Trade-union recognition, the Amazon case, and a New Deal

In this series of blogposts the Institute of Employment Rights explores various policy dimensions associated with the delivery of Labour’s reform agenda in the domain of workers’ rights. It argues that while several reform proposals contained in Labour’s Plan to Make Work Pay – Delivering A New Deal For Working People seek to address genuine regulatory failures and a number of major concerns faced by workers and unions alike, they sometimes lack in detail and/or occasionally appear to fall short of the policy objectives they appear to pursue.

Implementing in full Labour’s New Deal for Working people will require a significant policy delivery effort, spanning well beyond the quasi-emergency legislation that the new executive intends to present to Parliament in the first 100 days of government, in the form of an Employment Rights Bill. Ending the Conservatives’ ‘scorched earth approach’ to industrial relations will not be a quick fix. This blogpost explores the main reforms necessary to remove current barriers to trade unions being recognised in the workplace and workers being collectively represented by a trade union.

Introduction

Labour’s Plan to make Work Pay: Delivering A New Deal for Working People proposes to ‘simplify the process of union recognition and the law around statutory recognition thresholds, so that working people have a meaningful right to organise through trade unions’. This is in addition to ensuring that ‘workers in precarious and gig-economy sectors have a meaningful right to organise through trade unions’ by ‘modernising’ the rules to ‘ensure they are fit for an economy with growing platform sectors and a rise in remote and home working’. Both ‘simplification’ and ‘modernisation’ are clearly very welcome, given the limited effect of the statutory procedure which was introduced by the Blair government in 2000. It is perhaps testimony to the ineffectiveness of that procedure that for 14 years it survived the anti-union onslaught of Conservative-led governments. The procedure does not continue to exist because it empowers trade unions or because it imposes undue burdens on employers.

So far as ‘simplification’ is concerned, Labour is committed to:

• ‘review the process for statutory recognition claims, with existing thresholds presenting too high a hurdle in modern workplaces that are increasingly fragmented’;
• ‘remove the antiquated rule that means that unions must show that at least 50% of workers are likely to support their claim before the process has even begun’; and
• ‘modernise the rules governing the final ballot in which workers vote on whether to recognise a trade union, requiring unions to gain a simple majority to win’.[1]

In addition to new rights for gig-economy workers, in the abstract these three commitments are not insignificant. When set against the problems with the recognition procedure exposed by the recent Amazon case, however, they begin to look largely inconsequential, and certainly not enough. Amazon has created the opportunity for more radical surgery, and reminded us of the need for a statutory recognition procedure that moves away from the requirement to conduct a divisive and polarising ballot, which serves to focus the employer’s mobilisation against recognition rather than provide a true test of worker support for it.

The Amazon case

At the time of writing the GMB has made two requests for statutory recognition at Amazon’s Coventry site:  on 12 May 2023 and 4 March 2024.  On 25 April 2023 the GMB had 718 members in what was believed to be a workforce of 1,400 in the proposed bargaining unit.  This would have been enough not only to meet the admissibility requirements under the current legislation but perhaps also to secure recognition without a ballot (unless the company was able to persuade the CAC to order a ballot on industrial relations grounds).  However, the company refused the request for voluntary recognition, and when the union made an application under the statutory procedure it was told on 22 May 2023 that the employer had 2,749 workers in the proposed bargaining unit, which had apparently doubled in the course of a few months.  The union was now in a position whereby it would meet the first admissibility requirement of 10% membership but could not be confident at that stage that a majority of the inflated bargaining unit would support recognition.

If the union had continued with the claim and failed at the admissibility stage, it would have been barred for three years from bringing another claim in respect of the same group of workers. So the application was withdrawn.  Thereafter the union continued to grow in membership, reaching 1,255 members by March 2024. The company also continued to grow, claiming that at that stage there were 3,088 workers in the proposed bargaining unit.  Nevertheless, a second application was made by the GMB under the statutory recognition procedure, on this occasion successfully navigating the admissibility requirements.  In holding that the application was admissible, the CAC concluded that ‘the level of union membership within the bargaining unit stands at 35.62%’, that ‘membership of 35.62% in a proposed bargaining unit of some 3,088 is a relatively high proportion’, and that in a ballot ‘in such circumstances, support for recognition would be likely to surpass 35.62% by a margin which would take the total support to over 50%’.

Despite this, a majority of workers in the bargaining ballot voted to reject the union.  According to the CAC:

‘out of 3012 workers eligible to vote, 2601 ballot papers had been returned: 11 ballot papers were found to be spoilt. 1281 workers, that is 49.5% of those voting, had voted to support the proposal that the Union be recognised for the purposes of collective bargaining with the Employer. 1309 workers, that is 50.5% of those voting, voted to reject the proposal.  The number of votes supporting the proposal as a percentage of the bargaining unit was 42.5%’

The ballot result followed allegations (which would no doubt be contested) of unfair conduct by Amazon, which according to the GMB website consisted principally of the company creating ‘a culture of fear for low paid workers trying to improve their pay, terms and conditions’, with ‘relentless attacks’ on ‘their own workforce’.  Specifically, it was claimed first that:

‘union-busting tactics at the Coventry site were exposed when it was revealed workers had been bombarded with an unrelenting campaign of anti-union messages by company bosses, including multiple anti-union seminars’.

And secondly that:

‘We’ve seen workers pressured into attending six hours of anti-union seminars on top of the fortune spent by Amazon bosses to scare workers’, while

‘Workers have been told they will get no pay rise this year and will have to lose even more benefits if they vote for union recognition’.

These were in addition to the use of ‘one-click-to-quit QR codes’, in which posters throughout the workplace encouraged GMB members to click on a QR code which instructed the union on behalf of the member to cancel his or her membership.  As Foxglove Legal put it:

‘The QR code for the tool is displayed on each table in the canteen, in every meeting and training room, in every hallway, and every toilet, as well as shown every five minutes of the working day in screens at the centre of the warehouse.

If a worker scans the QR code and clicks the link, Amazon’s software automatically generates a letter from the worker and then emails it to the trade union, ending their membership.’

Following the ballot result, the GMB complained to the CAC. But the merits of the complaint could not be heard because the application to hear it had not been brought within the tight statutory time limits for bringing such complaints. Under existing law, claims must be brought on or before the first working day after the date of the ballot.  If the claims had been brought in time and if they had been substantiated, the CAC has the power (albeit rarely if ever exercised) to order the ballot to be re-run.

As explained above, the effect of the failure of the second statutory recognition application is that the GMB is barred from making a third application for three years. By that time many existing members will be working elsewhere and Amazon will be even better prepared to resist the claim. We now have a stalemate, which it is likely will only effectively be resolved by legislative intervention:  the law needs to change. In the meantime, it is reported that some of the tactics alleged to have been used by the company during the most recent recognition campaign are to be challenged in separate legal proceedings supported by the union. These relate to claims that Amazon induced workers not to be union members, which it is said by the union is contrary to the Trade Union and Labour Relations (Consolidation) Act 1992, section 145A. The claims, will of course, be vigorously contested. But if successful, they may bring some joy to the 900 or so workers on behalf of whom the union is litigating. And although each individual could recover £5,584 should these claims succeed, a successful outcome in those proceedings will not deliver recognition for the union.

Lessons from Amazon

As suggested above, the Amazon case exposes multiple critical failings of the statutory recognition procedure which the new government urgently needs to address.  As highlighted by a powerful report on the case by the GMB, these failings relate to (i) the feigned neutrality of the law, (ii) the high statutory thresholds that unions need to cross to bring a recognition application, (iii) the unfair and old-fashioned ballot process; (iv) the lightly controlled power of the employer to disrupt applications by threats and intimidation; and (v) the severe limitations on access to justice when employer wrongdoing is alleged.  The need to overcome these failings underpins the legal changes that are necessary if a deeply flawed procedure is to operate at all effectively as an instrument for promoting collective bargaining.

The first step that could be taken would be to reduce the need for a ballot, and with it the exposure of workers to hostile campaigning.  As already pointed out, in Labour’s Plan to make Work Pay the government pledged to remove the requirement at the admissibility stage for unions to show that at least 50% of workers are likely to support the application.  Building on that pledge, the GMB appears to propose that where a union meets the current admissibility criterion of 10% membership, plus evidence of majority support, then recognition should be granted automatically. If this welcome change had applied in the Amazon case, it would have been open to the union to seek support from non members in a petition and to have secured recognition automatically.  This would not, however, have precluded it, if it chose to do so, from proceeding to a ballot rather than seeking to show majority support by other means.

Where a ballot is to be held, the second step that needs to be taken is radically to revise the rules relating to the conduct of the ballot, in terms of the voting constituency, the ballot method, and the ballot outcome.  As persuasively argued by the GMB only those who are in employment at the time the recognition application is accepted should be permitted to vote, in order to stop the employer from ballot rigging.  In addition, the vote should be conducted electronically wherever the union requests, in order to reduce the risk of employer interference and intimidation.  Furthermore, recognition should be ordered where the majority of those voting in the ballot vote in favour, thereby dispensing with the additional requirement that those voting in favour should represent at least 40% of those eligible to vote.  The Labour Party has at least accepted the unfairness of the latter requirement.  But removal of the 40% threshold will not on its own deal with the defects of the existing law.

Thirdly, the Amazon case has emphasised the need to address the problem of employer power.  To this end, (i) in cases where recognition is not automatic the statutory recognition process needs greatly to be speeded up to reduce the opportunity for employer interference; (ii) access to the workforce needs to be organised on an equitable basis to ensure campaign fairness; and (iii) the opportunity to make a fresh claim should be reduced from three years to six months, to remove the reward for bad practice on the part of the employer.  Beyond that, however, there is a need to address employer behaviour.  Existing law in the form of unfair practices is addressed to specific activities of employers (such as bribes, inducements and threats of various kinds).  This needs to be strengthened so that it is unlawful as a matter of labour law generally applicable during a statutory recognition process and all at all other times for an employer to:

• interfere with the relationship between workers and their trade unions, for example by encouraging, facilitating or inducing them to resign from membership; and
• interfere with trade union recognition whether voluntary or statutory, for example by encouraging or seeking to persuade workers not to support it.

One final matter exposed by the Amazon case addressed by the GMB relates to the time limits within which unions must bring complaints of an unfair practice by the employer.  As we have explained above, the procedure allows the union to complain to the CAC about any breaches, but the complaint must be made within 24 hours of the ballot closing.  It is difficult to understand the purpose of this tight time scale which contrasts sharply with the leisurely pace at which much of the rest of the procedure is conducted where, coincidentally, delay suits the convenience of the employer.  It also difficult to understand how such a limitation period is consistent with fundamental constitutional principle:  a first principle of the rule of law is that citizens and organisations should have effective access to justice.    That principle is obviously undermined by time limits which expire before the existence and extent of unfair practices have yet come to light, particularly where they were covert.

Conclusion

The Labour Party’s proposals for the statutory recognition procedure are welcome, but as the Amazon case makes clear they do not go nearly far enough.   The Amazon case provides a compelling justification for serious reform.   A union should be entitled to recognition on showing 10% membership and evidence of majority support, if not – as in Australia – on showing evidence of majority support alone.   If a claim goes to a ballot, it should be on the basis of revised ballot rules to prevent ballot rigging and employer interference.  In addition, employer conduct should be more tightly regulated, and bad behaviour properly addressed, with appropriate remedies where such behaviour is exposed.   But although the lessons from Amazon are clear, there are other issues that also need to be addressed, beginning with the need to expand the mandatory bargaining issues under the statutory procedure to cover all terms and conditions of employment, and not only pay, hours and holidays.

Otherwise, a further problem is that the procedure enables a representative union to be shut out of the process if the employer has already voluntarily recognised a ‘sweetheart’ union – even if the union has no members, and even if the recognition is for peripheral subjects excluding pay, hours and holidays. This is clearly untenable, as is the fact that an independent trade union can be shut out by an agreement with a non-independent union, until steps are taken formally to derecognise the latter. Such agreements should be legally invisible, having no legal force or effect.  And in improving the recognition procedure, it should not be overlooked that it applies only to collective bargaining with a single employer, and that it is not permissible to seek recognition by multiple employers.   A claim can be made jointly by trade unions against a single employer:  it ought to be possible for a claim to be made by one or more unions against multiple employers for a single bargaining unit across company frontiers.

[1] Emphasis added.