstain. In this case, though more people have voted in favour of action, and no one voted against it, there would be no mandate for action because the ballot was one short of the turnout requirement. It is surely odd that the 251 members who support industrial action in the first example can succeed in achieving a mandate, while the 500 members voting for action in the second example do not. These two examples show that a turnout requirement is not a reliable indication of the level of popular support for action.
The crucial difference between these casesthe 251 with a mandate and the 500 withoutlies in the number of members who vote against action. In the latter case, the turnout threshold (501) was narrowly missed. But it would have been achieved had just one more person voted, even had that extra person voted against industrial action. Had one of those 501 voted against action, instead of abstaining, then the turnout threshold would be met and there would have been a mandate for action. This means that one person deciding to vote against action, rather than to abstain, has resulted in a mandate for action. So, voting against industrial action can bring it about that there is a mandate when there would not otherwise have been one. This is certainly counterintuitive.

Real-world examples
Of course, the hypothetical examples of the previous section merely illustrate logical possibilities. In practice, such extreme cases are unlikely. But there are real cases that involve similar issues. This can be seen in recent University and College Union (UCU) ballots.
UCU members were invited to vote in two separate disputes. The first dispute was over changes to the Universities Superannuation Scheme (USS) pension, affecting only some universities (since staff at post-1992 universities are generally in the Teachers' Pension Scheme). The second dispute concerned what union officials referred to as 'four fights': over pay, casualisation, equality, and workloads. In each dispute, members were asked to vote for or against strikes and simultaneously for or against action short of a strike (ASOS). Thus, many members had four separate ballots to complete. Further, some universities where thresholds were narrowly missed proceeded to re-ballot their members.
The figures below all relate to the first round of ballots concerning strike action over pensions, the results of which were announced on 4 November 2021. Newcastle University saw 496 members vote in favour of strike action, which was 80.1 per cent of the 619 votes cast. 3 This may look like overwhelming support, but it fell short of a legal mandate because overall turnout was only 49.5 per cent (based on 1,250 eligible voters), narrowly missing the required threshold. Had only seven more members voted, then they would have had a mandate for a strike. This would be so, even if all seven of those additional votes were against strike action. (In fact, support for a strike was so great amongst those that did vote that, even if 370 of the 631 abstainers had instead voted against a strike, this would not have been enough to change the outcome. There would still have been a majority in favour, by 496 votes to 493.) On the other hand, the University of Reading narrowly met the turnout threshold (370 votes out of 731, or 50.6 per cent), with 228 (61.6 per cent) of these favouring strike action. 4 Here a small difference in turnout (50.6 per cent, compared with 49.5 per cent at Newcastle) was crucial, because Reading was just over the required threshold. The result is that Reading UCU satisfied the legal requirements for strike action, despite lower levels of support for a strike amongst those who did vote. Indeed, Reading also had a smaller percentage of members supporting the strike. The 228 who voted in favour of strike action at Reading represent only 31.2 per cent of the 731 eligible members, whereas the 496 who voted for strike action at Newcastle comprise 39.7 per cent of the 1,250 eligible members. Thus, not only did Newcastle have a higher proportion of those who voted in favour of action, but it also had a higher proportion of union members in favour.
If we think that a mandate for industrial action should depend on levels of popular support, then it is puzzling that Reading should have a mandate, but not Newcastle. Note that this is a comparative claim. Even if you think that neither branch should have a mandate, or that both of them should, the point is that Newcastle surely has the stronger claim of the two. However, the 2016 act granted Reading, but not Newcastle, a mandate for action despite lower levels of support. The presence or absence of a legal mandate for action is not, therefore, a reliable indication of support for industrial action among branch members. As these examples show, the turnout threshold imposed by the act has surprising-and arguably undemocraticimplications in real-life cases and not only in hypothetical illustrations. The turnout requirement may block industrial action in some cases, even where it has relatively high levels of popular support amongst union members.
To be sure, the turnout requirement will never block action that a majority of members of voted for. If a majority of members vote in favour of action, then they will be sufficient to meet the turnout threshold too. However, there may be cases like Newcastle, where votes in favour of action significantly outnumber votes against, while falling narrowly short of a majority. Though this does not show that a majority of members support action, the claim that there is 'no majority for industrial action' in cases like this is misleading. By focussing on the whole membership, rather than those who vote, it effectively counts abstentions alongside those who vote against action. Failing to distinguish between abstention and opposition may misrepresent the views of many members, giving a misleading impression of the level of support for industrial action.
The Trade Union Act 2016 imposes ballot rules which give a mandate to Reading but not Newcastle, despite there being more support for industrial action in Newcastle. This mandate is not based on the wishes of union members. This is one reason why these ballots are undemocratic. This is bad enough, but the turnout requirement is also problematic for another reason. These rules can make it difficult for members to decide how they should vote, even if they know what it is that they want.

Strategic voting
Deciding whether to participate in a strike, or other industrial action, is often a difficult decision. Those involved must consider not only whether they have a just grievance, but also how likely action is to achieve anything and what effects it would have on other parties, including both those that they serve (customers, students, patients, and so on) and their own family members. Unsurprisingly, workers often feel torn over whether or not to join a strike. However, let us assume that we are dealing with workers who know what they want to do. Even for these workers, it is not always clear how they should vote in order to promote their preferred outcome.
It is clear what those who support industrial action should do. They should vote in favour of action. This makes a mandate more likely in two ways. First, they contribute to meeting the turnout threshold simply by voting, regardless of how they vote. Second, by voting in favour of action, they also make it more likely that there will be more votes in favour of action than against it. Thus, for these workers, there is no conflict.
Things are less obvious for those who oppose action though. One might naturally assume that they should vote against action. However, this is not always their best strategy. In fact, it may actually be counterproductive. Recall that unions sometimes fail to achieve a mandate because of insufficient turnout. In cases like that of Newcastle University, more votes against the strike may have led to the turnout threshold being reached, without changing the majority decision. So, someone voting against industrial action here could bring it about that there was a mandate for action, when there would not otherwise have been one had they abstained instead.
Based on this example, one might assume that abstention is always the best policy for opponents of industrial action. However, while abstention is sometimes an effective way to prevent action, it could backfire in other cases. Sometimes it may be very likely that the turnout threshold will be met, but the vote will be close between those for and against action. In those circumstances, those who oppose action would be better off voting against it. If the turnout threshold is met anyway, then abstention does no good and may instead result in supporters of action winning a majority of the votes.
While supporters of industrial action have a dominant strategy (vote in favour of action), it is not always clear what its opponents should do. In some cases, they may be better advised to abstain, hoping that the ballot fails owing to insufficient turnout, while in others their best option may be to vote against action. Because of this uncertainty, they may find that their actions-whichever of these options they choose-prove counterproductive, resulting in a legal mandate for action that would not otherwise have been achieved. Thus, they might end up regretting their action, because it has contributed to a mandate for industrial action that they opposed. The problem is that, without knowing how others will act, it may not be clear in advance whether it is better for them to vote or to abstain.
It might be replied that I have exaggerated the uncertainty, because it will often be possible for those who oppose strikes to predict how others will vote and to respond accordingly. For instance, if they know how likely others are to vote, or how they are likely to vote, then this can inform their voting strategy. No doubt this is sometimes the case. However, it assumes a certain amount of political sophistication amongst the voters in question. Some may vote against industrial action, without realising that their actions may actually make it more likely. Conversely, others may abstain, perhaps because they are ambivalent about industrial action, without realising that their abstention could be more effective than voting against the strike. In any case, even those who understand the voting system and can predict likely outcomes with reasonable accuracy may find themselves placed in an awkward situation. There may be circumstances in which they have to explain or justify to others how they voted. Since those others will not necessarily understand how the ballot works, they may not realise that voting against industrial action can make it more likely, or that abstention can be more effective than voting against it. They may expect someone to vote against action, even though doing so would actually make action more likely. This may make it difficult for someone who opposes action to explain why they did not vote against it.
However, the problem here is not simply the difficulty of explaining or justifying one's actions to others. This difficulty only arises because there is a potential conflict between voting in a way that accurately represents one's views on industrial action (which I assume one opposes) and voting in a way that best promotes one's preferred outcome (which might be abstaining, in the hope that the turnout threshold is not reached). Someone who votes against industrial action may succeed in expressing their opposition, but at the cost of making such action more likely. Conversely, someone who abstains may make the action less likely, but fails to express their opposition to it. Their abstention might be taken as a sign of apathy or indifference, rather than strategic opposition.
Thus, those union members who are opposed to industrial action may find themselves in a dilemma. It is not simply that they do not know how best to vote in order to promote their preferred outcome (no action), or that their actions may prove counterproductive (by making action more likely). Rather, they may be torn between voting in ways that are more likely to bring about their preferred outcome and voting in ways that clearly signal their attitudes towards the dispute in question. While conflicts between different reasons for voting are not unique to strike ballots, confronting voters with such dilemmas is an undesirable feature of voting systems. 5

Neutrality and abstention
So far, I have argued that both supporters and opponents of industrial action may have objections to the balloting rules. However, there is one other group to consider, namely those who neither support nor oppose action. By this, I mean those who genuinely have no preference either way, distinct from those who oppose the strike, but may abstain for strategic reasons (as discussed above).
There are various reasons why someone might have no preference, either for or against engaging in industrial action. For one, the issues involved are often very complicated. Those contemplating industrial action may have to consider many variables, including not only the justice of their cause, but also the likelihood that action will achieve anything and whether the costs that the action imposes are proportionate to the likely benefits. It might be that some people do not feel able to make an informed decision on these matters. Second, it might be that some people are ambivalent, because they feel the pull of reasons on both sides. In these cases, they may not come down on one side or the other. Third, there may also be cases where someone decides to abstain because they judge that the issue does not concern them. Someone close to retirement or already working out their notice period, for example, might be eligible to vote in the ballot though they will no longer be working in the occupation by the time that any action takes place. In such cases, they may feel that the dispute is no longer their business and that the decision ought to be left to others.
Whatever their reasoning, those who have no preference may want some way to excuse themselves from the decision making, leaving the decision to others. Ordinarily, one can do this by abstaining, since this usually has no effect either way on the outcome. However, when sufficient abstentions can serve to block a mandate for industrial action, then abstention is no longer a neutral option. Those who merely want to leave the decision to others may be lumped in with others who oppose industrial action or who abstain for strategic reasons. Thus, abstention does not clearly express their attitude either.
It is not clear whether there is any entirely neutral option open to these people. Perhaps it would be better if they could disenfranchise themselves, so they do not count either as voting on either side or towards the size of the electorate (and, consequently, the necessary turnout). However, while the Trade Union Act 2016 does allow some people to be excluded, this is not an option for everyone. Given the current balloting rules, the best option for others who neither support nor oppose action, but who do want to support democratic decision making, is arguably to cast a spoilt ballot. This means that they do not count as a vote for or against action, but would still count towards meeting the turnout threshold.
It might be objected that casting a spoilt ballot is not entirely neutral, since contributing towards meeting the turnout threshold still makes industrial action more likely. There is some truth to this. Because a mandate for industrial action requires sufficient turnout, supporters of action want to promote higher turnout, whereas its opponents want lower turnout. Thus, casting any vote can be seen as helping the supporters of industrial action to overcome one of the obstacles to achieving a mandate. However, this is true even if that vote is a vote against action. The fact that supporters of industrial action would sometimes prefer the undecided to vote against that action, rather than to abstain, further highlights the problems with these balloting rules.
If the turnout requirement is itself undemocratic, then casting a spoilt ballot contributes to ensuring that the decision-either to engage in industrial action or not-is made according to the wishes of union members. Thus, it might be argued that even those with no preference either way ought to return spoilt ballot, in order to resist the 2016 act's attempts to frustrate democratic decision making. By doing so, they help to ensure that the decision is made in a more democratic fashion, rather than the result being imposed by undemocratic legal requirements.

Conclusion
The Trade Union Act 2016 imposes various restrictions on ballots for industrial action. In particular, a mandate for action now requires that a majority of eligible union members vote in the ballot. This turnout requirement is intended to make it harder for unions to achieve a mandate for industrial action. Therefore, it is unsurprising that those who favour action have reason to oppose it. However, I have argued that others also have reason to object to democratic deficits in the resulting balloting procedure.
Given these rules, it is not always clear whether opponents of action are better off voting against it or abstaining. The first problem they face is that either of these options may be their best strategy in some circumstances, but counterproductive in others. Consequently, their attempts to prevent action may make only make it more likely. But a second problem is that they cannot clearly and accurately express their views. They may be in a situation where abstention is their best strategy for preventing industrial action, but this option-while instrumentally effective-does not accurately express their opposition to action.
A similar problem befalls those who are ambivalent or undecided. Ordinarily, such people might abstain from voting but, in this case, that is no longer a neutral option, because their abstention might prevent a mandate for action. Further, were they to abstain, it would not be clear whether this was because of ambivalence or strategic opposition to industrial action. Again, they may not be able to express their true preferences.
Democratic decisions are supposed to reflect the wishes of group members. But these wishes have to be communicated through votes. If people's votes do not reflect their true preferences, then the outcome may not be what people want. Where balloting rules make accurate votes potentially counterproductive and incentivise people to vote strategically in order to make their preferred outcome more likely, then people's votes may not express their true preferences. Consequently, the outcome determined by those votes may not be what people really want.
The current balloting rules not only create additional obstacles to achieving a mandate, but also place some potential voters in a situation where, first, it is not clear whether voting will promote their preferred outcome and, second, where-whatever they do-they might later find that their actions were instrumentally counterproductive and/or expressively misleading.