Judicial arbitration of unfair dismissal cases: The role of peer effects
Introduction
What factors, other than the intrinsic characteristics of cases, influence the decisions of judges arbitrating unfair dismissal cases? Answering this vexed question matters a great deal, not only for the parties involved in a dismissal dispute but also for policy makers evaluating how legislation designed to provide basic workers’ rights is enforced in practice. In the context of Australian labour courts, our previous research (Freyens and Gong (2017)) considered the influence of social values on judicial decisions. Specifically, we examined how judges’ own social values (proxied by their past history as labour lawyers) together with the values of the political party appointing them affected their decisions. We examined the extent to which these factors related to judges’ propensity to decide cases in favour of the plaintiffs (dismissed employees). We then interacted these values with the social values embodied in the law (legislated statutes and their reforms). We found significant effects from all three channels, particularly after the so-called WorkChoices reforms, which made access to unfair dismissal remedies more difficult for dismissed employees. Ceteris paribus, judicial decisions in dismissal cases varied significantly with (and were predictably signed by) the social values of judicial appointees and political appointors. Judges whose social values were unaligned with the WorkChoices reforms responded orthogonally, i.e. by meting out decisions that countered the reforms. In other words, we found significant evidence that labour court judges behave as activists.
In the present work we push our investigation further to check for the influence of other judges on individual decisions (peer effects). Our objective is to determine whether labour court judges operating in the same jurisdiction (State-based panels) influence one another's decisions. If spatial correlation among decisions was established, we could potentially infer that the significant political activism found in our previous study may spillover onto the behaviour of non-activist judges, providing an expansion channel for the spread of activism in labour courts. The presence of strong peer effects would have substantial implications for policy designs aimed at countering political activism in labour courts. For instance, attempts to take judicial nomination and appointment prerogatives away from the executive branch of government1 [64] would presumably be more effective if courtroom decisions are influenced by independently-minded peer groups rather than by vested interest and ideologically-motivated politicians.
There is much that researchers can do to isolate peer effects and we review a number of approaches and findings in our literature review. The spatial econometrics methodology pioneered by Manski (1993) is by far the most popular method, although it has too rarely been used to examine judicial decisions, and its scope for ill-advised attributions of causality has been a recurrent subject of controversy, which we will also touch on (Hanushek et al. (2003), Angrist (2014)). The Manski approach Manski (1993) rests on a decomposition of peer effects into two broad types of hypothesized interactions: endogenous (behavioural) interactions and exogenous (type) interactions. Endogenous effects arise when individual behaviour depends on the average (aggregate) behaviour of the group: in our context, individual judges may decide more (or less) often in favour of dismissed employees because that is what their peers do. Exogenous effects arise when individual behaviour depends on the exogenous characteristics of their reference group: individual judges may decide more often in favour of dismissed employees not because their peers do so but because the composition of their peer group is exclusively made out of former union lawyers. Endogenous effects are of particular interest because they can give rise to ‘social multipliers’: an intervention that alters behaviour in the reference group will tend to be more effective due to behavioural spillover effects within the reference group. Identifying social interactions with spatial econometrics methods is fraught with several difficulties that have to be overcome: there can be selection bias2 through endogenous group formation (e.g. if judges are only appointed when they have previously worked with currently serving judges), or there could be correlated effects that arise when members of a reference group behave similarly because of an unobserved institutional feature (e.g. if all judges had been instructed by the same expert). We will test for these potential issues and show why they do not present identification problems for our study (the autonomous political appointment process largely rules out the selection effect). However, the largest hurdle to overcome is addressing the so-called ‘reflection effect’, which arises from the simultaneous and reciprocal nature of the relation between outcome (individual judicial decision) and regressor (average or aggregate peer decisions). In our estimation section we present our approach to dealing with both the correlated effect issue (omitted variables) and the reflection problem.
In this paper, we use spatial econometric methods to analyze a panel of 82 judges and a data set of 2223 unfair dismissal cases arbitrated over a 15 year period. We test for the presence of endogenous and exogenous effects in judicial decisions. In doing so, we will address a number of the potential methodological issues that could invalidate our results, such as the need to establish the conditions of a randomized experiment and to address the reflection problem. Section 2 reviews the academic literature on the drivers of judicial decisions-making in employment disputes, with a particular focus on peer effects. Section 3 presents the context for this research, and the data used for the analysis. Section 4 presents a set of preliminary tests performed on the data to ensure it is suitable for use in our spatial econometric model. Section 5 presents our model specification, our estimation strategy, and our results, whereas the last section presents our conclusions.
Section snippets
Arbitration of employment disputes
There is a considerable literature examining the determinants of judicial behaviour particularly in the United States where this strand of research originated in the post-war period. This literature, published in law, economics, industrial relations and political science journals, focuses on establishing the likely personal and professional motivations for judges and arbitrators to decide cases and how changes in institutional design, political context, or socio-economic circumstances may
Institutional background
The Fair Work Commission (FWC) is the labour court in charge of conciliating and arbitrating a range of labour market disputes in Australia. It is divided into ten panels, one of the most prominent ones being the Termination of Employment Panel (TEP), which is composed of six State jurisdictions: Victoria, New South Wales, Queensland, South Australia, Western Australia and the Australian Capital Territory. The President of the Commission is a Federal Court judge holding the title of Justice.
Data set
Our data is collected from electronic transcripts documenting the decisions of labour courts (Fair Work Australia, FWA, and its predecessor, the Australian Industrial Relations Commission – AIRC) in unfair dismissal disputes. Transcripts are public domain information, which record factual information about the parties’ background and their respective allegations. Transcripts of decisions report the testimonies of witnesses and the outcome of the case. The information is sometimes difficult to
cases selection
We start by addressing two preliminary issues related to our use of tried cases: testing for selection effects and establishing the conditions of a randomized experiment. The selection effect (Priest and Klein (1984)) holds that tried cases are more complex than the mass of settled cases because settlement weeds out cases where the evidence compellingly backs the claims of one party to the dispute. Only those cases with conflicting and inconclusive evidence then proceed to trial. If this is the
Model specification
We have a panel of commissioners allocated multiple cases and making multiple decisions each year over a 15 years period. We assume that commissioners’ decisions are determined by the characteristics of the cases, of the disputing parties, and of the commissioners themselves. The outcome of the ith case in year judged by commissioner of panel is given by Y {0,1} (with Y=1 indicating that the employee wins the case). The outcome
Conclusions
We used a panel of Australian labour court judges arbitrating a large number of unfair dismissal disputes over a decade and a half to test for the presence of peer effects in their decisions. Using a spatial econometrics approach, we found no evidence of endogenous interactions: peer behaviour does not influence individual decisions. However, we found significant exogenous interactions: peers’ characteristics influence individual decisions. It would seem then that Labour court judges are not
Competing interests
The authors have no conflict of interest to declare.
Acknowledgments
We gratefully acknowledge seed funding from University of Wollongong and University of Canberra, which have enabled data collection efforts. We also thank two anonymous referees for helpful comments on a previous version of this paper.
References (78)
The perils of peer effects
Lab. Econ.
(2014)- et al.
Estimating models of complex FDI: are there third-country effects?
J. Econom.
(2007) - et al.
Labor courts, nomination bias, and unemployment in Germany
Eur. J. Polit. Econ.
(2011) - et al.
Peer effects in decision-making: evidence from corporate investment?
China J. Account. Res.
(2017) - et al.
Judicial decision making under changing legal standards: the case of dismissal arbitration
J. Econ. Behav. Org.
(2017) - et al.
The political economy of supreme court constitutional decisions: The case of roosevelt's court-packing plan?
Int. Rev. Law Econ.
(1992) - et al.
Are judges biased by labor market conditions?
Eur. Econ. Rev.
(2003) - et al.
Reputation and judicial decision-making?
J. Econ. Behav. Org.
(1994) Peer effects in education: how might they work, how big are they and how much do we know thus far?
(2011)- et al.
Peer effects in adolescent overweight?
J. Health Econ.
(2008)