SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Collective bargaining legislation in the public sector and conceptual framework
  5. 3. Methodology
  6. 4. Data
  7. 5. Empirical results
  8. 6. Concluding remarks
  9. Acknowledgement
  10. References

We study public-sector bargaining and contract outcomes using Canadian data from 1978 to 2008. We have a number of interesting results, but our principal findings are from our analysis of wage settlements. We find that the essential services designation, which only allows non-essential members of a bargain unit to strike, is associated with decreases in wages. Our estimates also suggest that there is an arbitration wage premium and that making adjustments to the ability to pay criterion used by arbitrators to determine awards does not affect this premium. We also discuss the implications of our estimates.

1. Introduction

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Collective bargaining legislation in the public sector and conceptual framework
  5. 3. Methodology
  6. 4. Data
  7. 5. Empirical results
  8. 6. Concluding remarks
  9. Acknowledgement
  10. References

The last several decades have been a time of transition and upheaval in the public sector. A combination of economic and political forces has initiated an era of structural change in public-sector employment arrangements in the USA and Canada (Hebdon et al. 2013; Swimmer 2001). This shift in the economic and political environment has led to a movement towards efficiency in delivering public-sector outputs and a ‘new public management’ (Hebdon and Kirkpatrick 2005). Some of these changes in the public sector include greater non-standard or contingent employment, privatization, transferring services to other levels of government, as well as restructuring and other cost-saving initiatives (e.g., among others, Beaumont 1996; Thompson 1995; Gunderson 2002; Jalette and Hebdon 2012).

Governments responded to the deep recession of the early 1990s by changing the collective bargaining process in various ways. For example, in Canada, some governments used legislation that imposed temporary controls on compensation and arbitration awards, while others changed arbitration criteria to emphasize ability to pay (Swimmer 2001). Unfortunately, many earlier studies studying public-sector contract and bargaining outcomes have relied on data collected prior to the start of this era of structural change. Consequently, it is not clear whether the findings from these earlier studies still hold in the new environment of structural change and hard bargaining. This suggests that it is important to have estimates that reflect the changing environment in the public sector and its effect on contract and bargaining outcomes.

In this article, we study public-sector bargaining and contract outcomes in Canada using data that cover the period between 1978 and 2008, which include contracts and strikes from the 10 Canadian provinces as well as the Federal jurisdiction. Our sample will thus overlap with earlier studies, but also includes the more recent negotiations that reflect the current realities in the public sector. Earlier papers have also emphasized the quality of the Canadian data, as well as the similarities in the industrial relations system between Canada and the USA (e.g., among others, Budd 1996; Currie and McConnell 1991; Gunderson and Melino 1990). We focus on three outcome measures: the probability of how the contract was settled (strike, legislated settlement, mediation, arbitration and bargained), strike duration and wage settlements. We use the term bargained to refer to settlements made without third-party assistance.

While we consider three outcomes measures, we focus our attention on wage settlements. Wage settlements are of particular interest in this era of restraint since various governments implemented wage controls in the public sector as one way to reduce their spending. Governments in many jurisdictions used direct interventions (e.g., temporary policies to limit wage settlements and arbitrated awards in the early 1990s, as well as legislating wage settlements and other contract terms instead of bargaining) in the bargaining process to control the growth of wages in the public sector. In addition, many jurisdictions also strengthened the ability to pay criterion used by arbitrators to attempt to limit the size of wage awards in the public sector. We will explore the effects of these changes on wage settlements in the public sector. Earlier research using Canadian data has found that compulsory arbitration is associated with an increase in wage settlements relative to other legal structures, and that the ability to pay criterion is not the principal driver of arbitrated settlements (Currie and McConnell 1991). One important policy implication of this earlier literature is that strengthening the ability to pay criteria of arbitrators could reduce the arbitration wage premium relative to other legal structures. Our estimates will show whether these conclusions and policy implications still hold in this new era of public-sector bargaining.

The next section provides some background information on collective bargaining legislation and a conceptual framework. Section 3 presents our empirical methodology for each of the outcome measures we consider. Section 4 discusses our data. We present our empirical findings in Section 5 and also discuss how they compare to previous findings. Section 6 concludes the article with a summary of our principal results and a discussion of their implications.

2. Collective bargaining legislation in the public sector and conceptual framework

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Collective bargaining legislation in the public sector and conceptual framework
  5. 3. Methodology
  6. 4. Data
  7. 5. Empirical results
  8. 6. Concluding remarks
  9. Acknowledgement
  10. References

Collective bargaining and labour relations legislation for the public sector in Canada is generally a provincial responsibility, although the public-sector workers for the Federal government fall under the Federal jurisdiction.1 Generally, these options include (a) separate and specific legislation for various elements of the public sector, for example, among others, police, firefighters and nurses; and (b) private-sector legislation for elements of the public sector not covered by specific legislation. For example, municipal workers are often not covered by specific legislation and so would be covered by private-sector legislation in the jurisdiction.

There are many aspects of labour relations legislation that impact collective bargaining, but our interest is primarily on the legal structures that exist in the public sector. The legal structures in the public sector in our data include compulsory arbitration, right-to-strike, essential services designation and choice of procedure. We consider the effects of these legal structures on settlement outcomes and disputes in the public sector from the perspective of the ‘joint cost’ model. The joint cost theory predicts that strikes are inversely related to the ‘joint cost’, that is the sum of costs, of a strike to both the firm and the union (Kennan 1980; Redder and Neumann 1980; Siebert and Addison 1981). The joint costs model of strikes posits that strikes should be less frequent when the costs of a strike (to both parties) are high. Similarly, the joint costs model also predicts that strikes should be shorter when the costs of a strike are higher. Consequently, factors that can increase the costs of strikes should reduce the incidence of strikes as well as their duration when they do occur. Moreover, the joint cost theory predicts that policy variables and other variables related to the bargaining environment can reduce strikes by reducing uncertainty, misinformation and divergent expectations.

The dispute resolution procedures for the legal structures in the public sector vary across jurisdictions, but fall into three broad categories. First, under compulsory arbitration, workers are not allowed to strike, and unresolved disputes must be settled through final and binding arbitration. The types of arbitration available are either conventional arbitration (where the arbitrator fashions the terms of the contract) or final offer selection (where the arbitrator selects the final offer of either the union or the employer). Second, in the right-to-strike legal structure, workers have the right to go on strike if they cannot negotiate a contract with their employer after they are in a legal strike position, with no restrictions on the sorts of workers who can participate in a job action. Alternatively, in the right-to-strike legal structure, the employer can lock out the workers if they cannot reach an agreement with the bargaining unit. However, the strike or lock-out cannot occur unless all legal requirements have been met (e.g., mediation has been completed). In legal structures where dispute resolution is by arbitration, the employer cannot lock out. Third, in an essential services designation, unions are allowed to strike, but some portion of the workers in the bargaining unit are legally obligated to continue providing designated services (as determined prior to a strike) during the job action. Procedures for the determination of essential services vary by province.

In a compulsory arbitration legal structure, strikes are illegal. The joint costs model suggests that the penalties for illegal strikes, for example fines and disciplinary action against some workers, would increase the costs of going on strike for the union, and make strikes less likely to occur and shorter when they do occur relative to a right-to-strike or the essential services designation.

Since the precedence set by previous awards helps guide an arbitrator's award, there might be some feedback from previous settlements to current wage settlements. This means that if there are large awards in the past, then the current awards will reflect this. This could drive up wages above those negotiated settlements if the awards being decided in an economic downturn are based in part on precedence set during economic boom years. However, if past awards show restraint, then current awards will also reflect this restraint and lead to lower wages.

In the right-to-strike legal structure, strikes/lock-outs are legal, and the joint costs theory of strikes suggests that work stoppages could be more frequent and longer when they do occur relative to the other legal structures in the public sector. The unrestricted strike threat in the right-to-strike legal structure could increase the bargaining power of the union because it can withhold all services. Thus, the strike threat in the right-to-strike legal structure could be associated with wage increases relative to legal structures that have restricted or no strike threats.

The essential services designation, like the right-to-strike legal structure, allows for strikes but prohibits some members of the bargaining unit from walking out or being locked out by the employer. This means that there are some basic services that continue to be provided by a bargaining unit even during a strike. From the perspective of the joint costs model, the costs of a strike are lower — but not less than those in the right-to-strike legal structure — so that there are likely to be fewer strikes and shorter strikes when they do occur relative to the right-to-strike legal structure, but more and longer strikes relative to the other legal structures. The effects of the essential services designation on wages are more difficult to determine than the other legal structures in the public sector. On the one hand, bargaining units with employees that are designated as essential often end up in arbitration, and this can lead to larger wage settlements. Alternatively, bargaining units with employees designated as essential cannot totally withhold services with the strike threat, unlike the right-to-strike legal structure, and this could reduce their bargaining power. If these two effects offset each other, then the essential services designation legal structure would be ambiguous on wages.

In a choice of procedure legal structure, the union chooses before negotiations whether it wants dispute resolution by strike or arbitration. This means that the ultimate effect of this legal structure on strike and bargaining outcomes depends on the union's choice, and so can range from the effects of a right-to-strike regime to those in a compulsory arbitration one. Gunderson (2002) noted that, in the past, bargaining units in the Federal jurisdiction with weak bargaining positions tended to choose the arbitration option in order to get the pattern in wage settlements set by more powerful bargaining units. However, bargaining units have increasingly opted for the strike option in the Federal jurisdiction in order to use a strike threat to bargain for larger wage settlements (Gunderson 2002).

Sitting alongside these designations, some unions in the public sector could face private-sector legislation, if they are not targeted by specific legislation. This legislation can include compulsory conciliation, which can involve both one-stage and two-stage procedures; cooling off periods, which occur after conciliation and before a strike; mandatory strike votes; employer called strike votes; compulsory dues check-offs, where all employees pay union dues; bans on replacement workers; reinstatement rights, for workers on strikes; and reopener clauses (negotiated reopeners and technical change reopeners). These legislative procedures and their potential effects are typically considered in analyses of private-sector data. We refer the reader to Campolieti et al. (2014) for a thorough discussion of these effects on collective bargaining outcomes from the perspective of the joint cost model as well as some other theories. While we include controls for these private-sector legislative variables in our empirical models, we will focus our discussion on the estimates of the effects of the legal structure and some of the other variables in our empirical specification.

3. Methodology

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Collective bargaining legislation in the public sector and conceptual framework
  5. 3. Methodology
  6. 4. Data
  7. 5. Empirical results
  8. 6. Concluding remarks
  9. Acknowledgement
  10. References

Our analysis focuses on several aspects of the bargaining process in the public sector. First, we consider the factors influencing how the contract is settled. More specifically, some contracts require an arbitrator or mediator/conciliator to help facilitate a contract. In other cases, the government intervenes and imposes a settlement or could use more general legislation to put limits on wage settlements in the public sector. The contract might also be settled with both parties coming to an agreement after bargaining. Finally, there could also be a work stoppage (strike or lock-out). Our first outcome measure captures these five potential avenues to contract settlement. Second, for the bargaining units that do experience strikes (or lock-outs), we also estimate the determinants of strike duration, that is how long do strikes last when they occur. Third, we also consider the factors affecting wages, which we use to measure the settlement costs of negotiations.

We estimate the probability of how the contract is settled with a multinomial logit model. Our dependent variable for this model indicates whether there was a strike (or lock-out), the contract had a legislated settlement, arbitration was used to settle the contract, mediation or conciliation was employed to settle the contract, or the contract was bargained without third-party assistance. Our specification of the multinomial logit model will include controls for the legal structure and other legislative variables, previous contract outcomes, political party in power, economic conditions, and bargaining unit characteristics, which we will detail in the next section. Rather than presenting coefficient estimates or odds ratios in our tables, we present the derivatives or marginal effects of each of the explanatory variables. The marginal effects will reflect the impact of a change in the explanatory variable on the probability of how the contract was settled for each of the five settlement categories.

We also examine the determinants of strike duration for the bargaining units that experience work stoppages using a Cox proportional hazard model (Cox 1972), which specifies the (settlement) hazard rate at time t as function of explanatory variables (e.g., legal structure and other legislative variables, political party in power, economic conditions, and bargaining unit characteristics). We present hazard rate ratios, which are easier to interpret. For example, if an explanatory variable has a hazard rate ratio of 1.10 (0.90) then a one-unit change in that variable is associated with a 10 per cent increase (decrease) in the hazard rate, so that exits occur more quickly (slowly) or, equivalently, that the variable is associated with a decrease (increase) in strike duration.

We also consider the factors affecting the settlement costs of a negotiation by estimating a regression with the natural log of the average real wage during the contract as a dependent variable on a vector of variables that includes the legal structure and other legislative variables, the political party in power, the characteristics of the bargaining unit, economic conditions, the log of wages in the previous contract, average wages in the province, controls for previous contract outcomes, as well as year dummies. We estimate this regression using a fixed-effect estimator. The fixed-effect estimator makes it possible to control for unobserved bargaining-pair heterogeneity that could have an effect on wage settlements. Like earlier papers (e.g., among others, Budd 1996; Cramton et al. 1999; Currie and McConnell 1991), we use wages as a measure of settlement costs since they are likely to account for the bulk of the settlement costs for a negotiation.

We compute the standard errors for all the models we estimate using the cluster-robust variance estimator. Our clustered standard errors incorporate a group-specific correlation term induced by the fact that the legislative variables and some of our other controls only vary at the jurisdiction and year level, while the control variables vary at the bargaining unit level (Moulton 1990). This means that our standard errors are more conservative than those in earlier papers studying collective bargaining outcomes in the public sector because they did not cluster their standard errors.

One potential concern about our estimates is that strike activity and other contract outcomes could influence the legislation that is put in place by a jurisdiction. Consequently, we follow previous literature (e.g., Currie and McConnell 1991, 1994), and include jurisdiction and year dummy variables in our empirical models to lessen the potential endogeneity biases in the legislative variables.

4. Data

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Collective bargaining legislation in the public sector and conceptual framework
  5. 3. Methodology
  6. 4. Data
  7. 5. Empirical results
  8. 6. Concluding remarks
  9. Acknowledgement
  10. References

Human Resources and Skills Development Canada (HRSDC) collects and distributes a number of datasets on wages, contract settlements and strikes that we use in this analysis. Our data cover the period from January 1978 — the first month in the datasets — to April 2008, which is the last month that wage and strike data could be matched to other supporting economic data when we began this project. The wages dataset only provides wage information for bargaining units that contain at least 500 workers (except for the Federal jurisdiction where the limit is 100 or more). The contract settlement database contains information on the means by which a collective agreement was settled. When we merge the contract outcome and wage database, we focus on contracts with more than 500 members in the bargaining unit to maintain a consistent mix of bargaining units between the Federal and other jurisdictions. The strikes database, which cannot be merged with either the contract or wage databases, contains information on bargaining units of all sizes. However, we restrict the strike data to contain strikes for bargaining units with 500 or more members for consistency with our wage and contract analysis. The data from the contract database are somewhat richer than the data in the strike database. Unfortunately, this means that we cannot include some of the controls we use in our contract and wage analyses (e.g., how the contract was settled in the previous negotiation) in our strike duration analysis.

Information on collective bargaining legislation comes from various sources at the Labour Law Analysis unit at the HRSDC. We also present a summary of the bargaining units that changed legal structures in Table 1. The rows of Table 1 are the legal structure at time t, while the columns are the legal structure at time t + 1. The main diagonals of Table 1 report the proportion of the bargaining units in our sample that remain in the same legal structure between two years, while the off-diagonal elements in the table represent the transitions to different legal structures between one year and the next.

Table 1. Changes in Legal Structure
Legal structure at time tLegal structure at time t + 1
Right-to-strikeChoice of procedureCompulsory arbitrationEssential service designation
  1. Notes: Entries in the table indicate the proportion of bargaining units in the sample that move from legal structures at time t to the legal structure at time t + 1. The main diagonals of the table indicate the proportion of bargaining units that remain in the same legal structure; the other entries denote the portion of bargaining units that change legal structures between time t (indicated in the row) and t + 1 (indicated in the column).

Right-to-strike0.75550.00060.01370.0101
Choice of procedure0.00040.00530.00.0038
Compulsory arbitration0.01600.00.12690.0006
Essential service designation0.00930.00380.00040.0535

Each collective agreement in the public-sector wages and strikes dataset is assigned a service sector, based on the broad industry category reported by the HRSDC. We rely on the existing literature for the dates that compulsory arbitration, choice of procedure and the right-to-strike were extended to specific sectors, as well as for general labour relations and collective bargaining legislation that applies to the private sector. We also identified when specific sectors are assigned an essential services designation.2

The wage database provides wage level and annual negotiated settlement data. We calculate the real hourly wage in the contract by applying the average annual increase of a contract to the previous wage level as reported in our dataset and then convert these into real values using the consumer price index (CPI).

Our data contain information on how the contract was settled. This includes whether the contract was settled by direct bargaining without third-party help, mediation or conciliation, arbitration, strike or lock-out, or was legislated. These five categories are mutually exclusive. For the bargaining units that experience strikes or work stoppages, the strike database contains information on the characteristics of the bargaining units as well as the length of the strike (in days). Our legislated settlements include two types. First, it includes settlements where the provincial or Federal government passed legislation that imposed a settlement on the bargaining unit, where the terms of the settlement are set by legislation and not negotiation between the bargaining unit and public-sector employer. This type of settlement is sometimes referred to as a ‘back-to-work’ order. The second type of legislated settlement, which was more prevalent in the 1990s, was some form of limited wage determination, such as wage freezes and/or guidelines on wage settlements, that was designed to control ‘bargained’ wage settlements as well as arbitrator awards. Unfortunately, we cannot distinguish between these two types of settlements in our data since we do not have adequate detail in the coding in the databases on the nature of the legislated settlement. Consequently, we refer to both of these types of settlements as legislated settlements and do not distinguish between these two types of interventions in our discussion.

Our data do not come with the details that allow us to distinguish between the different types of arbitration. Conventional interest arbitration allows arbitrators to fashion awards, while final offer arbitration forces the arbitrator to choose the final offer of either the union or the employer. There are no jurisdictions in Canada that mandate final offer arbitration, although there are a few jurisdictions where it is an option if it is mutually agreed to. While final offer arbitration is seldom used in the jurisdictions that allow it, there have been a couple of recent cases where the Government of Canada has mandated final offer selection in special back-to-work legislation. Unfortunately, the lack of detailed coding for the arbitration information in the HRSDC databases means that we cannot distinguish between conventional and final offer arbitration in the data. However, as noted earlier, most arbitration in Canada takes the form of conventional interest arbitration (Gunderson 2002). We also do not have a measure of the percentage of workers covered by essential services designation in a bargaining unit, although this percentage has been on the rise over the last few decades (Gunderson 2002).

In order to control for the economic environment that public-sector contract negotiations occur in, we retrieve information on provincial unemployment rates, consumer price indices and average provincial wages from Statistics Canada's CANSIM database and merged these with the HRSDC databases. The other variables we include in our empirical analyses are all coded from the information available in the HRSDC databases (e.g., among others, bargaining unit size, cost of living adjustment clauses, province, union affiliation, year, industry, contract duration and previous contract outcomes).

We also incorporate information on the political environment in our empirical analysis by including a set of indicator variables for the political party that was in power at the time of contract settlement. The Canadian political system is parliamentary in nature, so we classify the party in power as a majority government, where it holds a majority in the legislature and can pass legislation without the support of another party, or a minority government, where it needs the support of another party to pass legislation. There are three major political parties in Canada, which tend to be active at the Federal level and most provinces: the New Democrat Party (NDP), which is left of centre in the political spectrum; the Conservative party, which is right of centre in the political spectrum; and the Liberal party, which is centrist in its ideology. Our term Conservatives includes the Progressive Conservative party that is active in many provinces, as well as the Conservative party at the Federal level, which was a merger of the Progressive Conservatives and the Reform Party. We code the indicators for political party in power as Liberal majority government, Liberal minority government, Conservative majority government, Conservative minority government, NDP majority government, NDP minority government and other. Our ‘other’ category includes provincial political parties that are regional in nature, such as the Saskatchewan party in Saskatchewan, or the Parti Québécois in Quebec.

5. Empirical results

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Collective bargaining legislation in the public sector and conceptual framework
  5. 3. Methodology
  6. 4. Data
  7. 5. Empirical results
  8. 6. Concluding remarks
  9. Acknowledgement
  10. References

Table 2 contains the descriptive statistics for the disputes and contracts in our data. We present the statistics for duration separately since the strike and wage contract settlement databases are not linked. Mean strike duration is about 30 days in our sample, but there is some variation in the strike length by legal structure. Mean strike duration is about 12 days in the compulsory arbitration structure, while in the choice of procedure and essential services designation mean that strike duration is about 23–24 days. In contrast, in the right-to-strike legal structure, mean strike duration is about 32 days.

Table 2. Summary Statistics for Contracts, Disputes and Wages by Legal Structure, 1978–2008
  Legal structure
TotalRight-to-strikeChoice of procedureCompulsory arbitrationEssential service designation
(a) Strike database
Strike duration (days)30.532.524.112.322.7
(b) Settlement database
Number of contracts5,8314,56653791422
Number of strikes2492092434
Number of arbitrations489239822418
Number of mediations or conciliations97782058567
Number of legislated settlements61950794459
Mean real wage$18.88$18.92$20.66$19.85$16.33
Median real wage$18.47$18.49$20.64$20.55$15.73

Panel (b) in Table 2 presents summary statistics based on the settlements and wage database. Most of the contracts in our sample are concentrated in the right-to-strike legal structures. Strikes are not very frequent (4.3 per cent incidence rate) in our sample, and most occur in the right-to-strike legal structure. There are very few strikes in the choice of procedure and compulsory arbitration legal structures, which is consistent with the predictions of the joint cost model. Arbitration tends to be more common as a dispute resolution procedure, occurring in about 8.4 per cent of all the contracts in our sample. Not surprisingly, arbitration features much more prominently in the compulsory arbitration and choice of procedure legal structures as a dispute resolution method (28 and 15 per cent of contracts are settled using arbitration), but it is also used in the right-to-strike and essential service legal structures (5.2 and 4.3 per cent of negotiations are settled using arbitration). The overall dispute rate, strikes plus arbitrations, in our sample is about 12.7 per cent. As a percentage of contracts, the overall dispute rate tends to be lower in the right-to-strike (9.8 per cent of all contracts) and essential service designations (12.3 per cent of all contracts). Mediation tends to be used more in the right-to-strike and essential services designation, relative to the choice of procedure and compulsory arbitration legal structures. There are also a considerable number of contracts with legislated settlements. About 11 per cent of the contracts in our sample had legislated settlements, with most of these occurring in the right-to-strike (11.1 per cent of contracts), choice of procedure (17 per cent of contracts) and essential service legal structures (14 per cent of contracts). One interpretation of these patterns in disputes is that there are fewer disputes when the strike threat is an option and is not restricted, that is the right-to-strike legal structure. However, that interpretation is tempered by the fact that the government intervenes more in contract negotiations involving right-to-strike legal regimes.

One of the stylized facts of the public-sector bargaining literature is that the compulsory arbitration legal regime has higher wage levels than contracts in other legal structures. We find that the mean contract wage in the compulsory arbitration legal structure is higher than that in the right-to-strike and essential service designation. The mean wage level in the choice of procedure legal structure is about 80 cents larger than the mean wage in the compulsory arbitration legal structure. However, the mean could be sensitive to outlier contract settlements, so we also present median wages since the median is less sensitive to outliers. Median wages in the choice of procedure and compulsory arbitration legal structures are quite similar, and the differences between the median wages in the compulsory arbitration legal structure relative to the right-to-strike and essential services designation legal structures are much more pronounced.

5..1 Contract Settlement Analysis

We estimate a multinomial logit for the probability for how the contract was settled and present the marginal effects in Table 3. These marginal effects are for each of the different categories in the bargaining outcome measure, which includes work stoppages (i.e., strikes or lock-outs), legislated settlements, arbitration, mediation/conciliation and bargained.

Table 3. Multinomial Logit Estimates of How Contract Is Settled, 1978–2008
 Contract outcome
StrikeLegislatedArbitrationMediation or conciliationBargained
Marginal effectStandard errorMarginal effectStandard errorMarginal effectStandard errorMarginal effectStandard errorMarginal effectStandard error
  1. Notes: Excluded reference category is presented in square brackets. The specification also includes dummy variables for jurisdictions that have legislation for conciliation, cooling-off periods, mandatory strike votes, employer-initiated strike votes, reinstatement rights, compulsory dues check-off, bans on replacement workers, negotiated reopeners and technical reopeners, as well as controls for industry, year, province, average real provincial wage, log of previous contract duration, log of number of members in bargaining unit and inverse of unemployment rate, which are not presented. Standard errors are clustered by jurisdiction and year.

  2. *** denotes statistical significance at the 1 per cent level; ** denotes statistical significance at the 5 per cent level; * denotes statistical significance at the 10 per cent level.

  3. NDP, New Democrat Party.

Mean probability0.043 0.106 0.084 0.161 0.607 
Legal structure [Right-to-strike]
Choice of procedure−0.0060.008−0.0020.0020.0420.036−0.0550.0380.0220.054
Compulsory arbitration−0.017***0.003−0.003*0.0020.202***0.046−0.072***0.019−0.110**0.048
Essential service0.0040.005−0.0010.0010.0280.023−0.0040.031−0.0270.040
Previous contract outcome [Bargained]
Legislated settlement−0.0040.0050.0000.0030.0260.0320.249**0.101−0.271***0.099
Strike0.0130.0110.0030.0030.0130.0280.109**0.047−0.138***0.050
Conciliation or mediation0.018***0.005−0.0010.0010.031**0.0120.081***0.019−0.129***0.024
Arbitration−0.011***0.003−0.0010.0020.100***0.0220.0010.024−0.090***0.034
Political party in power [Liberal majority]
Liberal minority0.0080.008−0.0050.0030.043*0.0240.0250.043−0.071*0.040
Conservative majority0.0020.0050.0010.0020.034**0.0130.0060.026−0.0430.028
Conservative minority0.0070.010−0.007**0.0040.0100.0250.0130.055−0.0230.056
NDP majority−0.0030.005−0.008**0.003−0.0040.0180.097**0.044−0.083*0.050
NDP minority−0.0030.014−0.005**0.002−0.057***0.005−0.0550.0400.120**0.033
Other0.0140.014−0.0010.0020.0480.038−0.0180.072−0.0420.082
Value of log likelihood function−4,990.23         
Number of contracts5,831         

Our estimates on the indicator variables for choice of procedure and essential service designation tend to have marginal effects that are quite small in magnitude and not statistically significant. However, the dummy variable for compulsory arbitration tends to have a much larger and more precisely estimated effect on how the contract is settled. Interestingly, a compulsory arbitration legal structure is associated with a statistically significant decrease in the probability of bargained settlements relative to the right-to-strike legal structure. We also find that compulsory arbitration is associated with a decrease in settlements that involve strikes (a decrease of 1.7 percentage points). This estimate is consistent with the joint cost model of strikes, which suggests that strikes should be less frequent in the compulsory arbitration legal structure. This estimated effect is also quite large relative to the mean strike incidence rate. The estimates for the compulsory arbitration dummy variable are also associated with a decrease in the likelihood of legislated settlements (a decrease of 0.3 percentage points) and mediation (a decrease of 7.2 percentage points). In contrast, and not surprisingly, we find that compulsory arbitration is associated with a 20 percentage point increase in the likelihood of an arbitrated settlement.

One potential concern about our estimates on the effect of legal structure is that these effects could also be partly captured by the previous contract outcomes. To examine the sensitivity of our estimates in Table 3 to this, we exclude the previous contract outcomes variables from the specification. We find that our conclusions would be unchanged; all of the marginal effects would have the same sign and many the same values, although there are a few estimates that are slightly larger.

Previous outcomes can also have some important effects on the probability of how the contract is settled. In particular, our estimates indicate that previous outcomes can adversely impact the probability of a bargained outcome. More specifically, a legislated settlement in the previous contract is associated with a 27 percentage point decrease in the probability of a bargained contract.3 A strike in the previous contract is associated with a 14 percentage point decrease in the probability of a bargained contract. Conciliation or mediation in the previous contract is also associated with a 13 percentage point decrease in the probability of a bargained contract. Finally, arbitration in the previous contract is associated with about a 9 percentage point decrease in the probability of a bargained contract, which could reflect some of the chilling effects of arbitration (Ehrenberg and Schwarz 1987). A legislated settlement in the previous negotiation is associated with a 25 percentage point increase in the probability of mediated settlements. This result could reflect that bargaining units with legislated settlements in the past might want to use mediation to obtain more favourable settlements. However, it is also consistent with bargaining units wanting to obtain contract terms that they would have more control over or to avoid the stress (and potential costs) of a legislated settlement. A strike in the previous contract is associated with an 11 percentage point increase in the probability of a mediated settlement. This suggests that mediation might be more attractive when there have been work stoppages and bargaining difficulties that the union and public-sector employer may wish to avoid in the future. Finally, the past use of mediation is also associated with an 8 percentage point increase in the probability of a mediated settlement in the current contract.

Legislated settlements and strikes in the previous contract do not have a statistically significant effect on the probability of settling the current contract with strikes, legislated settlements or arbitration. In contrast, arbitration in the previous contract is associated with a 1 percentage point decrease in the likelihood of a strike. One interpretation of this finding is that it could reflect a greater reliance on arbitration since the number of contracts settled using arbitration has increased over time. Another plausible interpretation of this finding is that the past use of arbitration provides more information to both unions and employers, and this could reduce the number of strikes in the future as the likelihood of divergent bargaining positions is reduced. We also find that the past use of arbitration also increases the likelihood of arbitrated settlements in the current contract. Our results for arbitration are quite important from a policy perspective, since they are consistent with a narcotic effect associated with arbitration, so that the past use of arbitration may lead to the increased use of arbitration in the future (Ehrenberg and Schwarz 1987). Alternatively, it is also consistent with an unobserved factor that reflects a bargaining relationship with some persistence of past conflict.

Conciliation and mediation in the previous contract are associated with a statistically significant increase in the probability of a strike in the current contract. We also find that the past use of conciliation or mediation increases (about 3 percentage points) the probability of an arbitrated settlement in the current contract. A possible policy implication of these results is that conciliators and mediators ought to exploit relationship improvement opportunities to avoid conflict in subsequent rounds of bargaining or arbitrated settlements.

The estimates on the controls for political party in power at the time of contract settlement also present some interesting results. Liberal minority and Conservative majority governments tend to be associated with increases in the probability of arbitrated settlements and decreases in the probability of bargained contracts, relative to Liberal majority governments. Conservative minority governments are associated with a relatively small decrease in the probability of legislated settlements, but not with any of the other contract outcomes. This result is somewhat surprising since the conservatives tend to be a right-wing and pro-business political party, but this estimate could reflect that the conservative's coalition partners are either at the centre or the left of the political spectrum. NDP majority governments are associated with a decrease in the probability of legislated settlements and an increase in the probability of mediated settlements. NDP majority governments are also associated with 8 percentage point decrease in the probability of bargained contracts relative to Liberal majority governments. Our estimates indicate that NDP majority governments, who tend to be pro-union, are less likely to legislate settlements, but they may be more likely to use to use third parties to settle contracts. NDP minority governments, like NDP majority governments, are associated with decreases in the probability of legislated and arbitrated settlements, relative to Liberal majority governments. Moreover, NDP minority governments are also associated with a 12 percentage point increase in the probability of bargained contracts, relative to Liberal majority governments.

5..2 Strike Duration Analysis

We present the estimates from our Cox proportional hazard model in Table 4 for bargaining units with 500 or more employees. Our estimates indicate that relative to the right-to-strike legal structure, the other public-sector legal structures that we consider (i.e., choice of procedure, compulsory arbitration and essential service designation) are not associated with any statistically significant differences in the strike settlement hazard. While we find that strikes in the right-to-strike legal structure are longer by about 10 days in our descriptive analysis in Table 2, we are not able to obtain a great deal of precision in our estimates from the Cox proportional hazard model in Table 4 because there are very few strikes in the other legal structures.

Table 4. Hazard Rate Ratios for Cox Proportional Hazard Model of Strike Duration, 1978–2008
 Hazard rate ratiot-statistic
  1. Notes: Excluded reference category is presented in square brackets. The specification also includes dummy variables for jurisdictions that have legislation for conciliation, cooling-off periods, mandatory strike votes, employer-initiated strike votes, reinstatement rights, compulsory dues check-off, bans on replacement workers, negotiated reopeners and technical reopeners, as well as dummy variables for industry, province, year and union affiliation, but these estimates are not presented. Standard errors are clustered by jurisdiction and year.

  2. *** denotes statistical significance at the 1 per cent level; ** denotes statistical significance at the 5 per cent level; * denotes statistical significance at the 10 per cent level.

  3. NDP, New Democrat Party.

Public-sector designation [Right-to-strike]
Choice of procedure1.1340.36
Compulsory arbitration1.0040.01
Essential service designation1.0280.14
Strike in previous contract1.0640.57
Inverse of unemployment rate4.6810.77
log(bargaining unit size)1.0170.43
Political party in power [Liberal majority]
Liberal minority0.724−1.05
Conservative majority0.645***−2.94
Conservative minority1.1470.45
NDP majority0.918−0.36
NDP minority1.2470.25
Other0.868−0.65
Number of contracts637 
Value of log likelihood function−3,394.94 

We also find that our control for the business cycle, that is the inverse of the unemployment rate, is associated with an increase in settlement hazard, that is strikes would be shorter in good economic times, which is consistent with public-sector strikes being countercyclical (e.g., Harrison and Stewart 1989). However, this estimate is not statistically significant. While we find that the size of the bargaining unit is associated with an increase in the settlement hazard, as in the literature examining private-sector strikes (e.g., Campolieti et al. 2005, 2014), our estimate is not statistically significant. Most of our controls for the political party in power have estimates that are not statistically significant. However, we find that conservative majorities are associated with a statistically significant decrease in the settlement hazard, that is longer strikes, relative to Liberal majority governments.

5..3 Wage Analysis

We present the fixed-effect estimates from our analysis of wage settlements in Table 5. In the first set of estimates (for the 1978–2008 period), we find that relative to the right-to-strike legal structure, the choice of procedure legal structure is associated with a 2.7 per cent decrease in wage settlements, although the estimates are only statistically significant at the 10 per cent level. The coefficient estimate on the compulsory arbitration legal structure dummy variable is 1.4 per cent, but it is not statistically different from zero. We also find that the essential services designation is associated with a statistically significant 2.3 per cent decline in wages, relative to the right-to-strike legal structure. This suggests that in the essential services designation, restricting a bargaining unit's ability to strike, relative to the right-to-strike legal structure, translates into less bargaining power and lower wage settlements.

Table 5. Wage Regressions
 1978–2008Excludes changes to arbitration involving ability to pay from 1978 to 2008 sampleExcludes period of public-sector wage restraint (1990–1993) from 1978 to 2008 sampleExcludes changes to arbitration involving ability to pay and period of public-sector wage restraint (1990–1993) from 1978 to 2008 sample
Coefficient estimateStandard errorCoefficient estimateStandard errorCoefficient estimateStandard errorCoefficient estimateStandard error
  1. Notes: Dependent variable is the log of the real wage (mean wage in contract). Excluded reference category is presented in square brackets. The specification also includes dummy variables for jurisdictions that have legislation for conciliation, cooling-off periods, mandatory strike votes, employer-initiated strike votes, reinstatement rights, compulsory dues check-off, bans on replacement workers, negotiated reopeners and technical reopeners, as well as dummies for year, and controls for average real provincial wage, log of previous contract duration, log of number of members in bargaining unit and inverse of unemployment rate, which are not presented. The ability to pay was introduced as a criterion in arbitration hearings in a few jurisdictions and sectors. These include Newfoundland, police (1992); Quebec, municipal workers (1993); Quebec, fire and police (1996); Ontario, all public sector workers (1996); Manitoba, teachers (1996); and, Federal, all public sector workers (2003). The first set of estimates excludes all of these sectors from the analysis. Standard errors are clustered by jurisdiction and year.

  2. *** denotes statistical significance at the 1 per cent level; ** denotes statistical significance at the 5 per cent level; * denotes statistical significance at the 10 per cent level.

  3. NDP, New Democrat Party.

Public-sector designation [Right-to-strike]
Choice of procedure−0.027*0.015−0.027*0.016−0.0190.015−0.0200.016
Compulsory arbitration0.0140.0100.0140.0100.024**0.0100.025**0.011
Essential service designation−0.023***0.009−0.025***0.009−0.017**0.008−0.020**0.008
Previous contract outcomes
Strike0.010**0.0050.012**0.0050.012**0.0050.014**0.006
Conciliation/mediation0.003*0.0020.0020.0020.005**0.0020.0030.002
Arbitration0.0060.0040.008*0.0050.010**0.0040.012**0.005
Legislated settlements
Legislated settlement in current contract−0.0070.007−0.0060.007−0.0010.0080.0000.008
Legislated settlement in previous contract−0.018***0.006−0.017***0.007−0.019***0.007−0.018***0.007
Political party in power in jurisdiction [Liberal majority]
Liberal minority0.0050.0090.0140.0090.0050.0090.016*0.009
Conservative majority−0.011**0.005−0.0050.006−0.011**0.005−0.0030.006
Conservative minority0.0080.0090.0180.0110.0020.0090.0130.011
NDP majority0.011**0.0050.015***0.0050.0090.0060.015**0.007
NDP minority0.0050.012−0.0030.0120.0070.0140.0040.014
Other0.0100.0100.0080.0100.0300.0130.028**0.012
Lagged log of real wage0.436***0.0280.435***0.0300.432***0.0320.423***0.035
Cost of living adjustment0.0050.0040.0040.0040.009*0.0050.0070.005
Number of contracts5,831 4,941 4,844 3,952 

The estimates on the controls for previous contract outcomes indicate that conciliation or mediation and strikes are associated with small increases in wage settlements that lie between 0.3 and 1.0 per cent, which are statistically significant at the 10 per cent level or better. Our estimate on the indicator variable for arbitration in the previous contract is 0.6 per cent and not statistically different from zero. Our estimates also indicate that legislated settlements are associated with a decrease in wage settlements. More specifically, a legislated settlement in the current contract is associated with a decrease of 0.7 per cent in wages, but it is not statistically different from zero. The lagged effects of a legislated settlement tend to be much larger. We find that legislated settlements in the previous contract are associated with a 1.8 per cent decrease in wage settlements. The large lagged effect could reflect that the union's bargaining position is weakened in successive rounds of bargaining by past legislated settlements, or that union negotiators are willing to accept smaller wage settlements in order to avoid a legislated settlement in the future rounds of bargaining.

Previous analyses (Currie and McConnell 1991; Gunderson et al. 1996; Mitchell 1988; Olson 1980; Stern et al. 1975) have highlighted that the compulsory arbitration legal structure was associated with an increase in wages relative to other legal structures. Our fixed-effect regression analysis of wage settlements indicates that there are no statistically significant differences between the compulsory arbitration legal structure relative to the right-to-strike legal structure. However, the estimates in the previous literature that found arbitration was associated with higher wage settlements relative to other legal structures were based on data that only considered contracts up until the late 1980s. As we have emphasized earlier, the public-sector landscape has changed considerably in the last few decades. In order to investigate the effects of the changed environment in the public sector on our estimates, we also estimate some additional specifications in Table 5 to explore some of the possible explanations for our results. First, some jurisdictions added greater emphasis on the ability to pay criterion used by arbitrators for some sectors in the public sector, which could have reduced the size of the wage increases awarded by arbitrators relative to those in other legal structures in the public sector.4 Second, the early 1990s in Canada was a period when almost all jurisdictions controlled the wages of workers in the public sector. These wage restraints would have limited the size of bargained wage settlements as well as the size of arbitration awards.

The second set of estimates in Table 5 uses the sample that omits the sectors that had changes in arbitration criterion regarding ability to pay. The estimates on the indicator for compulsory arbitration legal structure, the other indicators for legal structure as well as the estimates on the other control variables in the second column are very much like those in the first column of Table 5. Consequently, these results indicate that the estimates are not sensitive to changes in the ability to pay criterion in arbitration. One implication of this set of estimates is that the ability to pay criterion of arbitrators is not likely a key driver of the size of arbitration awards relative to settlements in the right-to-strike legal structure.5

The third set of estimates in Table 5 excludes data from 1990 to 1993, the period of wage restraint in the public sector in Canada. While the estimates on the choice of procedure and essential service designation are slightly smaller than those in the first column, the estimate on the compulsory arbitration dummy variable is very different. Excluding the early 1990s, compulsory arbitration is associated with a statistically significant 2.4 per cent increase in wages, relative to the right-to-strike legal structure. This estimate is similar to those in the earlier literature. The coefficient estimate on the dummy variable for a legislated settlement in the current contract is about −0.1 per cent and not statistically significant. Our coefficient estimate on the legislated settlement in the previous contract dummy is about −1.9 per cent and still statistically significant. The estimates on the other controls for previous contract outcomes are relatively similar to those we presented in the first column. These additional analyses suggest that the wage restraint in the public sector during the early 1990s plays a large factor in our arbitration results in the first column of Table 5. To the extent that wage restraint was the policy goal of this period, our results confirm its effectiveness.

The last set of estimates in Table 5 excludes the sectors that had changes in arbitration criteria and the period of wage restraint in the early 1990s from the sample. These estimates are very much like those for the sample that only excludes the period of wage restraint in the early 1990s, and also indicate that the estimates are not sensitive to changes in the ability to pay criterion in arbitration.

Currie and McConnell (1991) hypothesized that the wage premium from arbitrated settlements relative to other legal structures in the Canadian data was not driven by the ability to pay criterion. Consequently, by strengthening the ability to pay criteria in arbitration, legislators could have neutralized this wage premium in arbitrated settlements relative to the right-to-strike legal structure. However, our estimates indicate that strengthening the ability to pay did not have an effect on the arbitration wage premium. Moreover, our estimates suggest that the most effective way to reduce the size of the arbitration wage premium, relative to the right-to-strike legal structure, is to impose direct wage controls (i.e., legislate wage settlements or wage freezes).

In summary, our findings on the effects of compulsory arbitration on wages for the whole sample (1978–2008) in Table 5 appear to be driven by the period of wage restraint in the public sector during the early 1990s. Consequently, we believe that there is still a wage premium associated with the compulsory arbitration legal structure relative to the right-to-strike legal structure. This suggests that while the dispute costs associated with compulsory arbitration legal structure are lower than those in the right-to-strike legal structure (Currie and McConnell 1991, 1994), this is likely offset by larger settlement costs (i.e., wage increases) in the compulsory arbitration legal structure.

Another difference between our results and the earlier literature is with respect to the essential service designation. Currie and McConnell (1991) and Gunderson et al. (1996) found in their empirical analysis of public-sector contracts from Canada that an essential service designation did not reduce bargaining strength relative to a right-to-strike legal structure, that is it is ambiguous for wages. In contrast, we find that an essential services designation would be associated with a statistically significant decrease in wages relative to a right-to-strike legal structure. As noted by Gunderson (2002), the proportion of workers designated as essential has been on the rise for a few decades. Our estimates could reflect this change, that is as more workers in bargaining units have been designated as essential the bargaining power of these unions derived from a strike threat has eroded and this is reflected in lower wage settlements.

Finally, we also find that the estimates on the controls for the political party in power at time of contract settlement also tend to be relatively robust across the additional subsamples we consider in Table 5, although the precision of the estimates changes somewhat. Overall, Conservative majority governments at the time of contract settlement are associated with a 1 per cent decrease in wage settlements relative to a Liberal majority, and NDP majority governments are associated with a 1 per cent increase in wage settlements relative to Liberal majority governments. Recall that the Liberal party is more centrist in its orientation, while the Conservative and the NDP are to the right and left of the political spectrum.

6. Concluding remarks

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Collective bargaining legislation in the public sector and conceptual framework
  5. 3. Methodology
  6. 4. Data
  7. 5. Empirical results
  8. 6. Concluding remarks
  9. Acknowledgement
  10. References

We use data from 1978 to 2008 to study contract settlement outcomes, strike duration and wages (settlement costs) in the Canadian public sector.

Our analysis of the factors influencing the probability of how a contract is settled indicates that previous contract disputes and legislated settlements reduce the likelihood of bargained contracts in the current round of negotiations. This suggests that there could be some persistence in conflict, so that past difficulties, for example strikes and legislated settlements, could result in repeated disputes. We also find that arbitration is also associated with a decrease in the likelihood of a bargained settlement. Moreover, the use of arbitration tends to increase the reliance on arbitration in the future.

One of our most interesting results is the lack of a statistical significant estimate on the compulsory arbitration legal structure dummy variable in our wage analysis. Most of the previous literature examining the effects of collective bargaining in the public sector has found that the compulsory arbitration legal structure is associated with statistically significant increases in wage settlements relative to the right-to-strike legal structure as well as other legal structures in the public sector. However, we find that the insignificant estimate for compulsory arbitration is driven by the period of wage restraint in the Canadian public sector during the early 1990s. Consequently, we believe that the compulsory arbitration legal structure is still associated with a wage premium relative to the other legal structures in the public sector. Moreover, we also find that changes in arbitration that strengthened the ability to pay criterion used by arbitrators did not have an effect on the size of wage settlements. Consequently, we also conclude that wage restraint was more effectively achieved by legislated controls on the size of wage settlements and the size of arbitration awards than by strengthening the ability-to-pay criterion of arbitrators.

We also found that the essential services designation is associated with a statistically significant decrease in wages relative to the right-to-strike legal structure. This suggests that restricting a bargaining unit's ability to strike reduces its bargaining power. This finding differs from earlier studies, which found that the essential services designation was ambiguous in terms of its effect on wage settlements. However, there has been an increase in the percentage of workers designated as essential in this legal structure over the last few decades, which may have eroded union bargaining power. This finding is particularly important in light of the recent changes in public-sector bargaining in Canada (Bill C4 in 2013), which changed the manner in which public-sector workers in the Federal jurisdiction can be designated as essential and restricted access to arbitration (only bargaining units with more than 80 per cent of their workers designated as essential will be allowed access to arbitration).

The recession following the financial crisis of 2008 ushered in another period of restraint in the public sector. While our study does not cover this new period of restraint, it provides a new baseline to consider the effects of the various channels that can be used to control public-sector bargaining outcomes as well as their potential costs and benefits.

Acknowledgement

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Collective bargaining legislation in the public sector and conceptual framework
  5. 3. Methodology
  6. 4. Data
  7. 5. Empirical results
  8. 6. Concluding remarks
  9. Acknowledgement
  10. References

Campolieti gratefully acknowledges the financial support of SSHRC. An earlier version of this paper appeared as Dachis and Hebdon (2010).

Notes
  1. 1

    We summarize the legislative options for public-sector workers in a table that is available on request.

  2. 2

    This is based on our interpretation of labour legislation created by HRSDC Labour Law Analysis through the annual ‘Highlights of Major Developments in Labour Legislation’ going back to 1980 and Adell et al. (2001).

  3. 3

    Note that the negative relationships between the bargained outcome and mediation or conciliation are simply an artefact of our definition of bargained outcome.

  4. 4

    These changes included Newfoundland, police (1992); Quebec, municipal workers (1993); Quebec, fire and police (1996); Ontario, all public sector workers (1996); Manitoba, teachers (1996); and Federal, all public sector workers (2003).

  5. 5

    When we estimated a model that included an interaction term between the indicator variable for compulsory arbitration and a variable indicating the sectors that had changes in the ability to pay criterion, we obtain a positive estimate on the interaction term that is statistically significant at the 10 per cent level of significance. This estimate is consistent with the second set of estimates in Table 5.

References

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Collective bargaining legislation in the public sector and conceptual framework
  5. 3. Methodology
  6. 4. Data
  7. 5. Empirical results
  8. 6. Concluding remarks
  9. Acknowledgement
  10. References
  • Adell, B., Grant, M. and Ponak, A. (2001). Strikes in Essential Services. Kingston: Queen's University Press.
  • Beaumont, P. B. (1996). ‘Public sector industrial relations in Europe’. In D. Belman, M. Gunderson and D. Hyatt (eds.), Public Sector Employment in a Time of Transition. Madison, WI: Industrial Relations Research Association, pp. 283307.
  • Budd, J. (1996). ‘Canadian strike replacement legislation and collective bargaining: lessons for the United States’. Industrial Relations, 35: 245260.
  • Campolieti, M., Hebdon, R. and Hyatt, D. (2005). ‘Strike incidence and strike duration: some new evidence from Ontario’. Industrial and Labor Relations Review, 58: 610630.
  • Campolieti, M., Hebdon, R. and Dachis, B. (2014). ‘The impact of collective bargaining legislation on strike activity and wage settlements’. Industrial Relations, 53: 394429.
  • Cox, D. (1972). ‘Regression models and life tables’. Journal of the Royal Statistical Society Series B, 34: 187220.
  • Cramton, P., Gunderson, M. and Tracy, J. (1999). ‘The effect of collective bargaining legislation on strikes and wages’. Review of Economics and Statistics, 81: 475487.
  • Currie, J. and McConnell, S. (1991). ‘Collective bargaining in the public sector: the effect of legal structure on dispute costs and wages’. American Economic Review, 81: 693718.
  • Currie, J. and McConnell, S. (1994). ‘The impact of collective bargaining legislation in the U.S. public sector: no policy may be the worst policy’. Journal of Law and Economics, 37: 519547.
  • Dachis, B. and Hebdon, R. (2010). ‘The Laws of Unintended Consequence: The Effect of Labour Legislation on Wages and Strikes’. C.D. Howe Institute Commentary #304.
  • Ehrenberg, R. G. and Schwarz, J. L. (1987). ‘Public-sector labor markets’. In O. Ashenfelter and R. Layard (eds.), Handbook of Labor Economics, Vol. 2. Amsterdam: North-Holland, pp. 12191269.
  • Gunderson, M. (2002). ‘Collective bargaining and dispute resolution in the public sector’. In C. Dunn (ed.), Oxford Handbook of Canadian Public Administration. Oxford: Oxford University Press, pp. 517532.
  • Gunderson, M. and Melino, A. (1990). ‘The effects of public policy on strike duration’. Journal of Labor Economics, 8: 295316.
  • Gunderson, M. and Melino, A., Hebdon, R. and Hyatt, D. (1996). ‘Collective bargaining in the public sector: comment’. American Economic Review, 86: 315326.
  • Harrison, A. and Stewart, M. (1989). ‘Cyclical fluctuations in strike duration’. American Economic Review, 79: 827841.
  • Hebdon, R. and Kirkpatrick, I. (2005). ‘Changes in the organisation of public services and their effects on employment relations’. In D. Ackroyd, R. Batt and M. Tolbert (eds.), Oxford Handbook of Work and Organization. Oxford: Oxford University Press, pp. 531553.
  • Hebdon, R. and Kirkpatrick, I., Masters, M. and Slater, J. (2013). ‘U.S. Public Sector Collective Bargaining: Tumultuous Times’. In H.R. Stanger, P.F. Clark, A.C. Frost (eds.), Collective Bargaining Under Duress: Case Studies of Major US Industries. Ithaca, NY: Cornell University Press, pp. 255295.
  • Jalette, P. and Hebdon, R. (2012). ‘Unions and privatization: opening the black box’. Industrial and Labor Relations Review, 65: 1735.
  • Kennan, J. (1980). ‘Pareto optimality and the economics of strike duration’. Journal of Labor Research, 1: 7794.
  • Mitchell, D. (1988). ‘Collective bargaining and compensation in the public sector’. In B. Aaron, J. M. Najita and J. L. Stern (eds.), Public Sector Bargaining. Washington, DC: Bureau of National Affairs, pp. 124159.
  • Moulton, B. R. (1990). ‘An illustration of a pitfall in estimating the effects of aggregate variables on micro units’. Review of Economics and Statistics, 72: 334338.
  • Olson, C. (1980). ‘The impact of arbitration on the wage of firefighters’. Industrial Relations, 19: 325339.
  • Redder, M. and Neumann, G. R. (1980). ‘Conflict and contract: the case of strikes’. Journal of Political Economy, 88: 867888.
  • Siebert, W. S. and Addison, J. T. (1981). ‘Are strikes accidental?Economic Journal, 91: 389404.
  • Stern, J. L., Rehmus, C. M., Loewenberg, J. J., Kasper, H. and Dennis, B. D. (1975). Final Offer Arbitration: The Effects on Public Safety Employee Bargaining. Toronto, ON: D. C. Heath and Co.
  • Swimmer, G. (2001). ‘Public sector labour relations in an era of restraint and restructuring: an overview’. In G. Swimmer (ed.), Public Sector Labour Relations in an Era of Restraint and Restructuring. Don Mills, ON: Oxford University Press, pp. 135.
  • Thompson, M. (1995). ‘The industrial relations effects of privatization: evidence from Canada’. In G. Swimmer and M. Thompson (eds.), Public Sector Collective Bargaining in Canada: Beginning of the End or the End of the Beginning. Kingston: IRC Press, pp. 164179.