Judicial Independence, Judicial Promotion and the ... - Springer Link

4 downloads 58 Views 126KB Size Report
defines substantive independence as the situation where judges do not decide .... tures are not overturned just because the “political winds” have changed.
European Journal of Law and Economics, 17: 209–235, 2004 c 2004 Kluwer Academic Publishers. Manufactured in The Netherlands. 

Judicial Independence, Judicial Promotion and the Enforcement of Legislative Wealth Transfers—An Empirical Study of the New Zealand High Court PUSHKAR MAITRA RUSSELL SMYTH [email protected] Department of Economics, P.O. Box 11E, Monash University, 3800 Victoria, Australia

Abstract This paper investigates the Landes-Posner thesis on judicial independence using data on public law decisions in which the government was the defendant decided in the New Zealand High Court over the period 1958–2001. We use survival analysis to examine whether successive New Zealand governments have promoted judges from the High Court to the Court of Appeal (which stands above the High Court) on the basis of political considerations, the quality of the judge’s decision-making or both. Our findings suggest that the quality of decision-making has generally been important. Consistent with the weak form of the Landes-Posner hypothesis we find no evidence that governments have used their powers to punish judges who decided cases against them. On the contrary, we find some support for the strong form of the Landes-Posner thesis that governments positively use their powers to secure judicial independence. Keywords: judicial independence, judicial promotion, interest groups, New Zealand JEL Classification: K2, K4

Since Landes and Posner (1975) first suggested that governments value judicial independence because it makes their promises to interest groups with which they contract credible, there have been several empirical studies of their argument for United States’ courts. Among these Anderson, Shughart and Tollison (1989) examine whether the legislature exercises its powers over judicial remuneration in order to motivate judges to act independently. Toma (1991) studies whether the legislature uses its control over the courts’ budget to reward or penalize judges for their decisions. Hanssen (1999, 2000) explores the relative independence of courts with appointed judges and elected judges and the effect this has on the rate of litigation. Cohen (1991) investigates judicial decision-making within the context of institutional and political constraints using cases deciding the constitutionality of the United States Sentencing Commission. Meanwhile other studies have examined different aspects of the LandesPosner interest group perspective through focusing more on the activities of the executive and legislature (see e.g. Crain and Tollison, 1979, 1979a; Crain, Shughart and Tollison, 1988). There have, however, been few studies of the Landes-Posner hypothesis for courts outside the United States. Ramseyer (1994) discusses the Landes-Posner thesis in the context of a review of judicial independence in Japan and the United States, while Salzberger (1993) considers the Landes-Posner argument in a comparative review of judicial independence

210

MAITRA AND SMYTH

in Israel, the United States and the United Kingdom. These studies, though, do not utilize rigorous statistical testing. The only systematic multivariate studies of judicial independence for courts outside the United States, of which we are aware, are a series of studies by Ramseyer and Rasmusen (1997, 1999, 2001a, 2001b) for Japanese courts and a study by Salzberger and Fenn (1999) for the English Court of Appeal. The lack of studies investigating the interest group perspective on judicial independence in courts outside the United States is symptomatic of a more general problem with the law and economics literature, which is that there is a general dearth of empirical studies for non-United States courts. As a result, “most theory and data developed by social scientists for understanding legal systems still remain very much the product of, and thus bound to, the inevitable peculiarities of the U.S. context” (Atkins, 1991). This paper seeks to partially redress this situation through using the same methodology as employed by Salzberger and Fenn (1999), to investigate whether successive New Zealand governments have promoted judges on the basis of political considerations, the quality of the judge’s decision-making or both. Dreyfuss (1995) suggests that the quality of judicial decision-making encompasses three inter-related concepts: accuracy, precision and coherence. We adopt this characterisation of judicial quality here. Accuracy refers to the extent to which the law produces the objectively correct result. Precision refers to the extent to which the court reaches the same result in equivalent cases. Coherence refers to the extent to which new decisions demonstrate continuity with prior law, rather than veering in unpredictable directions. The data that we use are public law decisions of judges in the New Zealand High Court (“the High Court”) over the period 1958–2001 in cases in which the government was the defendant. We investigate what effect the propensity of High Court judges to decide cases against the government has had on their chances of promotion to the New Zealand Court of Appeal (“the Court of Appeal”), which stands immediately above the High Court. In theory, Court of Appeal and High Court judges are appointed by the Governor-General in the name of the Queen. In practice, however, the Governor-General rubber-stamps the decision of the government. Thus, similar to the process in the United Kingdom, the procedure for promoting judges from the High Court to the Court of Appeal is under direct political control. New Zealand is one of the few countries in the world without a written constitution. This means that the role of the courts in New Zealand differ from the United States, because in New Zealand the ultimate source of constitutional power is the elected parliament. With no written constitution any decision the courts do make in New Zealand can always be overridden by an Act of parliament, however politically difficult it might be for the legislature to do so in practice (Allan, 1998). The Landes-Posner thesis was developed for the United States where there is a written constitution. And traditionally the question of the independence of the courts has been answered with reference to the formal institutional rules found in the constitution (see Stephenson, 2003, pp. 59–60). However, Salzberger (1993) shows that Israel and the United Kingdom which, like New Zealand, do not have written constitutions give their judiciaries more independence than is required by the relevant formal restrictions.1 Salzberger (1993) explains this through distinguishing between structural and substantive independence. Salzberger (1993, p. 352) defines substantive independence as the situation where judges do not decide cases according

AN EMPIRICAL STUDY OF THE NEW ZEALAND HIGH COURT

211

to the legislature’s or government’s will, while structural independence refers to the institutional arrangements that enable the existence of substantive independence. Because in New Zealand and the United Kingdom there are no restrictions on the legislative powers of parliament, the courts have a lower level of structural independence than in the United States. In the United States the answer to the question why do judges enjoy a high level of substantive independence might well be because there is a high level of structural independence which facilitates it. As Salzberger (1993, p. 352) puts it, “an American is likely to reply ‘we have an independent judiciary because we have a written constitution in which judicial independence is guaranteed”’. In contrast in New Zealand or the United Kingdom the question is why does Parliament, to whom the courts are firmly subordinate, choose to maintain a separate and independent branch? This makes countries such as New Zealand and the United Kingdom more interesting case studies of the Landes-Posner thesis than the United States because the focus is squarely on examining why it is that successive governments value substantive judicial independence when there are relatively few safeguards guaranteeing structural independence. The issue of how structural and substantive independence interact is topical in New Zealand for three reasons. First, while judges’ decisions have generally been free from controversy, there have been recent instances, which have been well publicized in New Zealand, where members of the Executive have openly criticized judges. For example, on two separate occasions in 1995 one Cabinet minister, John Banks, was reprimanded for criticizing judges of the District Court, which sits below the High Court, for giving what he considered lenient sentences in criminal trials.2 In one outburst on radio talkback he claimed most District Court judges “are second-rate lawyers that couldn’t make, can’t make and haven’t made a living in private practice and have little to offer the judicial system in this country” (quoted in Stockley, 1997, p. 167). The Chief Judge of the District Court wrote to the Attorney-General in protest and Banks issued a written apology.3 However, less than a month later Banks criticized District Court judge Richard Bollard on radio talkback for giving a lenient sentence to a Maori activist who had attacked a landmark tree in Auckland. Banks stated sarcastically of Bollard on radio, “So this outstanding individual that sits on the District Court in Auckland, Judge Bollard, a truly special New Zealander, gave a severe telling-off to the One Tree vandal and sentenced him to a dreadful punishment of six months periodic detention”.4 The Chief Justice of New Zealand protested to the Prime Minister complaining that Bank’s comments were a political attack on the independence of the judiciary and the Attorney-General later characterized Bank’s comments as unacceptable.5 A more recent high profile example concerned a decision of the Environmental Court relating to the subdivision of rural land in Queenstown, which was handed down in 2001. Before the Court’s final ruling, the Mayor of Queenstown, Warren Cooper, who was a former member of Parliament and Cabinet Minister, publicly referred to the decision as “absolutely unacceptable” and “Alice In Wonderland stuff”.6 He proceeded to state: “This will be a veritable gravy train for the lawyers . . . .. In 35 years of experience in central and local government, this decision takes first prize for the most dopey and illconceived solution to the problem of housing in rural areas”.7 Cooper’s comments resulted in a political furore and were regarded as an attempt to put political pressure on the Court (see Sutton and Bigwood, 2001).

212

MAITRA AND SMYTH

A second reason the issue of judicial independence is topical in New Zealand is that the government has indicated that it tends to overhaul the procedure for appointing and dismissing judges and dealing with judicial misbehavior. These issues were put on the agenda after a High Court judge, Justice Robert Fisher, was found to be viewing pornographic videos on his office computer in 2002. In the ensuing debate about what sanctions he should receive the government concluded that there was no correct procedure in place for dealing with such misbehavior. Under the proposed reforms, which are expected to be introduced into the Parliament in late 2003, a Judicial Appointments and Liaison Office would be established to advise the government on suitable judicial appointments and a Judicial Conduct Commissioner would be appointed to filter complaints against judges. With the Attorney-General’s approval, the commissioner could activate a Judicial Conduct Panel to consider a complaint; however, the final right to remove a judge whether for misbehavior or incapacity would rest with Parliament.8 The government has argued that the reforms will increase structural independence through creating agencies which provide checks and balances, but media commentary has been critical. For example, according to one editorial: “Such agencies . . . are not necessarily a check or balance. Their view will merely reflect the view of those who appoint them—if that is the desired outcome. Worse still, panels and suchlike can provide a ready-made cover. A government has to defend a judicial appointment made by itself, but can deflect the flak if that appointment is made by an agency”.9 Third, currently the highest court in New Zealand is the Judicial Committee of the Privy Council, composed of English Law Lords and other senior judges from British Commonwealth countries, which sits in London. The New Zealand government has introduced new legislation, the Supreme Court Bill, which is currently before the Parliament, to replace the Judicial Committee in the New Zealand judicial hierarchy with a Supreme Court of New Zealand, which would sit above the Court of Appeal. The main motivation for replacing the Judicial Committee is nationalistic. While the Judicial Committee was once the final court of appeal for several British Commonwealth countries including Australia and Canada, apart from New Zealand, only a few small countries retain it as their highest court. The New Zealand government sees replacing the Judicial Committee as a sign that along with countries such as Australia and Canada, New Zealand has “graduated from behind Britain’s founding legal skirts”.10 However, while it is inevitable that the Judicial Committee will be replaced, the Supreme Court Bill has been criticized on several fronts. One argument is that the quality of judicial decision making will suffer (Farmer, 2001; Auckland District Law Society, 2003). Farmer (2001, p. 238) suggests, “it is na¨ıve to expect that a small country like New Zealand can consistently match the quality of judges at the highest level in the United Kingdom, given the much greater size of the English Bar from which judges are drawn” (emphasis in original). The Supreme Court Bill proposes a new five-judge panel to be appointed by the government. The critics of this process argue that the political party which happens to be in power at the time will be able to select all of the judges of the highest court, therefore locking-in future generations. This differs from what happened in Australia and Canada when appeals to the Judicial Committee were abolished because the then existing highest domestic court became the highest court.11

AN EMPIRICAL STUDY OF THE NEW ZEALAND HIGH COURT

213

The paper is set out as follows. The next section reviews the conceptual arguments regarding whether governments value judicial independence. Section 2 discusses judicial independence in New Zealand in more detail and compares it with judicial independence in the United Kingdom. The empirical study is presented in section 3. Initially we provide an overview of the data, formulate the hypotheses to be tested and describe the variables used to explain duration to promotion. Following this we discuss the econometric methodology, present the results and compare the findings with those of Salzberger and Fenn (1999) for the English Court of Appeal. Finally Section 4 concludes. The main contribution of the paper is twofold. 1. Foreshadowing the main findings of the paper, we find no evidence that governments have used their powers to punish judges who decided cases against them. In contrast to Salzberger and Fenn’s results for the English Court of Appeal, this extends to judges who reverse lower court decisions that were initially in the government’s favour. Moreover, contrary to Salzberger and Fenn, we find some support for the strong form of the Landes-Posner thesis that governments positively use their powers to secure judicial independence. 2. The second contribution of the paper is more methodological. We use a Weibull distribution to parameterize the baseline hazard, which facilitates comparisons with Salzberger and Fenn’s findings for the English Court of Appeal. The use of the Weibull Distribution to characterise the baseline hazard is dictated by the fact that it gives the best fit (in terms of the lowest Akaike Information Criterion (AIC)). We also examine the robustness of the results using a Cox proportional hazard regression model where we do not make any distributional assumption regarding the baseline hazard. The results from the Cox proportional hazard model are weaker in terms of statistical significance, but the signs on the variables of interest are still generally the same and in this sense the results are robust. 1.

Conceptual perspectives on judicial independence

Landes and Posner’s (1975) basic argument is that governments value judicial independence, rather than see it as a constraint on government power. They argue that governments value judicial independence because an independent judiciary will resolve disputes in terms of the intent of the enacting legislature. Thus, legislative agreements enacted by previous legislatures are not overturned just because the “political winds” have changed. This role provides greater stability for long-term contracts between the legislature and interest groups and therefore increases the durability and, hence, present value of legislative wealth transfers. Other arguments have since been put forward to explain why governments might value judicial independence. One strand of literature suggests that judicial independence allows governments to minimize the risks associated with political competition. Ramseyer (1994) argues that although the government could increase its power while in office if it constrained judges that power would come at a cost. What the government does to the opposition, the opposition might do to them when it takes office. This can be interpreted as a cooperative equilibrium of an infinitely repeated game between two political parties, where each

214

MAITRA AND SMYTH

promotes judges from the other side in order to smooth their policy consumption (and that of their voters).12 Stephenson (2003) builds on the Ramseyer (1994) argument to propose a simple formal model where independent judicial review is only sustainable where the political system is sufficiently competitive, judicial doctrine is sufficiently moderate and the parties are sufficiently risk adverse and forward looking. A second strand of literature analyses judicial independence as a means for political decision-makers to control their bureaucracies (see e.g. McCubbins and Schwartz, 1984; McCubbins, Noll and Weingast, 1987; Ferejohn and Shipan, 1990; Ferejohn and Weingast, 1992). The issue this literature addresses is how does a representative democracy ensure that policy preferences are responsive to the preferences of citizens in circumstances when much delegated policy-making is done by unelected bureaucrats and elected officials have limited resources to monitor the activities of agencies. McCubbins and Schwartz (1984) and related studies argue that an independent judiciary helps the government to police the bureaucracy. By giving citizens the right to sue misbehaving bureaucrats through independent tribunals, the government can obtain information about how the bureaucrats are performing. This information can be used to provide incentives to improve bureaucratic performance and thus improve the government’s electoral advantage. A third explanation for why governments value an independent judiciary is that independent courts help to shift the blame for unpopular decisions on to the courts (Salzberger, 1993). However, as Stephenson (2003, p. 63) notes, a problem with the blame deflection explanation is that it assumes that the public is sophisticated enough to observe that the courts rather than the government made important policy decisions, while the public is not sophisticated enough to realize that the government could impede independence if it chose too. This criticism is particularly valid in countries such as New Zealand where the courts have little structural independence. Another strand of the literature suggests that government support for independent courts reflects public support for judicial independence and governments, which impair judicial independence, will suffer at the polls. For instance, Vanberg (2001) develops a game-theoretic model where governments that do not comply with the decisions of courts will suffer voter backlash at the next election. Thus, a fourth, even simpler story than blame deflection is that promoting judges from your own side of politics is politically unpopular—i.e. the opposition might highlight and publicize such appointments, leading to electoral costs. There might on the other hand be reasons why governments might want to curtail the independence of courts. Ramseyer and Rasmusen (1997) argue that politicians do not maximize votes simply through making their promises credible to interest groups and others in the electorate, but rather they also need to deliver on their promises. An independent judiciary can make it more difficult to deliver promises if it holds the relevant legislation invalid. The Landes-Posner thesis assumes that politicians are concerned with long-term credibility. In reality, politicians are likely to be myopic, focusing instead on the electoral cycle. This creates temptations for governments to cheat on judicial independence in pursuit of short-term electoral advantage. Hanssen (2001) develops and tests a model in which incumbent politicians choose the level of judicial independence by weighing up the gains from the increased durability of their preferred policy choices against the costs of policymaking by the judiciary. He finds that

AN EMPIRICAL STUDY OF THE NEW ZEALAND HIGH COURT

215

firmer single party control is associated with a less independent judiciary, closer competition between political parties with a more independent judiciary, greater differences in party policy positions with a more independent judiciary, and greater differences between the policy preferences of judges and politicians with attempts to curtail judicial independence. Rogers (2001) makes a related argument focusing on the informational advantages courts can provide to legislatures. Similar to McCubbins and Schwartz (1984) and the literature on using the courts to monitor agencies, Rogers (2001) argues that independent courts can provide informational benefits to legislatures. However, to realize these benefits policyoriented legislatures will face a trade-off where the court’s preferences diverge from those of the government. The conclusion from Rogers’ (2001) model is that the informational benefits to legislatures of independent courts will often be worth the policy costs when legislation is struck down. 2. 2.1.

Judicial independence in New Zealand Constitutional arrangements in New Zealand

As discussed in the introduction New Zealand does not have a written constitution. The main reason for this is that New Zealand’s constitutional heritage is English and the United Kingdom does not have a written constitution. The origins of parliamentary sovereignty in New Zealand are contained in the New Zealand Constitution Act (1852) passed by the United Kingdom parliament, which created a General Assembly for the colony of New Zealand to “make laws for the peace, order and good government of New Zealand” (s. 53). Subsequently, the Legislature Act (1908) transferred the privileges, immunities and powers of the United Kingdom parliament as at 1 January 1865 to the New Zealand parliament (Allan, 1998; Joseph, 1993; Palmer, 2000). Some other former British dependencies such as Australia and Canada have written constitutions. This is because, unlike New Zealand which is relatively small, the size of both countries necessitated a federal system of government, which in turn required a written document setting out, as higher law, the division of power between federal and state (Allan, 1998). In illustrating his distinction between substantive and structural independence, Salzberger (1993, pp. 352–353) makes two main points. First, there is no legal system which guarantees full structural independence. Second, there is always a gap between the degree of structural dependency and substantive independence in favour of the latter. While New Zealand does not have a formal written constitution it does have written documents and processes designed to provide structural independence to the courts (see Allan, 1998). Foremost among these is the Judicature Act (1908), which deals with the operation of the courts and the Constitution Act (1986) which is declaratory of New Zealand’s existing laws and institutions and includes, among other provisions, protection for judges against arbitrary removal from office. Section 13 of the Judicature Act sets the retirement age for a High Court judge at 68, which guarantees security of tenure until that age and prevents the executive from removing “troublesome” judges by forced early retirement (Goldfinch, 1993, p. 157).13 Section 23 of the Constitution Act (1986) provides that a judge of the High Court cannot be removed from office except by the Sovereign or the Governor General acting on the address of the House of Representatives.

216

MAITRA AND SMYTH

The address can only be moved on the grounds of that judge’s misbehavior or incapacity to discharge the functions of office. However, the Constitution Act (1986) is not entrenched. It is an ordinary Act of parliament and, as such, it can be repealed by a simple majority. One potentially important difference between New Zealand and the United Kingdom is that for the last decade of the study New Zealand has had a United States style Bill of Rights. The Bill of Rights Act (1990) is a statutory Bill of Rights which sets out, in a broad fashion, the place of the individual citizen vis-`a-vis state power. The various freedoms and rights proclaimed include electoral rights, freedom of expression, freedom of movement, freedom of religion and the right not to be deprived of life. The Act indirectly recognizes judicial independence. Section 25 of the Bill of Rights Act (1990) states that every person charged with an offence has the right to “a fair and public hearing by an independent and impartial court”. The Bill of Rights has facilitated a cathartic process in the New Zealand courts where they have become more policy-oriented. One prominent New Zealand legal academic has dubbed this development, “the legislative thesis”, the main characteristic of which is that judges have become more willing to make the law (Watts, 2001). The most vocal New Zealand judge advocating a readjustment of the role of the courts and parliament is Court of Appeal judge, Justice E.W. Thomas. Justice Thomas has a well-articulated view of the role of the courts and the proper relationship between the courts and the legislature based on Jaffe’s (1969) notion of positive jurisprudence (see e.g. Thomas, 1999, 2000, 2001).14 While Justice Thomas explicitly states he is not questioning the doctrine of parliamentary sovereignty, he argues that justice (or fairness) should be the dominant objective of any developed system of law, which gives the courts a wide scope for judicial review. While the Bill of Rights Act (1990) has given the courts more structural independence to judicially legislate, similar to the Constitution Act (1986) and unlike the United States Bill of Rights, it only has the status of an ordinary Act of parliament and can be repealed by parliament. 2.2.

Judicial appointment process

The system of judicial appointments in New Zealand and the United Kingdom are similar. In both instances, in Salzberger’s (1993) terminology, they are structurally dependent on the Parliament, or more accurately the government. In the United Kingdom, the judges of the Court of Appeal are appointed by the Queen on the advice of the Prime Minister after consultation with the Lord Chancellor. The Lord Chancellor, who has the real power to appoint judges is a member of the Cabinet and is a political appointment (Salzberger and Fenn, 1999, pp. 834–835). In New Zealand, judges of the High Court are appointed by “the Governor-General in the name and on behalf of Her Majesty” (s. 4(2) Judicature Act (1908)). In practice, the Governor-General, as the representative of the Queen, appoints the Chief Justice on the recommendation of the Prime Minister and the High Court and Court of Appeal judges on the recommendation of the Attorney-General (see Goldfinch, 1993; Stockley, 1997). In New Zealand, the Attorney-General, who serves the political role of the Lord Chancellor in Britain,15 and the Prime Minister hold the real power to appoint judges. Thus analogous to the system of judicial appointments in the United Kingdom, judicial appointments are under direct political control.

AN EMPIRICAL STUDY OF THE NEW ZEALAND HIGH COURT

217

The government is under no obligation to consult with interested parties before selecting candidates. The view has been expressed that the consultation process in New Zealand is narrower than in the United Kingdom (Hammond, 1991, p. 98). Sir Geoffrey Palmer who was Attorney-General from 1984 to 1989 and Prime Minister in 1989–90 records that he consulted widely with the Solicitor-General, the Chief Justice, President of the Court of Appeal and President of the Law Society before recommending judicial appointments (Palmer, 1995, p. 43). However, anecdotal evidence suggests that the level of consultation varies with the incumbent in office. For instance, Sir Thomas Eichelbaum, who is the immediate past Chief Justice of New Zealand and who served with six different AttorneysGeneral is on the record as stating that the nature of the consultative process varied widely from one Attorney-General to the next (Eichelbaum, 1993, p. 92). 2.3.

Judicial remuneration

The process for setting judicial remuneration in New Zealand provides more structural independence to the judiciary than in the United Kingdom. In the United Kingdom judicial salaries are determined by the Lord Chancellor with the concurrence of the Minister for Civil Service. The only structural safeguard is that judicial salaries cannot be reduced (Salzberger, 1993, p. 353). In New Zealand judicial salaries are determined by the Higher Salaries Commission, which is an independent statutory body (s. 9A Judicature Act (1908)). Its determinations are binding and thus can only be reversed by the legislature through enacting legislation (Winterton, 1995, p. 91). Section 24 of the Constitution Act (1986) prohibits reducing the salary of a judge of the High Court or Court of Appeal while he/she is in office. As Goldfinch (1993, p. 157) puts it, these provisions “remove salary determination from the hands of the executive or legislature and insulates the judiciary from indirect attacks upon their salary which could be carried out by a hostile legislature or executive through allowing salaries to erode with inflation”. 2.4.

Judicial promotion from the High Court to the Court of Appeal

The High Court is the second highest court in New Zealand. Appeals from the High Court are to the Court of Appeal, which is the highest court located in New Zealand. Before 1958 the Court of Appeal was staffed by High Court judges sitting in turn, but since 1958 the Court of Appeal has functioned as a permanent court with its own judges. Until the Supreme Court Act is passed, the Judicial Committee continues to stand above the Court of Appeal in the New Zealand court hierarchy as the highest New Zealand court, but, in part because of the cost, appeals to the Judicial Committee have been relatively rare. Less than 1 per cent of unsuccessful litigants in the Court of Appeal, appeal to the Judicial Committee.16 Given the small number of appeals to the Judicial Committee, the Court of Appeal increasingly regards itself as a final court of appeal for most purposes. Sir Ivor Richardson (1981), the immediate past President of the Court of Appeal, states: “Our court is an intermediate court of appeal. However, there have been only some thirty-five appeals to the Privy Council in the last twenty years. Thus, for practical purposes, the Court of Appeal is the court of last resort and exercises oversight of the administration of justice in New Zealand”. As

218

MAITRA AND SMYTH

discussed in the introduction the Judicial Committee is in the process of being replaced by a Supreme Court, to be located in New Zealand, as part of legislation which is before the New Zealand parliament. Similar to the situation in the United Kingdom, the New Zealand judiciary is largely a career-based judiciary. Most Court of Appeal judges are promoted from the High Court. Since the permanent Court of Appeal was established in 1958, 18 of the 23 judges appointed to the Court of Appeal were promoted from the High Court.17 The other Court of Appeal judges have been appointed straight from the Bar. In almost all cases, the Court of Appeal represents the pinnacle of a New Zealand judge’s career. There has only been one case in more than 40 years where a judge has been promoted from the Court of Appeal to a higher judicial office. The one case is Sir Robin Cooke who was President of the New Zealand Court of Appeal, before being bestowed the title Lord Cooke of Thorndon and being elevated to the House of Lords in London. As Salzberger (1993, p. 354) emphasizes the fact that judicial appointment and promotion is under direct political control has even more potential effect on judicial independence in light of the fact that the British and New Zealand judiciaries are career-based judiciaries. Up until 1985 High Court judges received the same remuneration as judges of the Court of Appeal in New Zealand. This reflected the practice in the United Kingdom where English Court of Appeal judges received the same remuneration as judges of the English High Court for more than a hundred years up until the 1970s (Winterton, 1995, p. 93). However, from the 1970s in the United Kingdom and from 1985 in New Zealand Court of Appeal judges have received higher salaries than their counterparts in the High Court. The differences in salaries between High Court and Court of Appeal judges in New Zealand are not very large,18 but the non-pecuniary difference in status is still substantial. This compensation structure is consistent with Greenberg and Haley’s argument with respect to the United States Supreme Court. Greenberg and Haley (1986, p. 624) suggest: Since an important attraction of judicial appointment is its power and prestige, it is likely that a position on the Supreme Court offers the largest non-pecuniary benefits. Consequently, the monetary compensation to a Supreme Court justice should represent a smaller proportion of his package of total benefits. However, judicial status does not just derive from which court a judge belongs to within a hierarchical court structure. It also has to be understood more broadly in relation to the prestige it confers to the individual judge in the eyes of the Executive and among his/her peers. There is a large law and economics literature which suggests that power, prestige and reputation provides a powerful non-pecuniary motive for many people (see e.g. McAdams, 1995; Lessig, 1995; Posner, 1998). There is also a large literature which suggests that judges are motivated by the same factors as other people (see e.g. Cooter, 1983; Cohen, 1991, 1992; Macey 1994; Miceli and Cosgel, 1994; Posner, 1994). Schauer (2000, p. 629) suggests judges, “like the rest of us, care about their reputation, care about the esteem in which they are held by certain reference groups and care enough such that at the margin, or even far from the margin, they seek to conform their behaviour to the demands of the relevant esteem-granting (or withholding) or reputation creating (or damaging) groups”. Several authors have argued that the esteem of other judges, along with the Bar, is a major reference

AN EMPIRICAL STUDY OF THE NEW ZEALAND HIGH COURT

219

point, particularly for courts such as the High Court, whose decisions do not attract the same level of law review comment or public interest as the Court of Appeal (Cooter, 1983; Posner, 1994; Schauer, 2000). Thus the impact of the production of independent decisions on the prestige of judges has to be appreciated not only in regard to politicians, but also in relation to other judges. This modifies the effect of prestige on judicial independence and on the chances of promotion to higher office. Another consideration is that judges will behave in a manner, which produces the highest likelihood of reward and, as Schauer (2000) emphasises, however circumspect judges are about admitting it, it is hardly implausible to suspect that many in intermediate courts desire higher judicial office. While “the prospects for promotion of any given judge are low, the well-documented tendency of people to exaggerate the probability of low probability events may suggest that strategizing for promotion may play a larger role in judicial thinking than the actual probability of promotion would otherwise suggest” (Schauer, 2000, p. 632). In this context, it is worth noting that the judges’ choice of independence is an equilibrium phenomenon. If it is known that governments will promote more independent judges, then judges will behave more independently than they otherwise would. Once judges have been promoted to the Court of Appeal, though, this incentive effect does not operate, so behaviour in the High Court might not be a good predictor for behaviour in the Court of Appeal for those that get promoted. 3. 3.1.

The empirical study Data, hypotheses and variables

The comparison in the last section between judicial independence in New Zealand and the United Kingdom suggests that there are strong similarities between the two systems at least in terms of judicial appointments. This justifies applying the Salzberger and Fenn (1999) model, originally developed to account for the British judicial reality, to examine the implications of New Zealand’s institutional arrangements for substantive judicial independence. In particular we study whether successive New Zealand governments have used the judicial appointment and promotion process which, because they are under direct political control, are convenient tools of the structural dependency of the judiciary on the government, to influence the level of substantive independence. Following Salzberger and Fenn (1999) we test both a strong and weak form of the LandesPosner hypothesis. The strong form of the hypothesis is that the government positively uses its powers to secure judicial independence. The weak form of the hypothesis is that the government does not use its powers to induce judicial loyalty, even when those powers are available at no cost. We follow Salzberger & Fenn who define judicial loyalty as the propensity to decide cases according to the government’s will. This is an appropriate definition of judicial independence for our purposes because, consistent with Salzberger and Fenn, our objective is to examine whether a judge’s propensity to decide cases against the government has any effect on his/her prospects of being promoted from the High Court to the Court of Appeal. We examine public law decisions in the High Court in which the government (typically the minister or his/her department) was the defendant. As Salzberger and Fenn

220

MAITRA AND SMYTH

note these cases are often regarded as a “battleground” between citizen and state and when the plaintiff succeeds in this type of case it is seen more often than not as a defeat for the government. We collected data on reported and unreported public law decisions of the High Court in which the government was the defendant contained in the Lexis database over the period 1958 to 2001. We started in 1958 because this was the first year of the permanent Court of Appeal and therefore it was the first year that High Court judges had a clear promotion path. We collected data on 604 cases in total. In most of the cases in the sample, the High Court used single-judge panels. In the remaining cases there was a mixture of two, three and four judge panels.19 In the case of single-judge panels we used actual decisions and in the case of multi-judge panels we used individual judgments or votes.20 From the raw data we constructed profiles for each of the 76 High Court judges who gave decisions in relevant cases over the period. There are two political parties in New Zealand which were in office at different points over the period; namely the National Party and the Labour Party.21 Of the 43 years covered in the study, the National Party was in office for 32 years and the Labour Party was in office for 11 years. Reflecting their longer period in office, the National Party appointed 67 of the 76, or just under 90 per cent of the judges who heard public law cases in which the government was the defendant over the period of the study. The profile for each judge contained information on (a) the period of time he/she had been on the High Court either prior to promotion or retirement or, in the case of judges still on the High Court, at the end of 2001 (TIMEHC); (b) proxies for loyalty to the government and (c) proxies for the quality of their decisions. We use the following proxies for judicial loyalty. 1. AGA is the percentage of cases the judge decided against the government. LABAGA and NATAGA are the percentage of cases the judge decided against Labour and National governments respectively.22 A positive sign associated with AGA, LABAGA and NATAGA implies that an increase in the percentage of cases decided against the government increases the hazard of promotion (decreases the time to promotion). Thus if AGA is positively related to the chances of promotion it implies that loyalty is penalized and disloyalty is rewarded and vice-versa if it has a negative sign. 2. REV is the percentage of decisions in which the judge reversed, either fully or partly, the decision of the lower court. If this variable is combined with the AGA variables multiplicatively, we can get the following interaction terms (AGAREV, AGALREV and AGANREV). These variables take account of the fact that if the judge affirms a lower court decision against the government, this is not as disloyal as if the judge reverses a lower court’s decision which was in the government’s favour. 3. GOVT is a dummy variable set equal to 1 if the same political party is in power when the judge was appointed as when he/she finished his/her career or, in the case of continuing judges, when the data was collected. If this variable is positively related to the probability of promotion, it suggests that political parties are more likely to promote their own appointees irrespective of merit. We use the following variables for the quality of decision-making.

AN EMPIRICAL STUDY OF THE NEW ZEALAND HIGH COURT

221

1. REVCA is the percentage of decisions in which the judge was reversed, partly or fully, in the Court of Appeal. As the Court of Appeal can be regarded, for all intents and purposes, as the final court of appeal in New Zealand, this is an indicator of the extent to which the judge got the law correct. Thus, in terms of Dreyfuss’ (1995) conceptualisation of judicial quality, discussed in the introduction, REVCA is a good proxy for measuring accuracy; the extent to which the judges produce the objectively correct result. In addition to Salzberger and Fenn (1999), Posner (2000) also uses the extent to which a judge is reversed on appeal as a measure of judicial quality. 2. APPOINT is the age at which the judge was appointed to the High Court. Salzberger and Fenn treat this variable as a measure of judicial quality. The rationale is: “The younger the judge at appointment, the higher the likelihood he is considered to be a ‘high flier’ marked out for promotion” (Salzberger and Fenn, 1999, p. 837). The interpretation of the APPOINT variable, though, is not clear-cut. Conceptually, the interpretation is qualified because it is also possible that those who are better connected politically get appointed at younger ages. If this is what is happening it is connections rather than talent which is being rewarded with judicial office. 3. POST is a dummy variable set equal to 1 if the judge has a postgraduate degree in law. Slightly less than 20 per cent of the judges in the sample had postgraduate degrees in law. Thus, it is a quantifiable measure of the heterogeneity between judges with respect to their level of human capital formation. There are several studies suggesting that productivity is positively related to human capital formation (see e.g. Ben-Porath, 1967; McDowell, 1982; Ryder, Stafford and Stephan, 1976). We hypothesize that judges with higher levels of human capital formation should perform better on the Bench. A rider on using postgraduate qualifications to measure human capital formation is that there are of course other measures of human capital formation, such as experience obtained at the Bar through litigating cases. Judges, who do not have higher degrees in law, might still have accumulated substantial human capital through on the job training, but such measures are difficult to quantify in studies such as this. To facilitate comparison between the results for this study and those of Salzberger and Fenn, as far as possible we used the same proxies for loyalty and quality as they employed in their study. Where we use the same variables as Salzberger and Fenn we construct them in the same manner as they do. There are, however, two differences between the variables we use and those used by Salzberger and Fenn. First, we include the variable POST (whether the judge has a postgraduate degree in law) as an additional proxy for the quality of decision-making. Second, Salzberger and Fenn use a variable BOPINION, which is the percentage of concurrences given by the judge. The rationale is that arguably less weight should be given to such decisions when measuring the quality of decision-making. We collected information on this variable and used it in preliminary regressions, but the standard errors were very high, which is likely to be a reflection of the fact that there were very few decisions that were in the form of concurrences (less than 1 per cent). As this variable is not central to the hypotheses being tested we omitted it prior to reporting the final results. Table 1 presents descriptive variables for TIMEHC, APPOINT, REVCA, AGA and REV.

222

MAITRA AND SMYTH Table 1.

3.2.

Descriptive statistics (N = 76 Judges).

Variable

Minimum

Maximum

Mean

SD

TIMEHC

1

26

10.17

5.68

APPOINT

8.14

28

76

51.89

REVCA

0

25

3.07

6.71

AGA

0

100

49.67

30.93

REV

0

100

39.36

35.58

Econometric methodology

We use survival analysis to model the conditional probability of promotion to the Court of Appeal as a function of the decision-making record of the judge, the political persuasion of the government and proxies for the quality of the judge’s decisions. These events in any given year are the outcome of a sequence of conditional probabilities (or hazards) of promotion or retirement in previous years. The time to promotion is modeled as a failure time process represented by a log hazard of the promotion equation. The hazard or risk is the conditional probability of being promoted at age t, provided that the judge has not been promoted until that date. The hazard rate of an event T can be defined as: h (t|u, η) ≡ Probability that the event T occurs in the time interval (t, t + dt) given that it has not occurred until t where U denotes the set of strategies the judge might adopt to facilitate promotion and let u ∈ U denote the actual strategy adopted and η denote the set of unobservable factors that affect the hazard of promotion. The proportional hazard model for the observed time to promotion is given by h (t|X, η) =  λ0 (t) e X β , where λ0 (t) denotes the baseline hazard function (λ0 (t) = h (t|u = 0, η = 0)) and X denote the set of co-variates whose values represent the information available to the judge at time t. The parameters (β) can be estimated using maximum likelihood methods once we specify an appropriate functional form for the baseline hazard function. One could choose between a number of alternative distributions to characterize the baseline hazard— Exponential, Weibull, Log-Normal, Log-logistic and Gompertz.23 We estimated the model using these alternative distributional assumptions, and ultimately the Weibull distribution proved to be the best because it gave the lowest AIC.24 We also estimate the log hazard of duration to promotion semi-parametrically using Cox’s proportional regression model (in this case we do not need to make any distributional assumption regarding the baseline hazard model). As indicated above, using the Weibull distribution has the added advantage that it facilitates direct comparison with Salzberger and Fenn’s findings, although we do not agree with their reason for use of this distribution. Salzberger and Fenn argue that since “the conditional probability of promotion increased over time for (their) sample of judges” that implied that the Weibull regression was the best parameterization of the baseline hazard model. We use a Weibull distribution because it is the best fit. It is worth noting, however, that our results also imply that the conditional probability of promotion increases over time. For the proportional Weibull regression model, λ0 (t) = pt p−1 , where p is the shape parameter estimated using data. Note that an estimated value of p exceeding one implies

AN EMPIRICAL STUDY OF THE NEW ZEALAND HIGH COURT

223

that the hazard of promotion increases over time while an estimated value of p less than one implies that the hazard of promotion decreases over time. In the Cox proportional hazard model formally the baseline hazard function is not estimated. One advantage of using a hazard model to estimate the time to promotion is that it allows us to account for censoring in the sample. The censoring arises from the fact that there are judges who have “not exited”—they have either retired before being promoted or are continuing (and have not yet retired or been promoted). Of the 76 judges on which we have data, at the end of 2001, which was the end date of the time frame for the study, 15 had been promoted, 40 had retired and 21 were still sitting on the High Court. Rather than ignoring altogether the continuing judges, for each hazard function we report two sets of results. For the promotion hazard these are “promotion versus retired”, where N (completed) are judges who are promoted and N (censored) are judges who have retired and “promotion versus retired and continuing” where N (completed) are judges who are promoted and N (censored) are judges who have either retired or are still continuing. For the retirement hazard these are “retired versus promoted”, where N (completed) are judges who are retired and N (censored) are judges who have been promoted and “retired versus promoted and continuing” where N (completed) are judges who have retired and N (censored) are judges who have either been promoted or are still continuing. We estimate two separate models. Model 1 treats the effects of promotion by “government” irrespective of political persuasion and retirement as the combined determinants of the length of service. Model 2 treats the effects of promotion by a Labour government, National government and retirement as the joint determinants of length of service. Model 1 consists of the effect of two independent hazard functions for promotion and retirement. The log hazard of promotion is: log h PROM (t | X, η) = log λ0 (t) + β0 + β1 APPOINT + β2 POST + β3 (REVCA) + β4 AGA + β5 REV + β6 AGAREV + β7 GOVT + η (1) The log hazard of retirement is: log h RETIRE (t | X, η) = log λ0 (t) + β0 + β1 APPOINT + η

(2)

Remember that in the Cox proportional regression model we cannot estimate λ0 (t) and β0 . In the Weibull regression model λ0 (t) = pt p−1 and we are able to estimate p and β0 . Model 2 considers the promotion decisions of Labour and National governments separately with separate decision-making variables reflecting the record of each judge when each of the parties was in office (NATAGA, LABAGA, AGANREV, AGALREV). log h NATPROM (t | X, η) = log λ0 (t) + β0 + β1 APPOINT + β 2 POST + β3 REVCA + β4 NATAGA + β5 REV + β6 AGANREV + β7 LABAGA + η (3) log h LABPROM (t | X, η) = log λ0 (t) + β0 + β1 APPOINT + β2 POST + β3 REVCA + β4 LABAGA + β5 REV + β6 AGALREV + β7 NATAGA + η

(4)

224

MAITRA AND SMYTH

and log h RETIRE (t | X, η) = log λ0 (t) + β0 + β1 APPOINT + η

3.3.

(5)

Results

Table 2 reports the results for the promotion hazard in model 1. We report the coefficient, hazard ratio and t-statistic for each variable. A negative sign on the coefficient decreases the hazard of promotion (increases the time to promotion) and a positive sign on the coefficient increases the hazard of promotion (decreases the time to promotion). The first set of results are for the “promotion versus retired” hazard and the second set of results are for the “promotion versus retired and continuing” hazard. In the following discussion we focus on the results for the Weibull regression to facilitate comparison with Salzberger and Fenn’s findings. At the end of this section we make some general comments about the Cox regressions. The Weibull parameter p is greater than 1 in both cases, which is consistent with the conditional probability of promotion increasing with the length of service. This is confirmed by figures 1 and 2 where we plot the fitted hazard function in the two cases. The results for the promotion versus retirement hazard provide good support for the hypothesis that the quality of decision-making is an important factor influencing promotion. The coefficient on APPOINT has a negative sign and it is statistically significant in the promotion versus retirement hazard.

.008

Hazard function

.006

.004

.002

0 0

10

20 analysis time

Weibull regression Figure 1.

Estimated hazard from promotion versus retired regression.

30

−1.97 1.41 −1.02

−0.0960

3.1388

2.4108

−1.2762

APPOINT

REVCA

AGA

REV −1.76

−0.9729

0.9280

1.4845

POST

P 40 −49.6573

40 −35.3934

Log likelihood

15

1.66

−1.54

0.00

−0.79

1.04

0.61

−1.90

15

0.9436

−0.8427

−34.7619

−0.9457

1.6936

1.9238

−0.0878

N (censored)

1.4845

2.5294

0.3780

2.55E-07

0.2791

11.1430

23.0761

0.9084



N (completed)

4.31

1.64

−0.01

−15.1831

GOVT

1.00

−0.36

−0.8440

AGAREV

Weibull regression

Cox proportional hazard

Promotion versus retired and continuing

2.5693

0.4305

3.63E-20

0.3884

5.4390

6.8471

0.9160

1.5451

0.8190

−0.6806

−15.5339

−0.3892

1.8053

2.8842

−0.8066

−2.1082

−38.7723

61

15

4.40

1.45

−1.24

−0.01

−0.34

1.17

0.86

−1.60

−0.86

1.5451

2.2683

0.5063

1.79E-07

0.6776

6.0815

17.8900

0.9225



0.8053

−0.5687

−38.8348

−0.2076

1.3263

1.5875

−0.0754

−53.1533

61

15

1.41

−1.04

0.00

−0.18

0.88

0.47

−1.57

2.2373

0.5663

3.70E-17

0.8125

3.7670

4.8913

0.9274

Coefficient t-statistic Hazard ratio Coefficient t-statistic Hazard ratio Coefficient t-statistic Hazard ratio Coefficient t-statistic Hazard ratio

Cox proportional hazard

Promotion versus retired

Weibull regression

Pormotion hazard.

Constant

Table 2.

AN EMPIRICAL STUDY OF THE NEW ZEALAND HIGH COURT

225

226

MAITRA AND SMYTH

.008

Hazard function

.006

.004

.002

0 0

10

20

30

analysis time

Weibull regression Figure 2.

Estimated hazard from promotion versus retired and continuing regression.

The other proxy for the quality of decision-making that is significant in the promotion versus retirement hazard is POST, which has a statistically significant positive coefficient. The associated hazard ratio (2.5294) in the Weibull regression implies that having a postgraduate degree in law more than doubles the hazard of promotion. If we interpret the coefficients in terms of elasticities, we find that a judge with a postgraduate degree in law is 92.8 per cent more likely to be promoted compared to a judge without a postgraduate degree in law. Interestingly in the Weibull regression the percentage of cases in which the judge was reversed in the Court of Appeal (REVCA) increases the hazard of promotion (twenty three times), though the effect is not statistically significant. Turning to the proxies for judicial loyalty, while none of the variables are significant in the promotion versus retired and continuing hazard, in the Weibull promotion versus retirement hazard GOVT is statistically significant with a negative coefficient, while AGA, REV and AGAREV are statistically insignificant.25 The results for GOVT in the promotion versus retirement hazard suggest that National (Labour) governments are statistically more likely to promote judges to the Court of Appeal who are appointed to the High Court by Labour (National) governments. This finding provides support for the strong form of the LandesPosner hypothesis that governments will positively use their powers to secure judicial independence. The statistically insignificant coefficients on AGA, REV and AGAREV are consistent with the weak form of the Landes-Posner hypothesis that governments do not use their powers to induce judicial loyalty. Judges who frequently decide against the government or reverse the decisions of lower courts that were in favour of the government are just as likely to be promoted early as other judges. Note that if we focus on the point estimate rather than statistical significance the positive sign on AGA implies that time to promotion, on average,

AN EMPIRICAL STUDY OF THE NEW ZEALAND HIGH COURT

227

decreases with disloyalty. Thus the sign on the AGA coefficient, although statistically insignificant, is consistent with the conclusion that governments reward disloyalty. Table 3 reports the results of the retirement hazard. In the Weibull regression the Weibull parameter p is greater than 1 for both retired versus promoted and retired versus promoted and continuing hazard functions. APPOINT has a positive sign and is statistically significant for both retirement hazard functions. The coefficient on APPOINT suggests that every additional year of age at appointment to the High Court leads to a statistically significant 7 per cent or 8 per cent reduction (depending on the hazard function) in the duration to retirement. Tables 4 and 5 present the promotion hazards for Labour and National governments in model 2. This separates out judicial promotions made during periods of National and Labour rule and therefore provides a more direct test of whether National and Labour governments reward or punish judges who decide cases in favour of one political party or the other. In Table 4, none of the parameters of the Labour hazard function are statistically significant. This could be due to the fact that there was only a small number of promotions to the Court of Appeal while Labour was in office. Turning to Table 5, the results for the proxies for the quality of decision-making in the National promotion hazard reinforce the results in Table 2. APPOINT is statistically significant with a negative coefficient, POST is statistically significant with a positive coefficient and REVCA is statistically insignificant in both the promotion versus retired and promotion versus retired and continuing hazard functions. In terms of the judicial loyalty variables, in the Weibull regressions LABAGA is statistically significant with a negative coefficient for both the promotion versus retired and promotion versus retired and continuing hazard functions in Table 5. This finding is again consistent with the strong form of the Landes-Posner hypothesis that governments use their powers to control promotion to positively secure judicial independence. The results for LABAGA suggest that each additional percentage point added to the percentage of cases which a judge decides against the Labour government significantly increases the duration to when they can expect to be promoted by a National government by over 200 per cent. This result differs from that of Salzberger and Fenn, who found that a propensity to reverse lower court decisions that were in favour of Conservative governments did reduce their chances of being promoted by a Conservative government. In Table 5 REV has a positive coefficient and is statistically significant, but NATAGA and AGANREV are insignificant in both hazards. This suggests that judges who decide cases against the National government or reverse lower court decisions that were in favour of the National government are treated the same as other judges by National governments in terms of promotion. This is consistent with the weak form of the Landes-Posner hypothesis that governments do not use their powers to reward judicial loyalty. This is the same result as in Salzberger and Fenn. It is worth pointing out again though that if we examine the point estimates on NATAGA, the positive sign on NATAGA implies that time to promotion, on average, decreases with disloyalty. The sign on the NATAGA coefficient, although statistically insignificant, is consistent with the conclusion that National governments reward disloyalty to National governments. There is a proviso to the conclusion that these results support the Landes-Posner hypothesis. As discussed above the National party was in power for most of the period under

15 −111.4950

−30.5679

Log likelihood

40

3.53

15

2.8104

0.0767

N (censored)

8.05

2.8104

– 1.0755

40

3.47

−6.83

0.0727

−11.4166

N (completed)

P

Weibull regression

Cox proportional hazard

Retired versus promoted and continuing

1.0797 2.9559

0.0780

−12.1679

−33.374

36

40

8.50

3.71

−7.32

2.9559

1.0812

– 0.0822

−114.3549

36

40

3.75

1.0857

Coefficient t-statistic Hazard ratio Coefficient t-statistic Hazard ratio Coefficient t-statistic Hazard ratio Coefficient t-statistic Hazard ratio

Cox proportional hazard

Retired versus promoted

Weibull regression

Retirement hazard.

APPOINT

Constant

Table 3.

228 MAITRA AND SMYTH

−0.35

−0.0276

3.0860

1.9945

REVCA

NATAGA

LABAGA

1.09 −0.88 −0.48 −0.25 2.83

2.6826

−2.5155

−2.1419

−0.2317

1.9355

REV

AGALREV

POST

0.82

0.56

−1.50

−6.4270

Constant



49 −19.9250

49 −17.8725

Log likelihood

6

−0.18

−0.43

−0.90

0.96

0.64

0.47

−0.48

N (censored)

−0.1704

−1.8256

−2.4206

2.3795

1.4801

2.5220

−0.0349

6

1.9355

0.7932

0.1174

0.0808

14.6228

7.3485

21.8885

0.9727

N (completed)

P

Weibull regression

Cox proportional hazard

Promotion versus retired and continuing

0.8433

0.1611

0.0888

10.7992

4.3934

12.4534

0.9657

2.0845

−0.2886

−1.7844

−1.8586

2.5107

1.1493

3.3321

−0.0033

−8.0363

−18.6225

70

6

3.01

−0.31

−0.43

−0.54

1.23

0.54

0.61

−0.04

−1.91

1.9355

0.7493

0.1679

0.1559

12.3141

3.1559

27.9977

0.9968



−0.3202

−2.3526

−1.9288

2.8574

1.0314

1.9699

−0.0258

−20.5858

70

6

−0.34

−0.56

−0.71

1.21

0.48

0.37

−0.35

0.7260

0.0951

0.1453

17.4165

2.8052

7.1701

0.9745

Coefficient t-statistic Hazard ratio Coefficient t-statistic Hazard ratio Coefficient t-statistic Hazard ratio Coefficient t-statistic Hazard ratio

Cox proportional hazard

Promotion versus retired

Weibull regression

Promotion hazard (LAB).

APPOINT

Table 4.

AN EMPIRICAL STUDY OF THE NEW ZEALAND HIGH COURT

229

−1.77 0.52 −1.86

−0.1202

3.4222

1.5373

−2.9177

REVCA

NATAGA

LABAGA

46 −26.8389

−24.1832

Log likelihood

9

2.18

−1.20

1.89

−1.69

0.35

0.48

46

1.3498

2.0295

−5.4635

5.0088

−2.7512

0.9564

2.4415

−1.68

N (censored)

3.39

1.3498

7.7656

0.0016

317.6320

0.0541

4.6520

30.6379

−0.1091

9

2.21

−1.35

2.0497

−6.4269

– 0.8867

N (completed)

P

POST

AGANREV

2.10

5.7609

0.70

−0.49

−1.4355

Constant

REV

Weibull regression

Cox proportional hazard

Promotion versus retired and continuing

7.6103

0.0042

149.7194

0.6385

2.6024

11.4906

0.8966

1.4133

1.9288

−6.4448

6.2781

−2.3320

0.9983

1.7998

−0.1214

−1.6999

−26.1166

67

9

3.46

2.09

−1.53

2.43

−1.64

0.40

0.37

−1.76

−0.55

1.4133

6.8815

0.0016

532.7802

0.9710

2.7137

6.0482

0.8856



1.8922

−5.7102

5.5932

−2.1062

0.6051

0.6942

−0.1121

−28.9082

67

9

2.06

−1.38

2.22

−1.44

0.26

0.14

−1.69

6.6339

0.0033

268.594

0.1217

1.8315

2.0021

0.8939

Coefficient t-statistic Hazard ratio Coefficient t-statistic Hazard ratio Coefficient t-statistic Hazard ratio Coefficient t-statistic Hazard ratio

Cox proportional hazard

Promotion versus retired

Weibull regression

Promotion hazard (NAT).

APPOINT

Table 5.

230 MAITRA AND SMYTH

AN EMPIRICAL STUDY OF THE NEW ZEALAND HIGH COURT

231

consideration. If this political pattern was predicted ex ante (or at least if there were no systematic errors in anticipation) then if the Landes-Posner hypothesis is an accurate explanation for what is happening, judicial independence should be more important to the Labour party. The Nationals are more likely to remain in power at any given point in time, so have less need to assure interest groups of the durability of their promises through judicial independence. The results do not support this position, but as discussed earlier, it is likely that there are just too few Labour appointments to tell. The results for the Cox proportional hazard regressions in Tables 2–5, where we do not make any distributional assumptions regarding the baseline hazard are generally consistent with the findings from the Weibull regressions. To summarize the basic results for the Cox regressions: 1. The coefficient estimate of APPOINT is always negative though not always statistically significant in the promotion hazard regressions and is positive and statistically significant in the retirement hazard regression (see Tables 2 and 3). 2. The coefficient estimate of POST is positive and statistically significant in the promotion versus retired hazard regression and positive but not statistically significant in the promotion versus retired and continuing regression (Table 2). Once again this implies that judges with a postgraduate degree in law are more likely to be promoted. 3. None of the loyalty variables are statistically significant in explaining the hazard of promotion in the Cox regression. 4.

Conclusion

This paper makes two contributions to the literature. The first is that our results can be interpreted as providing some support for both the weak and strong forms of the LandesPosner hypotheses for an important common law jurisdiction outside the United States. This adds to the paucity of empirical studies using non-United States data. As far as the weak form of the Landes-Posner hypothesis, we find no evidence that successive New Zealand governments have used their powers to punish judges who decided cases against them. In contrast to the English Court of Appeal, this extends to judges who reverse lower court decisions that were initially in the government’s favour. The results for GOVT in the promotion versus retired hazard in Table 2 and LABAGA in Table 5 provide support for the strong form of the Landes-Posner hypothesis that governments positively use their powers to secure judicial independence. The second contribution of the paper is methodological in nature. While we use the same general approach as Salzberger and Fenn, we believe that our reason for choosing a Weibull regression model is more convincing. We initially estimated the model using a number of alternative distributional assumptions, but ultimately the Weibull distribution proved to be the best because it gave the lowest AIC and therefore the best fit. We chose the Weibull regression on this basis and not because the conditional probability of promotion was increasing over time. We also check the robustness of the results using a Cox proportional hazard model. While the results are weaker in terms of statistical significance, the findings generally hold with the more general Cox form.

232

MAITRA AND SMYTH

We conclude with a few observations on the reasons why our results differ from Salzberger and Fenn (1999). The first point to emphasise is that New Zealand and the United Kingdom share several similarities in terms of structural independence. Among these (a) neither country has a written constitution; (b) judicial appointment is subject to direct political control in both countries; (c) both countries have essentially career-based judiciaries, which increases the importance of (b); and (d) in both countries one political party has been in power for long periods. In a couple of areas, however, New Zealand courts appear to have more structural independence. One is that in the United Kingdom judicial salaries are determined by the Lord Chancellor (although the safeguard does exist that judicial salaries cannot be reduced). In New Zealand judicial salaries are determined by an independent statutory body, the Higher Salaries Commission. The other is that since 1990 New Zealand has had a Bill of Rights which, while it is not enshrined, has helped to facilitate an increasing judicial activism, where judges are more willing to judicially legislate. This phenomenon is most often associated with the United States where the courts have a much higher level of structural independence. In terms of judicial activism, though, there are some similarities between New Zealand and Israel which has become “Americanised” in the absence of a written constitution (Salzberger, 1993, pp. 357–358). Overall the differences in results between this paper and those in Salzberger and Fenn (1999) suggest differences may exist among common law countries, which pleads in favour of a differentiated analysis between national heterogeneous situations. The results also suggest that subtle differences in structural judicial independence might generate differences in substantive independence and that more research is needed to explore the effects of the two concepts on judicial decision-making. Acknowledgments We thank Dhammika Dharmapala and two anonymous referees of the journal for helpful comments. We also thank Yael Furstenberg and David Goldberg who provided assistance with data collection. This project was funded by a Monash Small Grant.

Notes 1. In this paper, when discussing judicial review in the United Kingdom we ignore the effect of the European Convention for Human Rights. While this is giving British judges increasing powers to review and challenge acts of Parliament, this is a recent development and was much less important over the timeframe of the Salzberger and Fenn’ (1999) study. 2. The description of Banks’ attacks on the District Court draw heavily on Stockley, (1997, pp. 167–168). 3. “Banks regrets talkback jibe,” The Press, Christchurch, 25 May 1995. 4. “Talkback lands Banks in Strife,” The Press, Christchurch, 19 June 1995. 5. “Cabinet colleague roasts Banks,” The Press, Christchurch, 20 June 1995. 6. Otago Daily Times, 25 May 2001. 7. Ibid. 8. New Zealand Herald, 7 March 2003; New Zealand Herald 10 March 2003. 9. New Zealand Herald, 10 March 2003. 10. The Daily News (New Plymouth), 19 December, 2002. 11. The Dominion Post (Wellington), April 11, 2003.

AN EMPIRICAL STUDY OF THE NEW ZEALAND HIGH COURT

233

12. See Dixit, Grossman and Gul (2000), although these authors do not look explicitly at judicial appointments. 13. The retirement age for High Court judges was 72, until amended by s.3 of the Judicature Amendment Act (1980). 14. Justice Thomas has also given judicial effect to his views in Fulcher v Parole Board (1997) 15 CRNZ 222, 242–243 and R. v Hines (1997) 3 NZLR 529, 579–582. 15. In New Zealand the Attorney-General is not a member of the judiciary as the Lord Chancellor is in the United Kingdom. Their political roles, as members of the Executive, however are the same. 16. Collector of Customs v Lawrence Publishing Co. Ltd. (1986) 1 NZLR 404, 414–415. 17. This excludes the three initial Court of Appeal judges appointed in 1958 (none of which are considered in this study given its timeframe), but includes the Chief Justice of New Zealand, who is regarded as an ex officio member of the Court of Appeal. If we exclude individuals appointed Chief Justice 17 out of 20, or 85 per cent of appointees to the Court of Appeal were promoted from the High Court. 18. Winterton (1995, p. 94) gives figures for judicial salaries in 1994. The salary of a High Court judge was $166,500 and the salary of a Court of Appeal judge was $175,000 (salaries in New Zealand dollars). 19. Of the 604 cases, there were 589 cases (or 97.52 per cent of the total) on which only one judge sat. Of the remaining 15 cases, there were eight two-judge panels, five three-judge panels and two four judge panels. 20. In the multi-judge panels there were no dissenting judgments. Thus, the decision of the court and the decision of the individual judge are the same in all instances. 21. The National Party is a conservative party with policies similar to the Republicans in the United States or Tories in the United Kingdom. The policies of the Labour Party are broadly similar to the Democratic Party in the United States or Labour Party in the United Kingdom. 22. In most cases the issue of whether the court decided against the government was clear cut as there was only one issue at stake. In a small number of cases (less than 2 per cent of the total) there were multiple issues of law where the court found for the government on one issue and against the government on another issue. Invariably in these instances there was one major issue before the court and one or more side issues. Thus, in these cases we recorded the outcome for the major issue on which the court was deciding. 23. The most general parameterization of the baseline hazard function is the Gamma distribution. We could not use this distribution because of convergence problems. 24. The AIC is defined as: AIC = −2(Log Likelihood) +c(c+ p + 1), where c is the number of model covariates and p is the number of model specific ancillary parameters. The AIC is computed as 91.275, 90.787, 89.250, 90.585 and 92.485 for the Exponential, Log normal, Weibull, Log logistic and Gompertz baseline hazard parametrisations. 25. The standard errors for AGAREV are very large. This could have some effect on the precision of estimates, but because AGAREV is a central variable in testing the judicial loyalty hypothesis we retained it in the reported results.

References Allan, J. (1998). “Why New Zealand Doesn’t Need a Written Constitution.” Agenda. 5, 487–494. Anderson, G., Shughart, W. & Tollison, R. (1989). “On the Incentives of Judges to Enforce Legislative Wealth Transfers.” Journal of Law and Economics. 32, 215–228. Atkins, B. (1991). “Party Capability Theory as an Explanation for Intervention Behavior in the English Court of Appeal.” American Journal of Political Science. 35, 881–903. Auckland District Law Society. (2003). Submission to the Courts Committee of the Auckland District Law Society to the Justice and Electoral Committee on the Supreme Court Bill, April. Ben-Porath, Y. (1967). “The Productivity of Human Capital and the Life Cycle of Earnings.” Journal of Political Economy. 75, 352–365. Cohen, M. (1991). “Examining Judicial Behaviour or What’s Unconstitutional About the Sentencing Commission.” Journal of Law, Economics and Organization. 7, 183–199. Cohen, M. (1992). “The Motives of Judges: Empirical Evidence From Antitrust Sentencing.” International Review of Law and Economics. 12, 13–27.

234

MAITRA AND SMYTH

Cooter, R. (1983). “The Objectives of Private and Public Judges.” Public Choice. 41, 107–132. Crain, W. M. & Tollison, R. (1979). “Constitutional Change in an Interest Group Perspective.” Journal of Legal Studies. 8, 165–175. Crain, W. M. & Tollison, R. (1979a). “The Executive Branch in the Interest-Group Theory of Government.” Journal of Legal Studies. 8, 555–567. Crain, W. M., Shughart, W. &. Tollison, R. (1988). “Legislative Majorities as Non-Salvageable Assets.” Southern Economic Journal. 55, 303–314. Dixit, A., Grossman, G. & Gul, F. (2000). “The Dynamics of Political Compromise.” Journal of Political Economy. 108, 531–568. Dreyfuss, R. C. (1995). “Forums of the Future: The Role of Specialized Courts in Resolving Business Disputes.” Brooklyn Law Review. 61, 1–44. Eichelbaum, T. (Sir) (1993). “Judicial Independence — Fact or Fiction?”New Zealand Law Journal. March, 90–92. Farmer, J. (2001). “The New Zealand Court of Appeal: Maintaining Quality After the Privy Council.” In R. Bigwood (ed.) Legal Method in New Zealand, Auckland: Butterworths, pp. 237–250. Ferejohn, J. & Shipan, C. (1990). “Congressional Influence on Bureaucracy.” Journal of Law, Economics and Organization. 6, 1–20. Ferejohn, J. & Weingast, B. (1992). “A Positive Theory of Statutory Interpretation.” International Review of Law and Economics. 12, 263–279. Goldfinch, S. (1993). “Judicial Independence and the Administration of the Courts in New Zealand.” Political Science. 45, 153–169. Greenberg, P. E. & Haley, J. A. (1986). “The Role of the Compensation Structure in Enhancing Judicial Quality.” Journal of Legal Studies. 15, 417–426. Hammond, R. G. (1991). “The Judiciary and the Executive.” Journal of Judicial Administration. 1, 88–113. Hanssen, F. A. (1999). “The Effect of Judicial Institutions on Uncertainty and the Rate of Litigation: The Election versus the Appointment of State Judges.” Journal of Legal Studies. 28, 205–232. Hanssen, F. A. (2000). “Independent Courts and Administrative Agencies: An Empirical Analysis of the States.” Journal of Law, Economics and Organisation. 16, 534–571. Hanssen, F. A. (2001). “Is There a Politically Optimal Level of Independence?” The Hoover Institution, Stanford University, Working Paper, April. Jaffe, L. L. (1969). English and American Judges as Lawmakers. Joseph, P. (1993). Constitutional and Administrative Law in New Zealand. Sydney: Law Book Company. Landes, W. & Posner, R. (1975). “The Independent Judiciary in an Interest Group Perspective.” Journal of Law and Economics. 18, 875–902. Lessig, L. (1995). “The Regulation of Social Meaning.” University of Chicago Law Review. 62, 943– 1045. Macey, J. R. (1994). “Judicial Preference, Public Choice and the Rules of Procedure.” Journal of Legal Studies. 23, 627–646. McAdams, R.H. (1995). “Cooperation and Conflict: The Economics of Group Status Production and Race Discrimination.” Harvard Law Review. 108, 1003–1084. McCubbins, M., Noll, R. & Weingast, B. (1987). “Administrative Procedures as Instruments of Political Control.” Journal of Law, Economics and Organization. 3, 243–277. McCubbins, M. & Schwartz, T. (1984). “Congressional Oversight Overlooked: Police Patrols versus Fire Alarms.” American Journal of Political Science. 28, 165–179. McDowell, J. (1982). “Obsolescence of Knowledge and Career Publication Profiles: Some Evidence of Differences among Fields in Costs of Interrupted Careers.” American Economic Review. 72, 752–768. Miceli, T. & Cosgel, M. (1994). “Reputation and Judicial Decision-making.” Journal of Economic Behavior and Organization. 23, 31–51. Palmer, G. (Sir) (1995). “Judicial Selection and Accountability: Can the New Zealand System Survive?” In Gray & McClintock (eds.), Courts and Policy: Checking the Balance, Wellington: Brookers. Palmer, G. (Sir) (2000). “The Legal History and Framework of the New Zealand Constitution.” Unpublished manuscript, Chen & Palmer Solicitors, Auckland. Posner, E. (1998). “Symbols, Signals and Social Norms in Politics and the Law.” Journal of Legal Studies. 27, 765–798.

AN EMPIRICAL STUDY OF THE NEW ZEALAND HIGH COURT

235

Posner, R. (1994). “What Do Judges and Justices Maximize? (The Same Thing Everybody Else Does).” Supreme Court Economic Review. 3, 1–41. Posner, R. (2000). “Is the Ninth Circuit Too Large? A Statistical Study of Judicial Quality.” Journal of Legal Studies. 29, 711–718. Ramseyer, M. J. (1994). “The Puzzling (In)dependence of Courts: A Comparative Approach.” Journal of Legal Studies. 23, 721–747. Ramseyer, M. J. & Rasmusen, E. B. (1997). “Judicial Evidence in a Civil Law Regime: The Evidence From Japan.” Journal of Law, Economics and Organization. 13, 259–286. Ramseyer, M. J. & Rasmusen, E. B. (1999). “Why the Japanese Taxpayer Always Loses.” Southern California Law Review. 72, 571–595. Ramseyer, M. J. & Rasmusen, E. B. (2001). “Why is the Japanese Conviction Rate So High?” Journal of Legal Studies. 30, 53–88. Ramseyer, M. J. & Rasmusen, E. B. (2001a). “Why are Japanese Courts so Conservative in Politically Charged Cases?” American Political Science Review. 95, 331–344. Richardson, I. (Sir) (1981). “The Role of an Appellate Judge.” Otago Law Review. 5, 1–10. Rogers, J. R. (2001). “Information and Judicial Review: A Signaling Game of Legislative-Judicial Interaction.” American Journal of Political Science. 45, 84–99. Ryder, H., Stafford, F. & Stephan, P. (1976). “Labor, Leisure and Training Over the Life Cycle.” International Economic Review. 17, 651–674. Salzberger, E. (1993). “A Positive Analysis of the Doctrine of the Separation of Powers, or Why Do We Have an Independent Judiciary?” International Review of Law and Economics. 3, 349–379. Salzberger, E. M. & Fenn, P. (1999). “Judicial Independence: Some Empirical Evidence From the English Court of Appeal.” Journal of Law and Economics. 42, 831–847. Schauer, F. (2000). “Incentives, Reputation and the Inglorious Determinants of Judicial Behavior.” University of Cincinnati Law Review. 68, 615–636. Stephenson, M. C. (2003). “When the Devil Turns: The Political Foundations of Independent Judicial Review.” Journal of Legal Studies. 32, 59–88. Stockley, A. P. (1997). “Judicial Independence: The New Zealand Experience.” Australian Journal of Legal History. 3, 145–169. Sutton, R. & Bigwood, R. (2001). “Taking Stock: Legal Method in New Zealand Today (and for the Future?).” In R. Bigwood (ed.), Legal Method in New Zealand, Auckland: Butterworths, pp. 305–338. Thomas, E. W. (1999). “Fairness and Certainty in Adjudication: Formalism versus Substantialism.” Otago Law Review. 9, 459–488. Thomas, E. W. (2000). “The Relationship of Parliament and the Courts: A Tentative Thought or Two for the New Millennium.” Victoria University of Wellington Law Review. 31, 5–36. Thomas, E. W. (2001). “A Critical Examination of the Doctrine of Precedent.” In R. Bigwood (ed.), Legal Method in New Zealand, Auckland: Butterworths, pp. 141–174. Toma, E. (1991). “Congressional Influence and the Supreme Court: The Budget as a Signaling Device.” Journal of Legal Studies. 20, 131–146. Vanberg, G. (2001). “Legislative-Judicial Relations: A Game-Theoretic Approach to Constitutional Review.” American Journal of Political Science. 45, 346–361. Watt, P. (2001). “The Judge as Casual Law-maker.” In R. Bigwood (ed.), Legal Method in New Zealand, Auckland: Butterworths, pp. 175–213. Winterton, G. (1995). Judicial Remuneration in Australia. Melbourne: Australian Journal of Judicial Administration.