Corps de l’article

Introduction

This is a study of the relationship between institutions, corporate governance and union power based on large scale comparative survey evidence. An influential strand of the finance literature focuses on shareholder rights (see La Porta et al., 1997, 1998). More recently, this literature has been expanded to look at how other markets – including labour markets – are regulated, and the relationship between this and shareholder rights (see Botero et al., 2004; Djankov et al., 2003). For example, Botero et al. (2004) analyze the link between laws on collective action and investor protection. However, the mainstream finance literature remains dominated by rational hierarchical approaches that see institutional effects primarily in terms of constraints on rational actors, focusing on the relative strength of property rights, looking at the effects of law, constitutions and politics (Djankov et al., 2003: 596). In reaction, an emerging body of socio-economic literature has provided criticisms of such approaches on three fronts. First, owner and stakeholder behaviour may not simply reflect state-regulatory pressures or the absence thereof; institutional pressures operate at a wide range of levels and domains, encompassing markets, civil society, organizations (Boyer, 2006: 36–37). Second, institutional complementarity may not necessarily emerge from mutually supportive features but may, in fact, arise from contradictions (Crouch, 2005: 360–362); hence, strong employee rights may not necessarily preclude strong ownership rights, or, indeed, effective organizational performance. Mutually beneficial complementarities in turn, are likely to reinforce specific industrial relations practices even if, over time, both will be subject to redefinition and change (Hopner, 2005). Third, employees and managers are not passive subjects of external institutional forces, but make choices of their own, moulding organizational outcomes, and impacting on the relative powers of owners (Gospel and Pendleton, 2004). This reflects a key difference in underlying assumptions: rational-incentive approaches focus on the manner in which individuals seek to maximize their own benefit. In contrast, socio-economic accounts give a central place to the issue of relationships (encompassing both individuals and associations) and the manner in which such relationships are regulated (Hall and Gingerich, 2004).

The predictions of rational-incentive approaches therefore deserve further investigation and critique, given that they constitute the dominant strand of the mainstream economics and finance literature and have been particularly influential in guiding policy makers in liberal market economies. As Ezzamel and Reed (2008: 612–613) note, evaluating and critiquing the full range of different accounts on corporate governance and firm behaviour is of value, both in understanding what firms do (both broadly and in the area of industrial relations) and in developing alternative, critical and more explicitly multi-dimensional alternatives. In this article, we systematically assess and critique the predictions of rational hierarchical approaches to corporate governance on the relative strength of employees and their collectives in different national and organizational settings, and critically reappraise the implications thereof, and the relevance of alternative approaches within the heterodox socio-economic tradition.

Rational-Incentives Approaches

Rational-incentives approaches see institutions as constraints on the ability of rational profit maximizing individuals to make optimal decisions (Powell and DiMaggio, 1991; Shleifer and Vishny, 1997; North, 1990). As Ezzamel and Reed (2008: 611) note, such approaches combine “the neo-rationalist emphasis on rational calculation and action with more recent theoretical reinterpretations that give greater emphasis to the strategic role of political and cultural power in reshaping modes of collective regulation,” and, hence, individual choices and actions.

Of particular concern is the issue of property rights: approaches adopting this starting point have been particularly influential in the literature on finance and corporate governance (Djankov et al., 2003: 596). In practice, the rational-incentives approaches assign paramount importance to hierarchy, with dominant relations imposing isomorphism (Boyer, 2006). The role of owners is given particular attention; other social actors are treated as potential hinderers – or, at best, facilitators – on the effective operation of markets (see North, 1990). When hierarchies do not operate effectively, the system does not work optimally. Institutional settings create a particular process of decision making, leading to optimal or sub-optimal outcomes.

Rational Incentive Accounts of the Relationship between the Firm and its Stakeholders

Shleifer and Vishny (1997) see shareholders as the only stakeholders with sunk funds in the firm; hence they argue that the firm’s sole objective should be the maximization of their wealth. Similarly, Roe (2003) believes that improving the conditions of workers merely worsens the classical agency problem between managers and shareholders and prevents the separation of ownership and control.[1] In detail, the classical agency problem consists of managers’ tendency to focus on corporate growth rather than on shareholder value. Roe (2003) and Shleifer and Vishny (1997) both suggest that employers and employees prefer bigger firms, the former in the interests of empire building, and the latter as they tend to provide better working conditions and job security. Hence, in countries with left wing governments, ownership and control will remain concentrated to counteract managers’ tendency towards empire building. Conversely, countries with right wing governments will favour investors over professional managers and workers; ownership will separate from control as investors are well protected by law from abuses by corporate managers (Roe, 2003).

Stakeholder Power and Politics

Developments and extensions of rational-incentive approaches have factored in the impact of institutions but only in terms of being providers of incentives or disincentives on rational actors. Pagano and Volpin (2005) hypothesize that the type of electoral system – a proportional or majoritarian electoral system – is the main determining factor of the levels of protection that employees and investors have. Whereas a majority of votes ensures victory in a proportional system, one needs to win a majority of districts in a majoritarian system. Pagano and Volpin (2005: 1016) predict that the former produces strong workers’ rights but weak investor protection, as it makes politicians focus on social groups with homogeneous preferences, i.e. workers and their unions, and managers, at the expense of owners. A majoritarian system produces the exact opposite outcome.

Type of Legal System

La Porta et al. (1997) had previously distinguished between two main families of law, common law and civil law. Common law, the legal system of the UK, US and most Commonwealth countries, accords an important role to the judges who make law in the courts by setting precedents. In contrast, civil law, the legal systems of Continental Europe and most of the rest of the world, is based on extensive codes of law – having their origins in Roman law – and the role of the judges is limited to interpreting these codes in the courts.[2] La Porta et al. (1997) argue that, given its higher flexibility and better enforcement, common law provides stronger investor protection than civil law. Further, among the civil law family, French civil law provides the lowest levels of shareholder protection, followed by German civil law and then Scandinavian civil law. More recently, La Porta and colleagues have argued that legal systems shape the regulation of a wide range of markets, including labour (Botero et al., 2004: 1379), with civil law countries being characterized by the direct regulation of markets (including labour markets) by governments (ibid.: 1340). In particular, Botero et al. (2004) find that civil law markets have more extensive law governing employment relations as well as industrial relations, i.e. collective actions by the workers.

Summarizing Rational Hierarchical Approaches

Key similarities in these approaches can be summarized as per Figure 1; Table 1 further illustrates differences and similarities.

Figure 1

Rational Hierarchical Accounts of Institutional Setting, Property Owners and Outcomes

Rational Hierarchical Accounts of Institutional Setting, Property Owners and Outcomes
Boyer 2006

-> Voir la liste des figures

Table 1

Context and Practice – Alternative Perspectives

Theory / Characteristic

Roe

Pagano and Volpin

La Porta et al.

Determines strength of non-owner stakeholders

Government policy and ideology

Electoral system

Legal System

Measure

Left or right wing governments

Proportional representation (PR) or first past the post

Common or civil law

Path dependence

Path dependence

Limited number of alternatives. Electoral systems rarely changed

Path dependence

Predictions – Shareholder rights

Property owners stronger under right wing governments

Property owners stronger under majoritarian systems

Property owners stronger under common law

Predictions – Employee rights

Right wing governments are likely to make for weaker employee rights and, hence, fewer constraints on on corporate governance

First past the post electoral systems are likely to make for weaker employee rights, and hence, fewer constraints on corporate governance

Common law systems are likely to make for weaker employee rights and, hence, fewer constraints on corporate governance

Number of optimal arrangements

One

One

One

-> Voir la liste des tableaux

This literature, as noted, advocates the maximization of shareholder value, privileging the role of the owners of the business who have sunk funds in the firm. Owner and employee rights are seen as a zero-sum game: should the one be stronger, the other will be weaker and vice versa. Hence, it is assumed that “carefully devised laws and sanctions . . . as well as carefully drafted employment contracts are necessary to provide the necessary carrots and sticks for workers to act in the interests of owners” (Ezzamel and Reed, 2008). However, Goergen (2007) argues that workers and suppliers, similar to shareholders, have sunk investments in the firms they deal with and that they are equally likely to lose these investments when the firm disappears. Hence, the question of employee rights is central to the analysis of the nature of corporate governance not simply in terms of something that negatively or positively impacts on the manner in which the firm can be governed: labour is not simply another commodity that can be purchased and disposed of. Recent work on corporate governance has focused on this lacuna, and sought to explain how specific ownership regimes impact on both managers and labour and, indeed, the manner in which employees and their collectives influence both owners and managers (Johnstone, 2007: 94; Gospel and Pendleton, 2003: 559; Gospel and Pendleton, 2004).

Socio-Economic Alternatives

Recent critiques have suggested that institutional features that are seemingly at odds with each other may, in fact, generate complementarities in compensating or balancing contrasting components: for example, industry relevant skills may offset pressures towards low labour mobility in other areas (Crouch, 2005: 360–362); institutional pressures operate at a wide range of levels and domains (Boyer, 2006: 36–37).

These critiques raise two fundamental issues. First, there is the question as to whether rational-hierarchical approaches to analyzing institutional effects are sufficient for understanding managerial choices and, in turn, the relative power of employees and their collectives. Second, the “mixed” approach to understanding institutional effects highlighted by Crouch (2005: 360–362) emphasizes that strong shareholder powers do not necessarily make for weak unions. The relative strength of employee collectives may depend on complementarities elsewhere in the system; the relative strength of civil society, and a wide range of non-shareholder framed enablers and constraints on managerial behaviour. Managers and other stakeholders are not simply passive subjects of ownership behaviour, but active agents making strategic choices that may mould outcomes (Gospel and Pendleton, 2003: 559). These issues deserve closer investigation.

In contrast to rational-incentive approaches, and from a heterodox starting point, a burgeoning body of socio-economic literature (Dore, 2000; Whitley, 1999; Hall and Soskice, 2001) locates institutions in terms of webs of relationships (rather than individual selfish action) (Ezzamel and Reed, 2008). Hall and Gingerich (2004: 7) note that in order to effectively manage their people, firms “have to engage with multiple spheres of the political economy.”

The Varieties of Capitalism literature argues that a key distinction is between liberal market economies (LMEs) and the coordinated market alternatives (CMEs). Examples of the former would include Britain and the United States, and, of the latter, continental North Western Europe. Such literature does highlight the continued viability – and distinct paths – of alternatives to lightly regulated liberal markets. And, as relationships may make for mutually beneficial complementarities, there is no inherent reason why employee and owner well-being should be antagonistic or mutually exclusive; rather, it has been argued that in CMEs, a more sustainable basis for prosperity is provided through mutually beneficial trade offs and compromises between key stakeholders (see Lincoln and Kalleberg, 1990; Dore, 2000). Whilst the limits of rational-incentive approaches to understanding industrial relations may be immediately evident, the Varieties of Capitalism literature is not without problems.

Firstly, a two-archetype model discounts the important differences that exist between coordinated markets, and the importance of other types of capitalism, such as the Mediterranean variety(ies) (Amable, 2003; Boyer and Hollingsworth, 1997). This makes such approaches only of limited utility in seeking to understand the differences in industrial relations practice in continental Europe. A second limitation is that such approaches – in common with rational-incentive approaches – assume that national institutional frameworks are deeply embedded, and unlikely to change significantly over time (Boyer, 2006).

Alternative approaches to understanding institutions from regulationist and historical institutionalist starting points share the Varieties of Capitalism’s focus on relationships, associations, and informal and formal rules. However, the former has stressed the temporary, and inherently unstable nature of institutional mediation, and the uneven nature of systemic evolution and change (Boyer, 2006). Meanwhile, the latter sees institutions as flexible and dynamic, although at specific times, coherences will be evident at national and regional levels (Lane and Wood, 2009). From a range of different starting points, a number of different writers have also sought to broaden the range of capitalist archetypes, from a simple LME/CME dichotomy, inter alia to identify differences between European coordinated markets, and Far Eastern alternatives (Whitley, 1999), and between Scandinavia and continental Europe (Amable, 2003). Both these accounts give very much more detailed attention to the types of difference in industrial relations that are likely to be encountered in different contexts; again, both identify unionization, the central focus of this paper, as a key dimension.

Data and Methodology

In this paper, we aim to test the veracity of the claims of the rational incentive approaches, and explore the extent to which any limitations may be accounted for within the socio-economic literature on comparative capitalism. Our study benefits from longitudinal data sourced from the Cranet database. Cranet is a survey of HR managers. It relies mostly on closed-ended questions in order to preclude prevarication and/or ambiguous responses. The Cranet survey is conducted every 3 to 4 years. Over time, the number of countries surveyed has increased. The latest survey, the 2003/4 survey, covers a total of 41 countries from Western Europe, Eastern Europe, developed countries such as the USA, Japan and Australia as well as transitional economies such as South Africa, Turkey and Tunisia. Other surveys were conducted in 1991, 1995 and 1999/2000. The surveys aim to be representative of each economy at each point in time by covering all sectors of the target countries’ economies. For the vast majority of countries, they are full population surveys. For a smaller number of the larger countries (e.g., France, Germany, Italy and the UK), firms were selected randomly, but weighted for sector and size, from publicly available mailing lists. The surveys exclude smaller firms, i.e. those with less than 100 employees.

To ensure data is available across the various years, this paper is based on the 1995, 1999/2000 and 2003/2004 surveys and focuses on firms[3] from Western European countries only. In the empirical analysis, we use time dummy variables to identify observations from the three waves of surveys and to detect possible behavioural shifts across time. Due to the confidential nature of the data involved, firms forming the panel in a given year cannot be matched with those from other years. Since our interest is in developments in national economies over time, rather than in the effects on individual organizations, this detailed, representative data from three different points in time over more than a decade is appropriate for our purposes.[4]

We use two measures of the influence of trade unions.[5] The first one is the proportion of the total number of employees in each company that are members of a trade union. Central to the strength of unions is an ability to draw members into the fold, and organize them: “unions (only) exist wherever workers are unionized” (Hyman, 2002: 55; Kelly, 1998: 126–127); in turn, union membership provides the basis for collective bargaining and other forms of collective representation (Kelly, 2002: xli). In the survey questionnaire, the variable is ordinal in nature and takes on the following values: 0 if the proportion of employees who are members of trade unions is 0%; 1 if it is between 1–10%; 2 if it is between 11–25%; 3 if the proportion is between 26–50%; and 4 and 5 if it is between 51–75% and 76–100%, respectively. This variable measures the current power of unions within a given firm, and, reflects their ability to attract and retain workers (c.f. Checchi and Visser, 2005: 7–8).

The second measure captures the change in the perceived influence of unions, i.e. respondents’ views on the degree of change in the influence of trade unions in a firm during the previous three years. This potentially captures two alternative phenomena: the actual extent to which unions’ influence in the firm has increased or decreased, and/or the extent to which management believes (whether based on concrete evidence or not) that union strength is increasing or decreasing. Management also influences union strength. The relationship between firms and unions is essentially about power and the distribution of resources (Hyman, 2002: 62–63): if managers see unions as weak, they may seek to push them back even further, leading to a further weakening of unions (Kelly, 1998; Kelly, 2002: xli; Hyman, 2002). Hence, this question measures not just union strength, but the likelihood that managers will adopt a harder line towards them (c.f. Kelly, 1998: 103–104). This is also an ordinal variable which takes on a value of zero if unions have had no influence, 1 if their perceived influence has decreased, 2 if their influence has remained the same, and 3 if union influence has increased. As both measures of union influence are discrete and there is also a clear order of preference between the different values they can take, the appropriate estimation technique is an ordered logit. Each logit regression contains industry dummies[6] and year dummies.

For each measure of trade union influence, we run four separate ordered logit regressions. Each logit is estimated using the pooled company-level data for 1995, 1999/2000 and 2003/2004. The first logit is a test of La Porta et al.’s (1997) legal theory. The logit contains three dummies: one for common law countries, another one for countries with German law and a third one for countries with Scandinavian law. This implies that the base case will be that of the countries of French law, the group of countries with the weakest investor protection. The second and third logits test Roe’s (2003) politics theory. Both logits use politics indices from Cusack and Engelhardt (2006). The first ordered logit regression is based on RILE which ranges from –100 (extreme left) to +100 (extreme right). The index measures the political orientation of the party in power based on its views on a wide range of economic and social issues. This index is identical to that used by Roe (2003: 49–50) himself. The second logit uses MYRL as the main explanatory variable. This index is more specialized as it measures the political orientation of the party in power in terms of its views on ten different issues concerning the regulation of markets and wealth distribution. Similar to RILE, it ranges from –100 (extreme left) to +100 (extreme right). Finally, the fourth type of logit is based on Pagano and Volpin’s (2005) proportionality index. The index, which is available for 1985–2002, is equal to zero if none of the seats are assigned via a proportionality rule, 1 if a minority of seats are assigned by such a rule, 2 if the majority of seats are assigned by this rule, and 3 if all of the seats are assigned proportionally.[7] All the logits contain industry and year dummies.[8]

Findings

Tables 2 to 4 report the results for the various ordered logits. The tables report the regressions based on the proportion of employees who are members of a trade union. The equivalent regressions for the change in the perceived influence of trade unions are discussed in what follows, but not reported in tabular form. Table 2 contains the results for the logit explaining the proportion of employees in the trade union by La Porta et al.’s (1997) legal families. First, the threshold variables in the ordered logit are all highly significant, suggesting that an ordered logit is the appropriate estimation technique. The various types of pseudo R-squares are also relatively high, ranging from 0.131 to 0.338. Second, the dummies for the three families of common, German and Scandinavian law are all highly significant. In particular, the coefficient on the common law dummy is negative, suggesting that trade union penetration is lower in countries of the common law variety. However, contrary to La Porta et al. (1997, 1998, and 2000) and Botero et al. (2004), union membership is higher in countries with German law (the coefficient on the German law dummy is significant and positive) and highest in those with Scandinavian law (the coefficient on the Scandinavian law dummy significant and even higher) relative to countries of French law. Finally, it is also worth mentioning that virtually all of the industry dummies are significant and positive. This suggests that the influence of trade unions varies across industries as well as across the various legal families; in part, this could reflect industry specific variations in work organization, and in the relative centralization of bargaining (see Boyer and Hollingsworth, 1997). The year dummies are both significant and positive with the year dummy for 1995 having the highest coefficient. This would reflect the fact that the proportion of employees who are members of a trade union has dropped gradually over 1995 to 2003/4 in many national contexts. In turn, this would reflect the general weakening in the influence of unions that has followed on the long period of prosperity and growth that ended in the early 1970s, and the structural changes in the global economy that followed it; whilst unions have fared better in some national contexts than others, most have battled to stem declining memberships, revitalization strategies notwithstanding (Kelly, 1998; c.f. Checchi and Visser, 2005; Western, 1993).

Table 2

La Porta et al. – Proportion of Employees in Trade Union

 

 

Estimate

Std. Error

Wald

df

Sig.

95% Confidence Interval

 

 

 

 

 

 

 

Lower Bound

Upper Bound

Threshold

[propunion2 = .00]

–1.404

.123

129.208

1

.000

–1.646

–1.162

 

[propunion2 = 1.00]

.213

.122

3.056

1

.080

–.026

.452

 

[propunion2 = 2.00]

1.018

.122

69.110

1

.000

.778

1.257

 

[propunion2 = 3.00]

2.155

.124

301.184

1

.000

1.911

2.398

Location

Common_law

–.102

.048

4.472

1

.034

–.196

–.007

 

German_law

.306

.057

28.843

1

.000

.194

.417

 

Scan_law

2.663

.061

1887.214

1

.000

2.543

2.783

 

IND2

.421

.158

7.055

1

.008

.110

.731

 

IND3

.515

.146

12.508

1

.000

.230

.801

 

IND4

.309

.125

6.109

1

.013

.064

.555

 

IND5

.566

.126

20.269

1

.000

.320

.813

 

IND6

.069

.150

.214

1

.643

–.225

.363

 

IND7

–.907

.133

46.331

1

.000

–1.169

–.646

 

IND8

.257

.144

3.183

1

.074

–.025

.540

 

IND9

–.976

.130

56.080

1

.000

–1.232

–.721

 

IND10

–.750

.224

11.155

1

.001

–1.190

–.310

 

IND11

.929

.142

43.080

1

.000

.651

1.206

 

IND12

–.164

.157

1.091

1

.296

–.473

.144

 

IND13

1.026

.154

44.655

1

.000

.725

1.327

 

IND14

1.309

.145

81.307

1

.000

1.024

1.593

 

IND15

1.074

.165

42.596

1

.000

.751

1.396

 

IND16

–.201

.134

2.234

1

.135

–.464

.062

 

IND17

2.664

.645

17.067

1

.000

1.400

3.927

 

year95

.175

.049

12.573

1

.000

.078

.271

 

year99

.087

.051

2.938

1

.087

–.012

.186

 

Pseudo R-Square

 

 

 

Cox and Snell

.338

 

Nagelkerke

.353

 

McFadden

.131

The dependent variable in the ordered logit is an ordinal variable which equals 0 if the proportion of employees who are members of trade unions is 0%; 1 if it is between 1–10%; 2 if it is between 11–25%; 3 if the proportion is between 26–50%; 4 if it is between 51–75% and 5 if it exceeds 75%. Common_law, German_law and Scan_law are dummy variables which equal 1 if the firm in question is from a country with common law, German law and Scandinavian law, respectively, and equal 0 otherwise. The industry dummies are defined in endnote 6.

-> Voir la liste des tableaux

Regression results for the dependent variable showing the perceived change in the influence of trade unions over the three previous years are all highly significant (the results are not reported in a table).[9] However, the pseudo R-squares are much lower: it seems that, in most instances, managers’ views on the strength of unions are not founded on wishful thinking (perhaps to justify a harder line approach to unions in the future) but rather are realistic (and indeed, cautious) estimates of the position and potential of unions. The results suggest that, relative to the French law family, the base case, trade unions in common law countries have suffered the highest perceived drop in their influence, closely followed by those in countries with German law. Conversely, trade unions in countries of the Scandinavian law tradition have benefited from a slight increase in their influence. Fewer of the industry dummies are significant than in the previous table, and some are positive and others negative; this would reflect the complexities of industry effects within and between sectors, reflecting the relative importance of specific production paradigms (Boyer and Hollingsworth, 1997). The evidence here also suggests a clear cross-national decreasing trend in trade union influence.

Table 3 tests the validity of Roe’s (2003) politics theory using the Cusack and Engelhardt (2006) politics index RILE, which covers the government’s orientation in terms of a wide range of social and economic issues. All of the threshold variables are significant. The coefficient on RILE is significantly negative which suggests that the influence of trade unions, as measured by the proportion of employees who have membership, is lower in countries with right wing governments. This provides support for Roe’s (2003) politics thesis. Our findings also echo those of Checchi and Visser (2005: 14) that the efforts of right wing governments to curtail union power negatively impacted on union density; however, Western (1993: 277) suggests these effects may be lagged. Further, compared to the regressions based on La Porta et al., the values for the pseudo R-squares tend to be much lower (by as much as a factor of about 3). Compared to Table 2, most of the industry dummies are significant. Similar to Table 2, not all of the significant dummies are positive. In particular, the industry dummies for sectors providing services to public institutions are highly significant and positive. This may reflect the privatization of central and local government services over the period of study. Private sector organizations taking over the provision of public services may have inherited workers who have had traditionally a high trade union representation and who would now have to fight for their collective rights (Dibben et al., 2007). Again, the dummy for the year 1995 is significantly positive, suggesting a decreasing trend in union membership over 1995–2003/4. The results for the ordered logit explaining the proportion of employees in the trade unions by the more specialized politics index, MYRL, confirm the results from Table 4.[10] Again, trade unions seem to have less influence under right wing governments. Similar conclusions can also be drawn as to the industry and year dummies. Estimation results for ordered logits[11] explaining the perceived change in the influence of trade unions using the RILE and MYRL politics index, respectively, as the main explanatory variable, suggest that perceived trade union influence has decreased during conservative governments.

Table 3

Roe (Politics Index) – Proportion of Employees in Trade Union

 

 

Estimate

Std. Error

Wald

df

Sig.

95% Confidence Interval

 

 

 

 

 

 

 

Lower Bound

Upper Bound

Threshold

[propunion2 = .00]

–1.745

.117

222.482

1

.000

–1.974

–1.515

 

[propunion2 = 1.00]

–.297

.115

6.659

1

.010

–.522

–.071

 

[propunion2 = 2.00]

.370

.115

10.324

1

.001

.144

.595

 

[propunion2 = 3.00]

1.253

.116

117.340

1

.000

1.027

1.480

Location

RILE

–.003

.001

20.306

1

.000

–.004

–.001

 

IND2

.362

.152

5.694

1

.017

.065

.660

 

IND3

.281

.140

4.015

1

.045

.006

.556

 

IND4

.137

.120

1.291

1

.256

–.099

.373

 

IND5

.377

.121

9.738

1

.002

.140

.614

 

IND6

.051

.144

.127

1

.722

–.231

.334

 

IND7

–.731

.129

32.264

1

.000

–.983

–.479

 

IND8

.434

.138

9.881

1

.002

.163

.704

 

IND9

–.926

.126

53.987

1

.000

–1.173

–.679

 

IND10

–.386

.217

3.154

1

.076

–.812

.040

 

IND11

.640

.135

22.401

1

.000

.375

.905

 

IND12

–.073

.151

.233

1

.629

–.369

.223

 

IND13

.724

.146

24.576

1

.000

.438

1.010

 

IND14

1.262

.137

85.169

1

.000

.994

1.530

 

IND15

1.091

.154

50.004

1

.000

.788

1.393

 

IND16

–.188

.130

2.102

1

.147

–.442

.066

 

IND17

1.906

.648

8.653

1

.003

.636

3.176

 

year95

.364

.048

58.040

1

.000

.271

.458

 

year99

–.063

.049

1.649

1

.199

–.158

.033

 

Pseudo R-Square

 

 

 

Cox and Snell

.111

 

Nagelkerke

.116

 

McFadden

.037

The dependent variable in the ordered logit is an ordinal variable which equals 0 if the proportion of employees who are members of trade unions is 0%; 1 if it is between 1–10%; 2 if it is between 11–25%; 3 if the proportion is between 26–50%; 4 if it is between 51–75% and 5 if it exceeds 75%. RILE ranges from –100 (extreme left) to +100 (extreme right). The index measures the political orientation of the party in power based on its views on a wide range of economic and social issues. The industry dummies are defined in endnote 6.

-> Voir la liste des tableaux

Table 4 uses Pagano and Volpin’s (2005) proportionality index as the main explanatory variable of the proportion of employees in the trade unions.[12] As suggested, the power of workers seems to be higher in countries that allocate a higher proportion of parliamentary seats via the proportionality rule. The pseudo R-squares for the regression explaining the proportion of workers who are members of trade unions are relatively high. However, they are not more pronounced than those for the regression on the perceived change in the influence of trade unions, confirming the argument that managerial assessments of changes in union influence are generally realistic. Once more the figures reflect the decline in union power since 1995. Again, the level and evolution of trade union influence have been different across various industry sectors and these patterns hold across various national systems.

Table 4

Pagano and Volpin’s Proportionality Index – Proportion of Employees in Trade Union

 

 

Estimate

Std. Error

Wald

df

Sig.

95% Confidence Interval

 

 

 

 

 

 

 

Lower Bound

Upper Bound

Threshold

[propunion2 = .00]

–.784

.120

42.389

1

.000

–1.020

–.548

 

[propunion2 = 1.00]

.792

.120

43.854

1

.000

.558

1.027

 

[propunion2 = 2.00]

1.518

.120

159.092

1

.000

1.282

1.753

 

[propunion2 = 3.00]

2.487

.122

416.518

1

.000

2.249

2.726

Location

PROPVOT

.561

.015

1339.102

1

.000

.531

.591

 

IND2

.384

.153

6.316

1

.012

.085

.684

 

IND3

.301

.141

4.549

1

.033

.024

.578

 

IND4

.320

.121

6.991

1

.008

.083

.558

 

IND5

.617

.122

25.567

1

.000

.378

.856

 

IND6

.162

.145

1.244

1

.265

–.123

.447

 

IND7

–.555

.129

18.382

1

.000

–.809

–.301

 

IND8

.519

.139

13.865

1

.000

.246

.792

 

IND9

–.799

.127

39.714

1

.000

–1.048

–.551

 

IND10

–.386

.219

3.109

1

.078

–.816

.043

 

IND11

.933

.137

46.274

1

.000

.664

1.202

 

IND12

.254

.153

2.779

1

.095

–.045

.554

 

IND13

1.246

.149

69.652

1

.000

.953

1.538

 

IND14

1.537

.140

121.208

1

.000

1.263

1.810

 

IND15

1.231

.157

61.261

1

.000

.923

1.539

 

IND16

–.168

.130

1.654

1

.198

–.423

.088

 

IND17

3.153

.647

23.732

1

.000

1.884

4.421

 

year95

.311

.048

42.079

1

.000

.217

.405

 

year99

–.135

.050

7.387

1

.007

–.232

–.038

 

Pseudo R-Square

 

 

 

Cox and Snell

.225

 

Nagelkerke

.235

 

McFadden

.081

The dependent variable in the ordered logit is an ordinal variable which equals 0 if the proportion of employees who are members of trade unions is 0%; 1 if it is between 1–10%; 2 if it is between 11–25%; 3 if the proportion is between 26–50%; 4 if it is between 51–75% and 5 if it exceeds 75%. proportionality index. The index, which is available for 1985–2002, is equal to zero if none of the seats are assigned via a proportionality rule, 1 if a minority of seats are assigned by such a rule, 2 if the majority of seats are assigned by this rule, and 3 if all of the seats are assigned proportionally. The industry dummies are defined in endnote 6.

-> Voir la liste des tableaux

To summarize, based on the pseudo R-squares in the regressions, Pagano and Volpin’s (2005) theory seems to work best in terms of explaining both the level of trade union influence and the perceived change in trade union influence. Although the type of legal family has an influence on the power of trade unions, the effects are different from those that one would expect based on La Porta et al. (1997). In particular, French law, which according to La Porta et al. fares worst in terms of investor protection, does not seem to have the strongest trade unions even if, arguably, those unions have a disproportionate influence at national level. Roe’s (2003) politics thesis works, but relatively less than Pagano and Volpin’s theory.

Conclusions: The Limits and Possibilities of Theory

The survey revealed that the strongest indicator of union strength was national legal tradition, suggesting a significant degree of path dependence. However, it cannot be concluded that this difference is only due to investor protection, allowing owners more or less room to impose their wishes on the workforce. As noted earlier, in terms of La Porta et al.’s index of shareholder protection, and their later work on employee rights (Botero et al., 2004), we would expect countries of French law to have the highest union influence, followed by countries of German and Scandinavian law, when in fact the converse is the case.

Why would this be so? As suggested by the industrial relations literature, this could reflect both investor protection and the practical operation of labour law: in other words, the relationship between institutional setting and practice cannot be explained solely in terms of a single hierarchy. First, relatively weak shareholder rights do not necessarily make for strong unions or vice versa. For example, if shareholders are weak, managers may enjoy a relatively high degree of autonomy; the manner and scope in which they exercise this is likely to be constrained and enabled by wider civil society, inter-firm relations and state policies in a range of different areas. Conversely, in a number of civil law contexts, relatively strong ownership protection appears to have coexisted with strong employee rights (Johnstone, 2007: 96; Jackson, 2004: 284–309). Second, and as a result, investor protection constitutes an incomplete basis for understanding the effects of different national legal traditions. Different legal traditions will affect shareholder rights directly, but will also affect those of employees in a manner that transcends the simple operation of markets. The manner in which employees seek to enforce their rights is an open-ended process, with, especially in common law systems, not always predictable results.

This does not mean that common law systems will necessarily work better for employers. Whilst employees may have fewer rights that are harder to enforce, common law systems may result in excessive levels of litigation, constraining management flexibility in a range of areas concerning the employment contract (Colvin, 2006: 73); this is particularly the case, given that such systems are often associated with strong – and expanding – individual rights, in areas such as equality and anti-discrimination (c.f. ibid.: 91; Wood, Harcourt and Harcourt, 2004). Civil law systems are indeed diverse, though not necessarily in the manner predicted by La Porta and colleagues; indeed, it seems that in some instances, increases in investor rights have proved compatible with cooperative industrial relations involving strong unions (Johnstone, 2007: 96; Jackson, 2004: 284–309).

Whilst not providing quite such a strong fit, the Pagano and Volpin model also provides some measure of union strength: rentier power is greater in majoritarian systems, which may facilitate a harder line against unions. However, again, a caveat is in order: we found that the level and evolution of trade union influence have been different across various industry sectors which would reflect the operation of other factors at work. In other words, the relationship between electoral system and firm level practice did not appear to operate as a single hierarchy – there is more to understanding the relative strength of employees than simply as the opposite of shareholder rights. This unevenness would include the effectiveness of the operation – or lack thereof – of bargaining arrangements as a check on employer power; highly centralized bargaining – ultimately in the form of neo-corporatist deal making – may have uneven effects on sectoral practices (Harcourt and Wood, 2003). A neo-corporatist tradition equips unions better to survive the vicissitudes of right wing governments or, indeed, internal strategic blunders (c.f. Turner, 2004; Kelly and Frege, 2004). Proportional democracy is more likely also to lead to coalition governments and these may, in turn, facilitate neo-corporatism (Olsen, 1982). Indeed, almost all of the countries where sustainable neo-corporatist deals have been reached have had proportional representation or mixed systems; in turn, the former are likely to make for higher levels of union density (Western, 1993). There is also a strong correlation between participative structures at the workplace, and propensity to participate in the wider political life; a history of workplace co-determinist structures in many proportional representation systems is likely both to strengthen pluralism in the workplace and to make union members more likely to vote (D’Art and Turner, 2007: 107). Hence, collective employee representation at the workplace is likely to make pro-labour parties more successful in national level elections and vice versa (ibid.). In contrast, with a majoritarian system, the rentier grouping is likely to be relatively strong, driving trends towards patronage and elite based politics, strengthening the hand of the property-owning classes (Warner, 1997: 542).

A somewhat less strong predictor was the Cusack and Engelhardt index of right or left wing governments; as the literature suggests, right wing governments are worse for unions and collective representation at the workplace. However, there appeared more to this relationship than a simple hierarchy: union strength varied on a sectoral basis, reflecting the operation of other dynamics.

A further three caveats are in order. First, the weakening of the traditional model of union-party relations commonly encountered in social democracies (c.f. Valenzuela, 2002; McIlroy, 1998) makes it difficult to conclude that a left wing government will be consistently union friendly. Second, the role of right wing governments may vary from concentrating on protecting investor rights, to a more active role in curbing the organizing and mobilizing activities of unions (Gamble, 2002). Third, different state and institutional traditions may limit the powers of elected governments vis-à-vis a range of social collectives (Turner, 2004). And, institutional mediation is often open ended and incomplete (Ogden and Watson, 2008). These caveats may explain why legal traditions are a stronger predictor of union strength; however, it would be wrong to conclude that this means that owners are necessarily worse off. The literature on neo-corporatism highlights the extent to which, in the Scandinavian and Rhineland countries, an emphasis on consensus, compromise and mutual understanding may provide a more sustainable basis for prosperity (Harcourt and Wood, 2003; Arestis and Marshall, 1995). Under such regimes, firms that take a longer view regarding the setting of wages and human resource development are not undermined by short-termist less principled competitors poaching newly trained workers (Hall and Gingerich, 2004; Henley and Tsakalotos, 1992; Harcourt and Wood, 2003). Again, firms based in such countries are likely to have stronger incentives to invest in employee skills development, on account of the greater security of tenure and rights enjoyed by the latter (Gospel and Pendleton, 2004: 4; Regini, 1997). Hence, we would argue for an approach that sees corporate governance and behaviour as more than simply variations in property rights and industrial relations.