Address correspondence to: J. Mark Ramseyer (corresponding author) Harvard Law School Cambridge, Massachusetts 02138 FAX: 617-496-6118 Eric B. Rasmusen Indiana University Kelley School of Business Bloomington, Indiana 47405 FAX: 812-855-3354 December 5, 1999 Skewed Incentives: Paying for Politics as a Japanese Judge By J. Mark Ramseyer & Eric B. Rasmusen* Nineteen sixty-eight. Best of times, worst of times -- probably it all depended on what one wanted to do, much as it had for Therese Defarge and Sydney Carton two centuries before. At least for those with appropriately libertine preferences, they could be amusing times. Hair and I Am Curious (Yellow) did make standing in line for on-stage nudity and pornography respectable. LSD and anonymous sex did still promise risk-free entertainment. But they were grim times too. Already in 1967 the the Red Guards had sacked the British compound in Peking. Come 1968, the Vietcong launched the Tet offensive, American troups massacred the town of My Lai, and the Soviets trashed Czechoslovakia. James Earl Ray shot Martin Luther King, and 55,000 troops quieted the riots that followed. Students sacked Columbia and shut down Paris. They took over the Chicago streets. And when young instructors at San Francisco State told their students to bring guns to class, the university president just quit and went home. Life in Japan moved in parallel to all this. Over the course of 1968, New Left radicals staged protest after protest. They took over the prestigious University of Tokyo. They fought off the police with bricks and molotov cocktails. They assembled an 800,000-strong mob to march through town on October 21. And in true Jacobin style they saved the worst for their best friends: in just one year, their internecine brawls left 1,100 injured and two dead. It was in this frenzied world of late-1960s Japan that the issue of judicial independence came to the fore. As cases loosely tied to the polarized political disputes left the courts, radicals began to tell tales of leftist judges sentenced to derailed careers in the judicial outback. Again and again, they told of judges who flouted the ruling Liberal Democratic Party (LDP) to professional disgrace. According to Article 76 of the Constitution, all judges were to be "independent in the exercise of their conscience and ... bound only by this Constitution and the laws." Notwithstanding this nominal independence, argued the radicals, judges who ignored LDP preferences paid with their careers. Since then, tales of left-leaning judges punished for their politics have continued to reappear -- and continued to fizzle. They fizzled for a simple reason that goes to the heart of social scientific research: the need to hold constant a variety of factors before one can measure the impact of any one variable. Restated, they fizzled because the radicals never showed that some non-political factor -- temperament, for instance, or talent -- was not the true cause of the leftist career failure. Every time they paraded a favored judge with a spoiled career, the center could simply blame bad judgment, low IQ, or random bad luck. In a series of recent studies we have undertaken the systematic, multi-variate study the radical left never did.1 We first identify a large population of Japanese judges (several hundred), and locate proxies for a judge's ability and work habits. We then hold constant those proxies using regression techniques and ask whether judges who in a variety of ways indulged heterodox politics suffered in their careers. The results are striking. The left was wrong about many things, but they were right about judges. Notwithstanding the nominal independence of the Japanese courts, judges who flouted the LDP systematically took a hit in their careers. Judicial careers. -- To understand how judicial careers might suffer in Japan, consider the institutional structure of the courts. In Japan, a would-be judge began his legal education while still an undergraduate. During his last year, he took the entrance exam to the one national law school, the Legal Research & Training Institute (LRTI). If lucky, he passed it on his first try -- but few did. Given a pass rate that usually ranged between 1 and 4 percent, he more likely flunked. He then retook the exam every year until he passed or eventually despaired of passing. Most LRTI graduates became lawyers. A few became prosecutors, however, and annually some 70 to 130 became judges. Formally, they became judges upon appointment by the Cabinet. Generally, the Cabinet deferred to the Supreme Court Secretariat, the administrative office of the courts, and the Secretariat took most of those who applied. As judges, these men then served a series of 10-year terms subject to routine reappointment. Most retired in their late 50's or early 60's, and the rest retired at the mandatory age of 65. When a new judge joined the bench, the Secretariat named him to a specific court. Traditionally, it identified the stars in each class and started them at the Tokyo District Court. They then stayed on a fast track for most of their career. It also identified the dogs (albeit dogs who scored in the top 4 percent of the students taking the LRTI exam), and started them with a more tedious assignment in the provinces. Itself, the Secretariat staffed with the best and the brightest. The Secretariat regularly moved judges all around the country, and up and down the judicial hierarchy, usually at 3-year intervals. Some cities were more attractive than others, some jobs were more prestigious than others, and a posting to the Secretariat was perhaps the most prestigious of them all. As a result, by controlling periodic assignments, the Secretariat could reward and punish judges. It could also discriminate in pay: although it could not constitutionally reduce a judge's pay, it did not need to move judges up the pay scale at the same pace. The main use of these incentives was not political. Most of what judges anywhere do has nothing to do with politics. Rather, the Secretariat used career incentives to induce judges to work and to allocate talent to its most appropriate use. By rewarding judges for good performance, the system gave them an incentive to work hard and carefully. The question here is whether notwithstanding this desirable effect, the Secretarit also used the incentives for political ends. The Supreme Court. -- The fifteen justices on the Supreme Court stand apart from this system. The Cabinet appoints Supreme Court justices to serve until mandatory retirement at age 70. They are subject neither to 10-year terms nor periodic reassignment by the Secretariat. Although voters can remove them in any national election, this has never been a serious possibility. Japanese prime ministers generally appoint to the Supreme Court men in their early 60's. For most of the post-war years, the LDP faced very high odds of staying in power. As a result, no prime minister needed to try to extend his influence past the next election by appointing young justices. Instead, his incentive was to appoint older men to avoid the Harry Blackmun problem -- the risk that a politically reliable appointee would over the course of long tenure drift so far politically that he would end by sleeping with the enemy. The Chief Justice serves as administrative head of the lower courts. In this capacity he supervises the Secretariat. As such, he is not, like an elected minister, the titular head of a bureaucracy he does not understand. Usually he headed the Secretariat itself before joining the high court. Usually, in short, he knows exactly what buttons to push to make it respond. The Japanese military. -- To explore the potential political implications of this framework, turn first to the most famous Japanese judicial dispute of all: the constitutionality of the military. Back in 1947, Douglas MacArthur imposed on Japan a constitution that in Article 9 explicitly banned all military forces.2 By 1955, conservative politicians had formed the Liberal Democratic Party and solidified their control over the Diet. Eager to reassert national independence, they tried to eliminate Article 9. The Socialists and Communists, however, blocked this and declared the maintenance of Article 9 the cause of the day. The LDP did not wait for a constitutional amendment to rearm. Instead, it announced that Article 9 banned only offensive forces. Defensive weapons were constitutional, and Japan could validly have a "Self-Defense Force." Other than lacking nuclear weapons, ICBMs, and aircraft carriers, the Self-Defense Force looked much like any other ultra-sophisticated modern military machine. But whether it would survive now turned on what judges thought of the LDP's ok-as-long-as-it's-defensive theory. In the 1960s, nearly 200 citizens of the northern island of Hokkaido sued to injoin a planned missile base. Case assignment in Japanese district courts varies by court, but in order to prevent corruption most use some variant of random assignment. By the exigencies of that assignment process, the SDF case went to Judge Shigeo Fukushima. Fukushima was a member of the Young Jurists League (the YJL). By its own constitution, the YJL was dedicated to preserving the Constitution of Japan. As the discussion above should make clear, this strict constructionism did make the YJL the Japanese analog to the Daughters of the American Revolution. Rather, it was by many accounts a Communist Party affiliate fighting the LDP's attempts to amend the constitution and delete Article 9. Fukushima could have picked any number of administrative-law doctrines to duck the issue, but he chose to confront it head on instead, and enjoined the base. The case went up to the High Court, which promptly reversed. On remand, Fukushima forthrightly declared the Self Defence Force unconstitutional. The case made news in several ways, but for our purposes what matters is what happened to Fukushima. After a stint in Tokyo, he went to a couple of small provincial cities in the northeast. And there he stayed. By 1989, he was 59 and had served in provincial family courts for 12 years running. After writing an op-ed for a national newspaper complaining about how the Secretariat had sandbagged his career, he quit. Disentangling politics from other factors. Fukushima clearly had a bad career. The question is whether it was so bad that one cannot explain it either by bad luck or legal mediocrity. To answer the question, we need both a systematic data base on a wide variety of judges, and a multi-variate test. More specifically, we need a test that measures the effect on his career of his Article 9 opinion, holding all else constant: whether he did worse for his politics -- holding constant his seniority, IQ, work habits, and anything else that might affect career performance. Although we do not teach students how to do these statistical tests in law school, they are a staple of sex- and race-discrimination litigation. For example, suppose one wanted to know whether a law firm discriminated against female associates by pay. One would not want to know whether women earned the same pay as the men. After all, if the firm only started hiring women recently, they would be less experienced than the men, and their low pay could simply reflect that inexperience. Alternatively, even if they earned the same, if the firm's partners hired run-of-the-mill men but only (given their bias) super-star women, then the very fact that the women earned what the men made would be evidence of bias. The classic way to test for bias in this context is to use an "OLS regression" to estimate associate pay as a function of several factors such as sex (1 if male, 0 if female), IQ, hours billed, and seniority. Using data on all available factors and associates, one would estimate -- modern computer technology makes it astonishingly easy -- the coefficients to an equation: PAY = a + b1 SEX + b2 IQ + b3 HOURS + b4 SENIORITY + e, where e represents random error. If, as seems likely, pay rises with IQ, hours billed, and seniority, the coefficients b2, b3, and b4 would all be positive. If the firm paid men higher wages than women, holding constant IQ, hours, and seniority, the coefficient b1 would be positive too. If it paid men and women the same, b1 would equal 0. Now turn to Japanese judges and their political activity. Before asking whether Fukushima suffered for his opinion, however, consider how one might test whether YJL members suffered more generally. And as an index of career success, ask whether they earned salaries as high as their non-League colleagues. We do not have the pay data on individual judges (nor, having written several articles like this, is the Secretariat likely ever to give us that data). We do, however, know how many years it takes for a judge to obtain the status of "sokatsu," an administrative rank that generally comes at a certain senior step on the pay scale. The length of time it takes a judge to obtain "sokatsu" status thus correlates with how fast he has climbed the pay scale. Accordingly, for our dependent variable (the equivalent to PAY in the example above), we use the time it takes a judge to reach "sokatsu" status. To test for political discrimination, we introduce a variable, YJL, that takes the value 1 if a judge was a member and 0 otherwise. We then add several straightforward control variables: SEX, FLUNK (the number of times a judge failed the entrance exam to the LRTI -- a proxy for IQ and hard work), ELITEU (1 if a judge attended either of the two most prestigious universities -- again, a proxy for IQ and hard work), FIRST-TDC (whether the Secretariat thought the judge a star and assigned him initially to the Tokyo District Court -- again a proxy for ability), and FIRST-BO = 1 (whether the Secretariat thought the judge a loser and assigned him initially to a branch office).3 We attach a star, *, to any coefficient that is statistically significantly different from 0 at the 90 percent level or higher (that is, to any coefficient that is reliably different from zero). Based on data on the 500-odd judges hired from 1959 to 1968,4 we obtain the following coefficients: YJL 1.00* SEX -3.42* FLUNK -0.03 ELITEU 0.25 FIRST-TDC -1.06 FIRST-BO 1.45* Recall that we estimated a judge's time to sokatsu as a linear combination of these variables. That the coefficient on sex (1 = male, 0 = female) is negative and statistically significant signals classic sex discrimination: holding constant other indices of judicial ability, a male judge reaches sokatsu status (and hence a given step on the pay scale) more quickly than a woman. Indeed, the value of -3.42 means that a man reaches that point on the pay scale nearly 3-l/2 years before a similarly qualified woman. Likewise, a judge who begins his career at a branch office will reach that point about 1-1/2 years later than his peers. About those coefficients that are not statistically significant, we can say nothing. About YJL membership, however, the statistically significant coefficient of 1.00 indicates that a League member climbed the pay scale a year more slowly than otherwise, holding constant other indices of judicial ability. The military, again. -- Return now to Judge Fukushima. To test for discrimination against him and others who held the military unconstitutional, one would need first to collect all cases involving Article 9. Then, one would construct an index of the quality of judicial postings, and collect a variety of proxies for judicial ability. Finally, holding constant those proxies, one would run a regression similar to the one above on judicial pay, and ask whether judges who held Article 9 unconstitutional received worse career postings than their peers. In a recent paper, we do just that.5 Using data on 50 judges involved in Article 9 disputes, we regress the quality of jobs a judge gets during the decade after deciding an Article 9 case on (i) how he decided that case and (ii) a variety of control variables. We find that systematically the judges who hold the military unconstitutional receive worse postings thereafter. Other disputes. -- We can apply this method to a wide variety of possible political interventions.6 For example: a. Campaign law.7 Japanese electoral law bans door-to-door canvassing. Like most campaign restrictions, the ban helps incumbents and hurts challengers. As the LDP has always had the most incumbents, the ban has helped the LDP and hurt the opposition. The opposition parties have claimed that the ban violated the constitutional guarantee of free speech. The LDP has, of course, claimed otherwise. When opposition candidates challenged the ban in court, some judges held it unconstitutional. Hence the test: to compare the careers of those who held it constitutional with the careers of those who held it unconstitutional. Systematically, we find that judges who held it constitutional did better than judges who did not. b. Apportionment.8 During its first two decades in power, the LDP relied heavily on the rural vote. As time passed, the rural areas became increasingly overrepresented in the Diet. Again, the opposition parties regularly challenged the apportionment in court, and the LDP regularly insisted that all was proper. Here too we find political bias in judicial incentives. During the period when the LDP relied heavily on the rural vote, judges who struck down the electoral districts systematically had worse careers than those who upheld them, holding constant the usual controls. During the 1970s and 1980s, LDP leaders repositioned the party as an organization of urban consumers. Once they did so, judges who threw out the old districting no longer suffered. c. Injunctions.9 Japanese administrative law lets citizens challenge administrative dispositions in court. Even majority parties need to let constituents keep high-handed bureaucrats in line, and courts can come in handy toward that end. Predictably, judges who uphold such routine challenges do not suffer in their careers.10 That said, one would not expect LDP leaders to appreciate judges who injoin national bureaucrats. After all, LDP leaders directly control the national bureaucracy. Although they will not want to police every routine tax audit and license revocation, they will want to determine major issues of policy. If they want a bureaucrat to change policy, they can phone him and tell him to change it. As a result, if a court unilaterally injoins a bureaucrat against changing national policy, it is probably thwarting LDP goals. To test how LDP leaders see injunctions against national bureaucrats, we ask how well judges do who issue the injunctions. We find that systematically those who injoin them do worse in their careers than those who deny such challenges, holding constant the usual controls. Once again, judges face politically biased incentives. Historical circumstances also allowed us to tell whether judges were penalized by the Secretariat simply for being activist, rather than for being activist against the LDP. During this period, leftists increasingly took over local governments. Judges, however, answered to the national LDP-controlled government. As a result, under our theory of political influence one would not expect judges to suffer for injoining municipal or prefectural bureaucrats. There, the judge would be more likely to be injoining socialist or communist policy. And systematically, such judges did not suffer. Convictions.11 -- Japanese courts convict roughly 98 percent of criminal defendants. To test whether this too might result from biased judicial incentives, we compared judges who acquit sometimes with those who acquit never. Using our usual regression method, we find that judges who acquit have worse careers. Crucially, however, this turns out to be an example of why one needs to combine simpler forms of thinking about data with the computer printouts of regression analysis. The judges driving the punishment results are not judges who simply decided that the police had nailed the wrong man. They are -- once again -- political odd ducks: judges who acquitted communist candidates of electoral law violations on constitutional grounds, for example, or judges who acquitted demonstrators of beating a police officer on the grounds that the officer (while being beaten) had bungled his arrest procedures. Even here, the dynamic is fundamentally political, and everyday judicial decisionmaking shows no signs of political influence, or even of a Secretariat bent on imposing its judicial philosophy. Conclusions: Japan is a wealthy capitalist democracy. As befits such a country, most elections are fair, most police are clean, and most judges are honest. Not only are most judges honest, but most are exceptionally able, and for most decisions they have occasion to make, the constitutional protection of their independence seems to be protection in practice as well as in theory. Yet most judges are also politically conservative, for politics can intrude even where the Constitution guarantees independence. Given the Japanese democratic system, majority party leaders choose the cabinet; the cabinet chooses the Supreme Court; the Supreme Court supervises the Secretariat; and the Secretariat decides which judge works where for how long and at what pay. Such a system allows for merit selection of judges and for keen incentives for them to work hard and well, but by this very fact it also allows the Secretariat leeway to derail the careers of wayward judges. The LDP controlled the cabinet without interruption from 1955 to 1993. Inevitably, personnel policies in the courts have come to reflect the policy preferences of the LDP. Put differently, a college graduate who took his Marxism seriously was not likely to find a career in the courts much fun. # # 30 # # * Mitsubishi Professor of Japanese Legal Studies, Harvard Law School and Professor of Business Economics and Public Policy, Indiana University, Kelley School of Business. This research was funded in part by the John M. Olin Program in Law, Economics & Business, Harvard Law School. The working paper version can be found at http://Php.indiana.edu/~erasmuse/@Articles/Unpublished/judicat.pdf. 1 J. Mark Ramseyer & Eric B. Rasmusen, Judicial Independence in a Civil Law Regime: The Evidence from Japan, 13 J. Law, Econ. & Org. 259 (1997); J. Mark Ramseyer & Eric B. Rasmusen, Why the Japanese Taxpayer Always Loses, 72 S. Cal. L. Rev. 571 (1999); J. Mark Ramseyer & Eric B. Rasmusen, Why Are Japanese Judges so Conservative in Politically Charged Cases? (Working Paper, John M. Olin Center for Law, Economics, and Business, Harvard Law School, 1999,http://Php.indiana.edu/~erasmuse/@Articles/Unpublished/jappub.pdf). J. Mark Ramseyer & Eric B. Rasmusen, Why Is the Japanese Conviction Rate so High? (Working Paper, John M. Olin Center for Law, Economics, and Business, Harvard Law School, 1998,http://Php.indiana.edu/~erasmuse/@Articles/Unpublished/japcon.pdf). 2 "[L]and, see, and air forces, as well as other war potential, will never be maintained." 3 We also included dummies for the year of hiring to capture year fixed effects. We omit the coefficients here. 4 We perform a variety of adjustments to the data to avoid potential sample bias. These are detailed in Why Are Japanese Judges so Conservative, supra note 1. The adjusted R2 for the regression is 0.11. 5 Why Are Japanese Judges so Conservative, supra note 1 We use a dependent variable that represents the fraction of time during the decade after an opinion that a judge spent in various posts. Because the dependent variable is censored by 0 and 1, we use tobit regressions. 6 Again, we use a dependent variable that represents the fraction of time during the decade after an opinion that a judge spent in various posts. Because the dependent variable is censored by 0 and 1, we use tobit regressions. 7 Judicial Independence, supra note 1. 8 Why Are Japanese Judges so Conservative, supra note 1. 9 Why Are Japanese Judges so Conservative, supra note 1. 10 Why the Japanese Taxpayer, supra note 1. They are indeed punished, however, if the opinion is reversed on appeal. 11 Why Is the Japanese Conviction Rate, supra note 1. Ramseyer & Rasmusen: Page 18