(Page 2 of 24) - Judging Political Promotion of Judges: Survival Analysis, Split Population Model and Matching Method authored by Fukumoto, Kentaro.

INTRODUCTION

The Japanese Constitution stipulates that all judges shall be independent in the exercise

of their conscience and shall be bound only by the Constitution and the laws. At the same time,
it provides the Cabinet with the power to designate a Chief Supreme Court Judge (appointed by
the Emperor) and to appoint other Supreme Court Judges. The judges of the inferior courts
shall be nominated by the Supreme Court and appointed by the Cabinet, while no executive
organ is empowered to administer disciplinary action against judges. Ramseyer and Rasmusen
(2003) argue that, despite the Constitution, Japanese judges routinely validate what the
government has done, and question the conventional view in which the Japanese judicial system
is independent of political control. In order to substantiate their view, they have shown in a
series of statistical examinations that judges with leftist preferences do worse in their careers.
While agreeing on the importance of analyzing judicial independence, we suggest that their
methods are problematic.

To be concrete, Ramseyer and Rasmusen examine whether or not a judge who once

belonged to a leftist group took longer to reach the moderately prestigious administrative status
known as sokatsu, hereafter we call a “division chief judge (DCJ)”.

In order to regress the

time to DCJ in an OLS model, they drop some judges from the career data set and substitute
arbitrary values for those who even did not reach DCJ. The problem with this data
manipulation is not only to cause bias in the estimates, but also to ignore the time-dependence
of an event such as a judge obtaining a DCJ appointment.

Survival analysis provides an efficient solution to simultaneously deal with the issues of

censoring and time-dependence. In particular, we consider the possibility that some judges
may have a chance to become DCJ, while others may have no chance at all. To account for the
problem of incorrectly assuming all observations will eventually make a transition, we apply a
split population model to the data set of judicial career. Also, we attempt to deal with the issue
of what is called “left truncation,” an issue rarely addressed in the context of split population
modeling. At the same time, we take causal inference seriously and match every observation
with its similar one so that our survival analysis is less dependent on the particular models we
specify.

This paper proceeds as follows. We begin with a brief review of the argument made

by Ramseyer and Rasmusen (2003) and a replication of the statistical analysis reported in their
book. The second section reconsiders their analysis. First, we correct their career data and
report the OLS results using the corrected data set. Second, we turn to an explanation of

A judge with this status is the head of the division of a court and eligible to preside court for

the cases requiring consultation with associate judges.