Printer Friendly

Final offer arbitration in Great Britain: style and impact.

Within the last decade a number of British workplaces have signed agreements incorporating final-offer arbitration (FOA), a well known procedure in the US but unusual in British industrial relations. This is the first thorough study of recent British experience. Data are drawn from a postal survey of 72 plants, yielding information on nearly 300 wage bargaining rounds, and from interviews with eight managers at FOA plants which had disputes under the procedure. A full picture of the style and impact of FOA procedures in British workplaces is documented.

I. Context and forms of final-offer arbitration in Great Britain and the US

1. Introduction and context

Final-offer arbitration (FOA) is an impasse resolution procedure which involves an arbitrator being constrained to decide for one or the other sides' final offers in a dispute. This is in contrast to conventional arbitration where the arbitrator has the option of compromise, usually within the terms of reference specified by the parties. FOA is also known as pendulum, either-or, flip-flop and last offer arbitration. The theory behind this type of dispute procedure is that it provides an incentive for both parties to moderate their positions to such an extent that third party intervention is not required (see section 5 for details). It is specifically designed to be an impasse (sometimes known as 'failure to agree') deterrent.

Collective agreements incorporating an FOA procedure have been signed in a number of British private sector plants in the 1980s, forming some of the so-called 'new style deals'. The plants with such deals have been characterised as predominantly greenfield and owned by foreign, often Japanese, hi-tech companies (Bassett 1986, Rico 1987). Co-signatory to many of these agreements was the Electrical, Electronic, Telecommunication and Plumbing Union (EETPU).(1) Typically the agreements have also often incorporated single-union recognition, strike-avoidance or no-strike clauses, task flexibility and single status for white and blue collar employees. FOA is therefore usually only one element of a 'package' offered to the company by the union (Bassett 1986).

It might be noted in passing that these agreements are not the first examples of FOA in this country. Some FOA arrangements existed in the Victorian and Edwardian period, the first example being the conciliation board set up in the Nottingham hosiery industry in 1860 (Cuthbert 1960). Another example is the coal mining industry negotiations between 1894 and 1914 (Treble 1986, 1990). Finally, Wages Councils which set pay rates for 2.5m people in unorganised or weakly organised sectors (and which have been in place for almost the whole of the twentieth century) also resolve impasses via FOA. A Wages Council has three parties--employers, workers and independent members. The process of negotiating terms and conditions involves conciliation, mediation and, at the last resort, a straight choice between the two sides by the independent members (Wood 1985: 421).

To date the only studies of the recent UK experience with FOA have been by arbitrators, describing cases when FOA has been triggered (Kessler 1987, Gennard 1988, Lewis 1990). These studies have a generally consistent theme which is similar to that expressed by some American writers: that FOA is all very well in theory but when it actually fails to prevent a dispute, then there are problems in its application. Up to now there has been no attempt to assess the recent British experience of FOA on the basis of predictions generated by US scholars concerning incentives to negotiate and other outcomes.

This study of FOA in Britain examines its context, form, incidence and impact. The new style deals which do not have FOA are used as a control group. Data are drawn from a survey of 72 workplaces, providing information on nearly 300 annual bargaining rounds over five years (1986-90), and interviews with managers at eight plants where FOA procedures have been invoked. Final-offer arbitration has emerged in Great Britain in a very different context to that in which it operates in North America. It is important to be aware of these different environments since certain implicit assumptions about FOA theory have been made in the US because of its context there. In particular it is assumed that arbitration is compulsory, that is, it is triggered automatically once a certain stage in negotiations is reached without agreement. And second, arbitration is held to be a specific alternative for the right to strike. These assumptions are implicit in the sense that FOA was designed and developed with the resolution of disputes in the North American public sector in mind, where industrial action is legally prohibited for all public employees unless given the right to strike by the relevant state or national legislating authorities (Ashenfelter 1985: 2).

The situation is very different in the UK, such that these two important assumptions are not necessarily valid. All the FOA agreements in the UK were voluntarily signed, rather than legally imposed as in North America. Collective agreements in the UK are generally open-ended (not fixed term) and not legally enforceable, since the parties are 'conclusively presumed by the 1974 Trade Union and Labour Relations Act (section 18) and previously by the common law, not to intend to create legal relations' (Lewis 1990: 38). In contrast collective agreements in the US and Canada are normally legal contracts and are usually fixed-term (often three years). One consequence of the different legal status of UK and North American agreements is that any 'strike-avoidance' or 'peace' clause in the UK is not legally enforceable, so industrial action is not ruled out.

Furthermore not all UK agreements with provision for FOA have compulsory mechanisms for triggering its use. A number of the agreements provide for FOA by joint, voluntary reference. Both sides must agree to refer a dispute to the arbitration stage.

Another consequence of the different UK context of FOA is that the voluntary nature of the UK agreements means that there is scope to address an issue not examined in the US, namely the factors influencing the incidence of FOA agreements. There is no particular theory to test on this issue but the question of why certain plants have adopted FOA when others have not is intuitively interesting.

Note that whilst the distinction between disputes of interest and disputes of rights is of fundamental importance in the US, this is not so in Britain. Whilst in the US, FOA is primarily used for interest disputes, here most collective grievance procedures are designed for all collective disputes whether over the terms of a new contract or over the interpretation of the existing contract.

The remainder of the first part of the paper describes the sample and data employed in section 2, outlines the types of FOA procedure in place in section 3 and in section 4 assesses the question of FOA incidence. The impact of FOA is considered in part II. The theoretical predictions concerning its impact on impasse rates, incidence of industrial action and pay settlements are set out in section 5. And evidence on each respective issue is analysed in sections 6 to 8. Conclusions are presented in section 9.

2. The sample

The ad hoc way in which FOA agreements have sprung up in the UK and the relatively small number of plants with FOA, had implications concerning the selection of our sample. North American examples of FOA are almost wholly in the public sector, so studies there have been able to draw on secondary data, showing the outcomes and processes of public sector bargaining rounds, to test the effectiveness of different dispute resolution mechanisms. By contrast all recent UK cases of FOA agreements were signed in the 1980s, and are in private sector companies. Therefore most of our information is primary data which we collected. Furthermore because less than 40 UK workplaces actually have FOA agreements, all possible research sources and methods had to be used to illuminate the subject.

Information on the extent of FOA agreements in the UK is immersed in more general studies of new style, no-strike or single union deals signed in the 1980s. This helped to provide a ready made comparison group for the FOA plants. The non-FOA but no-strike or single union establishments are a convenient and appropriate control group for their counterparts with FOA, since this lessens the potential difficulty of confusing an FOA effect with a general cooperative industrial relations effect. It must be acknowledged, however, that because both the FOA and control groups comprise part of the new style consensual industrial relations sector this research may tell us little about the effect of putting FOA into old-style non-cooperative bargaining structures.

Plants with new style deals were identified from surveys which provide the names and some details of these companies.(2) In sum the reports record 27 different FOA agreements in a total of 84 new style deals. Trade unions involved in these deals were contacted to both clarify the status of existing agreements and supply details of any new agreements or any agreements not specified in the available reports. After trawling all the available sources our final list consisted of 101 different workplaces having either FOA or non-FOA new style deals.

Data gathering was structured into two phases. First, a postal survey of the 101 workplaces was undertaken to obtain substantive information on plant characteristics, the impasse procedure and three impact measures: use of impasse procedure; incidence of industrial action; and pay settlements, over the preceding five years. Second, we undertook long detailed interviews at eight of the FOA plants where an impasse was reached in at least one year, to try to obtain more detailed empirical and attitudinal information about FOA effectiveness generally and in particular bargaining rounds in these workplaces.

The postal survey examining the last five annual bargaining rounds was sent in February 1991 to our 101 plants. Of the 75 plants which responded, three did not actually recognise a trade union for bargaining and therefore must have either derecognised a union or were incorrectly identified in the trawl of sources. These three responses were therefore rejected from the sample on the grounds that collective bargaining over pay did not actually take place.(3)

The basic characteristics of the 72 valid respondents are summarised in Table 1. The sample shows a preponderance of plants which employ more than 250 workers, have high union density, are either UK or Japanese owned, operate in the electronics or engineering industries and were established on their present site in the 1980s. Forty-four of the plants have provision for arbitration, of which 27 provide for FOA and 17 have conventional arbitration arrangements. Two data sets were constructed from the responses: one comprising the 72 plants which recognise a union for bargaining; and one in which each bargaining round is a separate case. The latter file of 297 bargaining rounds (some respondents did not supply a full five years data) contains information on pay settlements, use of disputes procedure and industrial action.

3. Impasse Procedures

We have information from 72 workplaces with new style procedures. Of these 54 have some form of impasse procedure. These are set out in Table 2 and Figure 1. It is readily apparent that the voluntary nature of the British new style agreements have thrown up a rich array of configurations of pre-arbitration conciliation and mediation stages and types of arbitration. There is further heterogeneity in the reference to arbitration mechanism, but the arrangements for selecting the arbitrator are more standardised.
Table 1. Characteristics of the sample

Characteristics N %

1. Employment by size band

25 1 1.4
25-99 8 11.1
100-249 26 36.1
250-499 13 18.1
500+ 24 33.3

2. Country of company HQ

UK 25 34.7
Other EC 7 9.7
Europe non-EC 3 4.2
Japan 24 33.3
USA 12 16.7
Other 1 1.4

3. Union recognised

AEU 18 25.0
EETPU 36 50.0
ISTC 1 1.4
GMB 5 6.9
TGWU 8 11.1
Multi-union 4 5.6

4. Union density

0-24% 8 11.1
25-49% 12 16.7
50-74% 31 43.1
75-100% 18 25.0
Don't know 3 4.2

5. Location by region

North 12 16.7
Yorks and Humberside 5 6.9
East Anglia 2 2.8
South East 11 15.3
South West 2 2.8
Wales 28 38.9
Scotland 1 1.4
Northern Ireland 3 4.2

6. Industrial sector

Metals 4 5.6
Mechanical engineering 9 12.5
Electronics 31 43.1
Vehicles 4 5.6
Food & drink 6 8.3
Plastics 8 11.1
Other 10 13.8

7. Plant vintage: date established

Pre-1980 24 33.3
1980-85 15 20.8
1986 8 11.1
1987 6 8.3
1988 9 12.5
1989 7 9.7
1990 3 4.2

TOTAL 72 100.0

Source: LSE 'New Style' Industrial Relations Survey (see
section 2).

Notes: AEU--Amalgamated Engineering Union, EETPU--Electrical,
Engineering, Telecommunication and Plumbing Union (both now
merged into AEEU), ISTC--Iron and Steel Trades Confederation,
GMB--General, Municipal, Boilermakers and Allied Trades Union,
TGWU--Transport and General Workers Union.


Of the 54 workplaces with impasse procedures half have FOA, and of these 27 plants, just over half also have provision for either conciliation alone (11) or both mediation and conciliation (3) before a dispute is referred to arbitration, but 13 of the agreements provide for a move straight from failure to agree to arbitration. These two types of FOA agreement reflect the debate about whether it is better to have FOA unaccompanied by pre-arbitration procedures, to deter disputes (Feuille 1975), or to have pre-arbitration procedures, to prevent a large gap between final offers and therefore an unworkable award if FOA is reached (Gallagher and Pegnetter 1979). Interestingly the very first no-strike deal (at Toshiba) was pure FOA with no pre-arbitration procedures. Subsequently this purity has decayed. The effect of pre-arbitration procedures on the competence of FOA to deter disputes is examined in section 6.

The provision of conciliation is consistent with the general picture of dispute resolution in Britain where about two-thirds of agreements which provide for intervention by the statutorily established Advisory, Conciliation and Arbitration Service (ACAS), envisage both conciliation and arbitration (Millward and Stevens 1986: 190). Furthermore, the EETPU model collective agreement, used as the basis for many of these new style deals, specifically favours a penultimate stage of conciliation by ACAS, in the collective disputes procedure (the relevant sections of this agreement are reproduced in the Appendix).

A greater proportion of the conventional arbitration agreements have provision for pre-arbitration stages (77 per cent). Finally, there are ten agreements which, whilst not having any arbitration stage, instead have mediation only (2), conciliation only (5) or both conciliation and mediation (3).

Concentrating on the 44 plants with arbitration, perhaps the most interesting finding is that over a third provide for joint, voluntary reference to arbitration. The proportion of such reference procedures among the FOA group (41 per cent) is, surprisingly, even higher than those with conventional arbitration (24 per cent). This certainly adds some weight to Lewis's contention that FOA in the UK is not simply a transplanted version of the North American example (Lewis 1990). Joint, voluntary reference procedures are certainly alien to arbitration in North America which is overwhelmingly compulsory, but fit in with the general pattern of arbitration procedures in the UK. The ACAS guidelines on arbitration state that arbitration will usually only proceed with the express consent of both parties (Lewis 1990: 42).

Joint reference procedures raise practical questions such as what happens if one party will not agree to arbitration. But perhaps more interesting is what such a procedure tells us about the parties and what effect such a procedure will have on collective bargaining and the rate of impasse. It seems plausible to argue that a joint reference procedure implies that the parties do not really understand the theory of impasse deterrence behind FOA, because they are keeping open an option of vetoing its use, thereby lessening its threat effect.

Finally a substantial majority of all workplaces with arbitration and especially the FOA plants allow ACAS to choose the arbitrator. The fact that UK FOA agreements do not in general provide for any more elaborate means of choosing the arbitrator, unlike in North America (Lester 1984), is probably because ACAS is the generally accepted body for resolving industrial disputes. It is no doubt also a function of the fact that FOA is still quite novel in the UK. The topic of arbitrator behaviour and therefore alternative means of choosing the arbitrator has simply not arisen.

4. The incidence of FOA

Final-offer arbitration procedures exist in 27 (38 per cent) of our 72 workplaces with 'new style' industrial arrangements which recognise trade unions. The characteristics of workplaces with and without FOA is not something that has previously been studied. This is because there has been no information to analyse the issue in Britain. And in the US FOA procedures are typically introduced by legislation, in contrast to their voluntary provision in the UK.

Cross tabulations reveal that FOA impasse procedures are more likely to exist in workplaces: which are UK or Japanese owned; where the EETPU is the recognised union; which are located in the South East; where the impasse procedure was introduced between 1980 and 1986; in medium sized workplaces; and where union density is above 25 per cent.

The likelihood of a workplace having an FOA procedure was analysed more formally by a logit regression using the sub-sample of plants which have some sort of procedure (Table 3). We classified the dependent variable as one if an FOA agreement and zero otherwise. Therefore the non-FOA category includes both conventional arbitration and non arbitration procedures. Only three variables are significantly associated with the incidence of FOA agreements. These are the date when the impasse procedure was introduced, whether or not the EETPU has sole recognition at the plant and whether or not the workplace is located in the South East region. The probability of the workplace having an FOA agreement is significantly much higher where the collective agreement incorporating FOA was signed in the period 1980 to 1986, has the EETPU as co-signatory or if it is in the South East.(4)

This statistical result concerning the year the disputes procedure was introduced and the role of the EETPU can be brought to life by discussing a personality involved. Between 1980 and 1986 the electricians union National Officer for Engineering was Roy Sanderson. He was (and remains) a tireless advocate of new style industrial relations in general and FOA in particular.(5) The EETPU model agreement, the basis for many of their new style deals allows a choice of FOA, conventional arbitration or industrial action as the ultimate stage in the collective grievances procedure (see the Appendix). So there is certainly some scope for local EETPU officials and management to independently choose their preferred final stage when negotiating the terms of the collective agreement. Yet a substantial majority chose FOA.

It would seem therefore that it is the energy and beliefs of one man that holds the key to the introduction of FOA procedures in the first part of the 1980s in the UK. A majority of the managers we interviewed stated that they introduced FOA having 'heard about it' in other companies' agreements. Roy Sanderson seems to have sown the seeds of FOA with the Toshiba deal in the early 1980s, and his zeal coupled with word of mouth and other management information sources seem to have been the key to its growth up to 1986. Obviously seed can only take root on fertile ground. Contingent factors promoting fertility probably included the assertion of managerial prerogatives made possible by inter alia the new labour legislation and high regional unemployment.

II. The impact of FOA in Great Britain

5. Theory

Most of the theoretical work on the impact of FOA stems from a seminal article by Stevens (1966). His main prediction was that under a compulsory arbitration arrangement fewer impasses would result if the procedure did not allow the arbitrator to produce a compromise award, i.e. FOA. Under FOA, because the parties are uncertain about the arbitrator's behaviour if the dispute goes to arbitration, a contract zone of more favourable and negotiatable solutions to the dispute is created. Stevens contrasted this 'incentive to settle' with the 'chilling effect' of conventional arbitration. In conventional arbitration regimes, because the parties believe that the arbitrator will compromise between their final offers, they are encouraged to maintain an extreme position in the face of arbitration, actually increasing the probability of an impasse. Therefore the most important test of the impact of FOA is that, all else equal, the impasse rate should be lower in workplaces where FOA is available compared to that in workplaces with conventional arbitration.
Table 3. The incidence of FOA in new style deals

Independent variables Logit coefficient Standard error

UK 1.156 1.177
Japan 0.130 1.287

Union recognised
EETPU 2.349(*) 1.276

Time procedure introduced
1980-6 5.022(***) 1.681

Interaction: EETPU, 1980-6 -2.880 2.064

South East 2.008(*) 1.097

Workplace size
Less than 100 -0.893 1.582
100-499 -0.819 1.120

Union density %
0-24 0.379 1.928
25-74 0.416 1.234

Constant -3.067(**) 1.475

Sample size 54
Chi-squared (df) 30.21 (10)
Pseudo |R.sup.2~ 0.404

Significant at *** 1%, ** 5%, * 10%. Notes: (i) Sample = 54
plants with some form of dispute procedure. (ii) Dependent
variable: FOA procedure available in workplace agreement, Yes
(1) No (0), Mean = 0.5. (iii) The defaults (excluded
components) are: Non-UK or Japanese owned; EETPU not
recognised; procedure introduced before 1980 or after 1986;
rest of UK; plant size of 500 or more employees; and 75-100 per
cent union density.

There have been no explicit theoretical assessments of the impact of FOA on industrial action, because action is legally prohibited in the majority of American FOA jurisdictions. As this is not the case for the UK, it seems pertinent to examine relative industrial action rates under FOA and conventional procedures as a further test of FOA's impact.

One way of looking at this is through asymmetric information theories of strikes. These predict that industrial action is more likely when there are asymmetries of information between the parties about bargaining positions and strengths (Kennan 1986). One consequence of the chilling effect of conventional arbitration may be that information is withheld, in contrast to FOA's promotion of 'genuine' negotiations, which in theory engenders an open information approach. If asymmetries of information are greater under conventional arbitration than FOA, then industrial action may also be more likely under conventional arbitration.

Three issues are examined concerning the impact of FOA on pay settlements. First, does FOA availability have a 'strike like' effect on pay settlements? One criterion to evaluate FOA regimes concerns bias. FOA is designed to achieve--without industrial action being threatened or taken--the same rate of settlements as would exist under conventional bargaining with a strike threat. This is examined by comparing pay settlements under FOA regimes to those without any impasse procedure.

Second, in bargaining rounds where an impasse procedure exists, is there a significant difference between pay settlements where an FOA regime is in place and those where conventional arbitration prevails? Farber (1980) has suggested that FOA skews the distribution of settlements against the more risk averse party. If North American evidence (Ashenfelter and Bloom 1983) on the risk preferences of management and unions also holds in the UK we should expect to find that negotiated settlements under the FOA regime are lower than under conventional impasse procedures because unions are more risk averse than management.

Third, where FOA procedures exist, do negotiated settlements differ from arbitrated settlements? If the parties are to have confidence in FOA there should be no consistent evidence of management or unions 'winning' at arbitration. We examine this issue by comparing the pay outcomes under negotiation and arbitration.

It has to be recognised that there is a conceptual problem in what follows. FOA procedures are chosen by management. It is possible therefore that FOA is not a truly exogenous factor influencing impasse rates, industrial action and pay settlements. Rather FOA procedures, dispute rates and pay settlements may be all jointly determined by some other factor. However, this conceptual problem may not matter too much in practice. Two-thirds of FOA procedures were introduced prior to 1987. Therefore our information on impasses, industrial action and pay is not contemporaneous with the introduction of FOA. Certainly it would seem reasonable that where agreements were signed before this date they can be considered exogenous to the determination of industrial relations outcomes during the survey period.

6. Use of impasse procedures

FOA is specifically designed to promote in-house resolution of disputes, and therefore to prevent impasses (or failures to agree), in direct contrast to conventional arbitration and mediation/conciliation which in theory are more likely to encourage disputes. Therefore the most commonly used and important test of the impact of FOA is relative impasse rates with and without FOA.

The variety of dispute procedures within our sample provide both complexity and depth for our analysis of impasse rates. The taxonomy of procedures is repeated in flow chart form for impasse rates in Figure 2. Figure 2 shows that of the 212 wage rounds where a dispute procedure was available, 27 went to impasse, an overall impasse rate of 12.7 per cent. This compares with national private sector impasse rates of 12 per cent and 17 per cent, for 1980 and 1984 respectively, for pay and conditions disputes (Millward and Stevens 1986: Table 7.5).(6)

Figure 2 also illustrates the complexity within the impasse rate measure. Of the 27 impasses, 10 involved rounds where there was no provision for arbitration, 6 where there was no provision for mediation/conciliation, and 11 where there was provision for both. Of these 11, only four involved the use of arbitration as the other 7 were settled either during or after pre-arbitration meetings. This complexity provides five different impasse rates on which to test the impact of FOA:

1. Overall impasse rate: Using all 212 rounds and not distinguishing the degree of impasse.

2. Overall impasse rate within arbitration available rounds: A sub-sample of 170 rounds (b), looking at all 17 impasses in these rounds (f+g).

3. The use of arbitration within arbitration available rounds: The same sub-sample as (2), but looking at the 10 impasses involving the use of arbitration (g+j).

4. Overall impasse rate in the mediation/conciliation available rounds: Sub-sample of 144 rounds (c+d), looking at all 21 impasses (f+h).

5. Use of mediation/conciliation only: Sub-sample of 140 rounds, i.e. same as (4) but excluding the 4 rounds where arbitration is also used, and looking at remaining 17 impasses (h+i).

The hypothesis for each model is that bargaining rounds where FOA is available should be associated with a lower impasse rate, all else equal.

It will be seen from Table 4 that the availability of FOA is indeed associated with a lower rate of impasse on four out of five of the measures. However this impasse- reducing impact of FOA is only statistically significant for two of the measures (4 and 5).

The availability of FOA is associated with a significantly lower impasse rate in rounds where mediation and/or conciliation is also provided (models 4 and 5). It is not surprising that FOA deters disputes more effectively compared to where no arbitration provision exists (in 42 of the comparative rounds, i.e. comparing box d to box c). But it must be emphasised that it is FOA which is the TABULAR DATA OMITTED key element here, because it performs better than other measures of arbitration. When the model is repeated using all arbitration or conventional arbitration (coupled with mediation and/or conciliation) neither of these measures lower the impasse rate.

The results therefore support the idea that it is the interaction of FOA with mediation and/or conciliation that deters disputes. This does not mean that an FOA procedure not accompanied by pre-arbitration stages will be made more effective by the inclusion of pre-arbitration stages, but rather that an unaccompanied mediation/conciliation procedure will be made more effective through the addition of FOA than of conventional arbitration. We conclude this because using only the 114 rounds where FOA was available (not reported in the table) did not indicate that either mediation and/or conciliation had any impact on the impasse rate. It seems safe to conclude therefore that in this sample the existence of FOA in a multi-stage procedure deters impasses more effectively than conventional arbitration.

One question arising from these results is how to explain the apparent failure of FOA to significantly deter disputes more effectively than non-FOA procedures (models 1 and 2), and in particular its failure to deter the use of arbitration compared to conventional arbitration (model 3). We investigated this lower than expected impasse deterrence effect of FOA at the interview stage of the research. Evidence from these individual cases of FOA use suggests that three explanations are particularly important.

First, we are unable to properly control for the bargaining history of the parties. A number of the managers indicated that before introducing the agreement which incorporated FOA, the plant had quite dispute-prone industrial relations. Therefore although FOA had been used, it had accompanied a reduction in the level of disputes compared with the previous non-FOA regime. Unfortunately it is not possible to control for these bargaining histories because a large number of the plants in the sample, the greenfield sites, have no such bargaining history.

Second, recent experimental work by Babcock and others (1991) suggests that one cause of disputes can be over optimistic bargainers--over optimistic about their chances of winning a trial/hearing. This notionally simple concept is very hard to test except in the laboratory because the level of optimism and whether or not it is unrealistic are both extremely difficult to assess. Having said this in two cases of FOA use managers said that the employer's side was very confident of winning the award, yet lost. The risk of losing was simply not considered a possibility. Therefore, in these two cases, the either-or mechanism of FOA provided no greater incentive to settle the disputes without arbitration.

The effect of incentives and sanctions created by principal/agent situations are a third factor. They were examined with reference to strikes by Ashenfelter and Johnson (1969) and have since been applied to arbitration by McCall (1990). In one dispute from our sample there was a principal/agent problem on management side and in two others on the employees' side. One manager stuck to an offer which he knew to be unreasonable and which would be rejected in arbitration in order to appear tough to higher level managers at company headquarters. In one of the examples involving the employees, the union members had threatened to join another union if the local official did not pursue their claim. Therefore the official had to take the claim all the way to arbitration even though it was against national union policy and he was fairly sure of losing the award. These cases illustrate that disputes may get to FOA even if one of the parties has no expectation of winning.

Our control variables included various characteristics of the plants and the impasse resolution systems. No control variable was consistently significant in each model. But there is tentative evidence that some impasse rates are higher in UK owned plants, where multiple unions are recognised, in plants established before 1980 and where there are 250 employees or more. The evidence also hints that where arbitration exists the availability of conciliation may lower the impasse rate (model 3).(7)

7. Industrial action

Industrial action occurred in only 5 out of 297 bargaining rounds (1.68 per cent) covered by our sample. Given this lack of industrial action--just two strikes and three overtime bans--it is pointless and impossible to try to distinguish (for this sample) whether or not collective agreements incorporating no strike or no disruption clauses yield lower levels of industrial action than new style deals not containing such clauses. Still less is it possible to take a view on whether FOA is superior to conventional arbitration in minimising or eliminating industrial action.

However, it is possible to compare the extent of industrial action in this sample with analogous information covering more traditional industrial relations. The Confederation of British Industry (the employers' organisation) monitors annual pay settlements in over 1000 bargaining groups with collective bargaining in manufacturing industry. During 1985-9 industrial action took place in 4.6 per cent of these settlements (Ingram, Metcalf and Wadsworth 1991).(8) Thus the incidence of industrial action in workplaces with new style industrial relations arrangements is only a little over a third of that in more traditional collective agreements. There are two possible reasons for this lower incidence of industrial action. The characteristics of the two groups may be different and/or the impasse procedures in the newer agreements may have an impact.

We know, for example, that the incidence of industrial action is positively related to the size of the workplace (Prais 1978), multiple union recognition (Machin et al. 1993) and the workplace being a subsidiary of a larger organisation (Blanchflower and Cubbin 1986). Our sample has, on the one hand, a higher fraction of larger workplaces and subsidiaries than corresponding national samples such as the CBI Pay Databank or WIRS 1980 and 1984, so these characteristics are certainly not responsible for the low incidence of industrial action. On the other hand our sample has a very low rate of multiunionism. Only 5.6 per cent of our workplaces have multiple unions compared with a national level of 35 per cent in manufacturing in 1984 (Millward and Stevens 1986: Table 3.7).

Thus part of the explanation for the extraordinarily low rate of industrial action in our sample of workplaces may reflect the fact that new style deals typically involve single union recognition. But the procedures themselves have surely played a part too. The impasse rate in our sample is not significantly lower than the national impasse rate in more traditional collective agreements found in WIRS 1984 (Millward and Stevens 1986: Table 7.5), see section 6. Therefore it is plausible that some of the disputes that might otherwise have resulted in industrial action were instead resolved by the impasse procedures built into the new style industrial relations collective agreements.

Some of the plants where interviews took place were brownfield sites when the FOA agreement was signed. In all these the managers stated that there had been problems with industrial action at the plant before the agreement was signed. However, not all interviewees stated that the FOA procedure was the crucial element in the decline or lack of industrial action. Only two managers categorically claimed that had FOA not been available, there would have been a strike over the issue in dispute. Other managers said that either the dispute which involved FOA was not sufficiently important to have resulted in industrial action or that other aspects of their industrial relations arrangements were more important explanations, than the FOA procedure, for the lack of industrial action. Most often cited was management's consensual style of operations, but other factors included the non-strike clause and the impact of the law on industrial action.

8. Pay settlements

Nominal pay settlements were recorded for 258 annual bargaining rounds over 5 years. They were translated into real settlements by deflating by the annual rise in the retail price index at the time of the settlement. The range of real settlements was -4.7 per cent per annum to +8.2 per cent per annum.


To recap there are three issues here. Does FOA availability bias settlements compared to a no impasse procedure environment? Does the availability of FOA produce lower negotiated settlements than conventional procedures? Under FOA, do negotiated settlements differ from arbitrated settlements?

These issues are analysed in Table 5. We have information on 161 annual bargaining rounds incorporating FOA or with no impasse procedure. The mean real increase in these 161 negotiated settlements was 1.71 per cent per annum. There is no significant difference between settlements under an FOA regime and those permitting a strike threat: the coefficient on FOA in column 1 is not significantly different from zero. This suggests that FOA in the UK is unbiased. After controlling for the characteristics of the workplace and workforce, negotiated pay settlements are similar in the FOA workplaces to those plants with no impasse procedure.

The second issue concerns negotiated settlements under different types of arbitration practice. Our survey covered 149 bargaining rounds where either FOA or conventional procedures were available but not used. The mean real wage increase in such settlements was 1.92 per cent per annum. The results provide confirmation of Farber's conjectures. Settlements in workplaces with FOA are, on average, 0.812 percentage points lower than those with conventional procedures. This implies that, on average, the availability of conventional procedures yielded a pay rise of 2.3 per cent per annum while settlements in workplaces with FOA only averaged 1.5 per cent per annum.

If the mechanism behind such a result is the parties' relative risk preferences as Farber argues, then the ranking of FOA settlements below conventional procedure settlements could be reversed if relative risk preferences change.(9) Interestingly, not one of the managers interviewed believed that FOA availability had had any impact on pay settlements at their plant. When presented with this evidence, most managers still argued that other factors (such as single status or management style) determined the result but one did rationalise the result along the lines of FOA making both sides behave more reasonably in negotiations.

Where FOA exists the real pay settlements averaged 1.56 per cent per annum for the 92 rounds for which we have information. Arbitrated settlements yielded very slightly higher pay rises than negotiated settlements (the coefficient is 0.117) but the difference between the two is not statistically significant. This is a reassuring result. When FOA is invoked, the average level of settlement is much the same as when the parties reach agreement via negotiation. Thus there is no evidence of either management or unions consistently winning when FOA is used.

This raises an important methodological point. Some commentators have calculated 'win-loss' tallies for management and unions under FOA regimes (Feuille and Schwochau 1988). But as Ashenfelter and Bloom (1983) point out, if unions win on more occasions than management this may simply reflect their more reasonable final offers. We have provided an alternative way of looking at the issue. After controlling for other influences on pay settlements we find the outcomes of negotiation and arbitration to be similar, and therefore no evidence that on average one party gains at the expense of the other when FOA is invoked.

The interviews provide a gloss on this result. In two cases of FOA use, the union claim was amended quite drastically immediately before the arbitration hearing. This is direct evidence of unions compromising more readily than management, coming to arbitration with more reasonable final offers and in one case 'winning' the award because of it, but with a pay rise not that much higher than the employer's final offer. Further, although the arbitrated and negotiated settlements are similar, some of the cases of FOA being invoked in the sample turned out not to involve the level of the pay settlement but related factors such as grading and overtime hours.

Pay settlements were also associated with some of the control variables. On average (in at least one regression), settlements were lower in plants established before 1980, in plants not owned by a Japanese parent company, where the workplace is a subsidiary of a larger organisation, in 1986, 1989 and 1990, in Yorkshire and Humberside, the south east and south west.

9. Conclusions

Final-offer arbitration has emerged in the 1980s in a great variety of formats and situations in the UK. There has been no standard formula to which even the majority of agreements have adhered. The picture is one of considerable diversity within a small group of plants.

The voluntary nature of UK final-offer arbitration agreements has created an interesting issue in itself: what has determined the incidence of these agreements so far in the UK? One particular figure in the union movement, Roy Sanderson of the EETPU, was prominent but managers have also played an important part in the spread of the procedure.

Two results stand out concerning the impact of final-offer arbitration on workplace outcomes. First, FOA seems to out-perform conventional arbitration in deterring disputes when it is coupled with conciliation and/or mediation. Second, the availability of FOA compared with conventional procedures may have made union negotiators compromise more readily in pay negotiations, as predicted by one aspect of the relevant theory.

A fuller picture of the impact of final-offer arbitration may only be made clear in another five or more years' time, when more workplaces will have had greater experience with this relatively novel type of procedure.


(1) Note that the EETPU is no longer, having merged with the AEU to form the Amalgamated Engineering Electrical Union (AEEU) on 1 May 1992.

(2) Income Data Services (IDS) 1987, Trade Union Congress (TUC) 1988, Industrial Relations Review and Report (IRRR) 1989, and Labour Research (LR) 1989. These reports do not all concur on the details of the agreements. The number of agreements listed and discrepancies in the number of FOA deals is indicated in Milner (1992).

(3) See Milner (1992) for further discussion of the sample. The non-respondents were not biased in a way which discredits the representativeness of the sample.

(4) The South East and EETPU variables are only significant at the 10 per cent level. All other regional variables were included in alternative versions of the logit equation but none were significant. Combining the South East with other regions to produce a South of England variable also failed to provide a significant result.

(5) Bassett (1986) states that Sanderson got the idea of FOA from a convention of US arbitrators which he attended in the late-1970s. Sanderson (interviewed by us) states that although he had heard of FOA from the US, he adapted it considerably before including it in the package offered to Toshiba at Plymouth in 1981, the first new style deal.

(6) However, the two measures are not quite compatible since the national figures include interventions which would not figure in our sample, for example by senior management. Also our sample covers just manufacturing plants whereas the WIRS figures include the private service sector.

(7) See Milner (1993) for a fuller discussion of dispute theories and the control variables used to test them.

(8) Note that the rate in electronics at 7.5 per cent (1986-9) was even higher than the manufacturing average, so the sample's concentration in this industry does not seem to be a feasible explanation for the low rate of industrial action.

(9) FOA may well, on the basis of these results, sow the seeds of its own destruction in a voluntary system. A union leader, knowing that FOA may lead him or her to concede lower pay settlements, would presumably, not agree to its installation.


Extract from EETPU Model Memorandum of Agreement

Appendix V.3: Collective Grievance Procedure

1) General

Where an issue concerns a group of employees, the procedure for raising the grievance will be as follows:-

i) Stage One

The item will be raised on the agenda of the Advisory Board by any Board member or the Company, and a Board recommendation sought. If no resolution is found at Stage One, Stage Two will be implemented. Where it is mutually agreed to discuss an item without initial reference to the Advisory Board, for example, matters of interpretation of agreements, the item will be referred directly to Stage Two of the procedure.

ii) Stage Two

If the matter is not resolved after full consideration of the Board recommendation by both the Management and Union Members, it will be formally referred for resolution at this Stage, between the Company and internal Union representatives.

iii) Stage Three

If Stage Two fails to bring about a settlement, the local Full-time official will become involved and then the National Officer of the Union.

iv) Stage Four

If the matter is not resolved after every effort has been made by the Company and the Union officials, it will be jointly referred to ACAS for Conciliation. This joint reference will be based on a mutual understanding that such a reference can assist in the process of jointly finding a self-solution.

v) Stage Five

If Stage Four fails to produce a mutually satisfactory resolution, then either side can proceed to Arbitration. The reference will be submitted to ACAS for the appointment of a mutually agreed independent Arbitrator. The Arbitrator will decide the issue on a 'Pendulum Arbitration' basis. The brief to the Arbitrator will be to fully investigate evidence form both parties, relevant to the issue, and without compromise, to find for one party or the other. The Arbitrator's decision will be binding on both parties and will represent the final solution to the issue.

OR Stage Five Alternative (1)

If Stage Four fails to produce a mutually satisfactory resolution, then either side can proceed to Arbitration. The reference will be submitted to ACAS for the appointment of a mutually agreed independent Arbitrator to consider in full the relevant evidence submitted by both parties in order to find a solution. The Arbitrator's decision will be final and binding on both parties and will represent the final solution to the issue.

OR Stage Five Alternative (2)

If Stage Four fails to produce a mutually satisfactory resolution, the Union may recommend a withdrawal of labour. Any withdrawal of labour will only take place once the statutory ballot requirements have been observed and the action endorsed by the Union membership.

Source: EETPU.


Ashenfelter, O. (1985), 'Evidence on US experiences with dispute resolution systems', Industrial Relations Section, Princeton University, Working Paper No. 185.

Ashenfelter, O. and Bloom, D. (1983), 'The pitfalls in judging arbitrator impartiality by win-loss tallies under final-offer arbitration', Labor Law Journal, 34, pp. 534-39.

Ashenfelter, O. and Johnson, G. (1969), 'Bargaining theory, trade unions and industrial strike activity', American Economic Review, 59.1, pp. 35-49.

Babcock, L., Camerer, C., Issacharoff, S., and Loewenstein, G. (1991), 'Self-serving assessments of fairness and bargaining impasse', Carnegie Mellon University, mimeo.

Bassett, P. (1986), Strike Free: New Industrial Relations in Britain, London: MacMillan.

Blanchflower, D. and Cubbin, J. (1986), 'Strike propensities at the British workplace', Oxford Bulletin of Economics and Statistics, 48.1, pp. 19-39.

Cuthbert, N.H. (1960), The Lace Makers' Society: A Study of Trade Unionism in the British Lace Industry, 1760-1960, Nottingham: Amalgamated Society of Operative Lace Makers and Auxiliary Workers.

Farber, H.S. (1980), 'An analysis of final-offer arbitration', Journal of Conflict Resolution, 24.4, pp. 683-705.

Feuille, P. (1975), 'Final offer arbitration and the chilling effect', Industrial Relations, 14.3, pp. 302-310.

Feuille, P. and Schwochau, S. (1988), 'The decisions of interest arbitrators', The Arbitration Journal, 43.1, pp. 28-35.

Gallagher, D.G. and Pegnetter, R. (1979), 'Impasse resolution under the Iowa multistep procedure', Industrial and Labor Relations Review, 32.3, pp. 327-338.

Gennard, J. (1988), 'How pendulum arbitration agreements have worked in practice', University of Strathclyde, mimeo.

Income Data Services (IDS) (1987), 'Single union deals and dispute procedures', IDS Study, no. 509.

Industrial Relations Review and Report (IRRR) (1981), 'Single union deals', no. 442, 27 June.

Ingram, P., Metcalf, D. and Wadsworth, J. (1991), 'Strike incidence and duration in British manufacturing industry in the 1980s', Centre for Economic Performance, LSE, Discussion Paper No. 48.

Kennan, J. (1986), 'The economics of strikes', in Ashenfelter, O. and Layard, R. (eds.), Handbook of Labor Economics, Vol. II Amsterdam and New York: North Holland, Elsievr Publishers.

Kessler, S. (1987), 'The swings and roundabouts of pendulum arbitration', Personnel Management, December, pp. 38-42.

Labour Research Department (1989), 'Are single-union deals on their way out?', Labour Research, November, pp. 9-11.

Lester, R.A. (1984), Labor Arbitration in State and Local Government, Princeton: Princeton University.

Lewis, R. (1990), 'Strike-free deals and pendulum arbitration', British Journal of Industrial Relations, 28.1, pp. 32-56.

Machin, S., Stewart, M. and Van Reenen, J. (1993), 'Multiple unionism, fragmented bargaining and economic outcomes in unionised UK establishments', in Metcalf, D. and Milner, S. (1993), New Perspectives on Industrial Disputes, London: Routledge.

McCall, B.P. (1990), 'Interest arbitration and the incentive to bargain: A principal-agent approach', Journal of Conflict Resolution, 34.1, pp. 151-167.

Millward, N. and Stevens, M. (1986), British Workplace Industrial Relations 1980-1984: The DE/ESRC/PSI/ACAS Surveys, Aldershot: Gower.

Milner, S. (1992), 'Final-offer arbitration in the UK: Incidence, processes and outcomes', Department of Employment, Research Paper, forthcoming.

Milner, S. (1993), 'Dispute deterrence: evidence on final-offer arbitration', in Metcalf, D. and Milner, S. (1993), New Perspectives on Industrial Disputes, London: Routledge.

Prais, S.J. (1978), 'The strike-proneness of large plants in Britain', Journal of the Royal Statistical Society, Series A, 141.3, pp. 368-384.

Rico, L. (1987), 'The new industrial relations: British electricians' new-style deals', Industrial and Labor Relations Review, 41.1, pp. 63-78.

Stevens, C.M. (1966), 'Is compulsory arbitration compatible with bargaining?', Industrial Relations, 5, pp. 38-52.

Trades Union Congress (TUC) (1988), Meeting the Challenge; First Report of the Special Review Body, London: TUC.

Treble, J.G. (1986), 'How new is final offer arbitration', Industrial Relations, 25.1, pp. 92-94.

Treble, J.G. (1990), 'The pit and the pendulum: Arbitration in the British coal industry, 1893-1914', The Economic Journal, 100, pp. 1095-1108.

Wood, Sir J. (1985), 'Last offer arbitration', British Journal of Industrial Relations, 23.3, pp. 415-424.
COPYRIGHT 1992 National Institute of Economic and Social Research
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1992 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Metcalf, David; Milner, Simon
Publication:National Institute Economic Review
Date:Nov 1, 1992
Previous Article:Comparative productivity in British and American manufacturing.
Next Article:Has the North-South divide come to an end? - Prospects for regional unemployment.

Terms of use | Copyright © 2017 Farlex, Inc. | Feedback | For webmasters