The contents of partnership agreements in Britain 1990-2007

This article assesses the contents of the majority of employer-union partnership agreements signed in Britain from 1990 to 2007. Few agreements contain the expected partnership principles and most express modest overall aims and limited ambition. Typical agreements are substantively hollow with more than one-third containing no substantive provisions; and procedurally biased, with more than four-fifths offering unions greater involvement in employers’ decisions. Partnership agreements continue the procedural bias of traditional British collective agreements representing the lowest common denominator of agreement — unions work towards the success of the enterprise in return for involvement in employers’ decisions. The implications for New Labour’s Third Way approach and the policy of encouraging voluntary partnership agreements to promote fairness at work are considered.

The contents of partnership agreements in Britain 1990Britain -2007 Introduction A n important aspect of New Labour government policy in Britain since 1997 was to encourage employers and trade unions to sign partnership agreements that involved 'labor-management cooperation for mutual gains' (Kelly, 2004: 268). The theme of partnership formed an important part of New Labour's broader Third Way approach to industrial relations (Brown, 2000). This approach relied upon voluntary partnership agreements and represented a non-statutory approach to industrial relations as opposed to the statutory support for integrative relations found in the more coordinated market economies of Europe. New Labour's approach assumed that employers and unions would functionally adapt to more competitive economic circumstances without recourse to legislation (Heery, 2002). Subsequent debates evaluating partnership agreements, therefore, help to inform a wider assessment of New Labour's record in government and the potential of the Third Way approach to develop a progressive workplace consensus in Britain.
Latest estimates suggest New Labour's approach met with some success, with 248 partnership agreements signed between 1990 and 2007, covering almost ten percent of all British employees (Bacon and Samuel, 2009: 238). The implications of signing partnership agreements for trade unions and their members, however, have generated a lively debate between advocates and critics of partnership. Advocates argue that increased industrial relations cooperation may produce mutual gains. This outcome may occur because cooperation allows employers to develop efficient working practices that generate higher profits, and then share these gains with employees through improved wages and enhanced employment security (Kochan and Osterman, 1994). The positive impact of partnership may even extend further to help reverse the decline of British trade unions, if unions demonstrate their potential to make a positive contribution to organisational success (Terry, 2003). However, critics will have none of this. They argue, in contrast to the mutual gains view, that partnership agreements prioritise employer interests and employees will gain little (Kelly, 2004;Martínez Lucio and Stuart, 2005). Thus, critics expect partnership to accelerate the decline of British trade unions, highlighting a supposed inability to influence employers and failure to protect core terms and conditions of union members.
Given such contrasting claims, it is surprising that to date no studies have assessed the actual contents of a sizable and representative sample of partnership agreements in Britain. This omission is notable because partnership agreements are likely to express the terms on which employers and unions establish dialogue. This article partly addresses this gap by presenting the first comprehensive analysis of the contents of 126 partnership agreements; the largest sample of partnership agreements collected to date, more than half (51%) of all 248 agreements estimated to have been signed in Britain between 1990and 2007(Bacon and Samuel, 2009. The analysis shows that partnership agreements express modest aims and limited ambition, with many containing few partnership principles. These agreements are also substantively hollow, with more than one-third containing no substantive provisions; and are procedurally biased, with more than four-fifths offering unions greater involvement in employers' decisions.

The principles, provisions and balance of interests in partnership agreements
Advocates and critics of partnership have different views of the principles, provisions and balance of interests reflected in the contents of partnership agreements. In terms of principles, most commentators assume partnership agreements in Britain contain a set of shared values and beliefs that establish the behaviours, attitudes and expectations required for cooperative industrial relations. The most well known in Britain are the two sets of key partnership principles promoted by the employers' organisation, the Involvement and Participation Association (IPA, 1992; see Guest and Peccei, 2001), and the Trades Union Congress (TUC, 1997;see Martínez Lucio and Stuart, 2005;Terry, 2003). Both organisations have encouraged employers and unions to sign partnership agreements, and provided partnership principles to help them establish a shared understanding of what partnership should involve. Such principles underpin the substantive and procedural provisions contained in partnership agreements generally considered necessary for industrial relations cooperation and gain sharing.
However, it is not known how many partnership agreements in Britain contain these principles, and whether the substantive and procedural provisions of agreements reflect mutual interests. Indeed, the prevailing critical view suggests partnership agreements will reflect employers' interests given the contraction of collective bargaining and decline in trade union recognition since the 1980s. Dominant employers, it is argued, capture most, if not all, of any gains from such arrangements by avoiding disputes while seeking to restructure and reduce labour costs (Kelly, 2005). This view suggests that employer interests will dominate the contents of partnership agreements. This section reviews these issues.

Principles
Creating a partnership relationship requires strategic commitments from employers and unions to a shared set of aims (Guest and Peccei, 2001). According to Knell's (1999: i) early study of 15 British partnership cases, for example, organisations adopting a partnership approach shared a 'remarkable commonality of vision, value and purpose'. Common principles may also be expected to be expressed in formal partnership agreements because the IPA and the TUC widely publicised sets of principles for organisations developing partnership agreements. On this basis, advocates of partnership expect written partnership agreements to outline bold aims and high ambitions for partnership, including mutual gains, to establish the basis for a progressive workplace consensus.
Turning to these principles in detail, the IPA (1992) encourages employers and employee representatives to make three commitments to partnership values, covering 'the success of the enterprise; building trust through greater involvement; and respect for the legitimacy of other partners'. Four 'building blocks' should also be adopted, including commitments covering partnership processes -'informing and consulting staff about issues at workplace and company level; and the effective representation of people's views within the organisation' -and the expected outcomes from partnership -'recognition of employees' desire for security and the employers' need to maximise flexibility', and 'sharing success within the company' (IPA, 1992). The TUC (1997) advise employers and unions developing partnership agreements to sign up to six principles, covering values -'a commitment to the success of the organisation' and 'a recognition of and respect for the legitimate roles of the employer and the trade union', process -'openness and transparency', and expected outcomes -'a focus on the quality of working life', 'a commitment to employment security', and 'adding value to all concerned'.
It is not known, however, whether and how many partnership agreements include these IPA and TUC partnership principles. Indeed, there is some evidence that partnership agreements may be heterogeneous, reflecting specific organisational circumstances (cf. Johnstone et al., 2009;Oxenbridge and Brown, 2004). As partnership is an imprecise concept, agreements reached in the low-trust context of British industrial relations may include few of the ambitious principles of the IPA and TUC, and in reality they may 'represent the lowest common denominator of agreement' (Bacon and Storey, 2000: 410). This outcome may occur because corporate governance in Britain may deter employers from making promises they cannot keep (Deakin et al., 2002;Thompson, 2003). Critics of partnership also suggest that the contraction of collective bargaining and the decline in union recognition since the 1980s will result in partnership agreements reflecting employers' interests (Kelly, 2005). Partnership values are also believed to deny different class interests in the employment relationship, promote union moderation, and result in union concessions to employers' demands (Upchurch et al., 2008).
The first aim of this article, therefore, is to determine which, if any, IPA and TUC principles feature in British partnership agreements. It is important to establish whether employers and unions have made reciprocal commitments that may provide the basis for high-trust relations to support partnership, or whether employer interests predominate.

Provisions
The partnership principles discussed above are likely to shape the specific substantive and procedural provisions expressed in partnership agreements. It is claimed that partnership agreements extend joint regulation to address issues previously decided only by managers (Samuel, 2005;Wills, 2004). The most important substantive issue for unions is, arguably, an employer commitment to provide employment security, if not job security, often in exchange for increases in flexible working (so called 'flexicurity' arrangements) (Sisson et al., 1999). However, it is by no means clear that employers commit to providing employment security in partnership agreements. Thompson (2003) suggests increased market competition and shareholder pressure in liberal market economies will prevent employers from making meaningful substantive promises in partnership agreements because they will be unable to keep such promises. Evidence supporting this view includes Guest and Peccei's (2001: 224) finding that fewer than one-third of IPA member organisations made a formal commitment to employment security and fewer than one-quarter included a formal guarantee of no compulsory redundancies. This pattern is consistent with suggestions that employers prefer to make unilateral decisions rather than extend joint regulation to a broader range of issues.
As well as employment and job security, it is generally expected that partnership agreements will be substantively rich and feature provisions addressing workers' qualitative needs (Hyman, 1997). These provisions may include employee training and development, increased employee involvement, single status, fair treatment, and practices to improve work-life-balance (Haynes and Allen, 2001;Heery, 2002). Many of these provisions do not normally feature in traditional collective agreements (cf. Dunn and Wright, 1994). However, whether or not partnership agreements broaden the range of substantive issues subject to joint regulation remains an important issue to be addressed.
In addition, some commentators assume that, as well as being substantively rich, a prominent feature of partnership agreements is a procedural emphasis on joint consultation. The available evidence suggests this involves creating joint consultative committees to consider issues previously outside the traditional bargaining agenda (Marks et al., 1998;Samuel, 2007;Wills, 2004). Critics of partnership, however, suggest employers will not sign agreements with unions that limit their autonomy (Kelly, 2005). Few partnership agreements are likely, therefore, to provide the means for unions to influence management decisions in practice. Indeed, where joint consultation expands, critics suggest this should not be interpreted as extended joint governance because unions are merely coopted to support the employers' agenda.
Traditional collective agreements in Britain, signed during earlier periods of higher union density, are characterised by a procedural bias concerned with the distribution of power and authority, which attempts to prevent the parties defecting from their agreed roles (Dunn and Wright, 1994;Flanders, 1968). This bias reflects the low-trust relations typical in British industrial relations. Thus, it seems plausible that partnership agreements, being contemporary collective agreements, will be path dependent to a degree, displaying a similar procedural bias to traditional collective agreements rooted in low-trust relations. However, this possibility remains to be tested empirically. The second aim of this article, therefore, is to analyse the substantive scope and the procedural provisions expressed in British partnership agreements signed 1990-2007.

The balance of interests
The final question explored in this article is the extent to which partnership agreements in Britain contain expressions of mutual interests and whether this has changed over time. Kelly (2004) theorises partnership agreements as falling The contents of partnership agreements in Britain 1990-2007 Samuel and Bacon along a continuum of power. The distinction is made between 'employerdominant agreements marked by a balance of power favourable to the employer' at one end of the continuum; and at the other extreme, 'labour-parity agreements, where there is a more even balance of power ' (2004: 271, original emphasis). Based on Guest and Peccei's (2001) survey of IPA member organisations, it seems likely that most partnership agreements lie towards the employer dominant pole of this continuum and contain more provisions favouring employer interests than favouring those of labour.
Stronger criticisms of partnership go further, arguing that employers invariably dominate cooperative relationships in liberal market economies (Kelly, 2005). Thus few, if any, principles and provisions reflecting labour interests are expected to be found in British partnership agreements. Indeed, Thompson (2003) suggests this outcome is inevitable because employers cannot realistically keep promises to share financial gains with employees, or to provide employment security, let alone job security, given intensified product-market competition and shareholder pressure. However, such arguments play down potential variation in the relative balance of power between employers and unions. In certain instances, it seems likely that stronger unions may extract employer concessions before they sign partnership agreements.
It must be kept in mind therefore, that union strength could significantly affect the relative balance of interests expressed in partnership agreements. This point is particularly important when distinguishing between private and public sector partnership agreements in Britain (Bacon and Samuel, 2009). Private sector employers may sign partnership agreements merely to coerce weaker unions into legitimising managers' decisions (Kelly, 2005;Samuel, 2007). In contrast, stronger and potentially adversarial unions in the public sector may insist that partnership allows them to influence employers' plans to modernise public services (cf. Samuel, 2007). Thus, while public sector employers must consult with unions to facilitate reform, strong public sector unions seem unlikely to commit to unspecified substantive change without meaningful involvement in shaping reform plans. For reasons of relative union strength, therefore, it might be expected that the contents of public sector partnership agreements will reflect a relative balance of interests but will lack substantive scope given disagreements over the agenda to reform public services.
Once again, however, there is little empirical evidence with which to address these important issues. Thus, the final aim of this article is to evaluate the relative balance of interests in the contents of partnership agreements in Britain over time and by private and public sector.

Research method
The research analysed the contents of the largest sample of partnership agreements collected to date in Britain, involving 126 agreements drawn from an estimated total of 248 agreements signed between 1990 and 2007 (Bacon and Samuel, 2009). To gather as many partnership agreements as possible, both authors searched the internet for pronouncements of partnership deals between employers and independent trade unions or staff associations (for full details see Bacon and Samuel, 2009: 235-6). The authors then searched the websites of the IPA, the TUC, the former Department of Trade and Industry (including the now closed Partnership at Work fund) and all available British trade union websites. The next phase involved seeking confirmation and copies of partnership agreements from the employers and unions identified by this exhaustive search. Most provided copies for analysis.
The analysis is based on full copies of 126 partnership agreements and excludes the following types of agreements: arrangements for longstanding good industrial relations without a written partnership agreement between the primary parties; statements to work towards partnership; single-issue agreements couched in partnership terms (for example health and safety agreements, and learning agreements); non-union agreements; short-term Partnership Fund projects; and standard recognition agreements simply using the label 'partnership'. Excluding these agreements increases the likelihood that agreements have common contents and so this study does not intend to capture the full range of usage of the term partnership in British collective agreements.
Comparing our sample of 126 agreements with the 122 other known partnership agreements in Britain (N = 248) (Bacon and Samuel, 2009) suggests that the agreements examined here are representative in terms of sector, industry, region of operation, and the number of signatory unions. Therefore, this is the most representative analysis of partnership agreements in Britain to date, combining the largest independently collected sample of partnership agreements and the broadest range of organisations with these agreements.
Of the 126 agreements analysed, most (112) continued in operation by December 2007. Six were signed before 1997; 26 signed in New Labour's first term of office (1997)(1998)(1999)(2000); 48 signed in New Labour's second term (2001-4); and 46 signed in New Labour's third term (2005-7). Recent agreements are over-represented as almost three-quarters (74%) of our sample of agreements are five years old or less. However, this over-representation does not appear to affect the findings as statistical analysis (not reported here) shows no significant associations between the contents and age of agreements, or between the contents and survival of agreements.
Within the sample, most agreements are in health and social work (39%) and public administration (20%), with the remainder in other business services (9%), transport and communications (9%) and manufacturing (7%). The majority are public sector partnership agreements (three-fifths of the 126 agreements), which is representative of the estimated population of partnership agreements in Britain for 1990-2007 (three-fifths of 248 agreements) (see Bacon and Samuel, 2009: 237). Over four-fifths (88%) of these agreements involve large organisations with 500 employees or more. Once again, however, statistical analysis (not reported here) shows no significant associations between the contents of partnership agreements and sector or organisational size.

Findings: principles, provisions and the balance of interests
As discussed earlier, advocates assume partnership agreements in Britain will reflect IPA and TUC partnership principles, contain substantive commitments to exchange inter alia employment security for flexible working, and make procedural provisions for joint consultation over an extended range of issues. Thus, partnership agreements are expected to reflect the mutual interests of employers and labour. In contrast, critics argue that partnership agreements offer employers the opportunity to capitalise on relative union weakness and further employer interests. Thus, partnership agreements are expected to contain few, if any, employer concessions to labour. This section considers the extent to which our sample of 126 partnership agreements conforms to these competing expectations.

Principles
The percentages of partnership agreements containing IPA and TUC partnership principles are presented in Table 1. Nearly all agreements (91%) commit the parties to focus on the success of the organisation, a central partnership principle for both the IPA and the TUC. The findings show, however, that many partnership agreements do not include other key IPA or TUC principles. Only just over one-half of agreements (56%) include statements on building trust through greater involvement, possibly reflecting a mutual recognition of lowtrust relations typical in British industrial relations.
The IPA and TUC principle 'respect for the legitimacy of the other partners' is difficult to measure. The principle of explicitly recognising both parties' rights appears in only just over one-quarter of the agreements (28%), in statements recognising the union's right to represent their members and management's right to manage. The union's right to represent their members is recognised in just over one-half of agreements (52%). As the partnership debate suggests, union rights are at risk in these agreements, this is our preferred measure of respecting the legitimacy of the partners to these agreements. However, over two-fifths of agreements (46%) contain neither statement; just over onequarter (28%) recognise the union's right to represent their members but do not contain a clause on management's right to manage; and few (just 2%) recognise management's right to manage but do not contain a clause on the union's right to represent their members.
Moving from values to process (Table 1), two-fifths of agreements (41%) include content relating to the discussion of strategic level issues, and we take this as a commitment to the IPA principle of 'informing and consulting staff about issues at workplace and company levels'.
Turning to the outcomes expressed in partnership agreements, one-half of agreements include mutual benefits statements to share success with employees. Surprisingly, perhaps, just one in 10 agreements include specific statements that commit employers to sharing financial success within the organisation (11%).
As explained above, employers do not make a general commitment to sharing organisational success in one-half of agreements, so it is not surprising that only one-fifth of agreements contain a commitment to employment security (21%), and fewer (16%) contain a commitment to improving the quality of working life. Indeed, our sample of partnership agreements contains no employment guarantees and the few commitments to employment security that are expressed are normative. Significantly, only just over one in ten agreements (11%) include 'flexicurity' statements combining employment security for employees and maximising flexibility for employers. Thus, the lack of employer commitments on employment and job security may help to explain why job security does not appear to differ between employers with partnership agreements and those without (Kelly, 2004).
The typical British partnership agreement contains less than half of the six IPA principles (mean 2.8). Six agreements (5%) contain no IPA principles and only four agreements include all six. Seven in ten partnership agreements feature three or fewer IPA principles. Many partnership agreements also contain few TUC partnership principles. Out of five principles (mean 2.3), only four agreements contain all five, and three in five contain two or fewer. If genuine partnership agreements must include all IPA or TUC partnership principles, then just four agreements (3%) meet this strict criterion: Derby Hospitals NHS In relation to the first aim of this article, therefore, it would appear that most partnership agreements do not fully reflect key IPA or TUC principles. For example, few employers commit to providing employment security and few unions commit to flexible working. This finding is surprising given that the IPA and the TUC, as well as other agencies, have actively promoted these partnership principles in Britain for well over a decade. Partnership agreements in Britain, it seems, display only modest aims.

Substantive provisions
Given that most partnership agreements in Britain lack many of the partnership principles expected, it is questionable whether they will contain provisions representing a broadening of substantive scope. Nearly two-thirds (63%) of the sample of partnership agreements contain one or more substantive provisions, while just over one-third (37%) contain no substantive provisions at all (Table 2). Distinguishing between core and non-core substantive issues (Moore et al., 2004), one-quarter of partnership agreements include provisions on the core issue of pay (26% of agreements), and one-fifth include hours (21%) and working conditions (21%). As reported above, the sample of agreements contains no employment guarantees. The low incidence of provisions covering other, noncore, issues in the sample of agreements is as follows: training (29%), equal opportunities (22%), redundancies (20%), family friendly practices (18%), sick pay (14%), pensions (13%) and single status (10%). Most partnership agreements, therefore, make only limited substantive provisions. On this evidence, it would appear that as well as displaying modest aims, typical formal partnership agreements in Britain are substantively hollow.

Procedural provisions
Most partnership agreements (90%) in the sample include procedural provisions relating to: cooperative statements of intent; the unions party to agreements Over four-fifths of agreements (82%) contain explicit provisions to involve workplace union representatives in joint decision-making. More than twothirds of agreements (68%) establish new consultative forums and many detail their composition and remit. Where specified, partnership consultation covers strategic issues in two-fifths of cases (41%), policy issues in just under half of cases (47%), and workplace issues in just over one-third of agreements (38%).
Arguably, however, the emphasis on consultation over business-led agendas in partnership agreements may serve to weaken or even displace collective bargaining and so undermine union efficacy (Bacon and Storey, 2000;Danford et al., 2005;Kelly, 2004;Marks et al., 1998). If this is so, then partnership agreements should restrict, exclude or simply omit negotiation, and so effectively limit union rights and power (see Oxenbridge and Brown, 2005: 92-3). Yet our findings suggest partnership employers run parallel systems of consultation and negotiation. The majority of partnership agreements (61%) contain explicit provisions for negotiation over general terms and conditions of employment. Nevertheless, as explained, these agreements are substantively hollow, so precisely what is negotiable is rarely specified. One-third of the agreements analysed (33%) distinguish partnership consultation machinery from traditional bargaining by recognising that unions negotiate over terms and conditions in separate arenas. On this analysis, it seems, partnership agreements in Britain may not always be antithetical to collective bargaining and in some cases they may represent the maturation of employer and union policy beyond the traditional adversarialism of British industrial relations, although this possibility remains to be tested.
In relation to the second aim of this article, therefore, it would appear that the general lack of substantive content reflects the modest aims of the typical British partnership agreement. Most agreements, it seems, are substantively hollow. In addition, many display the procedural bias of traditional British collective agreements and promote joint consultation as an adjunct, rather than an alternative, to the collective bargaining process. Given that most unions have signed substantively hollow partnership agreements that do not appear to limit employers' autonomy, it seems plausible that the procedural bias of partnership agreements indicates continuing mutual distrust. The lowest common denominator in these partnership agreements is that employers promise to involve unions when making decisions, in return for a union commitment to the success of the organisation.

The balance of interests
The lack of partnership principles and substantive content in our sample of partnership agreements appears to support Kelly's (2004) argument that employer interests dominate labour interests. However, what has not yet been demonstrated here is the proportion of agreements where unions commit to employer success without employer reciprocation -this indicates constitutes a more direct test of employer dominance. Table 3 shows that employers and unions make reciprocal commitments in just over three-quarters (76%) of agreements. This is important because it shows that most employers make concessions when signing these agreements and unions as institutions make gains. Indeed, in two-fifths (41%) of cases, the largest group of agreements in our sample, employers commit to both sharing success and to union participation in employers' decisions, while unions commit to the success of the organisation. Arguably, these three key commitments express the basis for 'labour-parity' arrangements in that both parties may gain. Typical examples of such agreements include Avon Cosmetics, 2006, DHL, 2007, Ethel Austin, 2002, Eurotunnel, 2000, and nine separate NHS Scotland Area Health Boards, 2003-6. The different permutations of these three commitments suggest five further forms of partnership agreement.
The second most frequent form of partnership agreement (35% of cases) involves union commitments to the success of the organisation and employer commitments to union participation in employers' decisions but without an employer commitment to sharing the success of the organisation. These agreements appear designed to improve industrial relations cooperation; and unions, rather than employees, appear the immediate beneficiaries (Terry, 2003: 462). Typical examples of these agreements include the Driver and Vehicle Licensing Agency, 2004, National Air Traffic Services, 2003, S & A Foods, 2002, and Scottish Widows, 1999 In contrast to predictions that 'employer dominant agreements will be widespread in Britain' (Kelly, 2005: 195), employers appear to dominate the content of less than one in 10 partnership agreements (8%), whereby unions Mutual gains Union commitment to the success of the organisation; employer commitments to sharing success within the company, and union input into employers' decisions.

(41)
Cooperative industrial relations Union commitment to the success of the organisation; employer commitment to union input into employers' decisions.

(35)
Employer dominant Union commitment to the success of the organisation only. 10 (8) Union bypassing Union commitment to the success of the organisation; employer commitment to sharing success within the company.

(7)
Strong union Employer commitments to sharing success within the company, and union input into employers' decisions.

(6)
Shallow None of these three commitments 4 (3) a We classify partnership agreements according to the presence of three key commitments: success of the enterprise, sharing success and input into employers' decisions.
commit to the success of the organisation but employers do not reciprocate and commit to sharing success, or cede union participation in employers' decisions. Typical examples of these agreements include British Bakeries, 2000, North Bristol NHS Primary Care Trust, 2003, and Rolls-Royce (Bristol), 1999. A similar number of agreements (7%) do not include provision for union participation in employers' decisions but do contain union commitments to the success of the organisation and employer commitments to sharing the gains of organisational success. In this small group of agreements, it seems unions are less central to the partnership process and the emphasis may be on direct forms of employee involvement. Examples of these agreements include Accenture HR Services, 2004, Hays Personnel (Managed Solutions), 2004, and Kelly Services, 2005 Several agreements (6%) involve employer concessions but unions do not commit to the success of the enterprise. With the exception of the private sector agreements of Go North East, 2004 and the Offshore Contractors Association, 2006, these are multiple-union agreements in the public sector.
The six forms of partnership agreement are presented in Table 4 by private and public sector, and by government term of office. Note that five of the six Table 4 Types of agreement by sector and government term, 1990-2007 (n=126)

Number of Agreements Signed by Government Term
Sector Type of Agreement 1990Agreement -1997Agreement 1997Agreement -2000Agreement 2001Agreement -4 2005 employer dominant agreements in the private sector were signed in New Labour's first term of office ahead of the passing into statute of the 1999 Employment Relations Act that provided legal provisions for union recognition. Some employers, it seems, may have pre-empted this legislation to sign partnership agreements on their terms. However, the clearest pattern is the surge in mutual gains and cooperative industrial relations agreements in the public sector during New Labour's second and third terms, totalling four-fifths (29 out of 34) of public sector agreements signed in New Labour's second term, and just under four-fifths (27 out of 34) of public sector agreements in New Labour's third term.
In relation to the third and final aim of this article therefore, employers overtly dominate the contents of only a small proportion of agreements (8%). Indeed, three-quarters (75%) of this sample of partnership agreements are more appropriately categorised as mutual gains or cooperative industrial relations agreements, although many of these are public sector agreements signed since 2001. Kelly's (2004) assertion that partnership agreements permit employer interests to dominate those of labour is not a feature of this representative sample of agreements. Most partnership agreements exchange union input into employers' decisions in return for union commitment to the success of the enterprise. Arguably, the contents of most partnership agreements suggest that unions are securing, at least on paper, increased institutional centrality under such arrangements.

Conclusion
Since the mid-1990s, partnership at work was promoted in Britain by successive New Labour governments and organisations representing employers and trade unions. As part of a broader Third Way approach, employers and unions were encouraged to adopt partnership agreements containing a set of principles intended to advance fairness at work. Such agreements, therefore, should contain reciprocal employer-union commitments to deliver mutual gains and establish the basis for a progressive workplace consensus. However, the reliance on voluntary agreements, as opposed to state coercion (Adams, 1995), raises the broader question as to whether New Labour, when in government, succeeded in encouraging employers and unions to adopt partnership agreements to deliver mutual gains.
In order to explore this issue, this article presented findings from the largest and most representative analysis of the contents of partnership agreements signed in Britain between 1990 and 2007. The first aim was to assess the extent to which 126 partnership agreements contain partnership principles and establish the basis for mutual gains. The findings show that many partnership agreements do not contain the expected principles and most have only modest aims. The second aim was to analyse the substantive and procedural content of partnership agreements. The findings demonstrate that most partnership agreements are substantively hollow and procedurally biased. The third and final aim was to assess the relative balance of interests expressed in partnership agreements.
The findings show that employer interests are not dominant, as typical partnership agreements offer unions increased involvement in employers' decisions.
These findings have several implications for the Third Way approach to partnership in Britain, and for understanding the role of power, trust and path dependency for partnership relations. On the first of these issues, our analysis suggests the Third Way approach to partnership in Britain has delivered little in the absence of stronger legal and institutional support for employee participation and gain sharing. The modest aims of partnership agreements in Britain highlight clear limits to New Labour's policy of encouraging partnership agreements. Employers and unions appear remarkably unaffected by the partnership advice offered by successive New Labour government, the IPA, and the TUC. New Labour's attempt to recast Britain's industrial relations landscape was renegotiated downwards without fear of sanction, given New Labour's reluctance to confront employers (less so unions), its preference for self-regulation to state coercion, and the continuing British tradition of state abstentionism in industrial relations. Although the former government and other stakeholders encouraged employers to adopt partnership principles, this is a weak incentive in the private sector compared to other corporate governance pressures that may deter employers from making promises they cannot keep (Deakin et al., 2002;Thompson, 2003). Formal partnership agreements have been modest in number as well as content in the private sector. The findings may appear unsurprising, as Adams (1995) argues that progressive change in industrial relations typically requires coercive action by the state. Social partnership in several northern continental European countries was not the outcome of enlightened consensus, but followed from decisive state action and the imposition of legal obligations on employers to consult and bargain with worker representatives. Without such coercion, the Third Way has not materialised in British industrial relations.
Partnership is sometimes explained by reference to power (Kelly, 2004) and the balance of power between employers and labour helps explain the lack of substantive detail contained in partnership agreements. Since the mid-1990s, many of the formal institutions for joint regulation in unionised workplaces have remained in place but unions have found it increasingly difficult to influence employer policy. Partnership agreements appear to reflect this trend, with employers reluctant to make substantive concessions so retaining their prerogative to determine unilaterally key substantive provisions (see Bacon and Storey, 2000). As many employers do not make significant substantive commitments in partnership agreements it seems highly unlikely that partnership will have a discernable positive impact on employee terms and conditions (see Kelly, 2004).
As employers do not usually make substantive concessions in partnership agreements, the question arises why a significant number of employers signed partnership agreements providing for union input into employers' decisions. Perhaps the simplest explanation is that partnership agreements describe a relationship that most unionised employers should prefer to adversarial relations.
Partnership agreements rarely fail through employer defection, suggesting their interests are not best served by reversion to more adversarial relationships (Bacon and Samuel, 2009).
Employer motives towards unions probably differ in the British private and public sectors. Private sector employers, when signing partnership agreements, appear to seek a general commitment from unions to accept the business must be successful in return for consulting unions over change. It is clearly in the employers' interests to sign such limited agreements that do not restrict their autonomy. Thus, partnership agreements may not provide for actual union influence on employers' decisions and some case studies suggest this is often limited in practice (Upchurch et al., 2008). A different explanation is feasible in the public sector, where employers seek to engage stronger unions in discussion over market-based reforms to avoid industrial action. Partnership agreements in the public sector, arguably, represent 'an uneasy truce' between unions resisting market-based reforms and managers unable to modernise services without engaging staff in the process (Bacon and Samuel, 2009: 244). Historically, this situation seems to be business as usual for public sector industrial relations in Britain. Given that the growth in the number of agreements has been in the public sector and the NHS in particular, future research may usefully explore whether voluntarism adequately describes the government's approach in this sector.
Our findings also suggest unions make few concessions and therefore have little to lose by signing partnership agreements that are substantively hollow and procedurally biased. Unions in the private sector implicitly support business success and at least a formal partnership agreement may help remind employers to consult with unions. However, public sector unions signing these agreements do not formally commit to public sector reforms and so may sign partnership agreements to engage employers in debates over proposed changes to improve public services. Both private sector and public sector employers and unions, therefore, appear to make limited commitments in order to manage the perceived risks associated with partnership (Martínez Lucio and Stuart, 2005).
The absence of ambition in partnership agreements appears also to reflect the low-trust equilibrium in British industrial relations. Indeed, only two-fifths of agreements conform to an underlying mutual gains model. The absence of the expected partnership principles in most agreements suggests traditional low-trust relationships underpin these agreements. Both employers and unions have probably made optimum concessions in partnership agreements given the degree to which they feel reciprocation is possible. Thus, agreements fall a long way short of the promise of mutual gains and may reflect distrust between employers and unions. The limited ambition of partnership agreements in Britain is, perhaps, inevitable as employers in liberal market economies cannot realistically keep promises to share financial gains with employees, or to provide employment security, let alone job security, given increased productmarket competition and shareholder pressure (Sisson et al., 1999;Thompson, 2003). Clearly, high-trust relationships are difficult to establish in an institutional context encouraging low trust.
A final explanation of our findings is that partnership agreements appear path dependent. Partnership agreements reflect the lack of substantive content and contain the procedural bias previously noted in traditional collective agreements (Dunn and Wright, 1994;Flanders, 1968). The non-statutory approach to partnership in Britain by former New Labour governments does not appear to have changed traditional arms-length, low-trust approaches to industrial relations in Britain.
These findings are consistent with previous evidence that many organisations in Britain have made only limited progress towards a partnership approach (Bacon and Storey, 2000;Guest and Peccei, 2001). It appears that voluntary partnership agreements are weak instruments for extending fairness at work in the low-trust industrial relations contexts of liberal market economies, such as Britain. Despite the partnership rhetoric of New Labour's Third Way since 1997-2010, it seems that in the absence of stronger statutory and institutional supports for employee participation and gain sharing, voluntary partnership agreements in Britain express only modest aims of limited ambition.