The purpose of this study is to investigate the influence of Metropolitan and Provincial Offices of Education on school performance and predict how school performance dynamically changes with the variables of Metropolitan and Provincial Offices of Education. The subjects of this study are the followings:

First, how much can school performance be explained by the variables of Metropolitan and Provincial Offices of Education? Second, what kinds of variables of Metropolitan and Provincial Offices of Education can explain differences of school performance after controlling the influence of school-level variables? How much the power of explanation do they have? Third, does the dynamic change of school performance change with an effect of the variables of policy and finance of Metropolitan and Provincial Offices of Education?
Subordinate criteria of the school performance are academic achievement, personality, students'' dropouts, students'' satisfaction, parents'' satisfaction, and educational transition. The variables affecting the school performance are classified into two things: those of school and Metropolitan and Provincial Offices of Education. Background variables of school-level are establishment type and the total number of students. Policy variables are autonomous school, after school program, student counsel, and autonomy club. Moreover, the financial variable is student welfare·educational gap reduce cost. A background variable of Metropolitan and Provincial Offices of Education is types of cities and provinces. Policy variables are activation of teacher training, favorable conditions for education (that is, lightening administrative duties of teachers), and prevention and extermination of school violence. And its financial variables are a percentage of expansion of basic operating expenses of schools and percentage of investment of improvement cost of educational environment.

This study is designed for a multi-level analysis of level two under the hypothesis including variables of school and Metropolitan and Provincial Offices of Education that can affect school performance. In order to test the model, 16 Metropolitan and Provincial Offices of Education and 606 of 1,520 general high schools (40%) were used as a sample. In addition, system dynamics, one of the subordinate criteria of school performance, used to predict the dynamic change of academic achievement of general high schools.

The results of this study are the followings:

First, variances of Metropolitan and Provincial Offices of Education by subordinate criteria of school performance were academic achievement (12.330%), personality (16.564%), students'' dropouts (4.770%), students'' satisfaction (21.617%), parents'' satisfaction (17.993%), and educational transition (32.907%). This means that the difference of school performance is statistically significant to Metropolitan and Provincial Offices of Education.
Second, after controlling school-level variables, the difference of Metropolitan and Provincial Offices of Education by subordinate criteria of school performance was statistically significant. However, the difference was varied with other variables.
Third, types of dynamic change of academic achievement of general high schools were different with the amount of input of variables of Metropolitan and Provincial Offices of Education. The results of policy experiment are based on the main model are as follows: activation of teacher training was the most influential variable in academic achievement of general high schools. The next most influential variable was efficiency of the local educational finance and the third one was lightening teachers'' administrative duties.

Key words: school performance, Metropolitan and Provincial Offices of Education, general high school, dynamic change, academic achievement, personality, students'' dropouts, students'' satisfaction, parents'' satisfaction, educational transition, multi-level analysis, system dynamics